Accounting Quality: International Accounting Standards and US GAAP

Mary E. Barth* Stanford University

Wayne R. Landsman, Mark Lang, and Christopher Williams

University of North Carolina

November 2007 * Corresponding author: Graduate School of Business, Stanford University, 94305-5015, mbarth@stanford.edu. We appreciate funding from the Center for Finance and Accounting Research, Kenan-Flagler Business School and the Center for Global Business and the Economy, Stanford Graduate School of Business. We appreciate comments from Julie Erhardt, and workshop participants at George Washington University, Oklahoma State University, Shanghai University of Finance and Economics, Singapore Management University, Southern Methodist University, and the 2007 European Accounting Association Congress. We also thank Dan Amiram and Mark Maffett for assistance with data collection.

Accounting Quality: International Accounting Standards and US GAAP

Abstract

We compare accounting quality metrics for IAS firms to those for US firms to investigate whether US GAAP-based accounting amounts are associated with less earnings management, more timely loss recognition, and higher value relevance than IAS-based accounting amounts. We find that US firms exhibit higher accounting quality than IAS firms. Comparisons of accounting amounts for US and IAS firms before and after the IAS firms adopt IAS indicate that application of IAS has no systematic effect on the relative difference in accounting quality, although differences in value relevance decrease after IAS firms adopt IAS. US firms have higher accounting quality than IAS firms before and after 2005. The difference in accounting quality between US and IAS firms is largely attributable to firms in countries of German/French legal origin. We also find that the quality of IAS accounting amounts does not differ from that of US GAAP accounting amounts reconciled from domestic standards and reported by non-US firms that cross list on US markets. Our results suggest that although IAS accounting amounts are of lower quality than those of US GAAP reported by US firms, they are of comparable quality to reconciled US GAAP amounts reported by cross-listed firms.

2

1. Introduction

The question we address is whether application of US Generally Accepted Accounting

Principles (GAAP) results in accounting amounts that are of higher quality than those resulting

from application of International Accounting Standards (IAS).1 In particular, we investigate

whether accounting amounts of firms that apply US GAAP exhibit less earnings management,

more timely loss recognition, and higher value relevance than accounting amounts of firms that

apply IAS. The accounting amounts that we compare result from the interaction of features of

the financial reporting system, which include accounting standards, their interpretation,

enforcement, and litigation. We refer to the combined effect of the features of the financial

reporting system as the effect of application of US GAAP or IAS. As predicted, we find that

accounting amounts of US firms applying US GAAP generally are of higher quality than those

of non-US firms applying IAS. This finding applies before and after 2005, when application of

IAS became mandatory in many countries. Tests comparing the quality of accounting amounts

between US and IAS firms in four legal origin groups of countries indicate that our findings are

largely attributable to differences in quality between US and German/French legal origin firms.

In addition, we find no clear pattern of differences in quality of accounting amounts of non-US

firms cross-listing in the US and applying US GAAP in their Form 20-F reconciliations and

those of non-US firms applying IAS.

This study makes comparisons of accounting quality that are potentially relevant to

current policy debates relating to the possibility of permitting firms listed on US exchanges to

prepare financial statements based on IAS and that have not been made in prior literature. The 1 The International Accounting Standards Board (IASB) issues International Financial Reporting Standards (IFRS), which include standards issued not only by the IASB but also by its predecessor body, the International Accounting Standards Committee, some of which have been amended by the IASB. Because most of our sample period pre-dates the effective dates of standards issued by the IASB, following the convention of Barth, Landsman, and Lang (2008), throughout we refer to IAS rather than IFRS.

3

US Securities and Exchange Commission (SEC) presently is considering permitting US firms

and non-US firms cross-listing in the US to file financial statements based on IAS. A major

reason for this is the possibility that accounting amounts based on IAS may be similar in quality

to those based on US GAAP. Two contributing factors underlying this possibility are the

increasing use of IAS throughout the world and the restructuring of the International Accounting

Standards Committee (IASC) into the International Accounting Standards Board (IASB). These

two developments could lead to high quality IAS-based accounting amounts because they likely

will lead to higher quality standards and more consistent interpretation, application, and

enforcement of the standards.

