and ESEARCH · forward the concept of “investor awareness” which states that increased investor awareness for added stocks and a corresponding smaller drop in the awareness for

RThe International Journal of

Business and FinanceESEARCH

CONTENTS

New Evidence from S&P 500 Index Deletions 1Rashiqa Kamal

The Interplay between Director Compensation and CEO Compensation 11Dan Lin & Lu Lin

The Relationship between Brand Image and Purchase Intention: Evidence from Award Winning Mutual Funds 27Ya-Hui Wang & Cing-Fen Tsai

Time Series Modeling and Forecasting Inflation: Evidence from Nigeria 41Ikechukwu Kelikume & Adedoyin Salami

Capital Structure Determinants of Publicly Listed Companies in Saudi Arabia 53Turki SF Alzomaia

Cost Efficiency of Ghana’s Banking Industry: A Panel Data Analysis 69Kofi Adjei-Frimpong, Christopher Gan & Baiding Hu

Effects of Service Innovation on Financial Performance of Small Audit Firms in Taiwan 87Yi-Fang Yang, Lee-Wen Yang & Yahn-Shir Chen

The Association between Firm Characteristics and Corporate Financial Disclosures: Evidence from UAE Companies 101Khaled Aljifri, Abdulkareem Alzarouni, Chew Ng & Mohammad Iqbal Tahir

VOLUME 8 2014NUMBER 2

The International Journal of Business and Finance Research ♦ VOLUME 8 ♦ NUMBER 2 ♦ 2014

1

NEW EVIDENCE FROM S&P 500 INDEX DELETIONS Rashiqa Kamal, University of Wisconsin-Whitewater

ABSTRACT

Kamal, Lawrence, McCabe, and Prakash (2012) argue that information asymmetry exists in the financial markets and additions to S&P 500 Index convey new information about the added firms to the uninformed investors. They further argue that because of important changes and regulations in the financial markets, like, Regulation Fair Disclosure, Sarbanes-Oxley Act, and Decimalization of the exchanges, in or after the year 2000, information asymmetry has decreased. In support of their arguments, they find that for additions, the positive abnormal returns on announcement day have decreased, and added stocks’ liquidity changes have become marginal in the post-2000 period. We extend their work and for a sample of deletions between October 1989 and December 2011, we find that the negative abnormal returns on the announcement day are not significantly different in the post-2000 period, but the negative returns are reversed earlier in the post-2000 period. Contrary to our expectation, liquidity changes after deletion are significant in the post-2000 period. However, when we divide our sample into optioned versus non-optioned stocks and control for other factors that affect liquidity, we find that liquidity changes after deletion are insignificant in the post-2000 period. JEL: G12; G14 KEYWORDS: S&P 500 Deletions, Information Asymmetry, Liquidity Changes INTRODUCTION

esearch on additions to and deletions from the Standard and Poor’s (S&P) 500 Index has recorded significant abnormal returns (Shleifer, 1986, Beneish and Whaley, 1996, 2002, Lynch and Mendenhall, 1997, Dash, 2002, and Chen, Noronha and Singal 2004, 2006a, b) and changes in

liquidity (Beneish and Whaley, 1996, Erwin and Miller, 1998, and Hegde and McDermott, 2003) around these events. Several theories have been put forward to explain the price effects around index changes. The downward-sloping demand curve hypothesis (Shleifer, 1986) argues that index funds, which buy stocks added to the S&P 500 to replicate the market, drive price movements associated with S&P 500 inclusion announcements. However, the information hypothesis (Jain, 1987, Dhillon and Johnson, 1991, Denis, McConnell, Ovtchinnikov and Yu, 2003) states that inclusion in the index conveys positive information about the stock and thus, drives up the price. Cai (2007) argues that addition to the S&P 500 Index conveys favorable information about the stock and the industry and hence, can be considered a partial explanation for the positive price effect of additions to the S&P 500. Recently, Kamal, Lawrence, McCabe and Prakash (2012) argue that additions to the S&P 500 provide information about the performance and future prospects of a firm and hence, reduce the informational asymmetry amongst investors, resulting in a significant positive abnormal return and an increase in stock liquidity on the announcement of firm’s addition to S&P 500 (pg. 381). Kamal et al. (2012) further argue that the information environment has changed after the year 2000 due to important regulations and changes in the financial markets. They categorize the additions to the S&P 500 Index into two sub-periods, namely, pre- and post-2000 periods and show that for additions to the S&P 500, abnormal returns have reduced in the post-2000 period, whereas, liquidity of the added firms’ stock has become marginal. In light of their arguments, it would be worthwhile to test whether deletions from S&P 500 also experience similar changes in abnormal returns and liquidity around the announcement and effective days in pre- and post-2000 periods. Even though there is a plethora of literature around S&P 500 Index changes, Chen, Noronha and Singal (2004) rightly observe that most of the researchers have focused on

R

R. Kamal | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

2

additions. Hegde and McDermott (2003) argue that deletions have unique characteristics that make them noisier events than additions to study (pg. 452). Even though the two aforementioned studies have attempted to address deletions, as such, it is fair to say that there is still a void in the literature regarding the study of deletions from the S&P 500. A recent study by Ivanov (2010) uses a sample of discretionary deletions between October 1989 and December 2007 and shows that firms with analyst following, after deletion, experience an increase in earnings forecasts and actual forecasts, which is contrary to the predictions of the information hypothesis. If Kamal et al (2012)’s observation that the information environment has changed after the year 2000 is accurate, then it could be possible that Ivanov (2010)’s results are affected by this phenomenon. The purpose of this study is to extend the work of Kamal et al (2012) to deletions from the S&P 500 Index, in an attempt to fill the void in this area of research. We study deletions from the S&P 500 Index between October 1989 to December 2011, and divide the sample period into pre- and post-2000 periods. Researchers have found significant negative abnormal returns around the announcement day and effective day of stocks deleted from the S&P 500 (Lynch and Mendenhall, 1997; Dash, 2002; Beneish and Whaley, 2002), and significant decrease in liquidity after announcement, of deleted stocks (Beneish and Whaley, 1996, and Hegde and McDermott, 2003). Extending the argument of Kamal et al (2012) to deletions, due to reduced information asymmetry in the post-2000 period, we expect the negative abnormal return on the announcement day of deletions to be significantly smaller or marginal in the post-2000 period, as compared to the pre-2000 period. We also expect the increase in the relative bid-ask spread (or the decrease in liquidity) after deletion to be marginal in the post-2000 period, as compared to the pre-2000 period. Using standard event-study methodology, we calculate the average abnormal returns on announcement day for deletions and compare them across the pre- and post-2000 periods. We find that for the sample of deletions, the average abnormal return on the announcement day is negative and insignificant in the post-2000 period but it is not significantly different from the negative average abnormal return in the pre-2000 period. The same is true for the cumulative abnormal returns between the announcement day and the effective day. However, we find that in the post-2000 period the negative cumulative abnormal returns are reversed in 20 days after the effective day, as compared to in 60 days after the effective day in the pre-2000 period. We argue that the early reversal of abnormal returns in the post-2000 period can be attributed to reduced information asymmetry in the post-2000 period, as found for additions to S&P 500, by Kamal et al (2102). Then we calculate and compare the average relative bid-ask spreads before and after announcement of deletion, in the pre- and post-2000 periods. Contrary to our expectation, we find that even though the relative spread is smaller in magnitude in the post-2000 period, the increase in spread is significant in the post-2000 period. Erwin and Miller (1998) and Kamal et al (2012) argue that due to informational efficiencies already achieved by optioned stocks, optioned and non-optioned stocks behave differently on addition to the S&P 500 Index. Following them, we separate our sample of deleted stocks into optioned and non-optioned stocks, that is stocks that were trading options at the time of announcement of deletion, and stocks that were not trading options at the time of announcement of deletion, and compare the changes in relative spread before and after announcement, in the pre- and post-2000 periods, by setting up a multivariate regression. After controlling for price, volume, and return variance, as expected, we find that the changes in spread of non-optioned stocks are insignificant in the post-2000 period. However, we are cautious in evaluating these results because of the limited sample size for the pre-2000 period. Overall, our results show some evidence that the information environment has changed in the post-2000 period, with respect to deletions from the S&P 500 Index. The rest of the paper is organized in the following way: the next section reviews the literature briefly, followed by the section on data and methodology. Results are discussed after that, and the last section concludes.


3

LITERATURE REVIEW The abnormal returns followed by additions to or deletions from the S&P 500 Index have been a widely researched phenomenon. Several authors have put forward various theories to explain these returns. One of the explanations for the price changes around additions to and deletions from the S&P 500 index is the price pressure effect. Evidence of temporary price pressure is presented by Harris and Gurel (1986) who find a reversal of the initial price response, associated with additions to or deletions from the S&P 500. Lynch and Mendenhall (1997) show that a part of the price decrease resulting from deletions, remains permanent and that it cannot be explained by price pressure effect. Dash (2002) shows that the short-term price reactions, associated with deletions from the index are reversed within six days. Chen, Noronha and Singal (2004, 2006a) find a temporary price effect for deletions from the S&P 500 index and they put forward the concept of “investor awareness” which states that increased investor awareness for added stocks and a corresponding smaller drop in the awareness for deleted stocks can explain asymmetric price effect around additions to and deletions from S&P 500 Index. However, Beneish and Whaley (2002) present evidence that suggests that deleted firms do not fully recoup their losses, thus they might have information content. Denis, McConnell, Ovtchinnikov and Yu (2003) examine a sample of additions between 1987 and 1999 and find that these firms experience an increase in realized earnings per share and forecasted earnings per share. As such, their findings support the information hypothesis of the price reaction to index additions. Extending their work, Ivanov (2010) tests whether discretionary deletions from the S&P 500 display similar information content and finds that contrary to the predictions of the information hypothesis, the earnings forecasts and actual earnings of deleted firms increase, on average. Recently, Kamal, Lawrence, Prakash and McCabe (2012) analyze additions to the S&P 500 index by dividing them into pre- and post-2000 periods and find that the positive abnormal returns around the announcement day decreased significantly in the post-2000 period. They also find that the change in the liquidity of added stocks has become marginal in the post-2000 period. They attribute these findings to the changes in the information environment in the post-2000 period. They argue that passage of important regulations in the post-2000 period (namely, Regulation Fair Disclosure in October 2000, Decimalization of NYSE and NASDAQ in 2001, and Sarbanes-Oxley Act in October 2002) has decreased the information asymmetry in this period and hence, announcements of additions to S&P 500 are not as informative as they used to be in the pre-2000 period. Another possible explanation for the price effects is the liquidity hypothesis that states that inclusion in an index may have valuation consequences because it increases a stock's liquidity. The supporters of this view (like, Erwin and Miller, 1998) argue that inclusion may result in greater institutional interest in the stock leading to an increase in public information about it. As a result, the stock will be held more widely, will become more liquid and the bid-ask spread will fall which lowers the required rate of return on the stock and leads to a price increase. Erwin and Miller (1998) use a sample of 109 additions over the period 1984-1989 and examine the changes in stock liquidity when the stock is added to S&P500 and find a significant decrease in the bid-ask spread upon addition to S&P 500 for the stocks that were not trading listed options. They also find that these liquidity effects are mitigated for those stocks that were already trading listed options and the reduction in bid-ask spread is more prominent for the non-optioned stocks. Beneish and Whaley (1996) study a sample of 103 additions between 1989 and 1994 and find that spread decreases after announcement. Hegde and McDermott (2003) use a sample of 74 (27) firms over the period 1993-1998 and find a sustained increase (decrease) in the liquidity of added (deleted) stocks. The explanation for the change in liquidity is supported by sound theoretical arguments. According to Shleifer (1986), addition of stocks to S&P 500 may result in closer scrutiny of firm by analysts and investors leading to greater institutional interests, large trading volumes and lower bid-ask spreads. Hegde and McDermott (2003) argue that change in the composition of equity ownership may increase the


4

proportion of liquidity-motivated traders and greater competition amongst informed traders leading to improvement in the liquidity of added stock. DATA AND METHODOLOGY The S&P 500 index started pre-announcing index changes beginning in October 1989; hence, in this paper, the sample period covers deletions from October 1989 to December 2011. We examine the abnormal returns around announcement (AD) and effective days (ED) for the entire period and then split up the period into pre- and post-2000 years. Following Kamal et al (2012), we define deletions occurring up to September 2000 as the pre-2000 period and deletions occurring after that period as the post-2000 period. We exclude deletions announced in October 2000 when Regulation Fair Disclosure was implemented. Some of the reasons that S&P removes a company from the 500 Index are the filing of Chapter 11 by the company or approval of an alternative recapitalization plan by the shareholders of the company that changes the company’s debt/equity mix or cessation of the company in its current form due to reasons like mergers, acquisitions, and takeovers. Chen et al (2004) argue that due to requirement of post-announcement data, the sample will be biased because firms that cease to exist after the announcement will not be included in the sample. Hence, we follow the methodology of Chen et al (2004) to create a “survivorship bias” free sample, and after excluding deletions that resulted from mergers, acquisitions, spinoffs, bankruptcies, liquidation proceedings, and leveraged buyouts, our sample consists of 120 deletions. Fifty-one of these are announced in the pre-2000 period and 69 are announced in the post-2000 period. Daily returns required to calculate the abnormal returns are obtained from Center for Research in Security Prices (CRSP). Firms for which returns were not available for 245 days prior to announcement date were dropped because these returns were used to compute the beta for calculation of risk-adjusted returns. After imposing these data requirement restrictions, and one outlier, the final sample consists of total 115 deletions, out of which 51 were announced in the pre-2000 period and 64 were announced in the post-2000 period. For liquidity tests, to test for a change in the bid-ask spread when a stock is deleted from S&P 500 we obtain daily bid and ask closing quotes during the period 30 trading days before and 30 trading days after the announcement of deletion for each stock. We eliminate all firms with data less than 58 days. Daily stock spread data, stock price, return and trading volume data are obtained from the CRSP database. After excluding the firms with unavailable data there are 42 firms in our pre-2000 sample and 48 firms in the post-2000 sample, for liquidity tests. Information of option trading is obtained from “CBOE Equity Option Volume Archive”, http://www.cboe.com/data/AvgDailyVolArchive1998.aspx, and since this information is available only for the year 1998 and forward, our pre-2000 sample is reduced to only 15 deletions. Out of these 15 deletions, eight stocks trade options at the time of deletion announcement, and seven do not trade options. In the post-2000 sample, at the time stocks were deleted from the S&P 500 index, 42 were trading listed options while 6 were not. This paper basically follows the methodology of Kamal et al (2012). To calculate the average abnormal returns (AARs) and the cumulative abnormal returns (CARs), we use the standard event-study methodology, with the announcement date of deletion as the event date. To compare liquidity changes around announcement date of deletions, in the pre- and post-2000 period, we calculate the absolute spread as the difference between the ask and bid prices. The relative spread is the absolute spread divided by the mean of the ask and bid prices. Similar to Erwin and Miller (1998) and Kamal et al (2012), we test for changes in bid-ask spread while controlling for share price, trading volume and return variance. Return variance on day t is estimated using the variance of the stock’s return over the five-day period immediately preceding day t. The following multivariate model is estimated: 𝑆𝑝𝑟𝑒𝑎𝑑𝑖𝑡 = 𝜆0 + 𝜆1𝑇𝑖𝑚𝑒𝐷𝑢𝑚𝑚𝑦𝑖𝑡 + 𝜆2𝑃𝑟𝑖𝑐𝑒𝑖𝑡 + 𝜆3𝑉𝑜𝑙𝑢𝑚𝑒𝑖𝑡 + 𝜆4𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒𝑖𝑡 + 𝜀𝑖𝑡 (1)


5

where Spreadit, Priceit, Volumeit, and Varianceit, are the relative bid-ask spread, the closing share price, the trading volume and an estimate of return variance, respectively. TimeDummyit is a dummy variable which equals 1(0) in the 30 trading days before (after) announcement day for firm i on day t. Table 1 presents the summary statistics for 42 (48) firms deleted from the S&P 500 Index in the pre (post)-2000 period, 30 trading days before, and 30 trading days after the announcement of deletion. The table reports the mean, median and standard deviation for the relative spread, the closing share price, the adjusted volume, and the return variance. Table 1: Summary Statistics for 90 Firms Deleted from the S&P 500 Index between October 1989 and December 2011

Pre-2000 period (42 firms) Post-2000 period (48 firms) Mean Median Stdev Mean Median Stdev

30 days before deletion Relative Spread

0.030 0.025 0.022 0.005 0.002 0.010

Price 20.18 13.38 23.83 15.88 11.65 14.93 Volume 585,239 124,000 1,707,550 6,324,206 2,391,900 22,775,069 Return

variance 0.001 0.0004 0.001 0.004 0.001 0.15

30 days after deletion Relative Spread

0.032 0.024 0.026 0.006 0.002 0.010

Price 17.86 13.13 23.27 13.86 10.50 13.58 Volume 741,862 244,000 1,727,751 9,630,464 3,287,450 26,453,010 Return

variance 0.001 0.001 0.002 0.009 0.001 0.037

This table shows the mean, median and the standard deviation for the relative bid/ask spread, the closing share price, adjusted volume, and the return variance for 42(48) firms deleted in the pre(post)-2000 period, 30 trading days before and 30 trading days after the announcement of deletion. The relative spread is the absolute spread (difference between ask and bid prices) divided by the mean of the ask and bid prices. Return variance on day t is estimated using the variance of the stock’s return over the five-day period immediately preceding day t. RESULTS AND DISCUSSION Panel A of Table 2 shows that for the entire sample of 115 deletions from October 1989 to December 2011, the announcement day average abnormal return is -3.26% and it is significant at the 1% level. This result is consistent with previous research (like, Chen et al, 2004, find significant abnormal returns equal to -7.82% for 16 deletions between March 1990 and April 1995). Panel B of Table 2, reports the announcement day average abnormal return for the pre-2000 period sample as significant -4.34%, and the post-2000 sample has an insignificant return of -2.40%. However, these returns are not significantly different (t-stat=1.16). This shows that our results for deletions do not support the results for additions to S&P 500 as presented by Kamal et al (2012), and we find that the negative abnormal returns around the announcement days of deletions are not significantly different in the pre- and post-2000 periods. Table 2: Announcement Day Average Abnormal Return For Deletions from the S&P 500 Index

Panel A: For all deleted firms 1989-2011 (115 firms) AAR t-stat -3.26%*** 3.59 Panel B: Comparison of Pre- versus Post-2000 Pre-2000 (51 firms) Post-2000 (64 firms) Comparison of Pre- versus Post-2000 AAR t-stat AAR t-stat t-stat -4.34%*** 7.23 -2.40% 1.54 1.16

This table presents the announcement day average abnormal return (AAR) for firms deleted from the S&P 500 during October 1989-December 2011, in Panel A, and in the pre- and post-2000 period, in Panel B. Deletions occurring up to September 2000 are defined as the pre-2000 period and deletions occurring after that period as the post-2000 period. AARs are calculated using the standard event-study methodology. ***, **, and * indicate significance at the 1, 5 and 10 percent levels respectively. Next, we examine the permanence of the negative abnormal returns. Table 3 presents the cumulative abnormal return for deletions between the announcement day and the effective day, and 20 and 60 days


6

after the effective day. Previous research has shown that the negative announcement day average abnormal returns around deletions from the S&P 500 are not permanent and that the effect is reversed shortly (Chen et al, 2004, find that for 46 deletions between 1996 and 2001, the stocks recover their losses 60 days after the effective day). For this analysis, our sample is reduced to 106 firms because we need data after 60 days from the effective day. Panel A of Table 3 shows the cumulative abnormal return for the entire sample period and overall our results support the findings in the current literature. We find that the negative cumulative abnormal returns are reversed within 20 days (0.22%) after the effective day but it is significantly reversed only after 60 days (0.27%) of the effective day. Panel B of Table 3 presents the cumulative abnormal return for the deletions in the pre- and post-2000 period. The results show that the cumulative abnormal return between announcement day and effective day is negative and significant in both pre- and post-2000 periods, but they are not significantly different from each other. The interesting result here is that the negative abnormal returns are reversed earlier in the post-2000 period. Panel B of Table 3 shows that the cumulative abnormal return between announcement day and 20 days after effective day is significant -0.20%, and in the post-2000 period, it is significant 0.58%. These returns are significantly different from each other. Furthermore, the cumulative abnormal returns between the announcement day and 60 days after the effective day are insignificant in the pre-2000 period but positive and significant in the post-2000 period. This result suggests that in the post-2000 period, the negative returns around deletions do not last as long as in the pre-2000 period. This result indicates that probably the announcement of deletion does not convey much new information in the post-2000 period, or information asymmetry has decreased in the post-2000 period. Table 3: Cumulative Abnormal Return for Deletions from the S&P 500 Index

Panel A: For all deleted firms 1989-2011 (106 firms) CAR t-stat

AD -3.99%*** 3.69 AD to ED -2.54%*** 3.63

AD to ED+20 0.22% 1.42 AD to ED+60 0.27%*** 3.24

Panel B: Comparison of Pre- versus Post-2000 Pre-2000 (49 firms) Post-2000 (57 firms) Comparison of Pre- versus

Post-2000 CAR t-stat CAR t-stat t-stat

AD -4.21%*** 7.23 -3.79%** 1.94 0.20 AD to ED -2.20%*** 6.15 -2.83%*** 2.23 0.48

AD to ED+20 -0.20%*** 2.15 0.58%*** 2.15 2.73*** AD to ED+60 0.03% 0.53 0.47%*** 3.36 2.90***

This table presents the cumulative abnormal return (CAR) for firms deleted from the S&P 500 during October 1989-December 2011, in Panel A, and in the pre- and post-2000 period, in Panel B. Deletions occurring up to September 2000 are defined as the pre-2000 period and deletions occurring after that period as the post-2000 period. CARs are calculated using the standard event-study methodology. ***, **, and * indicate significance at the 1, 5 and 10 percent levels respectively. Overall, from the results presented in Tables 2 and 3, we do not find that the abnormal returns around the announcement of deletions from the S&P 500 have changed much in the pre- and post-2000 period. Even though our results do not show significant difference in the two sub-periods, for completeness’ sake, following Kamal et al (2012), we also estimate a multivariate regression with the announcement day average abnormal return as the dependent variable and a dummy variable for the pre-2000 period, a dummy variable for technology firms, a dummy variable for the exchange on which the stock is listed, a dummy variable for bull market, log of the relative size, and shadow cost, as independent variables (pg. 391). We did not find any significance in this regression; hence, the results are not reported here. Another way that we want to test whether the information content of deletions’ announcement has changed in the post-2000 period is to examine the change in liquidity of the deleted stocks before and after announcement, in the pre- and post-2000 period. Table 4 presents the results of this examination. We


7

calculate the relative bid-ask spread for the 30 days before and 30 days after the announcement of deletion for the entire sample, and for the pre- and post-2000 sample periods. Column 1 of Table 4 shows that for the entire sample, the relative bid-ask spread for deleted stocks significantly increases in the period after the announcement of the deletion. Columns 2 and 3 of Table 4 report the relative bid-ask spread in the pre- and post-2000 periods, respectively. We find that the relative spread increases (hence, liquidity decreases) in both the sub-samples but the spread increases significantly in the post-2000 period. This finding is contrary to our expectation that if information asymmetry has reduced in the post-2000 period, there should be insignificant increase in the spread of deleted stocks in the post-2000 period. Table 4: Relative Bid-Ask Spread before and after Announcement of Deletion from S&P 500 Index

Complete Sample (1989-2011) Pre-200 period Post-2000 period Spread before index deletion 0.0168

0.0303 0.0049

Standard Deviation 0.0208 0.0217 0.0100 Spread after index deletion 0.0181 0.0317 0.0061

Standard Deviation 0.0228 0.0257 0.0098 T-stat for equality 2.23 1.48 3.13

p-value 0.03 0.14 0.002 Number of stocks 90 42 48

This table presents the average relative bid-ask spreads before and after deletions from the S&P 500 index. Relative spread is the absolute spread divided by the mean of the ask and bid prices. Average spread is the difference between ask and bid prices. The spreads prior to deletion are calculated over the 30 trading days before announcement date of deletion and the spreads in the post deletion period are calculated over 30 trading days after the announcement date of deletion. Column 1 presents the relative bid-ask spread for the entire sample, and columns 2 and 3 present the spreads in the pre- and post-2000 periods, respectively. Research has shown that spread can also be affected by other factors, so next, we estimate regression equation (1) described in the previous section. 𝑆𝑝𝑟𝑒𝑎𝑑𝑖𝑡 = 𝜆0 + 𝜆1𝑇𝑖𝑚𝑒𝐷𝑢𝑚𝑚𝑦𝑖𝑡 + 𝜆2𝑃𝑟𝑖𝑐𝑒𝑖𝑡 + 𝜆3𝑉𝑜𝑙𝑢𝑚𝑒𝑖𝑡 + 𝜆4𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒𝑖𝑡 + 𝜀𝑖𝑡 (1) where Spreadit, Priceit, Volumeit, and Varianceit, are the relative bid-ask spread, the closing share price, the trading volume and an estimate of return variance, respectively. TimeDummyit is a dummy variable which equals 1(0) in the 30 trading days before (after) announcement day for firm i on day t. Results are presented in Table 5. The univariate results in Table 4 are also supported by the multivariate results presented in Table 5. TimeDummy is a dummy variable that is equal to 1 (0) in the 30 days trading day before (after) announcement of deletion from the S&P 500 Index. According to our hypothesis, we expect this time dummy variable to be negative (indicating an increase in spread after announcement of deletion) and insignificant in the post-2000 period. In Table 5, we find that the in the post-2000 period, the increase in the spread after deletion is significant at the 10% level, whereas, it is insignificant in the pre-2000 period. This is contrary to our expectation. Table 5: Multivariate Regression for Relative Spread of Deletions from the S&P 500 Index

Pre-2000 period Post-2000 period Intercept 0.032*** 0.008***

Time Dummy 0.0004 -0.0007* Price -0.0002*** -0.0002***

Volume -0.000*** 0.000 Variance 2.935*** 0.037***

Adjusted R2 0.1390 0.0742 This table presents the estimates for the following multivariate model: Spreadit=λ0+λ1TimeDummyit+λ2Priceit+λ3Volumeit+λ4Varianceit+εit where Spreadit, Priceit, Volumeit, and Varianceit, are the relative bid/ask spread, the closing share price, the trading volume and an estimate of return variance, respectively. TimeDummyit is a dummy variable which equals 1(0) in the 30 trading days before (after) announcement day for stock i on day t. Return variance on day t is estimated using the variance of the stock’s return over the five-day period immediately preceding day t. ***, **, and * indicate significance at the 1, 5 and 10 percent levels respectively. Erwin and Miller (1998) show that changes in liquidity after addition to the S&P 500 index can also be affected by option trading status of the stock at the time of announcement because they find that optioned


8

stocks show no change in liquidity while the non-optioned stocks show a significant increase in liquidity after addition. They argue that this happens because optioned stocks have already achieved informational efficiencies, while non-optioned stocks have not. Kamal et al (2012) argue that in the post-2000 period, if information asymmetry decreases then non-optioned stocks should show diminished or no liquidity changes. This argument should be true for deletions as well. We also extend our analysis and examine liquidity changes of optioned versus non-optioned stocks in the pre- and post-2000 period. Results are presented in Tables 6 and 7. However, we need to point out a limitation of these results. Since the data on option trading is obtained from Chicago Board of Option Exchange’s options archives page and it reports this information only for the year 1998 and forward, the number of deleted firms in the pre-2000 period (15 firms) is small as compared to in the post-2000 period (48 firms). The results should be viewed in light of this limitation. In Table 6, the results indicate that in the pre- and post-2000 periods, the spread for optioned stocks increases significantly, or the liquidity decreases, after the announcement of deletion. For non-optioned stocks, again contrary to expectation, the spread increases significantly after the announcement of deletion from the index. Table 6: Relative Bid-Ask Spread before and after Announcement of Deletion for Optioned and Non-Optioned Stocks

Pre-2000 period Post-2000 period Optioned Non-optioned Optioned Non-optioned

Spread before index deletion 0.0325 0.0316 0.0059 0.0033 Standard Deviation 0.0214 0.0346 0.0106 0.0053

Spread after index deletion 0.0391 0.0300 0.0061 0.0046 Standard Deviation 0.0315 0.0331 0.0105 0.0061 T-stat for equality 2.68 -0.74 2.32 2.81

p-value 0.008 0.46 0.02 0.005 Number of stocks 8 7 42 6

This table presents the average relative bid-ask spreads before and after deletion from the S&P 500 index, for optioned and non-optioned stocks, in the pre- and post-2000 periods. Relative spread is the absolute spread divided by the mean of the ask and bid prices. Average spread is the difference between ask and bid prices. The spreads prior to deletion are calculated over the 30 trading days before announcement date of deletion and the spreads in the post deletion period are calculated over 30 trading days after the announcement date of deletion. In Table 7, we estimate the multivariate regression equation (1) for optioned and non-optioned stocks in the pre- and post-2000 periods. As can be seen in Table 7, the liquidity change after deletion for non-optioned stocks in the post-2000 period is insignificant (the time dummy variable is insignificant). This is in accordance with our expectation. Table 7: Multivariate Regression for Relative Spread of Deletions for Optioned and Non-Optioned Stocks

Pre-2000 period Post-2000 period Optioned Non-optioned Optioned Non-optioned

Intercept 0.0594*** 0.0278*** 0.0075*** 0.0050*** Time Dummy -0.0044*** 0.0053*** -0.0004 -0.0005

Price -0.0022*** -0.0003*** -0.0002*** -0.0001*** Volume -0.0000*** 0.0000*** -0.0000 -0.0000 Variance 1.3578*** 1.9154*** 0.0381*** 0.5100***

Adjusted R2 0.335 0.239 0.061 0.1427 This table presents the estimates for the following multivariate model, for optioned and non-optioned stocks: Spreadit=λ0+λ1TimeDummyit+λ2Priceit+λ3Volumeit+λ4Varianceit+εit where Spreadit, Priceit, Volumeit, and Varianceit, are the relative bid/ask spread, the closing share price, the trading volume and an estimate of return variance, respectively. TimeDummyit is a dummy variable which equals 1(0) in the 30 trading days before (after) announcement day for stock i on day t. Return variance on day t is estimated using the variance of the stock’s return over the five-day period immediately preceding day t. ***, **, and * indicate significance at the 1, 5 and 10 percent levels respectively. CONCLUDING COMMENTS This paper revisits the abnormal returns and liquidity changes around deletions from the S&P 500 Index, in light of new research that argues that the information environment in the financial markets has changed


9

after the year 2000, because of major regulations passed in and after year 2000. By examining and comparing deletions in the pre- and post-2000 periods, we find some effect of decreased information asymmetry in the post-2000 period on deletions. Our results indicate that the average abnormal returns on the announcement day of the deletion are not significantly different in the post-2000 period, from in the pre-2000 period. However, we do find evidence that the negative abnormal returns after deletion from the S&P 500 Index are reversed earlier in the post-2000 period from the pre-2000 period. This could probably be because of reduced information asymmetry in the post-2000 period. We also find that contrary to our expectations, the average relative bid-ask spread after the announcement of deletion increases significantly (that is, liquidity after announcement of deletion decreases) in the post-2000 period. If there is reduced information asymmetry in the post-2000 period, we should see only a marginal increase in the spread in the post-2000 period. However, when we examine optioned versus non-optioned stocks and control for other factors that affect the relative spread, namely, closing price, trading volume, and return variance, we find insignificant increase in the spread after announcement of deletion, in the post- 2000 period, for non-optioned stocks. This is consistent with our expectations. As is true for past studies on deletions from the S&P 500 Index, the small sample size is obviously a limitation of this study. REFERENCES Beneish, M. D. and R. E. Whaley (1996) “An anatomy of the 'S&P 500 game': the effects of changing the rules,” Journal of Finance 51 (5), 1909-1930 Beneish, M. D. and R. E. Whaley (2002) “S&P 500 index replacements: A new game in town,” The Journal of Portfolio Management 29 (1), 51-60 Cai, J (2007) “What’s in the news? Information content of S&P 500 additions,” Financial Management 36 (3), 113-124 Chen, H., G. Noronha, and V. Singal (2004) “The price response to S&P 500 index additions and deletions: evidence of asymmetry and a new explanation,” Journal of Finance 59(4), 1901-1930 Chen, H., G. Noronha, and V. Singal (2006a) “S&P 500 Index Changes and Investor Awareness,” Journal of Investment Management 4 (2), 23-37 Chen, H., G. Noronha, and V. Singal (2006b) “Index Changes and Losses to Index Fund Investors,” Financial Analysts Journal 62 (4), 31-47 Dash, S. (2002) “Price Changes Associated with S&P 500 Deletions: Time Variation and Effect of Size and Share Prices,” Standard & Poor’s, July 9 Denis, D. K., J. J. McConnell, A. V. Ovtchinnikov, and Y. Yu. ( 2003) “S&P 500 Index Additions and Earnings Expectations,” Journal of Finance 58 (5), 1821-1840 Dhillon, U.S. and H. G. Johnson (1991) “Changes in the Standard and Poor’s 500 list,” Journal of Business 64 (1), 75-85 Erwin, G.R. and J.M. Miller (1998) “The Liquidity Effects Associated with Additions of A Stock to the S&P 500 Index: Evidence From Bid/Ask Spreads,” Financial Review 33 (1), 131-146 Harris, L.E. and E. Gurel (1986) “Price and Volume Effects Associated with Changes in the S&P 500 List: New Evidence for the Existence of Price Pressures,” Journal of Finance 41 (4), 815-829


10

Hegde, S. P. and J. B. McDermott (2003) “The Liquidity Effects Of Revisions to the S&P 500 Index: An Empirical Analysis,” Journal of Financial Markets 6 (3), 413-459 Ivanov, S. (2010) “Discretionary Deletions from the S&P 500 Index: Evidence on Forecasted and Realized Earnings,” The International Journal of Business and Finance Research 4 (4), 1-9 Jain, P.C. (1987) “The Effect on Stock Price of Inclusion in and Exclusion from the S&P 500,” Financial Analysts Journal 43 (1), 58-65 Kamal, R., E. Lawrence, G. McCabe, and A. Prakash (2012) “Additions to S&P 500 Index: Not So Informative Any More”, Managerial Finance 38 (4), 380-402 Lynch, A. W. and R. R. Mendenhall (1997) “New Evidence on Stock Price Effects Associated With Changes in the S&P 500 Index,” Journal of Business 70 (3), 351-383 Shleifer, A. (1986) “Do Demand Curves for Stocks Slope Down?,” Journal of Finance 41 (3), 579-590 BIOGRAPHY Dr. Rashiqa Kamal is an Assistant Professor in the Department of Finance and Business Law at the University of Wisconsin-Whitewater. She can be contacted at: Department of Finance and Business Law, College of Business and Economics, University of Wisconsin-Whitewater, 800 W. Main St., Whitewater, WI 53190. E-mail: [email protected]


11

THE INTERPLAY BETWEEN DIRECTOR COMPENSATION AND CEO COMPENSATION

Dan Lin, Takming University of Science and Technology Lu Lin, Takming University of Science and Technology

ABSTRACT

This paper empirically examines the determinants of director compensation and CEO compensation and investigates whether director compensation has an effect on CEO compensation. Based on 713 firms (or 2,852 firm-years) between 2007 and 2010, we find that CEO tenure is related to the ability of the CEO in influencing the board’s pay determination process. However, sitting on the board does not strengthen the CEO’s power over the board during the pay negotiation process. More importantly, we find evidence of a “mutual back scratching” relationship between CEO and the board of directors. Excess director compensation and CEO compensation are positively related. The results thus support Jensen’s (1993) argument that as the CEO is involved in the selection of directors, the monitoring role of the board of directors becomes less effective. JEL: J33, M52 KEYWORDS: Director Compensation, CEO Compensation, Board of Directors INTRODUCTION

ue to the conflicts of interests between outside shareholders and managers in the modern corporate structure, the board of directors has the fundamental role of monitoring managers to ensure that managers act in the interest of shareholders. However, as the CEO is often involved in

the selection of directors, Jensen (1993) argues that the board directors may not be an effective monitor. The board of directors may become more aligned with the CEO, thereby compromising the independence of the board. Brick et al. (2006) further suggest that when the board of directors is highly compensated, they are less likely to conduct critical monitoring of the CEO, referred to as “mutual back scratching”. According to Hermalin and Weisbach (1998), the CEO may also use barriers to monitoring, including large boards, inside directors, CEO duality, CEO tenure, and CEO membership in nominating committee, in an attempt to maximize his compensation. Therefore, one objective of this study is to examine whether director compensation has an effect on CEO compensation by utilizing the excess director compensation variable, which is the residual from the director compensation model. After the financial crisis of 2008, the “fat cat problem” highlighted the executive compensation issue. Recently there have been increasing concerns about the escalation in executive compensation (Dong and Ozkan, 2008). In particular, the substantial rises in executive pay have far exceeded the increases in underlying firm performance (Gregg et al., 2005). The review of CEO compensation by Frydman and Jenter (2010) shows that there was a dramatic increase in compensation levels from the mid‐1970s to the early 2000s in the US. Especially in the 1990s, the annual growth rates were more than 10% by the end of the decade. The increase in executive compensation is also evident in firms of all sizes while larger firms have experienced greater growth. The high level of CEO pay in the U.S. has therefore brought about considerable debate and a lot of attention from academia and policy makers regarding executive compensation, in particular, the pay-setting process and the effectiveness of the compensation contracts. The compensation packages of the top executives are set by the board of directors. After the financial crisis, the boards of collapsed firms are asked to hold full responsibility because they have not conducted

D

D. Lin and L. Lin | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

12

appropriate supervision over top executives. In this regard, this study incorporates the characteristics of the board of directors and the effect of director compensation, in addition to CEO characteristics, when examining the determinants of CEO compensation. In short summary, the objective of this study is twofold. First, we analyze the determinants of director compensation. Based on the director compensation model, we derive the residuals (i.e., “excess director compensation”). Secondly, we examine whether excess director compensation and a set of CEO and director characteristics (such as CEO tenure, CEO shareholdings and board size) are related to CEO compensation. While the determinants of CEO compensation and the pay-for-performance relationship (Jensen and Murphy, 1990; Main et al., 1996; Brick et al., 2006; Ozkan, 2007) have been extensively researched, the compensation structure of the board of directors as a governance mechanism has received less attention (Cordeiro et al., 2000; Gregg et al., 2005), in particular, the interplay between director compensation and CEO compensation (Brick et al., 2006). Accordingly, this study makes an important contribution by linking director compensation with CEO compensation and examines whether there is a “mutual back scratching” relationship between CEO and the board of directors. That is, whether the CEO receives higher compensation when the directors are paid more. Specifically, we include an “excess director compensation” variable in the CEO compensation model. If there is a positive relationship between excess director compensation and CEO compensation, then a “mutual back scratching” relationship between the board of directors and the CEO exists. If a negative relationship is observed, it means that the directors are effective monitors of the top management. In addition, this study contributes to the literature by adopting multiple measures when analyzing director compensation. This allows us to examine the director compensation from different perspectives. Unlike CEO compensation, as there is more than one person sitting on the board of directors, the board of director compensation may be measured by the total director compensation for the entire board, the average director compensation, and the compensation of the highest paid director. Most of previous studies rely on one single measure (for example, Becher et al., 2005; Fernandes, 2008) or differentiate compensation by cash and stock compensation only (for example, Cordeiro et al., 2000; Brick et al., 2006). These studies may suffer from the weaknesses inherent in a particular measure. For example, total director compensation for the entire board may be influenced by the size of the board. The average director compensation ignores the dispersion within each firm and may be distorted by extreme values. Using the compensation of the highest paid director may sometimes be measuring the compensation of the CEO. Therefore, it is important to consider different measures. Based on 713 firms (or 2,852 firm-years) between 2007 and 2010, we find support for the “mutual back scratching” relationship between the CEO and the directors. Specifically, excess director compensation and CEO compensation are positively related. The evidence thus suggests that the directors are not good monitors of the CEO. The results also support Jensen’s argument. As directors are selected by the CEO, the effectiveness of directors’ monitoring of the top management is weakened. The remainder of this paper is organized into five sections. In Section 2, we review the prior empirical literature on director and CEO compensation and develop the hypotheses tested in this study. In Section 3, we describe the data, methodology and sample characteristics. In Section 4, we present the results on director compensation and CEO compensation. A conclusion is provided in Section 5. LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT In modern economies, most companies are characterized by the separation of ownership and control where the ownership is held by diverse shareholders and the control is in the hands of top executives. As a


13

result, shareholders are not able to monitor managers’ actions directly. According to the agency theory, these companies are likely to suffer from agency problems. That is, managers as the agents may not always act in the interest of shareholders (i.e., the principals), thereby giving rise to the conflicts of interests. The governance structure of the firms, as argued by the agency theorists, can mitigate the potential agency problem between managers and shareholders arising from the separation of ownership and control, and therefore, influence the way firms set executive compensation packages (Murphy, 2009). In fact, the board of directors who is responsible for providing advice to the management and assisting with strategy development plays a key governance role in monitoring top management (Fama and Jensen, 1983). The board of directors also has an essential role in setting CEO compensation (Finkelstein and Hambrick, 1988; Boyd, 1994; Barkema and Gomez-Mejia, 1998; Carpenter and Sanders, 2002; Chhaochharia and Grinstein, 2009). Therefore, one objective of this study is to examine whether the board of directors has influences over CEO compensation. An early paper by Finkelstein and Hambrick (1988) provides a synthesis on CEO compensation and suggests that there are two main set of factors that affect CEO compensation: first, the market factors, including managerial labor market, marginal products of CEOs, CEO discretion, firm size, firm performance, and human capital; secondly, the power and preferences of the board and CEO. Consistent with this view, Ozkan (2007) finds that corporate governance mechanisms have a significant effect on the level of CEO compensation. Specifically, measures of board and ownership structures explain a significant amount of cross-sectional variation in CEO total compensation. Barkema and Gomez-Mejia (1998) propose a general research framework on the relationship between pay and performance. They argue that criteria, such as the market, peer compensation, individual characteristics, a firm’s governance structure (including ownership structure, board of directors, remuneration committee, and market for corporate control), and contingencies (such as a firm’s strategy, R&D level, market growth, industry concentration and regulation, and national culture), can enhance our understanding of the determinants of executive pay. Moreover, the managerial power theory argues that excessive CEO pay is due to the greater power of executives over directors that allows the former to set their own pay and extract rents (Bebchuk et al, 2002; Bebchuk and Fried, 2004). An implication of the theory is that enhancing the independence of the board can improve corporate governance and prevent managers from extracting rents in the form of higher pay (Guthrie et al., 2012). Therefore, the first objective of this study is to examine the determinants of director compensation. Then, we investigate if CEO characteristics and director characteristics, including excess director compensation, have influences over CEO compensation. Specifically, this study adds to the literature on executive compensation by investigating the effect of director compensation on CEO compensation and testing if there is a “mutual back scratching” relationship between the CEO and the board of directors. The hypotheses of this study are developed below. Director Compensation Following Hill and Phan (1991), this study uses CEO tenure to proxy for CEO’s ability to exercise influence over the board of directors. Previous studies (Hermalin and Weisbach, 1991; Shivdasani and Yermack, 1999) have suggested that CEOs can exert influence over the director selection process. Ryan and Wiggins (2001) argue that the level of CEO entrenchment and CEO power over the board of directors increase with CEO tenure. Specifically, they find that firms with long-tenured CEOs (i.e., more entrenched managers) discourage board scrutiny of management and provide weaker incentives to directors to monitor management. Therefore, CEO tenure is expected to be negatively associated with director compensation. That is, the following hypothesis is proposed.


14

H1a: CEO tenure will be negatively related to director compensation. CEO director is an important corporate governance variable that accounts for the CEO influence over the board. Previous studies (Boyd, 1995; Daily and Schwenk, 1996; Conyon and Peck, 1998; Cordeiro and Veliyath, 2003) have mostly used CEO chairman as the proxy; that is, whether the CEO is also the chairman of the board of directors. However, this study argues that even in the case where the CEO is not the chairman and is simply a board of director, he still has the ability to exert influence on the board. Hence, this study argues that using a broader definition, CEO director, is a better proxy. To test for the influence of CEO over the board of directors, we include a dummy variable, if the CEO also holds a board seat. When a CEO is also a board of director, the board is likely to be entrenched. Brick et al. (2006) find that directors of firms with a unitary leadership structure (that is, the CEO and the Chairman are the same person) receive higher total compensation than directors of firms with a dual leadership structure where the roles of CEO and the chairman are performed by different persons. They argue that this is because the unitary leadership structure reflects weak governance. Accordingly, we offer the following hypothesis.

H1b: CEO director will be positively related to director compensation.

Firms with larger boards are expected to be associated with higher director compensation for two reasons. Firstly, as the number of directors increases, the total board compensation will increase. Secondly, firms with larger boards are typically more complex firms and therefore should give higher pay to their directors. Therefore, a positive relationship between board size and director compensation is proposed. H1c: Board size will be positively related to director compensation. CEO Compensation As CEOs build a power base and gain voting control over time, they may exert greater influence over board composition. Consequently, CEOs may be able to demand compensation packages that serve their own interests rather than the shareholders’ (Hill and Phan, 1991; Cordeiro and Veliyath, 2003; Ozkan, 2011). Moreover, Finkelstein and Hambrick (1996) suggest that the tenure of an executive can affect and proxy for his attitudes towards risk. This is because long-tenured executives have established high firm-specific human capital and become less mobile (Hill and Phan, 1991). They will be unwilling to take on any unnecessary risks that are likely to bring more harms than benefits. Hill and Phan (1991) further argue that the positive relationship between pay and firm risk will be stronger the longer the tenure of the CEO. Hence, CEO tenure is expected to be positively associated with CEO compensation. H2a: CEO tenure will be positively related to CEO compensation. A CEO who is also a board of director is likely to obtain higher pay since he can not only participate in but also exert influence over the board’s pay determination process. Therefore, a positive relationship is expected between CEO compensation and CEO director.

H2b: CEO director will be positively related to CEO compensation. The level of CEO shareholdings shows the extent to which the wealth of the CEO is linked with firm value and is related to the extent of agency problems faced by companies (Dong and Ozkan, 2008). CEOs with greater shareholdings in the firm will have stronger incentives to boost the firm’s stock value. Therefore, less incentive compensation is needed for aligning the interests of CEO and shareholders. Accordingly, CEO shareholdings can act as a substitute for CEO compensation (Cordeiro and Veliyath, 2003) and a negative relationship is expected between CEO compensation and CEO shareholdings.


15

H2c: CEO shareholdings will be negatively related to CEO compensation. Male CEOs are expected to receive higher compensation than female CEOs given that the CEO market is predominated by males. Therefore, we offer the following hypothesis.

H2d: Male CEO will be associated with higher CEO compensation. The size of the board affects the effectiveness of the board in monitoring management. For example, when the board size grows large, more resource networks and professional views can be brought to board. However, these advantages may be overwhelmed by the efficiency losses in communication, decision-making and coordination between board members as the number of board members increases. In other words, a large board may in effect reduce the effectiveness of board monitoring and therefore be associated with higher CEO compensation. Consistent with the latter view, Core et al. (1999) report that larger boards pay more to their CEOs in terms of both cash compensation and total compensation. Based on a sample of 414 UK companies between 2003 and 2004, Ozkan (2007) also reports that firms with larger board size are associated with higher CEO compensation, measured by total compensation and cash compensation. Moreover, Guest (2010) examines a comprehensive and long period dataset of 1,880 UK firms over the period 1983-2002 and reports a positive relationship between board size and the rate of increase in executive compensation, providing support for the argument that large boards suffer from the problems of less efficient decision-making and poor communication. Therefore, this study expects a positive relationship between board size and CEO compensation.

