ESSAYS ON ACCOUNTING INFORMATION SYSTEMS AND …

ESSAYS ON ACCOUNTING INFORMATION SYSTEMS

AND ACCOUNTING CHOICE

By Jongkyum Kim

A dissertation submitted to the

Graduate School-Newark

Rutgers, The State University of New Jersey

in partial fulfillment of requirements

for the degree of

Doctor of Philosophy

Ph.D. in Management

Written under the direction of

Professor Bharat Sarath

and approved by

Dr. Bharat Sarath

Dr. Miklos A. Vasarhelyi

Dr. Divya Anantharaman

Dr. Steven B. Lilien

Newark, New Jersey

October 2016

© [2016]

Jongkyum Kim

ALL RIGHTS RESERVED

ii

ABSTRACT OF THE DISSERTATION

Two Essays on Accounting Information Systems and Accounting Choices

By Jongkyum Kim

Dissertation Director: Professor Bharat Sarath

The first essay investigates whether the implementation of enterprise resource

planning (ERP) systems affects the audit repot lag, the time period between a company’s

fiscal year end and the date of the audit report. Prior research has shown that the

implementation of ERP systems can significantly affect a firm’s business operations and

processes. However, scant research has been conducted on the relationship between ERP

implementation and the timeliness of external audits, such as audit report lags. While some

of the alleged benefits of ERP are closely related to removing impediments contributing to

audit report lags, others argue that the complex mechanisms of ERP systems create greater

complexity for control and audit. The test results indicate that overall a firm’s ERP

implementation is negatively associated with audit report lag. However, this negative

association is significant only at fourth and fifth years after initial ERP implementation.

These results imply that the use of ERP systems by client firms may help decrease the audit

report lag, but it takes time for the full impact of the firms’ accounting systems to be

realized.

The second essay examines the consequences resulting from the adoption of the

fair value option (FVO) of SFAS 159 and explore the intentions of firms that adopt FVO.

The primary objective of SFAS 159 is to mitigate the earnings volatility. However, there

are also arguments implying that FVO could in fact increase earnings volatility. Thus, I

iii

first examine whether adopting FVO helps improve earnings volatility. Also, prior studies

report the negative relationship between earnings volatility and the accuracy of earnings

prediction. Therefore, the second hypothesis tests whether adopting FVO has an impact on

the analysts’ forecast accuracy. The test results indicate that there is a positive association

between FVO and earnings volatility and the relation can be amplified depending on the

size of gains or losses resulting from the FVO. Also, the results indicate that the analysts’

forecast error of FVO adopter is greater than that of non-FVO adopters. I test two more

hypotheses to explore the intention of firms’ decision to adopt FVO: opportunistic intention

and informative intention. However, I do not find results which confirm my hypotheses

except for limited evidence for the opportunistic intention.

iv

DEDICATION

To my parents, my brother and sister, my wife, and my sons

v

ACKNOWLEDGEMENTS

I would never have been able to finish my dissertation without the guidance of my

committee members, assistance from friends, and support from my family.

I would like to express my deepest gratitude to my advisor, Dr. Bharat Sarath, for

his enormous and continuous support and guidance. Dr Sarath has been such a great mentor

not only as an academic advisor but also as a true teacher for my life. The conversations

with him have been a good opportunity to broaden my horizons on different cultures and

our society. I am very fortunate to have worked under the supervision of an outstanding

scholar and a great person like him.

I am also very thankful to Dr. Miklos Vasarhelyi for his warm advices and

continuous support throughout the Ph.D. program. I sincerely admire his passion for

research and students, and the lessons from him will be very important guidance for my

future career as a college professor.

I also would like to thank Dr. Divya Anantharaman for her insightful comments

and support. Whenever I stop by her office for advices, she has always welcomed me with

smile and given me very helpful advices. I deeply appreciate her help and kindness. I also

thank Dr Steven Lilien for serving on my dissertation committee and for his valuable

comments and suggestions for my proposal and defense.

I am also very thankful to my friends and colleagues at Rutgers Business School,

Seokyoon, Changhee, Kyunghee, Karina, Vicky, Li He, Yubin, Hyunsang, and … Sunjung.

Most importantly, I cannot imagine that I could have accomplished this work

without the love and support of my family. I would like to express my warmest appreciation

to my parents, sister, brother, and wife, Jiyoung.

vi

Table of Contents

ABSTRACT OF THE DISSERTATION.................................................................... ii

DEDICATION ............................................................................................................. iv

ACKNOWLEDGEMENTS ........................................................................................ v

Table of Contents ........................................................................................................ vi

List of Tables ............................................................................................................. viii

Chapter 1: The Impact of Enterprise Resource Planning (ERP) Systems on the Audit

Report Lag ............................................................................................... 1

I. Introduction ............................................................................................................. 1

II. Literature Review and Research Question ............................................................. 6

III. Research Design ................................................................................................. 12

IV. Results ................................................................................................................. 19

V. Analyses with ERP data of the 2000s ................................................................... 23

VI. Conclusion .......................................................................................................... 26

References ................................................................................................................ 31

Tables for Chapter 1 ................................................................................................. 36

Chapter 2: The Fair Value Option (FVO) under SFAS 159: Effectiveness of FVO and

Intention of FVO Adopters .................................................................. 49

I. Introduction ........................................................................................................... 49

II. Literature Review on Fair Value Accounting ....................................................... 51

III. SFAS 159 ............................................................................................................ 62

IV. Hypothesis Development .................................................................................... 72

vii

V. Research Design ................................................................................................... 78

VI. Results ................................................................................................................ 84

VII. Conclusion, Limitations, and Extensions .......................................................... 90

References ................................................................................................................ 93

Tables for Chapter 2 ................................................................................................. 97

Appendix A for Chapter 2 ....................................................................................... 120

viii

List of Tables

Table 1.1 Determinants of audit report lag ................................................................. 36

Table 1.2 Tangible and intangible benefits of ERP systems ........................................ 37

Table 1.3 Sample selection and identification of ERP firms ....................................... 38

Table 1.4 Distribution of firm years for audit report lags ............................................ 39

Table 1.5 Variable definitions ..................................................................................... 40

Table 1.6 Descriptive statistics .................................................................................... 41

Table 1.7 Multivariate test results (Model 1) ............................................................... 45

Table 1.8 Multivariate test results (Model 2) ............................................................... 46

Table 1.9 Multivariate test results ................................................................................ 47

Table 1.10 Multivariate test results .............................................................................. 48

Table 2.1 Distribution of FVO firms by FVO adoption year and industry .................. 97

Table 2.2 Variable definitions ..................................................................................... 98

Table 2.3 Descriptive statistics (for H1) .................................................................... 100

Table 2.4 OLS regression analyses of earnings volatility on adoption of FVO (full

sample) ...................................................................................................... 103

Table 2.5 OLS regression analyses of earnings volatility on adoption of FVO (financial

firms only) ................................................................................................. 104

Table 2.6 OLS regression analyses of earnings volatility on adoption of FVO (non-

financial firms only) .................................................................................. 105


Table 2.8 OLS regression analyses of analysts’ forecast error on adoption of FVO 108


ix

Table 2.10 OLS regression analyses of financing activity on adoption of FVO ....... 111

Table 2.11 Logistic regression analyses of financing activity on adoption of FVO .. 112

Table 2.12 Gains or losses?: number of firms which reposrt gains or losses resulting from

FVO in the first FVO adoption year .......................................................... 113

Table 2.13 Logistic regression analyses of financing activity on adoption of FVO with

only firms which adopted FVO in 2008 (financial vs non-financial firms

seperately) ................................................................................................. 114

Table 2.14 Logistic regression analyses of financing activity on adoption of FVO with

only firms which adopted FVO in 2008 (financial vs non-financial firms

together) ..................................................................................................... 115

Table 2.15 Descriptive statistics (for H4) .................................................................. 116

Table 2.16 OLS regression analyses of exotic features of bonds on adoption of FVO

(Model 4-2a) .............................................................................................. 118

Table 2.17 Logistic regression analyses of exotic features of bonds on adoption of FVO

................................................................................................................... 119

1

Chapter 1: The Impact of Enterprise Resource Planning (ERP) Systems on the Audit

Report Lag

I. INTRODUCTION

The primary purpose of this study is to examine the relationship between the

implementation of enterprise resource planning (ERP) systems and the timeliness of

audited financial statements. Specifically, this study investigates whether the

implementation of ERP systems affects the audit report lag, the time period between a

company’s fiscal year end and the date of the audit report.

It is argued that an ERP system is the most important development in the corporate

use of information technology in the 1990s (Davenport 1998). An ERP system is a

packaged business software system that enables a company to manage resources (material,

human, financial, etc.) more efficiently and effectively by providing an integrated solution

for the organization’s information-processing needs (Nah et al. 2001). Kumar and

Hillegersberg (2000) define ERP systems as information system packages that integrate

information and information-based processes within and across functional areas in an

organization. As a result, ERP systems collect and distribute information more timely and

thus help managers improve their ability to process and analyze information (Hitt et al.

2002). Therefore, ERP systems can radically change the way in which accounting

information is processed, prepared, audited, and disseminated (Brazel and Dang 2008).

The alleged benefits and widespread usage of ERP systems have motivated many

empirical research inquiries over a variety of issues such as success factors of ERP, post-

implementation firm performance, and market reaction. However, a relatively small

amount of research has been conducted on how ERP systems affect the external audit.

2

Because an ERP system is typically the most important investment in information

technology of corporations, this study can provide useful insight on how information

technology (ERP) of a client firm affects the efficiency of external audit, at least to the

extent that audit report lag is a good proxy for audit efficiency.

Our research is motivated by several factors concerning recent audit and IT

environment of corporations. First, the usage of ERP systems is now widespread. ERP

software packages have become popular especially for both large and medium-sized

organizations to overcome the limitations of fragmented and incompatible legacy systems

(Robey et al. 2002). Thus, it is important to thoroughly examine the impacts of ERP

systems. However, scant research has been conducted on the relationship between ERP

implementation and the timeliness of external audits, such as audit report lags. An

interesting aspect of ERP is that even though managers decide to implement ERP systems

for their own operational purposes, ERP implementation has also changed audit

environment affecting the efficiency of audit work. According to Behn et al. (2006),

auditors regard inefficient client closing process and consolidation process as important

impediments to reduce the audit report lag. Also, they point out that poor integration of

different systems within the same company is the biggest system impediment to eliminate

the audit lag. However, some prior studies which investigate the characteristics or benefits

of ERP imply that those problems can be improved by implementing ERP systems

(O’Leary 2004). Thus, it will be interesting to examine the relationship between ERP

implementation and the audit report lag.

Second, there are contradicting implications over how advancements in

technology or ERP systems affect audit report lag. Issuing new corporate filing

3

requirements in 2002 which reduce the number of days allowed for filings of Forms 10-K

and 10-Q (SEC 2002), the Securities and Exchange Commission (SEC) argued that

advancements in information technology have improved the ability of companies to capture,

process, and disseminate their financial information. On the other hand, some studies

suggest that ERP systems can deteriorate the internal control mechanism of corporations.

For example, Lightle and Vallario (2003) point out that integrated ERP systems provide a

single point of control segmentation for segregation of duties, but also provide

opportunities for inappropriately configured access privileges that violate internal control

guidelines. As poorer internal controls might be positively associated with audit delay

(Ashton et al. 1987), this control risk can be a factor which leads to longer audit delay.

Thus, whether ERP systems implementation affects audit delay positively or negatively is

a matter of empirical investigation.

Third, there has been a growing interest in continuous auditing (CA) over the past

twenty years, and CA is increasingly under consideration as a tool to enhance the external

audit (Kuhn and Sutton 2010). However, theoretically CA can be realized by reducing time

to audit by taking advantage of more advanced IT systems, and it is expected that the

movement to CA model will be evolutionary rather than revolutionary (Behn et al. 2006).

Even though this paper neither directly relates ERP systems to CA nor argues that ERP

systems will realize CA, we think that ERP systems are critical investments of firms in

their IT systems, and thus the development and pervasive use of ERP systems can provide

the significant infrastructure necessary for the evolution of the assurance function from the

traditional audit to CA (Kuhn and Sutton 2010). In this regard, examining the relationship

between ERP implementation and the audit report lag can add empirical evidence on

4

whether ERP systems or advances in information technology of client firms can actually

help reduce the time spent by external auditors to complete audit works and furthermore

whether they can make a contribution to the movement toward CA.

The results of this study suggest that ERP implementation is negatively associated

with audit report lag. The test results of the first model indicate that an increase in audit

report lags of post-ERP implementation period for ERP firms is significantly smaller than

that of matched firms. On the other hand, it is not likely that ERP systems are associated

with audit report lags immediately after those systems are implemented. According to the

results of the second model, only the fourth and fifth years after ERP implementation show

significant negative coefficients. This result may suggest that it takes time for ERP systems

to be fully utilized and to make a significant impact on the firm’s accounting systems.

This study contributes to the research in both the accounting information systems

and auditing literatures which examine the effects of ERP systems implementation on a

firm’s information environment. What is interesting in our findings is that implementing

ERP systems can have unintended effects. Managers decide to implement ERP systems in

order to improve operational performances of their firms. However, this paper indicates

that implementing ERP systems can affect not only various measures of firms’

performances but also the external audit which is generally regarded as something

independent of client firms’ decisions. This is the first paper to show empirical evidence

that investments in information technology of client firms can make an impact on the

efficiency of external audit even though the investment is motivated by the management’s

operational purposes. This finding also contributes to auditing literature in that it identifies

an additional determinant of the audit report lag, which is the quality of IT systems of client

5

firms. Identifying the determinants of audit delay is meaningful because understanding the

determinants of audit delay helps us to better explain the auditing processes or behaviors

of auditors. For example, given the results of this study, we may conjecture that external

auditors rely at least partially on the clients’ IT systems when they collect and process data

necessary to carry out their audit tasks. Although Brazel and Dang (2008) report a negative

association between ERP implementation and earnings announcement lags, the earnings

announcement lag cannot properly reflect the audit delay because it is influenced by

management’s incentives (Bamber et al. 1993). Thus, our study more directly examines the

association between ERP implementation and audit delay. In addition, our study indirectly

supports the argument that implementation of ERP systems is positively associated with

the effectiveness of internal controls. From the result that ERP implementation is associated

with shorter audit report lags, we may conjecture that as ERP systems get effectively

utilized external auditors are likely to regard ERP firms as having stronger internal controls

than non-ERP firms. That is because weak internal control indicates high control risk which

may force auditors to conduct more strict audits in order to decrease detection risk,

ultimately leading to an increased audit work.

The remainder of this paper is organized as follows: Section II discusses prior

research and develops our research question; Section III describes data collection process

and research methodology; Section IV presents empirical results; Section V discusses the

analyses with the 2000s’ ERP data; and Section VI concludes by discussing implications,

limitations, and further research.

6

II. LITERATURE REVIEW AND RESEARCH QUESTION

Audit Report Lag

Relevance, along with reliability, is a primary attribute that makes accounting

information useful for decision making (FASB 1980) and one of the critical ingredients of

relevance is timeliness. Being delayed in releasing financial statements affects the

timeliness of information provided (Ashton et al. 1989). Audit delay, which can be

measured by audit report lag, is likely to increase the level of uncertainty associated with

decisions for which the financial statements provide information, and thus can ultimately

impair the usefulness of these reports (Givoly and Palmon 1982). In general, the concept

of timeliness in financial reporting can be defined by two dimensions: the frequency of

reporting, which is the length of reporting period, and the reporting lag (Davies and

Whittred 1980). This study uses the latter one to examine whether ERP implementations

affect the timeliness of financial reports. Thus, the audit report lag in this study is defined

as the time period between a firm’s fiscal year end and the audit report date. For the audit

report dates, we have used the dates when auditors sign on the independent auditors’ reports

of 10-Ks.

There are numerous studies which investigate the determinants of the audit report

lag. One stream of the research on the determinants of the audit report lag is related to

auditors. Using Australian data, Whittred (1980) examines the effect of qualified audit

reports on the timeliness of annual reports. The results show that qualifications are

positively associated with delay in releasing preliminary profits and final annual reports.

They also indicate that the more serious the qualification is, the greater the audit report lag

is. He argues that this result is due to the increase in time spent for the year-end audit and

7

auditor-client negotiations. Newton and Ashton (1989) examine the relationship between

audit structure and audit delay for Canadian Big Eight firms, and find that auditors using

structured audit approaches tend to have greater audit report lag than auditors using

unstructured approaches. Knechel and Payne (2001) relate audit report lag to incremental

audit hours, resource allocation of audit team effort, and provision of non-audit services.

Their results indicate that the presence of controversial tax issues and the use of less

experienced audit staff are positively correlated with audit report lag.

Another research stream on the determinants of the audit report lag is related to

the characteristics of client firms. Using the U.S. firm data, Givoly and Palmon (1982)

show that the reporting lag of individual firms seems to be more closely related to industry

patterns and tradition rather than to firm attributes such as size and complexity of

operations. Using data from an international accounting firm, Ashton et al. (1987)

investigate the relationship between the audit report lag and fourteen different firm

attributes. Among these attributes, five variables show significant relationship with the

report lag: total revenue representing firm size, quality of internal control, public company,

operation complexity, and relative amount of interim work. On the other hand, Ashton et

al. (1989) examine cross-sectional variability in audit delay with a sample of Canadian

firms. They test eight variables and the results indicate that firm size (total asset), industry

(financial service), negative net income, and the existence of extraordinary items are

significantly related to audit delay. Bamber et al. (1993) try to propose a more

comprehensive model of audit report lag and emphasize the importance of the amount of

audit work required, the level of resources expended to complete the audit, and the audit

firm technology. The results show that the explanatory power of their model is almost three

8

times greater than that of previous models.

Besides these studies, Krishnan and Yang (2009) explore the trends in audit report

lags and earning announcement lags, and examine the quality of reporting for the period

from 2001 to 2006. The results show that although the filing lags decreased after 2003

when the new corporate filing requirements became effective, audit lags increased

substantially following the new rules particularly in 2004 and 2005 with the introduction

of SOX Section 404. On the other hand, the long audit report lags do not seem to be

associated with lower quality of reporting, measured by absolute value of discretionary

accruals and quality of accruals. Table 1.1 exhibits the determinants of the audit report lag

which have been identified as significant in some of the prior studies.

[Insert Table 1.1 about here]

Enterprise Resource Planning (ERP) Systems

Empirical research related to ERP systems can be broadly divided into four

streams. First, several studies explore the impacts of ERP systems on corporations. Deloitte

Consulting’s (1998) study suggests the rationale and specific benefits of implementing

ERP, and O’Leary (2004) identifies tangible and intangible benefits. Table 1.2 presents

parts of benefits reported on Deloitte Consulting (1998) and O’Leary (2004). However,

Granlund and Malmi (2002) argue that ERP systems have led to relatively small changes

in management accounting and control procedures.


Second, a significant amount of ERP research has identified the factors necessary

for a successful ERP implementation and an effective ongoing usage. Through an extensive

9

literature review, Nah et al. (2001) identify eleven critical factors for the success of ERP

projects; among those are the teamwork and composition of ERP project team, top

management support, business process reengineering, and effective communication.

Bradley (2008) examines success factors using the classical management theory. The

author suggests that a proper project manager, personnel training, and the presence of a

champion are important to implement ERP systems successfully. In addition, the

organizational culture and top management leadership, including strategic and tactical

conducts, appear to be significantly associated with successful ERP implementation (Ke

and Wei 2008).

Third, a stream of research investigates the relationship between ERP

implementation and post-implementation firm performance. Research findings show

mixed results on this association. For example, Poston and Grabski (2001) find no positive

relation between ERP implementation and overall post-implementation financial

performance, but Hunton et al. (2003) find that although the financial performance of ERP

adopters does not change much, non ERP adopters tend to decrease compared to the

performance of ERP adopters. Nicolaou (2004) indicates that ERP implementation has a

positive association with a long-term financial performance and this relation is stronger

when implementation characteristics are controlled. Nicolaou and Bhattacharya (2006;

2008) also show that both the timing of ERP customizations and carrying of such

modifications in conformity with the adopting organizations’ strategic objectives increase

post-implementation operational performance.

Finally, prior studies have found that in general the market reacts positively to

announcements of ERP implementation. Hayes et al. (2001) have found the overall positive

10

reaction to initial ERP announcements measured by standard cumulative abnormal returns.

This positive reaction seems to be more significant when firms are small and healthy and

when the ERP vendors are large. Hunton et al. (2002) show that analysts react to ERP

implementation plans by revising their earnings forecasts positively. Consistent with Hayes

et al. (2001), this tendency is stronger when firms are financially healthy.

Research Question

Key characteristics of ERP systems are integration, standardization, routinization,

and centralization. These attributes allow managers to access more comprehensive

information on a near real-time basis so that they can make better decisions. However,

implementation of ERP systems affects not only internal information processes but also

external audit environment. Using ERP systems, firms can integrate data from different

departments or business segments, standardize data, improve financial control, and reduce

financial closing cycles and data entry mistakes (O’Leary 2004). These benefits have

potential to remove the main impediments recognized by auditors for the audit delay, such

as poor integration of different systems, poor standardization of data, poor financial control,

and inefficient financial closing (Behn et al. 2006). Thus, the alleged benefits of ERP

systems can imply that the audit report lags are likely to decrease by implementing ERP.

In addition, Morris (2011) finds that ERP firms are less likely to report internal

control weaknesses under SOX Section 404. This may imply that external auditors are more

likely to regard ERP firms as having stronger internal controls compared to non-ERP firms.

According to the audit risk model suggested by the AU Section 312 (2007), audit risk

consists of inherent risk, control risk, and detection risk. Thus, if control risk is low,

11

auditors can increase detection risk, maintaining the same level of overall audit risk. Then,

ceteris paribus, auditors can carry out audit tasks less strictly, and thus the amount of end-

year audit work will decrease. As a result, the audit report lag is likely to decrease.

Therefore, we might conjecture that ERP implementation can lead to shorter audit report

lag. In fact, Ettredge et al. (2006) report an association between material weakness in

internal control and audit delay. With introduction of SOX section 404, attestation of

corporate internal control by auditors and managers became mandatory. Using data on

internal control assessments over financial reporting, Ettredge et al. (2006) find that the

presence of material internal control weakness is associated with longer audit delay. Also,

Pae and Yoo (2001) presents a model which shows that when the auditor’s legal liability is

large the client firm underinvests in the internal control system, and thus the auditor

overinvests effort, leading to an decrease in audit efficiency.

However, some others are against this argument. Grabski et al. (2011) say that the

hidden and complex mechanisms of ERP systems create greater complexity for control and

audit (Grabski et al. 2011). Lightle and Vallario (2003) point out that integrated ERP

systems provide a single point of control segmentation for segregation of duties, but also

provide opportunities for inappropriately configured access privileges to violated internal

control guidelines. These studies imply that ERP implementation can rather increase the

overall control risk. In fact, we have witnessed that several firms had experienced serious

problems in the process of implementing advanced information systems. For example,

building and implementing a new system known as ‘Pulse’, Oxford Health Plans

experienced critical problems which led to nearly out of control on its various business

processes (Hammonds and Jackson 1997). Thus, if we apply the same logic of the audit

12

risk model mentioned earlier, it might be expected that ERP implementation is associated

with longer audit report lag.

It is, therefore, an empirical question whether ERP implementation affects audit

report lag positively or negatively. This paper seeks to provide empirical evidence on this

issue. Thus, the research question examined in this study is:

RQ: Is there an association between ERP implementation and audit report lag?

III. RESEARCH DESIGN

Sample Selection

We used the original Nicolaou (2004) data set of ERP and non-ERP firms in order

to examine our research question. For collecting ERP data, a two-phase process was

employed to identify the sample of ERP firms. First, ERP firms were identified by

extracting ERP implementation announcements from the Lexis-Nexis Academic Universe

(News) Wire Service Reports for the period of January 1, 1990 through December 31, 1998.