Prior research finds that accounting amounts of non-US firms applying IAS and US firms

applying US GAAP generally are of higher quality than those of non-US firms applying

domestic standards. Prior research also finds that accounting amounts of US firms applying US

GAAP generally are of higher quality than the US GAAP-based amounts of non-US firms

presenting domestic to US GAAP reconciliations on Form 20-F. However, the relative quality of

US and IAS accounting amounts is an open question, as is the relative quality of IAS accounting

amounts and US GAAP accounting amounts of firms cross-listing in the US.

To operationalize accounting quality we follow prior research. In particular, we interpret

accounting amounts that exhibit less earnings management, more timely loss recognition, and

higher value relevance as being of higher quality. Our metrics for earnings management are

based on the variance of change in net income, the ratio of the variance of change in net income

to the variance of change in cash flows, the correlation between accruals and cash flows, and the

frequency of small positive net income. Our metrics for timely loss recognition are based on the

frequency of large losses and the association between bad-news returns and earnings, and our

4

value relevance metrics are the explanatory powers of income and equity book value for prices,

and stock return for earnings.

We address our research question by making two comparisons of accounting quality.

Our primary comparison is of the quality of accounting amounts for a sample of non-US firms

that apply IAS (IAS firms) to that for a matched sample of US firms applying US GAAP (US

firms). Our second comparison is of the quality of accounting amounts of IAS firms to that for a

sample of non-US firms that reconcile earnings and equity book value determined under

domestic standards, i.e., neither IAS nor US GAAP, to those determined under US GAAP on

Form 20-F (20-F firms). The IAS firms are from 24 countries and adopted IAS between 1995

and 2006.

Our empirical metrics of accounting quality also reflect the effects of the economic

environment that are unattributable to the financial reporting system. The economic

environment includes volatility of economic activity and information environment. We utilize

two research design features to mitigate these effects. First, for our primary comparison, for

each sample firm applying IAS we select a US firm that is of similar size and in the same

industry as the firm applying IAS. For our second comparison, because the number of 20-F

firms is substantially smaller than the number of IAS firms, we are unable to use a similar

matching procedure. However, we require 20-F firms to be from the same countries as IAS

firms. Second, when constructing several of our metrics, we include controls for economic

environment.

We find that US firms generally have higher quality accounting amounts than IAS firms.

In particular, US firms have a significantly higher variance of change in net income, a

significantly higher ratio of the variances of change in net income and change in cash flows, a

5

significantly less negative correlation between accruals and cash flows, and a significantly higher

frequency of large negative net income. In addition, US firms have significantly higher value

relevance of earnings and equity book value for share prices. However, US and IAS firms

exhibit no significant differences in value relevance of earnings for stock returns or frequency of

small positive net income.

To determine whether this difference in accounting quality diminishes with the

application of IAS, we repeat the accounting quality comparison in the period before IAS firms

adopted IAS, i.e., when they applied domestic standards. Findings suggest that application of

IAS reduces the difference in accounting quality between the IAS and US firms for some

metrics, but increases the difference for others. However, application of IAS significantly

reduces the difference in accounting quality as indicated by each of the value relevance metrics.

Findings from our first comparison indicate that accounting amounts of US firms are of

higher quality than accounting amounts of IAS firms, and prior research finds that reconciled US

GAAP accounting amounts of cross-listed firms are of lower quality than accounting amounts of

US firms. Thus, the natural question to address is that addressed by our second comparison.

That is, are reconciled US GAAP-based accounting amounts also of higher quality than IAS-

based accounting amounts? We find no clear pattern of differences in quality between

accounting amounts based on reconciled US GAAP and those based on IAS. In particular, of our

eight accounting quality metrics, four are consistent with higher quality for 20-F firms and four

are consistent with higher quality for IAS firms. Three of the differences are significant, with

two consistent with higher quality for 20-F firms and one with higher quality for IAS firms.