H2e: Board size will be positively related to CEO compensation. To examine the impact of director compensation on CEO compensation, we include the residuals from the director compensation model in the CEO compensation model, i.e., the excess director compensation. While the pay of the CEO is determined by the board of directors, the CEO is involved in the selection of the board of directors. Therefore, this study expects a “mutual back scratching” relationship between the CEO and the board of directors; that is, a positive relationship between excess director compensation and CEO compensation. Specifically, this study tests if CEOs receive a higher pay when directors are being paid higher. H2f: Excess director compensation will be positively related to CEO compensation. Control Variables To control for other variables documented in previous literature as important in determining compensation levels, the following variables are also included in the models. Firm size controls for the fact that larger firms which are typically more complex will require directors to spend more time and put more effort in monitoring managers. In other words, larger firms are associated with greater complexity and information processing demands and therefore, directors of larger firms are expected to receive higher compensation. Hence, a positive relationship is expected between director compensation and firm size. Similarly, CEOs of larger firms have greater responsibility, require more effort, and therefore are expected to be more highly compensated (Smith and Watts, 1992; Core et al. 2003). The study by Conyon (1997) has reported a significantly positive relationship between firm size and CEO compensation levels. Accordingly, a positive relationship is also expected between CEO compensation and firm size. Agency theory suggests that one way to align the interests of managers with that of shareholders is to tie the compensation contracts to firm performance (Firth et al., 2006; Chhaochharia and Grinstein, 2009); that is, to create a pay-for-performance linkage. In other words, to motivate directors to actively monitor managers on behalf of shareholders, directors should be rewarded when firm performance is high.


16

Therefore, we expect a positive relationship between director compensation levels and firm performance. Similarly, making the CEOs hold accountable for firm performance is essential for motivating the CEOs to initiate strategies that boost firm value. Hence, a positive relationship between CEO compensation and firm performance is also expected. The pay of directors and CEOs is likely to be set with reference to the pay of other directors and CEOs in the same industry. Hilburn (2010) reports that directors of technology companies have higher pay than their counterparts at general industry companies. Therefore, the differences in industry structures, complexity and industry customs are likely to affect the level of compensation (Hempel and Fay, 1994). Hence, this study includes a dummy variable for industry sectors to control for inter-industry differences in compensation levels. Year dummies are also included in our models to control for unobserved differences between years. The inclusion of these dummies can capture common factors that are driven by industry- and economy-wide effects. DATA AND METHODOLOGY The data used in this study are obtained from the Standard and Poor’s ExecuComp database. To be included in the sample, the sample firms must have all the required financial information, such as total assets, sales, ROA and ROE, CEO compensation, and director compensation data. As the information on director compensation in ExecuComp database is more complete from the year 2006 and onwards, the sample period for this study is set between 2007 and 2010. Previous literature has suggested that banks are likely to face greater potential conflicts of interests than industrial firms due to its distinct characteristics such as the existence of deposit insurance, high debt-to-equity ratios and asset-liability issues (Becher et al., 2005). Since the nature of financial services industry is different from that of industrial firms, firms belonging to the financial services industry are excluded from the sample. Therefore, our sample begins with a total of 940 firms (or 3760 firm-years). After eliminating 28 firms with missing data and 199 firms in the finance, insurance and real estate industries (that is, Division H of the SIC division structure), the final sample consists of 713 firms (or 2,852 firm-years). The hypotheses are tested using pooled time-series cross-sectional regression analysis. The two models tested in this study are outlined below. Model 1 is on director compensation and Model 2 is on CEO compensation. ( ) ( )

( ) ttti5ti

titi2tititi

YEARDUMINDUSDUMEPERFORMANCFSIZElnBSIZElnCEODIR CEOTENUREDIRCOMPln

++++

+++=

−1,,4

,3,,1,,

ββ

βββα (1)

The dependent variable (DIRCOMP) of Model 1 is measured in three ways, the total director compensation, the average compensation of directors, and the compensation of the highest paid director. Firstly, the total director compensation is the directors’ total compensation for the entire board, including cash fees, stock awards, option awards, non-equity incentive plan compensation, change in pension value and non-qualified deferred compensation earnings, and other compensation provided by ExecuComp database. The reason for measuring director compensation for the entire board is that it is the board collectively that monitors for and acts on behalf of the shareholders. Secondly, the average director compensation is the per capita compensation of directors (Fernandes, 2008), where the compensation is measured in total and includes cash fees, stock awards, option awards, non-equity incentive plan compensation, change in pension value and non-qualified deferred compensation earnings, and other compensation provided by ExecuComp database. One weakness with this measure is that measuring director compensation as an average ignores the dispersion within each firm.


17

As most studies focus on the CEO who holds the top paying job, this study also analyzes the highest paid person on the board; i.e., the third measure of director compensation in this study. Gregg et al. (1993) who examine the relationship between directors’ pay and corporate performance also adopt this measure. Formally, the compensation of the highest paid director is the total compensation of the highest paid director, where total compensation includes cash fees, stock awards, option awards, non-equity incentive plan compensation, change in pension value and non-qualified deferred compensation earnings, and other compensation provided by ExecuComp database. The definitions of independent and control variables are as follows. CEO tenure (CEOTENURE) is measured by the number of years the CEO had held the position in a given company. An alternative measure for CEO tenure is the age of the CEO (CEOAGE), which is expected to have strong positive correlation with CEO tenure and also proxies for CEO experience. CEO director (CEODIR) is a dummy variable that equals one if the CEO is also a board of director. Board size (BSIZE) is measured by the number of directors on the board. Firm size (FSIZE) is measured by total assets and sales. Firm performance (PERFORMANCE) is measured by the return on assets (ROA) and return on average equity (ROE), which are lagged one year in order to avoid measuring the effect of compensation on performance. The lagged performance measure can also account for the fact that director compensation paid in one year is usually determined by the firm performance in the previous year. ROA has been widely used in previous studies on executive compensation and corporate governance as a proxy for firm performance. ROA shows how efficient the firm is in utilizing its assets (Finkelstein and Hambrick, 1996; Finkelstein and Boyd, 1998; Carpenter and Sanders, 2002). On the other hand, ROE can better reflect firm performance from the shareholders’ point of view. Therefore, in this study, models are estimated separately using both measures. Industry (INDUSDUM) is determined by SIC division structure, ranging from Division A to J (Descriptions for the SIC division structure are outlined below. Division A: agriculture, forestry, and fishing; Division B: mining; Division C: construction; Division D: manufacturing; Division E: transportation, communications, electric, gas, and sanitary services; Division F: wholesale trade; Division G: retail trade; Division I: services; Division J: public administration.) Note that Division H, the finance, insurance, and real estate industries, is excluded from the sample. In this study we also include year dummies (YEARDUM). Based on Model 1, we derive the excess director compensation (EXDIRCOMP), which is the residual from the director compensation model when total director compensation is used as the dependent variable. The excess director compensation measures the extent of director under- or overpayment. This variable is then included in the second model on CEO compensation, as outlined below, to test the impact of director compensation on CEO compensation. ( )

( ) ( )ttti8

titititi

ti,3ti,2ti,1ti,ti,

YEARDUMINDUSDUMEPERFORMANCFSIZElnEXDIRCOMPBSIZElnCEOGENDER

CEOHOLDINGβCEODIRβ CEOTENUREβαCEOCOMPln

+++

++++

+++=

−1,

,7,6,5,4

β

ββββ (2)

The dependent variable (CEOCOMP) of Model 2 is measured in two ways, CEO total compensation and CEO cash compensation. Ozkan (2011) suggests that firm performance may affect cash and equity-based components of compensation differently. It is important to incorporate multiple measures for compensation. In this study, the CEO total compensation comprises salary, bonus, other annual payment, restricted stock grants, long-term incentive payouts, value of options granted and all other payments provided by ExecuComp database. The second measure, CEO cash compensation, consists of salary and bonus.


18

The additional variables introduced in the second model are defined as follows. CEO shareholdings (CEOHOLDING) is calculated as the shares owned by the CEO, excluding options that are exercisable or will become exercisable within 60 days, divided by the number of common shares outstanding. CEO gender (CEOGENDER) is a dummy variable that equals one if the CEO is male. Excess director compensation (EXDIRCOMP) is the residual from the director compensation model where the dependent variable is the total director compensation. Table 1 presents the descriptive statistics of CEO characteristics and CEO compensation for 713 sample firms. The average and median age of CEOs is both 55, ranging from 34 to 80. The mean CEO ownership is 1.53% and ranges from 0 to 75.8% of outstanding shares. CEO tenure, which measures the number of years the CEO had held the position in a given company, has an average of 7.2 years and ranges from 0 to 47 years. The mean (or median) value of cash compensation, which consists of salary and bonus, received by the CEOs of our sample firms is $1,116,474 (or $875,158). The total compensation has an average of $5,838,773 and ranges from $30,002 to $128,706,100. In our sample, about 96.6% of CEOs are male and 96.8% of CEOs also hold a board seat. Table 1: Descriptive Statistics of CEO Characteristics and CEO Compensation Mean Median Max Min SD CEO characteristics CEO age 55 55 80 34 6.67 CEO shareholdings (%) 1.53 0.29 75.80 0.00 4.91 CEO tenure (years) 7.22 5.00 47.00 0.00 6.68 CEO cash compensation ($'000) 1,116.5 875.2 77,926 7.1 2,466.3 CEO total compensation ($'000) 5,838.8 4,076.8 128,706 30.0 6,722.0 CEO gender Male 2754 96.56% Female 98 3.44% Total 2852 100.00% CEO is also a board of director Yes 2760 96.77% No 92 3.23% Total 2852 100.00% This table reports the descriptive statistics of CEO characteristics and CEO compensation for 713 firms (or 2,852 firm-years) between 2007 and 2010. CEO shareholdings is calculated as shares owned by the CEO, excluding options that are exercisable or will become exercisable within 60 days, divided by the total number of common shares outstanding. CEO cash compensation includes salary and bonus. CEO total compensation includes salary, bonus, other annual, total value of restricted stock granted, total value of stock options granted (using Black-Scholes), long-term incentive payouts, and all other compensation. The descriptive statistics for firm characteristics and director compensation are shown in Table 2. The average board size is 9 and ranges from 3 to 26 directors. The average firm size, measured by total assets, is $9,898 million and $7,849 million if measured by sales. Firm performance is measured by ROA and ROE. The average ROA and ROE are 3.99% and 9.99%, respectively. The mean and median “average director compensation per board” is $181,794 and $166,643, respectively. The mean “total director compensation per board” is $1,597,003 and ranges from $33,374 to $14,685,740.


19

Table 2: Descriptive Statistics of Firm Characteristics and Director Compensation Mean Median Max Min SD Firm characteristics Board size 9 8 26 3 2.46 Total assets ($m) 9,898.2 2,345.1 797,769 10.0 36,640 Sales ($m) 7,849.1 2,034.8 425,071 0.1 25,797 ROA (%) 3.99 5.17 52.85 -163.38 11.77 ROE (%) 9.99 12.24 524.38 -906.03 36.19 Director compensation per board DIRCOMP_Average ($'000) 181.8 166.6 1,796 3.6 114.6 DIRCOMP_Maximum ($'000) 305.6 221.4 7,779 13.6 430.3 DIRCOMP_Total ($'000) 1,597.0 1,402.6 14,686 33.4 1,081.1 This table reports the descriptive statistics of firm characteristics and director compensation for 713 firms (2,852 firm-years) between 2007 and 2010. DIRCOMP_Average is the average director compensation for each firm (or each board), that is, the per capita compensation of directors. DIRCOMP_Maximum is the compensation of the highest paid director in each firm. DIRCOMP_Total is the total director compensation for the entire board. Director compensation is defined to include cash fees, stock awards, option awards, non-equity incentive plan compensation, change in pension value and non-qualified deferred compensation earnings, and other compensation Table 3 reports the descriptive statistics for the components of director compensation. Between 2007 and 2010, there is a total of 24,604 director-years. The average cash fees paid to directors is $71,708,000. The directors in our sample receive an average of $73,103,000 in stock awards, $28,111,000 in option awards, and $515,000 in non-equity incentive plan. The total director compensation has an average of $185,118,000 and ranges from -$1,299,073,000 to $7,778,702. The negative total compensation can be attributed to the negative amounts in stock and option awards and the negative change in pension value and non-qualified deferred compensation earnings. Table 3: Descriptive Statistics of Director Compensation Mean Median Max Min SD Components of director compensation Cash fees ($'000) 71.71 68.39 777.2 0.0 44.35 Stock awards ($'000) 73.10 60.69 7,612.0 -362.1 94.70 Option awards ($'000) 28.11 0.00 4,939.6 -1,886.1 95.45 Non-equity incentive ($'000) 0.52 0.00 2,619.0 0.0 29.02 Pension change ($'000) 0.96 0.00 406.0 -805.3 12.23 Other compensation ($'000) 10.68 0.00 6,004.4 0.0 94.02 Total compensation ($'000) 185.12 165.50 7,778.7 -1,299.1 182.22 This table reports the descriptive statistics of director compensation for 24,604 director-years between 2007 and 2010. Director compensation is classified as cash fees, stock awards, option awards, non-equity incentive plan, change in pension value and non-qualified deferred compensation earnings, and all other compensation. Cash fees are director fees that are earned or paid in cash. Stock awards are measured by the value of stock-related awards (e.g. restricted stock, restricted stock units, phantom stock, phantom stock units, common stock equivalent units etc.) that do not have option-like features. Option awards are measured by the value of option-related awards (e.g. options, stock appreciation rights, and other instruments with option-like features). Non-equity incentive is measured by the value of amounts earned during the year pursuant to non-equity incentive plans. Pension change is composed of above-market or preferential earnings from deferred compensation plans and aggregate increase in actual value of defined benefit and actual pension plans during the year. Other compensation includes perquisites and other personal benefits, contributions to defined contribution plans, life insurance premiums, gross-ups and other tax reimbursements, discounted share purchases, consulting fees, awards under charitable award programs etc. Table 4 reports the correlations between variables. Overall, the CEO and director compensation are positively related to board size, firm performance, measured by ROA and ROE, and firm size, measured by total assets and sales. The CEO shareholdings are negatively associated with CEO compensation, suggesting a substitution effect between CEO shareholdings and CEO compensation (Cordeiro and Veliyath, 2003). Consistent with the expectation, the CEO tenure, a proxy for CEO power, is positively related to CEO compensation and negatively related to director compensation.


20

Table 4: Correlation Matrix

1 2 3 4 5 6 7 8 9 10 11 12 13

1.CEO age 1 2.CEO shareholdings 0.11 *** 1 3.CEO tenure 0.42 *** 0.38 *** 1 4.Board size 0.07 *** -0.24 *** -0.22 *** 1 5.Assets 0.03 -0.06 *** -0.05 ** 0.32 *** 1 6.Sales 0.05 *** -0.07 *** -0.07 *** 0.29 *** 0.67 *** 1 7.ROEt-1 0.00 -0.01 0.01 0.07 *** 0.05 *** 0.08 *** 1 8.ROAt-1 0.00 0.02 0.02 0.04 ** 0.03 * 0.07 *** 0.67 *** 1 9.CEOCOMP_CASH 0.10 *** -0.02 0.06 *** 0.10 *** 0.14 *** 0.11 *** 0.02 0.01 1 10.CEOCOMP_TOT 0.12 *** -0.12 *** 0.00 0.34 *** 0.28 *** 0.32 *** 0.11 *** 0.09 *** 0.65 *** 1 11.DIRCOMP_AVE 0.01 -0.15 *** -0.03 * 0.10 *** 0.16 *** 0.15 *** 0.09 *** 0.07 *** 0.15 *** 0.37 *** 1 12.DIRCOMP_MAX -0.03 * -0.07 *** -0.07 *** 0.10 *** 0.06 *** 0.06 *** 0.02 0.02 0.05 *** 0.16 *** 0.73 *** 1 13.DIRCOMP_TOT 0.05 *** -0.19 *** -0.10 *** 0.51 *** 0.33 *** 0.30 *** 0.11 *** 0.08 *** 0.18 *** 0.51 *** 0.85 *** 0.66 *** 1 This table reports the correlations of variables used in the regression analysis for a sample of 713 firms during the period 2007-2010. CEOCOMP_CASH denotes CEO cash compensation. CEOCOMP_TOT denotes CEO total compensation. DIRCOMP_AVE denotes the average director compensation. DIRCOMP_MAX denotes the compensation of the highest paid director. DIRCOMP_TOT denotes the total director compensation. RESULTS Table 5 reports OLS estimation results for director compensation. In Panel A, the dependent variable is total director compensation, measured by the directors’ total compensation for the entire board. In Panel B, the dependent variable is the average compensation of directors, which is measured as the per capita compensation of directors, where the compensation is measured in total. In Panel C, the dependent variable is the total compensation of the highest paid director. For each measure of director compensation (i.e., in each panel), Model 1 is estimated four times as we have adopted alternative measures for CEO tenure (i.e., CEO tenure and CEO age), firm size (i.e., total assets and sales), and firm performance (i.e., ROE and ROA). The regression estimates in Table 5 show that CEOs with shorter tenure or younger age are significantly associated with higher director compensation at the 1% level. This finding is consistent with our prediction that short-tenured CEOs have less ability to exercise influence over the board of directors. The result is consistent across three measures of director compensation. Inconsistent with our expectation, CEO director dummy variable is negatively associated with the director compensation, significant at the 1% level. In other words, the director compensation is higher when the CEO is not a member of the board. The result suggests that without the influence of CEO over the board, directors are able to set higher compensation to favor themselves. Board size is significantly positively related to total director compensation and the compensation of the highest paid director at the 1% level. However, it is significantly negatively related to the average compensation of directors. This is because as the number of board members increases, the total director compensation per board evens out, leading to a negative relationship. Firm size, measured by total assets and sales, are also are significantly positively related to director compensation. Interestingly, the study by Song and Xu (2007) based on a sample of Chinese listed companies finds that the total compensation received by board of directors is negatively associated with board size, CEO tenure and the proportion of inside directors. They suggest that when the board lacks independence, the executives will dominate over directors, resulting in less compensation to directors. Consistent with Song and Xu (2007), this study finds that directors of larger firms receive more compensation.


21

Table 5: Analysis of Director Compensation Panel A: Dependent variable: ln(DIRCOMP_total) Panel B: Dependent variable: ln(DIRCOMP_average)

1 2 3 4 1 2 3 4 Intercept 3.877 *** 5.074 *** 4.003 *** 3.864 *** 3.877 *** 5.074 *** 4.003 *** 3.864 *** (55.437) (21.725) (53.237) (51.494) (55.437) (21.725) (53.237) (51.494) CEOTENURE -0.008 *** -0.008 *** -0.008 *** -0.008 *** -0.008 *** -0.008 *** (-10.478) (-10.750) (-10.497) (-10.478) (-10.750) (-10.497) ln(CEOAGE) -0.339 *** -0.339 *** (-7.337) (-7.337) CEODIR -0.101 *** -0.108 *** -0.098 *** -0.101 *** -0.101 *** -0.108 *** -0.098 *** -0.101 *** (-5.935) (-6.132) (-5.891) (-6.275) (-5.935) (-6.132) (-5.891) (-6.275) ln(BSIZE) 0.843 *** 0.889 *** 0.941 *** 0.843 *** -0.157 *** -0.111 *** -0.059 ** -0.157 *** (34.922) (37.732) (33.208) (31.848) (-6.497) (-4.709) (-2.067) (-5.940) ln(ASSETS) 0.196 *** 0.198 *** 0.198 *** 0.196 *** 0.198 *** 0.198 *** (44.784) (44.852) (40.955) (44.784 ) (44.852) (40.955) ln(SALES) 0.172 *** 0.172 *** (36.447) (36.447) ROEt-1 0.001 *** 0.001 *** 0.001 *** 0.001 *** 0.001 *** 0.001 *** (4.032) (3.813) (2.609) (4.032) (3.813) (2.609) ROAt-1 0.001 * 0.001 * (1.754) (1.754) Industry and year dummies Yes Yes Yes Yes Yes Yes Yes Yes

Adjusted R2 0.546 0.544 0.521 0.545 0.303 0.299 0.264 0.301 Panel C: Dependent variable: ln(DIRCOMP_max)

1 2 3 4 Intercept 4.208 *** 6.150 *** 4.297 *** 4.202 *** (24.094) (14.841) (24.679) (23.425) CEOTENURE -0.015 *** -0.015 *** -0.015 *** (-8.503) (-8.603) (-8.465) ln(CEOAGE) -0.555 *** (-6.132) CEODIR -0.114 *** -0.126 *** -0.112 *** -0.115 *** (-3.472) (-3.854) (-3.364) (-3.481) ln(BSIZE) 0.178 *** 0.258 *** 0.245 *** 0.177 *** (10.925) (12.204) (11.519) (10.251) ln(ASSETS) 0.152 *** 0.155 *** 0.153 *** (68.617) (84.134) (59.841) ln(SALES) 0.136 *** (71.922) ROEt-1 0.000 0.000 0.000 (1.369) (0.904) (0.674) ROAt-1 0.000 (0.116) Industry and year dummies Yes Yes Yes Yes

Adjusted R2 0.208 0.198 0.195 0.207 This table presents the regression analysis of director compensation for 713 firms between 2007 and 2010, where the director compensation includes cash fees, stock awards, option awards, non-equity incentive plan, change in pension value and non-qualified deferred compensation earnings, and all other compensation. In Panel A, B and C, the dependent variables are total director compensation, average director compensation and the compensation of the highest paid director, respectively. CEOTENURE is measured by the number of years the CEO had held the position in a given company. CEOAGE is the age of the CEO. CEODIR is a dummy variable that equals one if the CEO is also a board of director. BSIZE is measured by the number of directors on the board. ASSETS and SALES are measures for firm size. ROE and ROA measure firm performance and are lagged one year. t-statistics (in parentheses) are calculated using White's (1980) heteroskedasticity-consistent standard errors. ***, ** and * indicate coefficient is significant at the 1, 5 and 10% level, respectively.


22

The results show that director compensation, measured by total director compensation and average compensation of director, is higher when firms have better past performance, supporting the argument that compensation contracts should be linked to firm performance. However, when the director compensation is measured by the compensation of the highest paid director, the significant relationship with past firm performance disappears. In other words, highly paid directors are often not paid based on their performance. This finding supports the recent call for reviewing the compensation packages of “fat cat” directors (Dong and Ozkan, 2008). The evidence also suggests that for highly paid directors, the pay-for-performance linkage often does not exist. In particular, Gregg et al. (2005) argue that the substantial rises in executive pay have far exceeded the increases in underlying firm performance. Moreover, in terms of the firm performance measures, we find that ROE is a better predictor of director compensation than ROA. This can be explained by the fact that ROE can better reflect how well a firm performs from the shareholders’ point of view. The OLS estimation results for CEO compensation by incorporating the effect of director compensation are presented in Table 6. The dependent variable is CEO total compensation for Panel A and CEO cash compensation for Panel B. The former measure comprises the CEO’s salary, bonus, other annual payment, restricted stock grants, long-term incentive payouts, value of options granted and all other payments. The latter measure consists of salary and bonus only. For each measure of CEO compensation (i.e., in each panel), Model 2 is estimated four times as we have adopted alternative measures for CEO tenure (i.e., CEO tenure and CEO age), firm size (i.e., total assets and sales), and firm performance (i.e., ROE and ROA).The results show that excess director compensation is significantly positively related to CEO compensation at the 1% level. This finding supports our hypothesis that CEOs receive higher pay when the directors are paid higher. Accordingly, the evidence is consistent with the argument of a “mutual back scratching” relationship between the CEO and the board of directors (Brick et al., 2006). The results also suggest that directors are not good monitors of the top management and support Jensen’s (1993) argument that the effectiveness of directors’ monitoring role can be weakened by the fact that directors are selected by the CEO. Consistent with the expectation, CEO tenure and CEO age are positively related to CEO compensation. Although the level of significance is weaker when the CEO compensation is measured by CEO total compensation, CEO tenure and CEO age are significantly related to CEO cash compensation at the 1% level. Inconsistent with our expectation, CEO director dummy variable is negatively related to CEO compensation at the 5% significance level. That is, CEO compensation is higher when the CEO does not hold the board seat. Therefore, the observed high compensation received by CEOs that we observe today cannot be explained by their presence on the board of directors. Moreover, the result does not support the argument that dual leadership where the roles of CEO and the chairman are performed by different people is associated with better governance and therefore lower CEO compensation. Additionally, the results demonstrate that CEO shareholdings are significantly negatively associated with CEO total compensation at the 1% level, providing support for the hypothesis that CEO shareholdings and CEO compensation contracts are substitute mechanisms for aligning the interests of CEO and shareholders (Cordeiro and Veliyath, 2003). However, CEO shareholdings are insignificantly associated with CEO cash compensation. This is because the cash component of CEO compensation contracts does not link CEO wealth with firm value, and therefore, does not have the substitution effect like CEO total compensation. Interestingly, we find that the gender of CEOs is significantly related to CEO cash compensation but not CEO total compensation. Specifically, the results show that male CEOs receive higher cash compensation. Board size and firm size are significantly positively related to CEO compensation at the 1% level, consistent with the hypothesis. Since larger firms are typically more complex and have larger boards, CEOs of larger firms are therefore more highly compensated. Interestingly, we find that both measures of firm performance, ROE and ROA, cannot explain CEO total compensation, therefore, providing evidence against the pay-for-performance linkage that have been


23

raised by the popular press. Consistent with the finding of this study, Ozkan (2007) does not find a significant relationship between CEO compensation and firm performance based a sample of large UK companies for the fiscal year 2003/2004. Table 6: Analysis of CEO Compensation Panel A: Dependent variable: ln(CEOCOMP_total) Panel B: Dependent variable: ln(CEOCOMP_cash)

1 2 3 4 1 2 3 4 Intercept 4.087 *** 3.095 *** 4.352 *** 4.088 *** 5.155 *** 2.946 *** 5.280 *** 5.146 *** (23.987) (12.027) (24.535) (24.673) (29.570) (15.665) (29.230) (28.565) CEOTENURE 0.002

0.003 * 0.002 0.007 *** 0.007 *** 0.007 ***

(1.636) (1.917) (1.635) (13.410) (14.502) (15.361) ln(CEOAGE) 0.261 *** 0.587 *** (4.819) (43.189) CEODIR -0.131 ** -0.125 ** -0.124 ** -0.130 ** -0.047 ** -0.033 -0.043 ** -0.049 ** (-2.172) (-2.084) (-2.094) (-2.142) (-2.286) (-1.641) (-2.241) (-2.575) CEOHOLDING -0.009 *** -0.009 *** -0.011 *** -0.009 *** -0.002 -0.001 -0.003 -0.002 (-6.668) (-5.787) (-7.335) (-6.661) (-1.269) (-0.405) (-1.471) (-1.228) CEOGENDER 0.050 0.041 0.038 * 0.047 0.028 *** 0.009 0.019 ** 0.034 *** (1.449) (1.145) (1.908) (1.304) (2.816) (0.854) (2.520) (3.410) ln(BSIZE) 0.143 *** 0.131 *** 0.310 *** 0.145 *** 0.120 *** 0.087 *** 0.191 *** 0.115 *** (3.671) (3.021) (7.351) (3.551) (17.770) (17.702) (15.542) (20.068) EXDIRCOMP 0.404 *** 0.407 *** 0.412 *** 0.405 *** 0.084 *** 0.092 *** 0.088 *** 0.083 *** (13.220) (12.899) (13.181) (13.130) (6.692) (8.069) (7.032) (6.865) ln(ASSETS) 0.433 *** 0.431 *** 0.433 *** 0.223 *** 0.219 *** 0.225 *** (109.245) (113.227) (118.890) (45.888) (46.982) (56.629) ln(SALES) 0.395 *** 0.208 *** (53.933) (37.799) ROEt-1 0.000 0.000 0.000 0.000 ** 0.000 ** -0.001 ** (0.885) (0.944) (0.147) (-2.332) (-2.327) (-2.383) ROAt-1

0.001 -0.002 ***

(1.594) (-10.616) Industry and year dummies Yes Yes Yes Yes Yes Yes Yes Yes

Adjusted R2 0.615 0.616 0.565 0.615 0.412 0.422 0.386 0.414 This table presents the regression analysis of CEO compensation for 713 firms between 2007 and 2010. In Panel A and B, the dependent variables are CEO total compensation and CEO cash compensation, respectively. CEOTENURE is measured by the number of years the CEO had held the position in a given company. CEOAGE is the age of the CEO. CEODIR is a dummy variable that equals one if the CEO is also a board of director. CEOHOLDING is calculated as shares owned by the CEO divided by the number of common shares outstanding. CEOGENDER is a dummy variable that equals one if the CEO is male. BSIZE is measured by the number of directors on the board. EXDIRCOMP is the residual from the total board compensation model where the dependent variable is the total board compensation. ASSETS and SALES are measures for firm size. ROE and ROA measure firm performance and are lagged one year. t-statistics (in parentheses) are calculated using White's (1980) heteroskedasticity-consistent standard errors. ***, ** and * indicate coefficient is significant at the 1, 5 and 10% level, respectively. CONCLUDING COMMENTS The global financial crisis in 2008 sheds light on the significance of reviewing the compensation packages of top executives. Based on a sample of 713 US firms between 2007 and 2010, this study examines the determinants of director and CEO compensation based on a number of board of director and CEO characteristics. We also investigates whether there is a “mutual back scratching” relationship between the CEO and the board of directors by analyzing the relationship between director compensation and CEO compensation. Specifically, this study proposes two empirical models. The first is on director compensation and the second is on CEO compensation.


24

The results show that CEOs with shorter tenure or younger age are associated with higher director compensation but lower CEO compensation. This finding provides support for the argument that CEO tenure or CEO age is related to CEO’s ability to influence the board’s pay determination process. Interestingly, we find that CEO who also holds a board seat is not associated with higher CEO compensation. The result thus indicates that sitting on the board of directors does not strengthen the CEO’s power over the board during the pay negotiation process. More importantly, the results suggest that CEOs receive higher pay when the director compensation is higher, supporting the “mutual back scratching” relationship between the CEO and the board of directors. There is also a substitution effect between CEO total compensation and the level of CEO ownership. Finally, firms with larger board size and firm size give higher pay to their directors and CEOs. One limitation of this study is that due to the constraint on the availability of board of directors’ data, the sample period of this study is limited to four years only. Future research could extend the sample period by dropping the board size variable to see if similar results can be reached. REFERENCES Barkema, H.G. and L.R. Gomez-Mejia1(1998) “Managerial compensation and firm performance: A general research framework,” Academy of Management Journal, vol. 41, p. 135-145. Bebchuk, L.A. and J. Fried (2004) Pay without performance: The unfulfilled promise of executive compensation, Cambridge, MA: Harvard Univ. Press. Bebchuk, L.A., J. Fried and D. Walker (2002) “Managerial power and rent extraction in the design of executive compensation,” University of Chicago Law Review, vol. 69, p. 751-846. Becher, D.A., T.L. Campbell II and M.B. Frye (2005) “Incentive compensation for bank directors: The impact of deregulation,” Journal of Business, vol. 78, p. 1753-1777. Boyd, B.K. (1994) “Board control and CEO compensation,” Strategic Management Journal, vol. 15, p. 335-344. Boyd, B.K. (1995) “CEO duality and firm performance: A contingency model,” Strategic Management Journal, vol. 16, p. 301-312. Brick, I.E., O. Palmon and J.K. Wald (2006) “CEO compensation, director compensation, and firm performance: Evidence of cronyism?” Journal of Corporate Finance, vol. 12, p. 403-423. Carpenter, M.A. and W.M.G. Sanders (2002) “Top management team compensation: The missing link between CEO pay and firm performance?” Strategic Management Journal, vol. 23, p. 367-375. Chhaochharia, V. and Y. Grinstein (2009) “CEO compensation and board structure,” Journal of Finance, vol. 64, p. 231-261. Conyon, M.J. (1997) “Corporate governance and executive compensation,” International Journal of Industrial Organization, vol. 15, p. 493-509. Conyon, M.J. and S.I. Peck (1998) “Board control, remunearion committees, and top management compensation,” Academy of Management Journal, vol. 41, p. 146-157. Cordeiro, J. and R. Veliyath (2003) “Beyond pay for performance: A panel study of the determinants of CEO compensation,” American Business Review, vol. 21, p. 56-66.


25

Cordeiro, J., R. Veliyath, E.J. Eramus and M. Tax (2000) “An empirical investigation of the determinants of outside director compensation,” Corporate Governance: An International Review, vol. 8, p. 268-279. Core, J.E., W.R. Guay and D.F. Larcker (2003) “Executive equity compensation and incentives: A survey,” Federal Reserve Bank of New York Economic Policy Review, vol. 9, p. 27-50. Core, J.E., R.W. Holthausen and D.F. Larcker (1999) “Corporate governance, chief executive officer compensation, and firm performance,” Journal of Financial Economics, vol. 51, p. 371-406. Daily, C.M. and C. Schwenk (1996) “Chief executive officers, top management teams, and boards of directors: Congruent or countervailing forces?” Journal of Management, vol. 22, p. 185-208. Dong, M. and A. Ozkan (2008) “Institutional investors and director pay: An empirical study of UK companies,” Journal of Multinational Financial Management, vol. 18, p. 16-29. Fama, E.F. and M.C. Jensen (1983) “Separation of ownership and control,” Journal of Law and Economics, vol. 26, p. 301-325. Fernandes, N. (2008) “EC: Board compensation and firm performance: The role of ‘independent’ board members,” Journal of Multinational Financial Management, vol. 18, p. 30-44. Finkelstein, S. and B. Boyd (1998) “How much does the CEO matter? The role of managerial discretion in the setting of CEO compensation,” Academy of Management Journal, vol. 41, p. 179-199. Finkelstein, S. and D.C. Hambrick (1988) “Chief executive compensation: A synthesis and reconciliation,” Strategic Management Journal, vol. 9, p. 543-558. Finkelstein, S. and D.C. Hambrick (1996) Strategic leadership: Top executives and their effects on organization, New York: West Publishing. Firth, M., P.M.Y. Fung and O.M. Rui (2006) “Corporate performance and CEO compensation in China,” Journal of Corporate Finance, vol. 12, p. 693-714. Frydman, C. and D. Jenter (2010) “CEO compensation,” SSRN Working Paper. Available at: http://ssrn.com/abstract=1582232. Gregg, P., S. Jewell and I. Tonks (2005) “Executive pay and performance in the UK 1994-2002,” Unpublished Working Paper, University of Bristol and University of Exeter. Gregg, P., S. Machin and S. Szymanski (1993) “The disappearing relationship between directors' pay and corporate performance,” British Journal of Industrial Relations, vol. 31, p. 1-9. Guest, P. (2010) “Board structure and executive pay: Evidence from the UK,” Cambridge Journal of Economics, vol. 34, p. 1075-1097. Guthrie, K., J. Sokolowsky and K.M. Wan (2012) “CEO compensation and board structure revisited,” Journal of Finance, vol. 67, p. 1149-1168. Hempel, P. and C. Fay (1994) “Outside director compensation and firm performance,” Human Resource Management, vol. 33, p. 111-133.


26

Hermalin, B.E. and M.S. Weisbach (1991) “The effects of board composition and direct incentives on firm performance,” Financial Management, vol. 20, p. 101-121. Hermalin, B.E. and M.S. Weisbach (1998) “Endogenously chosen boards of directors and their monitoring of the CEO,” American Economic Review, vol. 88, p. 96-118. Hilburn, W. (2010) “Trends in director compensation,” Bloomberg Businessweek, October 19, 2010. Hill, C.W.L. and P. Phan (1991) “CEO tenure as a determinant of CEO pay,” Academy of Management Journal, vol. 34, p. 707-717. Jensen, M.C. (1993) “The modern industrial revolution, exit, and the failure of internal control systems,” Journal of Finance, vol. 48, p. 831-880. Jensen, M.C. and K.J. Murphy (1990) “Performance pay and top-management incentives,” Journal of Political Economy, vol. 98, p. 225-264. Main, B.G.M., A. Bruce and T. Buck (1996) “Total board remuneration and company performance,” Economic Journal, vol. 106, p. 1627-1644. Murphy, K.J. (2009) “Compensation structure and systemic risk,” Working Paper FEB 34-09, USC Marshall School of Business. Ozkan, N. (2007) “Do corporate governance mechanisms influence CEO compensation? An empirical investigation of UK companies,” Journal of Multinational Financial Management, vol. 17, p. 349-364. Ozkan, N. (2011) “CEO compensation and firm performance: An empirical investigation of UK panel data,” European Financial Management, vol. 17, p. 260-285. Ryan, H.J. and R. Wiggins (2001) “The influence of firm- and manager-specific characteristics on the structure of executive compensation,” Journal of Corporate Finance, vol. 7, p. 101-123. Shivdasani, A. and D. Yermack (1999) “CEO involvement in the selection of new board members: An empirical analysis,” Journal of Finance, vol. 54, p. 1829-1853. Smith, C. and R. Watts (1992) “The investment opportunity set and corporate financing, dividend, and compensation policies,” Journal of Financial Economics, vol. 32, p. 263-292. Song, Z. and Y. Xu (2007) “The relationship between director compensation and board independence: Evidence from China,” Proceedings of 2007 International Conference on Management Science and Engineering. BIOGRAPHY Dan Lin is an assistant professor at the Takming University of Science and Technology, and can be reached at [email protected]. Lu Lin, the corresponding author of this paper, is an assistant professor at the Takming University of Science and Technology, and can be reached at [email protected].


27

THE RELATIONSHIP BETWEEN BRAND IMAGE AND PURCHASE INTENTION: EVIDENCE FROM AWARD

WINNING MUTUAL FUNDS Ya-Hui Wang, National Chin-Yi University of Technology Cing-Fen Tsai, National Chin-Yi University of Technology

ABSTRACT

Mutual funds represent one of the most popular investment instruments. Some institutions offer fund awards to recognize strong performing funds and fund groups that have shown excellent returns relative to their peers. Many fund companies also use awards won in their advertising and marketing material. This brings rise to the question: Do investors think award winning funds have a better brand image? Can awards increase investors’ purchase intention? The purpose of this study is to investigate the relationships and effects of brand image, perceived quality, perceived risk, perceived value, and purchase intention, as well as to examine the effects of demographic variables on these five dimensions. The research findings show significant relationships between brand image, perceived quality, perceived value, and purchase intention. In addition, some demographic variables may lead to significant differences in these five dimensions. Finally, the results from structural equation modeling show that there are positive and direct effects among brand image, perceived quality, perceived value, and purchase intention. Brand image indeed increases investors purchase intentions. The purchase intention is affected mainly by perceived quality, not by perceived risk. JEL: G1, M1, M5 KEYWORDS: Brand Image, Perceived Quality, Perceived Risk, Perceived Value, Purchase Intention INTRODUCTION

utual funds are one of the most popular investment instruments today. Many investors are interested in mutual funds, because they have many advantages: professional management, high liquidity, diversification, minimum amount of cash needed, etc. However, there exists a

vast array of mutual funds. The most important issue is how to choose a good fund for investment to increase one’s wealth. Some institutions hold fund awards to recognize strong performing funds and fund groups that have shown excellent yearly returns relative to their peers. Examples include, the TFF-Bloomberg Best Fund Awards, the Morningstar Fund Awards, and the Lipper Fund Awards. Many fund companies use awards won in their advertising and marketing material, bringing rise to a question: Do investors think awarded funds have a better brand image? Brand image is often used as an extrinsic cue when consumers evaluate a product before purchasing (Zeithaml, 1988; Richardson, Dick and Jain, 1994). As such, from the viewpoint of fund companies, does this branding work? Can it really increase investors’ purchase intention? Consumers are more likely to purchase well-known brand products with a positive brand image, because a brand with this image has the effect of lowering consumers’ perceived risks (Akaah and Korgaonkar, 1988; Rao and Monroe, 1988) or increasing consumers’ perceived value (Loudon and Bitta, 1988; Fredericks and Slater, 1998; Romaniuk and Sharp, 2003). Thus, will investors choose awarded funds as their investment target? How do investors feel about awarded funds? Does an awarded fund really see a better brand image? Higher perceived quality? Lower perceived risk? Higher perceived value? Previous studies on awarded funds have focused on performance persistence by taking secondary data from the financial market. Little or no research has investigated investors’ purchase intentions of

M

YH. Wang & CF. Tsai | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

28

awarded funds directly through questionnaires. Our study attempts to fill this gap. This paper studies relationships between awarded funds’ brand image, perceived quality, perceived risk, perceived value, and purchase intention using questionnaires. The aims of this study are: (1) to investigate the relationships and effects of brand image, perceived quality, perceived risk, perceived value, and purchase intention; (2) to analyze the differences between investors with different demographic variables in brand image, perceived quality, perceived risk, perceived value, and purchase intention; (3) to analyze the implications of these results. The rest of this paper is organized as follows. Section 2 reviews previous research on brand image, perceived quality, perceived risk, perceived value, and purchase intention. Section 3 describes the data and method we employ. Section 4 reports the empirical results, and section 5 concludes the paper. LITERATURE REVIEW The most popular fund awards in Taiwan include TFF-Bloomberg Best Fund Awards, Morningstar Fund Awards, and Lipper Fund Awards. The TFF-Bloomberg Best Fund Award is sponsored by the Taipei Foundation of Finance (TFF) and co-sponsored by Bloomberg LP. It is a well-known mark of recognition in the Taiwanese mutual fund industry and has been awarded for 15 years since 1998. Qualifying funds compete in both domestic and foreign categories. Under the category “Domestic Fund Award”, funds are recognized based on data and analytics provided by National Taiwan University professors. “Foreign Fund Award” winners are selected by TFF based on Bloomberg’s Fund Scoring Model, which analyzes overall performance and risk exposure of each qualifying fund. Morningstar sponsors The Morningstar Fund Awards, with the objective to recognize those funds and fund groups that have added the most value within the context of a relevant peer group for investors over the past year and over the longer term. Funds are scored by their total return percentile ranks in their Morningstar categories over one-, three- and five-year periods, with 30% of the total score on the one-year period, 20% on the three-year period, and 30% on the five-year period, for a total of 80% allocated to returns. The remaining 20% of a fund’s score is allocated to risk adjustment. The Morningstar Risk of each fund is a robust risk measure using utility theory to penalize funds more for downside variation in returns than for upside volatility thereby keeping with actual investor concerns. Lipper sponsors The Lipper Fund Awards, taking place in 23 countries in Asia, Europe, MENA, and the Americas. The Lipper Fund Awards program honors funds that have excellent consistent risk-adjusted returns relative to their peers. The program also recognizes fund families with high average scores for all funds within a particular asset class or overall. Lipper designates award-winning funds in most individual classifications for the three-, five-, and ten-year periods and fund families with high average scores for the three-year time period. The awards winners are formally announced between January and April every year. The American Marketing Association defines brand as “a name, term, sign, symbol, design or combination, intended to identify goods and services and to differentiate them from the competition”. Kotler (2000) claimed that “brand is a name, term, symbol, design or all the above, and is used to distinguish one’s products and services from competitors”. Keller (1993; 1998) defined brand image as “perceptions about a brand as reflected by the brand associations held in consumer memory”. Accordingly, brand image does not exist in the features, technology or the actual product itself, but rather it is something brought out by advertisements, promotions or users. Brand image is often used as an extrinsic cue when consumers are evaluating a product before purchasing (Zeithaml, 1988; Richardson, Dick and Jain, 1994). Perceived quality is the consumer’s judgment about a product’s overall excellence and superiority, not the actual quality of a product (Zeithaml, 1988; Aaker, 1991). Consumers often judge product quality via informational cues. They form beliefs on the basis of these informational cues (intrinsic and extrinsic), and then judge the quality of a product and make their final purchase decision based upon these beliefs


29

(Olson, 1977). According to Zeithaml (1988), intrinsic attributes are physical characteristics of the product itself, such as a product’s conformance, durability, features, performance, reliability, and serviceability. On the contrary, extrinsic attributes are cues external to the product itself, such as price, brand image, and company reputation. Garvin (1987) defined perceived quality to include five dimensions: performance, features, conformance, durability, reliability, serviceability, aesthetics, and brand image. Petrick (2002) developed a four-dimensional scale to measure the perceived quality of a product: consistency, reliability, dependability, and superiority. Bauer (1960) first proposed perceived risk to include two dimensions: uncertainty and adverse consequences. Dowling and Staelin (1994) defined risk as a consumer’s perceptions of the uncertainty and adverse consequences of engaging in an activity. Perceived risk was also defined as the unfavorable outcomes related to a product or service (Engel, Blackwell and Miniard, 1995), the subjective perception of possibility and severity of a wrong purchase (Sinha and Batra, 1999), or the uncertainty a consumer perceives about the outcome of his or her purchase (Hoyer and Macinnis, 2010). The measurement of perceived risk was not explicitly defined in Bauer’s (1960) original paper. Many researchers thus regarded perceived risk as a multi-dimensional concept (Roselius, 1971; Jacoby and Kaplan, 1972; Stone and Gronhaug, 1993). Jacoby and Kaplan (1972) defined perceived risk to include five components: financial, performance, social, psychological, and physical risk. These five components can explain 74% of variation in perceived risk (Kaplan, 1974). Peter and Tarpey (1975), and Murray and Schlacter (1990) expanded the components to include time risk. Stone and Gronhaung (1993) proved that 88% of perceived risk can be explained by these six components. Perceived risk increases as the probability of one or more negative outcomes increases (Dowling and Staelin, 1994). Consumer behavior is motivated to reduce risk (Bauer, 1960; Taylor, 1974). Researchers found factors that influence perceived risk: brand loyalty (Cunningham, 1967), store selection (Hirsh, Dornoff and Kernan, 1972), quality warranty (Terence and William, 1982), and some demographic variables such as age, household income, and education level (Spence et al., 1970). Consumers’ perceptions of value represent a trade-off between the perceived quality or benefits in a product relative to the perceived sacrifice by paying the price. Monroe and Dodds (1985) defined perceived value as a trade-off between buyers’ perceptions of quality and sacrifice. It is positive when perceptions of quality are greater than the perceptions of sacrifice. Zeithaml (1988) defined perceived value as “the consumer’s overall assessment of the utility of a product, based on perceptions of what is received (e.g., quality, satisfaction) and what is given (price, nonmonetary costs)”. Monroe and Dodds (1985) directly related perceived value to preferences or choice, whereby the larger the perceived value is, the more likely the consumer will express a willingness to buy or have a preference for the product. Perceived value has is the most important indicator to forecast purchase intentions and has been viewed is an important measures for gaining a competitive advantage (Zeithaml, 1988; Dodds et al., 1991; Cronin et al., 2000). Purchase intention is the likelihood that a customer will buy a particular product (Fishbein and Ajzen, 1975; Dodds et al., 1991; Schiffman and Kanuk, 2000). A greater willingness to buy a product means the probability to buy it is higher, but not necessarily to actually buy it. On the contrary, a lower willingness does not mean an absolute impossibility to buy. Bagozzi and Burnkrant (1979) defined purchase intention as personal behavioral tendency to a particular product. Spears and Singh (2004) defined purchase intention as “an individual’s conscious plan to make an effort to purchase a brand”. Purchase intention is determined by a consumer’s perceived benefit and value (Xua, Summersb, and Bonnie, 2004; Grwal et al., 1998; Dodds et al., 1991; Zeithaml, 1988). Brand Image’s Influence on Perceived Quality, Perceived Risk and Perceived Value Brand image is an important cue during the process of consumers’ purchase decision making. Favorable brand information positively influences perceived quality, perceived value, and consumers’ willingness to buy (Dodds, Monroe & Grewal, 1991; Monroe and Krishnan, 1985). Consumers are more likely to purchase well-known brand products with a positive brand image, because a brand with a more positive