Following Hayes et al. (2001), we employed a keyword search method using a combination

of the search terms "implement," "convert," and "contract" with the name of each of the

following ERP vendors: Adage, BAAN, Epicor, GEAC Smartstream, Great Plains,

Hyperion, Intentia Intemational, JBA Intemational, JD Edwards, Lawson, Oracle

Financials, Peoplesoft, QAD, SAP, SSA, or SCT. This key word search method yielded an

initial sample of 1,825 announcements. After eliminating announcements which are not

related to ERP implementation, a total number of 332 announcements remained. Second,

the Global Disclosure database was searched for any mention of ERP system

13

implementation in annual reports and SEC filings. The search term of “enterprise resource

planning” was applied to the full text of 10-Ks and 10-Qs for the time period from January

1, 1990 to February 28, 1999. The initial search identified 1,453 documents. Subsequently,

each of these documents was examined individually. Finally, this process resulted in 131

valid ERP implementation related disclosures. These 131 disclosures are all from firms that

are not included in the original Lexis-Nexis Newswires search. Thus, a total number of 463

firms were identified to have implemented ERP systems in the sample period through our

two-phase process. It is also important to notice that all of these announcements or SEC

filings are related to initial implementations of ERP systems, thus providing important

information on ERP adopters.

A subsequent cross-match with the “Global Vantage Key” (GVKey) file from

Research Insight database resulted in the elimination of 142 firms. Furthermore, 67 firms

were eliminated because financial data for these firms were not available in Compustat

database. In addition, five foreign firms were excluded, along with two other firms that

reported discontinued ERP projects. After these elimination processes, we had the final

sample of 247 firms. The paired t-test results indicate that the two groups from different

data sources are not significantly different in terms of size, measured by total assets and

net sales. Panel A of Table 1.3 presents the distribution of the final sample of 247 firms by

year and source, while Panel B presents the distribution of ERP firms by industry.


Identification of ERP Adoption and ERP Completion Year

ERP project completion time is defined as the system go-live date, the time point

14

at which the system gets ready to use. Of the 247 firms included in the final sample, only

105 firms disclosed both the start and completion years of the ERP implementation. For

the 142 sample firms that did not disclose a completion year, it was assumed that their

expected completion time was equal to the mean expected completion time of 9.92 months,

which is based on reporting by 72 firms from the original Lexis-Nexis newswires sample

and 89 firms from the original Global Disclosure database sample that reported expected

completion times. The year of completion was coded as the time period ‘t0’ to indicate that

this is the starting year of ERP use.

Matching Procedure

Following Barber and Lyon (1996), each ERP firm was matched with a control

group company on both industry and size at the year preceding the ERP adoption year (time

t-1). Firms were first matched by the four-digit Standard Industrial Classification (SIC)

code, and then matched by total assets. There were some instances where it was necessary

to go beyond the four-digit SIC code. For such cases, firms were matched using a three-

digit code. Also, an inability to find proper U.S. matches for certain unique firms resulted

in five lost firms; an example of one such firm is Boeing. As a result, the sample which has

been actually used for subsequent analyses consisted of only 242 firms.

In order to validate the soundness of the matching procedure, we performed a

search of the Lexis-Nexis Newswires and the websites of the major ERP vendors. Through

this procedure, we could find that 12 firms which were originally included in the non-ERP

matched group had been actually implemented ERP systems. These 12 firms were

substituted with other non-ERP firms. Furthermore, a graduate assistant called the IT

15

director or other person in charge of system implementation to make sure that the firms

selected as non-ERP matched firms are actually not ERP adopters. This procedure

eliminated 38 firms which either did not respond or responded that they had implemented

an ERP system in the recent past. These firms were also substituted with others.

Audit Report Lag Data

The data for audit report lags were manually collected with all available 10-Ks of

ERP firms and matched firms for the time period from January 1, 1994 to December 31,

2006. Because 10-Ks before 1994 were not available and the time period of our ERP

sample data is up to 1998, the time period of 1994 to 2006 was used. Initially, 4,330 firm

years were collected: 1,927 firm years for ERP firms and 2,403 firm years for matched

firms. Eliminating the firm years whose control variables used in our models were missing

resulted in the decrease of our sample size to 3,710 firm years. In addition, we deleted firm

years which are more than five years earlier or more than five years later than the ERP

completion year. This elimination gave us a final sample of 3,225 firm years. The final

numbers of firm years for ERP firms and for non-ERP firms are 1,616 and 1,609,

respectively. In case of ERP firms, the number of firm years for pre-ERP period is 863 and

that of post-ERP period is 753. As to non-ERP firms, the number of firm years for pre-ERP

period is 673 and that of post-ERP period is 936. Panel A and B of table 1.4 describes the

distribution of firm years for audit report lag data.


Model Specification

16

We have tested the association between ERP implementation and audit report lag

by estimating log-normal duration models. The standard log-normal model is based on the

assumption that the durations follow a log-normal distribution with density function:

𝑓(𝑡) =1

𝑏𝑡𝜙 (

log(𝑡)−𝑎

𝑏) , 𝑏 > 0

where 𝜙 and Φ denote the standard normal density and distribution functions respectively:

𝜙(𝑡) =1

√2𝜋exp (−

𝑡2

2) and Φ(t) = ∫ 𝜙(𝜏)𝑑𝜏

t

0.

The survivor function and transition rate functions are

𝐺(t) = 1 − Φ(log(𝑡)−𝑎

𝑏)

𝑟(𝑡) =1

𝑏𝑡

𝜙(𝑧𝑡)

1−Φ(𝑧𝑡) with 𝑧𝑡 =

log(𝑡)−𝑎

𝑏

where a=Aα and b=exp(Bβ). A and B denote covariate vectors introduced into the model.

It is assumed that the first component of each of the covariate (row) vectors A and B is a

constant equal to one. The associated coefficient vectors α and β are the model parameters

to be estimated (Blossfeld et al. 2007).

There are several reasons why we chose a log-normal model. First, a simple OLS

model is not appropriate for this study since the audit report lag is positively skewed

(Krishnan and Yang 2009), which is a violation of one of the OLS assumptions for

hypothesis testing. Second, transformation of data can be avoided if we use a log-normal

model. Some prior studies, such as Krishnan and Yang (2011) and Ashton et al. (1987,

1989), use the logarithmic transformation to address the normality issue. However, we try

to avoid this transformation since transformation of data can be a distortion of data. Third,

a log-normal model is a member of the class of accelerated failure time models which can

be interpreted in terms of time duration. Audit report lag is time duration measured by

17

number of days. Thus, it appears to be appropriate for this type of study. Lastly, the hazard

ratio graph of our data has the property that the transition rate initially increases with time

to the maximum and then decreases, which is consistent with the shape of log-normal

transition rate graph. A log-normal model is widely used when the transition rate at first

increases and then, after reaching a maximum, decreases (Blossfeld et al. 2007). For

example, Levinthal and Fichman (1988) examine auditor-client relationships to explore the

dynamics of interorganizational relations by estimating a log-normal model. They find that

interorganizaitonal attachments have positive duration dependence in the initial years of

attachment and negative duration dependence for longer durations. In particular, the results

show that the hazard rate of dissolution in the early stages of attachments increases with

time and after this honeymoon period the rate declines as the tenure of the attachment

increases. Bucklin and Sismeiro (2003) have also used the log-normal model to explore the

browsing behavior of visitors to a Web site. This study focuses on two basic elements of

browsing behavior: a user’s decision to continue browsing or to exit the site and the length

of time spent for each page. They specify a log-normal model for examining page-view

durations and the results indicate that the browsing patterns of the Web site users are

consistent with learning effects, site stickiness, time constraints, and cost-benefit trade-

offs.1

We have two models to provide empirical evidence on our research question. In

the first model, the independent variable of our interest is AfterERP which is an interaction

term of two indicator variables, ERP and After. The variable ERP is an indicator variable

1 Log-normal models have been also employed for studies in other disciplines such as medicine,

food science, and zoology (Strum et al. 2000; McCready et al. 2000; Gamel and McLean 1994;

Hough et al. 2003; Aragao et al. 2007; Tokeshi 1990; Tolkamp and Kyriazakis 1999).

18

which takes 1 if a firm-year is for an ERP firm and 0 otherwise. The variable After is also

an indicator variable which takes 1 if a firm-year belongs to post-ERP period and 0

otherwise. Thus, the variable AfterERP takes 1 only if a firm year is one of the post-ERP

firm years for ERP firms. Consequently, by using an interaction term, we can examine the

difference of differences, and the coefficient of AfterERP shows how much audit report

lags of ERP firms change compared to the change in audit report lags of non-ERP firms.

However, it is not likely that ERP systems make a significant impact on the firms

right away as soon as it becomes implemented. The results of Nicolaou (2004) suggest that

it takes time for ERP systems to significantly affect the accounting systems of companies.

Thus, in the second model, we divide the variable ‘AfterERP’ into five post-ERP years

(from AfterERP1 to AfterERP5) so that we can examine how long it takes for ERP systems

to get associated with the audit report lag.

In addition, both models include various control variables which seem to have

significant impact on audit report lag in prior research. Givoly and Palmon (1982) report

that report lag appears to be associated with industry patterns and tradition, and other

studies indicate that different industries are a significant factor which affects audit report

lag (Ashton et al. 1989; Newton and Ashton 1989; Bamber et al. 1993; Krishnan and Yang

2009). Thus, following Krishnan and Yang (2009), we included four indicator variables to

control different practices and accounting rules across industries: HighLit which represents

high litigation industries, HighGrowth which represents high growth industries, Financial

which represents financial services industry, and HighTech which represents high tech

industries. Prior studies found accounting or firm complexity to be a significant

determinant for audit report lag (Ashton et al. 1987; Ashton et al. 1989; Newton and Ashton

19

1989; Bamber et al. 1993; Krishnan and Yang 2009). Thus, to control the different degree

of complexity in the firm’s accounting and operations, we included variables which

represent the number of business segments (NumSegment), the existence of extraordinary

items (ExtraItem), and the existence of foreign operations (ForeignOper). We also included

Size and Auditor to control different firm size and audit firm effect. In addition, we

controlled different fiscal year end by including BusyEnd that is an indicator variable which

distinguish firms whose fiscal year end is December or January from firms whose fiscal

year end is the other months. Furthermore, financial distress seems to affect the audit report

lag (Ashton et al. 1989; Bamber et al. 1993; Krishnan and Yang 2009). Financial distress

is proxied by variables representing negative earnings (Loss) and debt to total asset ratio

(Leverage). Audit opinion is also found to be a significant determinant in several prior

studies (Bamber et al. 1993 and Krishnan and Yang 2009). We have captured audit opinion

with the variable of AuditOpi.2 Finally, we controlled the time trend of audit report lags by

including year dummies. Table 1.5 summarizes the definitions of variables used in this

study.


IV. RESULTS

Descriptive Statistics

Table 1.6 reports the descriptive statistics, where Panel A, B, and C display the

2 The indicator variable ‘AuditOpi’ takes 1 if the firm receives an opinion other than clean opinion,

0 otherwise. We used ‘AUOP’ (Auditor Opinion) item in Compustat for this variable. When

constructing this variable, we coded only ‘unqualified opinion’ as clean opinion and ‘unqualified

opinion with additional language’ was not coded as clean opinion because the additional language

can affect the time that auditors need to spend to complete the audit.

20

distributional properties of the variables used in this study. Panel A shows the descriptive

statistics of all samples including both ERP firms and non-ERP firms. The mean of audit

report lags is 46.2 days. This figure is similar to means of lags of year 2001 or year 2002

of Krishnan and Yang (2009) data and about 22 days shorter than one reported in Knechel

and Payne (2001) that use data of year 1991. Also, the descriptive statistics indicate that

about sixty percent of firm years have fiscal year end month of December or January and

more than ninety percent of firm years were audited by one of big six auditors. Also, less

than thirty percent of firm years received other than clean opinion from auditors. Panel B

exhibits the descriptive statistics of ERP firms versus non-ERP firms. The descriptive

statistics indicate that ERP firms are bigger than non-ERP firms, more likely to have

foreign operations and have fewer business segments. The audit report lags of two samples

do not seem to be different in average. Panel C presents descriptive statistics of pre- and

post- ERP implementation periods. Numerous variables of post-ERP implementation

period are significantly bigger than those of pre-ERP implementation period: number of

segments, incidence of extraordinary items, size, incidence of negative earnings, firms

audited by big six auditors, firms receiving other than a clean opinion, and firms operating

abroad. Also, the mean of audit report lags of post-ERP implementation period is

significantly longer than that of pre-ERP implementation period. Panel D reports the pair-

wise Pearson correlations and Spearman correlations. The table indicates that audit report

lag is negatively correlated with high litigation industry, high tech industry, size, big six

auditors, and foreign operations, and positively correlated with net loss and modified

opinions.


21

Multivariate Results

We examine how ERP implementation affects the audit report lag by estimating

two log-normal duration models. The test results of the first model are presented in Table

1.7. The results show a negative and significant coefficient for the interaction term,

AfterERP, which indicates that ERP implementation is negatively associated with the

length of audit report lag. The coefficients of other control variables are consistent with

prior studies in general. High litigation industry, high tech industry, and client size seem to

be negatively associated with the audit report lag. It seems that the negative association

between the audit report lag and high litigation and high tech industries is due to the

concern of the market on firms that belong to risky industries. It is expected that being

aware of this concern of the market, those firms try to relieve the market’s anxiety by

releasing annual reports more timely. On the other hand, presence of extraordinary items,

negative earnings, and modified audit opinion are positively associated with the audit

report lag. The dummy variable for year 2004 is positive and strongly significant. This can

be attributed to the SOX Section 404 of reporting requirements which became effective in

2004. The SOX section 404 requires managers and auditors to assess the soundness of

corporate internal control, an attestation which was not mandatory before the regulation.

The introduction of this new regulation probably leads to increased audit hours to be spent

in order to complete the audit. The result for year 2004 dummy is also consistent with

Ettredge et al. (2006) and Krishnan and Yang (2009) that report dramatic increases in audit

report lags in 2004, 19.8 days (from 50.3 to 70.1) and 23.3 days (from 49.3 to 72.6) in

average respectively.

22


Table 1.8 reports the test results for the second model. The coefficients of the post-

ERP 4th and 5th year interaction terms are negative and significant. Even though the

coefficients of the other three post-ERP period terms are not significant, they are all

negative. This result implies that ERP systems may help decrease the audit report lag, but

it takes time for the significant impact on the firms’ accounting systems to be realized. The

results for other control variables are mainly the same as those of the first model.


Additional Tests

There exists possibility that the auditors in the late 1990s might have more

knowledge or better understanding on ERP systems than in the early 1990s because ERP

systems began to be introduced in the early 1990s and as time passes the auditors may get

accustomed to those systems. Thus, we checked whether there is knowledge effect (or

education effect) of auditors by examining whether there is any difference in impacts of

ERP on ARLs between early ERP adopters and late ERP adopters. However, we couldn’t

find any significant differences between two groups (untabulated).3

In addition to two log-normal models, we ran OLS regressions using the natural

logarithm of audit report lag, rather than the audit report lag itself, as a dependent variable.

The results for both models are qualitatively the same as those of log-normal models

(untabulated). The coefficient of the interaction term in the first model is negative and

3 We divided ERP firms into two groups, early adopters and late adopters. However, because the

dividing point is not clear, we examined three pairs of groups: pre and post-1994, 1995, and 1996

pairs. Nevertheless, we couldn’t find significant results for all three pairs.

23

significant at five percent level. The coefficients of the post-ERP 4th and 5th year interaction

terms in the second model are also negative and significant. The results for control

variables are very similar to those of log-normal models.

Furthermore, we examined the VIF values of the independent variables to check

whether there is a serious multicollinearity problem. However, we could not find an

indication of multicollinearity. The VIF values for all variables do not exceed five.

V. ANALYSES WITH ERP DATA OF THE 2000S

ERP data of the 2000s

We tried to collect additional information on the firms that have implemented ERP

systems in the 2000s, from year 2000 to year 2008, in order to check whether our results

hold with more recent data. The data collection method is similar to that employed for

collecting our main ERP data. We used Lexis-Nexis Newswires to search for any

announcements relevant to ERP implementation and searched for any mentions of ERP

implementation in SEC filings. This search process yielded 786 unique firms which had

ERP-related announcements or mentions related to ERP implementation in SEC filings

during the search period. However, 151 firms were eliminated because those firms are not

available in Compustat. In order to collect information on ARLs, we employed the Audit

Analytics database instead of collecting manually. Additional 43 firms were eliminated

which are not available in Audit Analytics or whose data for ARLs are not available in

Audit Analytics. Consequently, we had the final sample of 592 firms. Finally, the ERP

firms were matched by the four-digit SIC code and total assets to form a non-ERP control

group. The firm years were selected as they had been for our main ERP data. These

24

processes gave us a final sample of 8,622 firm years.

Multivariate Results

The models estimated with the recent ERP data are the same as those used for the

main ERP data, and the test results are presented in Table 1.9 and 1.10. We could find

results which are consistent with those from our main ERP data, which makes the reliability

of the results even stronger. Table 9 reports the results of the first model. The results show

that the coefficient of the interaction term, AfterERP, is negative and significant at 1 percent

level. The results for other control variables are similar to those from the main data. In

general, the signs of coefficients are the same, but they tend to be more significant than

those of the previous corresponding model. For example, the coefficients of NumSegment

and BusyEnd which were positive and insignificant are now positive and significant. The

coefficients for year dummies are all significant, but the sign turns to positive from negative

in year 2004, which has a similar implication to that of previous results, the effect of SOX

section 404. One possible reason for the increase in statistical significance is probably due

to the increase in the number of observations estimated.


Table 10 reports the results for the second model using the recent ERP data. The

coefficients of all five interaction terms are negative, but in contrast to the results of the

previous corresponding model, they are all statistically significant. We can find one

possible reason for this result from the difference in (IT) audit environments between 1990s

and 2000s. In the 1990s when ERP systems were first introduced and both implementing

firms and auditors were not familiar with those systems, it might take much more time for

25

firms and auditors to learn about how they work and to assess their reliability. However, in

the 2000s ERP systems are not new anymore to auditors and probably even to

implementing firms. In other words, their knowledge level on ERP systems in the 2000s

may be much higher than that in the 1990s. Thus, we can conjecture that it might not take

much time for auditors to understand and effectively utilize ERP systems for their audit.

Another possible reason can arise from the possibility that those ERP announcements in

the 2000s are not indications of initial ERP implementations, rather just additions of some

more systems to or upgrades of existing ERP systems. This is more likely especially for

the 2000s’ data than 1990s’ data. The period of the 1990s is the booming period of

implementing ERP systems. The popularity of ERP did soar in the early 1990s and firms

began to invest billions in ERP systems (Chen 2001). Thus, it is likely that most major

companies were already using ERP systems in the early 2000s. Therefore, we cannot deny

the possibility that a significant portion of ERP-related announcements in the 2000s can be

related to additions of some more systems to existing ERP systems or upgrades of them. If

this is the case, auditors wouldn’t need to spend time in order to understand those

modifications to the systems as much as when the client firms are first implementing ERP

systems. Thus, it would also take less time for those modifications to make a significant

impact on the firms’ accounting systems.


Limitations of Data

In spite of our efforts to collect more recent ERP data, we should confess that this

recent ERP data can be noisy. As mentioned above, because it is highly likely that most

26

major firms had already implemented ERP systems until the early 2000s, it is more difficult

to find proper matched non-ERP firms in the 2000s than in the 1990s. Also, when collecting

ERP data, it is very difficult to distinguish between initial ERP implementations and

different sorts of system modifications such as upgrades, additions of more systems, or

system integrations of foreign branches. This problem is more pronounced in the 2000s’

data because the 1990s’ data includes only initial implementations (as mentioned in the

sample selection section) and it is more likely that ERP-related announcements in the 2000s

are just additions of some more systems to existing ERP systems or upgrades of them

because of the aforementioned reason.

Despite of these potential disadvantages of the 2000s’ ERP data, we however think

that the data still provide interesting information for a different time period. Also, the

results of this data make our main conclusion of negative association between ERP

implementations and ARLs even stronger in that overall they are quite consistent with those

of the 1990s’ data. Nevertheless, we need to be careful in interpreting the results from the

2000s’ data, especially for the second model, because as mentioned above the data can be

noisy and there is a discrepancy between the results of the second model and prior ERP

studies. Prior studies, in general, report that it takes time for ERP systems to make an

impact on implementing firms.

VI. CONCLUSION

Implications

We investigated whether ERP implementation is associated with the audit report

lag. The audit report lag is affected by numerous factors as shown in prior research. Recent

27

advance in technology has radically changed business operation and processes. Meanwhile,

it has also changed audit environment and can affect the determinants of the audit report

lag. This paper is the first study to show empirical evidence that investment in IT of client

firms can make an impact on the external audit even though the investment is motivated

by managers’ operational purposes. The test results of this study indicate that the ERP

implementation has negative association with the audit report lag. Given that an ERP

system is typically the most important investment in information technology for a company,

this result has an implication that advanced technology of client firms can have a positive

influence on the efficiency of external audit, at least to the extent that audit report lag is a

good proxy for audit efficiency. Also, the test result of our second model suggests that it

takes time for ERP systems to be fully utilized and to make a significant impact on the

accounting systems. Our findings are interesting in that implementing ERP systems can

make unintended effects on the external audit. Companies decide to implement ERP

systems to improve their own operational performances. However, our results indicate that

a decision to implement ERP systems can affect not only various measures of firms’

performances but also the external audit which is generally regarded as something

independent of client firms’ decisions.

We might be able to find possible reasons for these results from the characteristics

of ERP systems. In the accounting aspect, ERP systems can be characterized as integrated

and comprehensive enterprise-wide record keeping systems which embrace most activities

and processes of an organization. Thus, the firms with ERP systems can be equipped with

more efficient accounting closing and consolidation processes. Under the ERP

environment, auditors first need to examine the reliability of systems, which means the

28

adequacy of system design and compliance. Once the results of this examination are

acceptable, auditors can save time spent on collecting data from different departments or

business segments and on confirming management assertions because they can be provided

with comprehensive, integrated, and reliable information much more quickly. Our test

results might indicate that auditors are realizing these benefits of ERP systems.

The SEC mandated accelerated filing requirements for corporate 10-K and 10-Q

filings in 2003 (SEC 2002). The requirements reduced the 10-K filing period for

accelerated filers from 90 days to 75 days, and for large accelerated filers the filing period

was further reduced to 60 days. Concerning this filing environment, some might argue that

auditors might not have as much incentive as before to complete audit works sooner than

required periods and thus people might not be interested in audit delay as much as before.

Those arguments might be true. However, the main finding of this paper is that auditors

are more efficient for ERP firms, and this can mean that even if firms file at the same date

the audit quality will be higher for ERP firms. Also, the audit efficiency is closely related

to auditors’ costs. Thus, the audits can be more profitable for ERP firms as the audit costs

may be lower because of the increased efficiency. Therefore, the finding of this paper still

provides implications on different aspects of audit under the current filing rule. Research

questions related to audit fee or audit quality may be worth examining.

Prior studies focus on effects of ERP systems only on firm performances, such as

operational performances or market performances, which are managers’ intended

(expected) effects. However, this paper implies possibility that implementing advanced IT

systems can have unexpected impacts on different aspects including outsiders of the

implementing firms. Therefore, this study provides an implication that as more advanced

29

and sophisticated IT systems get introduced related parties should endeavor to understand

all potential effects of the systems more comprehensively so that they can maximize the

benefits of the systems.