Relating to the comparison of accounting quality metrics for US and IAS firms, we

conduct two additional analyses. First, because in 2005 IAS adoption became mandatory in

6

many countries, we repeat our accounting quality comparison separately between US firms and

IAS firms before 2005, and between US firms and IAS firms in 2005 or after. Findings for the

two comparisons generally are consistent with those based on the comparison using the

combined IAS sample. That is, US firms generally evidence higher accounting quality than IAS

firms in both periods. Second, because prior literature suggests that application of IAS can differ

based on a country’s legal origin, we repeat our accounting quality comparison separately

between US firms and IAS firms grouped into four major legal origin groups – English,

Germanic/French, Scandinavian, and Chinese. This analysis reveals that our findings for the full

IAS sample appear to be largely attributable to the German/French legal origin firms, which

comprise approximately half of our IAS sample. We find no systematic differences in

accounting quality between US and IAS firms in the other three groups. In addition, with the

exception of German/French legal origin firms, the findings reveal no clear pattern of differences

in accounting quality between IAS firms in the other three groups.

The remainder of the paper is organized as follows. The next section discusses

institutional background and related literature. Sections three and four develop the hypotheses

and explain the research design. Sections five and six describe the sample and data and present

the results. Section seven offers a summary and concluding remarks.

2. Institutional Background and Related Literature

2.1 INSTITUTIONAL BACKGROUND

The SEC currently requires that US firms file financial statements based on US GAAP,

and that non-US firms that list on US exchanges and file financial statements based on domestic

standards provide reconciliations to US GAAP-based earnings and equity book value. The

premise underlying these requirements is the belief that US investors are best served if firms

7

issue financial statements using a uniform set of accounting standards and that financial

statements based on non-US standards are of lower quality than those based on US GAAP.2 The

SEC permits non-US firms to file financial statements based on domestic standards with

reconciliation to US GAAP only for earnings and equity book value because of its desire not to

discourage non-US firms from listing their securities in the US.

The use of IAS is becoming widespread, with more than 100 countries requiring or

permitting financial statements to be based on IAS (www.iasplus.com). Many suggest that the

SEC should accept financial statements based on IAS for non-US firms because many US

investors are becoming familiar with IAS. However, the SEC remains concerned that the quality

of IAS-based accounting amounts is not yet high enough to achieve investor protection in the

US. In 2005, the SEC staff established a “roadmap” that would lead to a recommendation that

the SEC eliminate the reconciliation requirement for non-US firms applying IAS (Nicolaisen,

2005). The roadmap states that the SEC will evaluate the quality of the standards issued by the

IASB, as well as the quality of accounting amounts in IAS-based financial statements with a

focus on, among other things, the application and interpretation of IAS in financial statements

across countries and firms.3 The roadmap makes clear that the SEC’s decision to accept IAS-

based financial statements depends on its assessment of the comparability of financial statement

information based on IAS and US GAAP in an institutional setting where the two sets of

standards coexist. That is, if IAS-based information is comparable to US GAAP-based

information, the reconciliation requirement is unnecessary.

2 This concern stems from the belief that US investors can make better investment decisions regarding non-US firms if the investors have access to information about these firms that is “similar” and of “similar quality” to that available for US firms (Jenkins, 1999). 3 “Chairman Donaldson Meets with EU Internal Market Commissioner McCreevy,” Press Release 2005-62, U.S. Securities and Exchange Commission, Washington, D.C., April 21, 2005.

8

Consistent with the roadmap, in 2007, the SEC issued for public comment a proposal to

eliminate the reconciliation requirement for non-US firms applying IAS (SEC, 2007a). Also in

2007, the SEC issued a Concepts Release (SEC, 2007b) to permit US firms to file financial

statements based on IAS. Underlying this Concepts Release is the belief that IAS are of high

enough quality to achieve investor protection and are comparable to US GAAP.