30

image does have the effect of lowering consumers’ perceived risks (Akaah and Korgaonkar, 1988; Rao and Monroe, 1988) or increasing consumers’ perceived value (Loudon and Bitta, 1988; Fredericks and Slater, 1998; Romaniuk and Sharp, 2003; Aghekyan, Forsythe, Kwon, and Chattaraman, 2012). Thus, we note the first three hypotheses as follows. H1: Brand image has a significantly positive impact on investors’ perceived quality. H2: Brand image has a significantly positive impact on investors’ perceived risk. H3: Brand image has a significantly positive impact on investors’ perceived value. Influence of Perceived Quality on Perceived Value and Purchase Intentions Monroe and Krishnan (1985), Zeithaml (1988), Dodds et al. (1991), and Petrick (2004) stated that a higher perception of quality improves consumers’ perceived value that strengthens consumers’ purchase intention. Garretson and Clow (1999), Chaudhuri (2002) and Yee and San (2011) found perceived quality to have a significant impact on a consumer’s purchase intention. Tsiotsou (2006) investigated the effects of perceived quality on purchase intentions and showed that perceived quality has a direct effect and an indirect effect (through overall satisfaction) on purchase intentions. Thus, we set up the following two hypotheses. H4: Perceived quality has a significantly positive impact on investors’ perceived value. H5: Perceived quality has a significantly positive impact on investors’ purchase intention. Influence of Perceived Risk on Perceived Value and Purchase Intentions Consumer behavior is motivated to reduce risk (Bauer, 1960; Taylor, 1974). According to Bettman (1973), a consumer’s purchase intention is affected by perceived risk. Perceived risk exists in a consumer’s decision process when he or she cannot foresee the purchase outcome and then uncertainty takes place (Hoover et al., 1978). As a result, perceived risk is a critical factor influencing a consumer’s purchase decision (Garrestson and Clow, 1999; Yee and San, 2011; Chen and Chang, 2012). Sweeney, Soutar and Johnson (1999), and Snoj, Korda and Mumel (2004) also found that perceived risk has a significantly negative impact on perceived value. Thus, we offer the next two hypotheses as follows. H6: Perceived risk has a significantly negative impact on investors’ perceived value. H7: Perceived risk has a significantly negative impact on investors’ purchase intention. Influence of Perceived Value on Purchase Intentions Perceived value plays an important role in purchase or consumption decisions. Many scholars have note that perceived value is relevant to the emotional responses and consumption experiences of consumers, which can further influence the consumer’s purchase behavior (Dumana & Mattil, 2005; Petrick, 2004; Sweeney & Soutar, 2001). When other things remain unchanged, purchase intention is positively related to perceived value (Della, Monroe and McGinnis, 1981; Monroe and Chapman, 1987; Zeithaml, 1988; Chen and Chang, 2012; Yee and San, 2011; Wu, Chen, Chen, and Cheng, 2012). Accordingly, we propose the following hypothesis. H8: Perceived value has a significantly positive impact on investors’ purchase intention. Influence of Demographic Variables on Each Dimension Businesses have different marketing strategies for different consumer groups because market segmentation is helpful in finding target consumers and creating competitive advantages. Demographic segmentation means dividing the market into specific groups according to gender, marital status, age, education, occupation, income, religion, nationality or race (Assael, 2005). These characteristics are the link to buyers' wants and needs and affect purchasing behavior. Therefore, it is the most popular basis


31

for segmenting customer groups mainly because it is the easiest, most measurable and most widely used segmentation method (Plummer, 1974; Donthu and Garcia, 1999). Demographic variables such as age, gender, occupation, income, and so on, have significant impacts on investors’ buying behavioral pattern (Jani and Jain, 2013). Alexander et al. (1998) found that age has a significant impact on investor behaviors. Jianakoplos and Bernasek (1998), Sunden and Surette (1998) also found gender differences exist in investment decisions. Accordingly, we set up the following hypothesis. H9: There are significant differences in each dimension for investors with different demographic

variables DATA AND METHODOLOGY According to the research framework, we design the questionnaire items for six dimensions: brand image, perceived quality, perceived risk, perceived value, purchase intention, and demographic variables. These items are measured on Likert’s seven-point scale, ranging from 1 point to 7 points, denoting “strongly disagree”, “disagree”, “a little disagree”, “neutral”, “a little agree”, “agree”, and “strongly agree.” We administered the questionnaires to investors living in Taiwan using random sampling from October 5, 2012 to December 31, 2012. A total of 795 responses were distributed, and 691 usable responses were collected. An acceptable response rate of 87% was achieved. Figure 1 presents the research framework. This framework demonstrates the relationships and effects among “brand image”, “perceived quality”, “perceived risk”, “perceived value”, and “purchase intentions”. It also intends to measure the effects of “demographic variables” on brand image, perceived quality, perceived value, and purchase intentions. Figure 1: Research Framework

This figure shows the research design. It also shows how the hypotheses fit into the framework

Brand image

Perceived quality

Perceived risk

H1

H2

H3 H8

Perceived value

H4

H6

Purchase intention

H5

H7

Demographic Variables H9


32

We perform data analyses on SPSS 12.0 and AMOS 17.0. The methods adopted include descriptive statistics analysis, reliability and validity analysis, correlation analysis, one-sample t-test analysis, factor analysis, one-way ANOVA, and structural equation modeling (SEM) analysis. ANALYSES AND RESULTS Through descriptive statistics analysis in Table 1, we are able to understand the distribution of participants’ basic attributes. The gender data shows 44.6% of the subjects are male, and 55.4% are female. The results show 64% of participants unmarried and 36% married. The age categories show the main group is 21-30 years old, taking up 43.7%, followed by the group of 31-40 years old (24.9%), 41-50 years old (16.4%), and younger than 20 years old (10.3%). The education levels indicate university education is the main group, taking up 67.0%, followed by graduate school (14.5%) and high school education (14.0%). Income data shows: most subjects (37.8%) earned below NT$20,000 per month, 30.8% earned NT$20,001-NT$40,000, 16.5% earned NT$40,001-NT$60,000, and 14.9% earned more than NT$60,000. Some 66.6% of the subjects live in central Taiwan, followed by northern Taiwan (22.1%), southern Taiwan (9.0%), and eastern Taiwan (2.3%). Finally we collect data on occupation which show students form the major group (32.4%), followed by financial industry (20.0%), service industry (17.9%), manufacturing industry (6.9%), high-tech industry (6.4%), public servants (5.4%), and others (11%). Table 1: Descriptive Statistics

Variable Category Frequency Percent (%)

Gender Male 308 44.6 Female 383 55.4

Marital status Married 249 36.0 Unmarried 442 64.0

Age

Younger than 20 years old 71 10.3 21-30 years old 302 43.7 31-40 years old 172 24.9 41-50 years old 113 16.4 Older than 50 years old 33 4.8

Education level

Junior high school 21 3.0 Senior high school 97 14.0 University 463 67.0 Graduate school 100 14.5 Ph. D 10 1.4

Monthly income

Below NT$20,000 261 37.8 NT$20,001-NT$40,000 213 30.8 NT$40,001-NT$60,000 114 16.5 More than NT$60,000 103 14.9

Residential area

Northern Taiwan 153 22.1 Central Taiwan 460 66.6 Southern Taiwan 62 9.0 Eastern Taiwan 16 2.3

Occupation

Financial industry 138 20.0 Public servants 37 5.4 Manufacturing industry 48 6.9 High-tech industry 44 6.4 Service industry 124 17.9 Students 224 32.4 Others 76 11.0

This table shows the descriptive statistics analysis for the sample data. The first column is demographic variables in this study. The third and fourth column reveals the frequency and percentage of total number of observations in each category, respectively. As presented in Table 2, all the dimensions have a Cronbach’s α greater than 0.9, which complies with the criterion proposed by Guiedford (1965). Hence, the reliability coefficient (Cronbach’s α ) of the questionnaire is within the acceptable level. Factor analysis is also taken as a tool to verify the convergent validity of the questionnaire. This study adopts principal component analysis and uses the


33

Varimax to maximize the sum of the variance of the loading factors. We extract factors with an eigenvalue greater than 1, cumulative explained variation greater than 50%, and a factor loading greater than 0.5 (Kaiser, 1958). Table 2: Reliability and Validity Test

Dimensions (Factors) Eigen Value Explained Variance Cronbach’s Α

Brand image 6.587 65.867% 0.942

Perceived quality 6.933 69.328% 0.942

Perceived risk 6.743 74.927% 0.912

Perceived value 7.208 72.076% 0.957

Purchase intention 6.743 74.927% 0.958

This table shows the reliability and validity test of all factors in this study. The first and third figure in each cell is the Eigen value and the Cronbach’s α value, respectively. The second figure in each cell represents the explained variance of each factor.

According to the results in Table 2, the questionnaire has convergent validity. In addition, it has content validity, because our scale and item contents are constructed according to the literature review and passed the questionnaire pre-test. The questionnaire also has discriminant validity, because the correlation coefficient of each of the two factors in Table 3 is lower than the Cronbach’s α of each dimension. Table 3: Correlation Analysis

Dimensions Brand Image

Perceived Quality Perceived

Risk Perceived Value Purchase Intention

Brand image

1

Perceived quality 0.845*** (0.000)

1

Perceived risk 0.042

(0.276) 0.039

(0.307) 1

Perceived value 0.784*** (0.000)

0.866*** (0.000)

0.076** (0.046)

1

Purchase intention 0.768*** (0.000)

0.809*** (0.000)

0.103*** (0.006)

0.879*** (0.000)

1

This table presents the correlation analysis. The figures on the non-diagonal represent Pearson correlation coefficient between two factors. The figures in parentheses represent p-value. ***, **, and * indicate significance at the 1, 5 and 10 percent levels respectively. In this section we conduct the one-way ANOVA to investigate whether the demographic variables have significant effects on brand image, perceived quality, perceived risk, perceived value, and purchase intentions. As shown in Table 4, there are significant differences in these five dimensions for investors with different education levels and occupation. There are significant differences in perceived quality and perceived risk for gender. There are significant differences in brand image, perceived quality, perceived risk, and purchase intention for different residential areas, marital status only impact perceived value, and monthly income only impacts perceived risk. The results partial support our hypothesis H9. This section conducts structural equation modeling (SEM) analysis to test the fit of the factors (dimensions) of brand image, perceived quality, perceived risk, perceived value, and purchase intention. For a model with good fit, GFI (goodness of fit) should greater than 0.8 (Browne and Cudeck, 1993).


34

AGFI (adjust goodness of fit) should be greater than 0.8 and CFI (comparative fit index) greater than 0.9 (Hair et al., 1998; Gefen et al., 2000). RMSEA (root mean square error of approximation) should be under 0.05 (Bagozzi and Yi, 1988; Joreskong and Sorbom, 1992), and the ratio of chi-square value to degrees of freedom 𝜒2/𝑑𝑑 should be no greater than five (Wheaton et al., 1977). A stricter criterion is that 𝜒2/𝑑𝑑 should be smaller than three (Carmines and Maclver, 1981; Hair et al., 2006). The goodness-of-fit indices of the model are as follows: GFI is 0.888, AGFI is 0.856, CFI is 0.962, RMSEA is 0.043, and 𝜒2/𝑑𝑑 is 2.256. All these indices are within the acceptable range, meaning that the overall model fitness is good. Table 4: ANOVA of Demographic Variables

Gender Marital Status Education Monthly Income

Residential Area

Occupation

Brand image 2.607

(0.107)

0.184 (0.668)

4.009*** (0.003)

1.130 (0.341)

5.488*** (0.001)

4.463*** (0.000)

Perceived quality

4.925** (0.027)

0.762 (0.383)

3.663*** (0.006)

0.493 (0.741)

2.326* (0.074)

2.622** (0.016)

Perceived risk 4.644** (0.032)

0.084 (0.772)

2.737** (0.028)

3.854*** (0.004)

3.067** (0.027)

4.736*** (0.000)

Perceived value 0.057

(0.811)

2.997* (0.084)

3.902*** (0.004)

1.084 (0.363)

1.977 (0.116)

2.201** (0.041)

Purchase intention

1.978 (0.160)

0.579 (0.447)

3.463*** (0.008)

1.199 (0.310)

2.628** (0.049)

2.617** (0.016)

This table shows ANOVA of demographic variables on brand image, perceived quality, perceived risk, perceived value, and purchase intention. The figure in each cell represents the t-statistic or F-statistic. The figure in each parenthesis is the p-value. ***, **, and * indicate significance at the 1, 5 and 10 percent levels respectively. Table 5 presents the estimated values of the standardized parameters of the relationship model and the results from the hypotheses verified. According to Table 5 and the path analysis in Figure 2, we find that brand image has a significant positive influence on perceived quality (H1 is supported) and has an insignificant positive impact on perceived risk and on perceived value (H2 and H3 are not supported). Consistent with Monroe and Krishnan (1985), Zeithaml (1988), Dodds et al. (1991), and Petrick (2004), perceived quality has a significant positive influence on perceived value and purchase intention, respectively (H4 and H5 are supported). Perceived value also has a significant positive influence on purchase intention (H8 is supported). The results are consistent with Della, Monroe and McGinnis (1981), Monroe and Chapman (1987), Zeithaml (1988), Yee and San (2011), Chen and Chang (2012), and Wu, Chen, Chen, and Cheng (2012). On the other hand, perceived risk has an insignificant positive impact on perceived value (H6 is not supported) and has a significant positive impact on purchase intention (H7 is not supported, because the sign is not “negative” as expected). The results from SEM show that there are positive and direct effects among brand image, perceived quality, and purchase intention. However, ‘the brand image impact on perceived value’, ‘the brand image impact on perceived risk’, ‘the perceived risk impact on perceived value’, and ‘the perceived risk impact on purchase intention’ are all not significant. It means that investors’ purchase intentions are affected mainly by perceived quality, not by perceived risk.


35

Table 5: Estimated Values of Standardized Parameters and the AMOS Model Fit Test Results

Hypotheses (Paths) Standardized Factor Loadings

T-Value Results

H1: Brand image → perceived quality 0.895 21.556*** Supported

H2: Brand image → perceived risk 0.025 0.613 Not supported

H3: Brand image → perceived value 0.085 1.568 Not supported

H4: Perceived quality → perceived value 0.832 13.557*** Supported

H5: Perceived quality → purchase intention 0.125 2.247*** Supported

H6: Perceived risk → perceived value 0.022 1.096 Not supported

H7: Perceived risk → purchase intention 0.064 3.228*** Not supported

H8: Perceived value → purchase intention 0.851 13.383*** Supported

This table shows the estimated values of standardized parameters and the hypothesis test results. The first column represents our research hypotheses (paths). The figure in second column is the standardized factor loading of each path. The figure in third column is the t-statistic. ***, **, and * indicate significance at the 1, 5 and 10 percent levels respectively. CONCLUSION AND IMPLICATION The purpose of this study is to investigate the relationships and effects of brand image, perceived quality, perceived risk, perceived value, and purchase intention, as well as to examine the effects of demographic variables on these five dimensions. The research findings’ show significant relationships among brand image, perceived quality, perceived value, and purchase intention according to the correlation analysis. In the test of the effects of demographic variables on brand image, perceived quality, perceived risk, perceived value, and purchase intention, the one-way ANOVA result indicates significant differences in all five dimensions for investors with different education levels and occupation. There are significant differences in perceived quality and perceived risk for gender. There are significant differences in brand image, perceived risk, and purchase intention for different residential areas, and monthly income has an effect only on perceived risk. Finally, the results from SEM show that there are positive and direct effects among brand image, perceived quality, perceived value, and purchase intention - that is, brand image increases perceived quality, perceived quality increases perceived value, and perceived value increases purchase intention. However, ‘the brand image impact on perceived value’, ‘the brand image impact on perceived risk’, ‘the perceived risk impact on perceived value’, and ‘the perceived risk impact on purchase intention’ are all not significant. It means that brand image indeed increases investors’ purchase intentions, and purchase intention is affected mainly by perceived quality, not by perceived risk. This research discovered that brand image indeed increases investors’ purchase intention. Therefore, we suggest that fund managers should devote efforts to elevating and maintaining their brand images via advertising and marketing funds that have received awards. Once a positive image is established, fund companies may utilize the added values to promote their other funds that have not yet won an award. The results also show that brand image increases investors’ purchase intention, and purchase intention is affected mainly by perceived quality, not by perceived risk. Therefore, fund companies should pay attention to strategies that increase investors’ perceived quality when they are marketing their awarded funds. Finally, because there are significant differences in brand image, perceived quality, perceived risk, perceived value, and purchase intention for investors with different education levels and occupation, fund companies should provide different marketing strategies according to these characteristics of investors. Different product categories may lead to distinct results. The primary limitation of this study is that it explores only a one-product category (awarded funds), potentially limiting generalizability to other domains. Moreover, we did not classify the asset classes of awarded funds (e.g. equity, bond, and mixed


36

assets). Further research is recommended to do this and identify additional differences. We also only considered brand image, perceived quality, perceived risk, and perceived value in this study. Other determinants of purchase intention could be included in comprehensive models thereby potentially improving explanatory power. Finally, most of the respondents in our study are from the age group of 21-30 years old, or students or young persons who do not have much money to invest. Therefore, the potential for bias exists due to the different purchase behaviors among different age groups. Therefore, future studies might examine different age and education groups. REFERENCES Aaker, D. A (1991) “Managing Brand Equity: Capitalizing on the Value of a Brand Name,” New York: The Free Press. Aghekyan, M., Forsythe, S., Kwon, W. S., & Chattaraman, V (2012) “The Role of Product Brand Image and Online Store Image on Perceived Risks and Online Purchase Intentions for Apparel,” Journal of Retailing and Consumer Services, vol. 19(3), p. 325-331.

Akaah, I. P. & Korgaonkar, P. K. (1988) “A Conjoint Investigation of the Relative Importance of Risk Relievers in Direct Marketing,” Journal of Advertising Research, vol. 28(August/September), p. 38-44. Alexander, G. J, Jones, J. D., & Nigro, P. J. (1998) “Mutual Fund Shareholders: Characteristics, Investor Knowledge, and Sources of Information,” Financial Services Review, vol. 7(4), p. 301-316. Assael, H (2005) “A Demographic and Psychographic Profile of Heavy Internet Users and Users by Type of Internet Usage,” Journal of Advertising Research, vol. 45(1), p. 93-123. Bagozzi, R. P. & Burnkrant, R. E (1979) “Attitude Organization and Attitude-Behavior Relationship,” Journal of Personality and Social Psychology, vol. 37(1), p. 913-929. Bagozzi, R. P. & Yi, Y (1988) “On the Evaluation for Structural Equation Models,” Journal of the Academy of Marketing Science, vol. 16(1), p. 74-94. Bauer, R. A (1960) “Consumer Behavior as Risk Taking, Dynamic Marketing for a Changing World,” Proceedings of the 43rd Annual Conference of the American Marketing Association, 389-398.

Bettman, J. R (1973) “Perceived Risk and Its Components: A Model and Empirical Test,” Journal of Marketing Research, vol. 10 (May), p. 184-90. Browne, M. W., & Cudeck, R (1993) “Alternative Ways of Assessing Model Fit,” In K. A. Bollen, & J. S. Long (Eds.), Testing Structural Equation Models (p. 136-162). Newbury Park, CA: Sage. Carmines, E. G., & Maclver, J. P (1981) “Analyzing Models with Unobserved Variables,” In G. W. Bohrnstedt & E. F. Borgatta (Eds.), Social Measurement: Current Issues. Beverly Hills, CA: Sage. Chaudhuri, A (2002) “How Brand Reputation Affects the Advertising-Brand Equity Link,” Journal of Advertising Research, vol. 42, p. 33-56. Chen, Y. S., & Chang, C. H (2012) “Enhance Green Purchase Intentions: The Roles of Green Perceived Value, Green Perceived Risk, and Green Trust,” Management Decision, vol. 50(3), p.502-520. Cronin, J. J., Brady, M. K. & Hult, G. T. M (2000) “Assessing the Effects of Quality Value and Customer Satisfaction on Consumer Behavioral Intentions in Service Environments,” Journal of Retailing, vol. 76(2), p. 193-218.


37

Cunningham, S. (1967) “The Major Dimensions of Perceived Risk,” In Cox, D. F (Ed.), Risk-Taking and Information Handling in Consumer Behavior, Boston: Harvard University Press, p. 82-108.

Della, B., Monroe, K. B. & McGinnis, J. M (1981) “Consumer Perceptions of Comparative Price Advertisements,” Journal of Marketing Research, vol. 18 (4), p. 416-427.

Dodds, W. B., Monroe, K. B. & Grewal, D (1991) “Effects of Price, Brand, and Store Information on Buyers’ Product Evaluation,” Journal of Marketing Research, vol. 28, p. 307-319. Donthu, N. & Garcia, A (1999) “The Internet Shopper,” Journal of Advertising Research, vol. 39, p. 52-58.

Dowling, G. R. & Staelin, R (1994) “A Model of Perceived Risk and Intended Risk-Handling Activity,” Journal of Consumer Research, vol. 21(June), p. 119-134.

Dumana, T. & Mattilab (2005) “The Role of Affective Factors on Perceived Cruise Vacation Value,” Tourism Management, vol. 26, p. 321-323. Engel, J. F., Blackwell, R. D. & Miniard, P. W (1995) “Consumer Behavior (8th Ed.),” Fort Worth, TX: The Dryden Press.

Fishbein, M. & Ajzen, I (1975) “Belief, Attitude, Intention and Behavior: an Introduction to Theory and Research,” Reading, MA: Addison-Wesley.

Fredericks, J. O. & Slater, J. M (1998) “What Does Your Customer Really Want?” Quality Progress, vol. 31, p. 63-65. Garretson, J. A. & Clow, K. E (1999) “The Influence of Coupon Face Value on Service Quality Expectations, Risk Perceptions and Purchase Intentions in the Dental Industry,” Journal of Service Marketing, vol. 13(1), p. 59-70. Garvin, D. A (1987) “Competing on the Eight Dimensions of Quality. Harvard Business Review, Vol. 65(November/December), p. 101-109.

Gefen, D., Straub, D. W. & Boudreau, M. C (2000) “Structural Equation Modeling and Regression: Guideline for Research Practice. Communications of the Association for Information Systems, vol. 4(7), p. 1-70. Guielford, J. P (1965) Fundamental statistics in psychology and education, 4th Edition, New York: McGraw-Hill. Hair, J. R., Anderson, R. E., Tatham, R. L. & Black, W. C (1998) Multivariate Data Analysis (5th Ed). New York: Macmillan. Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E. & Tatham, R. L (2006) Multivariate data analysis (6th Ed.). New Jersey: Pearson Prentice Hall.

Hoover, R. J., Green, R. T. & Saeger, J (1978) “A Cross-National Study of Perceived Risk,” Journal of Marketing, vol. 42 (3), p. 102-108

Hoyer, D. & Macinnis, D. J (2010) Consumer Behavior, 5th ed., Cengage Learning Inc. Jacoby, J. & Kaplan, L (1972) “The components of perceived risk,” Proceedings of 3rd Annual Conference of the Association for Consumer Research, p. 382-393.


38

Jani, D. & Jain, R (2013) “A Comparative Analysis of Investors Buying Behavior of Urban Rural for Financial Assets Specifically Focused on Mutual Fund,” International Monthly Refereed Journal of Research In Management and Technology, vol. 2(3), p. 167-176. Jianakoplos, N. & Bernasek, A (1998) “Are Women More Risk Averse?” Economic Inquiry, vol. 36(3), p. 620-630.

Kaiser, H. F (1958) “The Varimax Criterion for Analytic Rotation in Factor Analysis,” Psychometrika, vol. 23, p. 187-200.

Kaplan, L. B., Szybillo, G. J. & Jacoby, J (1974) “Components of Perceived Risk in Product Purchase: A Cross-Validation,” Journal of Applied Psychology, vol. 59(3), p. 287-291. Keller, K. L (1993) “Conceptualizing, Measuring, and Managing Customer-Based Brand Equity,” Journal of Marketing, vol. 57(1), p. 1-22. Keller, K. L (1998) Strategic Brand Management, New York: Prentice Hall. Kotler, P (2000) Marketing Management, 10th ed., New Jersey: Prentice-Hall.

Loudon, D. & Bitta, A. J. D (1988) Consumer behavior: Concepts and Applications, 3rd ed., New York: McGraw-Hill.

Monroe, K. B. & Chapman, J. D (1987) “Framing Effects on Buyers Subjective Product Evaluations,” Advances in Consumer Research, vol. 14 (2), p. 193-197. Monroe, K. B. & Dodds, W. B (1985) “The Effect of Brand and Price Information on Subjective Product Evaluations,” In E. Hirschman and M. Holbrook (Eds.), Advances in Consumer Research (p. 85-90), Association for Consumer Research. Monroe, K. B. & Krishnan, R (1985) “The Effect of Price on Subjective Product Evaluation,” In Jacob and Jerry C. Olson (Eds.), Perceived Quality: How Consumers View Stores and Merchandise, MA: Lexington Books. Murray, K. B. & Schlacter, J. L (1990) “The Impact of Services versus Goods on Consumers’ Assessment of Perceived Risk and Variability,” Journal of the Academy of Marketing Science, vol. 18(1), p. 51-65. Olson, J. C (1977) “Price as an Information Cue: Effects in Product Evaluation,” In A. G. Woodside, J. N. Sheh, and P. D. Bennet (Eds.) Consumer and Industrial Buying Behavior, New York: North Holland Publishing Company, p. 267-286. Peter, J. P. & Tarpey, L. X (1975) “A Comparative Analysis of Three Consumer Decision Strategies,” Journal of Consumer Research, p. 29-37.

Petrick, J. F (2002) “An Examination of the Construct of Perceived Value for the Prediction of Golf Travelers’ Intention to Revisit,” Journal of Travel Research, vol. 41, p. 35-45.

Petrick, J. F (2004) “First Timers’ and Repeaters’ Perceived Value,” Journal of Travel Research, vol. 43 (1), p. 29-38. Plummer, J. T (1974) “The Concept and Application of Lifestyle Segmentation,” Journal of Marketing, vol. 38(1), p. 33-37.


39

Rao, A. R. & Monroe, K. B (1988) “The Moderating Effect of Prior Knowledge on Cue Utilization in Product Evaluations,” Journal of Consumer Research, vol. 15(2), p. 253-264. Richardson, P. S., Dick, A. S. & Jain, A. K (1994) “Extrinsic and Extrinsic Cue Effect on Perceptions of Store Brand Quality,” Journal of Marketing Research, vol. 58(4), p. 28-36.

Romaniuk, J. & Sharp, B (2003) “Brand Salience and Customer Defection in Subscription Markets,” Journal of Marketing Management, vol. 19, p. 25-44.

Roselius, T (1971) “Consumer Rankings of Risk Reduction Methods,” Journal of Marketing, vol. 35, p. 56-61. Schiffman, L. G. & Kanuk, L. L (2000) Consumer Behavior, 7th ed., Prentice-Hall. Sinha, I. & Batra, R (1999) “The Effect of Consumer Price Consciousness on Private Label Purchase,” International Journal of Research in Marketing, vol. 16, p. 237-251.

Snoj, B., Korda, A. P. & Mumel, D (2004) “The Relationships among Perceived Quality, Perceived Risk and Perceived Product Value,” Journal of Product and Brand Management, vol. 13(2), p. 156-67. Spears, N. & Singh, S. N (2004) “Measuring Attitude toward the Brand and Purchase Intentions,” Journal of Current Issues and Research in Advertising, vol. 26(2), p. 53-66.

Spence, H. E., James F. E. & Blackwell, R. D (1970) “Perceived Risk in Mail-Order and Retail Store Buying,” Journal of Marketing Research, vol. 7(August), p. 364-369. Stone, R. N. & Gronhaug, K (1993) “Perceived Risk: Further Considerations for the Marketing Discipline,” European Journal of Marketing, vol. 27(3), p. 39-50. Sunden, A. & Surette, B (1998) “Gender Differences in the Allocation of Assets in Retirement Plans,” American Economic Review, vol. 88, p. 207-221.

Sweeney, J. C., Soutar, G. N. & Johnson, L. W (1999) “The Role of Perceived Risk in the Quality-Value Relationship: A Study in a Retail Environment,” Journal of Retailing, vol. 75 (1), p. 77-105. Sweeney, J. C., and Soutar, G. N (2001) “Consumer Perceived Value: The development of a multiple item scale,” Journal of Retailing, vol. 77, p. 203-220. Taylor, J. W (1974) “The Role of Risk in Consumer Behavior,” The Journal of Marketing, vol. 38(2), p. 54-60. Terence, A. S. & William, O. B (1982) “Warranty and Other Extrinsic Cue Effects on Consumers' Risk Perceptions,” Journal of Consumer Research, vol. 9, p. 38-46. Tsiotsou, R (2006) “The Role of Perceived Product Quality and Overall Satisfaction on Purchase Intentions,” International Journal of Consumer Studies, vol. 30(2), p. 207-220. Wheaton, B., Muthén, B., Alwin, D. F. & Summers, G (1977) “Assessing Reliability and Stability in Panel Models,” Social Methodology, vol. 8, p. 84-136. Wu, L. Y., Chen, K. Y., Chen, P. Y., & Cheng, S. L (In press) “Perceived Value, Transaction Cost, and Repurchase-Intention in Online Shopping: A Relational Exchange Perspective,” Journal of Business Research, Available online 20 September, 2012.


40

Xua, Y., Summersb, T. & Bonnie, D. B (2004) “Who Buys American Alligator? Predicting Purchase Intention of a Controversial Product,” Journal of Business Research, vol. 57(10), p. 1189-1198. Yee, C. J., & San, N. C (2011) “Consumers' perceived quality, perceived value and perceived risk towards purchase decision on automobile,” American Journal of Economics and Business Administration, vol. 3, p. 47-57.

Zeithaml, V. A (1988) “Consumer Perceptions of Price, Quality, and Value: A Means-End Model and Synthesis of Evidence,” Journal of Marketing, vol. 52(July), p. 2-22. BIOGRAPHY Dr. Ya-Hui Wang is an assistant professor of Business Administration at National Chin-Yi University of Technology in Taiwan. She received her Ph.D. degree in financial management from the National Central University in Taiwan. She can be contacted at: Department of Business Administration, National Chin-Yi University of Technology, No. 57, Sec. 2, Zhongshan Rd., Taiping Dist., Taichung 41170, Taiwan, R.O.C. Phone: +886-4-2392-4505 ext. 7783. E-mail: [email protected]. Cing-Fen Tsai is an administrative assistant of Asian First Refrigeration Corporation in Taiwan. She received her MBA degree in business management from the National Chin-Yi University of Technology in Taiwan, R.O.C. She can be contacted with E-mail: [email protected].


41

TIME SERIES MODELING AND FORECASTING INFLATION: EVIDENCE FROM NIGERIA

Ikechukwu Kelikume, Lagos Business School Adedoyin Salami, Lagos Business School

ABSTRACT

A major concern of entrepreneurs and monetary authorities in Nigeria in the past decades was successful prediction general price level movements. The results allow successful planning on the part of monetary authorities and continued profit drive on the part of entrepreneurs and investors. This study uses a univariate model in the form of Autoregressive Integrated Moving Average model developed by Box and Jenkins and multivariate time series model in the form of Vector Autoregressive model to forecast inflation for Nigeria. This paper use changes in monthly consumer price index obtained from the National Bureau of Statistics and the Central bank of Nigeria over the period 2003 to 2012 to predict movements in the general price level. Based on different diagnostic and evaluation criteria, the best forecasting model for predicting inflation in Nigeria is identified. The results will enable policy makers and businesses to track the performance and stability of key macroeconomic indicators using the forecasted inflation. JEL: E3 E17, E31 KEYWORDS: Modeling Inflation, Forecasting, ARIMA, VAR INTRODUCTION

aintaining a reasonable degree of price stability and ensuring an adequate expansion of credit to foster steady and sustainable economic growth have been the primary goals of monetary policy. A challenging problematic macroeconomic economic issue confronting nation states and

monetary authorities today is tracking and predicting the movement in the general price level. Nigeria like most developing countries has had significant gaps between policy formulation, policy implementation and policy targets. In most cases, policy goals lag behind targets and are often unattainable due primarily to the prevalence of policy inconsistencies driven by the inability of monetary authorities to predict inflation and its real determinants. Inflation is a major monetary policy performance indicator and is a useful indicator in informing the public about trends in the movement of leading and lagging macroeconomic indicators. The knowledge of these indicators drives inflationary expectation and therefore serves as a nominal anchor for bargaining process and fixed contracts (Moser, Rumler and Scharler, 2004). Generally, a clear understanding of inflation forecasting techniques is crucial for the success of monetary policy in tracking the movement of macroeconomic aggregates and in maintaining stabile and sustainable economic growth. This paper compares Vector Autoregressive (VAR) model and Autoregressive Integrated Moving Average (ARIMA) model for forecasting the rate of change of the Nigerian Consumer Price Index (CPI). The main attraction to VAR modeling is that it has a natural basis for testing conditional predictability unlike the ARIMA model which is poor in predicting turning points but is relatively robust in generating short term forecast. Thus most empirical analysis on forecasting has focused on the use of VAR while ARIMA is used as a benchmark forecasting tool. Forecasts of the models with the highest predictive accuracy are then evaluated using a range of criteria that characterize optimal forecasts.

M

I. Kelikume & A. Salami | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

42

Following the introductory section, the rest of the paper is organized as follows: Section 2 summarizes the theoretical and empirical literature. Section 3 describes the models, methods and sources of data. Section 4 compares the forecasting performance of the models and evaluates the resulting models with the highest predictive accuracy. Section 5 concludes the paper. LITERATURE The central role of monetary policy in developed, emerging and developing economies is the maintenance of price stability and ensuring of adequate expansion of credit to foster economic growth and development. Generally, economists across economic divides differ in their analysis of the root causes of inflation and in the way and manner the inflationary spiral should be managed and controlled. While the monetarists hold the strong view that sustained growth in money supply not matched by corresponding growth in output will cause inflation at the long run (Milton Friedman 1956, 1960, 1971), structuralist economist explain the long run inflationary trend in developing countries in terms of structural rigidities, market imperfections and social tensions such as relative elasticity of food supply, foreign exchange constraints, protective measures, rise in demand for food, fall in export earnings, hoarding, import substitution, industrialization and presence of political instability ( Thirwell, 1974; and Aghevei and Khan 1977). Given the monetarists view and structuralist view on the root causes of inflation, it is increasingly difficult to forecast inflation in not only developed economies but also in OECD countries and in particular in emerging and developing economies. The empirical studies conducted by Olga, Kamps and Nadine (2009), and Stock and Watson (2008) found that over the longer term (3-year), forecasting horizon, and monetary indicators contain useful information for predicting inflation in most New Member States (NMS) countries of the European Union (EU). Models of inflation forecast and accuracy has evolved in several studies ranging from extrapolation to econometric modeling. The early study of inflation forecast by Landsman and Damodaran (1989) in which the univariate autoregressive integrated moving average method was used drew the conclusion that, ARIMA parameter estimator improves the forecast accuracy of the model because of its lower mean squared percentage error. Although, inflation forecasting with autoregressive integrated moving average method (ARIMA) compares favorably with other forecasting models such as the vector autoregressive method (VAR), and the Bayesian VAR, it has been shown that the ARIMA performs poorly forecasting turning points and yields poor forecast values when applied to volatile and high frequency data (Meyler, Kenny and Quinn (1998). Ho and Xie (1998), using the ARIMA framework, concluded that the ARIMA model is a viable alternative that gives satisfactory results in terms of its predictive performance. According to Wayne (1998), the use of the vector autoregressive model in forecasting exhibits significant degree of predictive accuracy when compared with other forecasting models. This same conclusion was reached by Meyler et al (1998). Applying the Bayesian VAR approach forecasting, they found the VAR approach improves forecasting performance. Black, Corrigan and Dowd (2000), comparing an AR (1) with the Mean Absolute Percentage Errors (MAPEs) of different models adding one variable at a time, found the money supply variable to improve the forecast values of inflation significantly, while the study by Jacobson, Jansson, Vredin and Warne (2001) shows that VAR model with long-run restrictions is useful for both forecasting inflation and for analyzing other issues that are central to the conduct of monetary policy. Using a VAR model, Gottschalk and Moore (2001) assessed the link between monetary policy instruments and inflation in Poland. The result showed that although the exchange rate was found to be effective with respect to output and prices, direct linkage between interest rate and inflation do not appear to be very strong. Toshitaka (2001), found mark-up relationship in estimating and forecasting inflation; excess money supply and the output gap were of importance in determining long run


43

equilibrium correlation model of inflation. Other studies on inflation forecasting in different region produced mixed results. Fritzer, Moser and Scharler (2002), found VAR models outperform ARIMA in terms of forecasting accuracy while Bokhari and Feridun (2006), indicate that VAR models do not perform better than the ARIMA model. Espasa, Poncela and Senra (2002), concludes that ARIMA models outperformed the VECM and dynamic factor models while Hubrich K. (2003), found that VAR models outperformed the autoregressive forecasting models. The study by Alnaa and Ahiakpor, (2005) followed the same pattern as other models proving the VAR modeling technique to be highly efficient in its predictive ability. But, the study by Binner, Bissoondeeal, Elger, Gazely and Mullineux (2005) drew a different conclusion. Using Neural Networks (NN) forecasting model-a nonlinear forecasting approach, they found the VAR and ARIMA modeling technique to be statistically inferior to the Neural Network model.Recent studies by Clausen and Clausen (2010), using Phillips curves showed that model forecasting based on ex post output gaps generally improve the accuracy of inflation forecasts compared to an AR (1) forecast model. The literature is rich in the support of the forecasting strength of ARIMA modeling technique in forecasting (Hill and Fildes, 1984; Libert, 1983; Poulos, Kvanli, and Pavur, 1987; Texter and Ord, 1989). Although, recent studies in Nigeria, have shown the VAR modeling technique to be highly useful in predicting short run forecast (Adebiyi, Adenuga, Abeng, Omanukwe and Ononugbo 2010, Uko and Nkoro 2012), there is however a need to revisit the forecasting ability of both ARIMA and VAR model in Nigeria. METHODOLOGY This study forecasts core inflation in Nigeria with the aid of a univariate time series model in the form of an Autoregressive Integrated Moving Average (ARIMA) model developed by Box and Jenkins and a multivariate time series Vector Autoregressive model (VAR). The choice of both models is linked to resent forecasting success in both ARIMA and VAR modeling. Our objective is to establish the best forecasting model in tracking price movements in Nigeria. The data used in this study is sourced from the Central Bank of Nigeria Statistical Bulletin and the National Bureau of Statistics. The frequency of the data is monthly and the period covered is 2003:01 to 2012:06. The variable used is the rate of change in the consumer’s price index, broad money supply (M2) and our focus is to forecast core inflation. The INF and M2 data gathered was estimated and analyzed with E-views 7 estimation software. Modeling and forecasting inflation with the Box-Jenkins methodology requires the following systematic steps. The first step is the data collection and examination stage, the second step is the identification of the data, while the third step is the estimation of the model. The fourth and fifth step is the diagnostic checks and forecasting stage respectively. ARIMA Model ARIMA entails the use of Box-Jenkins methodology which requires that the sample data be at least more than 50 observations (Meyler et al 1998) and even when sample observations is greater than 50 there is need to examine the data for the existence of structural breaks which if present in the data will necessitate only the examination of a sub-section of the data or the need to introduce a dummy variable but in this case, the data was stationary at levels as shown by Figure 1 and 2, this can easily be verified in the. augmented Dickey-Fuller test of unit root with 5 per cent level of significance reported in Table 1 below The estimation of a univariate time series variable with the autoregressive integrated moving average method ARIMA (p,d,q), requires identification of the appropriate value of p, d and q. Where p denotes the number of autoregressive term, d equals the number of times the series has to be differenced to obtain an I(0) series and q measures the moving average term. The chief identification tool is the plot of the autocorrelation function (ACF), the partial autocorrelation function (PACF), the correlograms and the augmented Dickey and the Fuller (1971, 1981) test for unit roots. From Figure 1 and 2, it can be seen that the series is stationary.


44

Figure 1: Autocorrelation Function Figure 2: Partial Autocorrelation Function

Figure 1 and 2 is the Autocorrelation and Partial Autocorrelation functions showing that the core inflation series is stationary at levels and that it is an Autoregressive Moving Average (ARMA) process. This is seen in the pattern. Table 1: Unit Root Teston Core Inflation with Intercept and a Linear Trend

T-Statistics Prob. Status Augmented Dickey-Fuller test Statistics

-3.4590** 0.0490 I(0)

Test Critical Values : 1% level 5% level 10% level

-4.0420 -3.4504

-3.1505 the table is the result of the augmented dickey-fuller (adf) which indicates the core inflation series is integrated or stationary at levels. ** indicates significance at 5 per cent level. Having obtained the order of integration (d), our next step is to obtain the ARMA pattern in the inflation series by considering the autocorrelation function (ACF), the partial autocorrelation function (PACF) and the associated correlograms. This process involves using the Box-Jenkins methodology where the ACF and the PACF plots are used to predict p and q in the ARMA model. The selection of p and q is usually based on the following characteristics of the ACF and the PACF plots. If data is purely AR (p), then ACF will decline steadily and PACF will cut off suddenly after p lags but if data is purely an MA (q), ACF will cut off suddenly after q lags and PACF will decline steadily. An ARMA (p, q) model usually exhibits a complex pattern in the ACF and PACF function. From the plot of the autocorrelation function and the partial autocorrelation function reported in Figure 1 and 2, we got a clear pattern to help predict p and q. The criteria used in this case are;

+

=

nknLog

nrssLogBIC *)(

(1)

+

=

nknLogLog

nrssLogHQC *))((*2

(2)

+

=

nk

nrssLogAIC *2

(3)

-1

-0.5

0

0.5

1

1.5

1 4 7 10 13 16 19 22 25 28 31

ρ kk

Lag length

PAC

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

1 4 7 10 13 16 19 22 25 28 31

ρ k

Lag length

AC


45

Where, SC = Schwarz criterion, HQC= Hannan-Quinn criterion and AIC = Akaike information criterion, K = the number of coefficient estimated, rss = residual sum of squares and n = the number of observations With the aid of the correlogram and the partial correlogram reported in Figure 1 and 2, we obtained the identified model which is seen clearly in the plot of the ACF and the PACF function reported in Figure 1 and 2. The selection of the ARMA (p, q) model in equation (4) is based on the ACF and the PACF function reported in Figure 1 and 2. Since the core inflation variable was stationary at levels as shown in the unit root test reported in Table 2, there was no need to difference the variable. The graph of the correlogram reported in Figure 1 and 2 reveals some interesting patterns. First, both the ACF function (Figure 1) and the PACF function (Figure 2) exhibits some form of exponential decay. The spikes from the graph shows the ACF to be statistically significant at lags 1, 2, and 14 while for the PACF, lags 1, 2, 12 and 14 appeared to be statistically different from zero The tentatively identified ARMA (p, q) model is specified as follows;

1414121222111414121211 −−−−−−− +++++++= ttttttt UUUUCINFCINFCINFCINFL ββββαααδ (4) Where; CINFL is inflation series at levels; 14121, ααα and are the coefficients of the AR (p) process while 141221 ,, ββββ and are the coefficients of the MA (q) process. AR (p) is autoregressive process while MA (q) is moving average process. VAR Model The seminal work by Sims (1980) brought a succor to the modeling of multivariate autoregressive models with the use of an unrestricted vector autoregressive model (VAR). Conventionally, in the VAR modeling technique, we consider several endogenous variables together with each endogenous variable explained by its lagged values and the lagged values of all other endogenous variables in the model. In VAR estimation and forecasting, a unit root test is not necessary because of the loss of information, observation (Sims 1980). In modeling and forecasting inflation with VAR, we used a univariate autoregressive framework, in which the model is specified describing the interdependence of money supply (broad money supply) and core inflation. In its simplest form, we express the set of n variables collected in the n x 1 vector Yt on their own lags and those of the other variables in the model. The model can be expressed as; Y t = α + i1β Y 1−t + i2β Y 2−t +…+ piβ Y pt− + tu (5) The coefficient matrix i1β are the n x n matrix (made up of the inflation and money supply variables) and

tu is an n x 1 vector of serially uncorrelated random error which is assumed to have a multivariate normal

distribution tu ~ ,0(iidN )∑ u . Assuming Y t is an n x 1 column vector composed of all the variables in our study, the VAR model simply relates current values of Y t to past values of Y t and an n x 1 vector of innovations U t . This can be written precisely as follows;

tttttt yyyyy 1221221111212111111 ∈+++++= −−−− ααβββ (6)

tttttt yyyyy 2221221211222112112 ∈+++++= −−−− ααββα (7)


46

The VAR equation can be written more compactly as;

tttt uyyy +++= −− 2211 ββα (8) Whereα , is an n x 1 vector and sj 'β are n x n metrics. RESULTS AND ANALYSIS After successful identification of the ARMA (p, q) process, we proceed to estimate the ARMA (p, q) process with EVIEW 7 estimation software. The result of the ARMA process is reported in Table 2. Thereafter, we proceed to check the reasonableness of the model fit to the data. This is done by simply obtaining the ACF and the PACF from the residual of the regression estimate reported in Table 2. Table 2: Estimated ARMA (p, q) Coefficient

Variables Coefficient Std. Error T-Statistic Prob. C AR (1) AR (12) AR (14) MA (1) MA (2) MA (12) MA (14)

10.210 1.0247 9.9637 0.0000 0.9709 0.0154 62.996*** 0.0000

-0.2555 0.0389 -6.5681*** 0.0000 0.1892 0.0385 4.9124*** 0.0000 0.8045 0.0750 10.727*** 0.0000 0.2108 0.0571 3.6923*** 0.0004

-0.0764 0.0301 -2.5338*** 0.0130 0.4434 0.0428 10.345 *** 0.0000

R-squared F-statistic

0.9959 Adjusted R-squared 0.9956 3268.8*** Durbin-Watson 1.7836

The table shows that 99 per cent of the systematic variation in core inflation is explained by it lag. The *** indicates that AR, MA and F-statistics are significant at the 1% level while the Durbin-Watson shows the absence of serial correlation. The model diagnostic check entails examining the graphical analysis of the residuals plot of the estimated model and the autocorrelogram plot of the residuals to verify whether the residuals of the estimated models are purely random. This is seen clearly in Figure 3 where the residual of the autocorrelation function at various the lags hover around zero with the exception of lags 2, 9 and 15 while in Figure 4 the residual of the partial autocorrelation function at various lags hover around zero with the exception of lags 2, 15 and 23. The estimated ARMA (p, q) model can therefore be accepted as a purely random walk hence there is need to look for another ARIMA model. We however, proceed to forecasting core inflation with the ARMA (p, q) model with the forecast shown in Table 5 below. To determine the appropriate lag length for the VAR model, we employ the Akaike information criterion (AIC) and Schwarz criterion (SC). This is determined precisely with the aid of E-view 7 estimating software shown in Table 3 below and the period with the lowest criterion was asterisked. Figure 3: Residual Autocorrelation Function Figure 4: Residual Partial Autocorrelation Function

Note: Figure 3 and 4 is the Autocorrelation and partial autocorrelation function of the residual showing that the residual is a random walk.