Limitations

This study has some limitations. First, because ERP related announcements, which

are the source of data for this study, are not detailed enough, we couldn’t take into

consideration different degrees of integration of a firm’s IT systems. If the ERP systems

are implemented only in financials and not integrated with logistics, MRP or CRM, or if

the subsidiaries are not included in the implementation, the benefits of ERP for the external

audit will be limited. However, we couldn’t distinguish different degrees of integration due

to the lack of information. Future research with more complete and thorough information

on the integration of ERP systems might be more convincing.

Additionally, more comprehensive and in-depth research is needed about how

advance in technology of client firms affects the behaviors of external auditors or about

what the true factors are which improve audit efficiency. There is a concern on the change

in audit environment caused by the advance in technology. It is argued that auditors do not

properly recognize the heightened degree of risk in ERP systems regarding internal control,

such as network security, database security, and application security (Hunton et al. 2004).

Hunton et al. (2004) say that auditors do not have enough knowledge on systems and ERP

audits should be performed by a cross-functional team which consists of both auditors and

IS experts. If this argument is true, it could be conjectured that auditors trust IT systems of

client firms without proper evaluation on them and reduce their audit work. Then, this is

30

not the realization of benefits of ERP, but rather they are increasing the efficiency of audit

at the expense of effectiveness of audit. Thus, future research that deals with this issue will

be interesting and meaningful.

Further Research

Morris and Laksmana (2010) indicate that ERP implementation is negatively

associated with earnings management activities. Earnings management by client firms

increases the detection risk of auditors and thus ultimately increases the audit risk. Pratt

and Stice (1994) and Simunic and Stein (1996) suggest that one of the responses of auditors

to risk (litigation risk) is audit fee adjustment. Also, it is obvious that the time which

auditors spend to carry out audit tasks is a critical factor which affects the level of audit

fees. Thus, jointly the results of Morris and Laksmana (2010) and of this study, we may

conjecture that ERP implementation is negatively associated with audit fees. By examining

this additional hypothesis, we may be able to understand more comprehensively how

advances in technology of client firms affect the environment of external audit.

31

REFERENCES

Aragao, G. M. F., M. G. Corradini, M. D. Normand, and M. Peleg. 2007. Evaluation of the

weibull and log normal distribution functions as survival models of Escherichia Coli

under isothermal and non isothermal conditions. International Journal of Food

Microbiology 119: 243-257.

American Institute of Certified Public Accountants (AICPA). 2007. Audit Risk and

Materiality in Conducting an Audit. AU Section 312. New York, NY: AICPA.

Ashton, R. H., J. J. Willingham, and R. K. Elliott. 1987. An empirical analysis of audit

delay. Journal of Accounting Research 25 (Autumn): 275–292.

Ashton, R. H., P. R. Graul, and J. D. Newton. 1989. Audit delay and the timeliness of

corporate reporting. Contemporary Accounting Research 5 (Spring): 657–673.

Bamber, E. M., L. S. Bamber, and M. P. Schoderbek. 1993. Audit structure and other

determinants of audit report lag: An empirical analysis. Auditing: A Journal of

Practice & Theory 12 (Spring): 1–23.

Blossfeld, H., K. Golsch, and G. Rohwer. 2007. Event history analysis with Stata.

Lawrence Erlbaum.

Bradley, J. 2008. Management based critical success factors in the implementation of

enterprise resource planning systems. International Journal of Accounting

Information Systems 9 (3): 175–200.

Brazel, J. F., and L. Dang. 2008. The effect of ERP system implementations on the

management of earnings and earnings release dates. Journal of Information Systems

22 (2): 1-21.

Behn, B. K., D. L. Searcy, and J. B. Woodroof. 2006. A within firm analysis of current and

expected future audit lag determinants. Journal of Information Systems 20 (1): 65-

86.

Bucklin, R. E., and C. Sismeiro. 2003. A model of web site browsing behavior estimated

on clickstream data. Journal of Marketing Research 40 (3): 249-267.

Chen, I. J. 2001. Planning for ERP systems: Analysis and future trend. Business Process

Management Journal 7 (5): 374-386.

Davenport, T. H. 1998. Putting the enterprise into the enterprise system. Harvard Business

Review 76 (July/August): 121-131.

32

Davis, B., and G. P. Whittred. 1980. The association between selected corporate attributes

and timeliness in corporate reporting: further analysis. A Journal of Accounting,

Finance and Business Studies 16 (June): 48-60.

Deloitte Consulting. 1998. ERP’s Second Wave. New York, NY: Deloitte Consulting.

Ettredge, M. L., C. Li, and L. Sun. 2006. The impact of SOX section 404 internal control

quality assessment on audit delay in the SOX era. Auditing: A Journal of Practice &

Theory 25 (2): 1-23.

Financial Accounting Standards Board (FASB). 1980. Qualitative Characteristics of

Accounting Information. FASB Concept No. 2. Norwalk, CT: FASB.

Gamel, J. W., and I. W. McLean. 1994. A stable, multivariate extension of the log-normal

survival model. Computers and Biomedical Research 27: 148-155.

Givoly, D., and D. Palmon. 1982. Timeliness of annual earnings announcements: Some

empirical evidence. The Accounting Review LVII (3): 486-508.

Grabski, S. V., S. A. Leech, and P. J. Schmidt. 2011. A review of ERP research: a future

agenda for accounting information systems. Journal of Information Systems 25 (1):

37-78.

Granlund, M., and T. Malmi. 2002. Moderate impact of ERPs on management accounting:

A lag or permanent outcome? Management Accounting Research 13 (3): 299–321.

Hammonds, K. H., and S. Jackson. 1997. Behind Oxford’s billing nightmare.

BusinessWeek. November 17. Available at:

http://www.businessweek.com/1997/46/b3553148.htm.

Hayes, D. C., J. E. Hunton, and J. L. Reck. 2001. Market reaction to ERP implementation

announcements. Journal of Information Systems 15 (1): 3–18.

Hitt, L. M., D. J. Wu, and X. Zhou. 2002. Investment in enterprise resource planning:

Business impact and productivity measures. Journal of Management Information

Systems 19 (Summer): 71-98.

Hough, G., K. Langohr, G. Gomez, and A. Curia. 2003. Survival analysis applied to

sensory shelf life of foods. Journal of Food Science 68 (1): 359-362.

Hunton, J. E., A. M. Wright, and S. Wright. 2004. Are financial auditors overconfident in

their ability to assess risks associated with enterprise resource planning systems?

Journal of Information Systems 1 (2): 7–28.

http://www.businessweek.com/1997/46/b3553148.htm

33

Hunton, J. E., B. Lippincott, and J. L. Reck. 2003. Enterprise resource planning systems:

Comparing firm performance of adopters and non-adopters. International Journal of

Accounting Information Systems 4 (3): 165–184.

Hunton, J. E., R. A. McEwen, and B. Wier. 2002. The reaction of financial analysts to

enterprise resource planning (ERP) implementation plans. Journal of Information

Systems 16 (1): 31–40.

Ke, W., and K. K. Wei. 2008. Organizational culture and leadership in ERP

implementation. Decision Support Systems 45 (2): 208–218.

Khun, J. R., and S. G. Sutton. 2010. Continuous audit in ERP system environments: The

current state and future directions. Journal of Information Systems 24 (1): 91-112.

Knechel, W. R., and J. L. Payne. 2001. Additional evidence on audit report lag. Auditing:

A Journal of Practice & Theory 20 (May): 137–147.

Krishnan, J., and J. S. Yang. 2009. Recent trends in audit report and earnings

announcement lags. Accounting Horizons 23 (3): 265-288.

Kumar, K., and J. V. Hillegersberg. 2000. ERP experiences and evolution.

Communications of the ACM (April): 22–26.

Levinthal, D. A., and M. Fichman. 1988. Dynamics of interorganizational attachments:

Auditor-client relationships. Administrative Science Quarterly 33 (3): 345-369.

Lightle, S. S., and C. W. Vallario. 2003. Segregation of duties in ERP. The Internal Auditor

60 (5): 27-31.

McCready, D. R., J. W. Chapman, W. M. Hanna, H. J. Kahn, D. Murray, E. B. Fish, M. E.

Trudeau, I. L. Andrulis, and H. L. A. Lickley. Factors affecting distant disease-free

survival for primary invasive breast cancer: Use of a log-normal survival model.

Annals of Surgical Oncology 7 (6): 416-426.

Morris, J. J. 2011. The impact of enterprise resource planning (ERP) systems on the

effectiveness of internal controls over financial reporting. Journal of Information

Systems 25 (1): 129-157.

Morris, J. J., and I. Laksmana. 2010. Measuring the impact of enterprise resource planning

(ERP) systems on earnings management. Journal of Emerging Technologies in

Accounting 7: 47-71.

Nah, F., J. Lau, and J. Kuang. 2001. Critical factors for successful implementation of

34

enterprise systems. Business Process Management 7 (3): 285–296.

Newton, J. D., and R. H. Ashton. 1989. The association between audit technology and audit

delay. Auditing: A Journal of Practice and Theory 8 (Supplement): 22-37.

Nicolaou, A. I. 2004. Firm performance effects in relation to the implementation and use

of enterprise resource planning systems. Journal of Information Systems 18 (2):79-

105.

Nicolaou, A. I. and S. Bhattacharya. 2006. Organizational performance effects of ERP

systems usage: The impact of post-implementation changes. International Journal of

Accounting Information Systems 7 (1): 18-35.

Nicolaou, A.I. and S. Bhattacharya. 2008. Sustainability of ERP performance outcomes:

The role of post-implementation review quality. International Journal of Accounting

Information Systems 9 (1): 43-60.

O’Leary, D. E. 2004. Enterprise resource planning (ERP) systems: An empirical analysis

of benefits. Journal of Emerging Technologies in Accounting 1: 63-72.

Pae, S., and S. Yoo. 2001. Strategic interaction in auditing: An analysis of auditors’ legal

liability, internal control system quality, and audit effort. The Accounting Review 76

(3): 333-356.

Poston, R., and S. Grabski. 2001. Financial impacts of enterprise resource planning

implementations. International Journal of Accounting Information Systems 2 (4):

271–294.

Pratt, J., and J. D. Stice. 1994. The effects of client characteristics on auditor litigation risk

judgements, required audit evidence, and recommended audit fee. The Accounting

Review 69 (October): 639-656.

Robey, D., J. W. Ross, and M. C. Boudreau. 2002. Learning to implement enterprise

systems: An exploratory study of the dialectics of change. Journal of Management

Information Systems 19 (1): 17–46.

Securities and Exchange Commission (SEC). 2002. Acceleration of Periodic Report Filing

Dates and Disclosure Concerning Website Access to Reports. Release No. 33-8128.

Washington, D.C.: SEC. Available at: http://www.sec.gov/rules/final/33-8128.htm.

Simunic, D. A., and M. T. Stein. 1990. Audit risk in a client portfolio context.

Contemporary Accounting Research 6 (Spring): 329-343.

http://www.sec.gov/rules/final/33-8128.htm

35

Strum, D. P., J. H. May, and L. G. Vargas. 2000. Modeling the uncertainty of surgical

procedure times. Anesthesiology 92 (4): 1160-1167.

Tokeshi, M. 1990. Niche apportionment or random assortment: Species abundance patterns

revisited. Journal of Animal Ecology 59 (3): 1129-1146.

Tolkamp, B. J., and I. Kyriazakis. To split behavior into bouts, log-transform the intervals.

Animal Behaviour 57 (4): 807-817.

Whittred, G. P. 1980. Audit qualification and the timeliness of corporate annual reports.

The Accounting Review LV (4): 563-577.

36

Tables for Chapter 1

Table 1.1

Determinants of Audit Report Lag

(1) (2) (3) (4) (5) (6)

Audit

client

related

Client size + - - - -

Net loss + + +

Poor financial condition +

Extraordinary Item + + + +

Operational complexity + +

Public company - -

Magmt incentive to provide timely reports -

Quality of Internal Control -

Industry (financial) - - - -

Auditor

related

Opinion qualification + +

Audit technology + +

Audit team allocation (partner or manager) -

Year end (Dec, or Dec and Jan) - +

Interim work - -

(1): Ashton et al. 1987, data of international accounting firm

(2): Ashton et al. 1989, data of Canadian firms

(3): Newton and Ashton 1989, data of Canadian firms

(4): Bamber et al. 1993, data of U.S. firms

(5): Knechel and Payne 2001, data of international accounting firm

(6): Kirishnan and Yang 2009, data of U.S. firms

37

Table 1.2

Tangible and Intangible Benefits of ERP Systems

Tangible Benefits Intangible Benefits

Inventory Reduction Information Access/Visibility

Personnel Reduction New Improved Processes

Productive Improvements Customer Responsiveness

Order Management Improvements Integration

Financial Close Cycle Reduction Standardization

IT Cost Reduction New Reports/Reporting Capability

Cash Management Improvement Sales Automation

Revenue/Profit Increases Financial Control

Maintenance Reductions No Redundant Data Entry

38

Table 1.3

Sample Selection and Identification of ERP firms

Panel A: Distribution of ERP implementations by year in sample period

Implementation

year Number of valid

ERP anncmts Number of ERP firms with

GVKey identification

Number of ERP firms

with available data in

Compustat

Source: Lexis-Nexis Newswires

98 141 69 52

97 83 55 44

96 43 30 21

95 19 15 11

94 32 19 17

93 3 4 2

92 4 4 4

91 3 2 2

90 4 2 2

Sum 332 200 155

Source: Global Disclosure

98 49 46 39

97 66 62 41

96 12 10 8

90-95 4 3 4

Sum 131 121 92

Grand Sum 463 321 247

Panel B: Distribution of ERP firms by industry

ERP firms represented by the following SIC codes Number of firms

0700 - Agricultural Services 3

1000 - Mining, Construction 5

2000 - Manufacturing 56

3000 - Manufacturing 127

4000 - Transportation, Communications, Utilities 12

5000 - Wholesale and Retail Trade 18

6000 - Finance, Insurance, Real Estate 10

7000 - Services 14

8000 - Health Services 2

9000 - Public Institutions 0

Total 247

39

Table 1.4

Distribution of firm years for audit report lags

Panel A: Distribution of firm years by year

Year Number of firm years

1993 22

1994 161

1995 237

1996 352

1997 425

1998 422

1999 368

2000 340

2001 294

2002 260

2003 219

2004 119

2005 4

2006 2

Sum 3225

Panel B: Distribution of firm years by ERP project completion year

Implementation

Year Number of firm years of audit report lags

ERP Firms Matched Firms

Pre-ERP

Years

-5 44 16

-4 87 45

-3 145 91

-2 196 145

-1 201 180

0* 190 196

Sum 863 673

Post-ERP

Years

+1 182 194

+2 156 194

+3 151 190

+4 139 185

+5 125 173

Sum 753 936

*denotes ERP project completion year.

40

Table 1.5

Variable Definitions

Audi Report Lag = the number of days from the fiscal year-end to the audit report date;

ERP = 1 if the firm is in the ERP firm data, 0 otherwise;

After = 1 if the firm year is an post-ERP implementation year, 0 otherwise;

AfterERP = Interaction term of ERP and After;

After1 to After5 = Indicator variables representing the first, second, third, fourth, and fifth year

after the ERP project completion year, respectively;

AfterERP1 to

AfterERP5 = Interaction terms between ERP and After1 to After5, respectively;

HighLit = 1 if the firm belongs to industries 28, 35, 36, 38, 60, 67, and 73 (two-digit

SIC code), 0 otherwise;

HighGrowth = 1 if the firm belongs to industries 35, 45, 48, 49, 52, 57, 73, 78, and 80 (two-

digit SIC code), 0 otherwise;

Financial = 1 if the firm belongs to industry 60-67 (two-digit SIC code), 0 otherwise;

HighTech = 1 if the firm belongs to industries 283, 284, 357, 366, 367, 371, 382, 384, and

737 (three-digit SIC code), 0 otherwise;

NumSegment = the number of business segments (Compustat historical segments file);

ExtraItem = 1 if the firm reports extraordinary items, 0 otherwise;

Size = natural log of total assets;

BusyEnd = 1 if the firm’s fiscal year ends in December or January, 0 otherwise;

Loss = 1 if the firm reports a loss before extraordinary items, 0 otherwise;

Leverage = debt to total assets ratio;

Auditor = 1 if the firm is audited by a Big 6 auditor, 0 otherwise;

AuditOpi = 1 if the firm receives an opinion other than clean opinion, 0 otherwise

(Compustat item ‘AUOP’);

ForeignOper = 1 if the firm has foreign operations, 0 otherwise; and

GC = 1 if the firm receives a going concern opinion, 0 otherwise (Audit Analytics).

41

Table 1.6


Panel A: Distributional Properties of Variables (All Sample), N=3225

Variable Mean Median Std Dev 1Q 3Q

Audit report lag 46.244 41 33.002 29 54

ERP 0.501 1 0.5 0 1

After 0.523 1 0.499 0 1

AfterERP 0.233 0 0.423 0 0

After1 0.116 0 0.32 0 0

After2 0.108 0 0.311 0 0

After3 0.105 0 0.307 0 0

After4 0.1 0 0.3 0 0

After5 0.092 0 0.289 0 0

HighLit 0.465 0 0.498 0 1

HighGrowth 0.222 0 0.415 0 0

Financial 0.026 0 0.16 0 0

HighTech 0.34 0 0.473 0 1

NumSegment 2.47 2 1.904 1 4

ExtraItem 0.143 0 0.35 0 0

Size 6.453 6.386 1.948 4.977 7.829

BusyEnd 0.622 1 0.484 0 1

Loss 0.267 0 0.442 0 1

Leverage 0.964 0.557 19.225 0.386 0.715

Auditor 0.908 1 0.287 1 1

AudOpi 0.294 0 0.455 0 1

Foreign 0.584 1 0.492 0 1

(continued on the next page)

42


TABLE 1.6 (continued)

Panel B: Distributional Properties of Variables (ERP firm sample vs Matched firm sample)a

ERP Firm Sample, N=1616 Matched Firm Sample, N=1609

Variable Mean Median Std Dev 1Q 3Q Mean Median Std Dev 1Q 3Q

Audit report lag 44.939 41 29.096 29 54 47.553 41 36.47 30 55

After 0.465 0 0.498 0 1 0.581 1 0.493 0 1

After1 0.112 0 0.316 0 0 0.12 0 0.325 0 0

After2 0.096 0 0.295 0 0 0.12 0 0.325 0 0

After3 0.093 0 0.291 0 0 0.118 0 0.322 0 0

After4 0.086 0 0.28 0 0 0.114 0 0.319 0 0

After5 0.077 0 0.267 0 0 0.107 0 0.309 0 0

HighLit 0.454 0 0.498 0 1 0.476 0 0.499 0 1

HighGrowth 0.219 0 0.414 0 0 0.224 0 0.417 0 0

Financial 0.038 0 0.192 0 0 0.014 0 0.118 0 0

HighTech 0.334 0 0.471 0 1 0.346 0 0.475 0 1

NumSegment 2.401 1 1.897 1 3 2.54 2 1.9089 1 4

ExtraItem 0.136 0 0.343 0 0 0.149 0 0.356 0 0

Size 6.534 6.474 2.019 5.007 7.953 6.372 6.309 1.87 4.916 7.752

BusyEnd 0.624 1 0.484 0 1 0.621 1 0.485 0 1

Loss 0.255 0 0.436 0 1 0.28 0 0.449 0 1

Leverage 0.562 0.562 0.265 0.379 0.718 1.368 0.55 27.215 0.39 0.714

Auditor 0.916 1 0.276 1 1 0.901 1 0.298 1 1

AudOpi 0.3 0 0.458 0 1 0.288 0 0.453 0 1

Foreign 0.621 1 0.485 0 1 0.546 1 0.497 0 1

43



Panel C: Distributional Properties of Variables (Pre-ERP vs Post-ERP)a

Pre-ERP Implementation, N=1536 Post-ERP Implementation, N=1689

Variable Mean Median Std Dev 1Q 3Q Mean Median Std Dev 1Q 3Q

Audit report lag 44.033 40 27.962 29 51 48.254 41 36.888 29 58

ERP 0.561 1 0.496 0 1 0.445 0 0.497 0 1

HighLit 0.439 0 0.496 0 1 0.488 0 0.5 0 1

HighGrowth 0.209 0 0.407 0 0 0.233 0 0.423 0 0

Financial 0.033 0 0.18 0 0 0.019 0 0.138 0 0

HighTech 0.317 0 0.465 0 1 0.361 0 0.48 0 1

NumSegment 2.008 1 1.553 1 3 2.891 2 2.088 1 4

ExtraItem 0.107 0 0.309 0 0 0.175 0 0.38 0 0

Size 6.27 6.153 1.866 4.873 7.521 6.62 6.56 2.005 5.094 8.139

BusyEnd 0.608 1 0.488 0 1 0.635 1 0.481 0 1

Loss 0.203 0 0.402 0 0 0.326 0 0.468 0 1

Leverage 0.6 0.555 0.991 0.391 0.708 1.295 0.56 26.548 0.381 0.722

Auditor 0.878 1 0.326 1 1 0.936 1 0.244 1 1

AudOpi 0.234 0 0.423 0 0 0.349 0 0.476 0 1

Foreign 0.55 1 0.497 0 1 0.615 1 0.486 0 1

44


Panel D: Pearson Correlations (top) and Spearman Correlation (bottom)b, N=3225

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Audit report lag 1 -0.04 0.06 -0.07 0.02 -0.02 -0.09 -0.02 0.02 -0.23 -0.01 0.20 0.03 -0.08 0.14 -0.13

ERP 2 -0.03 -0.12 -0.02 -0.01 0.08 -0.01 -0.04 -0.02 0.04 0.00 -0.03 -0.02 0.03 0.01 0.08

After 3 0.05 -0.12 0.05 0.03 -0.04 0.05 0.23 0.10 0.09 0.03 0.14 0.02 0.10 0.13 0.07

HighLit 4 -0.15 -0.02 0.05 0.32 -0.15 0.60 -0.11 -0.07 -0.16 -0.12 0.12 -0.02 -0.03 -0.03 0.25

HighGrowth 5 -0.04 -0.01 0.03 0.32 -0.09 0.08 -0.07 0.01 -0.07 -0.07 0.15 -0.01 -0.04 0.01 0.01

Financial 6 -0.01 0.08 -0.04 -0.15 -0.09 -0.12 0.03 -0.01 0.11 0.09 -0.04 0.00 0.05 0.00 -0.10

HighTech 7 -0.19 -0.01 0.05 0.60 0.08 -0.12 -0.13 -0.09 -0.18 -0.08 0.09 -0.02 -0.02 -0.08 0.18

NumSegment 8 0.00 -0.04 0.23 -0.12 -0.08 0.04 -0.16 0.12 0.37 0.08 -0.06 -0.01 0.11 0.10 0.16

ExtraItem 9 0.06 -0.02 0.10 -0.07 0.01 -0.01 -0.09 0.12 0.17 0.07 0.03 -0.01 0.06 0.19 -0.01

Size 10 -0.28 0.03 0.09 -0.15 -0.08 0.10 -0.16 0.35 0.17 0.20 -0.27 -0.07 0.20 0.10 0.29

BusyEnd 11 -0.01 0.00 0.03 -0.12 -0.07 0.09 -0.08 0.08 0.07 0.20 -0.03 0.02 0.02 0.05 0.04

Loss 12 0.20 -0.03 0.14 0.12 0.15 -0.04 0.09 -0.05 0.03 -0.26 -0.03 0.04 0.00 0.09 -0.04

Leverage 13 0.18 0.00 0.01 -0.30 -0.06 0.11 -0.33 0.26 0.20 0.28 0.17 0.17 0.00 0.04 -0.03

Auditor 14 -0.07 0.03 0.10 -0.03 -0.04 0.05 -0.02 0.11 0.06 0.19 0.02 0.00 0.02 -0.02 0.12

AudOpi 15 0.11 0.01 0.13 -0.03 0.01 0.00 -0.08 0.10 0.19 0.11 0.05 0.09 0.21 -0.02 0.04

Foreign 16 -0.16 0.08 0.07 0.25 0.01 -0.10 0.18 0.15 -0.01 0.30 0.04 -0.04 -0.03 0.12 0.04

aBold text indicates significant differences between two subsamples at the 0.05 level or better, one-tailed. Differences in means (medians) are asseseed using a t-test

(Wilcoxon rank sum test).

bBold text indicates significance at the 0.05 level or better, two-tailed.