2.2 RELATED LITERATURE

Several studies compare accounting amounts based on, and the economic implications of,

non-US firms applying IAS and domestic standards. Using a sample of IAS firms from 21

countries, not including the US, and metrics similar to those employed here, Barth, Landsman,

and Lang (2008) finds that accounting quality of firms applying IAS generally is higher than that

of firms applying domestic standards. Studies relating to German firms include Bartov,

Goldberg, and Kim (2005), Van Tendeloo and Vanstraelen (2005), Daske (2006), and Hung and

Subramanyam (2007). With the exception of Bartov, Goldberg, and Kim (2005), these studies

generally fail to find differences in accounting quality or economic implications, e.g., cost of

capital. Eccher and Healy (2003) finds differences in value relevance of accounting amounts

based on IAS and Chinese standards depending on whether firms’ shares can be owned by

Chinese or foreign investors. Daske et al. (2007b) analyzes the economic consequence of

mandatory application of IAS in 26 countries and generally capital market benefits. However,

the capital market benefits exist in countries with strict enforcement regimes and institutional

environments that provide strong reporting incentives.

Several studies compare accounting amounts based on, and the economic implications of,

US firms applying US GAAP and non-US firms applying domestic standards. Representative

studies include Alford, Jones, Leftwich, and Zmijewski (1993), Land and Lang (2002), and

9

Leuz, Nanda, and Wysocki (2003). Taken together, the findings from these and other studies

provide evidence of higher quality accounting amounts for US firms relative to non-US firms

applying domestic standards. Lang, Raedy, and Wilson (2006) obtains similar inferences for a

comparison of US GAAP-based accounting amounts of US firms and of non-US firms that cross-

list in the US and apply domestic standards but reconcile earnings and equity book value to US

GAAP on Form 20-F. That is, Lang, Raedy, and Wilson (2006) finds that US firms have higher

accounting quality than non-US firms applying US GAAP in their Form 20-F reconciliations.

In contrast to these other comparisons, there are relatively few studies that compare

accounting amounts based on, and the economic implications of, US firms applying US GAAP

and non-US firms applying IAS. Leuz (2003) compares measures of information asymmetry for

German firms trading on Germany’s New Market and finds little evidence of differences in

bid/ask spreads and trading volume for firms that apply US GAAP relative to those that apply

IAS. The fact that these firms do not differ in information asymmetry is indirect evidence

regarding whether the quality of their accounting amounts differs because bid/ask spreads and

trading volume reflect a variety of factors in addition to the quality of accounting amounts.

Bartov, Goldberg, and Kim (2005) documents that for German firms earnings response

coefficients are highest for those applying US GAAP, followed by those applying IAS, and

followed by those applying German standards. Although the study finds differences in earnings

response coefficients for the three samples of firms applying different accounting standards, such

evidence also is indirect regarding whether these differences are attributable to differences in the

quality of accounting amounts because earnings response coefficients are affected by factors,

e.g., risk, that are not necessarily dimensions of accounting quality. Also, because the samples in

both studies comprise only German firms, inferences from them are not generalizable to firms in

10

other countries. These studies’ lack of generalizability and use of indirect measures of

accounting quality leave open the question of whether IAS- and US GAAP-based accounting

amounts are of comparable quality.

Harris and Muller (1999) addresses a research question closely related to our second

research question. In particular, Harris and Muller (1999) provides evidence that US GAAP

reconciled amounts for 31 firms applying IAS are value relevant incremental to IAS-based

accounting amounts. An advantage of the Harris and Muller (1999) research design is that, by

construction because each firm is its own control, there are no economic differences between the

firms applying IAS and those reconciling to US GAAP. Another potential advantage is that to

the extent reconciled US GAAP amounts differ depending on whether firms apply IAS or

domestic standards in their financial statements, the Harris and Muller (1999) comparison is

most closely related to the SEC’s decision on removing the reconciliation requirement for IAS

firms. However, Harris and Muller (1999) makes this comparison only for firms that cross-list

on US exchanges, thereby limiting the generalizability of the study’s findings. Because cross-

listed firms are required to reconcile to US GAAP, the sample firms may make US GAAP-

consistent choices under IAS to minimize reconciling items (Lang, Raedy, and Wilson, 2006).