-0.2

-0.1

0

0.1

0.2

0.3

0.4

1 4 7 10 13 16 19 22 25

ρ k

Lag Length

ACF

-0.3-0.2-0.1

00.10.20.30.4

1 3 5 7 9 11 13 15 17 19 21 23 25

ρ kk

Lag Length

PACF


47

Table 3: VAR Lag Order Selection Criteria

Lag LogL LR FPE AIC SC HQ 0 -2063.8 0.0000 0.0000 39.727 39.778 39.747 1 -1607.2 886.70 0.0000 31.024 31.177 31.086 2 -1527.0 152.70 0.0000 29.559* 29.813* 29.662* 3 -1524.6 4.4521 0.0000 29.590 29.946 29.734 4 -1522.8 3.4246 0.0000 29.631 30.088 29.816 5 -1521.0 3.1998 0.0000 29.673 30.233 29.900 6 -1519.1 3.3489 0.0000 29.713 30.374 29.981 7 -1515.9 5.4911 0.0000 29.728 30.491 30.038 8 -1513.1 4.5463 0.0000 29.753 30.618 30.103

Note: * indicates lag order selected by choosing the lowest AIC and SC. From the VAR estimate in Table 4 below, the intercept (C) for the inflation model was positive and statistically significant at the 5 per cent level showing that inflation cannot be zero at any point in Nigeria. Furthermore, while the parameters for lag inflation was significant at the 1 per cent level as shown by the T-statistics although with different direction of impact, the lag broad money supply was not significant in explaining the variation in current inflation. While the immediate past period inflation will increase current inflation, the two periods past inflation will cause current inflation to fall. Table 4: Vector Autoregression Estimate

Variables CINF M2 C 0.4187

[2.3901]** 1.7473

[1.7737] CINF(-1) 1.8475

[38.893]*** 20649

[-0.7730] CINF(-2) -0.8744

[-18.335]*** 15022

[0.5601] M2(-1) 0.0000

[-0.0780] 0.9273

[9.5022] M2(-2) 0.0000

[-0.0000] 0.0740

[0.7550]

R2 Adjusted- R2

F-statistics Durbin-Watson Stat

0.9941 0.9939

4,495.8*** 2.0591

0.9947 0.9945

4,963.3*** 2.0292

Note: CINF is the core inflation series and M2 is the broad money supply, while numbers in brackets are the lag length, numbers in parenthesis are the T-statistics. ** and *** indicate 5 and 1 per cent level of statistical significance respectively. Furthermore, the adjusted R2 shows that more than 99 per cent of the systematic variation in inflation is explained by lag inflation and money supply although money supply is not significant. Overall, the model was significant at the 1 per cent level with a high F-statistic of 4,495.8 and the Durbin-Watson statistic shows the absence of serial autocorrelation in the model. The result of the ARIMA and the VAR model shows a good fit as shown by the Adjusted R-squared. While current inflation is explained by over 99 percent systematic variation in the independent variables in the ARIMA model, the VAR model also showed predictive power given the value of its coefficient of determination and adjusted coefficient of determination values of 0.9941 and 0.9939 respectively. The F-statistics for both models show an overall significance at the 1 per cent level. While the AR and MA’s were all significant at the 1 per cent level in the ARIMA model, the VAR model only lag inflation was significant at the 1 per cent level while all the lags of money supply was not significant in determining variation in inflation confirming the earlier findings of Salami and Kelikume (2012) that inflation is not always and everywhere a monetary phenomenon. Since our aim is to predict and forecast inflation and compare the forecast values of the ARIMA with that obtained from the VAR model, we generated the forecast values directly using the Eview 7 estimating software.


48

Table 5: Forecast Comparism of the VAR and ARIMA Model

months actual cinf arima forecast var forecast 2012m05 12.40 10.235 12.197 2012m06 12.70 10.227 12.102 2012m07 13.00 10.219 11.929 2012m08 13.30 10.212 11.689 2012m09 13.50 10.205 11.397 2012m10 - 10.199 11.064

Table 5 show a comparison of the ARIMA and VAR forecast with actual core inflation published by the Central Bank of Nigeria. The forecast result shows the VAR forecast to be closer to the actual inflation values. CINF is core inflation. CONCLUSION Current inflation in Nigeria which is 11.3 per cent in September 2012 (All Item Consumers Price Index) is expected to increase in the last quarter of 2012 following severe flooding and the washing away of farm lands in the earlier part of the year. Predicting price movements under periods of volatile food price increases has been made much more difficult. This study forecasts inflation in Nigeria using monthly data over the periods 2003:01 to 2012:06. Two methods used extensively in the literature for forecasting inflation are the Vector Autoregressive Method (VAR) and the Autoregressive Integrated Moving Average Method (ARIMA). This paper uses these methods for study with the sole objective of comparing both forecasting method. While the ARIMA model was a univariate time series model, the VAR model was a multivariate model that incorporates the interdependency amongst several endogenous variables. The result of our estimate from both ARIMA and VAR model tracks actual inflation values for the period 2012: 06 to 2012: 09. However, the VAR model had smaller errors in terms of the minimum square error and is the closest approximate to current inflation in Nigeria. The study forecasted core inflation using VAR for the month of 2012:10 to be 11.06 percent. A major limitation of this study is that it focused on two major forecast tool the VAR method and the ARIMA method and neglected the use of neural network analysis. In addition only core inflation was used as a measure of inflation. Subsequent studies on inflation forecasting in Nigeria should attempt to forecast inflation across a wider spectrum of inflation measures. REFERENCE Adebiyi M.A, Adenuga, A.O., Abeng, M.O., Omanukwe, P.N. and Ononugbo, M.C. (2010) “Inflation Forecasting Models for Nigeria,” Research Department, Central Bank of Nigeria Occasional Paper No. 36 Aghevei, B.B. and Khan, M.S. (1977) ‘‘Inflationary Finance and Economic Growth,’’ Journal of Political Economy, Vol. 85 (4) Alnaa, S. E. and Ahiakpor, F. (2011) ‘‘ARIMA approach to predicting inflation in Ghana,’’ Journal of Economics and International Finance, Vol. 3(5), p. 328-336 Aliyu S. U. R. and Englama A. (2009)‘‘Is Nigeria Ready for Inflation Targeting’’? Journal of Money, Investment and Banking Issue 11, p. 27-44 Binner, J. M., Bissoondeeal, R. K., Elger, T., Gazely, A. M. and Mullineux, A. W. (2005) ‘‘A comparison of linear forecasting models and neural networks: an application to Euro inflation and Euro Divisia,’’Applied economics, Vol. 37 (6), p. 665-680


49

Black, D. C., P. R. Corrigan, and Dowd, M. R (2000) ‘‘New dogs and old tricks: Do money and interest rates provide information content for forecasts of output and prices’’? International Journal of Forecasting Bokhari, S. M. and Feridun, M. (2006) ‘‘Forecasting Inflation through Econometric Models: An Empirical Study on Pakistani Data,’’ Dogus universitesi, Vol. 7(1), p. 39-47 Box, G. E. P. and Jenkins, G (1970) “ Time Series Analysis, Forecasting and Control, Holde-Day, San Francisco Clausen, B. and Clausen, J. R. (2010), “Simulating Inflation Forecasting in Real-Time: How Useful is a Simple Phillips Curve in Germany, the UK and the US”? International Monetary Fund working paper 52 Durevall, D. and Ndung'u (2001) ‘‘A Dynamic Inflation Model for Kenya, 1974-1996’’Journal of African Economies, Vol.10 (1) Espasa, A., Poncela, P. and Senra, E. (2002) ‘‘Forecasting monthly US consumer price indexes through a disaggregated I(2) Analysis, Statistics and Economic Series 1, Universidad Carlos III de Madrid working paper 02-03 Rumler, F., Moser, G. and Scharler, J. (2002), ‘‘Forecasting Austrian HCPI and its components using VAR and ARIMA Models’’ Oesterreichische National Bank, working paper 73 Gali, J., Mark, G. and David L. (2005) “Robustness of the Estimates of the Hybrid New Keynesian Phillips Curve”, Journal of Monetary Economics, Vol. 52, p. 1107-1118 Gottschalk, J. and Moore, D. (2001) ‘‘Implementing inflation targeting regimes: the case of Poland,’’ Journal of comparative economics, Vol. 29(1) Hill, Gareth and Fildes, Robert (1984) “The accuracy of extrapolation methods: An Automatic Box-Jenkins package (SIFT),” Journal of Forecasting, vol. 3, p. 319-323 Landsman, W.R. and Damodaran, A. (1989) ‘‘A comparison of quarterly earnings per share forecasts using James-Stein and unconditional least squares parameter estimator’’ International Journal of Forecasting, Vol. 5, p.491– 500 Milton Friedman (1956) ‘‘the quantity theory of money: A restatement’’Studies in the Quantity Theory of Money, University of Chicago Press, Chicago Milton Friedman (1960)“A Program for Monetary Stability,” The Millar Lectures, No. 3, Fordham University Press, New York Milton Friedman (1971) ‘‘The Theoretical Framework of Monetary Analysis,” National Bureau of Economic Research, Occasional paper 112 Moser, G., Rumler, F. and Scharler, J. (2004.) "Forecasting Austrian Inflation," Working Papers 91, Oesterreichische National bank (Austrian Central Bank). Ho, S. L and Xie, M. (1998) “The Use of ARIMA Models for Reliability Forecasting and Analysis”, Computers and Industrial Engineering - An International Journal, Vol. 35, (1-2), p. 213-221


50

Hubrich Kirstin (2003) ‘‘forecasting Euro area inflation: Does aggregating forecasts by HICP component improve forecast accuracy’’? International journal of Forecasting, Vol. 21(1), p. 119-136 Jacobson, T., Jansson, P., Vredin, A. and Warne, A. (2001)‘‘Monetary Policy Analysis and Inflation Targeting in a Small Open Economy: A VAR Approach. Journal of Applied Econometrics, Vol.16, p.487–520 Libert, G. (1983) “the M-competition with a Fully Automatic Box- Jenkins Procedure,” International Journal of Forecasting, Vol. 2, p. 325-328 Meyler A., Kenny, G.and Quinn, T. (1998) “Forecasting Irish Inflation using ARIMA Models,” Central Bank of Ireland Technical Paper, 3 Meyler A., Kenny, G. and Quinn, T. (1998) “Bayesian VAR Models for Forecasting Irish Inflation,” Central Bank of Ireland Technical Paper, 4 Mahamudu Bawumia, (2002) ‘‘Monetary Growth, Exchange Rates and Inflation in Ghana,’’ West African Journal of Monetary and Economic Integration,vol. 2(2) Olga A., Kamps C. and Leiner-Killinger, N. (2009) ‘‘Inflation Forecasting in the New EU Member States,’’European Central Bank Working Paper1015 Poulos, L., Kvanli. A. and Pavur, R (1987) ‘‘A comparison of the accuracy of the Box-Jenkins method with that of automated forecasting model,’’ International Journal of forecasting, Vol. 3, p. 261-267 Salami, A. and Kelikume, I. (2012), ‘‘Is Inflation Always and Everywhere a Monetary phenomenon?’’ The case of Nigeria,’’ The International Journal of Business and Finance Research Vol. 7(2), p. 105-114 Sims, C. A. (1980) “Macroeconomics and Reality,” Econometrica, Vol. 48(1), p. 1-48 Stock, J. H. and Watson, M. W. (2008)‘‘Phillips Curve Inflation Forecasts’’, National Bureau of Economic Research, Working Papers 14322 Texter, P. A and Ord, J. K.(1989) “Forecasting Using Automatic Identification Procedures: A comparative Analysis,” International Journal of Forecasting, Vol. 5, p. 209-215 Theodore, M. C, Neil, N., Khettry, K., Murray, Devine and Company, Mester, L. J., Novak, A. J. (2011) “ core measure of inflation as predictors of total inflation” Research Department, Federal Resrve Bank of Philadelphia, working paper No. 11-24 Thirlwell, (1974) inflation, savings and growth in developing economics, London: the macorellan press ltd. Toshitaka Sekine (2001) modeling and forecasting inflation in Japan,’’ International Monetary Fund working paper 82 Uko, A. K and Nkoro, E (2012)‘‘Inflation Forecasts with ARIMA, Vector Autoregressive and Error Correction Models in Nigeria,’’European Journal of Economics, Finance and Administrative Sciences issue 50


51

Valle, S. (2002) “Inflation Forecasting with ARIMA and Vector Autoregressive Models in Guatemala,” Economic Research Department, Banco De Guatemala Working Paper. Wayne R. (1998)‘‘Forecasting Inflation using VAR Analysis’’Bank of Jamaica ACKNOWLEDGE The Author would like to thank the editor of the International Journal of Business and Finance Research, the anonymous reviewers and the management of the Lagos Business School for funding this research. BIOGRAPHY Ikechukwu Kelikume is currently a doctoral student of the Swiss University of Economics (SMC) Switzerland and leads sessions in Microeconomic and macroeconomic environment of business at the Lagos Business School (LBS), Pan-African University. He researches and consults in areas which include macroeconomic modeling, financial and monetary economics as well as econometrics and quantitative methods in economics. +234 813 7978 069, [email protected] Adedoyin Salami holds a doctoral degree of Queen Mary College, University of London. He is a full time faculty member at the Lagos Business School (LBS), Pan-African University. He leads sessions in economic environment of business and had served as director of programs for five years until January 2005. He is a member of the Monetary Policy Committee of the Central Bank of Nigeria and had been a member of the Federal Government’s Economic Management Team. Dr. Salami’s research interest include issues in corporate long term financial management; macroeconomic policy; corporate competitiveness and risk management; and characteristics of small and medium enterprise (SMEs). +234 803 5767 562, [email protected]


53

CAPITAL STRUCTURE DETERMINANTS OF PUBLICLY LISTED COMPANIES IN SAUDI ARABIA

Turki SF Alzomaia, King Saud University

ABSTRACT

This paper investigates the capital structure of listed firms in Saudi Arabia, using firm specific data to study the determinants of leverage. The study is based on an analysis of the capital structure of 93 Saudi listed companies. The study extends from 2000 to 2010 and employs cross-sectional pool data methodology. The results suggest there exists a positive relationship between size, growth of the firm and leverage. On the other hand, the results show there are negative relationships between tangibility of assets, profitability, risk and leverage. JEL: G32 KEYWORDS: Capital Structure Determinants, Leverage, Saudi Arabia, Tradeoff Theory,

Pecking Order Theory INTRODUCTION

he capital structure decision is one of the most controversial subjects in corporate finance and has been extensively researched since the seminal paper of Modigliani and Miller (1958). A huge body of financial literature exists relaxing many assumptions of the Modigliani and Miller paper. From

that, several competing theories of capital structure choice were formed including trade-off theory, agency theory, and pecking order theory. Nonetheless, the capital structure decision is an empirical concern as well. Numerous scholarly papers examine the financing decision of public companies theoretically and empirically. In the early stage, the majority of empirical papers examined the case of US companies (Warner 1977, Castanias 1983, Altman 1984, Bradley et al., 1984, Titman and Wessels 1988, Crutchley and Hansen 1989, Harris and Rivav 1991). Rajan and Zingales (1995) extend the analysis of capital structure to G-7 countries focusing on four factors as determinants of leverage: tangibility of assets, the market to book ratio, profitability, and size. Moreover, Booth et al. (2001) extend the analysis of capital structure decision across 10 developing countries. The paper finds that the determinants of capital structure in developed countries are also significant in these 10 developing countries. Since then many financial researchers investigate capital structure decisions in individual countries around the world (Shah and Hijazi 2004, Gaud et al. 2005, Correa et al.2007, Gajural 2005, Waliullah and Nishat 2008). This paper attempts to explain the capital structure decision and its determinants in listed companies of Saudi Arabia. One main characteristic of the Saudi financial market environment is the absence of a corporate tax, a vague and general bankruptcy law, and a undersize and illiquid bond market. Our focus will be trying to determine factors that affect capital structure decisions in a unique institutional environment such as the Saudi Arabia case. We assume that the macroeconomic variables such as inflation and economic growth play minimal role in capital structure decision for Saudi Companies. Thus, for our analysis we consider only specific company factors such as size, growth, tangibility, profitability and risk. Our results indicate that factors affecting capital structure decision in developed and developing countries prevail for the Saudi public companies as well. Size and growth opportunities are found to be positively related to leverage while risk, profitability and tangibility are found to be negatively related to leverage.

T

T. SF. Alzomaia | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

54

Moreover, profitability and risk were the most important independent variables as determinants of the leverage ratio. This paper proceeds as follows. Section 2 briefly reviews the relevant theoretical and empirical capital structure literature. Section 3 is a brief discussion of the Saud Capital Markets and Institutional factors. Section 4 discusses the dataset and the hypotheses. Section 5 briefly explains the methodology. The results are discussed in section 6. Section 7 discusses briefly the decomposition of leverage ratio. Section 8 provides a summary and conclusions.

LITERATURE REVIEW Theoretical Literature Review The publication of Modigliani and Miller (1958) is the most important development in financial economics dealing with capital structure. Modigliani and Miller(henceforth M&M) make the following assumptions: Capital markets are perfectly competitive and frictionless, firms and individuals can borrow and lend at the risk free rate (implying that there is no bankruptcy cost), investors are with homogenous expectations, all cash flow streams are perpetuities (no growth), all firms are assumed to be in the same risk classes, firms issue only risk free debt and risky equity, no agency cost (managers always maximize shareholders wealth) and there exists no signaling opportunity (insiders and outsiders have the same information. Under these specific set of assumptions, M&M argued that in the absence of taxes, the capital structure of the firm is irrelevant to its value. In their 1963 paper, M&M extend the basic propositions in their original article by allowing for a corporate profit tax under which interest payments are deductible. They conclude that the value of the firm is a function of leverage and the tax rate. There are two extreme conclusions of the above theories. On the one hand; M&M (1958) suggest that capital structure is irrelevant while, on the other hand, in (1963) theorize the optimal structure is all debt. Miller (1977) extends the M&M model to consider the effects of personal taxes. Miller argues the M&M model with corporate taxes overstates the advantages of corporate debt financing. Personal taxes offset, to some extent, the benefits from the tax deductibility of corporate interest payments. Therefore, in the equilibrium, the value of the firm will still be independent of its capital structure. The following we will discuss briefly the four main theories of Capital Structure. Modigliani and Miller (1958) assumed implicitly that there are no bankruptcy costs. With relaxing this assumption, many researches argue that with the existence of bankruptcy costs an optimal debt-equity ratio will exist. This is referred to as the trade-off theory. The optimal debt to equity ratio is determined by increasing the amount of debt until the marginal tax gain from leverage is equal to marginal expected loss from bankruptcy costs. In providing the capital structure irrelevancy theorem, M&M implicitly assume no agency cost and mangers will act in the best interest of the firm's shareholders. Jensen and Meckling (1976), however, furnish an agency cost-based rationalization for optimal capital structure determination. Separation of ownership and control as well as conflict of interest between corporate managers, shareholders, and bondholders give rise to agency costs. Thus, the optimal capital structure mix of the firm is established through the efforts of all parties involved (agents, and investors) to minimize total agency-related costs. Therefore, it is possible to establish an optimal financial mix in a world without taxes or bankruptcy cost. Myers (1977) also provides an agency type of argument for the determination of a firm's capital structure. In Myer's model, a firm’s capital structure decision is influenced by the value of its underlying real options (in the form of growth opportunities). The greater this value, the less likely that a firm will take


55

on risky debt. As the proportion of risky debt rises, there is an incentive for managers to take on suboptimal investment strategies, because good investments will tend to benefit bondholders, rather than shareholders. The M&M approach to capital structure irrelevance also assume that the market possesses full information about the activities of a firm. Ross (1977), however, proposes an alternative formulation for the firm's capital structure determination that is based on the existence of symmetric information between the firm's insiders and outsiders. Ross argues that if managers possess inside information, the managerial decisions about the financial structures signal information to the market. Thus, managerial decisions to alter financial structure will alter the market's perception of the firm. Consequently, the value of the firm will rise with leverage. Myers (1984) noted that if we relax the homogenous expectation assumption, asymmetric information by different groups of market participants is admitted. Myers' work resulted in the symmetric information theory of capital structure. In world with asymmetric information, corporations should issue new shares only if they have extraordinary profitable investments that cannot be postponed or financed by debt, or if management thinks the shares are overvalued. Moreover, investors recognize this tends to reduce the firm's share price when it announces plans to issue new shares (signaling bad news). Finally, Myers suggests a pecking order theory of capital structure. Firms are said to prefer retained earnings as their main source of funds, next in order of preference is debt, and last comes external equity financing. Empirical Literature Review Warner (1977) discussed the role of bankruptcy costs in capital structure decisions and presents evidence of the direct costs of bankruptcy for a number of US railroad firms. Warner collects data for 11 railroad bankruptcies that occurred from 1933 to 1955. The study shows that direct bankruptcy costs may not be large enough to be a determinant factor in capital structure decisions. Castanias (1983) and Altman (1984) follow Warner's research of bankruptcy costs. Castanias analyzes the relation between failure and leverage in small firms. The study finds that firms with high rates of failure tend to have low debt-equity ratios. Although Castanias' results indicate the possibility of an optimal capital structure, the study focuses on industry data and does not account for indirect bankruptcy costs. Altman, in contrast, provides evidence of indirect costs. Altman compares expected profits with actual profits and shows that indirect costs are 8.1% of the value of the firm three years prior to bankruptcy and 10.5% the year of bankruptcy. The study indicates that total bankruptcy costs are not trivial. Bradley, Jarrell and Kim (1984) use cross-sectional, firm specific data to test for the existence of an optimal capital structure. BJK analyze three firm specific factors that influence the optimal capital structure: the variability of firm value, the level of non-tax shields and the magnitude of the cost of financial distress. Bradley et.al. find that firm leverage ratios are related inversely to earnings volatility provided there are significant cost of financial distress. However, BJK’s results indicate a strong positive relationship between leverage and non-tax shields. Titman and Wessels (1988) analyze the explanatory power of various factors that have been proposed by a number of capital structure theories as attributes that influence the choice of optimal capital structure. Crutchley and Hansen (1989) present an empirical test of the Agency theory. They focus on equity agency costs that result from the conflict of interest between managers and stockholders. C&H identify five proxies for agency costs; i.e., earning volatility, discretionary investment (advertising expenses and R&D), flotation costs, diversification loss to managers from holding firm's common stock, and firm size. The results are consistent with the Agency theory. An increase in earnings volatility will have a significant negative impact on leverage. Also, if discretionary expense increased, the firm uses less debt. Moreover, the authors find that large firms tended to rely more on debt. Thies and Klock (1993) provide


56

some support for Pecking Order theory. They suggest that the pecking order theory provides one explanation for the inverse relationship found in their study between profitability and all forms of leverage. Rajan and Zingales (1995) examines the capital structure of G-7 countries (US, UK, Japan, Germany, France, Italy, and Canada). The authors focus on four factors as determinants of leverage: tangibility of assets, the market to book ratio, profitability, and size. The results of the study indicate that tangibility of assets is positively correlated with leverage in all countries. The results also indicate that leverage increase with size in all countries except Germany. On the other hand, the market to book ratio is negatively correlate with leverage in all countries except Italy where it is positively correlated. Furthermore, profitability is negatively correlated with leverage in all countries except Germany. However, Bevan and Danbolt (2002), based on analysis of capital structure of 822 UK firms, examine the sensitivity of Rajan and Zingales' results to variation in leverage measures. They find that Rajan and Zingales' results are highly dependent upon the precise definition of leverage being examined. Thus, the authors argue that the determinant of leverage vary significantly depending on the nature of the debt sub- component being studied. Booth et al. (2001) analyzed capital structure decisions of firm across 10 developing countries (Brazil, Mexico, Jordan, Indi, Pakistan, Turkey, Zimbabwe, Korea, Thailand, and Malaysia) for the period 1980-1990, utilizing both firm specific and institutional factors. The authors find that related factors for explaining capital structure in developed countries are also relevant in developing countries. In general, the results show that for developing countries profitability was the most successful independent variable and negatively related to leverage. Size and tangibility of assets are positively related to the leverage ratio. Shah and Hijazi (2004) analyze the determinants of capital structure in listed firms in Pakistan for the period 1997 to 2001. They follow Rajan and Zingales (1995) of selecting only four independent variables: size, tangibility of assets, growth, and profitability. The results show that asset tangibility and size are positively correlated with leverage. In contrast, growth and profitability are negatively correlated with leverage. Gaud et al. (2005) analyses the determinants of the capital structure for 104 Swiss listed companies from 1991-2000, employing a dynamic panel framework. The results show that size and tangibility of assets are positively related to leverage, whereas profitability and growth are negatively related to leverage. Following the same methodology of dynamic panel framework, Correa et al. (2007) examines the determinants of capital structure decisions of the largest 500 Brazilian companies for the period 1999-2004. The results show that profitability and tangibility of assets are negatively related to leverage, while business risk is positively related to leverage. Gajural (2005) investigates the pattern and determinants of capital structure of Non-financial Nepalese firms for the period 1992-2004. The analysis shows that asset structure and size are positively related to leverage ratio. While liquidity, growth opportunities, profitability, and non-debt tax shield are negatively related to the leverage ratio. Frank and Goyal (2009) investigate the relative importance of several factors in the capital structure decision of listed US companies for the period of 1950-2003. Among these factors they found a core of six reliable factors that correlated with cross-sectional differences in leverage. The results of the study indicate that leverage is positively related to firm size, tangible assets, median industry leverage, and expected inflation. On the other side, leverage is negatively related to profits and market-to-book ratio. According to the authors all six factors, except profit, have the sign predicted by the static tradeoff theory in which the tax saving of debt are traded-off against deadweight bankruptcy costs.


57

Waliullah and Nishat (2008) examines capital structure determinant choices of 533 non- financial firms publicly listed on Karachi Stock Exchange (KSE) for the period from 1988 to 2005. Employing autoregressive distributed lag (ARDL) methodology, the paper divided the determinants of financiering behaviors into firm’s specific characteristics, reforms and industry characteristics. The results indicate that size of the firm and growth opportunities are positively related to the debt ratio. On the other hand, the results suggest that profitability and liquidity are negatively correlated with debt financing. Furthermore, the results show that firms with high risk and more tangible assets will rely more on equity financing and use less debt. SAUD CAPITAL MARKETS AND INSTITUTIONAL FACTORS Equity Market As of the end of 2010 there are 146 listed companies in Saudi Arabia with a market capitalization of about 80 percent of GDP. Market Capitalization is dominated by petrochemical companies (36.6 percent), financial companies (27.6 percent) and telecoms (10 percent). In April 2008, the Capital Market Authority restructured the Saudi stock market sectors based on the nature of business of each listed company, its income, and earnings structure. After the new market structure, the Saudi stock market consists of 15 sectors instead of its previous eight sectors. Since the new industry coding established only at the end of the period for our study, we will not include the average leverage of the industry as an explanatory variable in the study. The following table shows Saudi capital market indicators over the period 2000-2010. Table 1 illustrates some important characteristics of the Saudi Equity market during the period of the study. For instance, the number of listed company increased from 75 companies at the end of year 2000 to 111 companies at the end of year 2007 and reached 146 at the end of year 2010. Furthermore, Table 1 indicates the importance of the equity market in the Saudi economy, which can be approximated by market capitalization of listed companies to the GDP. The ratio of market capitalization to GDP was 32 percent at the end of 2000, then reaches its peak of 208% at the end of 2005, then fell to 79.6 percent at the end of 2010. The main characteristic of the stock market during the period of study, 2000-2010, is the high volatility of the market. The main index, the Tadawul All Share Index (TASI) was only 2,258 point at the end of 2000. Then from the year 2003 on it started to accelerate rapidly until reaching its peak of 20,635 points on February 25, 2006. Thus, between 2003 and its peak the index gained a staggering 700 percent. From that peak, the correction started and the market collapsed reaching 7,933 points at the end of 2006. Another collapse occurs during the world financial crises of 2008 when the Saudi index reach its bottom at the end of the year 2008 of 4,803 points. For the years 2009-2010 the index swings between 6,000-7,000 points. In general, even with this very obvious fluctuation, the equity market becomes an important financing tool for Saudi companies during the period of study. Bond Market Bond market development in Saudi Arabia traces its roots back to mid-1988, when government securities were issued in the domestic market to fund government fiscal deficits. The market stagnated until 2009 when the Capital Market Authority (CMA) approved the trading of Sukuk (Islamic bond) and traditional bonds for the first time in Saudi Arabia. This is an important step towards launching a second regulated market. However, the Saudi bond market is still viewed as illiquid and thin. The total amount of issued Sukuks and Bonds since the foundation of the market to end of 2010 stood at only at SRs 35.7 billion with 7 issuances by 3 companies. Thus, with such undersized bond and Sukuk market companies continue to rely heavily on short term bank loans as a the main debt instrument.


58

Table 1: Saudi Equity Market Indictors

End of Period Year

Listed Companies Market Capitalization of issued shares (Billion RLs)

Market Capitalization to

GDP (%)

Share Price Index (1985= 1000)

No. Annual % Change Value Annual % Change Index

Annual % Change

2000 75 9 255 11.3 32.2 2258.29 65

2001 76 1 275 7.8 40.5 2430.11 8

2002 68 -11 281 2.5 40.2 2518.08 4

2003 70 3 590 100.1 74 4437.58 76

2004 73 4 1149 94.7 123 8206.23 85

2005 77 5 2439 100.12 208 16712.64 104

2006 86 12 1226 -49.7 92.5 7933.29 -53

2007 111 29 1946 58.8 136 11038.66 39

2008 127 14 924.5 -52.5 52.2 4802.99 -56

2009 135 6 1195 29.3 82.8 6121.76 27

2010 146 8 1325 11 79.6 6620.75 8

This table shows some indicators of the Saudi equity market for the period under the study (2000-2010). These indicators include: number of listed companies, market capitalization of issued shares, market capitalization to GDP, and share price index. The number of listed companies increases from 75 companies in year 2000 to 146 companies in year 2010. The ratio of market capitalization to GDP was only 32 percent at the end of the year 2000, reaches its peak of 208% at the end of the year 2005, then dropped to 79.6 percent at the end of year 2010. The main index (TASI) was only 2,258 point at the end of 2000, then reaches its peak of 20,635 point in February 25, 2006. During the world financial crises of 2008 the Saudi index reaches its bottom at the end of 2008 at 4,803 points. The Sources of the data are: Saudi Stock exchange Company (Tadawul) and Saudi Arabian Monetary Agency (SAMA).

Bank Lending Historically commercial bank loans have been the main source of financing corporations in Saudi Arabia. According to the Saudi Arabian Monetary Agency (SAMA), at the end of year 2010, there were 21 commercial banks operating in Saudi Arabia including branches of five foreign banks. During the 1980s and 1990s, bank financing and lines of credit dominated corporate financing channels. Bank credit continues to be the most popular financing channel, catering to more than 80% of the total funding needed. The main characteristic of bank loans is their short-term nature. For example, 59% of total loans to companies were short-term loans with less than one-year maturity. This is in a line with Booth et al. (2001) findings that for ten developing countries the amount of long-term debt is much lower in comparison with developed countries. During the 1970’s the Saudi Government created five major lending institutions namely; Public Investment Fund, Saudi Credit Bank, Saudi Industrial Development Fund, Saudi Agricultural Bank, and the Real Estate Fund. These government institutions provided direct credit programs to major business sectors in Saudi Arabia. These programs are medium and long-terms credit programs. They charge minimal fees. The total loans distributed by these institutions since their inception up to the end of 2010 is SRs 414.3 billion. Tax System Saudi Public companies are not subject to income tax. Instead, they are subject to an Islamic Tax called 'Zakat', which is a religious tax based on Islamic law (the Sharia) and is assessed on earnings and holdings. Zakat is levied at a flat rate of 2.5% and is chargeable on the total of the company's capital resources and income that are not invested in fixed assets. These include the company's capital, net profits, retained earnings and reserves not created for specific liabilities. Moreover, loans used to finance


59

acquisition of capital assets, investments, and inventory are added to Zakat bases. Only resources (including income) which have been held for at least 12 months are subject to Zakat. Thus, we presume that there are no obvious tax advantages for debt financing for Saudi Companies and therefore the tax will not be considered as a factor for determining capital structure decisions for Saudi companies. DATA AND HYPOTHESES The sample consists of non-financial Public Saudi Firms over the years 2000-2010. The data are annual and the data source is Gulf Base (Zughaibi and Kabbani Financial Consultants (ZKFC)). The database contains balance sheet, profit and loss, and cash flow statement information for all Saudi public companies. The exclusion of financial firms was motivated by the fact that these firms have to comply with very strict legal requirements pertaining to their financing (Gaud et al., 2005). There were 146 listed companies in the Saudi market by the end of the year 2010. However, after excluding financial firms (11 banks and 31 insurance companies) the number of companies in the study is 104 companies. Moreover, we omitted any company with less than 3 years of available data. As a result we exclude all listed IPOs companies in the years 2009 and 2010 (11 companies). These procedures resulted in smaller number of 93 companies in our sample, with a total of 967 observations available for analysis. Table 2 shows basic statistics of selected financial statement items of Saudi companies for the period under the study. One important element from Table 2 is that almost 36% of total observations have no long term debt, and the value long term debt to total assets is around 20%. This assures the notion that Saudi companies depend heavily on short-term bank loans as a main source of leverage. The low long-term debt ratio are consistent with Booth et al. (2001) findings that companies in developing countries have substantially lower long term debt compared with companies in developed countries. Table 2: Statistics of Selected Financial Statement Items

Variable Observations (N)

Median (SR Million)

Mean (SR Million)

Standard Deviation

Percentile 0.10 0.90

Total Assets 967 1,122,984 6,897,762 26,195,388 183,157 9,641,749 Current Liabilities 967 173,286 1,284,519 4,910,702 19,750 1,418,089 Long Term Debt 625 19,202 1,347,604 7,001,614 0 1,337,799 Book Equity 967 659,070 3,150,109 10,241,793 117,073 5,177,969 Book Liabilities 967 272,080 3,747,312 16,374,737 28,269 3,327,414 Profit 967 61,117 374,113 2,218,699 -13,492 538,288

This table shows the number of observations, the median, mean, standard deviation, the .10 percentile and the .90 percentile of some key financial statement items for Saudi listed companies (excluding the financial institutions) covering the period 2000-2010. The number of companies under the study is 93 companies. From the total of 967 observations, only 625 of them have some form of long term debt. In accordance with previous studies concerning capital structure decision, proxies of the variables covered were used for analysis of leverage determinants. Numerous definitions of leverage have been suggested in the literature. In this study, the leverage ratio is defined as the ratio of book value of total debt divided by book value of total assets. We consider the book leverage rather than market leverage since we think the Saudi stock market was very volatile during the period of the study. Thus, using market leverage will be unreliable since there will be stock mispricing across the stock market over the period of the study. Furthermore, many empirical studies use long term debt only in calculating the leverage ratio. However, as mentioned before, looking carefully at the data we notice that many Saudi companies have zero long term debt which can be attributed to the new and illiquid bond market in Saudi Arabia. Thus, many of those companies depend mainly on short commercial banking loans as the only source of debt. Therefore, we consider the total debt (short + long term debt) in the measurement of the leverage ratio. In this study we define the dependent variable (leverage ratio) as follows:


60

Leverage Ratio = 𝑇𝑜𝑡𝑎𝑙 𝐿𝑖𝑎𝑏𝑙𝑖𝑡𝑖𝑒𝑠 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠

(1) For the independent variables we extend Rajan and Zingales' model (1995) to include business risk. Thus, our independent variables include: size, growth opportunities, tangibility of the assets, profitability, and business risk. Large firms are usually more diversified and have more stable cash flow. Therefore, they are less risky. This results in lower cost of debt as well as easier access to the external debt markets. Accordingly, we predict a positive relationship between size and leverage. In this study, firm size is measured by the natural log of sales. Size = ln(Sales) (2) Hypothesis 1: Firm size will have a positive relationship with leverage. Due to the agency cost of debt firms with high growth opportunities are expected to rely more on retained earnings and stakeholders co-investment than debt financing. Thus, we expect a negative relationship between growth opportunities and leverage. While the majority of empirical studies employ the market-to-book value as a proxy for growth opportunities, we measured it by the change in log of sales. Even though many studies employ log assets as a proxy of the firm growth, we employ log of sales as a proxy of growth. This is will not affect the analysis since there exists high correlation between change in assets and change in sales. The main reason for not using the market-to-book value is that, as mentioned before, the Saudi Stock market witnessed great volatility during the period under study. Thus using any market value will be unreliable. Therefore, following Titman and Wessels (1988), the growth rate of sales will be used as a proxy for growth opportunities. 𝐺𝑟𝑜𝑤𝑡ℎ = 𝑆𝑎𝑙𝑒𝑠𝑡−𝑆𝑎𝑙𝑒𝑠𝑡−1

𝑆𝑎𝑙𝑒𝑠𝑡−1 (3)

Hypothesis 2: The percentage change of sales will have a negative relationship with leverage. Tangible assets can be used as collateral and are less subject to information asymmetries. As a result, tangible assets minimize the agency cost of debt. According to agency cost and information asymmetry theories, firms with high tangible assets tend to depend more on debt financing. Tangibility of assets is defined as the ratio of fixed assets to total assets. We expect a positive relationship between tangibility of assets and leverage. 𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝐴𝑠𝑠𝑒𝑡𝑠 = 𝐹𝑖𝑥𝑒𝑑 𝐴𝑠𝑠𝑒𝑡𝑠

𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 (4)

Hypothesis 3: The greater the proportion of tangible assets the higher the leverage. The relationship between profitability and leverage is an unresolved issue in capital structure theories. In one hand, according to pecking order theory, firms prefer retained earnings as their main source of funds. Next in order of preference is debt, and last comes external equity financing. On the other hand, trade-off theory suggests that profitable firms prefer debt financing to benefit from the tax shield. However, in the case of Saudi Arabia where there is no tax advantage of debt and most profitable companies usually maintain large retained earnings, we believe that Saudi companies will exploit retained earnings as the first source of fund before turning to raise debt. Profitability will be measured by return on assets and we anticipate a negative relationship between profitability and leverage.


61

𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑙𝑖𝑡𝑦 = 𝑁𝑒𝑡 𝑃𝑟𝑜𝑓𝑖𝑡𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠

(5) Hypothesis 4: Profitability of the firm will have a negative relationship with leverage. Firms with high volatility of earning might find some difficulty of honoring the payment of debt obligations, which will result in high probability of bankruptcy. Thus, firms with high volatility of cash flow can lower their risk by reducing debt levels. We measure risk by variability of the return on assets (standard deviation of return on assets) and anticipate a negative relationship between risk and leverage. 𝑅𝑖𝑠𝑘 = 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑎𝑠𝑠𝑒𝑡𝑠 = σ (𝑅𝑂𝐴) (6)

Hypothesis 5: The variability of the return on assets will have a negative relationship with leverage. Table 3 shows a large difference for the leverage ratio for the Saudi companies which range from only 9.4% for the 10th percentile to 63.2% for the 90th percentile. The average debt ratio is 33.6% for Saudi public companies, which is comparable to the debt ratio of some of developing countries (Booth et al, 2001) such as Brazil 30.3%, Mexico 34.7%. However, the debt ratio is much lower in comparison to debt ratios of other developing countries included in Booth et al. study. Examples include the debt ratio for South Korea 73.4%, India 67.1%, Pakistan 65.5%, and Turkey59.1%. Furthermore, the debt ratio of Saudi Companies is much lower than the debt ratio of developed countries. Rajan and Zingales (1995) find debt ratio for listed companies in Germany 73%, France 71%, Italy 70%, Japan 69%, US 58%, Canada 56%, and UK 54%. Table 3: Descriptive Statistics for Leverage Measure and Explanatory Factors

Variable Observations (N) Median Mean Standard Deviation Percentile .10 .90

Leverage Ratio 967 0.292 0.336 0.205 0.094 0.632 Firm Size 967 12.726 12.281 3.073 9.714 14.815 Firm Growth 967 0.069 0.441 3.827 -0.172 0.474 Tangibility of Assets 967 0.681 0.651 0.211 0.330 0.896 Profitability 967 0.055 0.066 0.103 -0.025 0.185 Risk 967 0.0455 0.055 0.047 0.017 0.097

This table presents descriptive statistics of the leverage ratio and five independent variables: firm size, firm growth, tangibility of assets, profitability, and risk. The sample contains 93 companies listed in the Saudi stock exchange (TASI). The data covers 2000-2010.We define the leverage ratio as total liabilities divided by total assets. We measure size as the natural logarithm of sales. We define growth as the parentage change of sales. We measure asset tangibility by fixed assets divided by total assets. Profitability is net profit divided by total assets. Risk is defined as standard deviation of return on assets. The leverage ratio for Saudi Companies, with a mean of 33.6 percent, is low in comparison to the leverage ratio of most developed and developing countries. For the independent variables the table shows the size of Saudi companies generally rang between mid-size companies to large-size companies, ranging from 9.7 to 14.8. The growth opportunities demonstrate significant variability ranging from negative 17.2% to positive 47.4%. The value of the mean of the growth opportunities is about 44.1% which is much higher than the value of the median 6.9%. The table also shows high mean and median of tangibility of assets (65.1% and 68.1%), in which reflect the intense use of fixed assets for the Saudi's public companies. Table 4 shows the correlation coefficients between and among leverage ratio and each of the expletory variables, as well as the correlation among the independent variables.

Table 4 shows the leverage ratio has a positive and significant correlation with size and growth. Conversely, the leverage ratio has a negative and significant correlation with tangibility, profitability and risk. The correlations among independent variables show that growth has non-significant correlations with any of the explanatory variables. Size has a positive significant correlation with tangibility and


62

negative but non-significant correlation with risk. Moreover, we examine the variance inflation factor (VIF) to evaluate for the presence of multicolinearity among the independent variables. The VIF statistics are substantially lower than 10 indicating no multicolinearity between the independent variables. This implies we do not need to eliminate any independent variables for reasons of multicolinearity.

Table 4: Correlations between Individual Variable and VIF Coefficients

Leverage Ratio

Size Growth Tangibility Profitability Risk VIF

Leverage Ratio

Pearson Correlation

1 0.302*** 0.083** -0.152*** -0.117*** -0.116***

Size Pearson Correlation

1 -0.034 0.256*** 0.279*** -0.002 1.134

Growth Pearson Correlation

1 0.047 0.001 -0.001 1.004

Tangibility Pearson Correlation

1 -0.246*** 0.027 1.113

Profitability Pearson Correlation

1 -0.134*** 1.145

Risk Pearson Correlation

1 1.020

This table presents Pearson correlation coefficients for the variables used in the analysis and VIF (variance inflation factor) tests between independent variables. The sample contains 93 companies listed in the Saudi stock exchange (TASI). The data cover the period 2000-2010. Leverage ratio is defined as total liabilities divided by total assets. Size is defined as the natural logarithm of sales. We define growth as the parentage change of sales. We measure asset tangibility by fixed assets divided by total assets. Profitability is net profit divided by total assets. Risk is the standard deviation of return on assets. ***, **, and * indicate significant at the 1, 5 and 10 percent level respectively. METHODOLOGY We follow the literature by using a cross-sectional pooled data model to study capital structure decision determinant factors of Saudi Companies. The firm's debt ratio will be regressed against the natural log of its sales, the change in log of total sales, the tangibility of its assets, its return on assets, and the standard deviation of its return on assets. The coefficients are estimated using ordinary least square (OLS). For the outliers in our data sample we follow Bevan and Danbolt (2002), eliminate them by winsorising the dependent variable and all independent variables at the one percent level. The regression equation is:

𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 = ƒ �𝑠𝑖𝑧𝑒,𝑔𝑟𝑜𝑤𝑡ℎ 𝑜𝑝𝑝𝑜𝑟𝑡𝑢𝑛𝑖𝑡𝑖𝑒𝑠, 𝑡𝑎𝑛𝑔𝑖𝑏𝑙𝑖𝑡𝑦 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡𝑠, 𝑝𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑙𝑖𝑡𝑦, 𝑏𝑢𝑠𝑖𝑛𝑒𝑠𝑠 𝑟𝑖𝑠𝑘 � (7)

𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒�𝑓𝑖𝑟𝑚𝑖,𝑡� = α + β1 𝚕𝚗 𝑠𝑎𝑙𝑒𝑠𝑖,𝑡 + β2 ∆𝚕𝚗 𝑠𝑎𝑙𝑒𝑠𝑖,𝑡 + β3 𝑡𝑎𝑛𝑔𝑖𝑏𝑙𝑖𝑡𝑦 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 +β4 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + β5σ 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + ε𝑖,𝑡 (8)

Where i denote firm and t denotes the time, α is the intercept and ε𝑖,𝑡 is error term.

EMPIRICAL RESULTS From the result of our analysis we construct our regression model as follows:

𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 = 0.207 + 0.023𝚕𝚗 𝑠𝑎𝑙𝑒𝑠 + 0.004 ∆𝚕𝚗 𝑠𝑎𝑙𝑒𝑠 − 0.125𝑡𝑎𝑛𝑔𝑖𝑏𝑙𝑖𝑡𝑦 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡𝑠

− 0.521𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 − 0.638 σ 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 + ε𝑖,𝑡

Tables 5 shows the regression model summery as well as the output of the regression analysis. For our model in general, the R2 is 17.2%, which means these five independent variables account for only 17.2% of the variation in leverage ratios for listed Saudi companies. This value is close to the R2 of Frank and


63

Goyel (2003) of 17.5%. The F-statistics shows the validity of the model with a value of 41.140 which is significant at the one percent level meaning the model is capable of determining variation of the total debt ratio of Saudi listed companies. Table 5: The Model Summary and Cross Sectional Regression Results

95.0% Confidence Interval for B

Coefficients B Std. Error t-value Lower Bound Upper Bound (Constant) 0.207*** 0.036 5.718 0.136 0.278

Size 0.023*** 0.002 10.885 0.019 0.027

Growth 0.004*** 0.002 2.640 0.001 0.007

Tangibility -0.125*** 0.030 -4.174 -0.184 -0.066

Profitability -0.521*** 0.062 -8.399 -0.643 -0.399

Risk -0.638*** 0.128 -4.975 -0.889 -0.386

R-Square 0.176 MSE 0.035

Adjusted R- Square 0.172 Durbin-Watson 0.535 F 41.140 AIC -3243.835 Prob (F Statistic) 0.0001

This table shows results of the estimates from the Ordinary Least Square (OLS) Model. The sample contains 93 Saudi Firms listed in the Saudi Stock Exchange for which there is a minimum of 3 consecutive years of data for the 2000-2010 period. The leverage ratio was regressed against five independent variables: size, growth, tangibility, profitability, and risk. The estimated model is: 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒�𝑓𝑖𝑟𝑚𝑖,𝑡� = α + β1 𝚕𝚗 𝑠𝑎𝑙𝑒𝑠𝑖,𝑡 +β2 ∆𝚕𝚗 𝑠𝑎𝑙𝑒𝑠𝑖,𝑡 + β3 𝑡𝑎𝑛𝑔𝑖𝑏𝑙𝑖𝑡𝑦 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + β4 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + β5σ 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + ε𝑖,𝑡 . We define the leverage ratio as total liabilities divided by total assets. Size is the natural logarithm of sales. We define growth as the parentage change of sales. We measure asset tangibility by fixed assets divided by total assets. Profitability is net profit divided by total assets. Risk is defined as standard deviation of return on assets. ***, **, and * indicate significant at the 1, 5 and 10 percent level respectively. The results of the study show that size has a positive and significant relationship with leverage, though the size of the coefficient tends to be small. This suggests that size of the company has limited impact on the capital structure of Saudi Companies. Growth has a significant and positive relationship with leverage, contrary to our expectations, though the size of the coefficient tends to be small. This finding is consistent with the pecking order theory which predicts that growth companies accumulate more debt over time. One the other hand, this finding is contradictory to the agency theory prediction where firms with greater growth opportunities are expected to use less risky debt. Since the coefficient of growth is small, growth has very little effect of the capital structure of Saudi Companies. Tangibility has a negative and significant relationship with leverage, opposite from what we anticipated. This negative relationship is in accordance with the pecking order theory which asserts that because of low asymmetric information, large tangible assets makes equity issuance less costly. Another explanation for this unanticipated relationship between tangibility and leverage is that, as Beger and Udell (1994) argue, firms with close relationships with creditors need to provide less collateral because the relationship substitutes for physical collateral. With only 11 commercial banks in Saudi Arabia, at the time of the study, the close relationship between banks and listed companies is obvious. Furthermore, this outcome confirms the results of Booth et al. (2001) that total debt ratios decrease with the tangibility of assets. Profitability has a significant and strong negative relationship with leverage, with a size of a coefficient of -0.521. This result is consistent with the pecking order theory where profitable firms are predicted to use less debt. Booth et al. (2001) argue that the strong negative relationship can be related to agency and information asymmetry problems as well as the underdeveloped nature of the long-term bond market, which we believe is the case in Saudi Arabia. Risk has a significant and strong negative relationship with leverage. This means firms with more volatile cash flow will use less debt. This result is consistent with agency theory which predicts that an increase in


64

earnings volatility will have a significant negative impact on leverage. In summary, it seems that risk and profitability are the strongest explanatory powers of capital structure determinants for Saudi companies. DECOMPOSITION OF LEVERAGE RATIO Bevan and Danbolt (2002) suggest that the determinants of leverage are sensitive to the components of debt being analyzed. In addition, since we found that almost 36% of the study observations have no long-term debt, we think it more accurate if we divide the debt ratio to long term debt ratio and short term debt ratio. Thus, we decompose the leverage ratio into its sub-component as long and short term debt ratios, and then estimate the extent to which each of these ratios might be related to our five explanatory variables. The long-term debt ratio is defined as total liabilities minus current liabilities divided by total assets. The short-term debt ratio is defined as current liabilities divided by total assets. As discussed in the main body of the paper all five explanatory variables have significant relations with the total debt ratio. However, growth opportunities and tangibility of assets appeared with signs contrary to expectations. As noted earlier, total debt ratio risk and profitability, both negatively related to leverage, are the major factors determining the capital structure for Saudi companies. However, when we decompose the total debt ratio into long-term ratio and short-term ratios we get different results for some of coefficients as shown in Table 6. For the long-term debt ratio model, size is negatively related to leverage instead of positively related to leverage with total debt model, still for both models size have very small effect on the capital structure of Saudi companies. Growth is positively related to long term leverage but with a small effect. Tangibility of assets becomes positively related to the long-term leverage. Both profitability and risk have the same sign as before but with less effect when measuring long-term debt than the total debt ratio. Adjusted R2 and F-statistic are a little lower with long term debt ratio with values of 0.165 and 23.932 respectively. For the short-term debt model, all coefficients are significant and have the same signs as the total-debt model. However, tangibility becomes the most important factor for explaining the capital structure, followed by profitability and risk. Thus, the order of the importance of these three factors reverses. Additionally, the short-term model comes with the best explanatory power compared with the other two models. The adjusted R2 increased to 28.7 which mean these five independent variables account for 28.7% of the variation in short-term leverage ratios for listed Saudi companies. The F-statistics shows better validity of the model with a value of 78.678 in comparison to 41.140 for the total-debt ratio and only 23.932 for the long-term debt model. The short-term model is best fits the data set of listed companies in Saudi Arabia. These results assure the claim of Bevan and Danbolt (2002) that the determinants of leverage are significantly sensitive to the components of debt being analyzed. CONCLUSION This paper presents a study of capital structure determinants for 93 listed companies in Saudi Arabia for the period 1999-2010. The analysis is conducted using a cross-sectional pooled model. The study suggests size and growth opportunities are positively related to leverage. Tangibility, profitability and risk are negatively related with leverage. Moreover, the results indicate that risk and profitability are the major factors driving capital structure decisions for listed companies in Saudi Arabia. Our results provide some unexpected signs for some coefficients namely growth opportunities and tangibility of assets. In general, most empirical results of the study support the pecking order theory. This study can be extended by considering ownership structure and median industry leverage as explanatory variables for capital structure decision of Saudi companies.