45

Table 1.7

Multivariate Test Results (Model 1)

Variable Coefficient Std Error z-value p -value

Constant*** 4.192 0.090 46.37 0.000

AfterERP** -0.042 0.019 -2.21 0.027

HighLit** -0.078 0.020 -3.95 0.000

HighGrowth -0.030 0.019 -1.63 0.104

Financial -0.018 0.046 -0.4 0.692

HighTech*** -0.167 0.019 -8.62 0.000

NumSegment 0.005 0.004 1.06 0.291

ExtraItem*** 0.088 0.022 4.03 0.000

Size*** -0.071 0.005 -15.28 0.000

BusyEnd 0.023 0.015 1.52 0.130

Loss*** 0.154 0.018 8.76 0.000

Leverage 0.000 0.000 -0.3 0.763

auditor -0.034 0.026 -1.32 0.188

AudOpi*** 0.117 0.018 6.36 0.000

Foreign -0.026 0.016 -1.57 0.116

d_1994 -0.027 0.088 -0.3 0.763

d_1995 -0.059 0.086 -0.69 0.492

d_1996 -0.039 0.085 -0.46 0.649

d_1997 -0.029 0.085 -0.34 0.737

d_1998 0.023 0.085 0.27 0.791

d_1999 0.025 0.086 0.29 0.773

d_2000 0.046 0.086 0.53 0.594

d_2001 0.009 0.086 0.1 0.921

d_2002 -0.009 0.087 -0.1 0.920

d_2003 0.131 0.087 1.5 0.134

d_2004*** 0.455 0.091 5 0.000

d_2005 0.373 0.302 1.24 0.216

d_2006 0.048 0.418 0.11 0.909

Numer of Obs 3225

LR chi2(27) 852.84

Prob > chi2 0.000

***, **, * denote significant at the 0.01, 0.05, and 0.1 levels, respectively.

46

Table 1.8

Multivariate Test Results (Model 2)


Constant*** 4.185 0.090 46.3 0.000

AfterERP1 -0.003 0.033 -0.1 0.921

AfterERP2 -0.038 0.036 -1.07 0.283

AfterERP3 -0.030 0.037 -0.83 0.407

AfterERP4 * -0.066 0.039 -1.72 0.085

AfterERP5 ** -0.107 0.043 -2.51 0.012

HighLit*** -0.077 0.020 -3.91 0.000

HighGrowth -0.030 0.019 -1.59 0.113

Financial -0.017 0.046 -0.37 0.715

HighTech*** -0.167 0.019 -8.6 0.000

NumSegment 0.005 0.004 1.07 0.287

ExtraItem*** 0.089 0.022 4.09 0.000

Size*** -0.071 0.005 -15.21 0.000

BusyEnd 0.024 0.015 1.55 0.120

Loss*** 0.153 0.018 8.73 0.000

Leverage 0.000 0.000 -0.3 0.767

Auditor -0.033 0.026 -1.27 0.206

AudOpi*** 0.116 0.018 6.34 0.000

Foreign -0.026 0.016 -1.6 0.111

d_1994 -0.025 0.088 -0.29 0.774

d_1995 -0.058 0.086 -0.67 0.501

d_1996 -0.038 0.085 -0.44 0.656

d_1997 -0.027 0.085 -0.32 0.752

d_1998 0.023 0.085 0.27 0.788

d_1999 0.021 0.086 0.24 0.807

d_2000 0.041 0.086 0.47 0.637

d_2001 0.011 0.087 0.13 0.899

d_2002 -0.001 0.087 -0.01 0.989

d_2003* 0.151 0.088 1.71 0.086

d_2004*** 0.485 0.093 5.24 0.000

d_2005 0.421 0.303 1.39 0.165

d_2006 0.114 0.420 0.27 0.786

Numer of Obs 3225

LR chi2(31) 857.03

Prob > chi2 0.000


47

Table 1.9

Multivariate Test Results


Constant*** 4.227 0.037 114.75 0.000

AfterERP*** -0.054 0.011 -4.87 0.000

HighLit*** -0.039 0.014 -2.88 0.004

HighGrowth 0.001 0.010 0.06 0.953

Financial * -0.056 0.033 -1.72 0.085

HighTech -0.015 0.013 -1.14 0.253

NumSegment*** 0.016 0.002 7.1 0.000

ExtraItem*** 0.094 0.019 4.86 0.000

Size*** -0.040 0.003 -14.58 0.000

BusyEnd *** 0.031 0.009 3.46 0.001

Loss*** 0.092 0.010 9.48 0.000

Leverage 0.001 0.003 0.23 0.821

Auditor** -0.030 0.013 -2.34 0.019

GC*** 0.170 0.025 6.89 0.000

Foreign -0.006 0.010 -0.58 0.564

d_2000*** -0.288 0.039 -7.34 0.000

d_2001*** -0.285 0.038 -7.59 0.000

d_2002*** -0.194 0.037 -5.31 0.000

d_2003*** -0.104 0.035 -2.93 0.003

d_2004*** 0.177 0.035 5.02 0.000

d_2005*** 0.230 0.035 6.52 0.000

d_2006*** 0.246 0.036 6.94 0.000

d_2007*** 0.215 0.036 5.99 0.000

d_2008*** 0.177 0.036 4.89 0.000

d_2009*** 0.175 0.037 4.73 0.000

d_2010** 0.158 0.038 4.2 0.000

d_2011** 0.155 0.038 4.07 0.000

d_2012* 0.078 0.044 1.77 0.076

Number of Obs 8610

LR chi(27) 2011.8

Prob > chi2 0.000


48

Table 1.10



Constant*** 4.227 0.037 114.75 0.000

AfterERP1*** -0.066 0.019 -3.49 0.000

AfterERP2** -0.044 0.020 -2.22 0.026

AfterERP3*** -0.056 0.020 -2.72 0.007

AfterERP4** -0.046 0.023 -2.01 0.045

AfterERP5* -0.052 0.029 -1.81 0.071

HighLit*** -0.039 0.014 -2.88 0.004

HighGrowth 0.001 0.010 0.05 0.957

Financial * -0.056 0.033 -1.72 0.085

HighTech -0.015 0.013 -1.14 0.253

NumSegment*** 0.016 0.002 7.09 0.000

ExtraItem*** 0.094 0.019 4.87 0.000

Size*** -0.040 0.003 -14.58 0.000

BusyEnd *** 0.032 0.009 3.47 0.001

Loss*** 0.093 0.010 9.48 0.000

Leverage 0.001 0.003 0.22 0.825

Auditor** -0.030 0.013 -2.33 0.020

GC*** 0.170 0.025 6.89 0.000

Foreign -0.006 0.010 -0.58 0.565

d_2000*** -0.288 0.039 -7.34 0.000

d_2001*** -0.285 0.038 -7.58 0.000

d_2002*** -0.194 0.037 -5.3 0.000

d_2003*** -0.103 0.035 -2.93 0.003

d_2004*** 0.177 0.035 5.01 0.000

d_2005*** 0.231 0.035 6.53 0.000

d_2006*** 0.246 0.036 6.92 0.000

d_2007*** 0.215 0.036 6.01 0.000

d_2008*** 0.177 0.036 4.88 0.000

d_2009*** 0.178 0.037 4.79 0.000

d_2010*** 0.154 0.038 4.05 0.000

d_2011*** 0.155 0.039 4.01 0.000

d_2012* 0.075 0.045 1.66 0.097

Number of Obs 8610

LR chi(31) 2012.65

Prob > chi2 0.000


49

Chapter 2: The Fair Value Option (FVO) under SFAS 159: Effectiveness of FVO and

Intention of FVO Adopters

I. INTRODUCTION

In recent years, many countries have adopted the International Financial Reporting

Standards (IFRS) whose critical measurement principle is fair value accounting. In addition,

the FASB has issued several accounting standards which help facilitate the movement

toward fair value accounting from historical cost accounting. However, considerable

debate still exists on the advantages and disadvantages of fair values in financial statements.

While proponents claim that fair values help financial statements reflect more relevant

information in a timely manner, opponents argue that fair value accounting violates the

traditional accounting principle of conservatism, which may reduce the reliability of

reported financial information and thus decrease the usefulness of accounting information.

In 2007, the Financial Accounting Standards Board (FASB) issued Statement of

Financial Accounting Standards No. 159 (SFAS 159) – The Fair Value Option (FVO) for

Financial Assets and Financial Liabilities – which provides firms with an option to measure

and report selected financial instruments at fair value. The objectives of this statement are

for mitigating unnecessary earnings’ volatility caused by mixed-measurement accounting,

and for expanding the use of fair value measurement. However, SFAS 159 has drawn

widespread scrutiny and significant criticism from different parties including media,

regulators, and academic researchers. Critics argue that even though FVO may help reduce

the problem of mixed-measurement accounting, it may not help reduce earnings volatility

due to the inherent feature and estimation errors of fair values. In fact, there are several

studies which report evidence that fair value based earnings are more volatile than earnings

50

under the current GAAP (Barth et al. 1995; Hodder et al. 2006). Also, providing managers

with full discretion on selecting financial instruments to which FVO is applied might

induce other problems, such as reduced reliability and comparability. Furthermore, the

option of implementing FVO on an instrument by instrument basis suggested in SFAS 159

can even increase possibility that managers do use FVO not in accordance with the intent

of SFAS 159, and there exist prior studies which report the possibility of managers’

opportunistic election of FVO (Henry 2009; Hsu and Lin 2016).

Thus, the primary purpose of this study is to examine whether the FVO of SFAS

159 has achieved its intended objective which is reducing unnecessary earnings’ volatility.

In addition, this study also tries to explore the motivation of firms’ decision to adopt FVO.

The results indicate that the earnings’ volatility did not improve after adopting FVO. It

seems that there is even a positive association between FVO and earnings volatility and the

relation is being amplified by the size of gains or losses resulting from the FVO. The results

also indicate that the analysts’ forecast error of FVO adopter is greater than that of non-

FVO adopters, which is consistent with negative relation between earnings volatility and

analysts’ forecast accuracy. Given the results of even higher earnings volatility after

adopting FVO, I have tried to explore the intention of firms’ decision to adopt FVO using

two more hypotheses: opportunistic intention and informative intention. However, I could

not find strong results for either hypothesis but there seems to be some limited evidence

for opportunistic intention.

Prior studies in general examines whether FVO adopters are more likely to meet or

beat analysts’ forecast consensus using unrealized gains resulting from elected financial

instruments in order to find evidence of opportunistic election of FVO. This study

51

contributes to FVO-related literature by examining the change in earnings volatility around

adoption of FVO using a sample of all US firms that adopted FVO over a long time period

(eight years). Fiechter (2011) examines whether FVO adoption affects earnings volatility

He examines the effect of FVO of IAS 39, not SFAS 159, on earnings volatility of

international banks over two year time period (using quarterly data). That study suggests

a different type of opportunistic motivations of firms to adopt FVO related to financial

activity. By using a different method of identifying FVO adopters and by extending the

sample period, this study covers a significantly higher number of FVO adopters across

industries, and thus provides more comprehensive evidence of the consequences of FVO

adoption

This paper proceeds as follows. Section II introduces prior research related with

fair value accounting, and Section III discusses SFAS 159 and related research. Section IV

develops hypotheses, and section V describes data collection and research methodology.

Section VI presents test results and section VII offers a summary and conclusion.

II. LITERATURE REVIEW ON FAIR VALUE ACCOUNTING4

Is fair value relevant?

The primary interest of policy makers and academic researchers is in whether fair

value accounting improves information of financial statements that interested users use

compared to the information of historical cost accounting. Thus, a significant number of

prior studies examine the value relevance of fair value accounting. Financial statements

4 The words of fair value, mark-to-market, market value-based, and market value accounting have

been often used as synonyms (Barth 1994).

52

can be considered value-relevant if the information they reflect helps financial statement

users to assess the firm value (Barth and Landsman 1995). Thus, a common way to assess

the value relevance of fair value information recognized or disclosed could be assessing its

incremental association with share prices (Landsman 2007). In general, prior literature on

fair value accounting indicates that fair values provide more relevant information to

financial statement users than the amounts under historical cost accounting.

Barth (1994) examines disclosed fair value estimates of banks’ investment

securities and securities gains and losses. She reports that investment securities’ fair values

provide incremental explanatory power for equity values of banks beyond that of historical

costs, implying that the fair value information disclosed is value relevant. However, the

results for securities gains and losses are mixed. These mixed results on securities gains

and losses seem to be attributable to too much measurement error in the fair value estimates.

Eccher et al. (1996), Nelson (1996), and Barth et al. (1996) examine the value

relevance of fair value estimates required to be disclosed by SFAS No. 107. Using a sample

of U. S. banks, Eccher et al. (1996) investigate the association between share prices and

disclosures of those fair value estimates. They first regress market-to-book (MB) ratios on

the differences between the fair value estimates and their corresponding book values in

order to examine whether the fair value disclosures are value-relevant after controlling

historical cost book values. Similar to Barth (1994), they find that fair values of investment

securities have association with MB ratios. However, they find a weak association for fair

values of net loans, mixed results for fair values of off-balance-sheet instruments, and no

association for fair values of deposit. They also examine whether fair value disclosures

have incremental value relevance over the amounts from traditional historical cost

53

accounting. In order to do so, the study first develop a benchmark model using the

information only form historical cost accounting, and then augment the benchmark model

with fair value disclosures to assess their incremental value relevance. They find that the

R-square increased significantly when the fair value data is included in the model,

indicating that fair value disclosures provide incremental explanatory power in explaining

the variation of MB ratios.

Like Eccher et al. (1996), Nelson (1996) also examines the value relevance of fair

value estimates under SFAS 107 by investigating the association between share prices of

commercial banks and those fair value estimates. She finds an incremental explanatory

power for fair values of investment securities, but no evidence of incremental explanatory

power for fair value estimates of loans, deposits, long-term debt and off-balance sheet

financial instruments. However, it is also found that when additional value relevant

variables, ROE and growth in book value, which can be captured by the accrual accounting

system and are omitted from prior studies on fair value disclosures are incorporated into

the research model, the fair value estimates of investment securities lose their incremental

power to explain share prices.

On the other hand, unlike Eccher et al. (1996) and Nelson (1996), Barth et al. (1996)

find results indicating that differences between disclosed fair values and book values of

securities and fair value estimates of loans and long-term debt provide significant

incremental explanatory power for bank share prices beyond that of book values, but those

of deposits and off-balance sheet items do not. The primary difference between Barth et al.

(1996) and those two studies is that Barth et al. (1996) include additional explanatory

variables for incorporating institutional features of the bank industry: nonperforming loans,

54

interest-sensitive assets and liabilities, a proxy for the intangible asset attributable to core

deposits, net pension assets and nonfinancial assets and liabilities. Focusing on loans’ fair

value estimates, they also find that the coefficient of loans’ fair value is higher for banks

with higher regulatory capital ratios, implying that the market discounts unrealized gain on

loans disclosed by banks which are financially less healthy.

Venkatachalam (1996) investigates the value relevance of disclosed fair values of

banks’ off-balance sheet derivative financial instruments under SFAS No. 119. The

findings indicate that the fair values of off-balance sheet derivatives help explain cross-

sectional variations in bank share prices beyond the contractual (notional) amounts of such

derivatives. He also finds that fair value gains and losses for on-balance sheet financial

instruments are negatively associated with the fair value gains and losses of off-balance

sheet derivatives, which would be evidence that banks use derivatives to reduce their risk

exposures.

Hodder et al. (2006) investigate the properties of net income, comprehensive

income, and a self-constructed measure of full-fair-value (FFV) income for a sample of

U.S. commercial banks. They find that the volatility of FFV income is more than three

times that of comprehensive income and more than five times that of net income even

though average levels of the three income measures are similar in magnitude over the

analyzing period. They also find that the incremental volatility of FFV income is positively

associated with various market-based risks, such as standard deviation in returns, market

model beta, and long-term interest-rate beta, beyond the effects of volatility in net income

and comprehensive income. These findings imply that volatility of FFV income reflects

risk factors which are not captured by volatility of GAAP net income or comprehensive

55

income, and thus fair value accounting might help investors assess the denominator,

discount rate, in a valuation context.

Opponents of fair value accounting express a concern that fair value accounting

will deteriorate reliability of accounting amounts even though it may enhance their

relevance. However, it is in fact very difficult to test separately relevance and reliability of

accounting amounts and value relevance tests are likely to be joint tests of relevance and

reliability. In general, extant literature examines value relevance of accounting amounts by

investigating their associations with equity values to measure how well accounting

amounts reflect information used in financial markets. However, accounting amounts will

be used by market participants only when the amounts are not only relevant but also reliable

enough. An association between equity values and fair value amounts might indicate that

those amounts are relevant and also reliable enough to be used in the market. However,

likewise it is also difficult to declare that a failure to find such an association is attributable

to lack of only relevance or only reliability (Barth et al. 2001). Thus, examining relevance

of information should not be viewed as separate research from study on reliability.

Disclosure vs recognition:

According to Statement of Financial Accounting Concepts No. 5 (1984),

information is relevant when it is capable of making a difference in user decisions, and

reliable when it is representationally faithful, verifiable, and neutral. However, information

provided by financial statements does not have to be new to financial statements users to

be relevant. Rather, an important role of accounting is to summarize or aggregate

information that might be available from other sources (Barth et al. 2001). This would

56

imply that how information is presented or provided might be important. Financial

statements can provide information through recognition or disclosure. There are several

studies which examine and compare fair value estimates recognized and those disclosed.

In their experimental study, Hirst et al. (2004) construct two alternative income

measures for commercial banks, piecemeal (partial)-fair-value (PFV) income measure and

full-fair-value (FFV) income measure, and examine whether risk and value judgements of

bank-specialist analysts are affected by banks’ income measurement method under exposed

and hedged risk conditions of those banks. According to their research setting, while fair

value gains and losses of only investment securities are recognized and those of all other

financial assets or liabilities are disclosed in footnotes under PFV income measurement,

fair value gains and losses of all financial assets or liabilities are recognized under FFV

income measurement. Thy find that analysts’ risk assessments are higher and value

assessments are lower under FFV measurement than under PFV income measurement only

for banks exposed to interest-rate risk, but not for hedged banks. In addition, they also find

that analysts’ risk assessments are higher and value assessments are lower for exposed

banks than for hedged banks only under FFV measurement, but not under PFV income

measurement. These findings indicate that analysts can better differentiate between risky

firms and less risky firms when more fair value changes are recognized than when they are

just disclosed, and under risker conditions, recognizing fair value changes, rather than

disclosing them, may help analysts to get informed and to react to the risk.

Ahmed et al. (2006) examine if share prices are associated with whether the fair

value of derivative financial instruments is recognized or disclosed. Using a sample of

banks that simultaneously hold both recognized and disclosed derivatives prior to SFAS

57

No. 133, they find that while net fair values of recognized derivatives are positively

associated with share prices, coefficients on disclosed derivatives are not significant. Also,

using a sample of banks that have only disclosed derivatives prior to SFAS No. 133, which

are recognized after SFAS No. 133, the study finds no evidence that the fair value of

disclosed derivatives is associated with share prices in the pre-SFAS 133 period, but finds

significantly positive coefficients on derivatives recognized after SFAS 133.

Barth (2004) states that although disclosure is not a substitute for recognition, it can

be complement to recognition. Prior studies seems to indicate that recognized fair value

estimates are better perceived by market participants. However, recognizing those fair

value estimates only may not be enough to provide investors with useful information to

help them make a better and complete decision. There is an inherent risk involved in fair

value estimates which is injected by the inevitable estimation process. Thus, it would be

useful to investors to provide additional information through disclosures about the fair

value estimates recognized such as values at high risk, inherent past volatility of estimates,

and history of realized amounts. Also, how those estimates are presented would also matter.

With the advent of complex financial instruments, recognizing or disclosing those items

makes financial statements to be complex, and thus it gets more difficult to comprehend

the information presented in financial statements. Therefore, disclosing the additional

information in a way or format where information processing costs are lower would be

important for the information to be truly useful to information users.

Earnings management

In such cases where active markets for assets or liabilities do not exist, managers

58

have to rely on their own assumptions for the fair values of the assets or liabilities. This

reality inherently introduces the general problem of information asymmetry between

managers and investors. The information asymmetry can create the moral hazard problem

which may results in manipulation of information provided to the market (Landsman 2007).

There exists prior literature which examines issues of earnings management associated

with fair value accounting. Aboody et al. (2006) examine four key option pricing model

inputs and find that firms understate estimates of option values and thus stock-based

compensation expense (SFAS 123 expense). Specifically, firms tend to adjust expected

option life, expected stock price volatility, and expected dividend yield which are firm-

specific, but not the risk-free interest rate which is less subject to managers’ discretion. The

findings of this study imply that when managers have incentives to manage earnings and

they need to estimate fair values using their own assumptions, they are likely to adjust

those assumptions and thus manage earnings. Dechow et al. (2010) investigate the

securitization gains or losses in the income statement under SFAS 140. They find that

reported securitization gains are larger when pre-securitization earnings are low or below

last year’s level, indicating that managers may take advantage of considerable discretion

over assumptions used in estimating fair values to manage earnings. Also, they find

evidence that those gains are treated as a regular earnings component and thus managers

are rewarded for the gains. In addition, the board appears to be ineffective in monitoring

this earnings management by managers.

In contrast, there also exist prior studies implying that fair value accounting does

not exacerbate or might help mitigate the problem of earnings management. In a study

using Danish banks, Bernard et al. (1995) find no evidence that mark-to-market accounting

59

numbers are managed to avoid regulatory intervention. Their results also indicate that the

reported numbers under Danish mark-to-market accounting system are more reliable

indicators of firm values than those in U.S. system. They point out that the rigid regulatory

intervention policies of Denmark may have contributed to those results. Furthermore, it is

generally argued that the historical cost regime can be inefficient especially in good times

because firms cannot recognize the increase in fair value of assets under historical cost

accounting system.5 Thus, firms have incentives to time asset sales to manage earnings

under the historical cost regime. Proponents of fair value accounting argue that fair value

accounting helps mitigate this problem. The results of Black et al. (1998) are consistent to

this argument. They find evidence that when fair value measurement is used for

recognizing gains or losses in reported income, firms reduce the practice of timing asset

sales for earnings management purposes.

Broader perspectives on fair value accounting

On the other hand, there exist several theoretical studies which imply that we may

need to consider broader economic consequences of fair value accounting when we move

toward the complete fair value accounting regime from historical cost accounting regime.

Plantin et al. (2005, 2008a, 2008b) state that we live in an imperfect world where markets

are not always fully liquid and there are multiple sources of market imperfection. In such

an imperfect world, true market prices which would be available in a perfect world are not

available. Rather, market prices in an imperfect market play a dual role. They are not only

a reflection of the underlying economic fundamentals and actions taken by market

5 Plantin et al. (2008b) state that while the mark-to-market regime leads to inefficient sales in bad

times, the historical cost regime is particularly inefficient in good times.