Generalizing the study’s inferences is also difficult because the findings differ across empirical

specifications, the sample size is small, and the sample period, 1992-1996, pre-dates substantial

changes in IAS after 1996. The study’s lack of generalizability and use of cross-listed firms

leave open the question of whether IAS- and reconciled US GAAP-based accounting amounts

are of comparable quality.

3. Hypothesis Development

3.1 ACCOUNTING QUALITY OF IAS AND US GAAP

11

The quality of accounting amounts reflects the interaction of features of the financial

reporting system, which include accounting standards, their interpretation, enforcement, and

litigation. Relating to standards, the IASB has adopted an approach in developing standards

different from the Financial Accounting Standards Board (FASB) that could result in different

quality of accounting amounts. In particular, the IASB’s approach relies more on principles,

whereas the FASB’s approach relies more on rules.4 Reliance on principles specifies guidelines,

but requires judgment in application. Reliance on rules specifies more requirements that leave

less room for discretion. Ewert and Wagenhofer (2005) develops a rational expectations model

that shows that accounting standards that limit opportunistic discretion result in accounting

earnings that are more reflective of a firm’s underlying economics and, therefore, are of higher

quality. The inherent flexibility IAS principles-based standards afford can allow firms to

manage earnings, thereby decreasing accounting quality. This flexibility has long been a

concern of securities markets regulators, especially in international contexts (e.g., Breeden,

1994).

Relating to other features of the financial reporting system, Ball (1995, 2006), Lang,

Raedy, and Wilson (2006), Bradshaw and Miller (2007), and Daske et al. (2007a) observe that

both accounting standards and the regulatory and litigation environment are important to the

application of accounting standards. In particular, Cairns (1999), Street and Gray (2001), and

Ball, Robin, and Wu (2003) suggest that lax enforcement can result in limited compliance with

IAS, thereby limiting their effectiveness. Thus, because of the inherent flexibility of principles-

based standards and potential weakness in other features of the financial reporting system, we

4 The distinction here is more relative than absolute. IAS and US GAAP include both general principles and rules, depending on context (Schipper, 2003). However, the FASB has generally provided more detailed guidance on application of accounting principles than has the IASB.

12

predict that accounting amounts resulting from application of US GAAP are of higher quality

than those resulting from application of IAS.

Although we predict that application of US GAAP is associated with higher accounting

quality than IAS, this may not be true. Discretion in accounting can be opportunistic and

possibly misleading about the firm’s economic performance, but it can be used to reveal private

information about the firm (Watts and Zimmerman, 1986). The IASB attempts to limit

allowable alternative accounting practices and provides a consistent approach to accounting

measurement for the purpose of having a firm’s recognized amounts faithfully represent its

underlying economics. Thus, it is possible that IAS-based accounting amounts are of

comparable quality to US GAAP-based amounts. If this is the case and enforcement in countries

in which IAS are applied is comparable to that in the US, then it possible that accounting

amounts resulting from the application of the two sets of standards are of equal quality.5

For the same reasons that we predict US GAAP-based accounting amounts are of higher

quality than IAS-based amounts, we also predict that US GAAP-based accounting amounts

reconciled from those based on application of domestic standards on Form 20-F are of higher

quality than those based on IAS. However, because 20-F firms do not necessarily face the same

legal and regulatory environment as US firms applying US GAAP (Bradshaw and Miller, 2007),

we do not expect the difference in quality of accounting amounts of 20-F and IAS firms to be as

pronounced as the difference in quality of accounting amounts between US and IAS firms.

3.2 MEASURES OF ACCOUNTING QUALITY

Following prior research, we operationalize accounting quality using earnings

management, timely loss recognition, and value relevance metrics. We predict that higher

5 Because other features of the financial reporting system, e.g., enforcement and litigation environment, are likely to differ across countries in which IAS are applied, we report findings from our comparisons separately for groups in which enforcement is likely to be homogenous within the groups and heterogeneous across the groups.