65

Table 6: Cross-sectional Results of Decomposed Leverage Ratio

Model Coefficient St. Error t-Value 1. Total Debt Ratio Model Constant 0.207*** 0.036 7.718 Size 0.023*** 0.002 10.885 Growth 0.004** 0.002 2.64 Tangibility -0.125*** 0.03 -4.174 Profitability -0.521*** 0.062 -8.399 Risk -0.638*** 0.128 -4.975 Adjusted R2 0.172 F-statistic 41.14 Prob. of (F-Stat.) 0.0001 2. Long Term Debt Ratio Model Constant 0.108** 0.035 3.115 Size -0.006** 0.002 -3.087 Growth 0.003* 0.001 2.498 Tangibility 0.194*** 0.028 6.96 Profitability -0.158** 0.051 -3.064 Risk -0.467*** 0.095 -4.921 Adjusted R2 0.165 F-statistic 23.932 Prob. of (F-Stat.) 0.0001 3. Short Term Debt Ratio Model Constant 0.263** 0.024 10.781 Size 0.013*** 0.001 9.344 Growth 0.000** 0.001 0.314 Tangibility -0.306*** 0.02 -15.174 Profitability -0.227*** 0.042 -5.434 Risk -0.112*** 0.086 -1.304 Adjusted R2 0.287 F-statistic 78.678 Prob. of (F-Stat.) 0.0001

This table shows estimates from Ordinary Least Square (OLS) models. The sample contains 93 Saudi Firms listed in the Saudi Stock Exchange. for 2000-2010. Model 1 defines the total debt ratio as total liabilities divided by total assets. Model 2 defines the long-term debt ratio as total liabilities minus current liabilities divided by total assets. Model 3 defines short term debt ratio as current liabilities divided by total assets. In each model the leverage ratio was regressed against five independent variables: size, growth, tangibility, profitability, and risk. The estimated model is: 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒�𝑓𝑖𝑟𝑚𝑖,𝑡� = α + β1 𝚕𝚗 𝑠𝑎𝑙𝑒𝑠𝑖,𝑡 + β2 ∆𝚕𝚗 𝑠𝑎𝑙𝑒𝑠𝑖,𝑡 + β3 𝑡𝑎𝑛𝑔𝑖𝑏𝑙𝑖𝑡𝑦 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + β4 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 +β5σ 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑖,𝑡 + ε𝑖,𝑡 . ***, **, and * indicate significance at the 1, 5 and 10 percent level respectively. REFERENCES

Altman, E.I. (1984) “A Further Empirical Investigation of the bankruptcy Cost Question,” Journal of Finance, vol. 39, p. 1067-1089. Berger, A., and G. Udell (1994) “Lines of Credits, Collateral, and relationship lending in small firm finance,” Mimeo, Board of Governance of the Federal Reserve System. Bevan, A., and J. Danbolt (2002) “Capital Structure and its Determinants in the United Kingdom A decompostional Analysis,” Applied Financial Economics, vol. 12, p.159-170. Booth, L., V. Aivzin, A. Demirgue-Kunt and V. Mskimovic (2001) “Capital Structure in Developing Countries,” Journal of Finance, vol. 56, p. 87-130. Bradley, M., G.A. Jarrell and E.H. Kim (1984) “In the Existence of an Optimal Capital Structure: Theory and Evidence,” The Journal of Finance, vol.39, p 857-878. Castanias, R. (1983) “Bankruptcy risk and optimal Capital Structure,” Journal of Finance, vol. 38, p.1617-1635.


66

Correa, C., L. Basso and W. Nakamura (2007) “What Determines the Capital Structure of the Largest Brazilian Firms? An Empirical Analysis Using Panel Data,” Working paper. Crutchley, C.E. and R.S. Hansen (1989) “A test of the Agency of Managerial ownership, corporate Leverage, and Corporate dividends,” Financial Management, vol.18, p. 36-46. Frank, M.Z., and V.K. Goyal (2003) “Testing the Pecking order theory of capital structure,” Journal of Financial Economics, vol. 67, p. 217-248. Frank, M.Z., and V.K. Goyal (2007) “Trade-off and Pecking order theories of debt,” Handbook of Corporate Finance: empirical Corporate Finance, Vol.2. In: handbook of Finance Series, Chapter 12, - (Elsevier/north-Holland, Amsterdam). Frank, M.Z., and V.K. Goyal (2009) “Capital structure Decisions: Which factors are reliably important?,” Financial Management, vol. 38, p. 1-37. Gajural, D.P. (2005) “Capital structure Management in Nepalese Enterprises,” Corporate Finance Journals, Working paper series. Gaud, P., E. Jani, M. Hoesli and A. Bender (2005) “The Capital Structure of Swiss Companies: An empirical analysis using dynamic panel Data,” European Financial Management, vol. 11, p. 51-69. Harris, M. and A. Riviv (1991) “The Theory of Capital structure,” Journal of Finance, vol. 46, p. 297-356. Jensen, M.C. and W.H. Meckling (1976) “Theory of the firm: Managerial Behavior, Agency Costs, and Ownership structure,” Journal of Financial Economics, vol.7, p. 305-360. Miller, M.H. (1977) “Debt and Taxes,” Journal of Finance, vol. 32, p. 261-276. Modigliani, F., and M. Millar (1958) “The cost of Capital, Corporate Finance and the theory of Investment,” American Economic Review, vol. 48, p. 261-297. Modigliani, F., and M. Millar (1963) “Corporate Income taxes and the Cost of Capital: A Correction,” American Economic Review, vol.53, p. 433-443. Myers, S.C. (1977) “Determinants of Corporate Borrowing,” Journal of Finance, vol. 5, p. 147-175. Myers, S.C. (1984) “The Capital Structure Puzzle,” Journal of Finance, vol. 39, p. 575-592. Ross, S.A. (1977) “The Determination of Financial Structure: The incentive signaling Approach,” The Bell Journal of Economics, vol. 8, p.23-40. Nishat, M., and W. Allah (2008) “Capital structure choice in an Emerging Market: Evidence from Listed firms In Pakistan,” 21st Australian finance and Banking Conference 2008 paper. Rajan, R.G., and L. Zingales (1995) “What Do We knows About Capital Structure? Some Evidence From international Data,” Journal of Finance, Vol. 50, p. 1421-1460. Saudi Arabian Monetary Agency (2011) “47th Annual Report,”


67

Shah, A., and T. Hijazi (2004) “The Determinants of Capital Structure of Stock Exchange-Listed Non-Financial Firms in Pakistan,” Pakistan Development Review, vol. 43, p. 605-618. Titman, S., and R. Wessels (1988) “The Determinants of Capital Structure Choice,” Journal of Finance, vol. 43, p. 1-19. Thies, C.F., and M. Klock (1993) “Determinants of Capital Structure,” Review of Financial Economics, vol. 3, p. 40-52.

BIOGARPHY Dr. Turki Alzomaia is an assistant professor of finance at College of Business Administration, King Saudi University, Riyadh, Saudi Arabia. He received a PhD in finance and investment from the George Washington University. His research interests lie in the area of corporate finance. His email address is: [email protected]


69

COST EFFICIENCY OF GHANA’S BANKING INDUSTRY: A PANEL DATA ANALYSIS

Kofi Adjei-Frimpong, Lincoln University Christopher Gan, Lincoln University

Baiding Hu, Lincoln University

ABSTRACT

This study analyzes the efficiency of the banking industry in Ghana over the period of 2001–2010 using the data envelopment analysis. The study investigates the impact of size, capitalization, loan loss provision, inflation rate and GDP growth rate on Ghana’s bank efficiency using both static and dynamic panel data models. The static model is estimated by the fixed effects estimator whereas the dynamic mdoel is estimated by the two step system GMM estimator. The results suggest that Ghana banks are inefficient. This study reveals that well-capitalized banks in Ghana are less cost efficient. In addition, bank size has no influence on bank cost efficiency suggesting that larger banks in Ghana have no cost advantages over their smaller counterparts. The findings also exhibit that loan loss provision ratio has no effect on bank efficiency in Ghana. This study finds GDP growth rate negatively influences bank cost efficiency and that lagged cost efficiency tends to persist from year to year. JEL: E44, E50, E60 KEYWORDS: Data Envelopment Analysis, Bank Efficiency, Cost Efficiency INTRODUCTION

he banking industry in Ghana has changed considerably since 1988 as a result of the gradual and steady implementation of financial services deregulation, globalisation and the emergence of communication and information technologies. The financial deregulation was undertaken as part of

the structural economic adjustment and stabilization program launched in 1983 with the assistance of the International Monetary Fund and World Bank. These financial sector reforms are aimed at increasing banks competitiveness, efficiency and performance in Ghana’s banking system that could then contribute in greater measure to stimulate economic growth and ensure financial stability. During the pre-reform era, Ghana banking system was dominated by the state owned banks and totally controlled by the government. Ghana’s economic performance declined and its banking system was in distress. Banks were characterised by inadequate capital, insufficient loans loss provisions, high operating costs due to inefficient operations, a large portfolio of nonperforming loans and endured enormous political influence (International Monetary Fund, 1999; World Bank, 1989). The financial system was distorted by interest rate controls and selective credit policies, lack of competition, and weak supervision by the Bank of Ghana (World Bank, 1989). As a result, financial reforms were undertaken and most restrictions on foreign entry, interest rates and exchange rates were removed. The results have increased the capacity of financial institutions to mobilise domestic savings, enhanced efficiency among banks, and strengthened economic growth. The central bank set up the payments system infrastructure and appropriate measures that facilitate a competitive and efficient banking sector. The Ghana banking sector has shown considerable improvements in communication and computing information technology, as banks modernized their distribution networks and introduced new banking services such as Automated Teller Machines (ATMs), telephone banking, mobile banking and internet banking are now prevalent in Ghana. The Ghana banking sector is reasonably efficient, financially innovative, competitive, profitable, and growing quite quickly (Acquah, 2009). The

T

K. Adjei-Frimpong et al | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

70

sector has seen some structural changes with reduced concentration and strong competition for market shares, increase in branch network and provision of various new banking products in Ghana. For example, the number of banks actively operating in Ghana has grown from 7 in 1987 to 27 in 2010. Most of the new entrants were foreign banks. During the same period, the number of foreign banks in Ghana increased from 3 to 15. The bank concentration based on the Herfindahl-Hirschman index (HHI) has dropped considerably from 1,065.9 points in 2000 to 600.0 points in 2010 (Bank of Ghana, February 2009) representing a decrease in market concentration of 30.2 percent, as a result of the increase of the number of banks. During the period 2001 to 2010 the real gross domestic product has grown between 4.5 percent and 8.4 percent (International Monetary Fund, 2011). Ghana’s financial sector reforms policies have long been pursued with great enthusiasm and consistency than in some other African countries. Despite the considerable progress for the past 12 years as a result of the financial reforms, no study has been conducted to evaluate the level and determinants of bank efficiency in Ghana. This paper attempts to fill this gap in literature by providing empirical evidence on efficiency in the Ghana’s banking industry. In addition, better understanding of the factors affecting Ghana banks’ efficiency is vital to both bank regulators and policy makers because improvements in efficiency in the banking industry are essential prerequisite for providing a more efficient system of asset allocation in the financial system which then facilitates lower cost of capital to firms and accelerates capital accumulation and productivity growth (McKinnon, 1973)..The aim of this study is to determine whether deregulation has improved the level of bank cost efficiency of Ghana’s banking sector and examine the determinants of bank cost efficiency using both static and dynamic models. The rest of the paper is organized as follows. Section 2 discusses the relevant literature on bank efficiency. Section 3 provides the methodology and data employed. Section 4 presents the empirical results; and Section 5 concludes the paper. LITERATURE REVIEW Many studies have used various methods to estimate bank efficiency as well as different econometric approaches to determine the factors that affect bank efficiency. Many of the previous studies on bank efficiency have been conducted on developed economies (Pasiouras, 2008 and Delis et al., 2009 on Greek banks; Mukherjee et al., 2001 on US banks; and Girardone et al., 2004 on Italian banks). However, the recent resurgence of economic and financial reforms across the developing countries has also raised the awareness of the importance of bank efficiency (Tecles & Tabak, 2010 on Brazilian banks; Ariff & Can, 2008 on Chinese banks; Altunbas et al., 2007 on banks from 15 European countries and; Ataullah & Le, 2006 on India and Pakistan banks). Previous studies revealed mixed results regarding the relationship between financial reforms and efficiency. Casu & Molyneux (2003) use a sample of 530 banks from five European Union countries covering the period 1993 to 1997 to investigate the existence of productive efficiency across the European banking markets since the introduction of the Single Internal Market. Their results show an evidence of a small improvement in bank efficiency levels. Similarly, Ataullah & Le (2006) analyze the efficiency of the Indian banking sector during 1992–1998 using the data envelopment analysis (DEA) method and find evidence of efficiency gain in the Indian banking industry during the post-economic reforms era. A recent study by Loukoianova (2008) who uses DEA to investigate the cost and revenue efficiency of Japanese banks from the period 2000-2006 finds enhancement in efficiency for the period between 2001 and 2006. Staub et al. (2010) estimate cost, technical and allocative efficiencies for Brazilian banks for the period 2000–2007 and conclude that banks in Brazil are inefficient. In assessing the determinants of bank efficiency, the relationship between efficiency, on one hand, and bank size, bank capitalization, loan loss provisions ratio and GDP growth, on the other hand, is


71

ambiguous. The results of previous studies on the relationship between bank size and bank efficiency are inconsistent. Some previous studies have found that larger banks are more efficient (e.g. Miller & Noulas, 1996; Ataullah & Le, 2006; Tecles & Tabak, 2010). In contrast, Isik & Hassan (2002), Girardone et al. (2004) and Altunbas et al. (2007) studies have documented a significantly negative effect of bank size on bank efficiency. Other studies have observed insignificant influence of bank size on bank efficiency (e.g. Berger & Mester, 1997, Ariff & Can, 2008; Staub et al., 2010). Some previous studies such as Casu & Girardone (2004) , Ataullah et al.(2004), Staikouras et al. (2008) and Yildirim & Philippatos (2007) reported a negative impact of loan loss provisions ratio on bank efficiency. However, Altunbas et al. (2007) find a positive relationship between loan loss provision and bank efficiency while Staub et al. (2010) observed an insignificant relationship. The relationship between bank capitalization and bank efficiency clearly show mixed results. For example, some studies have reported a positive relationship between bank capitalization and bank efficiency (see Casu & Girardone, 2004; Pasiouras, 2008, Yildirim & Philipatos, 2007; Staikouras et al., 2008). On the other hand, Kwan & Eisenbeis (1997), Altunbas et al. (2004), Altunbas et al. (2007) and Kablan (2010) studies reveal a negative relationship. A negative relationship can be attributed to the fact that financial capital influences costs through its use as a source of financing loans (Berger & Mester, 1997; Ariff & Can, 2008; Staikouras et al., 2008). Thus, raising capital that involves higher costs than taking deposits, for example issuing shares, could generate a negative relationship between bank capitalization and bank efficiency. Others studies such as Ariff & Can (2008), Casu & Molyneux (2003) and Staub et al. (2010) find no significant impact of bank capitalization on bank efficiency. In regards to the macroeconomic factors on bank efficiency, Maudos et al. (2002) study 10 European Union countries for the period 1993–1996 and report that GDP growth rate has a positive correlation with profit efficiency but a negative correlation with cost efficiency. Yildirim & Philippatos (2007) investigate cost and profit efficiency of 12 transition economies of Central and Eastern Europe (CEE) banks from 1993 to 2000. The authors investigate the determinants of bank efficiency employing the generalized least squares fixed-effects estimators and find that economic growth has a positive relationship with bank cost efficiency but a negative relationship with profit efficiency. Previous empirical studies on bank efficiency have mostly employed static panel data methods to analyze the determinants of bank efficiency. However, many financial processes exhibit dynamic adjustment over time so failing to incorporate dynamic aspect of the data can lead to serious misspecification biases in the estimation and results. De Jonghe & Vennet (2008) report that most banking studies failed to consider the time it takes for the impacts of bank efficiency to materialize. However, there is gradual awareness of the need to include lagged efficiency such as in the studies of Ataullah & Le (2006), Staub et al. (2010) and Fiordelisi et al. (2011). METHODOLOGY Data Envelopment Analysis as Measure of Bank Efficiency Due to the small number of banks in Ghana, this paper employs the data envelopment analysis to determine efficiency scores of Ghana’s banks. This is because DEA works well with small sample size as opposed to parametric methods which require large sample size to generate reliable estimate (Isik & Hassan, 2002; Ariff & Can, 2008). DEA also does not specify any functional form of the underlying production relationship (Berger & Humphrey, 1997). Further, DEA is used extensively in studying the banking industry of developed and developing economies; for individual countries as well as cross-country comparisons (Aly et al., 1990; Chen & Ye, 1998; Sathye, 2001; Casu & Girardone, 2006. Following Aly et al. (1990), Sathye, (2001), Casu & Girardone (2006) and Tecles & Tabak (2010), this study uses variable return to scale (VRS) model (Banker et al., 1984) as constant returns-to-scale


72

(Charnes et al., 1978) assumption is unlikely to prevail since Ghana banks operate in an imperfect competitive environment and are also subject to financial constraints and regulatory requirements (see Coelli et al., 1998). These factors might compel or cause the banks not to operate at optimal scale. Following Elyasiani & Mehdian (1990), Drake (2001), Goddard et al. (2001) and Berger (2007) studies, this study assumes that bank management has more control over costs rather than over outputs (and with high bank operating costs in Ghana) and adopts an input-orientation approach. Estimating Bank Cost Efficiency Cost efficiency measures how close a bank’s cost is to the minimal cost (or best practice bank’s cost) for producing a certain level of output with given input prices and technology. Consider N banks that employ a vector of input quantities xi for the i-th bank, given the prices of input wi and the levels of output yi, the cost efficiency model for bank i can be expressed in a linear programming as follows: Minimizeλ,xi

∗ wi′xi∗ (1)

Subject to −yi + Yλ ≥ 0 xi∗ − Xλ ≥ 0 NI′λ = I 𝜆 ≥ 0 i = 1, …N where xi∗ is the frontier or cost-minimizing vector of input quantities for the i-th bank and λ is a Nx1 vector of constants. To estimate cost efficiency the optimal values xi∗ are estimated by solving the linear programming (equation 1), where X and Y are the matrix of observed inputs and outputs for all the banks. The cost efficiency of the i-th bank is calculated as the ratio of minimum cost to actual cost:

CE =wi′xi∗

wi′xi

(2)

The measure of cost efficiency is bounded between zero and one. A cost efficiency score of one represents a fully cost efficient bank and are also known as best practice banks in the sample, whereas inefficient cost banks exhibit a value less than one. However, those inefficient cost banks with a value of zero are considered worst practice banks. Inputs and Outputs for the DEA In order to estimate cost efficiency, inputs, input prices and outputs must be calculated. Table 1 shows the description of the variables used in the computation of bank efficiency. The choice of the inputs and outputs is essential for measuring the relative efficiencies in banks. The two most widely used approaches in the banking literature for the selection of bank inputs and outputs are the production and intermediation approaches. This study employs a variation of the intermediation approach originally developed by Sealy & Lindley (1977) which views banks as financial intermediaries, producing intermediation services through the collection of deposits and other liabilities and use them to generate interest-earning assets such as loans, securities and other investments. This study identifies two outputs, namely total loans and other earning assets and three inputs, that is, labour (proxy by personnel expenses), capital-related expenses and deposits. Deposits are the most important input resources for Ghana banks to perform their banking activities such as lending and investing. The choice of labour (personnel expenses) and capital expenses are other input resources used in the production of bank products and services. In the case of output, loans and investments securities (especially government securities) constitute the major activities


73

Table 1 Variables used in the Computation of Bank Efficiency

Variable Description Inputs: Deposits Customers deposits Labour Personnel expenses of bank staff such as salaries, wages and benefits Outputs: Loans Total customers’ loans Other earning assets Banks’ investments in different types of securities (e.g. government securities,

bonds, Treasury bill and equity investment) Input prices: Price of deposits Interest expenses divided by total deposits Price of labour Personnel expenses divided by the total assets Price of capital Capital-related expenses (operating expenses - personnel expenses) divided by

total fixed assets.

(especially government securities) constitute the major activities of the banks that channel their funds into investment or lending for profits. In Ghana, loans and other earning assets account for about two thirds of the bank assets and are important generator of revenues. The inputs prices are estimated as proxies since data on the number of personnel and input prices are not available. The production approach is not considered because it is difficult to obtain detailed bank information relating to transactions and financial documents which are required in the approach. Determinants of Bank Efficiency Empirical Model This study investigates the underlying relationship between the estimated efficiency levels and a variety of bank-specific and macroeconomic factors. In the second stage, both the static and dynamic panel data models are estimated with the DEA cost efficiency scores as the dependent variable and bank-specific and macroeconomic factors as the explanatory variables. Many banking studies have examined the factors that affect the efficiency of banks. In the banking literature some studies investigate only bank-specific factors while others assess both bank-specific and external factors. The widely used bank-specific factors are size, profitability, capitalization, loans to assets, loan loss provision to total loans (see Casu & Molyneux, 2003; Casu & Girardone, 2004; Ataullah & Le, 2006; Ariff & Can, 2008). The inflation and real GDP growth rates are commonly used to control for the macroeconomic conditions (see Salas & Saurina, 2003; Girardone et al., 2004; Yildirim & Philippatos, 2007). In this study, bank size, bank capitalization, loan loss provision to total loans, inflation and real GDP growth rates considered as the factors influencing bank cost efficiency in Ghana. The static panel data model used to determine the bank-specific and macroeconomic factors that affect bank cost efficiency in Ghana is given as follows: EFFit = α1CAPit + α2SIZEit + α3LLPit + α4INFit + α5GDPit + +η𝑖 + µit (3)

where i represents the individual bank and t denotes time, α are the parameters to be estimated, ηi is the individual bank specific-effect, EFFit is cost efficiency scores, CAPit is bank capitalization, SIZEit is bank size, LLPit is loan loss provision ratio representing credit risk, GDPit is real gross domestic product growth rate, INFit is inflation rate and µit is the random error term. A dynamic panel data model is specified by including one-year lagged efficiency among the explanatory variables to capture the dynamic nature of the efficiency of banks. This study attempts to test whether bank efficiency tends to persist over time in the Ghanaian banking context. According to Staub et al. (2010), banks that are more efficient in a specific year tend to be efficient in the following year. On the


74

other hand, Ataullah & Le (2006) suggest that the one-year lagged efficiency indicates accumulation of knowledge and technological endowment that may assist banks to produce higher outputs with their inputs or reduce cost by adjusting comparatively quickly to the financial reforms. Ataullah & Le (2006) and Staub et al. (2010) studies find significant and positive relationship between the efficiency of the previous year and that of the current year. Furthermore, early banking studies have confirmed the persistence of efficiency over time (Berger & Humphrey, 1991; Kwan & Eisenbeis, 1997). Following the procedure of Ataullah & Le (2006), Solis & Maudos, (2008) and Staub et al. (2010) the dynamic panel model specification for the determinants of bank cost efficiency in Ghana is given as follows: EFFit = β1EFFit-1+ β2CAPit + β3SIZEit + β4LLPit + β5INFit + β6GDPit + ηi + ϵit (4)

where i represents the individual bank and t denotes time, β are parameters to be estimated, ηi is the individual bank specific-effect, EFFit is cost efficiency scores, EFFi,t-1 is one-year lagged cost efficiency, CAPit is bank capitalization, SIZEit is bank size, LLPit is loan loss provision ratio representing credit risk, GDPit is real gross domestic product growth rate, INFit is inflation rate and ϵit is the random error term. The logit method has been used in recent studies on bank efficiency (see for example, Ataullah & Le, 2006; Maudos & Fernandez de Guevara, 2007; Solís & Maudos, 2008). Since the estimated values of DEA efficiency (EFFR) range between 0 to 1, logistic specification is used to transform the efficiency scores into natural log odds ratio as follows: Ln � EFFR

1−EFFR�. (5)

However, the transformed efficiency score is undefined when the efficiency score, EFFR is zero or one. This problem reduces the total observations by the number of undefined efficiency scores, causing some loss of the data. Consequently, as in Cox (1970 p.33), Voos & Mishel (1986), Campbell et al. (2007) and Kader et al. (2010), the logit transformation is modified by adding 1/2N to both numerator and denominator, where N represents the number of observations for the efficiency. The advantage of this modified logit transformation is that there is no reduction or elimination of the observations when the efficiency score is equal to zero or one (Maddala, 1983 p.30). The transformed efficiency score, EFF, is employed as the dependent variable for the evaluation of the determinants of efficiency. DEA-Solver Pro is used to estimate the efficiency scores. Model Variable Definitions The variable definitions are presented in Table 2. Table 2: Definition of Model Variables Variable Symbol Description Expected

Signs Cost efficiency EFF Estimated using data envelopment analysis Size Size Natural logarithm of total assets (+/-) Credit risk LLP Loan loss provisions over total loans (-) Capitalization CAP Total value of shareholders equity over total assets (+/-) Macroeconomic Factors: Inflation rate INF Change in consumer price index (+) GDP growth rate GDPG GDP growth rate between two consecutive years (+)


75

Estimation Techniques The study employs the fixed effect model to estimate the coefficients in the static equation (3). In terms of the static model, the regression equation for the determinants of bank efficiency assumes exogeneity of the explanatory variables and account for unobservable heterogeneity. The fixed effect model is estimated using robust standard errors (White/Huber (1980) test) to control for potential heteroscedasticity. Under these assumptions, the fixed effect estimator generates efficient parameter estimates and it is considered better than the GMM estimator. However, with a lagged dependent variable and endogenous explanatory variables in the dynamic panel estimation, the GMM estimator is more superior to fixed effect estimator which generates inconsistent estimates (Baltagi, 1995). Following Arellano and Bover (1995) and Blundell and Bond (1998) we developed the system GMM estimator that was designed to overcome potential bias and imprecision associated with first difference GMM estimator when the explanatory variables are persistent (or the sample size is small, as in this study) to estimate the coefficients in equation (4). The first-difference GMM estimator may suffer from the weaknesses of its instruments as the lagged levels of persistent explanatory variables are weak instruments for the equation in first-difference (Blundell & Bond, 1998; Bond, 2002). Particularly, in this study, a two-step system GMM estimator with Windmeijer (2005) corrected standard error is used because it is more efficient and robust to autocorrelation and heteroscedasticity and provides the least bias in small samples. In addition, forward orthogonal deviation is used in place of first-difference as recommended by Roodman (2006 pp. 20, 2009) because first-difference enlarges gaps in unbalanced panel data as it uses only lags variables as observations (Roodman, 2006 pp.19) that can produce biased results especially in small sample. This approach preserves sample size in panels with gaps (Roodman, 2009) . Forward orthogonal deviations approach subtracts the mean of all future available observations of a variable instead of subtracting the past value of observations of a variable. Based on previous banking studies loan loss provision and bank capitalization are assumed to be endogenous to efficiency is instrumented with their own lags. In this study, the second and third lags of loan loss provision and bank capitalization are used as instruments for the system GMM estimates as well as collapsing instruments (Roodman, 2006, 2009) . The use of these techniques allows us to considerably reduce the number of instrument counts in order to avoid over-fitting of the endogenous variables to have more reliable estimations. In terms of the static equation (3), the F-test is used to test the null hypothesis that the overall significance of the coefficients of the explanatory variables is jointly equal to zero. This must be rejected to ensure the model is correctly specified. On the other hand, the following tests must be satisfied under the system GMM estimation. First, the Hansen (and difference-in-Hansen) test should not be rejected suggesting that instruments in the system GMM estimation are valid. Second, it is imperative that the second order autocorrelation test under the null hypothesis of no second order autocorrelation is not rejected. This leads to the conclusion that the original error term is serially uncorrelated. The regressions are estimated by employing the Hansen and second order autocorrelation tests to select an appropriate set of instruments for estimation. Data The study covers Ghana banks during the period 2001 to 2010. The data used in this study depend on the amount of information available for each bank involved. The data exclude banks which have less than three years of operation during the study period. There were very few mergers and acquisitions and exit during the study period. The data were analyzed for inconsistencies, reporting errors, and outliers. In addition, the years with zero or missing values on input and output variables are omitted. With these restrictions, the sample data for this study is an unbalanced panel data of 25 banks with 211 annual observations, which accounts for more than 99% of bank assets in the time period under consideration.


76

The choice of an unbalanced panel is due mostly to entry during the study period. The number of banks in each year varied between 14 and 25. The data are based on balance sheets and income statements of the banks’ annual reports. The data are obtained from PricewaterhouseCoopers. The macroeconomic variables are obtained from International Monetary Fund's World Economic Outlook. The 25 banks consist of 4 state-owned banks, 8 domestic private banks, and 13 foreign-owned banks. A bank is identified as foreign-owned in Ghana if the foreign ownership share in its assets exceeds 50%. EMPIRICAL RESULTS Descriptive Statistics and Correlation Analysis Table 3 shows large variation across banks shown by the minimum and maximum values of the factors during the study period 2001 to 2010. The rate of inflation depicts a minimum figure of 10.2 percent and a maximum of 32.9 percent with an average of 16.4 percent from 2001 to 2010. The loan loss provision ratio exhibits a worrying trend. On average, 8.8 percent of the total loans in Ghana’s banking industry exhibits a minimum of zero percent and a maximum of 64 percent. Table 3: Summary Statistics of the Determinant Factors

Variable Observations Mean Standard Deviation Minimum Maximum size 211 11,935 1,388 7,910 14,560 inf 211 0.164 0.067 0.102 0.329 llp 211 0.088 0.084 0.000 0.640 gdpg 211 0.057 0.012 0.045 0.084 cap 211 0.136 0.113 -0.150 0.980

This table presents the descriptive statistics including the sample size, mean, standard deviation, minimum and maximum values for the 25 banks used in this study. inf, llp, gdpg and cap are in ratios size is in million cedis. The range is overwhelmingly substantial during the study period. Even though the loan loss provision ratio has been decreasing steadily, it is still considered relatively high. However, Ghana banks are well-capitalized. The average bank in the sample has a capital ratio of 13.6 percent. There are also noticeable differences in bank size during the study period. The average GDP growth is 5.7 percent during the study period. The Ghanaian economy has enjoyed a sustained economic growth from 2001 to 2010. However, the inflation rate continues to be high despite the economic and financial reforms. Table 4 presents summary statistic of the bank specific factors exhibiting yearly values of mean and standard from 2001 to 2010. The dispersion of bank specific factors (measured by standard deviation) is high, indicating that the factors are dispersed around the average. This suggests that Ghana’s banks are heterogeneous. The introduction of universal banks policy in 2003 in Ghana could reduce the heterogeneity across banks. In order to avoid multicollinearity problems in the determinant factors of bank efficiency, pairwise correlations of the explanatory variables used in the regressions are examined. Table 5 reports the results of the correlation matrix of the factors. The result shows low correlation among the variables and allays the fear of multicollinearity problems. This suggests that there is no significant correlation between the explanatory variables.


77

Table 4: Summary Statistic of Bank Specific Factors (2001-2010)

Size llp Cap Year Mean Standard

Deviation Mean Standard

Deviation Mean Standard

Deviation 2001 10.486 1.474 0.1 0.094 0.134 0.073 2002 10.859 1.29 0.122 0.111 0.104 0.075 2003 11.268 1.179 0.116 0.088 0.097 0.074 2004 11.563 1.085 0.097 0.063 0.119 0.043 2005 11.502 1.248 0.086 0.063 0.192 0.232 2006 11.835 1.124 0.067 0.046 0.14 0.076 2007 12.287 0.995 0.066 0.067 0.103 0.045 2008 12.479 1.218 0.059 0.058 0.143 0.153 2009 12.797 1.117 0.082 0.071 0.152 0.11 2010 13.152 0.858 0.106 0.13 0.158 0.091

This table presents the descriptive statistics showing the mean and standard deviation of the bank specific factors for the period under study. llp and cap are in ratios size is in million cedis. Table 5: Correlation Coefficients of Determinants of Bank Efficiency

Variable size inf llp gdpg cap size 1.0000 inf -0.3089 1.0000 llp -0.0149 0.0821 1.0000 gdpg 0.2117 -0.2290 -0.1298 1.0000 cap -0.2362 -0.0347 -0.1229 0.0336 1.0000

This table presents correlation coefficients of determinants of bank efficiency. inf, llp, gdpg and cap are in ratios and size is in million cedis. Bank Efficiency in Ghana Average Bank Efficiency Scores by Year Table 6 presents the results of the yearly and overall efficiency of Ghana’s banking system over the period 2001 to 2010. The results show that the overall average cost efficiency score for Ghana’s banking industry is 0.505. This implies that Ghana’s banks wasted 49.5 percent (half) of its costs relative to the “best-practice” banks. In other words, on average, the industry could reduce their cost by 49.5 percent and still produce the same amount of output. The results suggest Ghanaian bank managers did not use their inputs efficiently over the study period. Overall, the results show relatively low average efficiency scores during the study period, which suggests that Ghana banks are operating far from the efficiency frontier. On the contrary, Fang et al. (2011) in their study reported a relatively higher efficiency score of 76.95 percent for the Croatian banking sector over the period 1998 to 2008. Similarly, Ariff & Can (2008) and Maudos & Pastor (2003) studies reported an average cost efficiency score of 79 percent for the Chinese banking industry during the period 1995-2004 and 87.1 percent for the Spanish banking sector during 1985-1996. However, high levels of inefficiency in some emerging countries such as India, Turkey and Brazil have also been reported (Das & Ghosh, 2006; Denizer et al., 2007; Tescles & Tabak, 2010). In terms of yearly results, the cost efficiency of Ghana’s banking industry improved considerably from 0.452 in 2001 to 0.661 in 2010, an increase of 46.2 percent. In early years, from 2002 to 2005, cost efficiency increases from 0.416 in 2002 to 0.486 in 2005, showing improvement in input utilization, but then declines to 0.469 in 2006 and eventually starts to show a steady improvement in input utilization from 2007 to 2010. The trend in cost efficiency from 2007 to 2010 suggests that banks managers in Ghana have begun to use their inputs more efficiently that is, the managers are able to control the underutilization or wastage of valuable input resources. Nevertheless, more effort is still required. The high interest rates in the Ghana confirm the high financial costs of the capital, and high non-performing loan problems which result in low cost efficiency of the banks. Casu & Girardone (2009) in their study of five European countries banking sector report an increase in input waste from 2000-2001 onwards leading


78

to lower average bank efficiencies. They attribute the input waste to reduction in costs facilitated by bank deregulation and increased competition leading to many mergers and acquisitions that may have increased bank costs leading to a decline in their cost efficiency. The authors further explain that decreases in bank efficiency can be the cause of bank consolidation which allows managers to exploit market power. Table 6: Average Efficiency Scores of Ghana’s Banking Industry (2001-2010)

Year Number Ce of Banks Mean Standard Deviation 2001 17 0.452 0.263 2002 18 0.416 0.253 2003 18 0.451 0.201 2004 18 0.484 0.188 2005 20 0.486 0.174 2006 22 0.469 0.201 2007 23 0.453 0.196 2008 25 0.526 0.206 2009 25 0.577 0.250 2010 25 0.661 0.276 Mean 0.505 0.231

This table provides the average efficiency scores. The table shows the number of banks, mean and standard deviation scores, Ce represents cost efficiency. Our results show bank cost efficiency is relatively unstable over the study period. The results also show the low level of the efficiency scores in Ghana’s banks. However, since 2007 there has been a remarkable improvement in the efficiency scores in Ghana’s banks. For instance, the average cost efficiency score increased from 0.577 in 2009 to 0.661 in 2010 representing a yearly increase of 14.6 percent, also the biggest during the study period. Composition of Efficient Frontier Banks Table 7 describes the composition of the Ghana’s bank efficiency frontier, which is the input and output combination of the ‘best-practice’ banks in Ghana. The data in Table 7 shows a total of 62 of the 211 bank observations are regarded cost efficient over the study period. Based on individual years, only 7 out of 25 banks are on the cost efficiency frontier in 2010. Table 7: Number of Efficient Frontier Banks (2001-2010)

Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Total All Banks 3 5 5 6 7 7 9 7 6 7 62 Number of banks

17

18

18

18

20

22

23

25

25

25

This table shows the number of efficient frontier banks for the period 2001-2010. All banks indicate the banks under study. Ce represents cost efficiency With the exception of 2001 and 2007, the rest of the study period indicates a fairly distributed cost efficiency frontiers. The results indicate that 36 of the 62 efficient observations are recorded from 2006 to 2010 representing 58 percent. This shows the weakness of Ghana’s banks in regards to cost efficiency. The low bank cost efficiency apparently reflects the high operating and financial costs of managing a bank in Ghana. Even though the financial reforms have improved the bank efficiency in Ghana in comparison to pre-reforms period, there is more room for improvement, especially in terms of bank cost efficiency.


79

Determinants of Bank Cost Efficiency Table 8 presents the result of the determinants of cost efficiency in Ghana banks. The F-test is statistically significant at 1 percent level for all the explanatory variables. This indicates that the factors used are relevant in explaining the cost efficiency. The results indicate that bank size and bank capitalization are the most important factors in determining bank cost efficiency in Ghana. The analysis of the residuals indicates the presence of heteroskedasticity and as a result White/Huber robust standard error is applied. In terms of the system GMM, the p-value of Arellano-Bond test statistics AR(1) is 0.042 which shows that AR(1) test rejects the null hypothesis of no existence of first-order serial autocorrelation. However, the Arellano-Bond test statistics for the second order serial correlation AR (2) in the residuals do not reject the specification of the error term, since the p-value of AR (2) is 0.948. Thus, there is no second order serial correlation in the error term. The p-value of the Hansen test is 0.881. Accordingly, the Hansen test of over-identification reports that the instruments used in the system GMM estimation are valid. The difference-in-Hansen test of exogeneity indicates that the instruments used for the equation in levels are exogenous which strengthens the validity of instruments employed in the system GMM estimation. There is no evidence of correlation between the instruments and error terms. Hence, the dynamic cost efficiency equation is correctly specified. In addition, since the loan loss provision and bank capitalization are endogenous variables, the results in this study are based on the two-step system GMM instead of the static fixed effect estimator. Impact of Bank Specific Factors on Bank Cost Efficiency The system GMM results in Table 8 show that lagged cost efficiency, GDP growth rate and bank capitalization are the important factors in determining bank cost efficiency in Ghana. The lagged cost efficiency is significant and has a positive effect on the bank efficiency in the current year. This implies that bank cost efficiency tends to persist from year to year. This suggests that an increase in lagged cost efficiency could help increase the current year’s cost efficiency. The positive lagged cost efficiency may constitute some accumulated knowledge and technologies that may help banks to reduce their costs (see Ataullah & Le, 2006). This implies that the financial services in Ghana’s banking industry have encouraged banks to improve their cost efficiency. The result is consistent with the study of Staub et al. (2010) and Manlagnit (2011) which reveal lagged cost efficiency to have positive and significant effect on the current year efficiency. Table 8 shows that bank size is positive but has no significant impact on cost efficiency. This result implies that larger banks in Ghana have no cost advantages over their smaller counterparts. Similarly, some previous studies did not observe any significant efficiency advantage for large banks. For instance, Girardone et al. (2004) study on the Italian banking sector indicates no evidence of correlation between size and bank efficiency suggesting that larger banks are not more cost efficient than the smaller banks. Similarly, Staub et al. (2010) study on the Brazilian banking system in the period 2000 to 2007 find that bank size is not an important factor in determining bank cost efficiency. The bank capitalization coefficient is negative and statistically significant at 10 percent level. Ghana banks have been recapitalized by the Bank of Ghana, first in 2003 and then in 2009. This result suggests that well-capitalized banks are less cost efficient in Ghana. This could be due to a higher shareholders' leverage which forces banks to sacrifice costs in exchange for achieving better results. This finding is similar to the results reported by Tabak et al. (2011) on 495 Latin American banks operating in 17 countries over the period 2001-2008, Sufian (2009) on Malaysian banks from 1995 to 1999 and Ariff & Can (2008) on 28 Chinese commercial banks from 1995 to 2004. Based on the results, bank cost efficiency decreases with increases in the level of bank capitalization. This suggests that well-capitalized banks incur higher costs in providing banking products and services due to high level of non-performing


80

loans and higher cost of capital resulting from the increase in minimum regulatory capital requirement (PricewaterhouseCoopers, 2011). Financial capital affects costs through its use as a source of financing loans (Berger & Mester, 1997; Ariff & Can, 2008, Manlagnit, 2011). However, raising equity capital involves higher costs than raising deposits leading to increase in financial costs and may lead to decrease cost efficiency. In addition, bank capitalization may likely increase moral hazard incentives and is more likely to increase costs (Ariff & Can, 2008; Fiordelisi et al, 2011). This may reduce cost efficiency. Thus, bank capitalization, on the one hand may reduce bank capital risk, but on the other hand, may increase moral hazard incentives leading to increase in costs and therefore decline in cost efficiency (Ariff & Can, 2008).. The culture of risk management is not well-developed in Ghana’s banking industry (Amissah-Arthur, 2010). Intuitively, the level of bank capitalization may not be adequate to cover increases in bank risk taking that may contribute to bank insolvency which could lead to reduction in bank efficiency (Soedarmono et al., 2011). Table 8: Determinants of Bank Cost Efficiency

Variable Ce

Fixed Effect Model Estimates

System GMM Estimates

Ce t-1 - 0.269* (1.85) size 0.511** 0.033 (2.42) (0.07) inf 1.722 0.794 (1.33) (0.38) llp 0.916 -5.445 (0.34) (-1.03) gdpg -8.665 -42.910*

(-1.30) (-1.97) cap 3.501** -28.743* (2.18) (-1.81) Trend 0.356 (1.36) Constant -6.138 3.815 (-2.49)** (0.59) R-squared 0.109 F-Statistic (p-value) 0.000 0.004 Wald Test Heteroscedasticity (p-value) 0.000 Number of observations 211 186 Number of banks 25 25 Number of instruments 14 Hansen J test (p-value) 0.881 Arellano-Bond test: AR(1) p-value 0.042 AR(2) p-value 0.948 Difference-in-Hansen test (p-values): GMM instruments for levels 0.784

This table presents the regression estimates of the static equation: 𝐸𝐹𝐹𝑖𝑡 = 𝛼1𝐶𝐴𝑃𝑖𝑡 + 𝛼2𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛼3𝐿𝐿𝑃𝑖𝑡 + 𝛼4𝐼𝑁𝐹𝑖𝑡 + 𝛼5𝐺𝐷𝑃𝑖𝑡 + 𝜂𝑖 +𝜇𝑖𝑡 using fixed effect estimator and the dynamic equation: 𝐸𝐹𝐹𝑖𝑡 = 𝛽1𝐸𝐹𝐹𝑖𝑡−1+ 𝛽2𝐶𝐴𝑃𝑖𝑡 + 𝛽3𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽4𝐿𝐿𝑃𝑖𝑡 + 𝛽5𝐼𝑁𝐹𝑖𝑡 + 𝛽6𝐺𝐷𝑃𝑖𝑡 +𝜂𝑖 + 𝜖𝑖𝑡 applying two-step system GMM estimator. t-statistics are in parentheses below the estimates. *, ** and *** indicate level of significance at 10%, 5% and 1% respectively. The first column shows the variables entered into the equations. Ce represents cost efficiency. The loan loss provision coefficient has a negative effect but does not appear to have a significant influence on bank cost efficiency in Ghana during the study period. This result supports the finding of Yildirim & Philippatos (2007) and Brissimis et al. (2008) who find loan loss provision to be negatively related to bank cost efficiency. In addition, Staikouras et al. (2008) assess the cost efficiency of banks operating in six emerging South Eastern European countries and finds a negative relationship.