60

participants, but also an amplifier of those actions. That is, when actions affect prices, the

change in prices also affects actions in turn. If price changes are reflected continuously and

immediately in financial statements under the fair value accounting system, it would be

possible that market participants manage their financial statements more actively and

immediately to maintain their financial objectives, such as leverage ratio.6 This may induce

additional endogenous source of volatility injected by accounting principle. The market

participants’ reaction to the financial statements which are based on fair value can be more

problematic in that it can amplify the financial cycle. As an attempt to maintain their

financial objectives, market participants may need to purchase assets when prices increase

and sell when prices decrease, which results in greater demand during booms and greater

supply during recession (Plantin et al. 2008a). Plantin et al. (2008b) state that accounting

can be seen as a mere detail of measurement leaving the economic fundamentals unaffected

only in the context of completely frictionless competitive markets, implying that

accounting can be relevant only in the imperfect world. The paper argues that we are living

in an imperfect world and shows that accounting amounts matter. It develops a model

which can compare the real effects of a historical cost accounting and fair value accounting,

and theoretically shows that real decisions can be distorted due to different accounting

measurement systems. The main results of the paper indicate that fair value accounting is

more inefficient (lower informational content of prices and suboptimal real decisions) than

historical cost accounting when claims are long-lived, not sufficiently liquid, or senior.7

6 Especially, financial institutions, which are entities that influence financial markets significantly,

tend to adjust their balance sheets more actively in accordance with the price changes than

households (Plantin et al. 2008a) 7 This explains why banking and insurance industries voice strong objections while equity investors

are enthusiastic proponents for fair value accounting. For those financial institutions, a large

proportion of balance sheets consist of items that are of long duration, illiquid, and senior (Plantin

61

Allen and Carletti (2008) develop a model in the context of banking and insurance sector

and illustrate an important disadvantage of mark-to-market accounting. In times of crisis,

prices in illiquid markets may not properly reflect future payoffs, but rather reflect the

amount of liquidity available. It shows that in such circumstances using mark-to-market

accounting can lead to distortions of financial institutions’ portfolio and contract choices,

contagion, and ultimately unnecessary liquidations which would not arise if historical cost

accounting were used. This result implies that fair value accounting may not be beneficial

when markets are not liquid and historical cost accounting may be preferable in such

circumstances. Gorton et al. (2006) examine the issue of mark-to-market in the context of

a security market. Using a theoretical principal-agent setting, they show that mark-to-

market compensation contracts can introduce an externality among traders. The study

indicates that such contracts have feedback effects on asset prices. Similar to Plantin et al.

(2008a, 2008b), this suggests that while prices are dependent on traders’ actions, the prices

also affect the behavior of traders whose compensation is based on asset prices. If this is

the case, prices may not be exogenous indicators of asset value. Consequently, the results

indicate that traders may rationally herd, trading on irrelevant information, which results

in less informative asset prices than they would be without the mark-to-market

compensation scheme.

The issues aforementioned on fair value accounting seem to ultimately stem from

a question about how much we can trust the reliability of fair values estimated. The

proponents of fair value accounting argue that fair value provides more relevant

information and it is also better in terms of corporate governance than historical cost.

et al. 2008a; 2008b).

62

However, this argument may lose its persuasiveness to the extant which the fair values

estimated are noisy. Plantin et al. (2008b) state that the choice between historical cost

accounting and fair value accounting is a choice between obsolete information and

distorted version of current information. While historical cost accounting system ignores

price signals and induces excessive conservatism, fair values available are based on

incomplete information. If we admit that we are living in an imperfect world, we also

cannot help but admit that market prices reflect not only economic fundamentals but also

other sources of market imperfections, which may lead to distortions or transitory

fluctuations of prices. Thus, reflecting immediately the changes in fair value on financial

statements may induce unnecessary volatility of financial statement amounts and may not

be the best way to provide information users with the most relevant information on

underlying economics in order to help them make a better decision. Therefore, it appears

that enhancing reliability of fair value is certainly one of the most important issues to be

addressed before we move to the complete fair value accounting system, and developing a

better measurement basis to mitigate those problems of fair values would be worthwhile.8

III. SFAS 159

Background of SFAS 159

The advent of innovative financial instruments has caused concerns to many,

including accounting profession, regulating bodies, creditors and investors, and preparers

8 For example, Plantin et al. (2008a) suggest using an average value over some interval of time as

the accounting value of an asset and also using an average of past observed discount factors as the

discount factors for illiquid assets. Glover et al. (2005) proposes intertemporal financial statements

which can show facts separately from forecasts by reporting three different columns: Fact, Forecast,

and Total.

63

of financial statements. As a response, the FASB introduced a project on financial

instruments and off-balance-sheet financing in 1986, and issued three Statements of

Financial Accounting Standards (SFAS) in order to deal with complex issues regarding

financial instruments between 1990 and 1994: No. 105, “Disclosure of Information about

Financial Instruments with Off-Balance-Sheet Risk and Financial Instruments with

Concentrations of Credit Risk,” No. 107, “Disclosures about Fair Value of Financial

Instruments,” and No. 119, “Disclosure about Derivative Financial Instruments and Fair

Value of Financial Instruments.” SFAS 107 is the first standard which requires all firms to

disclose the fair value information of all financial instruments, both recognized and off-

balance-sheet, for which it is practical to estimate fair value. SFAS 119 requires disclosures

about fair value estimates for derivative financial instruments. The statement also requires

firms to distinguish financial instruments held for trading from financial instruments held

for purposes other than trading, and to disclose estimated holding gains and losses for

instruments held for trading. However, those three statements are not without criticisms.

The primary problem seems to be that they focus on “disclosure” issues rather than

“recognition” or “measurement” issues. With those three statements, information users are

provided with much more information that is highly technical, and thus more burden would

rest with users to interpret which information is relevant and useful and to understand the

underlying risk of firms using the information disclosed (Li 2006).

Meanwhile, the FASB appears to notice that recognition and measurement issues

should be resolved in order to address the criticisms on those three statements. As an effort,

the FASB issued SFAS No. 115, “Accounting for Certain Investments in Debt and Equity

Securities,” in 1993. The statement requires that all debt and equity securities be classified

64

into one of the three categories: trading, available for sale, and held to maturity. However,

it retains the uses of both fair value measures and historical cost measures. Securities

classified as held to maturity are carried at amortized cost and securities classified as

trading or available for sale are carried at fair values. This statement is criticized in that it

can provide the opportunity for earnings manipulation because categorizing securities or

transferring securities between categories is subjective and unrealized holding gains and

losses from different categories are treated differently. In 1998, the FASB issued SFAS No.

133, “Accounting for Derivative Instruments and Hedging Activities,” as an effort to

achieve the objective of measuring all financial assets and liabilities at fair value. The

statement establishes accounting and reporting standards for derivatives instruments and

hedging activities. It supersedes SFAS No. 105 and No. 119 and amends SFAS No. 107.

The major issue on SFAS 133 is its complexity. Under SFAS 133, the special hedging

accounting is applied only to qualifying items, not all items held for the purpose of hedging.

However, it is so complicated to follow the requirements for qualification that many

instruments end up not being carried at fair value. Thus, it is not uncommon that financial

assets and liabilities are accounted for differently depending on whether they are qualified

for a hedging relationship or not under complex SFAS 133.

In 2006, the FASB issued SFAS 157, “Fair Value Measurements.” The statement is

aimed at increasing the consistency and comparability of fair value measurements and

expanding disclosures about fair value measurements by improving both disclosure and

measurement issues of fair value. More specifically, the statement tries to establish a single

authoritative definition of fair value (“exit-price,” not “entry price”) 9 , establish a

9 SFAS 157 defines fair value as the price that would be received to sell the asset or paid to transfer

the liability (an exit price), not the price that would be paid to acquire the asset or received to

65

framework for measuring fair value, and expand financial statement disclosure

requirements for fair value measurements. SFAS 157 also simplifies and codifies the

related guidance that exists for developing fair value measurements so that it can eliminate

differences between the guidance that have added to the complexity in GAAP. Although

SFAS 157 does not require any new fair value measurements, it clarifies the meaning of

fair value used under the U.S. generally accepted accounting principles (GAAP). Also, the

statement led to several changes in practice related to certain valuation techniques. For

example, with the adoption of SFAS 157, companies are required to maximize the use of

relevant observable inputs and minimize the use of unobservable inputs when they use

valuation techniques for determining fair values. 10 The FASB also emphasizes in the

statement that fair value is a market-based measurement, not an entity-specific

measurement. 11 Prior to the adoption of SFAS 157, entities may have used internal

assumptions that management believed to be better than available observable inputs

(Deloitte 2008). As to the expanded disclosure, SFAS 157 requires firms to make tabular

disclosure for fair value hierarchy by major category of assets and liabilities on a recurring

basis. The fair value hierarchy prioritizes the inputs used to measure fair value into three

broad levels: level 1, level 2 and level 3. The level in the fair value hierarchy within which

assume the liability (an entry price) (FASB 2006). 10 Inputs refer to the assumptions that market participants would use in pricing the asset or liability.

Observable inputs are inputs that reflect the assumptions based on market data obtained from

sources independent of the reporting entity. Unobservable inputs are inputs that reflect the reporting

entity’s own assumptions based on the best information available to the entity in the circumstances

(FASB 2006). 11 The estimated cash flows used for determining fair value are the cash flows that market

participants could obtain from an asset or liability, not estimates of entity-specific cash flows.

Likewise, the discount rate used for fair value is the rate of interest that market participants would

expect to receive for bearing the risks inherent to those cash flows, not the reflection of entity-

specific risk preferences (Barth 2004).

66

each fair value measurement falls is determined on the basis of the lowest level input that

is significant to the fair value measurement.12 Critics of SFAS 157 argue that the exit value

does not reflect properly the value of an asset (or liability) to a company because it fails to

consider how the asset is used within the company and the value is not assessed in

conjunction with other assets. In addition, because in many cases, active markets for

identical assets or liabilities do not exist, many fair value estimates are based on level 2 or

level 3 inputs. Level 2 or 3 inputs are less reliable and difficult to verify (Landsman 2007).

SFAS 159

In January 2006, the Financial Accounting Standard Board (FASB) issued an

exposure draft, The Fair Value Option for Financial Assets and Financial Liabilities. After

receiving responses to the draft from researchers and practitioners, it finally issued FASB

Statement No. 159 (SFAS 159), The Fair Value Option for Financial Assets and Financial

Liabilities – Including an amendment of FASB Statement No. 115, in February 2007. The

statement became effective for fiscal years beginning after November 15, 2007, with early

adoption permitted (FASB 2007).

SFAS 159 affords reporting entities the option to measure certain financial assets

and financial liabilities at fair value, which was not previously available under U.S. GAAP.

Under SFAS 159, companies have the irrevocable option to measure most financial assets

and financial liabilities at fair value (FVO) without any justification and changes in fair

12 Level 1 inputs are quoted prices (unadjusted) in active for identical assets and liabilities. In

general, level 2 inputs are quoted prices in active markets for similar or related assets and liabilities,

or quoted prices in inactive markets for identical assets and liabilities. Level 3 inputs are

unobservable inputs which are based on the reporting entity’s own assumptions. (FASB 2006).

Level 1 inputs are the most preferred.

67

value of those items chosen for FVO are recognized in current-period earnings. In addition,

the difference between beginning book value and fair value of financial instruments

selected for FVO should be reported as a cumulative-effect adjustment (effect of re-

measurement to fair value) to the beginning balance of retained earnings in the first FVO

adoption year. Also, it states that the FVO is applied on an instrument by instrument basis.

Furthermore, because the Financial Accounting Standard Board (FASB) supported a broad

application of the FVO, it decided to keep scope exceptions to a minimum and thus the

FVO can be applied to most financial assets and liabilities. The financial assets and

liabilities excluded from the scope of SFAS 159 include investments in subsidiaries that

would be consolidated, assets and liabilities associated with pension and other

postretirement benefits, financial assets and liabilities recognized under lease contracts,

deposit liabilities of financial institutions, and financial instruments classified as

shareholder’s equity. Appendix A presents examples of FVO adoptions.

The primary objective of SFAS 159 is to mitigate volatility in reported earnings by

reducing artificial earnings volatility caused by measuring related assets and liabilities

differently. Before SFAS 159, firms could offset the change in fair value of financial

instruments against the change in fair value of other derivatives held for the purpose of

hedging for those instruments by following the hedging accounting of SFAS 133. However,

the hedging accounting was too complicated to apply and thus the costs to comply with the

requirements of SFAS 133 were high. Therefore, to the extent that costly hedging

accounting prevents firms from applying hedging accounting, firms would record related

financial assets and liabilities differently and this discrepancy could induce unnecessary

volatility in earnings (Guthrie et al. 2011). However, under SFAS 159 firms don’t have to

68

follow this costly and complex rules for hedging and they now can measure and record

almost all financial instruments at fair value. Thus, SFAS 159 is expected to help entities

to avoid the time, effort, and systems needed to document fair value hedging relationships

and demonstrate their effectiveness to qualify for continued hedge accounting.

Another purpose of SFAS 159 is to expand the use of fair value measurement which is

consistent with the Board’s long-run measurement objective for accounting for financial

instruments. Even though SFAS 159 permits FVO for only financial assets and liabilities,

the Board has a plan to expand FVO even to certain nonfinancial assets and liabilities that

are similar to financial items. Lastly, the statement is also an attempt by the FASB for

international convergence in financial reporting standards. Prior to SFAS 159, the

International Accounting Standards Board (IASB) issued the amended International

Accounting Standard 39 (IAS 39), Financial Instruments: Recognition and Measurement,

which describes conditions under which firms can elect fair value measurement for

financial instruments, which is called fair value option (IASB 2005).13 The purpose of this

amended IAS 39 is to simplify accounting for financial instruments as well as to provide

an opportunity to reduce accounting mismatches (Fiechter 2011).

Issues of SFAS 159

Complexity: The FASB argues that FVO will help firms to reduce artificial

earnings volatility caused by accounting mismatches because they do not have to apply

complex hedge accounting provisions presented in SFAS No. 133. SFAS No. 133 has been

13 While the IASB incorporates some eligibility requirements for application of FVO, SFAS 159

dose not impose any eligibility requirements primarily for the reason that it would reduce the use

of fair value measurement and increase complexity (FASB 2007, paras. A20 and A21).

69

criticized for its complexity and it has not been effective to mitigate earnings volatility

caused by measuring related assets and liabilities differently. However, fair value

measurement standard is also complex. It is complex and difficult to measure fair values

especially in the absence of quoted market prices in an active market. When quoted market

prices are not available, fair value measurement is based on subjective assumptions, and

thus it can provide an opportunity to manipulate earnings (Dechow et al 2010). In fact,

some respondents to the exposure draft for SFAS 159 believe that there is insufficient

guidance for determining fair values and many of the assumptions used in determining fair

values are subjective and complex.

Comparability: The comparability within and between entities may be impaired

because firms are allowed to choose different measurement attributes on a contract-by-

contract basis (FASC 2007; Li 2006). It is possible that entities with similar financial assets

and financial liabilities apply different measurement attributes and different measurement

attributes are applies to similar financial assets and financial liabilities within the same

entity due to the contract-by-contract basis FVO. This mixed-attribute measurement may

deteriorate investors’ ability to comprehend financial information.

Volatility: The primary rationale for SFAS 159 is to reduce the earnings volatility

caused by different accounting rules for financial assets and liabilities. However, critics

claim that the FASB’s argument is counterintuitive in that traditionally, managers have

resisted market value measurement to avoid volatile earnings and auditors have been

hesitant to use fair values to minimize the risk of attesting to amounts that shortly go up or

down (Li 2006).

Instrument by instrument approach: Under SFAS 159, firms are allowed to

70

irrevocably apply FVO by an instrument by instrument basis and managers do not have to

justify their decisions to apply FVO to certain financial assets or liabilities. This

unconstrained contract by contract election of fair value might reduce the reliability and

transparency of financial information because there is high possibility of management

discretion involved. And furthermore, the reduced reliability and transparency may result

in additional information processing costs for information users (FASC 2007). Also, it is

generally believed that fair value measurement is more representative of entities’

underlying economics and provides more relevant information than cost-based measures.

However, the instrument by instrument option will result in the reporting at fair value of

only some financial instruments and this discretionary mixed-attribute measurement may

rather significantly reduce the relevance and comparability of financial reporting. Thus,

critics argue that the statement does not meet the objective of financial reporting because

it may deteriorate usefulness of information and increase costs for users even though it

might help reduce earnings volatility, which is not an objective of financial reporting.

Therefore, they believe that FASB should require fair value accounting for all financial

instruments (FASB 2007).

Interim step toward fair value accounting: As mentioned earlier, one of the

objectives of SFAS 159 is to expand the use of fair value measurement. However, some

critics believe that even though the FASB argues that SFAS 159 is a cost-beneficial interim

step toward measuring all financial instruments at fair value, the statement may rather

result in a further delay in the broader requirement for fair value for financial instruments.

If SFAS 159 produces its intended effect, which is reduced earnings volatility, the preparer

community would have less incentive to support wider adoption of fair value measurement.

71

In other words, if prepares can manage earnings volatility with FVO, they would be less

supportive of a fair value measurement requirement. Thus, this would result in delay or

resistance to requiring fair value measurement (FASB 2007).

Research on FVO

Prior studies which examine the possibility of opportunistic adoption of FVO yield

mixed results. Henry (2009) explores the possibility of managers’ opportunistic usage of

FVO by examining commercial banks which adopted FVO before 2008 (early adopters).

By analyzing the disclosures of banks which rescinded or revised their decision of early

FVO adoption, she finds evidence that their initial adoption of FVO was not in compliance

with the intent of SFAS 159. Also, Hsu and Lin (2016) find that firms with more level 3

assets and liabilities and with weak corporate governance are more likely to recognize

unrealized gains by adopting FVO to meet or beat analysts’ forecast consensus. However,

Guthrie et al. (2011) find no evidence of systematic earnings management through FVO.

They identify FVO adopters among S&P 1500 companies and investigate whether those

firms adopt FVO with opportunistic motivation. In order to find evidence of opportunistic

election of FVO, they examine firms whose reported earnings meet or beat the analysts’

forecast consensus only with unrealized gains from elected financial instruments (current

earnings management) and firms which recognize prior unrecognized losses as cumulative-

effect adjustment to retained earnings to avoid recognizing losses in earnings in later years

(future earnings management). They find evidence of earnings management only from a

small number of early adopters, but no evidence of systematic opportunistic election of

FVO.

On the other hand, using an international sample of 222 banks from 44 countries,

72

Fiechter (2011) examines whether FVO of IAS 39 affects earnings volatility and whether

different levels of regulatory quality of different countries affect the likelihood of adopting

FVO. He hand collects information about the intentions of those banks to adopt FVO from

their disclosures, and finds that the earnings volatility of banks which adopt FVO to

mitigate the mixed measurement problem is lower than that of other banks in the cross-

section during the time period from 2006 to 2007.

IV. HYPOTHESIS DEVELOPMENT

Volatility

Barth (2004) identifies and discusses three additional sources of volatility of

financial statement amounts associated with fair value accounting, compared to historical

cost accounting. The first source of volatility is the estimation error volatility. In many

cases where deep and liquid markets do not exist, firms need to estimate fair values, and

the estimation error is inherent and inevitable in the process of estimation. This estimation

error can increase volatility of recognized fair values. The second is inherent volatility of

fair values. It is widely accepted that fair value amounts would fluctuate more from period

to period than amounts based on historical cost. However, this inherent volatility results

from economic reality that a firm faces, not from accounting measurement. The better

reflection of economic reality is the primary rationale for fair value accounting, and thus

this inherent volatility is not evitable. The last source of volatility identified is mixed-

measurement volatility. Mixed-measurement volatility is purely a result of accounting

measurement problem. Currently, some assets and liabilities are measured at fair value,

some are measured at historical cost (amortized historical cost), some are measured at the

73

lower of cost or market, and some others are not recognized at all. Thus, under this mixed-

measurement accounting, changes in market conditions do not affect the values of assets

and liabilities in the same way, which induces additional volatility of financial statement

amounts. Mixed-measurement volatility is different from the other two in that this is an

artifact of the accounting measurement system and does not reflect a risk that should be

priced. In addition, she states that estimation error impairs reliability of information,

inherent volatility can help increase relevance, and mixed measurement in general impairs

both reliability and relevance.

One of the FASB’s rationales for SFAS 159 is to mitigate the earnings volatility

caused by a problem of different accounting rules for financial assets and financial

liabilities, which is the mixed-measurement volatility. For example, before SFAS 159,

firms could offset the change in value of financial instruments against the change in value

of other financial instruments held for the purpose of hedging by following the hedging

accounting of SFAS 133. However, the hedging accounting was too complicated to apply

and thus the costs to comply with the requirements of SFAS 133 were high. Therefore, to

the extent that costly hedging accounting prevents firms from applying hedging accounting,

firms would record related financial assets and liabilities differently and this discrepancy

could induce unnecessary volatility in earnings (Guthrie et al 2011). However, under SFAS

159 firms don’t have to follow this costly and complex rules for hedging and they now can

measure and record almost all financial instruments at fair value.

On the other hand, because SFAS 159 provides all firms with an option to measure

financial assets and liabilities at fair value, even firms without a problem of unnecessary

earnings volatility caused by the mixed-measurement volatility can decide to record

74

financial instruments at fair value. This means that managers can decide to record financial

instruments at fair value for purposes other than the objective of reducing unnecessary

earnings volatility. In fact, Henry (2009) finds evidence that there were firms which had

decided to implement SFAS 159 opportunistically. In addition, according to SFAS 159, the

FVO can be applied instrument by instrument and thus there is a room for management

discretion in applying SFAS 159. Furthermore, the objective of SFAS 159 is

counterintuitive in that traditionally managers have resisted market values to avoid volatile

earnings and auditors have been hesitant to apply fair value to minimize the risk of attesting

to amounts that might go up or down temporally (Li 2006), which seems to be associated

with estimation error volatility and inherent volatility of fair values. In fact, Barth et al.

(1995) find evidence using a sample of banks that fair value based earnings are more

volatile than historical cost based earnings, and Hodder et al. (2006) reports that their self-

constructed full fair value income is much more volatile than comprehensive income or net

income under the current GAPP. The fair value option of SFAS 159 is primarily intended

to eliminate mixed-measurement volatility for financial instruments. However, increased

use of fair values is likely to increase the other two sources of earnings’ volatility if FVO

is not used in accordance of the intent of SFAS 159. Thus, it would be an interesting

empirical question to examine whether the earnings’ volatility of FVO adopters have in

fact decreased or not in order to examine the effectiveness of SFAS 159. Therefore, I first

examine the following null hypothesis:

H1: The SFAS 159 does not help improve earnings volatility.

Analysts’ forecast accuracy

75

Accurate earnings forecasts are important in that a number of applications of

accounting data, such as in valuation research and practice, require the prediction of

earnings. Reliably projected future earnings are the key for the valuation models and

investments in capital market. However, it is widespread managerial belief that earnings

volatility reduces earnings predictability (Dichev and Tang 2009). In fact, a number of

studies report a negative relationship between earnings volatility and earnings prediction.

For example, Graham et al (2005) state that earnings volatility is associated with more

disagreement among analysts about earnings forecasts. Considering the reported

relationship between earnings volatility and earnings predictability, if the SFAS 159 has an

impact on earnings volatility, we can also reasonably conjecture that it can affect earnings

predictability. However, because the effectiveness of SFAS 159 is not determinable in

advance, I test the following null hypothesis:

H2: The SFAS 159 does not have an impact on analyst forecast accuracy.