13

quality accounting amounts exhibit less earnings management, more timely loss recognition, and

higher value relevance of earnings and equity book value. Therefore, we predict that US GAAP-

based accounting amounts of US firms applying US GAAP, i.e., US firms, or non-US firms

applying domestic standards but reconciling earnings and equity book value to US GAAP, i.e.,

20-F firms, exhibit less earnings management, more timely loss recognition, and higher value

relevance of earnings and equity book value than do IAS-based accounting amounts of firms

applying IAS, i.e., IAS firms.

We examine two manifestations of earnings management – earnings smoothing and

managing towards positive earnings. Regarding earnings smoothing, following prior research,

we expect firms that smooth earnings less exhibit more earnings variability, after controlling for

other economic determinants of earnings volatility (Lang, Raedy, and Yetman, 2003; Leuz,

Nanda, and Wysocki, 2003; Lang, Raedy, and Wilson, 2006; Barth, Landsman, and Lang, 2008).

We predict that US and 20-F firms exhibit more variable earnings than do IAS firms.6 To test

our prediction, we use two earnings variability metrics, variability of change in net income and

variability of change in net income relative to variability of change in cash flow.

Although we predict that firms applying US GAAP have less earnings management and,

thus, higher earnings variability, some studies (e.g., Healy, 1985) suggest that, in the case of “big

baths,” managers may use discretion in ways that result in higher earnings variability. Thus,

firms applying IAS could have more discretion for this form of earnings management and thus

could exhibit higher earnings variability. Also, higher earnings variability could be indicative of

lower earnings quality because of error in estimating accruals. Thus, higher quality accounting

can result in lower earnings variability.

6 Our prediction is supported by Ewert and Wagenhofer (2005), which shows that applying accounting standards that limit management’s discretion should result in higher variability in accounting earnings. See Barth, Landsman, and Lang (2008) for further discussion.

14

We also expect firms with less earnings smoothing to exhibit a more negative correlation

between accruals and cash flows (Lang, Raedy, and Yetman, 2003; Leuz, Nanda, and Wysocki,

2003; Ball and Shivakumar, 2005, 2006; Lang, Raedy, and Wilson, 2006; Barth, Landsman, and

Lang, 2008). Because accruals reverse over time, accruals and cash flows tend to be negatively

correlated even in the absence of earnings management (Dechow, 1994). However, Land and

Lang (2002) and Myers and Skinner (2007), among others, posit that a more negative correlation

indicates earnings smoothing because managers respond to weak (strong) cash flow outcomes by

increasing (decreasing) accruals. In addition, Ball and Shivakumar (2005, 2006) show that

timely gain and loss recognition, which is consistent with higher earnings quality, attenuates the

negative correlation between accruals and current period cash flow. Consistent with Land and

Lang (2002), Myers and Skinner (2007), and prior studies testing for earnings smoothing, we

predict that US and 20-F firms exhibit a less negative correlation between accruals and cash

flows than do IAS firms.

Regarding our second manifestation of earnings management, prior research identifies

positive earnings as a common target of earnings management. Evidence of managing towards

positive earnings is a larger frequency of small positive earnings (Burgstahler and Dichev, 1997;

Leuz, Nanda, and Wysocki, 2003). The notion underlying this target is that management prefers

to report small positive earnings rather than negative earnings. If applying US GAAP reduces

managerial discretion relative to applying IAS, we predict that US firms and 20-F firms report

small positive earnings with lower frequency than do IAS firms.

Regarding timely loss recognition, we expect firms with higher quality earnings to

exhibit a larger frequency of large losses. This is consistent with Ball, Kothari, and Robin

(2000), Lang, Raedy, and Yetman (2003), Leuz, Nanda, and Wysocki (2003), and Lang, Raedy,

15

and Wilson (2006) that suggest one characteristic of higher quality earnings is that large losses

are recognized as they occur rather than deferred to future periods. This characteristic is closely

related to earnings smoothing in that if earnings are smoothed large losses should be relatively

rare. Thus, we predict that US firms and 20-F firms report large losses with higher frequency

than do IAS firms.