81

Furthermore, Staub et al. (2010) study on the Brazilian banking system in the period 2000 to 2007 show that loan loss provision ratio has a negative and insignificant impact on cost efficiency. Impact of Macroeconomic Factors on Cost Efficiency The GDP growth rate has a negative and significant effect on bank cost efficiency. This shows that economic growth reduces the banks’ cost efficiency. This finding is consistent with the studies of Fries & Taci (2005) and Chan & Karim (2010) on the Middle Eastern/North African banks, but opposite to the findings of Maudos et al. (2002) on 10 European countries’ banks, Grigorian & Manole (2006) on 17 Eastern European countries’ banks and Lozano-Vivas & Pasiouras (2010) on 87 countries’ banks, where real GDP growth rate is positively related to bank cost efficiency. On the contrary, it is hypothesized that economic growth will positively influence cost efficiency in Ghana’s banks. One possible explanation is that during higher economic growth (and therefore increased demand for bank financing) the banks lower their operating standards, such as relax evaluation of borrowers and monitoring of credit (reduce their capital ratio through aggressive lending resulting in higher costs) and thereby become less cost efficient. Thus, higher economic growth leads to greater risk taking (in less competitive banking markets) resulting in reduction in bank efficiency (Soedarmono et al., 2011). Generally an increase in inflation rate leads to increase in bad debts which reduces bank cost efficiency because the banks incur more costs in managing bad debts indicating a negative relationship between inflation rate and cost efficiency. Contrary to our expectation, the results show that the inflation coefficient is positive but statistically insignificant, implying that inflation has a weak influence on efficiency. In other words, the evidence suggests that high inflation in Ghana does not contribute to bank cost efficiency. The positive relationship revealed in this study indicates that Ghana’s banks are able to charge higher rates in a high inflationary environment to compensate for their returns (see Chan & Karim, 2010). This finding supports the study of Kasman &Yildirim (2006) who find no relationship between inflation and cost efficiency. The cost inefficiency in Ghana’s banking industry reflects the higher cost of operation mainly due to inadequate credit monitoring (and hence high non-performing loans) and inefficient control of operating expenses particularly high staff cost and cost of funds. This implies that banks operating in a less competitive banking market such as Ghana are able to charge higher prices and surprisingly, are not under any pressure to control their costs (see Maudos et al, 2002) and therefore become less cost efficient. In general, banks encounter problems of adverse selection and moral hazard caused by asymmetric information between the bank and its customers. Banks can reduce adverse selection by screening and monitoring borrowers to reduce moral hazard behavior (Vennet, 2002) in order to reduce bad debts (non-performing loans) and therefore total costs leading to increase in cost efficiency. CONCLUSIONS Ghana banking industry has undergone considerably transformation over the last 20 years. Using 25 banks over the period 2001-2010, this paper examines the cost efficiency of banks in Ghana using the DEA. In addition, fixed effect and two-step system GMM estimators are to investigate the determinants of bank cost efficiency. The findings reveal relatively low average efficiency scores for Ghana’s banks during the study period, suggesting that Ghana banks are operating far from the efficiency frontier. This finding is attributed to underutilization or waste of input resources. The cost efficiency scores show variance over time. The findings reveal that bank capitalization has negative and significant effect on bank cost efficiency suggesting that well-capitalized banks are less cost efficient. Similarly, GDP growth rate negatively impacts bank cost efficiency. The findings also show that lagged cost efficiency is an important factor in determining bank cost efficiency in Ghana. The level of bank cost efficiency is low in


82

Ghana, but it persists from year to year. Loan loss provision ratio, bank size and rate of inflation, however, are not important factors in influencing bank cost efficiency in Ghana. The findings of this study offer important implications for bank regulation, policy decisions and bank management in Ghana. The results indicate that GDP growth negatively influences bank cost efficiency. This suggests that banks lower their evaluation standards of borrowers or reduce their monitoring of loan performance during the boom period. Therefore, regulators and policymakers should pay attention to risk management and control procedures of Ghana banks (e.g., loan review, collateral appraisal). Bank of Ghana has twice increased the minimum capital requirement in 2003 and 2009, but the findings indicate that bank capitalization reduces cost efficiency. The bank cost efficiency in Ghana persist from year to year which indicates bank management ability and quality (knowledge) and technologies assist the banks to lower costs (see Ataullah & Le, 2006). The persistent cost efficiency should encourage banks to focus on reducing cost efficiency in order to reduce financial and operating costs which would help increase the bank’s profits. The small number of Ghana banks prevents this study from employing more determinant factors such as bank profitability, liquidity, interest rate, market share and bank concentration (measured by the HHI) for both bank efficiency and competition for the dynamic system GMM estimations. This is because increasing the determinant factors will increase the number of instruments in the system GMM estimation which may invalidate the system GMM results. The increase in the number of instruments could become large relative to the number of banks in the regression. This could generate too many instruments (over-fitting endogenous variables) in the system GMM estimations which will weaken the specification tests and bias the results (Roodman, 2007, 2009). Thus, when the instrument count is high, the Hansen test of validity of the instruments weakens (Roodman, 2009). This could mean accepting a model as valid when the problem of endogeneity is partially solved. Recommendations for Further Research Profit efficiency essentially captures the efficiencies (or inefficiencies) using both input and output variables, unlike cost efficiency which involves only input variables. Computing profit efficiency, therefore, constitutes a more important source of information for bank management. Therefore, investigating the profit efficiency of Ghana banks in future research would enrich the banking literature. This study only examines Ghana’s banking industry and we suggest that future research could use cross-country studies including other African states such as Nigeria, Kenya, Zambia, Tanzania and Uganda which have also undertaken similar financial reforms. Such a study may provide useful information about cross-country comparison of bank efficiency and competition in other countries with banks in Ghana. REFERENCES Acquah, P. A. (2009). Enhancing confidence in the Ghanaian financial system in the midst of the financial crisis. BIS Review, 60. Altunbas, Y., Carbo, S., Gardener, E. P. M., & Molyneux, P. (2007). Examining the relationships between capital, risk and efficiency in European banking. European Financial Management, 13(1), 49-70. Altunbas, Y., De Bondt, G., & Marques-Ibanez, D. (2004). Bank capital, bank lending, and monetary policy in the euro area. Centre for Banking and Financial Studies, School of Accounting , Banking and Economics, University of Wales. Aly, H. Y., Grabowski, R., Pasurka, C., & Rangan, N. (1990). Technical, scale, and allocative efficiencies in U.S. banking: An empirical investigation. The Review of Economics and Statistics, 72(2), 211-218.


83

Amissah-Arthur, K. B. (2010). Risk management in well-capitalized banks. BIS Review 92. Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics, 68(1), 29-51. Ataullah, A., Cockerill, T., & Le, H. (2004). Financial liberalization and bank efficiency: a comparative analysis of India and Pakistan. Applied Economics, 36(17), 1915-1924. Ataullah, A., & Le, H. (2006). Economic reforms and bank efficiency in developing countries: the case of the Indian banking industry. Applied Financial Economics, 16(9), 653-663. Baltagi, B. (1995). The Econometric Analysis of Panel Data New York, NY: Wiley. Bank of Ghana. (February 2009). Financial stability report. Volume 5 no. 1/2009, Accra, Ghana. Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some Models for Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis. Management Science, 30(9), 1078-1092. Berger, A. N. (2007). International comparisons of banking efficiency. Financail Market Institutional Instruments, 16, 119–144. Berger, A. N., & Humphrey, D. B. (1991). The dominance of inefficiencies over scale and product mix economies in banking. Journal of Monetary Economics, 28, 117-148. Berger, A. N., & Mester, L. J. (1997). Inside the black box: What explains differences in the efficiencies of financial institutions? Journal of Banking & Finance, 21(7), 895-947. Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115-143. Bond, S. R. (2002). Dynamic panel data models: A guide to micro data methods and practice. Institute for Fiscal Studies Working Paper, CWP09/02. Brissimis, S. N., Delis, M. D., & Papanikolaou, N. I. (2008). Exploring the nexus between banking sector reform and performance: Evidence from newly acceded EU countries. Journal of Banking & Finance, 32(12), 2674-2683. Campbel, S., Reeves, D., Kontopantelis, E., Middleton, E., Sibbald, B., & Roland, M. (2007). Quality of primary care in England with the introduction of pay for performance. The New England Journal of Medicine, 357(2), 181-190. Casu, B., & Girardone, C. (2004). Financial conglomeration: efficiency, productivity and strategic drive. Applied Financial Economics, 14(10), 687-696. Casu, B., & Girardone, C. (2006). Bank competition, concentration and efficiency in the single European market. The Manchester School, 74(4), 441-468. Casu, B., & Molyneux, P. (2003). A comparative study of efficiency in European banking. Applied Economics, 35(17), 1865-1876.


84

Chan, S.-G., & Karim, M. Z. A. (2010). Bank efficiency and macro-economic factors: The case of developing countries. Global Rconomic Review, 39(3), 269 — 289. Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research 6, 429-444. Chen, T., & Yeh, T. L. (1998). A Study of efficiency evaluation in Taiwan’s banks. International Journal of Service Industry Management, 9(5), 402-415. Coelli, T. J., Rao, D. S. P., & Battese, G. E. (1998). An Introduction to Efficiency and Productivity Analysis. Boston: Kluwer Academic Publishers. Cox, D. R. (1970). The Analysis of Binary Data. Methuen, London. Das, A., & Ghosh, S. (2006). Financial deregulation and efficiency: An empirical analysis of Indian banks during the post reform period. Review of Financial Economics 15, 193-221. Denizer, C., Dinc, M., & Tarimcilar, M. (2007). Financial liberalization and banking efficiency: evidence from Turkey. Journal of Productivity Analysis 177–195. Drake, L. (2001). Efficiency and productivity change in UK banking. Applied Financial Economics, 11(5), 557-571. Elyasiani, E., & Mehdian, S. (1990). Efficiency in the commercial banking industry, a production frontier approach. Applied Economics, v22(n4), p539(513). Fang, Y., Hasan, I., & Marton, K. (2011). Bank efficiency in transition economies: recent evidence from South-Eastern Europe. Bank of Finland Research Discussion Papers 5/2011. Fiordelisi, F., Marques-Ibanez, D., & Molyneux, P. (2011). Efficiency and risk in European banking. Journal of Banking & Finance, 35(5), 1315-1326. Girardone, C., Molyneux, P., & Gardener, E. P. M. (2004). Analysing the determinants of bank efficiency: the case of Italian banks. Applied Economics, 36(3), 215(213). Goddard, J. A., Molyneux, P., & Wilson, J. O. S. (2001). European Banking: Efficiency, Technology and Growth John Wiley & Sons. International Monetary Fund. (1999). Ghana: Selected issues. IMF Staff Country Report No.99/3, International Monetary Fund, Washington, D.C. International Monetary Fund. (2011). World Economic Outlook Database. IMF World Economic Outlook, International Monetary Fund, Washington, D.C. Isik, I., & Hassan, M. K. (2002). Cost and profit efficiency of the Turkish banking industry: An empirical investigation. The Financial Review, 37(2), 257-279. Kablan, S. (2010). Measuring bank efficiency in developing countries: the case of WAEMU (West African economic monetary union). African Economic Research Consortium.


85

Kader, H. A., Adams, M. B., & Hardwick, P. (2010). The cost efficiency of Takaful insurance companies. The Geneva Papers on Risk and Insurance - Issues and Practice 35(1), 161-181. Kwan, S., & Eisenbeis, R. (1997). Bank risk, capitalization and operating efficiency. Journal of Financial Services Research 12, 117–131. Loukoianova, E. (2008). Analysis of the efficiency and profitability of the Japanese banking system. IMF Working Paper No. 08/63. Maddala, G. S. (1983). Limited dependent and qualitative variables in econometrics. Cambridge: Cambridge University Press. Manlagnit, M. C. V. (2011). Cost efficiency, determinants, and risk preferences in banking: A case of stochastic frontier analysis in the Philippines. Journal of Asian Economics 22 22, 23–35. Maudos, J., & Fernández de Guevara, J. (2004). Factors explaining the interest margin in the banking sectors of the European Union. Journal of Banking & Finance, 28(9), 2259-2281. Maudos, J., & Fernández de Guevara, J. (2007). The cost of market power in banking: Social welfare loss vs. cost inefficiency. Journal of Banking & Finance, 31(7), 2103-2125. Maudos, J., & Pastor, J. M. (2003). Cost and profit efficiency in the Spanish banking sector (1985–1996): a non-parametric approach. Applied Financial Economics, 13(1), 1-12. Maudos, J., Pastor, J. M., Pérez, F., & Quesada, J. (2002). Cost and profit efficiency in European banks. Journal of International Financial Markets, Institutions and Money, 12(1), 33-58. McKinnon, R. I. (1973). Money and Capital, The Brookings Institution, Washington. Miller, S. M., & Noulas, A. G. (1996). The technical efficiency of large bank production. Journal of Banking & Finance, 20(3), 495-509. PricewaterhouseCoopers. (2011). Ghana Banking Survey. Accra: PricewaterhouseCoopers Roodman, D. (2006). How to do xtabond2: An introduction to difference and system GMM in Stata. Center for Global Development Working Paper No. 103. Roodman, D. (2009). How to do xtabond2: An introduction to difference and system GMM in Stata. The Stata Journal 9(1), 86–136. Roodman, D. (2009). A note on the theme of too many instruments Oxford Bulletin of Economics and Statistics, 71(1), 135-158. Salas, V., & Saurina, J. (2003). Deregulation, market power and risk behavior in Spanish banks. European Economic Review, 47, 1061–1075. Sathye, M. (2001). X-efficiency in Australian banking: An empirical investigation. Journal of Banking & Finance, 25(3), 613-630. Sealey, C., & Lindley, J. (1977). Inputs, outputs and a theory of production and cost of depository financial institutions. Journal of Finance, 32, 1251–1266.


86

Solís, L., & Maudos, J. (2008). The social costs of bank market power: Evidence from Mexico. Journal of Comparative Economics 36, 467–488. Staikouras, C., Mamatzakis, E., & Koutsomanoli-Filippaki, A. (2008). Cost efficiency of the banking industry in the South Eastern European region. Journal of International Financial Markets, Institutions and Money, 18(5), 483-497. Staikouras, C., Mamatzakis, E., & Koutsomanoli-Filippaki, A. (2008.). Cost efficiency of the banking industry in the South Eastern European region. Journal of International Financial Markets, Institutions and Money 18(5), 483–497. Sufian, F. (2009). Determinants of bank efficiency during unstable macroeconomic environment: Empirical evidence from Malaysia. Research in International Business and Finance 23, 54–77. Tabak, B. M., Fazio, D. M., & Cajueiro, D. O. (2011). Profit, cost and scale effciency for Latin American banks: Concentration-performance relationship. Central Bank of Brazil Working Paper No. 244, 1-37. Tecles, P. L., & Tabak, B. M. (2010). Determinants of bank efficiency: The case of Brazil. European Journal of Operational Research, 207(3), 1587-1598. Vennet, R. V. (2002). Cost and profit efficiency of financial conglomerates and universal banks in Europe. Journal of Money, Credit and Banking, 34(1), 254-282. Voos, P. B., & Mishel, L. R. (1986). The union impact on profits: Evidence from industry price-cost margin data. Journal of Labour Economics, 4(1), 105-133. Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step GMM estimators. Journal of Econometrics, 126, 25–51. World Bank. (1989). Ghana: Structural adjustment for growth. Report No. 7515-GH, World Bank, Washington, DC. Yildirim, H. S., & Philippatos, G. C. (2007). Efficiency of banks: Recent evidence from the transition economies of Europe, 1993–2000. The European Journal of Finance, 13(2), 123-143. BIOGRAPHY Kofi Adjei-Frimpong, Former PhD Student, Faculty of Commerce, Department of Accounting, Economics and Finance, PO Box 84, Lincoln University, Canterbury, New Zealand, Tel: 64-3-325-2811, Fax: 64-3-325-3847, Email: [email protected] Christopher Gan, corresponding author, Professor of Accounting and Finance, Faculty of Commerce, Department of Accounting, Economics and Finance, PO Box 84, Lincoln University, Canterbury, New Zealand, Tel: 64-3-325-2811, Fax: 64-3-325-3847, Email: [email protected] Baiding Hu, Senior Lecturer, Faculty of Commerce, Department of Accounting, Economics and Finance, PO Box 84, Lincoln University, Canterbury, New Zealand, Tel: 64-3-325-2811, Fax: 64-3-325-3847, Email: [email protected]


87

EFFECTS OF SERVICE INNOVATION ON FINANCIAL PERFORMANCE OF SMALL AUDIT FIRMS IN

TAIWAN Yi-Fang Yang, Chang Jung Christian University

Lee-Wen Yang, Chaoyang University of Technology Yahn-Shir Chen, National Yunlin University of Science and Technology

ABSTRACT

This study examines the effects of service innovation on financial performance of proprietorship audit firms in Taiwan. This study divides total sample into three business strategy categories, including conventional, non-conventional, and general firms. Non-conventional firms have the highest degree of service innovation followed by general firms. Conventional firms have the lowest degree of service innovation. Empirical results indicate that non-conventional firms financially outperform general firms, and the latter outperforms conventional firms. JEL: M42 KEYWORDS: Service Innovation, Financial Performance, Audit Firms INTRODUCTION

hen society becomes more complex, the investing public needs reliable information to make economic decisions, including whether to invest in an organization. Reliable accounting and financial reporting aid society in allocating resources in an efficient manner. Independent

auditors provide credibility to the information, reducing information risk. Auditors practice by establishing audit firms in the forms of proprietorships, partnerships, or corporations (Elder, Beasley and Arens, 2008). A dual market structure exists in the audit market with a few larger audit firms and a large number of smaller firms (Bröcheler, Maijoor and Witteloostuijn, 2004). Most prior studies explore topics relating to large audit firms, especially big international audit firms (e.g., Iyer and Iyer, 1996; McMeeking, Peasnell and Pope, 2007; Minyard and Tabor, 1991). Small audit firms, such as proprietorship firms, experience less investigation due primarily to limited data availability. A study of proprietorship audit firms appears warranted. To fulfill their social role, audit firms traditionally provide audit and non-audit services. Audit related services include audits of financial statements and income tax returns, corporate registration, and accounting and bookkeeping. Non-audit related services typically refer to management advisory services (MAS). As an innovative service, MAS range from a simple suggestion for improving the clients’ accounting system to advising in risk management, information technology, e-commerce system design, mergers and acquisitions, and actuarial benefit consulting (Elder et al., 2008). Audit firms provide audit services for years but their clients increasingly demand MAS due to global competition and rapid technological changes in recent years. Prior studies designate audit related services as traditional practices and MAS non-traditional practices (Banker, Chang and Natarajan, 2005; Rescho, 1987). The environment-strategy-performance (ESP) perspective posits that specific environmental conditions have a corresponding preferred strategic response (e.g. Luo, Tan and Shenkar, 1998; Tan and Tan, 2005). In terms of s resource-based view of the firm, audit firms provide different services to satisfy clients’ demands and thereby adopt different business strategies. Some audit firms adopt a conservative business strategy to provide traditional practices.. Some take an aggressive business strategy and focus on providing MAS practices, which have the highest degree of service innovation. Furthermore, more audit firms adopt a strategy to offer both traditional and MAS practices, the moderate innovative service

W

YF. Yang et al | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

88

provision firms. Of interest is whether the financial performance of proprietorship audit firms taking varied degree of service innovation differs? To answer the question constitutes our motivation. This study obtains empirical data of proprietorship audit firms from the 1989-2009 Survey Report of Audit Firms in Taiwan. Focusing the research on proprietorship audit firms adds research homogeneity (Fasci and Valdez, 1998). In terms of the degree of service innovation audit firms take, this study divides total sample firms into three categories: conventional, non-conventional, and general firms. Conventional firms are defined as proprietorship audit firms which adopt a conservative business strategy to provide traditional practices only. In contrast, non-conventional firms refer to proprietorship audit firms which take an aggressive business strategy and focus on providing MAS practices. If audit firms adopt a moderate strategy to offer both traditional and MAS practices, they are general firms. The main results indicate that non-conventional firms financially outperform general firms, and the latter outperforms conventional firms. In short, service innovations have positive effects on financial performance, the higher the service innovation degree, the better the operating results of audit firms. Findings of this study add knowledge to service business-related literatures. The rest of this paper proceeds as follows. The next section presents a literature review and hypothesis development, followed by the depiction of research methodology. The subsequent section reports empirical results. Finally, this study concludes in the last section. Literature Review and Hypothesis Development The environment-strategy-performance (ESP) perspective posits that specific environmental conditions have a corresponding preferred strategic response (e.g. Tan and Tan, 2005; Tang and Tang, 2012; Volberda and Lewin, 2003). Companies seek to respond to the external environment effectively to gain competitive forces (Porter, 1990). Strategies serve to exploit the companies’ capability as a weapon to achieve their missions and objectives. A clear strategy can play an important role in the companies’ success. Strategies link external market requirements with internal organizational and technological resources, and capabilities (Sun and Hong, 2002). Discussions on strategies at different levels include corporate, business, and functional strategies. Corporate strategies describe a company’s overall direction. Business strategies occur at a business unit level. Functional strategies develop a distinctive competence to provide a company or a business unit with a competitive advantage (Hunger and Wheelen, 2001). The three strategies are not mutually exclusive, and link in implementing a particular strategy (Miles, Kastrinos, Flanagan, Bilderbeek, Hertog, Huntink and Bouman, 1995). Business strategies are a set of decisions about the direction of a company. Companies select a business strategy according to evaluations the companies make about their distinctive competencies and the competing environment (Mintzberg, 1990). Because audit firms take different strategies as a means of organizational adaptation, a strong relationship exists between strategy type and performance (Rescho, 1987). Facing varied degrees of market competition and regulation, proprietorship audit firms provide different services to fulfill the business strategies they adopt. A typical proprietorship audit firm may provide different practices, including auditing financial statements of privately held companies, auditing financial statements for granting a bank loan, auditing financial statements for other purposes, auditing an income tax return, corporate registration, accounting, and MAS (Elder et al., 2008). For long, audit firms have provided the preceding four audit services, corporate registration, and accounting services. Prior studies thus designate them as traditional practices, and MAS as non-traditional practices (Banker et al., 2005; Rescho, 1987). Traditional practices are law-protected and statutory and regulated by the Generally Accepted Auditing Standard (GAAS). Specifically, audit services are required by the Company Act, Business Accounting Act, and the Securities and Exchange Act. However, MAS practices require a diverse product line, customization, and service innovation. As the traditional practices are a long-standing service, auditors offer them with standardized procedures to relatively stable clients. Audit firms offer MAS such as personal financial planning, integrated tax planning, information technology (IT) and electrical commerce advisory services, mergers and acquisitions (M&A), budgeting


89

and forecasting services, business valuation, and pension funds actuarial advisory services (Arens, Elder and Mark, 2012). MAS practices require a diverse product line, customization, and service innovations. Proprietorship audit firms adopt varied business strategies according to their capacity and proprietors’ distinctive competencies, such as academic background, professional experience and expertise, and the customer network. Audit firms may adopt a conservative business strategy to provide traditional practices only. In contrast, audit firms can adopt an aggressive business strategy and focus on providing MAS practices. The third type of business strategy audit firms take falls between aggressive and conservative business strategies, a moderate strategy with which audit firms offer both traditional and MAS practices. In terms of service innovations, this study defines proprietorship audit firms only providing traditional practices as conventional firms. In contrast, this study terms proprietorship audit firms focusing on MAS practices as non-conventional firms. When proprietorship audit firms offer both traditional and MAS practices, this study names them general firms. In Taiwan, related laws and regulations require companies’ financial statements to be audited by audit firms, resulting in law-protected and statutory traditional practices. Because traditional services are a general requirement by various governmental agencies, some accounting educators and accounting practitioners view them as services that clients need but do not necessarily want (Istvan, 1984). Early entrants gain competitive advantage more easily than subsequent ones. However, beginning in 1988, Taiwanese authorities have raised the passing rate of the Certified Public Accountant (CPA) uniform examination, leading to substantial increases in the number of qualified CPAs and in market competition. In 1998, the authorities abolished the long-standing audit fee standard to ensure fair audit market competition. Cancelling the audit fee standard adversely impacts the traditional practice market. Since then, a rumor of price-cutting strategy for client solicitation has prevailed in the industry and the audit market competition has enhanced. Furthermore, the tax authorities established a tax agent system and legalized the provision of corporate registration and accounting services by tax agents to small and medium-sized entities (SMEs) in 2004. Proprietorship audit firms have provided the same traditional practices to the SMEs for years. Tax agent legalization negatively affects proprietorship audit firms because of the competitive advantages the tax agents possess for a relatively lower service fees and easy service access by the clients. Facing recent worldwide competition and business globalization, companies consult with a professional management advisor concerning business administration and information technology to advance their international competitiveness. In practice, auditors have provided services to the same clients for years and are familiar with the clients’ daily operation and financial condition. Under the situation of long-term partnership and close client relations, audit firms gain a more favorable position in providing MAS than an ordinary professional consulting firm, such as McKinsey & Company. Further, joint provisions of audit services and non-audit (MAS) theoretically create synergy and knowledge spillover effects for audit firms (Beck, Frecka and Solomon, 1988; Simunic, 1984). Auditors devote more involvements and communications in providing MAS to meet clients’ demand for specific services, resulting in more flexible service provisions in format, timing, and place. As a tailor-made and innovative practice, MAS generally brings higher profits, higher growth potential and industry expansion rather than predatory competition (Rescho, 1987). Beginning in the 1990s, auditors have begun shifting their human resources from traditional, low-margin revenue product areas of auditing and accounting into relatively new, high-margin revenue product areas of MAS (Banker et al., 2005). In public accounting profession, different business strategies adopted by auditors lead to provision of varied services. A typical proprietorship audit firm provides either audit or non-audit services or both. Audit services include, but not limited to, attestation of financial statements for granting a bank loan, and attestation of an income tax return. Non-audit services comprise provisions of tax planning, administrative remedy of internal taxation, other tax operations, consultation, corporate registration, bookkeeping and accounting services. In practice, attestation, corporate registration, bookkeeping and accounting services have been provided for years. These services are referred to as traditional businesses, which are offered to relatively stable customers with standardized serving procedures. Auditors providing traditional services adopt a relatively conservative and moderate business strategy. In contrast, tax planning, administrative remedy of internal taxation, other tax


90

operations, and consulting services are referred to as non-traditional businesses, which require diverse product line, customization, and service innovation. Auditors focusing on non-traditional business tend to adopt service differentiation as their business strategy. Porter (1990) utilizes methods of gaining or sustaining competitive advantages to develop three general business strategies: low-cost producer, product differentiator and focused operation. Miles and Snow (1984) identify three types of strategies, including prospector, defender, and analyzer. The prospector pursues market expansion and innovation, the defender strives to maintain market position, and the analyzer seeks some combination of market expansion/innovation while endeavoring to preserve stability in existing markets. Although the classifications of business strategies differ, underlying concepts in Miles, Snow, Meyer and Coleman (1978) and Porter (1990) are qualitatively the same. The defender, prospector, and analyzer business strategies in Miles et al. (1978) essentially equate the low-cost producer, product differentiator and focused operation in Porter (1990) in terms of the overall strategic orientations (Miles and Snow, 1984). A typical proprietorship audit firm may provide traditional services only, non-traditional services only, or both. Following Miles et al. (1978) and based on auditing industrial peculiarity, we define proprietorship audit firms providing traditional services only, non-traditional services only, and both services as conventional firms, non-conventional firms, and general firms, respectively. In sum, the traditional practice market is saturated and increasingly competitive but MAS practice market exists potentially unlimited opportunities, resulting in low-margin profits for conventional firms but high-margin profits for non-conventional firms. Because general firms situate between traditional and MAS practice markets, they have moderate-margin profits. As a result, this study establishes the following hypotheses to distinguish the financial performance effects of proprietorship audit firms taking varied business strategies. H1: Financial performance of non-conventional audit firms is better than that of general audit firms H2: Financial performance of general audit firms is better than that of conventional audit firms H3: Financial performance of non-conventional audit firms is better than that of conventional audit firms. METHODOLOGY Data Empirical data are from the 1989-2009 Survey Report of Audit Firms in Taiwan, published by the Financial Supervisory Commission (FSC) annually except in 1991 due to the year’s inseparable data from other industries’ statistics. To collect business information on the public accounting profession for macro-economic analysis and industrial policy formation, the FSC administers the survey over all registered audit firms annually. Contents of the survey include quantitative information of total revenues and their compositions, total expenses and their compositions, demographics of various levels of employees, and ending amounts of and changes in fixed assets. An open questionnaire collects qualitative information by asking about operating difficulties audit firms encounter and future business orientation audit firms take. Because the FSC administers the survey pursuant to the Statistics Act, it require audit firms surveyed to fill out the questionnaire correctly within the due time. Thus, the Survey Report reveals an annual response rate of over eighty percent. As the sample period of this study is 20 years, this study deflates all monetary variables by the yearly Consumer Price Index to account for inflation. This study deletes firm-year observations that newly established in the survey year and that with dependent variables having values more or less than three standard deviations away from their means. The final number of observations is 9,220, including 123 non-conventional firms, 5,016 general firms and 4,081 conventional firms. This information indicates that most proprietorship audit firms, 54.40 percent (5,016/9,220), provide both traditional practices and MAS.


91

The percent of audit firms only providing traditional services is 44.26 percent (4,081/9,220) and that of firms exclusively focusing on MAS is 1.33 percent (123/9,220). Taken together, over half of the proprietorship audit firms, 55.74 percent (5,139/9,220), render MAS. Model Specification This study obtains empirical data of registered audit firms from Taiwanese public accounting industry. From the perspective of industrial economics and based on the structure-conduct-performance theoretical framework (Cowling and Waterson, 1976), this study establishes the following linear regression equation to test our hypotheses. 𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = α0 + α1 𝐷𝐷_𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + α2 𝑃𝐼𝑃𝑃𝑃𝑆𝑃𝐼𝐶𝑃 + α3 𝑃𝐷𝑈𝐶𝐴𝑇𝑆𝑃𝐼 + α4 𝐶𝑃𝑃

+ 𝛼5 𝑆𝑆𝑆𝑃 + 𝛼6 𝑆𝐼𝐷𝑃𝐼 + 𝑠 (1) Definitions of Variable Accounting defines financial performance as total revenues minus total expenses, net income or net profit. Sole proprietors are the owner and residual interest claimant of proprietorship audit firms and their annual income comprises salaries received from the firms and share of operating profit of the firms. Salaries of the sole proprietors, weekly or monthly, are a part of total expenses. The more the salaries of the sole proprietors, the less the operating profit of the firms. It makes no difference to the sole proprietors whether they receive salaries or not in terms of their total annual income. In addition, the criteria for salary payments to the sole proprietors vary across firms. Based on prior studies (Chen, Chang and Lee, 2008), this study adds their salaries back to net income to reduce such an artificial noise and has the following operational definition. Hence, the financial performance is net profit of the audit firms. 𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = 𝑇𝑇𝑠𝑠𝑇 𝑃𝑠𝑅𝑠𝑅𝑅𝑠 𝑇𝑜 𝐴𝑅𝐴𝐴𝑠 𝑃𝐴𝑠𝐹𝑠 − 𝑇𝑇𝑠𝑠𝑇 𝑃𝐸𝐸𝑠𝑅𝑠𝑠𝑠 𝑇𝑜 𝐴𝑅𝐴𝐴𝑠𝑠 𝑃𝐴𝑠𝐹𝑠

+ 𝑆𝑠𝑇𝑠𝑠𝐴𝑠𝑠 𝑃𝑠𝐴𝐴 𝑠𝑇 𝑠ℎ𝑠𝐴𝑠 𝑆𝑇𝑇𝑠 𝑃𝑠𝑇𝐸𝑠𝐴𝑠𝑠𝑇𝑠𝑠 (2) One of the research variables in this study is a dummy variable of business strategy (DV_strategy). In terms of the business strategies audit firms take, this study classifies total sample into three categories: conventional, non-conventional, and general firms. This study defines conventional firms as audit firms that have positive revenues from traditional practices but have no revenue from MAS. In contrast, if audit firms have positive revenues from MAS but have no revenue from traditional practices, this study term them as non-conventional firms. General firms refer to audit firms that have positive revenues from both traditional practices and MAS. This study employs the dummy variable of business strategy (DV_strategy) to distinguish among the conventional, non-conventional, and general firms. Apart from the research variables, this study includes other influences on financial performance as control variables. After acquiring academic qualifications in accounting, most professionals enter their careers as assistants in audit firms. They continue to learn and gain experience and expertise through learning by doing. The average years of experience for partners, managers, seniors, in-charge auditors, and assistants are over 10 years, 5-10 years, 2-5 years and 0-2 years, respectively (Elder et al., 2008). Previous studies find a positive association between employee experience and job performance (e.g., Schmidt, Hunter and Outerbridge, 1986), and point out that work experience relates positively to the performance of proprietorship audit firms (Fasci and Valdez, 1998; Collins-Dodd, Gordon and Smart, 2004; Chen et al., 2008). Therefore, this study expects a positive association between work experience of auditors and financial performance. Practitioners note that auditors older than 35 years have worked in audit firms for more than 5 years and have accumulated much practical experience. Thus, this study defines work experience of auditors (EXPERIENCE) as number of auditors older than 35 years as a percent of number of total auditors. Adequate technical training and proficiency as auditors require a


92

college or university education in accounting and auditing. Presumably, auditors with higher academic education level possess more and better knowledge, and have higher intellectual potential in learning and accumulating skills and expertise. Some prior studies report that auditors with a higher level of education improve audit firm performance (Bröcheler et al., 2004), but some find insignificant association between educational level of auditors and performance (Collins-Dodd et al., 2004; Fasci and Valdez, 1998). Hence, this study does not specify a directional prediction on the relationship between education level of auditors and financial performance. This study measures education level of auditors (EDUCATION) by a mean number of years auditors need to obtain an academic qualification. To remain knowledgeable about the endless stream of changes in accounting and auditing standards, tax laws, information technology, and consulting skills, auditors must comply with a requirement of taking part in continuing professional education. Prior researches on training for public accounting industry indicate that professional training enhances auditors’ competency and audit performance (Bonner and Pennington 1991; Grotelueschen, 1990; Thomas, Davis and Seaman, 1998). Further, continuing professional education positively relates to financial performance of audit firms (Chen, Chen and Lee, 2002; Chen et al., 2008). This study expects a positive association between financial performance and continuing professional education of auditors (CPE) which is defined as expenditures on professional training of audit firms. Size of a company might substitute for many omitted variables and its inclusion as a control variable enhances the accuracy of model specification (Becker, DeFond, Jiambalvo and Subramanyam, 1998). Prior studies estimate audit firm size by either the number of full-time employees (Collins-Dodd et al., 2004) or market share of the individual firms (Chen et al., 2002; Chen et al., 2008), and report a positive relationship between audit firm size and performance (Chen et al., 2002; Chen et al., 2008; Collins-Dodd et al., 2004; Rescho, 1987). This study defines audit firm size (SIZE) as natural logarithm of total revenues of the firms and expects a positive relationship to financial performance. The sample period of this study is 20 years and spans over two centuries. As a professional organization, audit firms are affected by the local economy or environment factors (e.g., Reynolds and Francis, 2001). Economic indicator, Taiwan Gross Domestic Product, is included to control for local economy effects. However, auditors provide services to the same clients for years and most of their practices are statutory, making the effects of environment factors on financial performance limited. Accordingly, this study does not specify a directional prediction on the relationship between economic indicator (INDEX) and financial performance. RESULTS Descriptive Statistics Table 1 displays the descriptive statistics for variables used in regression model. Panel A of Table 1 shows descriptive statistics for non-conventional firms. Mean financial performance (PERFORM) is $590,761. Work experience of auditors (EXPERIENCE), on average, is 0.700 which represents that 70 percent auditors are older than 35 years. Education level of auditors (EDUCATION) is 15.569, meaning that average education level of auditors lies between junior college degree and bachelor degree. Average expenditures on professional training of non-conventional firms (CPE) are $3,768. Mean non-conventional firm size (SIZE) is 13.116. Panel B presents the descriptive statistics of general firms. Mean financial performance (PERFORM) is $841,549. Work experience of auditors (EXPERIENCE) indicates that 45.8 percent auditors are older than 35 years. Education level of auditors (EDUCATION) of general firms is 15.539. Average expenditures on professional training of general firms (CPE) are $21,582. Mean general firm size (SIZE) is 14.903. Panel C indicates the descriptive statistics of conventional firms. Mean financial performance (PERFORM) is $553,822. Mean experience of auditors (EXPERIENCE) represents that 49.8 percent auditors are older than 35 years. Average education level of auditors (EDUCATION) is 15.187. Average expenditures on professional training of conventional firms (CPE) are $12,556. Mean conventional firm size (SIZE) is 14.423. The untransformed figure indicates that average total revenues of the firms are between $9,390,321 and 9,773,998.


93

Table 1: Descriptive Statistics

mean std. dev. mini. maxi. q1 median q3 panel a descriptive statistics of non-conventional firms (n=123) perform 590,761 1,522,701 -1,053,259 7,375,596 5,081 73,940 425,019 experience 0.700 0.394 0 1 0.364 1 1 education 15.569 2.983 9 34 14.333 16 17 cpe 3,768 12,225 0 76,938 0 0 275 size 13.116 1.513 8.294 16.119 12.300 13.305 14.046 index 9,547,341 2,473,120 4,974,759 13,070,681 7,536,283 9,570,584 11,612,093 panel b descriptive statistics of general firms (n=5,016) perform 841,549 929,706 -2,926,033 16,216,676 242,007 642,899 1,213,963 experience 0.458 0.293 0 1 0.235 0.400 0.667 education 15.539 6.738 0 154 14 14.667 15.500 cpe 21,582 94,738 0 5,097,905 0 2,000 9,500 size 14.903 0.932 5.687 17.182 14.408 15.025 15.521 index 9,773,998 2,357,859 4,974,759 13,070,681 7,953,510 9,731,208 12,243,471 panel c descriptive statistics of conventional firms (n=4,081) perform 553,822 720,160 -2,697,432 8,791,606 83,898 387,778 812,748 experience 0.498 0.325 0 1 0.250 0.429 0.750 education 15.187 4.812 7 198 14 14.667 15.500 cpe 12,556 41,804 0 862,400 0 0 6,000 size 14.423 1.127 6.234 16.968 13.845 14.615 15.202 index 9,390,321 2,398,608 4,974,759 13,070,681 7,536,283 9,570,584 11,612,093

Table 1 shows the descriptive statistics for variables used in regression model. PERFORM is equal to financial performance of audit firms. DV_ strategy is a dummy variable of business strategy. EXPERIENCE represents the work experience of auditors. EDUCATION stands for the education level of auditors. CPE is continuing professional education of auditors. SIZE represents the audit firm size. INDEX is an economic indicator. PERFORM, CPE, and INDEX are expressed in new Taiwan dollars. Correlation Analysis This study analyzes the Pearson correlation coefficients between dependent and independent variables used in regression models. The empirical results show the high correlation coefficients between financial performance (PERFORM) and size of audit firm (SIZE). However, the variance inflation factors (VIFs) are less than 10 (un-tabulated), implying that no serious multi-collinearity exists among the independent variables. Table 2: Correlation Matrix

variables perform experience education cpe size index perform 1 experience -0.107 1 education 0.009 0.028 1 cpe 0.207 -0.044 0.031 1 size 0.591 -0.283 -0.007 0.185 1 index 0.041 0.257 0.125 0.046 0.093 1

Table 2 shows the correlation for variables used in regression model. The number of total observations is 9,220. Regression Results Table 3 displays the OLS regression results of financial performance comparisons between general and non-conventional firms, conventional and general firms, and conventional and non-conventional firms in Columns (A), (B), and (C). The three regression models have good model specification with explanatory power of model (adjusted R2) lying between 0.332 and 0.394. This study uses White (1980) robust standard errors to calculate all t-statistics of coefficients to correct for heteroscedasticity. As a check on the multi-collinearity between independent variables, this study estimates the variance inflation factors (VIF). In econometrics, VIF greater than 10 implies serious multi-collinearity existing among independent variables. In the regression models of Table 3, the variable VIFs are less than 1.2. In addition, this study estimates the standardized regression coefficients (Beta) for each independent variable to ease comparison between variables. Standardized coefficients possess attributes similar to correlation coefficient with values lying between -1 and +1. Higher absolute value of standardized coefficients predicts more variations in dependent variable. In the OLS standardized regression model, no intercept exists. Column (A) shows that the coefficient on the dummy variable of business strategy (DV_strategy) is significantly positive (t = 11.046 and p < 0.01). Consistent with expectation, this


94

indicates that financial performance of non-conventional firms is better than that of general firms, which supports H1. Column (B) displays a significantly positive coefficient on the dummy variable of business strategy (DV_strategy) (t = 3.762 and p < 0.01), indicating that general firms financially outperform conventional firms and H2 receives a support. Column (C) reports a significantly positive coefficient on the dummy variable of business strategy (DV_strategy) (t = 8.273 and p < 0.01), indicating that non-conventional firms are superior in financial performance to conventional firms and H3 receives a support. As a differentiated and less-competitive market exists for MAS, audit firms offering MAS are able to generate more revenues from their human resources than other firms that continue to focus on more labor-intensive audit and assurance engagements (Banker et al., 2005). The above findings document that audit firms adopting different business strategies lead to varied operating results. Specifically, both the non-conventional firms and general firms providing MAS outperform the conventional firms providing no MAS, an evidence of a natural extension of prior studies (e.g., Banker et al., 2005). Table 3: Regression Results for Comparing Financial Performance between Audit Firms Adopting Different Business Strategies

Independent Variables Dependent Variable：PERFORM

(Predicted Sign) (A)

Non-Conventional v.s. General Firms

(B) General v.s. Conventional

Firms

(C) Non-Conventional v.s.

Conventional Firms

Std. coef. (t-statistic) Std. coef. (t-statistic) Std. coef. (t-statistic)

DV_strategy ( + ) 0.125 (11.046)*** 0.032 (3.762)*** 0.106 (8.273)*** EXPERIENCE ( + ) 0.099 (8.290)*** 0.069 (7.651)*** 0.055 (4.013)*** EDUCATION ( ? ) 0.006 (0.535) 0.010 (1.192) 0.022 (1.712)** CPE ( + ) 0.084 (7.575)*** 0.102 (12.044)*** 0.114 (8.789)*** SIZE ( + ) 0.653 (54.587)*** 0.596 (65.345)*** 0.567 (40.944)*** INDEX ( ? ) -0.070 (-6.004) *** -0.042 (-4.763)*** -0.016 (-1.237) Adjusted R2 0.394 0.373 0.332 F-value 557.86*** 904.01*** 348.72*** Number of observations 5,139 9,097 4,204

Column (A) of Table 3 displays the OLS regression results of financial performance comparisons between non-conventional and general firms. It indicates that financial performance of non-conventional firms is better than that of general firms. Column (B) shows the empirical results of financial performance comparisons between general and conventional firms. It demonstrates that general firms financially outperform conventional firms. Column (c) indicates the empirical results of financial performance comparisons between non-conventional and conventional firms. It indicates that non-conventional firms are superior in financial performance to conventional firms.*, **, *** Denote one-tailed significance at the 10 %, 5 % and 1 % levels Results of Control Variable and Model Fitness of Research Variables With respect to the results of control variables shown in Tables 3, both work experience of auditors (EXPERIENCE) and size of audit firm (SIZE) are consistent with expectation and reveal a positive relationship to financial performance in all regression models. However, education level of auditors (EDUCATION), continuing professional education of auditors (CPE), and economy indicator (INDEX) indicate mixed results. Further analyses indicate that size of audit firm (SIZE) is the most important independent variable in explaining variation of dependent variable, agreeing with prior studies (Collins-Dodd et al., 2004; Chen et al., 2008). In addition, this study conducts hierarchical regression to verify the incrementally explanatory power contributed by our research variables in Tables 3. The changes in the multiple squared correlation coefficients (ΔR2) for regression models are 0.1493, 0.1262 and 0.1370 with F-statistic of 45.66, 38.60 and 41.90. All F-statistics are statistically significant at the 1 percent level. In sum, the hierarchical regression results agree with those obtained by OLS regression model, which demonstrates that our research variables explain dependent variable with both econometric and economic implications.


95

Additional Test In the regression results shown in Table 3, this study defines financial performance as net profit of the audit firms. Apart from it, another kind of performance measure is net profit per employee which is more feasible due to its consideration of firm size. Do the results in Table 3 still hold if the dependent variable is net profit per employee? In this section, we replace the net profit of the audit firms with net profit per employee and rerun the OLS regressions to examine our three hypotheses with results displayed in Table 4. The dependent variable, net profit per employee (Productivity), is defined as net profit of the audit firms divided by ending number of employees. Similar to Table 3, the comparisons between non-conventional and general firms, general and conventional firms, and non-conventional and conventional firms are listed in in Columns (A), (B), and (C). The three regression models have good model specification with explanatory power of model (adjusted R2) lying between 0.121 and 0.135. Column (A) shows that the coefficient on the dummy variable of business strategy (DV_strategy) is significantly positive (t = 16.819 and p < 0.01). Consistent with expectation, this indicates that financial performance of non-conventional firms is better than that of general firms, which supports H1. Column (B) displays a significantly positive coefficient on the dummy variable of business strategy (DV_strategy) (t = 5.491 and p < 0.01), indicating that general firms outperform conventional firms in financial performance and H2 receives a support. Column (C) reports a significantly positive coefficient on the dummy variable of business strategy (DV_strategy) (t = 18.234 and p < 0.01), indicating that non-conventional firms are superior in financial performance to conventional firms and H3 is supported. In sum, the regression results of Table 4 are similar to those in Table 3. Table 4: Regression Results for Audit Firms Adopting Different Business Strategies

Independent Variables (Predicted Sign) Dependent Variable：Productivity

(A)

Non-Conventional v.s. General Firms

(B) General v.s. Conventional

Firms

(C) Non-Conventional v.s.

Conventional Firms Std. coef. (t-statistic) Std. coef. (t-statistic) Std. coef. (t-statistic) DV_strategy ( + ) 0.228 (16.819)*** 0.056 (5.491)*** 0.267 (18.234)*** EXPERIENCE ( + ) 0.240 (16.807)*** 0.226 (20.993)*** 0.172 (11.041)*** EDUCATION ( ? ) 0.015 (1.133) 0.014 (1.430) 0.029 (1.958)** CPE ( + ) 0.069 (5.198)*** 0.109 (10.835)*** 0.006 (0.422) SIZE ( + ) 0.290 (20.308)*** 0.289 (26.719)*** 0.282 (17.868) *** INDEX ( ? ) -0.079 (-5.624) *** -0.055 (-5.272)*** -0.014 (-0.965) Adjusted R2 0.135 0.121 0.131 F-value 134.39*** 208.80*** 106.22*** Number of observations 5,139 9,097 4,204

Column (A) of Table 4 displays the OLS regression results of productivity comparisons between non-conventional and general firms. It indicates that productivity of non-conventional firms is better than that of general firms. Column (B) shows the empirical results of productivity comparisons between general and conventional firms. It demonstrates that general firms financially outperform conventional firms. Column (c) indicates the empirical results of productivity comparisons between non-conventional and conventional firms. It indicates that non-conventional firms are superior in productivity to conventional firms. Productivity = (total revenues of audit firms-total expenses of the audit firms+ salaries paid to their sole proprietors)/number of employees. *, **, *** Denote one-tailed significance at the 10 %, 5 % and 1 % levels. CONCLUSION This study first examines the financial performance differences for proprietorship audit firms taking varied business strategies. Empirical data are from the 1989-2009 Survey Report of Audit Firms in Taiwan, published by the Financial Supervisory Commission (FSC). One of the main results indicates that non-conventional firms financially outperform general firms, and the latter outperforms conventional firms. This study contributes to the resource-based view of the firm by the following knowledge. In practice, larger audit firms render services to large companies (e.g., Francis, Maydew and Sparks 1999). Proprietorship audit firms serve small and medium-sized enterprises and provide more homogeneous practices due to relatively simple accounting treatments in their clients. When audit market is less competitive, core resources of proprietorship audit firms are expertise and experience accumulated from providing traditional practices. When audit market becomes increasingly competitive,


96

such as the entry of more qualified auditors or establishment of tax agents, the preceding core resources of proprietorship audit firms fade. Proprietorship audit firms expand service scopes into MAS and form new core resources obtained from joint provision of traditional and non-traditional practices. As a result, the core resources concept suggested in the resource-based view of the firm adapts in order to survive and sustain competitiveness. In the past three decades, traditional practice market has become increasingly competitive in the Taiwanese auditing industry due to either an increase in the number of qualified practicing public accountants, cancellation of the audit fee standard, or the tax agent legalization. Table 3 reports that financial performance of proprietorship firms only providing traditional services (conventional firms) is inferior to the other two sub-samples, non-conventional and general firms. This finding suggests that practitioners of proprietorship audit firms, especially the conventional firms, aggressively expand their scope of services into MAS. Banker, Chang, and Natarajan (2005) state that the profitability of audit firms has been sustained in recent years largely by the impact that MAS has had on their productivity. The conventional firms expanding their services to MAS enlarge their revenues, improve their traditional practice productivity, and thereby enhance their financial performance. For years, considerable debate rages among academics, practitioners, regulators, and legislators on the potential conflict of interest that may arise when auditors are also a management advisor to their audit clients. Namely, joint provision of audit service and MAS to the public company audit clients impairs auditor independence. The U.S. Sarbanes-Oxley Act of 2002 poses more stringent restrictions on the types of MAS auditors may perform for their public company audit clients. Proprietorship audit firms are not allowed to provide audit services to public companies by the Taiwanese Securities and Exchange Act. As a result, the problem of auditor independence is relatively trivial for proprietorship audit firms offering MAS. This provides an additional justification for proprietorship audit firms to expand their services into MAS. This study addresses the effects of business strategy on financial performance of proprietorship audit firms. This study uses the OLS regression to test our hypotheses. After controlling other factors affecting financial performance, this study obtains the following main results. First, proprietorship firms only providing MAS (non-conventional firms) financially outperform those providing both traditional practices and MAS (general firms), and the latter financially outperforms those only offering traditional practices (conventional firms). Due to data availability, this study employs a cross-sectional data, which may suffer violations of the assumption of independent observations under the OLS regression model. Additionally, practitioners argue that audit firms, especially small and medium-sized firms, establish coalition with consulting firms to save personal income taxes for partners or sole proprietors. Future studies may extend this study and reexamine the financial performance effects of coalition between audit firms and consulting firms from the income tax saving perspective. REFERENCES Arens, A.A., R.J. Elder and B. Mark (2012) “Auditing and Assurance Services: An Integrated Approach,” NJ: Prentice Hall. Banker, R.D., H. Chang and R. Natarajan (2005) “Productivity Change, Technical Progress, and Relative Efficiency Change in the Public Accounting Industry,” Management Science, vol. 51(2), p. 291-304. Beck, P.J., T.J. Frecka and I. Solomon (1988) “A Model of the Market for MAS and Audit Services: Knowledge Spillovers and Auditor-Auditee Bonding,” Journal of Accounting Literature, vol. 7, p. 50-64. Becker, C.L., M.L. DeFond, J. Jiambalvo and K.R. Subramanyam (1998) “The Effect of Audit Quality on Earnings Management,” Contemporary Accounting Research, vol. 15(1), p. 1-24.