Management discretion (Opportunistic decision)

The results for the first hypothesis indicate that SFAS 159 is not effective in

reducing earnings volatility. Then, the next question we might naturally have would be

about why firms have decided to adopt FVO. Thus, I have developed and examined two

more hypotheses to explore the question. Under SFAS 159, firms are allowed to apply FVO

by an instrument on instrument basis without justification for their decisions. This

instrument by instrument option result in reporting at fair value only part of financial

instruments and this discretionary accounting measurement may provide managers with

opportunities for earnings management. The office of the Chief Accountant of the SEC

76

states on its report for a Congressional committee that the benefit of requiring fair value

accounting for all financial instruments is to reduce opportunities for earnings management

and complexity of financial reporting created by the mixed attribute accounting model

(SEC 2005). However, SFAS 159 does not require all financial instruments to be recorded

at fair value and thus does not improve the problem of the mixed attribute accounting, but

rather increases the managements’ discretion in measuring financial instruments. Also, fair

value itself is inherently less reliable especially when measuring fair values of financial

assets or liabilities for which active markets do not exist. In such cases, managers should

estimate fair values with their own assumptions that can be subject to discretion or

manipulation (Landsman 2007). On the other hand, it should be also noted that FVO is

irrevocable and it is very difficult to accurately forecast the changes in values of financial

instruments in the long term. Therefore, it would be more reasonable to conjecture that the

decision for adopting FVO would be for a temporary one-time event in the near future if

the decision is made with an opportunistic intention. One potential candidate could be the

needs for external financing in the near future. For example, if a company has a plan for

external financing in a short period of time, the company may have more incentives to

manage earnings in order to send a more positive signal to the market and to make its

financial position look better at least temporarily before the external financing. Thus, I

conjecture and test the following hypothesis:

H3: Firms are more likely to adopt FVO in the year preceding the year when they

have needs for external financing.

Provision of more relevant information on financial instruments

77

The decision to adopt FVO might be viewed from a perspective of providing better

(more relevant) information on firms’ financial position. It is generally agreed that fair

value accounting provides more relevant information than historical cost accounting. As

the complexity of financial instruments owned by a company increases, the demand from

the market for more relevant information on them may also increase. Thus, firms with more

complex financial instruments have greater needs to provide information on their current

fair values. I try to examine this conjecture using exotic features of bonds issued. While

some bonds are issued without any additional features, other bonds are issued with some

exotic features, such as convertible, redeemable, and asset-backed. It is difficult to assess

exotic features of bonds and to properly reflect their values on financial statements under

historical cost accounting. Therefore, firms with more bonds with exotic features may have

greater motivation to adopt fair value accounting to provide better information on them. In

this regard, I conjecture and test the following hypothesis:

H4-1: Firms which have bonds with more exotic features are more likely to adopt

FVO.

In addition, if the motivation of firms to adopt FVO is for providing more relevant

information on complex financial instruments, it could mean that FVO may expedite

transactions of complex financial instruments. If firms adopt FVO because they care the

needs of investors for relevant information on financial statements, those firms might be

more willing to acquire complex financial instruments than before FVO is available

because the burden to provide relevant information could be lower with FVO. Thus, I also

conjecture and test the following hypothesis:

78

H4-2: FVO adopters are more likely to issue bonds with exotic features after

adopting FVO than firms that do not adopt FVO.

V. RESEARCH DESIGN

Sample selection

In order to identify the firms that have adopted the FVO, I took an advantage of the

item ‘TFVCE’ in COMPUSTAT. The ‘TFVCE’ stands for ‘total fair value change in

earnings’ and it reflects the gains or losses resulting from the change in fair value of

financial assets or liabilities which are included in reported earnings of income statement

only by SFAS 159. I have also cross-checked the reliability of this item by examining actual

10-Ks of ten firms, and the numbers provided by COMOUSTAT were reliable. Thus, I

assume that the first firm-year whose ‘tfvce’ item has a number other than ‘0’ is the first

year for the firm to adopt FVO. Also, the firms that have not had any numbers other than

‘0’ for ‘tfvce’ item for any years are classified as non-FVO firms. Through this process, a

total number of 331 unique firms were identified to adopt FVO. However, the sample size

was reduced to 216 because data for 8 consecutive years surrounding the first FVO

adoption year are not available, which is necessary for computing standard deviations of

earnings, or information for total assets at the year preceding the first FVO adoption year

(time t-1), which is necessary for identifying matched non-FVO firms, is not available for

115 firms. Standard deviation of net income was used as a proxy for earnings’ volatility

and it was computed using four years’ net incomes before and after FVO. In order to

conduct difference in differences analyses, each FVO firm was matched with a non-FVO

firm on industry and size at the year preceding the first FVO adoption year. Firms were

79

first matched by the four-digit Standard Industrial Classification (SIC) code, and then

matched by total assets.14 Also, in the matching procedure, non-FVO firms were allowed

to be selected more than once. The matching procedure resulted in unique 216 FVO firms

and 195 non-FVO firms, but 216 pairs of firms. In addition, 14 firms were lost due to the

unavailability of information for control variables for the first hypothesis testing which is

about volatility of earnings. Therefore, the final sample size includes 202 pairs of FVO

firms and non-FVO firms. Panel A of Table 2.1 presents the distribution of 202 FVO firms

by the first FVO adoption year and Panel B presents the distribution of the firms by industry.

While it seems that there are some early adopters who adopted FVO in 200715, most firms

adopted FVO for the first time in 2008 which is the first year when FVO became available.

Also, many FVO adopters belong to financial industry, and a significant number of

manufacturing firms have also adopted FVO.


Model Specification

I have tested the association between FVO and earnings volatility by estimating

multivariate OLS regressions using difference in differences methodology. Difference in

differences analysis has advantages over single cross-sectional difference analysis or single

time-series difference analysis in that the single cross-sectional difference estimator is

biased if there are any unobservable differences between the treatment and control groups

14 If selected non-FVO firms do not have 8 consecutive years’ data, those firms were replaced with

other non-FVO firms in order to minimize the number of firms lost due to data unavailability. 15 Henry (2009) identifies 35 early adopters (commercial banks), among which 11 firms rescinded

or revised their FVO elections. She identifies early adopters by searching 10-Qs and financial press.

The 202 FVO firms presented in Table 1 are the only firms used for hypothesis testing. Please refer

to Table 12 to see the number of all FVO firms identified.

80

and the single time-series difference estimator is biased if there exist time trends affecting

the outcome variable that are due to forces other than the treatment. The difference in

differences analysis combines those two estimators to avoid the problems of those methods

(Roberts and Whited 2011). The dependent variable, earnings volatility, is measured as four

year moving standard deviation of net income. The independent variable of my interest is

FVO_After which is an interaction term of two indicator variables, FVO and After. The

variable FVO is an indicator variable which differentiates FVO firms form non-FVO firms.

The variable After is also an indicator variable which differentiates pre-FVO period from

post-FVO period. Thus, the variable FVO_After takes ‘1’ if a firm year is one of the post-

FVO period years for FVO firms. Consequently, by using an interaction term, I examine

the difference of differences, and the coefficient of FVO_After indicates how much

standard deviations of net income for FVO firms change compared to those for non-FVO

firms. In addition, I conjecture that the greater the size of gains or losses resulting from

FVO is, the more significantly the earnings’ volatility is affected. Thus, I also include an

independent variable ‘FVO_Size_sum’ which is a four-year moving sum of ‘tfvce’ in the

second model in order to examine the effect of the size of gains or losses resulting from

FVO on the earnings volatility. Also both models include several control variables which

are likely to have impact on earnings volatility. Table 2.2 summarizes the definitions of

variables used in this study. The resulting models are as follows:

Model 1-1:

𝑁𝐼_𝑠𝑡𝑑𝑖,𝑡 = 𝛽0 + 𝛽1𝐹𝑉𝑂𝑖,𝑡 + 𝛽2𝐴𝑓𝑡𝑒𝑟𝑖,𝑡 + 𝛽3𝐴𝐹𝑡𝑒𝑟_𝐹𝑉𝑂𝑖,𝑡 + 𝛽4𝑇𝐴_𝑎𝑣𝑒𝑖,𝑡

+ 𝛽5𝐷𝑒𝑏𝑡_𝑎𝑣𝑒𝑖,𝑡 + 𝛽6𝐵𝑇𝑀_𝑎𝑣𝑒𝑖,𝑡 + 𝛽7𝑅𝑂𝐴_𝑎𝑣𝑒𝑖,𝑡

+ 𝛽8𝑆𝑝𝑒𝑐𝑖𝑎𝑙 _𝑠𝑢𝑚𝑖,𝑡 + 𝛽9𝐿𝑜𝑠𝑠_𝑠𝑢𝑚𝑖,𝑡 + 𝜀𝑖,𝑡

Model 1-2:

81

𝑁𝐼_𝑠𝑡𝑑𝑖,𝑡 = 𝛽0 + 𝛽1𝐹𝑉𝑂𝑖,𝑡 + 𝛽2𝐴𝑓𝑡𝑒𝑟𝑖,𝑡 + 𝛽3𝐴𝐹𝑡𝑒𝑟_𝐹𝑉𝑂𝑖,𝑡 + 𝛽4𝐹𝑉𝑂_𝑆𝑖𝑧𝑒_𝑠𝑢𝑚𝑖,𝑡

+ 𝛽5𝑇𝐴_𝑎𝑣𝑒𝑖,𝑡 + 𝛽6𝐷𝑒𝑏𝑡_𝑎𝑣𝑒𝑖,𝑡 + 𝛽7𝐵𝑇𝑀_𝑎𝑣𝑒𝑖,𝑡 + 𝛽8𝑅𝑂𝐴_𝑎𝑣𝑒𝑖,𝑡

+ 𝛽9𝑆𝑝𝑒𝑐𝑖𝑎𝑙 _𝑠𝑢𝑚𝑖,𝑡 + 𝛽10𝐿𝑜𝑠𝑠_𝑠𝑢𝑚𝑖,𝑡 + 𝜀𝑖,𝑡


In order to test the second hypothesis, I also estimate a multivariate OLS regression.

The information of analyst forecasts were extracted from IBES. For the forecasted EPS, I

used the latest mean forecast for each forecast period end date. The forecast error,

dependent variable (Per_fe), is a forecast error in percentage which is measured by dividing

mean forecast minus actual EPS by actual EPS multiplied by 100. The independent

variables of my interest in the model for examining the association between the FVO and

analyst forecast accuracy are FVO_After and FVO_size. The variable FVO_Size is the

magnitude of gains or losses from FVO, and it can be also regarded as another interaction

term of FVO_After and size of gains or losses from FVO in that only firm years of the

post-FVO period for FVO firms can have numbers for variable FVO_size. The size of gain

or loss used for the second hypothesis is measured by dividing the value of item ‘tfvce’ by

income before extraordinary items multiplied by 100. I also include several control

variables which have been identified in prior studies to have significant impact on analyst

forecast accuracy. Thus, the resulting model is as follows:

Model 2:

𝑃𝑒𝑟_𝑓𝑒𝑖,𝑡 = 𝛽0 + 𝛽1𝐹𝑉𝑂𝑖,𝑡 + 𝛽2𝐴𝑓𝑡𝑒𝑟𝑖,𝑡 + 𝛽3𝐴𝐹𝑡𝑒𝑟_𝐹𝑉𝑂𝑖,𝑡 + 𝛽4𝐹𝑉𝑂_𝑆𝑖𝑧𝑒𝑖,𝑡

+ 𝛽5𝐿𝑜𝑠𝑠𝑖,𝑡 + 𝛽6𝐷𝑒𝑏𝑡𝑖,𝑡 + 𝛽7𝐿𝑁_𝑇𝐴𝑖,𝑡 + 𝛽8𝑁𝑢𝑚𝐸𝑠𝑡𝑖,𝑡 + 𝜀𝑖,𝑡

82

I have constructed the dependent variable, Financing, for the third hypothesis

following Bradshaw et al. (2006). They develop a measure for external financing to

examine relations between external financing and future earnings performance and analysts’

forecasts. The measure for net external financing includes both equity financing and debt

financing. In addition to this variable, I also used two more indicator variables to test the

hypothesis: Financing_d and High_Financing. Financing_d is an indicator variable to

capture whether a firm has a net external financing activity or not, and High_Financing is

also an indicator variable to capture whether the magnitude of external financing is greater

than median within the same year and industry. Using those three dependent variables, I

estimated OLS and logistic regression models to test the hypothesis. The independent

variables of my interest are After_FVO and Lag_FVO_Size. Because I conjecture that

firms may have incentives to adopt FVO for external financing in a better condition in the

near future, I regressed the external financing on lagged variables. The model used for

hypothesis testing is as follows:

Model 3:

𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑛𝑔𝑖,𝑡 = 𝛽0 + 𝛽1𝐹𝑉𝑂𝑖,𝑡 + 𝛽2𝐴𝑓𝑡𝑒𝑟𝑖,𝑡 + 𝛽3𝐴𝐹𝑡𝑒𝑟_𝐹𝑉𝑂𝑖,𝑡

+ 𝛽4𝐿𝑎𝑔_𝐹𝑉𝑂_𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽5𝐿𝑎𝑔_𝐿𝑁_𝑇𝐴𝑖,𝑡 + 𝛽6𝐿𝑎𝑔_𝐷𝑒𝑏𝑡𝑖,𝑡

+ 𝛽7𝐿𝑎𝑔_𝑅𝑂𝐴𝑖,𝑡 + 𝛽8𝐿𝑎𝑔_𝑆𝑝𝑒𝑐𝑖𝑎𝑙𝑖,𝑡 + 𝛽9𝐿𝑎𝑔_𝐿𝑜𝑠𝑠𝑖,𝑡

+ 𝛽10𝐿𝑎𝑔_𝐵𝑇𝑀𝑖,𝑡 + 𝜀𝑖,𝑡

The two hypotheses on relevant information were tested using exotic features of

bonds issued. Three variables related with bonds’ exotic features were constructed: Exotic,

Exodic_d, and High_Exotic. The variable Exotic measures how many exotic features are

83

attached to the bonds issued. Exotic_d is an indicator variable which differentiates bonds

with exotic features from bonds without exotic features. High_Exotic is also an indicator

variable which differentiates bonds with more exotic features from bonds with fewer exotic

features. The information for bonds was extracted from Mergent Fixed Income Securities

Database. The model for H4-1 is Model 4-1, where FVO is dependent variable and

variables for exotic features are independent variables. The model for H4-2 is Model 4-2,

where variables for exotic features are dependent variables and FVO and After_FVO are

included as independent variables. While Model 4-1 use only observations of pre-FVO

period, Model 4-2 use observations of both pre-FVO and post-FVO periods. The control

variables included in Model 4-2 are lagged variables because firm features in the previous

year are more likely to affect bond issuance of the current year than firm features of the

current year. The resulting models for last two hypotheses are as follows:

Model 4-1:

𝐹𝑉𝑂𝑖,𝑡 = 𝛽0 + 𝛽1𝐸𝑥𝑜𝑡𝑖𝑐𝑖,𝑡 + 𝛽4𝐿𝑁_𝑇𝐴𝑖,𝑡 + 𝛽5𝐷𝑒𝑏𝑡𝑖,𝑡 + 𝛽6𝑅𝑂𝐴𝑖,𝑡

+ 𝛽7𝑆𝑝𝑒𝑐𝑖𝑎𝑙𝑖,𝑡 + 𝛽8𝐿𝑜𝑠𝑠𝑖,𝑡 + 𝛽9𝐼𝐶𝑊𝑖,𝑡 + 𝛽10𝐵𝑇𝑀𝑖,𝑡 + 𝜀𝑖,𝑡

Model 4-2

𝐸𝑥𝑜𝑡𝑖𝑐𝑖,𝑡 = 𝛽0 + 𝛽1𝐹𝑉𝑂𝑖,𝑡 + 𝛽2𝐴𝑓𝑡𝑒𝑟𝑖,𝑡 + 𝛽3𝐴𝐹𝑡𝑒𝑟_𝐹𝑉𝑂𝑖,𝑡 + 𝛽4𝐿𝑎𝑔_𝐿𝑁_𝑇𝐴𝑖,𝑡

+ 𝛽5𝐿𝑎𝑔_𝐷𝑒𝑏𝑡𝑖,𝑡 + 𝛽6𝐿𝑎𝑔_𝑅𝑂𝐴𝑖,𝑡 + 𝛽7𝐿𝑎𝑔_𝑆𝑝𝑒𝑐𝑖𝑎𝑙𝑖,𝑡

+ 𝛽8𝐿𝑎𝑔_𝐿𝑜𝑠𝑠𝑖,𝑡 + 𝛽9𝐿𝑎𝑔_𝐼𝐶𝑊𝑖,𝑡 + 𝛽10𝐿𝑎𝑔_𝐵𝑇𝑀𝑖,𝑡 + 𝜀𝑖,𝑡

Model 4-2 was also estimated using OLS or logistic regression models in accordance with

the features of dependent variables.

84

VI. RESULTS

1. Results for tests on earnings’ volatility


Table 2.3 reports the descriptive statistics of the sample of 808 observations, where

Panel A, B, and C display the distributional properties of the variables used for tests on

earnings volatility. Panel A shows the descriptive statistics of all observations including

FVO firms and non-FVO firms. Panel B exhibits the descriptive statistics of FVO firms

versus non-FVO firms. Panel C presents descriptive statistics of pre-FVO and post-FVO

periods. Panel B indicates that the standard deviation of net income of FVO firms is greater

than that of matched non-FVO firms and FVO firms reported losses more frequently than

non-FVO firms during the sample period. Panel C indicates that there is no statistical

difference in standard deviation of net income between pre-FVO period and post-FVO

period. However, firms seem to have reported special items and losses more frequently in

the post-FVO period than in the pre-FVO period. Panel D of Table 3 reports the pair-wise

Pearson correlations and Spearman correlations. The correlation table indicates that

earnings volatility is correlated with firm size, debt ratio, book to market ratio, and

reporting special items.



The multivariate test results for earnings volatility are presented in Table 2.4. The

coefficient of the interaction term, After_FVO, in the first model is positive, but not

statistically significant. In the second model where the variable ‘FVO_Size_sum’ is added

which is a variable for the size of gains or losses from FVO, the coefficient of

85

FVO_Size_sum is strongly significant and positive while the coefficient of After_FVO is

insignificant. In the both models, the coefficients for firm size and reporting loss are

positive and strongly significant. These results imply that even though the adoption of FVO

may affect the earnings volatility, the impact can be amplified by the size of gains or losses

by FVO.16


I also conducted the same analyses after dividing the sample into financial firms

and non-financial firms. The reason for this analysis is that because unlike non-financial

firms, financial firms’ financial statements are primarily comprised of financial instruments,

the motivation of non-financial firms for adopting FVO could be different from that of

financial firms and thus the effect of FVO on earnings’ volatility could be also different

between financial firms and non-financial firms. Table 2.5 and 2.6 report the test results for

financial firms and non-financial firms respectively. However, the results are qualitatively

the same to those of the full sample and I could not find any significant differences between

the two groups.

[Insert Tables 2.5 and 2.5 about here]

2. Results for tests on forecast accuracy


Table 2.7 reports the descriptive statistics of the sample of 2,387 firm years, where

Panel A shows the distributional properties of the variables used for tests on forecast

16 I also conducted the same analyses using different non-FVO firms matched by the propensity

score matching method, but the results are qualitatively the same to the results reported in Table 4.

86

accuracy and Panel B reports the pair-wise Pearson correlations and Spearman

correlations.17 Like the sample used for the first hypothesis, the sample for testing on

forecast accuracy includes observations which are not four years earlier or not four years

(including adoption year) later than the FVO adoption year.18 Panel A indicates that the

mean of forecast error is about 20 percent of actual EPS, about 25 percent of sample firm

years reported loss, and the mean of number of forecasts for each firm year is around 8.4.

Panel B indicates that forecast error is positively correlated with reporting loss, debt ratio,

and negative change in EPS, and negatively correlated with number of forecasts.



The test results for analyst forecast accuracy are presented in Table 2.8. The

variables of my interest are After_FVO and FVO_Size. Similar to the results for earnings’

volatility, the coefficient of FVO-Size is positive and strongly significant while the

coefficient of After_FVO is not significant. Reporting loss, firm size, and negative change

in EPS seem to be positively associated with forecast error, and the number of forecasts is

negatively associated with forecast error. The implication from these results is similar to

that of analyses for earnings’ volatility. That is, rather than FVO adoption itself, the

magnitude of gains or losses from FVO can affect the analysts’ forecast accuracy.


3. Results for tests on financing activity

17 Because data sets used for each hypothesis testing are different in terms of sample period and

types of variables, descriptive statistics of the samples are separately presented for each hypothesis

test. 18 216 pairs of FVO firms and non-FVO firms were used for last three hypotheses testing.

87


Table 2.9 reports the descriptive statistics of the sample of 1,645 firm years used

for the third hypothesis, where Panel A shows the distributional properties of the variables

used for tests on financing activity and Panel B reports the pair-wise Pearson correlations

and Spearman correlations. The sample size decreased significantly because I deleted firm

years which are more than two years earlier or more than one year later than the FVO

adoption year. Because the third hypothesis is for testing the managers’ opportunistic

motivation of adopting FVO for the near future (the first year after adopting FVO), I have

shorten the sample period. The mean of Financing_d indicates that about 44 percent of firm

years had positive net financing activity. The variables for financing activity appear to be

positively correlated with reporting loss, and negatively correlated with firm size, ROA,

and book to market ratio.



Table 2.10 presents results of OLS regressions using the variable Financing as a

dependent variable, and Table 2.11 presents results of logistic regressions using the

variables Financing_d and High_Financing. I could not find significant results for

After_FVO and Lag_FVO_Size from OLS models. The logistic regression models also

yield no significant results for those variables. Next, I divided the sample into financial

firms and non-financial firms as I did for volatility analyses, and ran the same regressions

for each of the subsamples. However, no significant results for After_FVO and

Lag_FVO_Size were found, and also no significant differences between financial firms

and non-financial firms were found (untabulated).

88


However, I was curious about why some firms adopt FVO as soon as it becomes

available while other firms delay adopting FVO. Thus, I took a look at distribution of

number of firms adopting FVO by the first FVO adoption year and checked whether there

is any difference in number of firms reporting gains and losses. As displayed in Table 2.

12, there exists a significant difference between financial firms and non-financial firms that

adopted FVO in 2008, which is the first year when the FVO became effective for regular

adopters. In 2008, the number of non-financial firms reporting gains is significantly higher

than number of firms reporting losses while the statistics is opposite for financial firms.


Therefore, I conducted similar analyses using only firms which adopted FVO in

2008. Table 2.13 reports the results of logistic regressions which use financial firm and

non-financial firm samples separately. Non-financial firm sample yields a positive and

significant coefficient for After_FVO while no significant results were found from the

financial firm sample. Furthermore, using both financial firms and non-financial firms

together, another regression analysis was conducted which includes different indicator

variables: NFFirm, an indicator variable which differentiates financial firms from non-

financial firms, and After_NFFirm, an interaction of NFFirm and After. The results are

presented in Table 2.14. The coefficient of After_NFFrim is positive and significant, which

indicates that financing activity of non-financial firms after adopting FVO is higher than

that of financial firms.


89

4. Results for tests on exotic features of bonds


Table 2.15 reports the descriptive statistics of the sample of 410 firm years for

testing last two hypotheses, where Panel A shows the distributional properties of the

variables used for tests on exotic features of bonds and Panel B reports the pair-wise

Pearson correlations and Spearman correlations. In many cases, there are several bond

issues for the same firm and same year. In those cases, only one bond issue was kept and

others were deleted to avoid over-representation of one firm with many bond issues for one

year. In cases where one firm issues both bonds with exotic features and bonds without

exotic features, only one bond issue from bonds with exotic features and another one bond

issue from bonds without exotic features were kept. As described in Table 2.2, the variable

Exotic is a numerical variable which captures the number of exotic features of bonds issued

and the variable Exotic_d is an indicator variable which takes 1 if a bond has at least one

exotic feature and zero otherwise. Panel A indicates that about 75 percent of bond issues in

the sample have at least one exotic feature. Also, Panel B indicates that variables for exotic

features of bonds are positively correlated with reporting loss in the previous year, and

negatively correlated with previous year’s firm size and debt ratio.



For the hypothesis 4-1, the variables, Exotic, Exotic_d, and High_Exotic, were

included in turn as an independent variable in Model 4-1. However, none of three logistic

regressions yielded significant results (untabulated). As to the hypothesis 4-2, as I did for

financial activity analyses, I also ran both OLS and logistic regressions using those three

90

variables for exotic features as a dependent variable in turn. Table 2.16 reports the results

of an OLS regression and Table 2.17 presents results of logistic regressions. No significant

results for the variable After_FVO were found from all three regression models.