Although we predict higher quality accounting results in a higher frequency of large

losses, the opposite could be true. In particular, a higher frequency of large losses could be

indicative of big bath earnings management. Also, a higher frequency of large losses could

result from error in estimating accruals. Thus, higher quality accounting can result in a lower

frequency of large losses.

Regarding value relevance, we expect firms with higher quality accounting amounts to

have a higher association between stock price and earnings and equity book value because higher

quality accounting amounts better reflect a firm’s underlying economics (Barth, Beaver, and

Landsman, 2001). First, higher quality accounting amounts are the product of applying

accounting standards that require recognition of amounts that are intended to faithfully represent

a firm’s underlying economics. Second, higher quality accounting amounts are less subject to

opportunistic managerial discretion. These two features of higher quality accounting are linked

together by Ewert and Wagenhofer (2005), which shows that accounting standards that limit

opportunistic discretion result in accounting earnings that have higher value relevance. Third,

higher quality accounting has less non-opportunistic error in estimating accruals. Consistent

with these three features of higher quality accounting, prior research also suggests that higher

quality accounting amounts are more value relevant (Lang, Raedy, and Yetman, 2003; Lang,

Raedy, and Wilson, 2006; Barth, Landsman, and Lang, 2008). Accordingly, we predict that US

16

firms and 20-F firms exhibit higher value relevance of earnings and equity book value than do

IAS firms.7

We examine whether the quality of accounting amounts of US firms and 20-F firms is

higher than that of IAS firms by conducting tests relating to earnings management, timely loss

recognition, and value relevance. Following Barth, Landsman, and Lang (2008), we infer higher

quality from a consistent pattern of evidence provided by the portfolio of tests.8

4. Research Design

4.1 OVERVIEW

To test our predictions, we first compare accounting quality metrics for non-US firms that

apply IAS (IAS firms) to those of a matched sample of US firms that apply US GAAP (US

firms). To compare IAS and US firms, following Barth, Landsman, and Lang (2008), we

identify each IAS firm’s industry and IAS adoption year. We then select as the matched US firm

a US firm in the same industry as the IAS firm whose size as measured by equity market value is

closest to the IAS firm’s at the end of the year of its adoption of IAS. Our analyses include all

firm-years for which the IAS firm and its matched US firm both have data. For example, if the

IAS firm has data from 1994 through 2000, and its matched US firm has data for 1995 through

2002, then our analysis includes data from 1995 through 2000 for the IAS firm and its matched

US firm. We employ this matching procedure to mitigate the effects on accounting quality of

economic differences unattributable to the financial reporting system, including country-level

differences in economic activity and size-related differences, such as information environment.

7 Examining value relevance in this context is subject to at least two caveats. First, it presumes the pricing process is similar across firms and across countries. Second, earnings smoothing can increase the association between earnings and share prices. See Wysocki (2005) for a discussion of various approaches to assessing accounting quality. 8 See Barth, Landsman, and Lang (2008) for a fuller discussion of advantages of using several metrics.

17

We next compare accounting quality metrics for IAS firms to those relating to the US

GAAP amounts of non-US firms applying domestic standards and reconciling accounting

amounts to US GAAP on Form 20-F, 20-F firms. When making this comparison we limit 20-F

firms to those that do not apply IAS. We require 20-F firms to be from the same countries as the

IAS firms and observations to be from 1989 to 2006. Ideally, we would follow a matching

procedure for IAS and 20-F firms, i.e., matching on industry, size, and country. However, data

limitations preclude us from doing so because the resulting matched sample would be too small

to conduct meaningful tests.

Following Barth, Landsman, and Lang (2008), for both comparisons we include controls

for firm characteristics that could affect our metrics but are unattributable to the financial

reporting system because it is possible that our matching procedures fail to control for these

differences.9 However, because matching and including controls could also control for some

effects attributable to the financial reporting system, such as enforcement and litigation, it is

unclear that they are desirable design features. The SEC is concerned with the quality of

reported accounting amounts, which is affected by all features of the financial reporting system,

not just accounting standards. If the SEC decides to accept IAS-based financial statements

without reconciliation to US GAAP from non-US firms, such financial statements would be

subject to the influence of all features of the financial reporting systems to which the firms are

subject. If the SEC decides to accept IAS-based financial statements from US firms, such

financial statements would be subject to the influence of the US financial reporting system.