97

Bonner, S.E. and N. Pennington (1991) “Cognitive Processes and Knowledge as Determinants of Auditor Expertise,” Journal of Accounting Literature, vol. 10: 1-50. Bröcheler, V., S. Maijoor and A. Witteloostuijn (2004) “Auditor Human Capital and Audit Firm Survival: The Dutch Audit Industry in 1930-1992,” Accounting, Organizations and Society, vol. 29, p. 627-646. Chen, A.L., R.Y. Chen and W.C. Lee (2002) “The Effect of Passing Rate of CPA Examination on the Industrial Structure of Accounting Firms in Taiwan,” PanPacific Management Review, vol. 5(2), p. 155-170. Chen, Y.S., B.G. Chang and C.C. Lee (2008) “The Association between Continuing Professional Education and Financial Performance of Public Accounting Firms,” The International Journal Human Resources Management, vol. 19(9), p. 1720-1737. Collins-Dodd, C., I.M. Gordon and C. Smart (2004) “Further Evidence on the Role of Gender in Financial Performance,” Journal of Small Business Management, vol. 42(4), p. 395-417. Cowling, K. and M. Waterson (1976) “Price-Cost Margins and Market Structure,” Econometrica, vol. 43(1), p. 267-274. Elder, R.J., M. Beasley and A. Arens (2008) “Auditing and Assurance Services: An Integrated Approach,” NJ: Prentice Hall. Fasci, M.A. and J. Valdez (1998) “A Performance Contrast of Male- and Female-Owned Small Accounting Practices,” Journal of Small Business Management, vol. 36(3), p. 1-7. Francis, J.R., E.L. Maydew and H.C. Sparks (1999) “The Role of Big 6 Auditors in the Credible Reporting of Accruals,” Auditing: A Journal of Practice and Theory, vol. 18(2), p. 17-34. Grotelueschen, A.D. (1990) “The Effectiveness of Mandatory Continuing Education for Licensed Accountants in Public Practice in The State of New York,” special report by the Mandatory Continuing Education Study Committee, New York State Board for Public Accountancy, Albany, NY: New York State Education Department. Hunger, J.D. and T. L. Wheelen (2001) “Essentials of Strategic Management,” NJ: Prentice-Hall. Istvan, D.F. (1984) “The Future of The Accounting Profession: Will Your Firm Survive Until 1990?” The Practical Accountant, vol. 17 (April), p. 71-74. Iyer, V. and G.S. Iyer (1996) “Effect of Big 8 Mergers on Audit Fees: Evidence from the United Kingdom,” Auditing: A Journal of Practice and Theory, vol. 15(2), p. 123-132. Luo, Y., J. Tan and O. Shenkar (1998) “Strategic Responses to Competitive Pressure: The Case of Township and Village Enterprises In China,” Asia Pacific Journal of Management, vol. 15, p. 33-50. McMeeking, K.P., K.V. Peasnell and P.F. Pope (2007) “The Effect of Large Audit Firm Mergers on Audit Pricing in the UK,” Accounting & Business Research, vol. 37(4), p. 301-319. Miles, I., N. Kastrinos, K. Flanagan, R. Bilderbeek, P. Hertog, W. Huntink and M. Bouman (1995) “Knowledge-Intensive Business Services: Users, Carriers, and Source of Innovation,” Luxembourg: EIMS Publication No.15, Innovation Program, Directorate General for Telecommunications, Information Market and Exploitation of Research, Commission of the European Communities. Miles, R.E., C.C. Snow, A.D. Meyer and H.J. Coleman. (1978) “Organizational Strategy, Structure and


98

Process,” The Academy of Management Review, vol. 3(3), p. 546-562. Miles, R.E. and C.C. Snow (1984) “Designing Strategic Human Resources System,” Organizational Dynamics, vol. 13(1), 36-52. Mintzberg, H. (1990) “Mintzberg on Management: Inside Our Strange World of Organizations,” NY: The Free Press. Minyard, D.H. and R.H. Tabor (1991) “The Effect of Big Eight Mergers on Auditor Concentration,” Accounting Horizons, vol. 5(4), p. 79-90. Porter, M.E. (1990) “The Competitive Advantage of Nations,” NY: The Free Press. Rescho, J.A. (1987) “Public Accounting Firm Strategy and Innovativeness: A Study of the Adoption of Product, Technical, and Administrative Innovations Using A Trategic Typology,” Ph.D. dissertation, University of Mississippi. Reynolds, J.K. and J.R. Francis (2001) “Does size matter? The Influence of Large Clients on Office-Level Auditor Reporting Decision,” Journal of Accounting and Economic, vol. 30, p. 375-400. Schmidt, F.L., J.E. Hunter and A.M. Outerbridge (1986) “Impact of Job Experience and Ability on Job Knowledge, Work Sample Performance, and Supervisory Ratings of Job Performance,” Journal of Applied Psychology, vol. 71(3), p. 432-439. Simunic, D.A. (1984) “Auditing, Consulting, and Auditor Independence,” Journal of Accounting Research, vol. 22(2), p. 679-702. Sun, H. and C. Hong (2002) “The Alignment between Manufacturing and Business Strategies: Its Influence on Business Performance,” The International Journal of Technological Innovation and Entrepreneurship, vol. 22, p. 699-705. Tan, J. and D. Tan (2005) “Environment-Strategy Co-Evolution and Co-Alignment: A Staged Model of Chinese SOEs under Transition,” Strategic Management Journal, vol. 26(2), p. 141-157. Tang, Z. and J. Tang (2012) “Entrepreneurial orientation and SME performance in China’s changing environment: the moderating effects of strategies,” Asia Pacific Journal of Management, vol. 29(2), p. 409-431. Thomas, C.W., E. Davis and S.L. Seaman (1998) “Quality Review, Continuing Professional Education, Experience and Substandard Performance: An Empirical Study,” Accounting Horizons, vol. 12(4), p. 340-362. Volberda, H.W. and A.H. Lewin (2003) “Co-Evolutionary Dynamics Within and Between Firms: From Evolution to Co-Evolution,” Journal of Management Studies, vol. 40(8), p. 2111-2136. White, H.A. (1980) “Heteroskdeasticity-Consistent Covariance Matrix Estimator and Direct Test for Heteroskdeasticity,” Econometrica, vol. 48, p. 817-838. BIOGRAPHY Yi-Fang Yang is an Assistant Professor of the Department of Accounting and Information System at the Chang Jung Christian University. She has published scholarly articles in International Journal of Human Resources Management, Human Systems Management, and International Research Journal of Finance and Economics. She can be reached at No.1, Changda Rd., Gueiren Dist., Tainan City, 71101, Taiwan,


99

[email protected]. Lee-Wen Yang, corresponding author, is an Assistant Professor of accounting at the Chaoyang University of Technology. She has published scholarly articles in Human Systems Management, Middle Eastern Finance and Economics, International Research Journal of Finance and Economics, and International Journal of Human Resources Management. She can be reached at No. 168, Jifong E. Rd., Wufong Dist., Taichung City, 41349, Taiwan, [email protected]. Yahn-Shir Chen is a Professor of the Department of Accounting at the National Yunlin University of Science and Technology. His research appears in journals such as The International Journal of Human Resources Management, Asia-Pacific Journal of Accounting & Economics, and Economics Bulletin. He can be reached at No. 123, Sec. 3, University Rd., Douliou, Yunlin County, 64002, Taiwan, [email protected].


101

THE ASSOCIATION BETWEEN FIRM CHARACTERISTICS AND CORPORATE FINANCIAL DISCLOSURES: EVIDENCE FROM UAE COMPANIES

Khaled Aljifri, United Arab Emirates University Abdulkareem Alzarouni, National Bank of Abu Dhabi

Chew Ng, Griffith University Mohammad Iqbal Tahir, The University of Faisalabad and Griffith University

ABSTRACT

This paper provides empirical evidence of the impact of firm specific characteristics on corporate financial disclosures amongst UAE companies. A total of 153 public, joint-stock companies, listed and unlisted, were incorporated at the time of study. Both descriptive statistics and multiple regression analyses are used to test the relationship between the characteristics of UAE firms and the extent of their financial disclosure. Eight hypotheses were established to examine the relationship between a number of explanatory variables (namely, type of industry, listing status, return on equity, liquidity, market capitalization, foreign ownership, non-executive directors, and audit committee) and the extent of disclosure in corporate annual reports. The results of this study show that listing status, industry type, and size of firm are found to be significantly associated with the level of disclosure. This finding not only provides support for previous studies, but also is of relevance to those in the UAE who want to understand corporate disclosure and should also be of interest to UAE user-groups. Conclusions drawn from this study may be of interest to policy makers and regulators who want to improve corporate financial disclosure in their countries. JEL: M4, M48, M49 KEYWORDS: Corporate Financial Disclosure, Firm Characteristics, UAE Firms,

Corporate Annual Reports INTRODUCTION

he quality of information disclosed in corporate annual reports has received a great deal of attention in the last four decades, mostly in developed countries. The relationship between the extent/quality of disclosure in corporate annual reports and the characteristics of the firm has been

extensively examined in the literature. Most of the studies in this area have used an index methodology, which is based on developing a general index and relating it to a number of explanatory variables (e.g., asset size, number of shareholders, profitability, listing status) in order to explain cross-sectional variation in the extent of disclosure in such corporate annual reports. It is essential to have high-quality standards and reporting practices to provide users of financial information with what they need (Biobele et. al., 2013). Deficiencies in such standards and practices cause inconsistency, incomparability, reduced transparency and a lack of trust in the information provided, which lead to higher costs of capital and increased risks for different user-groups. As Jenkins (2002, p. 2) stated, ‘High-quality financial reporting is essential to maintaining an efficient capital market system. A highly liquid capital market requires the availability of transparent and complete information so that all participants can make informed decisions as they allocate their capital among competing alternatives’.

T

K. Aljifri et al | IJBFR ♦ Vol. 8 ♦ No. 2 ♦ 2014

102

The above-mentioned perceived benefits accrue to economically advanced nations. However, financial reporting is even more essential for developing countries which seek to build a strong economy by regulating financial practices, protecting the national economy from the control of a handful of influential investors, and encouraging citizens to invest locally. The purpose of this study is to examine the relationship between the extent of disclosure in corporate annual reports and selected firms’ characteristics in The United Arab Emirates (UAE). UAE, which was established in 1971, is a new country that relies heavily on oil as its main source of income. Since its establishment, the UAE has adopted an open economic strategy, and it is one of the fastest growing countries in the world on various socioeconomic indicators, such as GDP per capita (Wikipedia, 2008). The country has witnessed remarkable progress and development in different economic aspects. However, the accounting profession is not well developed (Khasharmeh & Aljifri, 2010). The government of the UAE has, since 1980, examined the potential benefits of establishing an official securities market. The market was established in 2000. Different groups of participants in the UAE securities market (investors, brokers, financial analysts and businessmen) have expressed dissatisfaction with the practice of financial disclosure among UAE firms, and have complained about variations in disclosure. The research problem is, therefore, related to corporate disclosure practices in corporate annual reports. This study seeks to examine the corporate disclosure in annual reports of a sample of UAE firms and to determine the factors responsible for the variation, if any, in financial disclosure. There are few previous studies of UAE financial reporting. Al-Shayeb (2003a) attempted to examine factors that influence the general level of information disclosed by UAE companies in 2000. He found that overall compliance in the UAE was low since none of his sample companies complied with statutory requirements on disclosure. Aljifri (2008) studied the extent of disclosure by public companies listed on the Abu Dhabi Securities Market and the Dubai Financial Market in 2003. Using denominator-adjusted disclosure-indices, he compared the extent of corporate disclosure between companies, sectors, and the two financial markets. The results of his study indicate that significant differences are found between sectors. However, the size, the debt equity ratio, and profitability of a company were found to have no significant association with the level of disclosure. Unfortunately, the sample sizes in both these studies were very small, and so their conclusions about the level of disclosure may not be generalisable. Also the regulations governing disclosure were very new at the time of the studies. Although the findings of this study are specific to the UAE, the results of this research are relevant to other countries in the region with similar socio-economic environments. Conclusions drawn from this study may be of interest to policy makers and regulators who want to improve corporate financial disclosure in their countries. The findings of this study are also likely to benefit researchers and users of annual reports in other parts of the world. The rest of this paper is organized as follows: Section 2 presents the background of the UAE securities market and Section 3 provides a literature review. Research hypotheses are presented in Section 4. Section 5 reviews empirical studies which employ various index methodologies to assess and explain disclosure variations in corporate annual reports. The empirical results are presented in Section 6, and discussion and conclusions are provided in Section 7. Background In the UAE, five forces have shaped financial reporting requirements and practice in the UAE, namely, the Ministry of Economy, the Central Bank of the UAE (CBUAE), Emirates Securities and Commodities Authority (ESCA), Dubai International Financial Centre, and Abu Dhabi Accountability Authority. The Ministry of Economy issued Companies Law No.8/1984 and its amendment No.13/1988, both of which require firms to maintain records of their operations and to provide audited financial statements to the Ministry and to other authorities concerned. The Companies Law and its amendments do not specify any


103

particular standards, format, or information items that should be reported in financial statements. However, Article 190 of the Law states that the board of directors should prepare the company balance sheet, a profit and loss account, a report on company activities during the previous financial year, and the proposal for net profit distribution. The CBUAE issued Provision No.445/1988 which requires financial institutions to prepare their audited financial statements in accordance with the format prepared by the CBUAE. Later, the CBUAE issued Circular No.20/1999 which requires all banks and financial institutions to adopt the IAS/IFRS in their annual reports. Since 1999, all firms reporting to the CBUAE prepare their financial statements in accordance with the IFRS. ESCA was established by Federal Law No. 4/2000 on the 29th January 2000. ESCA requires all listed firms to report their reviewed interim financial statements quarterly as well as their audited financial statements at the end of their financial year. Articles Nos. 29, 31, and 36 of Regulation No. 3/2000 stipulate that listed firms and those applying to be listed have to report to ESCA and to make their financial statements public. Although there is an accounting body in the UAE called the UAE Accountants and Auditors Association (AAA), this association has not issued any national standards and it has no official role in regulating the profession. Hence, the accounting profession is not well-organized and there are no specific professional standards with which UAE firms and auditors must comply. It can therefore be concluded that the legal and regulatory frameworks for financial reporting in the UAE are imprecise and limited in scope. LITERATURE REVIEW Previous Studies and Disclosure Index Methodology Historically, Cerf (1961) was the first researcher who conducted an empirical study using a quantifiable measure of disclosure and relating it to certain financial and non-financial corporate variables. Cerf’s study was based on a sample of 527 US firms listed on the New York Stock Exchange (NYSE), on other exchanges or traded over the counter (OTC). He developed an index consisting of 31 items, each of which was scored on a scale of 1 to 4 on the basis of interviews with financial analysts. The index was then related to four corporate variables. He found a significant positive correlation between the level of disclosure and a firm’s asset size for firms that were not listed on the NYSE, and between the level of profitability and disclosure for firms listed on the NYSE and those traded OTC. He also found that firms listed on the NYSE disclosed more information than other firms. Cerf’s (1961) approach, with extensions and modifications, has been used widely in many other studies to examine the adequacy of corporate financial disclosure in different countries. Studies using disclosure index methodology can be classified into three groups: those in developed countries, those in developing countries, and international studies where data from several countries was included. However, the present literature survey in the present study is restricted to developing countries and especially to those countries that have similar socio-economic environments. Studies in Developing Countries Using a general disclosure index, previous studies suggest that the extent of corporate financial disclosures is a function of financial and non-financial characteristics of firms (Imhoff, 1992; Malone et al., 1993; Lang & Lundholm, 1993; Wallace et al., 1994; Inchausti, 1997; Cooke, 1989a, 1989b, 1989c; Patton & Zelenka, 1997; Priebjrivat, 1991; Abu-Nassar, 1993; Suwaidan, 1997; Hooks et al., 2002; Naser & Nuseibeh, 2003; Prencipe, 2004; Alsaeed, 2006; Aljifri, 2008, Hossain and Hammami, 2009; Bhayani,


104

2012; Ahmed, 2012). While some studies found that firm size, listing status, leverage, and industry type were significantly associated with higher disclosure levels, results for other variables (profitability, size of audit firm, and liquidity) were inconclusive. These findings could be attributed to differences in socio-economic and political environments between countries, organizational structure, construction of disclosure indices, and sampling error (Cooke & Wallace, 1990; Ahmed & Courtis, 1999). In Saudi Arabia, Abdel-Salam (1985) investigated the relationship between the extent of disclosure and some specific corporate variables. He found a negative results with respect to size of firm measured by either capital or assets. For the other variables (growth, government subsidy, government ownership, audit firm size) the results were not clear. In Jordan, in a study of 45 Jordanian firms, Solas (1994) found that firm size, number of shareholders, rate of return, and earnings margin had no significant relationship with the quality of financial reporting. These results contradict Abu Nassar’s (1993) study, which investigated the relationship between the level of disclosure of 96 firms listed on the Amman Stock Exchange and seven corporate variables. He used five models of regression analysis to overcome the problem of multicollinearity between the independent variables. The results revealed that among the independent variables, total dividends were found to be the most important influence on disclosure. However no relationship was found with equity ratio or the number of shareholders. In Bangladesh, Ahmed and Nicholls (1994) investigated the extent of corporate compliance with local disclosure requirements. By developing an index and applying it to 63 firms listed on the Bangladesh Stock Exchange, the researchers found no significant association between firm size and the level of disclosure. However, they reported a positive and significant relationship with the status of firms as subsidiaries of multinational firms. Hannifa and Cooke (2002) examined whether the extent of voluntary disclosure in annual reports of 167 Malaysian listed firms was associated with 31 corporate characteristics, divided into three groups of variables: corporate governance, cultural and firm specific (control) variables. A scoring sheet of 65 voluntary disclosure items, selected on the basis of previous research, was developed and applied to the annual reports of the selected sample. Using regression analysis, the results indicated a significant association between the extent of voluntary disclosure and two corporate governance variables (chair who is a non-executive director and domination of family members on boards) and with one cultural variable (proportion of Malay directors on the board). Naser et al. (2002) investigated changes in corporate disclosure in Jordan after the introduction of the International Accounting Standards (IAS). The results, applying regression analysis, indicated a slight improvement in the depth of disclosure after the introduction of IAS. In addition, the depth of disclosure was found to be associated with corporate size, audit firm status, liquidity, gearing, and profitability. In another study, Naser and Nuseibeh (2003) tried to assess the quality of information disclosed, in the years 1992-1999, by a sample of non-financial Saudi firms listed on the Saudi Stock Exchange. The researchers used two indices (weighted and un-weighted). The results indicated a relatively high level of compliance with mandatory requirements in all industries except the electricity sector. However, the level of voluntary disclosure was relatively low. Alsaeed (2005) also examined the effect of specific characteristics on the extent of voluntary disclosure by a sample of 40 Saudi firms. He reported that while large firms tend to present more voluntary information than small firms, the other characteristics (debt ratio, ownership dispersion, firm age, profit margin, return on equity, liquidity, industry type, and audit firm size) were found not to be significantly associated with the level of disclosure. Al Zoubi and Al Zoubi (2012) examined the opinions of accounting academics and investors on the adequacy of the quality and quantity of information disclosed by Jordanian listed companies in the


105

circumstances of the global financial crisis. They used a sample of the two groups of respondents (i.e., academics and investors), consisting of 90 respondents from each category. The analysis of the data gathered by questionnaire revealed that while accounting academics perceived the quantity of disclosed information to be sufficient, investors perceived the quantity of accounting disclosure to be inadequate. Research Hypotheses There has been a great deal of empirical work regarding the relationship between firm-specific characteristics and the extent of corporate disclosure. This research has used a variety of theoretical frameworks, such as agency theory, signalling theory, capital market theory and cost-benefit theory (Haniffa & Cooke, 2002). While the characteristics examined may be classified into various categories, they are not mutually exclusive. In this study, the characteristics of a company were divided into four categories (market-related, performance-related, structure-related, and corporate governance variables, see Table 1) to explain the relationship between company characteristics and their disclosure in the UAE. These characteristics (i.e., the variables) were selected on the basis that they met the following three preconditions: (1) the variable encompasses sound theoretical reasons for explaining the association between the variable and corporate disclosure, (2) the variable is relevant to the socio-economic environment of the UAE; and (3) sufficient data about the variable was available. Based on these criteria, eight firm-specific variables were selected for this research: (1) industry type, (2) listing status, (3) profitability, represented by return on equity ratio (ROE), (4) liquidity, (5) firm size, (6) foreign ownership, (7) composition of board of directors, and (8) audit committee. The theoretical and empirical support for these variables are discussed in the following subsections. Industry Type Accounting policies and techniques may vary between firms because of their industry-specific characteristics. Firms from a particular industry may adopt disclosure practices that differ from firms in other industries (Wallace 1989; Dye & Sridhar 1995). Some industries are highly regulated because of their overall contribution to a country’s national income. These industries are subject to more rigorous control, which may affect the level of disclosure (Owsus-Ansah, 1998b). A disclosure differential may also be associated with the scope of business operations. Firms with multi-production lines may have more information to disclose than those with small or single line production (Owsus-Ansah, 1998b). Finally, a dominant firm with a high level of disclosure within a particular industry may lead other firms to “follow the leader” (Belkaoui & Kahl, 1978) in that industry to adopt the same level of disclosure (Wallace & Naser, 1995). Therefore a positive association can be assumed between the industry type and the extent of disclosure: H1: There is a significant association between the type of industry and the extent of disclosure. Listing Status The level of disclosure may vary between listed and unlisted firms. Not only do firms have to comply with the listing rules imposed by the securities market in which they are listed (e.g., Leftwich et al., 1981; Cooke 1989a; Wallace et al. 1994), but they also seek funds and hope to reduce the cost of capital by disclosing more information (Cooke 1989a). Moreover, listed firms are much more in the public eye (Al-Mulhem, 1997) than unlisted ones, so they tend to disclose more information. It can be assumed that there is an association between listing status and increased disclosure: H2: Firms listed in the UAE securities market disclose more information than unlisted firms.


106

Profitability Profitability has been used to explain the variation of disclosure between firms. When profitability is high, management is more willing to disclose detailed information (Inchausti, 1997; Lang & Lundholm, 1993; Wallace & Naser, 1995; Suwaidan, 1997). Unprofitable firms will be less inclined to release more information to hide their poor performance. There are different measures of profitability such as net income, profit margin, return on assets, and return on equity. In this study return on equity was chosen as a proxy for profitability. Hence, it is hypothesised that return on equity is associated with the extent of disclosure: H3: Firms in the UAE securities market with a higher return on equity disclose more information than firms with lower return on equity. Liquidity The assessment of a firm’s liquidity is an important issue for those who use financial statements to judge a firm’s solvency. Liquidity is of interest to regulatory bodies as well as to investors and lenders. The inability of a firm to meet its current obligations may mean a default in payment of interest and principal to the lenders and may lead to bankruptcy. To alleviate these fears, firms are willing to disclose more information (Wallace & Naser, 1995). Also, liquidity is perceived to be associated with a strong financial position and firms with high liquidity ratios are expected to disclose more information (Belkaoui & Kahl, 1978; Cooke, 1989a, 1989b; Wallace et al., 1994). Therefore, an association between liquidity, as measured by current assets divided by current liabilities, and the extent of financial disclosure is hypothesised as follows: H4: Firms in the UAE securities market with high liquidity tend to disclose more information than firms with low liquidity. Firm Size In the literature, size has been found to be an influential variable in explaining differences in disclosure practices among firms (Cerf 1961; Singhvi & Desai, 1971; Buzby 1974a; Lang & Lundholm, 1993; Wallace et al., 1994; Zarzeski 1996; Naser, 1998; Archambault & Archambault, 2003). There are several reasons for a positive association between firm size and the extent of disclosure. Disclosing detailed information is costly, and thus may not be affordable for small firms. Large firms are usually diverse in the scope of their business, the types of products and geographical coverage. A considerable amount of information is required for management purposes and can be generated internally. Consequently, the marginal cost of disclosing the information publicly is low (Cooke 1989c). Also, large firms go to financial markets to raise funds more often than small ones. These large firms are aware that selling new securities and a low cost of capital depend on disclosing more information to users (Choi, 1973a, 1973b; Spero, 1979; Dhaliwal, 1979, 1980b; Barry & Brown, 1985). On the other hand, disclosure of detailed information may place small firms at a competitive disadvantage with other large firms in the same industry (Buzby, 1975). Different variables have been used in previous studies as proxies for firm size, including total assets, market capitalization, and net sales. (Wallace & Naser, 1995; Naser et al., 2002; Hanifa & Cooke, 2002). In this research, market capitalization is chosen as it is more objective than other variables and is an externally determined measure, set by choices that are made by the investing public (Wallace & Naser, 1995). A positive association is expected between a firm’s size (measured by market capitalisation) and the extent of disclosure:


107

H5: Firms in the UAE securities market with a large market capitalization tend to disclose more information than firms with small market capitalization. Foreign Ownership Based on agency theory, where there is a separation between owners (shareholders) and management of a firm, the potential for agency costs arise because of conflicts of interest between the principal and the agent (Jensen & Meckling, 1976; Watts, 1977; Fama & Jensen, 1983; Chau & Gray, 2002). Shareholders will be more inclined to increase monitoring of management behaviour in order to alleviate the agency problems. Monitoring costs affect both profitability and management remuneration, and consequently management can reduce monitoring costs by providing more information to shareholders.

Shareholders, according to Wallace (1989), also vary in their information needs. Some shareholders could be interested in profitability, others might be looking for forecast information, while others want information about social responsibility. Accordingly, corporate financial disclosure is likely to be higher in widely-held firms and in this sense the demand for information is expected to be higher from foreign investors due to the geographical separation between management and owners (Bradbury, 1992; Craswell & Taylor, 1992). Diffusion of ownership has been empirically found to be an important variable in explaining the variability of corporate financial disclosure (Leftwich et al., 1981; Craswell & Taylor, 1992; Hossain et al., 1994), and the demand for information is expected to be greater when a high proportion of shares is held by foreign investors. Therefore, it is hypothesized that: H6: UAE firms with a higher proportion of foreign ownership disclose more information than those without such ownership. Composition of Board of Directors Board composition is defined by Shamser and Annuar (1993, p. 44) as ‘the proportion of outside directors to the total number of directors’. The role of the board of directors in monitoring management behaviour and corporate financial disclosure may be a function of the composition of the board (Gibbins et al., 1990). Having a higher proportion of outside non-executive directors on the board may result in better monitoring of the behaviour of management by the board and limit managerial opportunism (Fama, 1980; Fama & Jensen, 1983). Also, non-executive board members are less aligned with management, and they may be more inclined to encourage and support more disclosure to the users of financial reporting (Mak & Eng, 2003). A positive relationship between the proportion of independent directors and disclosure has been found empirically in other capital market settings (Chen & Jaggi, 2000). Therefore, it is hypothesised that: H7: There is a positive association between the proportion of outside directors and the level of disclosure made by UAE firms. Audit Committee Recent high-profile accounting scandals, such as that involving Enron, have shed light on the effectiveness of audit committees. Such problems have led some countries to impose more regulations on audit committee functions, including independence, composition, expertise and disclosure activities (e.g., Sarbane-Oxely Act 2002). The structure of the audit committee determines the level of monitoring and thereby the level of financial disclosure. It has been argued that an effective and independent audit committee has an influential role in the financial reporting process (Kreutzfeldt & Wallace, 1986; Abbott


108

et al., 2004). An independent audit committee provides greater monitoring of the financial discretion of management and ensures the credibility of corporate financial disclosure (The Blue Ribbon Committee, 1999). The level of expertise is another factor which enhances the effectiveness of the audit committee. This expertise should be based on members of the audit committee who possess knowledge and experience in accounting and finance (Beasley & Salterio, 2001; Abbott et al., 2004). The US (Sarbane-Oxely Act 2002) and Malaysia (Bursa Malaysia listing requirements 2001) require that at least one member of the audit committee possesses a background in accounting and finance. Hence, it is assumed that having independent and qualified audit committee members enhances the quality of a firm’s financial disclosure. In this research, since information on the independence and qualifications of audit committee members is not available in the UAE, the effect of the existence of an audit committee on the extent of corporate disclosure is assessed by testing whether a firm has an audit committee or not. Therefore, it is hypothesized that: H8: There is a positive association between the existence of an audit committee and the level of financial disclosure made by UAE firms. Table 1 presents the research hypotheses and the predicted signs for each explanatory variable associated with each hypothesis. Table 1: Research Hypotheses and Predicted Signs of the Coefficients

Independent Variable Hypothesis Expected Sign Market Variables:

Industry Type H1 +/- Listing Status H2 +

Performance Variables: Profitability (ROE) H3 +

Liquidity H4 +

Structure Variables: Firm Size (Market Capitalization) H5 +

Foreign Ownership H6 +

Corporate Governance Variables: Composition of B.O.D. H7 +

Audit Committee H8 +

This table lists the hypotheses with their expected signs. It shows that market variables (with an exception for industry type), performance variables, structure variables, and corporate governance variables are predicted to have a positive relationship with the level of financial disclosure made by UAE firms. METHODOLOGY Selection of Information Items The most important step in constructing a disclosure index is the selection of information items that could be found in corporate annual reports. It should be noted that there is no consensus on the number or selection of items to be included in a disclosure index (Wallace et al., 1994; Al-Hussaini, 2001). Also, the number of information items used in previous studies has varied considerably, which may reflect differences in the settings where the studies were conducted (Wallace, 1993; Patel, 2003; Ngangan et al., 2005).


109

As Marston and Shrives (1991) noted, the number of items that could be disclosed by a company is very large, if not infinite. Wallace (1988) argued that there was no agreed criterion on which to select an item of information, and that to overcome the selection bias, an extensive list of disclosure items be developed. The number of items included in the scoring sheet in previous studies varied from a minimum of 17 items (Barrett, 1976) to a maximum of 530 (Craig & Diga, 1998). The disclosure index may include both mandatory and voluntary information items since both forms of disclosure result from social-system processes (Archambault & Archambault, 2003). Mandatory disclosure is required by statute, professional regulations and listing requirements of stock exchanges. The extent to which firms comply with legal and regulatory requirements depends on the strictness or laxity of the government, professional and other regulatory bodies (Marston & Shivers, 1991; Salawu, 2012). Voluntary disclosure on the other hand, in excess of the minimum, may arise where corporate perceptions of the benefits arising outweigh the costs (Gray & Roberts, 1989; Chakroun & Matoussi, 2012). In this research, the focus is on mandatory items because financial reporting and disclosure practice in the UAE are not well-organized and free-market mechanisms are immature (Owsus-Ansah, 1998a). In order to avoid penalizing a firm for not disclosing an item that does not apply to it, the list of items was based on the limited and specific requirements set by the UAE regulators, in addition to IFRS, to which all firms in the UAE claim to comply. The initial list of information items (index) consisted of 405 items, of which only six items were related to ESCA requirements. To ensure that a complete, relevant and applicable list of information items was included in the index, two control measures were adopted. First, the list was cross-checked with the disclosure checklists of three of the big-four accounting firms: KPMG, Ernst and Young and Deloitte. Second, the list was discussed with three senior external auditors working for three big auditing firms in the UAE (Deloitte, KPMG, and Talal Abu Ghazala). The purpose of this step was to refine the list and to determine the suitability of the items for firms operating in different sectors in the UAE. Based on these measures, 88 items were excluded as they related to 9 standards that were irrelevant or inapplicable to the UAE or the dates when they came into effect were after 2005 (see Table 2). These standards were: IFRS 1 First-time Adoption of IFRS; IFRS 6 Exploration for and Evaluation of Mineral Assets; IFRS 7 Financial Instruments Disclosure; IFRS 8 Operating Segments; IAS 12 Income Taxes; IAS 26 Accounting and Reporting by Retirement Benefit Plan; IAS 29 Financial Reporting in Hyperinflationary Economies; IAS 34 Interim Financial Reporting; and IAS 39 Financial Instruments Recognition and Measurement (see Table 2 for the reasons for exclusion). Table 2: Excluded IAS & IFRS Items

IFRS/IAS Title Reasons for Exclusion

IFRS 1 First-time Adoption of IFRS All firms claimed adoption of IFRS before 2004 IFRS 6 Exploration for and Evaluation of Mineral Assets It is effective on January 2006 onward IFRS 7 Financial Instruments: Disclosures It is effective on January 2007 onward IFRS 8 Operating Segments It is effective on January 2009 onward IAS 12 Income Taxes UAE firms pay no income tax IAS 26 Accounting and Reporting by Retirement Benefit Plan The UAE has its own benefits and retirement plan

IAS 29 Financial Reporting in Hyperinflationary Economies Inflation has been relatively low (1.5% to 4%) over the past ten years (GCC 2003)

IAS 34 Interim Financial Reporting The objective of this research is to assess disclosure in corporate annual reports

IAS 39 Financial Instruments: Recognition and Measurement All disclosure requirements in IAS 39 were covered in IAS 32

This table reports the excluded IFRS/IAS with the reason of exclusion. These standards were excluded as they that were irrelevant or inapplicable to the UAE or the dates when they came into effect were after 2005.


110

The final list included a total of 317 items of information. This final list was then pilot tested on 20 companies from different sectors to ensure that items peculiar to each selected sector were taken into account. Sample Selection, Data Collection, and Variables Measurements Based on the constitution of the UAE, the Ministry of Economy is responsible for the orderly operation of the country’s economy. Companies Law No. 8/1984 is the main act which governs the incorporation, control, and management of different types of firms. There are seven types of firms that may be incorporated in the UAE: general partnership, simple limited partnership, joint participation (venture); public joint stock company, private joint stock company, limited liability company, and partnership united with shares. All public joint stock firms have to lodge their annual reports with the Ministry. Therefore, the Companies Department at the Ministry of Economy was approached to provide copies of the 2005 annual reports of all public joint stock companies as this was the best and quickest way to obtain the required data. The year 2005 was selected as it was the most recent data available at the time of the request. A total of 153 public joint stock companies, listed and unlisted, were incorporated at the time of study. Seven firms were excluded from the sample as they were incorporated in either 2005 or later and had very little information in their annual reports. Another 33 firms were excluded because their annual reports were not available. Twenty of these 33 companies were solely owned by the UAE government, and while they had been registered as joint stock companies, their annual reports were not accessible. For the other 11 companies, annual reports were not available to the researchers, despite numerous efforts to obtain them. Hence, they were also excluded from the sample. Consequently, only 113 corporate annual reports were collected representing approximately 75% of the total population. Table 3 provides a summary of the total sample. The dependent variable, total disclosure index (TDI), for each firm was the disclosure made by the firm through its annual report and it was measured by the total disclosure index (TDI) as explained in in the next subsection. Data for these variables was obtained from the annual reports of the selected firms, and from the annual guide of firms published by ESCA. Table 4 summarises data collection and variables measurement. Table 3: Population and the Sample of Public Joint Stock Companies in the UAE

Sector Companies Annual Reports Available/Not Available Total Available

% Listed Unlisted Total Listed Unlisted

Available Established late 2005

Not Available

Available Established Late 2005

Not Available

Banking 22 3 25 22 0 0 3 0 0 25 100% Insurance 23 1 24 23 0 0 1 0 0 24 100% Services 36 32 68 34 2 0 0 4 28 34 50% Industries 27 9 36 27 0 0 3 1 5 30 83% Total 108 45 153 106 2 0 7 5 33 113 74% This table presents a summary of the population and the sample of Public Joint Stock Companies in the UAE. A total of 153 public joint stock companies, listed and unlisted, were incorporated at the time of study, however, only 113 corporate annual reports were collected representing approximately 75% of the total population


111

Table 4: Data Collection and Measurement of Variables

Independent Variable Proxy Sources of Information

Market Variables: Industry Type Dummy variable coded: Industry 1(Banking) = 1, otherwise = 0

Industry 2 (Insurance) = 1, otherwise = 0 Industry 3 (Services) = 1, otherwise = 0 Industry 4 (Industrial) = 1, otherwise = 0

ESCA

Listing Status Dummy variable coded 1 = listed, 0 = unlisted

ESCA

Performance Variables: Profitability (ROE) Return on Equity = earnings/total shareholders' equity Firms’ Annual

reports

Liquidity Current assets/ current liabilities Firms’ Annual reports

Structure Variables: Firm Size (Market Capitalization)

Number of shares at year-end date X market value per share ESCA

Foreign Ownership Percentage of non-UAE national investors’ ownership ESCA Corporate Governance Variables: Composition of B.O.D. Proportion of outside directors to the total number of directors ESCA Audit Committee Dummy variables coded: 1 = if a company has an audit committee, 0 = otherwise ESCA

This table reports data collection of the independent variables and their proxies. Data for these variables was obtained from the annual reports of the selected firms, and from the annual guide of firms published by ESCA. Statistical Procedures One of the methodological problems associated with the scoring procedures of the disclosure index is whether or not an item is applicable to a particular firm (Owsus-Ansah, 1998a). To overcome this problem, following Cooke’s (1989a) recommendation, the entire annual report of every company was read first by the researcher to understand clearly the scope of the disclosure practice and to determine whether an undisclosed item was in fact applicable to that particular company. This procedure led to the creation of a relative index for each sample firm. The relative index, which includes information items a firm was expected to disclose, was adopted by several previous studies (Buzby, 1975; Wallace, 1987; Firth, 1980; Babbie, 1994; Ahmed & Nicholls, 1994; Inchausti, 1997; Owsus-Ansah, 1998a, Aljifri, 2008). Consequently, the risk of penalising a firm for not disclosing an item, inapplicable to that firm, was significantly reduced. The next step was to apply the index to each firm using the dichotomous procedure of matching the firm’s annual report disclosures to the index. Following the unweighted index scoring scheme, an item scored one if it was disclosed and zero if it was not. Based on the calculated scores for each firm, the descriptive statistics were calculated to form a judgement about the current level of disclosure in the UAE. The scores of the disclosure index for each firm were calculated by dividing the total scores by the maximum score (M) (based on the relative index of the firm), as follows:

∑=

=n

idiM

1

(1)

where,

id = expected item of disclosure n = the number of items applicable to a firm, i.e. 317≤n

The total disclosure score (TD) for a firm was calculated as follows:


112

∑=

=M

idiTD

1

(2)

where:

id = 1 if the item id is disclosed

0 if the item id is not disclosed

nM ≤

The total disclosure index (TDI) for each firm was therefore calculated as:

TDI = TD (3)

Statistical Tests Different statistical approaches and methodologies have been adopted in previous studies to test the relationship between the extent of disclosure and various firm-specific variables. Earlier studies (Singhvi & Desai, 1971; Buzby, 1975; Stanga, 1976; Belkaoui & Kahl, 1978; Buckland et al., 1998) used a matched-pair statistical procedure to test for difference between the mean disclosure indexes of two or more groups of sample firms. More recent studies, starting with Chow and Wong-Boren (1987), have used a variation of a multiple regression procedure. For example, some researchers use dummy variable manipulation procedures within a stepwise multiple Ordinary Least Squares (OLS) regression analysis (Cooke, 1989a, 1989b; Haniffa & Cooke, 2002), while others use rank regression procedures (Land and Lundholm, 1993; Wallace et al., 1994; Naser, 1998; Chen & Jaggi, 2000; Chau & Gray, 2002) or a meta-analysis technique (Ahmed & Courtis, 1999). Wallace et al. (1994), Haniffa and Cooke (2002), and Archambault and Archambault (2003) used two regression models, reduced regression and full regression, to deal with the possibility of collinearity. In this research, both descriptive statistics and multiple regression analyses are used to test the relationship between the characteristics of UAE firms and the extent of their disclosure. Descriptive Statistics Two types of descriptive statistics were conducted. The first was a descriptive analysis which includes the mean and standard deviation. The second test was a correlation test which highlights the relationship between a single explanatory variable and the extent of disclosure by UAE firms. Multiple Regression Analysis One of the main problems that confronts researchers using the OLS regression is when the dependent variable is constrained to range between zero and one. The estimation of the regression model can, however, lie outside this range. This is because the standard OLS presumes an unconstrained dependent variable. Hence, the standard OLS estimates may be unreliable (Hanushek & Jackson, 1977; Ahmed & Nicholls, 1994; Greene, 2003). Consequently, the suggestion to transform the dependent variable was adopted (Hanushek & Jackson, 1977; Wallace e. al., 1994; Ahmed & Nicholls, 1994; Fox, 1997; Inchausti 1997; Naser & Al-Khatib, 2000; Greene, 2003; Makhija & Patton, 2004; Al-Shaimmari 2005). This approach also has the advantage that such transformation might result in normally distributed errors (Cooke, 1998). The transformation was done by taking the logarithm of the total disclosure index for each


113

firm. Then, the transformed total disclosure index (TTDI) was regressed against the eight specified variables by applying OLS regression procedures. The regression model is expressed as follows:

TTDI = ∝ + β1 indtypei + β2 indtypei + β3 indtypei + β4 indtype + β5 listingi + β6 ROEi + β7 liquidi + β8

marcapi + β9 foriegni + β10 boardi + β11 auditi + ε (4)

where: TTDI = Transformed Total disclosure index (actual score awarded to each

firm/maximum score) ∝ = Intercept indtype = Industry type (indtype1 = Banking; indtype2 = Insurance; indtype3 = Services; indtype4 =

Industrial) listing = Listing status ROE = Return on equity liquid = Liquidity marcap = Market capitalization foreign = Foreign ownership board = Board composition Audit = Audit committee. ε = Error term β1, β2, β3, β4, β5, β6, β7, β8, β9, β10, β11, β12, = slope coefficients of the model.

RESULTS Descriptive Analysis Table 5 provides descriptive analysis for the dependent variable and the explanatory variables. In general, the extent of corporate disclosure varies from 23% to 70%. The overall mean value of disclosure is 57% with standard deviation of 9%, reflecting a low to moderate level of disclosure among the 113 sample firms. Table 5: Descriptive analysis of the Dependent and Independent Variables

Variables N Mean Std. Deviation Minimum Maximum Total Disclosure 113 0.57 0.09 0.23 0.70 Listing Status 113 0.94 0.24 0.00 1 Net Income/Equity (ROE) 113 0.21 0.13 - 0.01 0.55 Liquidity 113 3.85 8.66 0.39 86.33 Market Capitalization (million dirham) 113 7,894 17,836 20 133,811 Foreign Ownership 113 0.26 0.32 0.00 100.00 % of outside directors 113 0.93 0.10 0.60 100.00 Audit Committee 113 0.53 0 1 Banks 25 0.22 0 1 Industrial 30 0.27 0 1 Insurance 24 0.21 0 1 Services 34 0.30 0 1

This table provides a descriptive statistics of the dependent and independent variables. It shows the extent of corporate disclosure varies from 23% to 70%. The overall mean value of disclosure is 57%.