I also ran the same regressions using all available bond issues without deleting any

observations during the sample period. However, even though the significance increased

in general probably due to the increase in sample size (the sample size was 1,257), I could

not find results which are consistent with the developed hypothesis (untabulated).

VII. CONCLUSION, LIMITATIONS, AND EXTENSIONS

In this study, I have first examined the effectiveness of FVO of SFAS 159 in

mitigating earnings’ volatility and its effect on analysts’ forecast accuracy. The test results

indicate that the earnings’ volatility of FVO adopters have in fact increased and the impact

is amplified in accordance with the size of gains or losses resulting from the FVO. Also,

the test results of the second hypothesis indicate that analysts’ forecast error of FVO

adopters increased more than that of non-FVO adopters, which is consistent with the

positive relation reported in prior studies between earnings’ volatility and analysts’ forecast

error. Given the results for the first two hypotheses, I have also tried to explore the

motivation of firms to adopt FVO. I conjectured two more hypotheses to try to identify

potential motives underlying FVO adoption (i) an opportunistic intention and (ii) an

informative intention. Unfortunately, I could not find results which support either

hypothesis other than some limited evidence for opportunistic intention.

The results of this study cast doubt on the effectiveness of SFAS 159. This study

91

implies that we need to examine more thoroughly whether SFAS 159 provides managers

with excessive discretion without achieving its intended purpose. Granting full discretion

to managers on which financial instrument is reported at fair value and thus whether

changes in fair values of specific financial instruments are included in net income might

deteriorate reliability and comparability of financial statements without lowering earnings’

volatility which is not even a main purpose of financial reporting. This study provides some

limited evidence for managers’ opportunistic motivation of adopting FVO. Further studies

with more refined data and more robust research methodologies which explore the

intentions of FVO adopters would be interesting.

Limitations and Extensions

Testing hypotheses 4-1 and 4-2 using exotic features of bonds issues may not be

appropriate. I hypothesize that firms with more complex financial instruments are more

likely to adopt FVO. However, I didn’t not test the hypothesis using actual features of

financial instruments owned by firms. Rather, I used information on bonds issued for

several years before adopting FVO. The recently issued bonds may not properly represent

the characteristics of all financial instruments held by the firm.

Examining bond covenants can be a possible future research topic. For example,

investigating whether firms adopt FVO in order to reduce the likelihood of violating bond

covenants or whether bond covenants are adjusted to reflect the effects of FVO would be

interesting. In fact, Demerjian et al. (2016) reports evidence that covenant definitions of

private loan contracts are modified to exclude the effects of FVO.

92

Examining the association between adopting FVO and CEO tenure would also be

interesting. As mentioned in the hypothesis development section, if managers adopt FVO

opportunistically, they would do so for a short term effect as it is very difficult to forecast

accurately the fair values of financial instruments in the long term; because CEOs who

are going to retire in the near future may have more incentives to manage short term

performances, those CEOs are more likely to adopt FVO.

Another possibility is that even if we suppose that results were found indicating

that earnings’ volatility decreased after adopting FVO, it would not necessarily mean that

the FVO was used in accordance with the intent of SFAS 159. For example, if firms use

FVO for smoothing earnings rather than for mitigating the problem of mixed-measurement

accounting, earnings’ volatility would also decrease. This practice is obviously not in

accordance with the intent of SFAS 159, and can be regarded as an example of

opportunistic usage of FVO. This implies that when there is too much discretion allowed

or not enough guidance provided for certain accounting standards, it can often be difficult

to assess whether the standards work in compliance with their intended objective. Thus,

given the current movement toward principle-based accounting from rule-based

accounting, it is also an important issue to determine how much additional guidance should

be provided in order to achieve each standard’s intended objective while simultaneously

avoiding the difficulties that arise from the specific guidance that reduce the effectiveness

of principle-based standards.

93

REFERENCES

Aboody, D., M. E. Barth, and R. Kasznik. 2006. Do firms understate stock-based

compensation expense disclosed under SFAS 123? Review of Accounting Studies 11:

429–461.

Ahmed, A. S., E. Kilic, and G. J. Lobo. 2006. Does recognition versus disclosure matter?

Evidence from value-relevance of banks’ recognized and disclosed derivative

financial instruments. The Accounting Review 81 (3): 567–588.

Allen, F., and E. Carletti. 2008. Mark-to-market accounting and liquidity pricing. Journal

of Accounting and Economics 45: 358–378.

American Accounting Association (AAA), Financial Accounting Standards Committee

(FASC). 2007. Response to FASB Exposure Draft, “The Fair Value Option for

Financial Assets and Financial Liabilities, Including an Amendment of FASB

Statement No. 115”. Accounting Horizon 21 (2): 189-200.

Barth, M. E. 1994. Fair value accounting: evidence from investment securities and the

market valuation of banks. The Accounting Review 69 (1): 1–25.

Barth, M. E. 2004. Fair values and financial statement volatility. In The Market Discipline

across Countries and Industries, edited by C. Borio, W. C. Hunter, G. G. Kaufman,

and K. Tsatsaronis, 323-333. Cambridge, MA: MIT Press.

Barth, M. E., W. H. Beaver, and W. R. Landsman.1996. Value-relevance of banks’ fair value

disclosures under SFAS 107. The Accounting Review 71 (4): 513–537.

———, ———, and ———. 2001. The relevance of the value relevance literature for

accounting standard setting: another view. Journal of Accounting and Economics 31:

77–104.

Barth, M. E. and W. R. Landsman. 1995. Fundamental issues related to using fair value

accounting for financial reporting. Accounting Horizons 9 (4): 97–107.

———, ———, and J. M. Wahlen. 1995. Fair value accounting: Effects on banks’ earnings

volatility, regulatory capital, and value of contractual cash flows. Journal of Banking

& Finance 19: 577–605.

Bernard, V. L., R. C. Merton, and K. G. Palepu. 1995. Mark-to-market accounting for banks

and thrifts: lessons from the Danish experience. Journal of Accounting Research 33

(1): 1–32.

94

Black, E. L., K. F. Sellers, and T. S. Manly. 1998. Earnings management using asset sales:

an international study of countries allowing noncurrent asset revaluation. Journal of

Business Finance and Accounting 25 (9/10): 1,287–1,317.

Bradshaw, M. T., S. A. Richardson, and R. G. Sloan. 2006. The relation between corporate

financing activities, analysts’ forecasts and stock returns. Journal of Accounting and

Economics 42: 53–85.

Dechow, P., L. Myers, and C. Shakespeare. 2010. Fair value accounting and gains from

asset securitizations: A convenient earnings management tool with compensation

side-benefits. Journal of Accounting and Economics 49 (1): 2–25.

Demerjian, P. R., J. Donovan, and C. R. Larson. 2016. Fair value accounting and debt

contracting: Evidence from adopting of SFAS 159. Journal of Accounting Research

54 (4): 1041-1076.

Dichev, I. D., and V. W. Tang. 2009. Earnings volatility and earnings predictability. Journal

of Accounting and Economics 47: 160-181.

Eccher, E. A., K. Ramesh, and S. R. Thiagarajan. 1996. Fair value disclosures by bank

holding companies. Journal of Accounting and Economics 22: 79–117.

Fiechter, P. 2011. The effects of the fair value option under IAS 39 on the volatility of bank

earnings. Journal of International Accounting Research 10 (1): 85-108.

Financial Accounting Standards Board (FASB). 1984. Recognition and measurement in

financial statements of business enterprises. Statement of Financial Accounting

Concepts No. 5. Stamford, CT: FASB.

———. 1991. Disclosures about Fair Value of Financial Instruments. Statement of

Financial Accounting Standards No. 107. Norwalk, CT: FASB.

———. 1993. Accounting for Certain Investments in Debt and Equity Securities.

Statement of Financial Accounting Standards No. 115, Norwalk, CT: FASB.

———. 1994. Disclosure about Derivative Financial Instruments and Fair Value of

Financial Instruments. Statement of Financial Accounting Standards No. 119.

Norwalk, CT: FASB.

———. 1998. Accounting for Derivative Instruments and Hedging Activities. Statement

of Financial Accounting Standards No. 133. Norwalk, CT: FASB.

———. 2006. Fair Value Measurements. Statement of financial Accounting Standards No.

95

157. Norwalk, CT: FASB.

———. 2007. The Fair Value Option for Financial Assets and Financial Liabilities-

Including an Amendment of FASB Statement No. 115. Statement of financial

Accounting Standards No. 159. Norwalk, CT: FASB.

Glover, J. C., Y. Ijiri, C. B. Levine, and P. J. Liang. 2005. Separating facts from forecasts

in financial statements. Accounting Horizons 19 (December): 267–282.

Gorton, G., P. He, and L. Huang. 2006. Asset Prices When Agents Are Marked to Market.

NBER Working Paper No. 12075, National Bureau of Economic Research (NBER).

Graham, J. R., C. R. Harvey, and S. Rajgopal. 2005. The economic implications of

corporate financial reporting. Journal of Accounting and Economics 40: 3-73.

Guthrie, K., J. H. Irving, and J. Sokolowsky. 2011. Accounting choice and the fair value

option. Accounting Horizon 25 (3): 487-510.

Henry, E. 2009. Early adoption of SFAS No. 159: Lessons from games (almost) played.

Accounting Horizon 23 (2): 181-199.

Hirst, D. E., P. E. Hopkins, and J. E. Wahlen. 2004. Fair values, income measurement, and

bank analysts’ risk and valuation judgments. The Accounting Review 79 (2): 453–

472.

Hodder, L. D., P. E. Hopkins, and J. M. Wahlen. 2006. Risk-relevance of fair-value income

measures for commercial banks. The Accounting Review 81 (2): 337-375.

Hsu, P., and Y. Lin. 2016. Fair value accounting and earnings management. Eurasian

Journal of Business and Management. 4 (2): 41-54.

International Accounting Standards Board (IASB). 2005. Financial Instruments:

Recognition and Measurement-The Fair Value Option. International Accounting

Standard 39. London, U.K.: IASB.

Landsman W. R. 2007. Is fair value accounting information relevant and reliable? Evidence

from capital market research. Accounting and Business Research Special Issue:

International Accounting Policy Forum: 19-30.

Li, F. 2006. Fair Value Option: A review and assessment. Tennessee CPA Review (October):

4-6.

Nelson, K. K. 1996. Fair value accounting for commercial banks: an empirical analysis of

SFAS No. 107. The Accounting Review 71 (2): 161–182.

96

Plantin, G., H. Sapra, and H. S. Shin. 2005. Marking to market, liquidity and financial

stability. Monetary and Economic Studies (Special Edition/October): 133–155.

———, ———, and ———. 2008a. Fair value accounting and financial stability.

Financial Stability Review 12 (October): 85-94.

———, ———, and ———. 2008b. Marking-to-market: Panacea or Pandora’s Box?

Journal of Accounting Research 46 (2): 435-460.

Roberts M. R. and T. M. Whited. 2011. Endogeneity in empirical corporate finance.

Working paper, University of Rochester.

Securities and Exchange Commission (SEC). 2005. Report and Recommendations

Pursuant to Section 401(c) of the Sarbanes-Oxley Act of 2002 on Arrangements with

Off-Balance Sheet Implications, Special Purpose Entities, and Transparency of

Filings by Issuers. Washington, D.C.: SEC. Available at:

https://www.sec.gov/news/studies/soxoffbalancerpt.pdf.

Venkatachalam, M. 1996. Value-relevance of banks’ derivatives disclosures. Journal of

Accounting and Economics 22: 327–355.

https://www.sec.gov/news/studies/soxoffbalancerpt.pdf

97

Tables for Chapter 2

Table 2.1

Distribution of FVO firms by FVO adoption year and industry

Panel A: Distribution of FVO firms by the first FVO adoption year

Year Number of FVO firms

2007 30

2008 108

2009 35

2010 14

2011 12

2012 3

Sum 202

Panel B: Distribution of FVO firms by industry

FVO firms represented by the following SIC codes Number of firms

01-09: Agriculture, Forestry, Fishing 1

10-14: Mining 2

15-17: Construction 4

20-39: Manufacturing 57

40-49: Transportation, Communications, Utilities 5

52-59: Retail 7

60-67: Finance, Insurance, Real Estate 100

70-89: Services 24

91-99 Public Administration 2

Total 202

98

Table 2.2 Variable Definitions

NI_std = Four year moving standard deviation of net income, winsorized at ± 1;

FVO = 1 if the firm is a FVO firm, 0 otherwise;

After = 1 if the firm year is an post-FVO adoption year, 0 otherwise;

After_FVO = Interaction term of FVO and After;

FVO_Size_sum = Four year moving sum of absolute value of the item 'tfvce'19;

TA_ave = Four year moving average of natural log of total assets;

Debt_ave = Four year moving average of debt to total assets ratio;

BTM_ave = Four year moving average of book to market ratio;

ROA_ave = Four year moving average of return on assets ratio;

Special_sum = Four year moving sum of a dummy variable for special item;

Loss_sum = Four year moving sum of a dummy variable for loss;

Per_FE = Forecast error in percentage, absolute value of mean forecast less actual eps

divided by actual eps times 100, winsorized at ± 1;

FVO_size = Proportion of gain or loss from FVO in earnings (‘tfvce’ divided by income

before extraordinary items times 100);

Loss = 1 if the firm reports a net loss, 0 otherwise;

Special = 1 if the firm reports special items, 0 otherwise;

Debt = Debt to total assets ratio;

LN_TA = Natural log of total assets;

NumEst = Number of forecasts for each firm year.

Per_EPS = Change in eps in percentage, absolute value of current eps less previous year's

eps divided by previous year's eps times 100;

Neg_Ch_EPS = 1 if change in eps is negative, 0 otherwise;

Per_EPS_Neg_Ch = Interaction term of per_eps and neg_ch_eps.

Financing = Net external financing, sum of equity financing and debt financing scaled by

average total assets, winsorized at ± 1

*Net equity financing: sale of common and preferred stock less purchase of

common and preferred stock less cash payments of dividends

*Net debt financing: issuance of long-term debt less payments for long-term

debts reductions plus net changes in current debt;

19 The ‘tfvce’ stands for ‘total fair value change in earnings’, and it reflects the gains or losses

resulting from the change in fair value of financial assets or liabilities which are included in reported

earnings of income statement only by SFAS 159.

99

Financing_d = 1 if variable 'financing' is positive, 0 otherwise;

High_Financing = 1 if 'financing' is greater than median 'financing' within the same year and

industry (1 digit sic code) group, 0 otherwise;

Lag_FVO_Size = Absolute value of the amount of 'tfvce' divided by net income at the end of

the previous year;

ROA = Return of assets

ICW = 1 if the firm is identfied as an internal control weakness firm, 0 otherwise

(Compustat item 'AUOPIC');

BTM = Book to market ratio.

Lag_LN_TA = Natural log of total assets at the end of the previous year;

Lag_Debt = Debt to total assets ratio at the end of the previous year;

Lag_ROA = Return of assets at the end of the previous year;

Lag_Special = 1 if the firm reports special items in the previous year, 0 otherwise;

Lag_Loss = 1 if the firm reports a net loss in the previous year, 0 otherwise;

Lag_ICW = 1 if the firm is identfied as an internal control weakness firm in the previous

year, 0 otherwise (Compustat item 'AUOPIC');

Lag_BTM = Book to market ratio at the end of the previous year.

NFFirm = 1 if the firm is a non-financial firm (firm whose sic code doesn’t start with 6),

0 otherwise;

After_NFFirm = Interaction term of NFFirm and After;

Exotic = Numerical variable for exotic features of bonds issued.

*The variable gets 1 point for each of the following features: asset backed,

convertible, redeemable, puttable, refundable, soft call, enhancement, and

exchangeable.

Exotic_d = 1 for bonds issued with exotic features

High_Exotic = 1 if 'exotic' is greater than median 'exotic' within the same year and industry

(1 digit sic code) group, 0 otherwise;

100

Table 2.3

Descriptive Statistics (for H1)

Panel A: Distributional Properties of Variables (All Sample), N=808 Variable Mean Median Std Dev 1Q 3Q

NI_std

404.488

40.434

1113.580

9.275

228.799

NI1_std

7.514

1.718

20.002

0.366

5.741

FVO

0.500

0.500

0.500

0.000

1.000

After

0.500

0.500

0.500

0.000

1.000

After_FVO

0.250

0.000

0.433

0.000

0.500

FVO_Size_sum

267.513

0.000

2366.390

0.000

0.001

tfvce1_sum

0.637

0.000

1.271

0.000

0.500

tfvce2_sum

51.145

0.000

655.740

0.000

0.003

TA_ave

7.696

7.483

2.758

5.861

9.353

Debt_ave

0.320

0.132

3.135

0.039

0.292

BTM_ave

0.246

0.630

7.963

0.384

0.982

ROA_ave

-0.169

0.008

3.786

-0.011

0.041

Special_sum

2.423

3.000

1.377

1.000

4.000

Loss_sum

1.134

1.000

1.396

0.000

2.000

101


Panel B: Distributional Properties of Variables (FVO vs Matched firm sample)a

FVO Firm Sample, N=404

Matched Firm Sample, N=404

Variable

Mean

Median

Std Dev

Mean

Median

Std Dev

NI_std 529.165 42.165 1351.900 279.812 36.662 790.185

NI1_std 7.390 1.930 20.368 7.637 1.468 19.652

After 0.500 0.500 0.501 0.500 0.500 0.501

TA_ave 7.795 7.573 2.805 7.596 7.391 2.709

Debt_ave 0.439 0.139 4.424 0.201 0.122 0.289

BTM_ave -0.257 0.624 11.214 0.750 0.637 0.855

ROA_ave -0.298 0.008 5.346 -0.039 0.009 0.301

Special_sum 2.473 3.000 1.348 2.374 2.000 1.406

Loss_sum 1.243 1.000 1.411 1.025 0.000 1.375

Panel C: Distributional Properties of Variables (Pre-FVO vs Post-FVO)a

Before FVO, N=404 After FVO, N=404

Variable Mean Median Std Dev Mean Median Std Dev

NI_std 330.310 32.292 849.743 478.667 44.828 1322.920

NI1_std 6.824 1.679 17.218 8.204 1.780 22.443

FVO 0.500 0.500 0.501 0.500 0.500 0.501

TA_ave 7.592 7.375 2.715 7.799 7.609 2.799

Debt_ave 0.215 0.137 0.300 0.425 0.128 4.423

BTM_ave 0.584 0.541 0.542 -0.092 0.830 11.245

ROA_ave -0.053 0.010 0.559 -0.284 0.006 5.326

Special_sum 2.208 2.000 1.368 2.639 3.000 1.354

Loss_sum 0.948 0.000 1.344 1.319 1.000 1.425

102


Panel D: Pearson Correlation (top) and Spearman Correlation (bottom) b

1 2 3 4 5 6 7 8 9 10 11 12 13 14

NI_std 1

-0.05 0.11 0.07 0.12 0.53 0.17 0.02 0.57 0.00 -0.14 0.02 0.11 -0.01

NI1_std 2 0.03

-0.01 0.03 0.02 -0.02 -0.01 -0.01 -0.42 0.27 -0.20 -0.30 0.01 0.41

FVO 3 0.06 0.05

0.00 0.58 0.11 0.50 0.08 0.04 0.04 -0.06 -0.03 0.04 0.08

After 4 0.08 0.04 0.00

0.58 0.11 0.50 0.08 0.04 0.03 -0.04 -0.03 0.16 0.13

After_FVO 5 0.08 0.06 0.58 0.58

0.20 0.87 0.14 0.04 0.06 -0.10 -0.05 0.11 0.13

FVO_Size_sum 6 0.09 0.06 0.53 0.47 0.85

0.29 0.09 0.24 0.01 -0.10 0.01 0.09 0.03

tfvce1_sum 7 0.07 0.05 0.68 0.44 0.90 0.76

0.19 0.11 0.02 -0.15 0.00 0.11 0.13

tfvce2_sum 8 0.08 0.03 0.49 0.49 0.85 0.74 0.79

0.03 0.01 0.01 0.00 0.00 0.00

TA_ave 9 0.75 -0.58 0.03 0.04 0.04 0.04 0.05 0.06

-0.15 -0.02 0.19 0.19 -0.42

Debt_ave 10 0.26 -0.03 0.03 0.01 0.02 0.01 0.04 0.05 0.25

-0.17 -0.99 -0.03 0.06

BTM_ave 11 0.11 -0.28 0.04 0.27 0.19 0.17 0.16 0.19 0.26 -0.02

0.12 -0.01 -0.11

ROA_ave 12 0.07 -0.20 -0.03 -0.09 -0.05 -0.06 -0.07 -0.07 0.18 -0.13 -0.23

0.04 -0.08

Special_sum 13 0.32 0.12 0.03 0.18 0.12 0.11 0.09 0.12 0.17 0.12 0.05 -0.05

0.07

Loss_sum 14 -0.01 0.57 0.09 0.14 0.13 0.13 0.13 0.10 -0.39 0.08 0.08 -0.75 0.09

a Bold text indicates significant differences between two subsamples at the 0.05 level or better, two-tailed. Differences in means are assessed

using a t-test.

b Bold text indicates significance at the 0.05 level or better, two-tailed.

103

Table 2.4

OLS Regression Analyses of Earnings Volatility on Adoption of the FVO

(full sample)

Variable Model 1-1 Model 1-2

Estimate p-value Estimate p-value

Intercept -1390.86*** 0.0014 -1147.30*** 0.003

FVO 81.61 0.3194 100.04 0.1679

After 348.29* 0.0575 198.38 0.2218

After_FVO 88.98 0.4427 -91.34 0.3765

FVO_Size_sum 0.17*** <.0001

TA_ave 343.90*** <.0001 298.96*** <.0001

Debt_ave -101.82 0.1456 -81.44 0.1883

BTM_ave -12.94*** 0.0012 -9.91*** 0.005

ROA_ave -112.40* 0.0521 -92.25* 0.0716

Special_sum -64.53*** 0.0051 -70.09*** 0.0006

Loss_sum 198.68*** <.0001 164.70*** <.0001

Industry dummies Included Included

Year dummies Included Included

Adj. R-sq 0.4576 0.5756

F-statistic 26.22 40.08

n 808 808

Dependent variable NI_std NI_std

*, **, *** indicate statistical significance at the 10, 5, 1 percent level respectively.