114

An examination of Table 5 shows that 94% of the sample companies are listed on the UAE securities market and the mean value of return on equity (ROE) for the 113 companies is 21%. The overall liquidity (current ratio) is 3.85 which is well above the rule of thumb of 2:1. The mean value of foreign ownership among the firms is 26% while non-executive directors constitute about 93% of the sampled boards. Fifty-three percent of the firms have audit committees, which indicates that good corporate governance is not widely practised and that further enforcement is needed from the UAE authorities. Correlation Analysis To assess the relationship between the total disclosure index (TDI) and the characteristics of the firms, a Pearson Product-Moment correlation matrix was used to examine the correlation between the dependent variable (TDI) and each of the independent variables used in this study. The Pearson Product-Moment correlation matrix for the dependent and independent variables is presented in Table 6. The statistical results show that significant positive relationships were found to exist between the extent of disclosure and banking industry (r = 0.347) and listing status variables (r = 0.288) at the P < 0.01 level. These results support hypotheses H1, which states that there is a significant relationship between corporate disclosure and industry type, and H2, which states that there is a significant relationship between corporate disclosure and listing status. However, no significant correlation was found between other industry types and the extent of disclosure. No other statistically significant correlations were found. Return-On-Equity (r = -0.074, P = 0.434) and liquidity (r = -0.001, P = 0.992) appear to be negative but not influentially correlated with the extent of disclosure. The most interesting result is that size measured by market capitalization was found to be negative but not significantly correlated with corporate disclosure (r = -0.028, P = 0.772). On the other hand, the relationship between the extent of disclosure and foreign ownership (r = 0.079, P = 0.408), composition of board of directors (r = 0.107, P = 0.258), and audit committee (r = 0.111, P = 0.241) was found to be positive but also not significantly correlated. These results do not appear to support hypotheses H3, H4, H5, H6, H7, and H8. Multiple Regression Analysis This subsection describes the results of running the ordinary least square (OLS) regression with log transformation analysis with all company variables using SPSS. All variables were entered into the model simultaneously. The purpose is to test whether the specified independent variables contribute significantly to the prediction of the disclosure level of firms in the UAE. The results of the multiple regression analysis are presented in Table 7. Results of the multiple regression analysis of the association between company variables and the extent of disclosure in the annual reports of the sample companies are shown in Table 7. As can be seen, the coefficient of determination R2 is equal to 33% and the adjusted R2 is equal to 26% where the P–value < 0 .01 and the F test statistics (10, 102) = 4.95. Table 6: Correlation Analysis

Tota

l D

iscl

osur

e

Listi

ng S

tatu

s

RO

E

Liqu

idity

Mar

ket

Cap

italiz

atio

n

Fore

ign

Ow

ners

hip

% o

f out

side

dire

ctor

s

Aud

it C

omm

ittee

Ban

ks

Indu

stria

l

Insu

ranc

e

Serv

ice

Total Disclosure

Person Correlation 1 0.288* -0.074 -0.001 -0.028 0.079 0.107 0.111 0.347* -0.115 -0.091 -0.122

Sig. (2 tailed)

0.002 0.434 0.992 0.772 0.408 0.258 0.241 0.000 0.224 0.339 0.200

N 113 113 113 113 113 113 113 113 113 113 113 113


115

Listing Status

Person Correlation 0.288 1 0.198* 0.042 0.110 0.143 0.053 0.053 -0.128 -0.095 0.044 0.169

Sig. (2 tailed) 0.002

0.035 0.661 0.245 0.130 0.578 0.579 0.175 0.317 0.646 0.074

N 113 113 113 113 113 113 113 113 113 113 113 113

ROE Person Correlation -0.074 0.198* 1 -0.142 0.056 -0.064 -0.139 0.107 -0.03 -0.184 .291* -0.055

Sig. (2 tailed) 0.434 0.035

0.132 0.559 0.499 0.142 0.259 0.751 0.051 0.002 0.565

N 113 113 113 113 113 113 113 113 113 113 113 113

Liquidity Person Correlation -0.001 0.042 -0.142 1 -0.044 0.090 -0.002 -0.015 -0.136 -0.020 -0.049 0.186*

Sig.(2 tailed) 0.992 0.661 0.132

0.643 0.343 0.982 0.877 0.152 0.837 0.605 0.049

N 113 113 113 113 113 113 113 113 113 113 113 113

Market Capitalization

Person Correlation -0.028 0.110 0.056 -0.044 1 0.019 -0.033 0.282* 0.197* -0.169 0.186* 0.150

Sig. (2 tailed) 0.772 0.245 0.559 0.643

0.839 0.729 0.003 0.036 0.074 0.048 0.113

N 113 113 113 113 113 113 113 113 113 113 113 0.113

Foreign Ownership

Person Correlation 0.079 0.143 -0.064 0.090 0.019 1 -0.127 -0.044 -0.183 0.103 -0.270* 0.306*

Sig. (2 tailed) 0.408 0.130 0.499 0.343 0.839

0.179 0.647 0.053 0.276 0.004 0.001

N 113 113 113 113 113 113 113 113 113 113 113 113

% of outside directors

Person Correlation 0.107 0.053 -0.139 -0.002 -0.033 -0.127 1 0.067 0.082 0.000 0.087 -0.152

Sig. (2 tailed) 0.258 0.578 0.142 0.982 0.729 0.179

0.484 0.387 0.999 0.359 0.108

N 113 113 113 113 113 113 113 113 113 113 113 113

Audit Committee

Person Correlation 0.111 0.053 0.107 -0.015 0.282** -0.044 0.067 1 0.287** -0.238 -0.119 0.075

Sig. (2 tailed) 0.241 0.579 0.259 0.877 0.003 0.647 0.484

0.002 0.011 0.210 0.428

N 113 113 113 113 113 113 113 113 113 113 113 113

Banks Person Correlation 0.347** -0.128 -0.03 -0.136 0.197* -0.183 0.082 0.287* 1 -0.320* -0.277* -0.350*

Sig. (2 tailed) 0.000 0.175 0.751 0.152 0.036 0.053 0.387 0.002

0.001 0.003 0.000

N 113 113 113 113 113 113 113 113 113 113 113 113

Industrial Person Correlation -0.115 -0.095 -0.184 -0.020 -0.169 0.103 0.000 -0.238* 0.320* 1 -0.312*

-0.394*

Sig. (2 tailed) 0.224 0.317 0.051 0.837 0.074 0.276 0.999 0.011 0.001

0.001 0.000

N 113 113 113 113 113 113 113 113 113 113 113 113

Insurance Person Correlation -0.091 0.044 0.291* -0.049 -0.186 -0.270* 0.087 -0.119 -0.277* -0.312* 1

-0.341*

Sig. (2 tailed) 0.339 0.646 0.002 0.605 0.048 0.004 0.359 0.210 0.003 0.001

0.000

N 113 113 113 113 113 113 113 113 113 113 113 113

Service Person Correlation -0.122 0.169 -0.055 0.186* 0.150 0.306* -0.152 0.075 -0.350* -0.394* -0.341* 1

Sig. (2 tailed) 0.200 0.074 0.565 0.049 0.113 0.001 0.108 0.428 0.000 0.000 0.000

N 113 113 113 113 113 113 113 113 113 113 113 113 This table presents the Pearson Product-Moment correlation matrix for the dependent and independent variables. The statistical results show that significant positive relationships were found to exist between the extent of disclosure and banking industry (r = 0.347) and listing status variables (r = 0.288) at the P < 0.01 level. The correlations are for selected variables used in the analysis. *, ** indicate significance at the 1 and 5 percent. levels respectively. These results show that approximately 26% of the variation in disclosure level scores between the companies can be explained by the eight independent variables included in the model. According to Anderson et al. (1993) and Abd-Elsalam and Weetman (2003), an explanatory y power of 20% is considered useful in social science research.


116

The results show that listing and industry type (banking sector) variables were positively and significantly associated with the extent of disclosure of the sample companies at the 1% level. Also, market capitalization was significant but with a negative coefficient at the 5% level. The model indicated that the other variables did not seem to be the main determinants of variations in the extent of disclosure of the sample companies. Table 7: Summary of the Results of OLS Regression with Log Transformation

Variable β t Sig. t VIF

Listing 0.3268 5.03 0.000 1.13

ROE -0.1650 -1.34 0.183 1.24

Liquidity 0.0004 0.23 0.820 1.07

Market Capitalization -1.8500 -2.06 0.042 1.17

Foreign Ownership 0.0850 1.65 0.102 1.21

Non-Executive Directors 0.0622 0.41 0.685 1.09

Audit Committee 0.0174 0.53 0.594 1.21

Banks 0.1939 4.28 0.000 1.62

Industrial 0.0055 0.13 0.896 1.56

Insurance 0.0502 1.04 0.299 1.78

Constant -0.9863 -6.43 0.000 -

R2 = 0.3268 F (10, 102) = 4.95 Mean VIF = 1.31

Adj. R2 = 0.2608 Sig. F = 0.0000

This table summarizes the results of OLS regression with Log Transformation. It shows that listing and industry type (banking sector) variables were positively and significantly associated with the extent of disclosure of the sample companies at the 1% level. DISCUSSIONS AND CONCLUSIONS This paper analyses the possible impacts of eight specific company variables on the extent of disclosure by UAE firms. The descriptive analysis revealed that the overall mean value of disclosure in the UAE is 57%, reflecting a low to moderate level of disclosure. The major conclusion that can be drawn from the regression analysis is that the industry type, listing status and the size of firm (market capitalization) are the most powerful explanatory variables when related to the variation in compliance with regulations that specify mandatory disclosure on the part of UAE firms. With regard to the other variables, the model showed that these variables did not seem to be the main determinants of variations in the extent of disclosure of the sample companies. This finding can be explained on the grounds that listed firms are exposed to more disclosure requirements stipulated by the UAE securities market. Similarly, unlike the other sectors, the banking sector is the most regulated sector by the Central Bank of the UAE. The results also showed that market capitalization was significant at the 5% level, but with a negative coefficient. This result provides unexpected evidence and is inconsistent with H5, which states that big firms tend to disclose more information than firms with small firms. This can be attributed to the fact that big firms have more social and political influence to avoid compliance with mandatory disclosure requirements. The results of this study have extended the understanding of how characteristics of a firm help to explain the variability in disclosure. The extent of disclosure was found to be significantly associated with listing status, industry type, and size of firm. This finding not only provides support for previous studies, but is also of relevance to those in the UAE who want to understand corporate disclosure. A possible policy implication of this finding is that unlisted large firms need closer scrutiny by the regulatory authorities.


117

The UAE authorities need to evaluate the efficacy of the regulatory requirements and also to introduce more effective monitoring and enforcement mechanisms. The findings of the study should also be of interest to UAE user-groups. As firms with small market capitalization, banking sector firms, and listed firms on the UAE securities market disclose more information in their annual reports, users should be cautious when dealing with large and unlisted firms and may have to consider different sources of information in addition to annual reports. This study also reveals that external auditors in the UAE provide unqualified reports without mentioning any departure from compliance with mandatory requirements of the UAE securities market or IFRS. This raises a question about the quality of auditing practices in this country and requires more attention from the authorities concerned. Currently the UAE Accountants and Auditors Association (AAA) has no authority to regulate the profession. Cooperation between government authorities and this accounting body is crucial in order to regulate financial reporting effectively (Craig & Diga, 1996). Consequently, UAE authorities should give more responsibility and support to the AAA, which can play an important role in increasing awareness among its members about disclosure requirements, as well as ensuring that only qualified auditors are licensed. This study provides some insights into the determinants of disclosure level in the UAE. However, the findings of this study should be interpreted with care as several limitations are associated with this kind of research. The first limitation relates to the use of the chosen index to measure the extent of disclosure. Although the disclosure index is considered the most suitable methodology to test the extent of disclosure (Marston & Shrives, 1991; Botosan, 1997; Prencipe, 2004), the interpretation of these results is constrained by the validity and reliability of the disclosure index used in the study. The level of corporate disclosure may be affected by the subjective selection of items for information disclosure. While the disclosure items included in the index were carefully selected, they may not fully encompass all possible items that need to be included in the assessment of corporate reporting practices. Wallace and Naser (1995) pointed out that the results of using a disclosure index may be different if the number or nature of items was changed. Also, the evaluation process was limited to mandatory items only as it was not possible to include voluntary disclosure items. Consequently, using a disclosure index with non-mandatory items may reveal quite different results from the present study. The subjectivity problem inherent in scoring the annual reports of the sample companies may not have been completely eliminated and hence, there is an unavoidable subjectivity in the scoring process (Owsus-Ansah, 1998b). Consequently, the comprehensiveness of corporate disclosures may not have been fully and/or properly captured by the disclosure index used in the study. As the economy of the UAE advances through time, more research will be needed in the future in order to gain further understanding of corporate disclosure. The current study examined the relationship between mandatory disclosure and certain company characteristics in a single period. Further research might attempt to extend this examination to include two or more periods, such as before and after the establishment of the official securities market, or a comparative study of firms before and after they are listed. In examining the explanatory power of company characteristics, it is possible that other variables, which have an impact on the extent of disclosure in the UAE, have not been included in the present study. Consequently, future research should investigate the effect of other variables, such as the qualifications of audit committee members and cultural factors, on the level of disclosure. Moreover, as the government moves towards the privatization of government-owned companies, a future research project could examine the impact of privatization on the disclosure behaviour of such companies.


118

REFERENCES Abbott, L, Parker, S., & Peters, G. (2004). “Audit Committee Characteristics and Restatements”, Auditing: A Journal of Practice and Theory, vol. 23(1), p. 69 – 87. Abd-Elsalam, O., & Weetman, P. (2003). “Introducing International Accounting Standards to an Emerging Capital Market: Relative Familiarity and Language Effect in Egypt”, Journal of International Accounting, Auditing & Taxation, vol. 12, p. 63 – 84. Abdulsalam, M. (1985). “General Disclosure Requirements and Measuring Its Existence in the Saudi Corporate Annual Reports”, College of Management Science Research Center, King Saud University, Riyadh. Abu Nassar, M. (1993). The Development of Financial Reporting in Jordan: A Survey of Prepares’ and Users’ Attitudes and Reporting Practices, Unpublished Ph.D. Dissertation, University of Kent at Canterbury UK. Ahmed, A., (2012). “Disclosure of Financial Reporting and Firm Structure as a Determinant: A Study on the Listed Companies of DSE”, ASA University Review, vol. 6(1). Ahmed, K. & Courtis, J. (1999). “Association between Corporate Characteristics and Disclosure Levels in Annual Reports: A Meta-Analysis”, British Accounting Review, vol. 31, p. 35 – 61. Ahmed, K., & Nicholls, D. (1994). “The Impact of Non-Financial Characteristics on Mandatory Disclosure Compliance in Developing Countries: The Case of Bangladesh”, International Journal of Accounting Education and Research, vol. 29(1), p. 62 – 77. Al-Hussaini, A. (2001). The Quality of Financial Reporting Practice in Kuwait, Unpublished Ph.D. Thesis, University of Wales. Aljifri, K., (2008). “Annual Reports Disclosure in a Developing Country: the Case of UAE”, Advances in Accounting, Volume 24(1), p. 93 – 100. Al-Mulhem, A. A. (1997). An Empirical Investigation of the Level of Financial Disclosure by Saudi Arabian Corporations, Unpublished PhD Dissertation, Hull University, UK. Alsaeed, K. (2005). “The Association between Firm-Specific Characteristics and Disclsoure: The Case of Saudi Arabia”, The Journal of American Academy of Business, Cambridge, vol. 7(1), p. 310 – 321. Al-Shammari, B. (2005). Compliance with International Accounting Standards by Listed Companies in the Gulf Co-Operation Council Member States: An Empirical Study, Unpublished PhD Dissertation, UWA, Australia. Al-Shayeb, A. (2003). “Corporate Financial Reporting in the UAE”, Arab Journal of Administrative Science, vol. 10(1), p. 109-125. Al Zoubi, A. & Al Zoubi, N. ( 2012). “The Adequacy of Accounting Mandatory Disclosure Under the Global Financial Crisis: a Field Study in Jordan”, Accounting and Management Information Systems, vol. 11(30, p. 424–441.


119

Anderson, D. Sweeney, D., & Williams, T. (1993). Statistics for business and economics. West Publishing Company, St. Paul. Archambault, J., & Archambault, M. (2003). “A Multinational Test of Determinants of Corporate Disclosure”, The International Journal of Accounting, vol. 38, p. 173 –194. Babbie, E. R. (1994). The Practice of Social Research, Wadsworth, Belmont, California, US. Barrett, M. E. (1976). “Financial Reporting Practice: Disclosure and Comprehensiveness in an International Setting”, Journal of Accounting Research, vol. 14(1), p. 10 – 26. Barry, C. B., & Brown, S. J. (1985). “Differential Information and Security Market Equilibrium”, Journal of Financial and Quantitative Analysis, vol. 20(4), p. 407 – 422. Beasley, M., & Salterio, S. E. (2001). “The Relationship between Board Characteristics and Voluntary Improvements in Audit Committee Composition and Experience”, Contemporary Accounting Research, vol. 18(4), p. 539 – 570. Belkaoui, A., & Kahl, A. (1978). Corporate Financial Disclosure in Canada, Research Monograph No. 1 of Canadian Certificate General Accountants Associations, June. Biobele, S.B., Igbeng, I.E. & John, E.F., (2013). “The Significance of International Corporate Governance Disclosure on Financial Reporting in Nigeria”, International Journal of Business and Management, vol. 8(8), P. 100-106. Bhayani, S. (2012). “The relationship between Comprehensiveness of Corporate Disclosure and Firm Characteristics in India, Asia-Pacific Finance and Accounting Review, vol. 1(1), p. 52-66. Blue Ribbon Committee (BRC) (1999). Report and Recommendations on Improving the Effectiveness of Corporate Audit Committees, New York: The New York Stock Exchange and the National Association of Securities Dealers. Botosan, C. (1997). “Disclosure Level and the Cost of Equity Capital”, The Accounting Review, vol. 72(3), p. 323 – 349. Bradbury, M. E. (1992), “Voluntary Disclosure of Financial Segment Data: New Zealand Evidence”, Accounting and Finance, vol. 32(1), p. 15 – 26. Buckland, R., Suwaidan, M., & Thomson, L. (1998). Companies’ Voluntary Disclosure Behaviour and Raising Equity Capital in the Amman Financial Market, Working Paper, Aberdeen Papers in Accountancy, Finance & Management. Buzby, S. L. (1974). “The Nature of Adequate Disclosure”, Journal of Accountancy, April, 38 – 47. Buzby, S. L. (1975), “Company Size, Listed Versus Unlisted Stock, and the Extent of Financial Disclosure”, Journal of accounting Research, vol. 13(1), p. 16 – 37. Chakroun, R. & Matoussi, C. (2012), “Determinants of the Extent of Voluntary Disclosure in the Annual Reports of the Tunisian Firms”, Accounting and Management Information Systems, vol. 11(3), p. 335–370.


120

Chau, G. K., & Gray, S. J. (2002). “Ownership Structure and Corporate Voluntary Disclosure in Hong Kong and Singapore”, Internationals Journal of Accounting, vol. 37(2), p. 247 – 265. Chen, C. L. P., & Jaggi, B. (2000). “Association between Independent Non-executive Directors, Family Control and Financial Disclosure in Hong Kong”, Journal of Accounting and Public Policy, vol. 19, p. 285 – 310. Choi, F. D. S. (1973)a. “Financial Disclosure in Relation to a Firm’s Capital Costs”, Accounting and Business Research, Autumn, p. 282 – 292. Choi, F. D. S. (1973)b. “Financial Disclosure and Entry to the European Capital Market”, Accounting Research, vol. 11(2), p. 159 – 75. Cooke, T. E., & Wallace, R. S.O. (1990). “Financial Disclosure Regulation and its Environment: A Review and Further Analysis”, Journal of Accounting and Public Policy, vol. 9, p. 79 – 110. Cooke, T. E. (1989)a. “Disclosure in the Corporate Annual Reports of Swedish Companies”, Accounting and Business Research, vol. 19(74), p. 113 – 124. Cooke, T. E. (1989) b. “Voluntary corporate Disclosure by Swedish Companies”, Journal of International Financial Management and Accounting, vol. 2, p. 171 – 195. Cooke, T. E. (1989)c. An Empirical Study of Financial Disclosure by Swedish Companies. New York: Garland Publishing Inc., New York & London. Cooke, T. E. (1998). “Regression Analysis in Accounting Disclosure Studies”, Accounting and Business Research’, vol. 28(3), p. 209-224. Craig, R., & Diga, J. (1998). “Corporate Accounting Disclosure in ASEAN”, Journal of International Financial Management and Accounting, vol. 9(3), p. 246 – 274. Craswell, A. T., & Taylor, S. L. (1992). “Discretionary Disclosure of Reserves by Oil and Gas Companies: An Economic Analysis”, Journal of Business Finance and Accounting, vol. 19(2), p. 295 – 308. Dhaliwal, D. S. (1979). “Disclosure Regulations and the Cost of Capital”, Southern Economic Journal, p. 785 – 794. Dye, R. A. & Sridhar, S. S. (1995). “Industry-Wide Disclosure Dynamics’, Journal of Accounting Research, vol. 33(1), p. 157 – 174. Fama, E. (1980). “Agency Problem and Theory of the Firm”, Journal of Political Economy, vol. 88, no. 2, pp. 288 – 307. Fama, E., & Jensen, M. C. (1983). “Separation of Ownership and Control”, Journal of Law and Economics, vol. 25, p. 301 – 325. Firth, M. (1980). “Raising Finance and Firm’s Corporate Reporting Policies”, Abacus, vol. 16, December, p. 100 – 115.


121

Fox, J. (1997). Applied Regression Analysis, Linear Models, and Related Methods, Sage Publications, California. Gibbins, M., Richardson, A., & Waterhouse, J. (1990). “The Management of Corporate Financial Disclosure: Opportunism, Ritualism, Polices, and Processes”, Journal of accounting Research, Spring, p. 121 – 143. Gray, S. J. & Roberts, C. B. (1989). “Voluntary Information Disclosure and the British Multinationals: Corporate Perceptions of Costs and Benefits”, in A.G. Hopwood, International Pressures of Accounting for Change, Prentice-Hall International. Greene, W. (2003). Econometric Analysis, 5th ed, Prentice Hall, New Jersey, US. Haniffa, R. M., & Cooke, T. E. (2002). “Culture, Corporate Governance and Disclosure in Malaysian Corporations”, ABACUS, vol. 38(3), p. 317 – 349. Hanushek, E., & Jackson, J. (1977). Statistical Methods for Social Scientists, Academic Press, New York Hooks, J., Coy, D., & Davey H. (2002). “The Information Gap in Annual Reports”, Accounting, Auditing, and Accountability Journal, vol. 15(4), p. 501 –522. Hossain, M., & Hammami, H. (2009). “Voluntary Disclosure in the Annual Reports of an Emerging Country: the Case of Qatar”, Advances in Accounting, vol. 25, 255-265. Hossain, M., Tan, L. M., & Adams, M. (1994). “Voluntary Disclosure in an Emerging Capital Market: Some Empirical Evidence from Companies Listed on the Kuala Lumpur Stock Exchange”, International Journal of Accounting, vol. 29(4), p. 334 – 351. Imhoff, E. A. (1992). “The Relation between Perceived Accounting Quality and Economic Characteristics of the Firm”, Journal of Accounting and Public Policy, vol. 11, Summer, p. 97 – 118. Inchausti, B. G. (1997). “The Influence of Company Characteristics and Accounting Regulation on Information Disclosed by Spanish Firms”, The European Accounting Review, vol. 6(1), p. 45 – 68. Jensen, M. C., & Meckling, W. H. (1976). “Theory of the Firm: Management Behavior, Agency Costs and Ownership Structure”, Journal of Financial Economics, vol. 3(4), p. 305 – 360. Khasharmeh, H., & Aljifri, k. (2010). “The Timelines of Annual Report in Bahrain and the United Arab Emirates: An Empirical Comparative Study”, International Journal of Business and Finance Research. Kreutzfeldt, R., & Wallace, W. (1986). “Error Characteristics in Audit Population: Their Profile and Relationships to Environment Factors”, Auditing: Journal of Practice and Theory, vol. 6(1), p. 20 – 43. Lang, M., & Lundholm, R. J. (1993). “Cross-sectional Determinants of Analysts Ratings of Corporate Disclosure”, Journal of Accounting Research, vol. 312, Autumn, p. 45 – 271. Leftwich, R., Watts, R., & Zimmerman, J. (1981). “Voluntary Corporate Disclosure: The Case of Interim Reporting”, Journal of Accounting Research, vol. 19, Supplement, p. 50 –77. Makhija, A., & Patton, J. (2004). “The Impact of Firm Ownership Structure on Voluntary Disclosure: Empirical Evidence from Czech Annual Reports”, Journal of Business, vol. 77(3), p. 457 – 491.


122

Malone, D., Fries, C., & Jones, T. (1993). “An Empirical Investigation of the Extent of Corporate Financial Disclosure in the Oil and Gas Industry”, Journal of Accounting, Auditing and Finance, 83, Summer, p. 249 –273. Marston, C. L., & Shrives, P. J. (1991). “The Use of Disclosure Indices in Accounting Research: A Review Article”, British Accounting Review, vol. 23(3), p. 195 – 210. Naser, K., & Al-Khatib, K. (2000). “Determinants of the Depth of Voluntary Disclosure in the Directors Statement in a Sample of Jordanian Listed Companies”, Advanced in International Accounting, vol. 13, p. 99 – 118. Naser, K., & Nuseibeh, R. (2003). “Quality of Financial Reporting: Evidence from the Listed Saudi Non-financial Companies”, International Journal of Accounting, vol. 38, p. 41 – 69. Nasser, K. (1998). “Comprehensiveness of Disclosure of Non-Financial Companies Listed on The Amman Financial Market”, International Journal of Commerce and Management, vol. 8(1), p. 88 – 119. Naser, K., Al-Khatib, K., & Karbhari, Y. (2002). “Empirical Evidence on the Depth of Corporate Information Disclosure in Developing Countries: the Case of Jordan”, International Journal of Commerce and Management, vol. 12(3-4), p. 122 – 155. Owsus-Ansah, S. (1998)a. The Adequacy of Corporate Mandatory Disclosure Practices on Emerging Markets: A Case Study of the Zimbabwe Stock Exchange, Unpublished Thesis, Middlesex University, London, UK. Owsus-Ansah, S. (1998)b. “The Impact of Corporate Attributes on the Extent of Mandatory Disclosure and Reporting by Listed Companies in Zimbabwe”, International Journal of Accounting, vol., 33(5), p. 605 – 631. Patton, J., & Zelenka, I. (1997). “An Empirical Analysis of the Determinants of the Extent of disclosure in Annual Reports of Joint Stock Companies in the Czech Republic”, European Accounting Review, vol. 6(4), p. 605 – 626. Prencipe, A. (2004). “Proprietary Costs and Determinants of Voluntary Segment Disclosure: Evidence form Italian Listed Companies”, European Accounting Review, vol. 13(2), p. 319 – 340. Priebjrivat, A. (1991). Corporate Disclosure: A Case of Securities Exchange of Thailand, Unpublished Ph.D. Thesis, New York University, New York, USA. Salawu, M. (2013). “The Extent and Forms of Voluntary Disclosure of Financial Information on Internet in Nigeria: An Exploratory Study”, International Journal of Financial Research, vol. 4(1), p. 110-119. Shamser, M., & Annuar, M. N. (1993). “Management versus Shareholders’ Interest: Board Composition, Market Risk and Shareholders Returns of Malaysian Listed Firms”, Malaysian Management Review, June, p. 44. Singhvi, S. S., & Desai, H. (1971). “An Empirical Analysis of the Quality of Corporate Financial Disclosure”, Accounting Review, vol. 46(1), p. 129 –138.


123

Solas, C. (1994). “Financial Reporting Practice in Jordan: An Empirical Test”, Advances in International Accounting, vol. 7, p. 43 – 60. Stanga, K. (1976). “Disclosure in Published Annual reports”, Financial Management, vol. 5(4), p. 42 – 52. Suwaidan, M. (1997). Voluntary Disclosure of Accounting Information: The Case of Jordan, Unpublished Ph.D. thesis, University of Aberdeen, UK. Wallace, R. S. O. (1988). “Intra-national and International Consensus on the Importance of Disclosure Items in Financial Reports: a Nigerian case study”, British Accounting Review, vol. 20(2), p. 223 – 265. Wallace, R. S. O. (1989). Accounting and Financial Reporting in Nigeria, The Institution of Chartered Accountants in England and Wales, Libre Print, Norwich, UK. Wallace, R. S. O., & Naser, K. (1995). “Firm-specific determinants of the comprehensiveness of mandatory disclosure in the corporate annual reports of firms listed on the stock exchange of Hong Kong”, Journal of Accounting and Public Policy, vol., 14(4), p. 311 – 368. Wallace, R. S. O., Naser, K. & Mora, A. (1994). “The Relationship between the Comprehensiveness of Corporate Annual Reports and Firm Characteristics in Spain”, Accounting and Business Research, vol. 25(97), p. 41 – 53. Watts, R. (1977). “Corporate Financial Statements: A Product of the Market and Political Process”, Australian Journal of Management, vol. 2(10), p. 53 –75. Wikipedia. (2008). Economy of the United Arab Emirates, viewed on 2nd May 2008,< http://en.wikipedia.org/wiki/Economy_of_the_United_Arab_Emirates>. Zarzeski, M. T. (1996). “Spontaneous Harmonization Effects of Culture and Market Forces on Accounting Disclosure Practice”, Accounting Horizons, vol. 10(1), p. 18 – 37. ACKNOWLEDGMENT The authors wish to thank the two anonymous reviewers for their valuable comments and suggestions. Biography Dr. Khaled Aljifri is an Associate Professor of Accounting at United Arab Emirates University. He can be contacted at: College of Business and Economics, Accounting Department, UAEU, P.O. Box 15551, Al Ain, United Arab Emirates. Email: [email protected]. Dr. Abdulkarim Alzarouni is the Deputy General Manager of Group Internal Audit in the National Bank of Abu Dhabi. He can be contacted via email: [email protected]. Dr. Chew Ng is Professor of Financial Accounting at the Griffith Business School, Griffith University. She can be contacted at: Department of Accounting, Finance and Economics, Griffith Business School, Griffith University, Nathan Campus, Queensland, Australia 4111. E-mail: [email protected]. Dr. Mohammad Iqbal Tahir. He can be contacted at: School of Management Studies, the University of Faisalabad, Pakistan, and Department of Accounting, Finance and Economics Griffith University, Nathan Campus, Queensland, Australia 4111. Email: [email protected].

REVIEWERS The IBFR would like to thank the following members of the academic community and industry for their much appreciated contribution as reviewers.

Hisham Abdelbaki, University of Mansoura - Egypt

Isaac Oluwajoba Abereijo, Obafemi Awolowo University

Naser Abughazaleh, Gulf University For Science And Technology

Nsiah Acheampong, University of Phoenix

Vera Adamchik, University of Houston-Victoria

Iyabo Adeoye, National Horticultural Research Instittute, Ibadan, Nigeria.

Michael Adusei, Kwame Nkrumah University of Science And Technology

Mohd Ajlouni, Yarmouk University

Sylvester Akinbuli, University of Lagos

Anthony Akinlo, Obafemi Awolowo University

Yousuf Al-Busaidi, Sultan Qaboos University

Khaled Aljaaidi, Universiti Utara Malaysia

Hussein Al-tamimi, University of Sharjah

Paulo Alves, CMVM, ISCAL and Lusofona University

Ghazi Al-weshah, Albalqa Applied University

Glyn Atwal, Groupe Ecole Supérieure de Commerce de Rennes

Samar Baqer, Kuwait University College of Business Administration

Susan C. Baxter, Bethune-Cookman College

Nagib Bayoud, Tripoli University

Ahmet Bayraktar, Rutgers University

Kyle Brink, Western Michigan University

Giovanni Bronzetti, University of Calabria

Karel Bruna, University of Economics-Prague

Priyashni Chand, University of the South Pacific

Wan-Ju Chen, Diwan College of Management

Yahn-shir Chen, National Yunlin University of Science and Techology, Taiwan

Bea Chiang, The College of New Jersey

Te-kuang Chou, Southern Taiwan University

Shih Yung Chou, University of the Incarnate Word

Caryn Coatney, University of Southern Queensland

Iyanna College of Business Administration,

Michael Conyette, Okanagan College

Huang Department of Accounting, Economics & Finance,

Rajni Devi, The University of the South Pacific

Leonel Di Camillo, Universidad Austral

Steven Dunn, University of Wisconsin Oshkosh

Mahmoud Elgamal, Kuwait University

Ernesto Escobedo, Business Offices of Dr. Escobedo

Zaifeng Fan, University of Wisconsin whitewater Perrine Ferauge University of Mons

Olga Ferraro, University of Calabria

William Francisco, Austin Peay State University

Peter Geczy, AIST

Lucia Gibilaro, University of Bergamo

Hongtao Guo, Salem State University

Danyelle Guyatt, University of Bath

Zulkifli Hasan, Islamic University College of Malaysia

Shahriar Hasan, Thompson Rivers University

Peng He, Investment Technology Group

Niall Hegarty, St. Johns University

Paulin Houanye, University of International Business and Education, School of Law

Daniel Hsiao, University of Minnesota Duluth

Xiaochu Hu, School of Public Policy, George Mason University

Jui-ying Hung, Chatoyang University of Technology

Fazeena Hussain, University of the South Pacific

Shilpa Iyanna, Abu Dhabi University

Sakshi Jain, University of Delhi

Raja Saquib Yusaf Janjua, CIIT

Yu Junye, Louisiana State University

Tejendra N. Kalia, Worcester State College

Gary Keller, Eastern Oregon University

Ann Galligan Kelley, Providence College

Ann Kelley, Providence college

Ifraz Khan, University of the South Pacific

Halil Kiymaz, Rollins College

Susan Kowalewski, DYouville College

Bamini Kpd Balakrishnan, Universiti Malaysia Sabah

Bohumil Král, University of Economics-Prague

Jan Kruger, Unisa School for Business Leadership

Christopher B. Kummer, Webster University-Vienna

Mei-mei Kuo, JinWen University of Science & Technology

Mary Layfield Ledbetter, Nova Southeastern University

John Ledgerwood, Embry-Riddle Aeronautical University

Yen-hsien Lee, Chung Yuan Christian University

Shulin Lin, Hsiuping University of Science and Technology

Yingchou Lin, Missouri Univ. of Science and Technology

Melissa Lotter, Tshwane University of Technology

Xin (Robert) Luo, Virginia State University

Andy Lynch, Southern New Hampshire University

Abeer Mahrous, Cairo university

Gladys Marquez-Navarro, Saint Louis University

Cheryl G. Max, IBM

Romilda Mazzotta, University of Calabria

Mary Beth Mccabe, National University

Avi Messica, Holon Institute of Technology

Scott Miller, Pepperdine University

Cameron Montgomery, Delta State University

Sandip Mukherji, Howard University

Tony Mutsue, Iowa Wesleyan College

Cheedradevi Narayanasamy, Graduate School of Business, National University of Malaysia

Dennis Olson, Thompson Rivers University

Godwin Onyeaso, Shorter University

Bilge Kagan Ozdemir, Anadolu University

Dawn H. Pearcy, Eastern Michigan University

Pina Puntillo, University of Calabria (Italy)

Rahim Quazi, Prairie View A&M University

Anitha Ramachander, New Horizon College of Engineering

Charles Rambo, University Of Nairobi, Kenya

Prena Rani, University of the South Pacific

Kathleen Reddick, College of St. Elizabeth

Maurizio Rija, University of Calabria.

Matthew T. Royle, Valdosta State University

Tatsiana N. Rybak, Belarusian State Economic University

Rafiu Oyesola Salawu, Obafemi Awolowo University

Paul Allen Salisbury, York College, City University of New York

Leire San Jose, University of Basque Country

I Putu Sugiartha Sanjaya, Atma Jaya Yogyakarta University, Indonesia

Sunando Sengupta, Bowie State University

Brian W. Sloboda, University of Phoenix

Smita Mayuresh Sovani, Pune University

Alexandru Stancu, University of Geneva and IATA (International Air Transport Association)

Jiří Strouhal, University of Economics-Prague

Vichet Sum, University of Maryland -- Eastern Shore

Qian Sun, Kutztown University

Diah Suryaningrum, Universitas Pembangunan Nasional Veteran Jatim

Andree Swanson, Ashford University

James Tanoos, Saint Mary-of-the-Woods College

Jeannemarie Thorpe, Southern NH University

Ramona Toma, Lucian Blaga University of Sibiu-Romania Alejandro Torres Mussatto Senado de la Republica & Universidad de Valparaíso

Jorge Torres-Zorrilla, Pontificia Universidad Católica del Perú

William Trainor, East Tennessee State University

Md Hamid Uddin, University Of Sharjah

Ozge Uygur, Rowan University

K.W. VanVuren, The University of Tennessee – Martin

Vijay Vishwakarma, St. Francis Xavier University

Ya-fang Wang, Providence University

Richard Zhe Wang, Eastern Illinois University

Jon Webber, University of Phoenix

Jason West, Griffith University

Wannapa Wichitchanya, Burapha University

Veronda Willis, The University of Texas at San Antonio

Bingqing Yin, University of Kansas

Fabiola Baltar, Universidad Nacional de Mar del Plata

Myrna Berrios, Modern Hairstyling Institute

Monica Clavel San Emeterio, University of La Rioja

Esther Enriquez, Instituto Tecnologico de Ciudad Juarez

Carmen Galve-górriz, Universidad de Zaragoza

Blanca Rosa Garcia Rivera, Universidad Autónoma De Baja California

Carlos Alberto González Camargo, Universidad Jorge Tadeo Lozano

Hector Alfonso Gonzalez Guerra, Universidad Autonoma De Coahuila

Claudia Soledad Herrera Oliva, Universidad Autónoma De Baja California

Eduardo Macias-Negrete, Instituto Tecnologico De Ciudad Juarez

Jesús Apolinar Martínez Puebla, Universidad Autónoma De Tamaulipas

Francisco Jose May Hernandez, Universidad Del Caribe

Aurora Irma Maynez Guaderrama, Universidad Autonoma De Ciudad Juarez

Linda Margarita Medina Herrera, Tecnológico De Monterrey. Campus Ciudad De México

Erwin Eduardo Navarrete Andrade, Universidad Central De Chile

Gloria Alicia Nieves Bernal, Universidad Autónoma Del Estado De Baja California

Julian Pando, University Of The Basque Country

Eloisa Perez, Macewan University

Iñaki Periáñez, Universidad Del Pais Vasco (Spain)

Alma Ruth Rebolledo Mendoza, Universidad De Colima

Carmen Rios, Universidad del Este

Celsa G. Sánchez, CETYS Universidad

Adriana Patricia Soto Aguilar, Benemerita Universidad Autonoma De Puebla Amy Yeo, Tunku Abdul Rahman College

Vera Palea, University of Turin

Fabrizio Rossi,University of Cassino and Southern Lazio

Intiyas Utami , Satya Wacana Christian University

Ertambang Nahartyo, UGM

REVIEWERS The IBFR would like to thank the following members of the academic community and industry for their much appreciated contribution as reviewers.

Haydeé Aguilar, Universidad Autónoma De Aguascalientes

Bustamante Valenzuela Ana Cecilia, Universidad Autonoma De Baja California

María Antonieta Andrade Vallejo, Instituto Politécnico Nacional

Olga Lucía Anzola Morales, Universidad Externado De Colombia

Antonio Arbelo Alvarez, Universidad De La Laguna

Hector Luis Avila Baray, Instituto Tecnologico De Cd. Cuauhtemoc

Graciela Ayala Jiménez, Universidad Autónoma De Querétaro

Albanelis Campos Coa, Universidad De Oriente

Carlos Alberto Cano Plata, Universidad De Bogotá Jorge Tadeo Lozano

Alberto Cardenas, Instituto Tecnologico De Cd. Juarez

Edyamira Cardozo, Universidad Nacional Experimental De Guayana

Sheila Nora Katia Carrillo Incháustegui, Universidad Peruana Cayetano Heredia

Emma Casas Medina, Centro De Estudios Superiores Del Estado De Sonora

Benjamin Castillo Osorio, Universidad Del Sinú-Sede Monteria

Benjamín Castillo Osorio, Universidad Cooperativa De Colombia Y Universidad De Córdoba

María Antonia Cervilla De Olivieri, Universidad Simón Bolívar

Cipriano Domigo Coronado García, Universidad Autónoma De Baja California

Semei Leopoldo Coronado Ramírez, Universidad De Guadalajara

Esther Eduviges Corral Quintero, Universidad Autónoma De Baja California

Dorie Cruz Ramirez, Universidad Autonoma Del Estado De Hidalgo /Esc. Superior De Cd. Sahagún

Tomás J. Cuevas-Contreras, Universidad Autónoma De Ciudad Juárez

Edna Isabel De La Garza Martinez, Universidad Autónoma De Coahuila

Hilario De Latorre Perez, Universidad Autonoma De Baja California

Javier De León Ledesma, Universidad De Las Palmas De Gran Canaria - Campus Universitario De Tafira

Hilario Díaz Guzmán, Universidad Popular Autónoma Del Estado De Puebla

Cesar Amador Díaz Pelayo, Universidad De Guadalajara, Centro Universitario Costa Sur

Avilés Elizabeth, Cicese

Ernesto Geovani Figueroa González, Universidad Juárez Del Estado De Durango

Ernesto Geovani Figueroa González, Universidad Juárez Del Estado De Durango

Carlos Fong Reynoso, Universidad De Guadalajara

Ana Karen Fraire, Universidad De Gualdalajara

Teresa García López, Instituto De Investigaciones Y Estudios Superiores De Las Ciencias Administrativas

Helbert Eli Gazca Santos, Instituto Tecnológico De Mérida

Denisse Gómez Bañuelos, Cesues

María Brenda González Herrera, Universidad Juárez Del Estado De Durango

Ana Ma. Guillén Jiménez, Universidad Autónoma De Baja California

Araceli Gutierrez, Universidad Autonoma De Aguascalientes

Andreina Hernandez, Universidad Central De Venezuela

Arturo Hernández, Universidad Tecnológica Centroamericana

Alejandro Hernández Trasobares, Universidad De Zaragoza

Alma Delia Inda, Universidad Autonoma Del Estado De Baja California

Carmen Leticia Jiménez González, Université De Montréal Montréal Qc Canadá.

Gaspar Alonso Jiménez Rentería, Instituto Tecnológico De Chihuahua

Lourdes Jordán Sales, Universidad De Las Palmas De Gran Canaria

Santiago León Ch., Universidad Marítima Del Caribe

Graciela López Méndez, Universidad De Guadalajara-Jalisco

Virginia Guadalupe López Torres, Universidad Autónoma De Baja California

Angel Machorro Rodríguez, Instituto Tecnológico De Orizaba

Cruz Elda Macias Teran, Universidad Autonoma De Baja California

Aracely Madrid, ITESM, Campus Chihuahua

Deneb Magaña Medina, Universidad Juárez Autónoma De Tabasco

Carlos Manosalvas, Universidad Estatal Amazónica

Gladys Yaneth Mariño Becerra, Universidad Pedagogica Y Tecnológica De Colombia

Omaira Cecilia Martínez Moreno, Universidad Autónoma De Baja California-México

Jesus Carlos Martinez Ruiz, Universidad Autonoma De Chihuahua

Alaitz Mendizabal, Universidad Del País Vasco

Alaitz Mendizabal Zubeldia, Universidad Del País Vasco/ Euskal Herriko Unibertsitatea

Fidel Antonio Mendoza Shaw, Universidad Estatal De Sonora

Juan Nicolás Montoya Monsalve, Universidad Nacional De Colombia-Manizales

Jennifer Mul Encalada, Universidad Autónoma De Yucatán

Gloria Muñoz Del Real, Universidad Autonoma De Baja California

Alberto Elías Muñoz Santiago, Fundación Universidad Del Norte

Bertha Guadalupe Ojeda García, Universidad Estatal De Sonora

Erika Olivas, Universidad Estatal De Sonora

Erick Orozco, Universidad Simon Bolivar

Rosa Martha Ortega Martínez, Universidad Juárez Del Estado De Durango

José Manuel Osorio Atondo, Centro De Estudios Superiores Del Estado De Sonora

Luz Stella Pemberthy Gallo, Universidad Del Cauca

Andres Pereyra Chan, Instituto Tecnologico De Merida

Andres Pereyra Chan, Instituto Tecnologico De Merida

Adrialy Perez, Universidad Estatal De Sonora

Hector Priego Huertas, Universidad De Colima

Juan Carlos Robledo Fernández, Universidad EAFIT-Medellin/Universidad Tecnologica De Bolivar-Cartagena

Natalia G. Romero Vivar, Universidad Estatal De Sonora

Humberto Rosso, Universidad Mayor De San Andres

José Gabriel Ruiz Andrade, Universidad Autónoma De Baja California-México

Antonio Salas, Universidad Autonoma De Chihuahua

Claudia Nora Salcido, Universidad Juarez Del Estado De Durango

Juan Manuel San Martín Reyna, Universidad Autónoma De Tamaulipas-México

Francisco Sanches Tomé, Instituto Politécnico da Guarda

Edelmira Sánchez, Universidad Autónoma de Ciudad Juárez

Deycy Janeth Sánchez Preciado, Universidad del Cauca

María Cristina Sánchez Romero, Instituto Tecnológico de Orizaba

María Dolores Sánchez-fernández, Universidade da Coruña

Luis Eduardo Sandoval Garrido, Universidad Militar de Nueva Granada

Pol Santandreu i Gràcia, Universitat de Barcelona, Santandreu Consultors

Victor Gustavo Sarasqueta, Universidad Argentina de la Empresa UADE

Jaime Andrés Sarmiento Espinel, Universidad Militar de Nueva Granada

Jesus Otoniel Sosa Rodriguez, Universidad De Colima

Edith Georgina Surdez Pérez, Universidad Juárez Autónoma De Tabasco

Jesús María Martín Terán Gastélum, Centro De Estudios Superiores Del Estado De Sonora

Jesus María Martín Terán Terán Gastélum, Centro De Estudios Superiores Del Estado De Sonora

Jesús María Martín Terán Gastélum, Centro De Estudios Superiores Del Estado De Sonora

Maria De La Paz Toldos Romero, Tecnologico De Monterrey, Campus Guadalajara

Abraham Vásquez Cruz, Universidad Veracruzana

Angel Wilhelm Vazquez, Universidad Autonoma Del Estado De Morelos

Lorena Vélez García, Universidad Autónoma De Baja California

Alejandro Villafañez Zamudio, Instituto Tecnologico de Matamoros

Hector Rosendo Villanueva Zamora, Universidad Mesoamericana

Oskar Villarreal Larrinaga, Universidad del País Vasco/Euskal Herriko Universitatea

Delimiro Alberto Visbal Cadavid, Universidad del Magdalena

Rosalva Diamantina Vásquez Mireles, Universidad Autónoma de Coahuila

Oscar Bernardo Reyes Real, Universidad de Colima

HOW TO PUBLISH

Submission Instructions

The Journal welcomes submissions for publication consideration. Complete directions for manuscript submission are available at the Journal website www.theIBFR.com/journal.htm. Papers may be submitted for initial review in any format. However, authors should take special care to address spelling and grammar issues prior to submission. Authors of accepted papers are required to precisely format their document according to the journal guidelines.

There is no charge for standard paper reviews. The normal review time for submissions is 90-120 days. However, authors desiring a quicker review may elect to pay an expedited review fee, which guarantees an inditial review within two weeks. Authors of accepted papers are required to pay a publication fee based on the manuscript length and number of authors. Please see our website for current publication and expedited review rates.

Authors submitting a manuscript for publication consideration must guarantee that the document contains the original work of the authors, has not been published elsewhere, and is not under publication consideration elsewhere. In addition, submission of a manuscript implies that the author is prepared to pay the publication fee should the manuscript be accepted.

Subscriptions

Individual and library subscriptions to the Journal are available. Please contact us by mail or by email to: [email protected] for updated information.

Contact Information

Mercedes Jalbert, Executive Editor The IBFRP.O. Box 4908Hilo, HI [email protected]

Website

www.theIBFR.org or www.theIBFR.com

Review of Business & Finance Studies Review of Business & Finance Studies (ISSN: 2150-3338 print and 2156-8081 online) publishes high-quality studies in all areas of business, finance and related fields. Empirical, and theoretical papers as well as case studies are welcome. Cases can be based on real-world or hypothetical situations.

All papers submitted to the Journal are double-blind reviewed. The Journal is listed in Cabell’s, Ulrich’s Periodicals Directory The Journal is distributed in print, through EBSCOHost, ProQuest ABI/Inform and SSRN.

The journal accept rate is between 15 and 25 percent

Business Education & AccreditationBE A

AT Accounting

Taxation&

Accounting and Taxation (AT)

Accounting and Taxation (AT) publishes high-quality articles in all areas of accounting, auditing, taxation and related areas. Theoretical, empirical and applied manuscripts are welcome for publication consideration.

All papers submitted to the Journal are double-blind reviewed. AT is listed in Cabell’s and Ulrich’s Periodicals Directory. The Journal is distributed in print, through EBSCOHost, ProQuest ABI/Inform and SSRN.

The journal acceptance rate is between 5 and 15 percent.

Business Education and Acreditation (BEA)Business Education & Accreditation publishes high-quality articles in all areas of business education, curriculum, educational methods, educational administration, advances in educational technology and accreditation. Theoretical, empirical and applied manuscripts are welcome for publication consideration.

All papers submitted to the Journal are double-blind reviewed. BEA is is listed in Cabell’s and Ulrich’s Periodicals Directory. The Journal is distributed in print, through EBSCOHost, ProQuest ABI/Inform and SSRN.

The journal acceptance rate is between 15 and 25 percent.

PUBLICATION OPPORTUNITIES

DE NEGOCIOSREVISTA GLOBAL R

GN

Revista Global de Negocios

Revista Global de Negocis (RGN), a Spanish language Journal, publishes high-quality articles in all areas of business. Theoretical, empirical and applied manuscripts are welcome for publication consideration.

All papers submitted to the Journal are double-blind reviewed. RGN is distributed in print, through EBSCOHost, ProQuest ABI/Inform and SSRN.RGN will be submitted to Ulrich’s Periodicals Directory, colciencia, etc. The Journal is distributed in print, through EBSCOHost, ProQuest ABI/Inform and SSRN

The Journal acceptance rate is 20 percent.

Documents

and ESEARCH · forward the concept of “investor awareness” which states that increased investor awareness for added stocks and a corresponding smaller drop in the awareness for