104

Table 2.5


(financial firms only)



Intercept -3162.95*** <.0001 -2630.53*** <.0001

FVO 168.02 0.1953 211.50* 0.0599

After 634.51** 0.0234 491.41** 0.0425

After_FVO 160.66 0.3764 -139.34 0.3817

FVO_Size_sum 0.15*** <.0001

TA_ave 392.34*** <.0001 340.93*** <.0001

Debt_ave -79.13 0.7836 -304.97 0.2229

BTM_ave -16.66*** 0.0017 -13.14*** 0.0043

ROA_ave -1389.05 0.4949 -705.66 0.6886

Special_sum -2.88 0.9398 -11.84 0.7195

Loss_sum 168.84*** 0.007 161.03*** 0.0029



Adj. R-sq 0.5505 0.6639


n 400 400

Dependent variable NI_std NI_std


105

Table 2.6


(non-financial firms only)



Intercept -344.96 0.265 -336.18 0.2502

FVO 46.05 0.4782 42.45 0.4889

After -131.23 0.4682 -149.70 0.3812

After_FVO 0.18 0.9985 -44.72 0.6061

FVO_Size_sum 0.38*** <.0001

TA_ave 104.64*** <.0001 111.12*** <.0001

Debt_ave -25.63 0.6131 -40.06 0.4035

BTM_ave -3.33 0.4527 -5.14 0.2214

ROA_ave -29.41 0.4802 -41.74 0.2894

Special_sum -3.46 0.8611 -18.48 0.3256

Loss_sum 41.54** 0.026 44.69** 0.0113



Adj. R-sq 0.6814

0.7157 F-statistic

23.32

26.62 n

408

408 Dependent variable NI_std NI_std


106

Table 2.7


Panel A: Distributional Properties of Variables (All Sample), N=2387 Variable Mean Median Std Dev 1Q 3Q

Per_FE 20.199 4.393 49.730 1.397 14.286

FVO 0.506 1.000 0.500 0.000 1.000

After 0.530 1.000 0.499 0.000 1.000

After_FVO 0.266 0.000 0.442 0.000 1.000

FVO_size 6.278 0.000 51.300 0.000 0.000

LOSS 0.247 0.000 0.431 0.000 0.000

Special 0.627 1.000 0.484 0.000 1.000

Debt 0.195 0.114 0.236 0.027 0.283

LN_TA 8.245 8.038 2.503 6.428 9.830

NumEst 8.419 6.000 6.852 3.000 13.000

Per_EPS 142.281 33.014 907.202 14.103 82.609

Neg_Ch_EPS 0.389 0.000 0.488 0.000 1.000

Per_EPS_Neg_Ch 66.080 0.000 597.990 0.000 23.767

107


Panel B: Pearson Correlation (top) and Spearman Correlation (bottom) a

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

Per_FE 1 0.039 0.107 0.077 0.095 0.239 0.028 0.075 0.002 -0.151 0.026 0.192 0.036

FVO 2 0.039 -0.008 0.595 0.121 0.064 0.030 0.049 0.009 0.049 -0.022 -0.001 0.008

After 3 0.107 -0.008 0.568 0.115 0.121 0.120 -0.016 0.010 0.008 0.068 0.098 0.062

After_FVO 4 0.077 0.595 0.568 0.203 0.125 0.077 0.018 0.003 0.022 0.010 0.040 0.037

FVO_size 5 0.095 0.121 0.115 0.203 0.028 -0.011 0.024 0.051 -0.009 -0.004 0.053 -0.003

LOSS 6 0.239 0.064 0.121 0.125 0.028 0.099 0.147 -0.275 -0.156 0.078 0.242 0.162

Special 7 0.028 0.030 0.120 0.077 -0.011 0.099 0.092 0.080 0.114 0.031 0.033 0.015

Debt 8 0.075 0.049 -0.016 0.018 0.024 0.147 0.092 0.160 -0.065 0.052 0.054 0.047

LN_TA 9 0.002 0.009 0.010 0.003 0.051 -0.275 0.080 0.160 0.260 -0.011 -0.011 -0.004

NumEst 10 -0.151 0.049 0.008 0.022 -0.009 -0.156 0.114 -0.065 0.260 -0.062 -0.134 -0.052

Per_EPS 11 0.026 -0.022 0.068 0.010 -0.004 0.078 0.031 0.052 -0.011 -0.062 0.024 0.650

Neg_Ch_EPS 12 0.192 -0.001 0.098 0.040 0.053 0.242 0.033 0.054 -0.011 -0.134 0.024 0.138

Per_EPS_Neg_Ch 13 0.036 0.008 0.062 0.037 -0.003 0.162 0.015 0.047 -0.004 -0.052 0.650 0.138

a Bold text indicates significance at the 0.05 level or better, two-tailed.

108

Table 2.8

OLS Regression Analyses of Analysts’ Forecast Error on Adoption of the FVO


Estimate Pr > |t| Estimate Pr > |t|

Intercept -22.73 0.3685 -20.97 0.4059

FVO 4.94* 0.08 4.95* 0.0789

After 1.04 0.8052 0.58 0.8903

After_FVO -3.72 0.3342 -5.21 0.1785

FVO_size 0.06*** 0.0013

LOSS 22.69*** <.0001 22.69*** <.0001

Special 0.23 0.9146 0.39 0.8536

Debt -0.98 0.8473 -1.32 0.7958

LN_TA 1.99*** 0.0021 1.98*** 0.0021

NumEst -0.99*** <.0001 -0.99*** <.0001

Per_EPS 0.00 0.542 0.00 0.5587

Neg_Ch_EPS 9.92*** <.0001 9.75*** <.0001

Per_EPS_Neg_Ch 0.00 0.1489 0.00 0.1659



Adj. R-sq 0.116 0.1195


n 2387 2387

Dependent variable Per_FE Per_FE


109

Table 2.9




Financing_d 0.4365 0.0000 0.4961 0.0000 1.0000

Financing 0.9747 -0.3117 15.9051 -4.0187 2.8812

High_Financing 0.5058 1.0000 0.5001 0.0000 1.0000

FVO 0.5064 1.0000 0.5001 0.0000 1.0000

After 0.5052 1.0000 0.5001 0.0000 1.0000

After_FVO 0.2547 0.0000 0.4358 0.0000 1.0000

Lag_FVO_Size 0.0585 0.0000 0.6239 0.0000 0.0000

Lag_LN_TA 7.7920 7.5290 2.7112 5.9038 9.6123

Lag_Debt 0.2098 0.1192 0.2639 0.0214 0.2995

Lag_ROA -0.0488 0.0104 0.6806 -0.0034 0.0475

Lag_Special 0.5994 1.0000 0.4902 0.0000 1.0000

Lag_Loss 0.2748 0.0000 0.4465 0.0000 1.0000

Lag_BTM 0.6155 0.5808 1.9854 0.3376 0.8853

110



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

Financing_d 1 0.537 0.794 0.026 -0.068 -0.031 -0.019 -0.044 0.024 -0.087 -0.028 0.054 -0.042

Financing 2 0.859 0.515 -0.005 -0.063 -0.047 -0.034 -0.138 0.011 -0.149 -0.050 0.157 -0.041

High_Financing 3 0.794 0.818 0.026 0.009 0.023 -0.012 -0.038 -0.009 -0.078 -0.044 0.072 0.001

FVO 4 0.026 -0.005 0.026 -0.004 0.577 0.093 0.024 0.047 -0.032 0.014 0.071 -0.039

After 5 -0.068 -0.056 0.009 -0.004 0.579 0.093 0.012 0.028 -0.013 0.054 0.127 0.015

After_FVO 6 -0.031 -0.042 0.023 0.577 0.579 0.161 0.025 0.043 -0.011 0.039 0.115 -0.036

Lag_FVO_Size 7 -0.070 -0.067 -0.014 0.376 0.376 0.651 -0.018 0.004 0.001 -0.005 0.023 0.022

Lag_LN_TA 8 -0.076 -0.049 -0.054 0.019 0.011 0.023 0.015 0.119 0.221 0.104 -0.294 0.026

Lag_Debt 9 -0.023 -0.062 -0.050 0.025 0.033 0.033 0.040 0.310 -0.169 0.068 0.195 -0.201

Lag_ROA 10 -0.151 -0.220 -0.159 -0.043 -0.122 -0.083 -0.086 0.107 -0.147 0.015 -0.233 0.150

Lag_Special 11 -0.028 -0.054 -0.044 0.014 0.054 0.039 0.042 0.096 0.073 -0.104 0.106 0.017

Lag_Loss 12 0.054 0.076 0.072 0.071 0.127 0.115 0.117 -0.318 0.051 -0.763 0.106 -0.031

Lag_BTM 13 -0.102 -0.046 -0.016 0.020 0.224 0.145 0.146 0.210 0.053 -0.264 0.085 0.073


111

Table 2.10

OLS Regression Analyses of financing activity on Adoption of the FVO



Intercept -3.078 0.598 -3.088 0.597

FVO -0.120 0.909 -0.117 0.912

After 1.063 0.428 1.048 0.434

After_FVO -0.604 0.683 -0.470 0.752

Lag_FVO_Size -0.582 0.337

Lag_LN_TA -0.409* 0.056 -0.418* 0.051

Lag_Debt -3.596** 0.039 -3.589** 0.039

Lag_ROA -2.601*** <.0001 -2.597*** <.0001

Lag_Special -0.882 0.276 -0.887 0.273

Lag_Loss 4.155*** <.0001 4.140*** <.0001

Lag_BTM -0.129 0.512 -0.124 0.528



Adj. R-sq 0.1161 0.1161


n 1645 1645

Dependent variable Financing Financing


112

Table 2.11

Logistic Regression Analyses of financing activity on Adoption of the FVO


Estimate Pr > ChiSq Estimate Pr > ChiSq

Intercept -148.900 0.964 8.929 0.989

FVO 0.075 0.318 0.043 0.551

After 0.111 0.254 0.009 0.926

After_FVO -0.030 0.781 0.038 0.711

Lag_FVO_Size 0.009 0.916 0.043 0.608

Lag_LN_TA -0.025 0.428 -0.015 0.617

Lag_Debt 0.542* 0.039 0.652** 0.011

Lag_ROA 1.452*** 0.000 1.468*** 0.000

Lag_Special 0.125 0.279 0.189* 0.090

Lag_Loss -0.020 0.902 -0.073 0.635

Lag_BTM 0.005 0.861 -0.020 0.487



LR ch2 (DF: 45) 207.682 113.800

LR Pr>ch2 <.0001 <.0001

n 1645 1645

Dependent variable Financing_d High_Financing


113

Table 2.12

Gains or Losses?: Number of firms which report gains or losses resulting

from FVO in the first FVO adoption year

Year Financial firms Non-financial firms

Gains Losses Gains Losses

2006 0 0 1 0

2007 26 12 4 3

2008 27 44 54 26

2009 9 7 17 20

2010 7 6 5 3

2011 6 4 7 2

2012 5 0 3 4

2013 2 3 3 2

2014 4 2 3 0

2015 5 3 1 1

114

Table 2.13

Logistic Regression Analyses of financing activity on Adoption of the FVO with only

firms which adopted FVO in 2008

(Financial vs non-financial firms separately)

Variable Non-financial firms only VS Financial firms only


Intercept -8.566 0.980 0.213 0.667

FVO -0.232 0.111 0.003 0.985

After -0.109 0.542 0.176 0.327

After_FVO 0.416** 0.039 -0.303 0.138

Lag_LN_TA 0.184** 0.014 -0.018 0.732

Lag_Debt 0.572 0.171 -0.558 0.172

Lag_ROA 0.680 0.187 9.717** 0.011

Lag_Special 0.517** 0.025 -0.064 0.762

Lag_Loss -0.597** 0.033 1.037*** 0.003

Lag_BTM 0.370** 0.045 -0.050 0.297

Industry dummies Included Not included


LR ch2 (DF: 27, 11) 91.772 19.673

LR Pr>ch2 <.0001 <.0001

n 482 413

Dependent variable High_Financing High_Financing


115

Table 2.14

Logistic Regression Analysis of financing activity on Adoption of the FVO with

only firms which adopted FVO in 2008

(Financial and non-financial firms together)

Variable Estimate Std error Wald Chisq Pr>ChiSq

Intercept -19.848 782.900 0.001 0.980

NFFirm -0.138 0.563 0.060 0.807

After -0.285 0.187 2.312 0.128

After_NFFirm 0.422** 0.206 4.198 0.041

Lag_LN_TA 0.056 0.068 0.673 0.412

Lag_Debt 1.220** 0.523 5.448 0.020

Lag_ROA 1.008 0.661 2.330 0.127

Lag_Special -0.114 0.224 0.261 0.610

Lag_Loss 0.019 0.292 0.004 0.948

Lag_BTM 0.007 0.033 0.047 0.829

Industry dummies Included

Year dummies Included

LR ch2 (DF: 31) 64.483

LR Pr>ch2 <.0004

n 455

Dependent variable High_Financing


116

Table 2.15




Exotic

1.017

1.000

0.829

1.000

1.000

Exotic_d

0.754

1.000

0.431

1.000

1.000

High_Exotic

0.746

1.000

0.436

0.000

1.000

FVO

0.573

1.000

0.495

0.000

1.000

After

0.559

1.000

0.497

0.000

1.000

After_FVO

0.322

0.000

0.468

0.000

1.000

Lag_LN_TA

10.294

10.458

2.634

8.425

12.172

Lag_Debt

0.272

0.207

0.248

0.071

0.370

Lag_ROA

0.002

0.010

0.136

0.002

0.026

Lag_Special

0.722

1.000

0.449

0.000

1.000

Lag_Loss

0.207

0.000

0.406

0.000

0.000

Lag_ICW

0.049

0.000

0.216

0.000

0.000

Lag_BTM

1.055

0.665

4.516

0.429

1.004

117

TABLE 15 (continued)


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

Exotic 1

0.703 0.696 0.018 0.054 0.017 -0.332 0.059 -0.159 -0.027 0.200 0.091 0.001

Exotic_d 2 0.837

0.981 -0.070 0.085 -0.006 -0.248 -0.052 -0.044 -0.001 0.111 0.024 0.042

High_Exotic 3 0.827 0.981

-0.072 0.103 0.006 -0.230 -0.044 -0.036 0.001 0.105 0.028 0.043

FVO 4 -0.016 -0.070 -0.072

0.007 0.595 0.173 0.117 0.037 -0.040 0.101 0.035 0.059

After 5 0.064 0.085 0.103 0.007

0.613 0.123 0.041 -0.064 0.216 0.176 0.019 0.073

After_FVO 6 0.002 -0.006 0.006 0.595 0.613

0.227 0.115 0.032 0.043 0.201 0.014 0.103

Lag_LN_TA 7 -0.341 -0.261 -0.243 0.180 0.114 0.220

0.034 0.199 0.136 -0.257 0.022 0.057

Lag_Debt 8 0.002 -0.107 -0.098 0.182 0.084 0.192 0.081

-0.305 0.081 0.179 0.072 -0.046

Lag_ROA 9 -0.077 -0.018 -0.013 -0.151 -0.176 -0.215 -0.127 -0.173

-0.110 -0.450 -0.084 0.279

Lag_Special 10 -0.016 -0.001 0.001 -0.040 0.216 0.043 0.136 0.116 -0.174

0.156 0.090 0.052

Lag_Loss 11 0.183 0.111 0.105 0.101 0.176 0.201 -0.234 0.118 -0.700 0.156

0.080 0.026

Lag_ICW 12 0.076 0.024 0.028 0.035 0.019 0.014 0.020 0.030 -0.054 0.090 0.080

0.209

Lag_BTM 13 -0.025 0.048 0.054 0.069 0.214 0.201 0.230 -0.103 -0.340 0.090 0.177 -0.037


118

Table 2.16

OLS Regression Analysis of exotic features of bonds on Adoption of the FVO

(Model 4-2a)

Variable Estimate Std Error t-value p -value

Intercept*** 1.934 0.330 5.860 <.0001

FVO 0.107 0.106 1.000 0.318

After 0.002 0.163 0.010 0.989

After_FVO 0.007 0.142 0.050 0.962

Lag_LN_TA -0.026 0.020 -1.260 0.207

Lag_Debt* -0.347 0.182 -1.900 0.058

Lag_ROA** -0.823 0.327 -2.510 0.012

Lag_Special 0.042 0.085 0.500 0.618

Lag_Loss 0.018 0.110 0.170 0.867

Lag_ICW 0.114 0.168 0.680 0.498

Lag_BTM -0.014 0.009 -1.510 0.131

Industry dummies Included

Year dummies Included

Adj. R-sq 0.3474

F-statistic 6.73

n 410

Dependent variable Exotic


119

Table 2.17

Logistic Regression Analyses of exotic features of bonds on Adoption of the

FVO

Variable Model 4-2b Model 4-2c


Intercept -12.417 0.950 -1.408 0.306

FVO 0.141 0.517 0.161 0.446

After -0.002 0.996 -0.042 0.903

After_FVO -0.144 0.639 -0.131 0.663

Lag_LN_TA 0.098 0.239 0.095 0.248

Lag_Debt 1.581 0.052 1.287 0.107

Lag_ROA 4.872 0.489 0.971 0.803

Lag_Special -0.182 0.602 -0.138 0.687

Lag_Loss 0.859 0.148 0.615 0.271

Lag_ICW 0.526 0.476 0.480 0.513

Lag_BTM -0.973 0.031 -0.815 0.052



LR ch2 (DF: 38) 151.044 139.299

LR Pr>ch2 <.0001 <.0001

n 410 410

Dependent variable Exotic_d High_Exotic


120

Appendix A for Chapter 2

Examples of FVO adoptions

SFAS 159 provides a fair value option that allows firms to irrevocably elect fair

value for the initial and subsequent measurement of certain financial assets and liabilities

on a contract-by-contract basis. If the fair value option is elected for an instrument, SFAS

159 specifies that all subsequent changes in fair value for that instrument shall be reported

in the income statement. Also, SFAS 159 requires that the difference between the carrying

value before election of the fair value option and the fair value of these instruments be

recorded as a cumulative-effect adjustment to the opening balance of retained earnings in

the period of adoption. In this Appendix, I present examples of FVO adoptions by several

firms.

Community Central Bank Corp. (Financial firm)

Community central bank adopted FVO in 2007 for various financial instruments

such as investment securities (including agency debentures and short callable bank qualifi

ed tax exempt municipal bonds), interest rate swap hedging securities, Federal home loan

bank advances, subordinated debentures, interest rate swap hedging subordinated debentu

res, and redeemed subordinated debentures. It made an adjustment of $420,000 to the

beginning balance of retained earnings and recognized a gain of $1,392,000 in earnings.

The following statements and notes are excerpts from the 2007 annual report of

Community central bank.

121

“While not required to adopt the new standard until 2008, the Corporation elected

to adopt it in the first quarter of 2007. …. As a result of the Corporation's adoption

s, certain financial instruments were valued at fair value using the fair value optio

n. The cumulative reduction to opening retained earnings from adopting these stan

dards was approximately $420,000. …. The following table shows the balance

sheet effect of the adoption of SFAS 159.” (p. 39)

122

“The table below contains the fair value measurement at December 31, 2007 using

the identified valuations. Additionally, the changes in fair value for the twelve

month period ended December 31, 2007 for items measured at fair value pursuant

to election of the fair value option.” (p. 41)

NorthStar Realty (financial firm)

NorthStar adopted FVO in 2008 for third party available for sale securities,

exchangeable senior notes, bonds payable and liabilities to subsidiary trust issuing

preferred securities. It made an adjustment of $190.766 million to the retained earnings and

recognized a gain of $782.3 million in earnings. The following statements and notes are

excerpts from the 2008 annual report of NorthStar realty.

123

“We adopted SFAS No. 159 effective January 1, 2008 and have elected to fair value

for our third party available for sale securities, our exchangeable senior notes, our

N-Star bonds payable and our liabilities to subsidiary trust issuing preferred

securities.” (p. 33)

124

“Our available for sale securities that serve as collateral for our term debt

transactions, which we have elected the fair value option pursuant to SFAS 159, are

carried at fair value with the net unrealized gains or losses reported as a component

of earnings in the statement of operations.” (p. 34)

“With respect to derivative instruments that have not been designated as hedges, or

are hedges on debt that is remeasured at fair value pursuant to SFAS 159, any net

payments under, or fluctuations in the fair value of, such derivatives are recognized

currently in income. …. In January 2008, we adopted SFAS 159 and elected the fair

value option for our bonds payable and its liability to subsidiary trusts issuing

preferred securities. …. As a result of this election, the interest rate swap

agreements associated with these debt instruments no longer qualify for hedge

accounting, in accordance with SFAS 133 ‘‘Derivatives and Hedging Activity’’,

since the underlying debt is remeasured with changes in the fair value recorded in

earnings. The unrealized gains or losses accumulated in other comprehensive

income, related to these interest rate swaps, will be reclassified into earnings ….”(p.

36)

“Unrealized Gain (Loss) on Investments and Other: Unrealized gain (loss) on

investments and other increased by approximately $756.6 million for the year

ended December 31, 2008 to a gain of $752.3 million, compared to a loss of $4.3

million for the year ended December 31, 2007. The unrealized gain on investments

for the year ended December 31, 2008 consisted primarily of unrealized gains

related to SFAS 159 mark-to-market adjustments of $958.2 million on various N-

Star bonds payable, $103.2 million on exchangeable senior notes and $95.2 million

on liability to subsidiary trusts issuing preferred securities offset partially by

unrealized losses related to SFAS 159 mark-to-market adjustments of $338.0

million on various N-Star available for sale securities, $30.0 million on our

corporate lending joint venture (which is an equity investment that is marked to

market) and $36.3 million on interest rate swaps as a result of these swaps no longer

qualifying for hedge accounting under SFAS 133.” (p. 42)

“The following table presents a summary of existing eligible financial assets and f

inancial liabilities for which the fair value option was elected on January 1, 2008 a

nd the cumulative-effect adjustment to retained earnings recorded in connection w

ith the initial adoption of SFAS 159.” (p. F-25)

125

ADM Tronics (non-financial firm)

ADM adopted FVO in 2008 for investments in associated companies which are

subject to the equity method of accounting. As a result of the FVO adoption, it recognized

a loss of $10,660,000 from changes in the fair value of the investments in Ivivi in 2008 and

made a pre-tax cumulative-effect adjustment of $ 9,220,483 to the beginning balance of

retained earnings. The following statements and notes are excerpts from the 2008 annual

report of ADM Tronics.

“As a result, our investment in Ivivi was reported during the period from October

18, 2006 until March 31, 2008 under the equity method of accounting, whereby we

126

recognized our share of Ivivi’s earnings or losses as they are incurred. Effective

April 1, 2008 (“the Adoption Date”), we have adopted SFAS No. 159 “The Fair

Value Option for Financial Assets and Liabilities” with respect to our investment in

Ivivi, whereby we report our investment in Ivivi at fair value. …. The fair value of

our investment in Ivivi at the adoption date was approximately $11,375,000. The a

doption of SFAS No. 159, with respect to our investment in Ivivi, resulted in the re

cognition of the following:

Pre-tax cumulative-effect adjustment to retained earnings: $ 9,220,483

Deferred tax liability: 2,425,188

Post-tax cumulative-effect adjustment to retained earnings: $ 6,795,295.” (p. 34)

Five star quality care, Inc. (non-financial firm)

Five Star Quality Care adopted FVO in 2008 for a put right for auction rate

securities. No adjustment to retained earnings was made because the put right was obtained

during 2008. Five Star Quality Care recognized a gain of $11,081,000 as a result of the

FVO adoption. The following statement and notes are excerpts from the 2008 annual report

of Five star quality care.

“The put right is recorded at estimated fair value under the Financial Accounting

Standards Board’s, or FASB, No. 159, ‘‘The Fair Value Option of Financial Assets

and Financial Liabilities’’, or SFAS 159. We calculated this value based on the

127

difference between the fair value for the ARS and the par value of the ARS, which

is the amount we expect to receive from UBS beginning June 2010.” (p. 54)

“During the fourth quarter of 2008, we elected the fair value option available under

SFAS 159 solely for valuing the put right received from UBS related to our

investments in ARS.” (p. F-12)

“We have accounted for our put right as a freestanding financial instrument and

elected to record it at its estimated fair market value under the fair value option of

SFAS 159. As a result, we recorded an $11,081 gain related to receipt of this put

right and recorded its value in other long term assets.” (p. F-22)

Discussions

The actual examples of FVO adoptions presented above indicate that firms adopt

FVO for various financial instruments. Some might argue that FVO may not significantly

affect the asset side of balance sheet especially for non-financial firms since investments

in equity securities had been already measured at fair value even before FVO. However,

because FVO can be applied to almost all financial instruments and there are no qualifying

criteria imposed, it is possible that adopting FVO for certain financial instruments, such as

available for sale securities, held to maturity securities, or various liabilities, makes a

significant impact on earnings or retained earnings. The examples presented above confirm

the possibility. Therefore, by adjusting the timing of adopting FVO, firms might try to

affect the current or future earnings.

Documents

ESSAYS ON ACCOUNTING INFORMATION SYSTEMS AND …