Although matching and including controls can mitigate differences in the financial reporting

system, there is no obvious means to include the effects of features of the US financial reporting

9 Untabulated findings indicate that inferences are insensitive to inclusion of the controls.

18

system on accounting quality for non-US firms. Thus, our findings cannot provide direct

evidence regarding accounting quality for US firms that would result if they were to apply IAS.

With the exception of the tests for the frequency of small positive earnings and large

negative earnings, we test for differences in each metric using a t-test based on the empirical

distribution of the differences. Specifically, for each test, we first randomly select, with

replacement, firm observations that we assign to one or the other type of firm, depending on the

test. For example when comparing IAS firms and US firms, we assign firm observations as

either IAS or US firms. We then calculate the difference between the two types of firms in the

metric that is the subject of the particular test. We obtain the empirical distribution of this

difference by repeating this procedure 1,000 times. An advantage of this approach for testing

significance of the differences is that it requires no assumptions about the distribution of each

metric. Another advantage is that it can used for all of our metrics, even those with unknown

distributions, e.g., the ratio of variability of change in net income to variability of changes in

cash flow.10

Our comparison of accounting quality for IAS and US firms is relevant to the SEC’s

decisions to permit non-US firms to file financial statements based on IAS without reconciliation

to US GAAP and US firms to file financial statements based on IAS. The comparison is relevant

to the SEC’s reconciliation requirement decision because the comparison provides evidence on

the relative quality of accounting amounts of US firms and those of non-US firms based on IAS.

We limit IAS firms to those that do not cross-list in the US to eliminate effects on the IAS

accounting amounts associated with the reconciliation requirement (Harris and Muller, 1999;

Lang, Raedy, and Wilson, 2006). Studying the quality of accounting amounts of firms that are

10 When applicable, we also test for significance using the Cramer (1987) test. In every case, that test results in the same inferences as the empirical distribution test.

19

potential candidates for cross-listing is relevant because, as Pownall and Schipper (1999) notes,

one of the SEC’s goals in permitting non-US firms to file financial statements based on IAS is to

attract US listings by removing impediments. However, although our IAS firms are large, global

firms, we cannot be certain that they would cross-list in the US if there were no reconciliation

requirement.

The comparison of accounting quality for IAS and US firms also is relevant to the SEC’s

decision to permit US firms to apply IAS because the comparison provides evidence on the

relative quality of IAS-based and US-based accounting amounts. However, because no US firms

apply IAS, our IAS firms are not US firms. Thus, as noted above, we cannot be certain that our

findings would obtain for a sample of US firms. Nonetheless, it is presently the only comparison

of quality of IAS- and US GAAP-based accounting amounts that can be made, and our study is

the first to do so for a broad sample of non-US firms.

Our comparison of accounting quality for IAS and 20-F firms also is relevant to the

SEC’s decision to remove the reconciliation requirement because the comparison provides

evidence on the relative quality of accounting amounts in the reconciliation and in financial

statements based on IAS. However, US GAAP-based accounting amounts of firms that have

made the decision to cross-list in the US could be affected by their incentives to cross-list in the

presence of a reconciliation requirement. As a result, their US GAAP-based amounts may not be

representative of those of firms that would cross-list in the absence of the reconciliation

requirement. Nonetheless, our study is the first to provide a comparison of the quality of IAS-

based and reconciled US GAAP-based accounting amounts of a broad sample of non-US firms.

4.2 ACCOUNTING QUALITY METRICS

4.2.1 Earnings Management

20

Our first earnings management metric is based on the variability of change in net income

divided by total assets, NIΔ (Lang, Raedy, and Wilson, 2006; Barth, Landsman, and Lang,

2008).11 A smaller variance of change in net income is evidence consistent with earnings

smoothing. However, net income is likely to be sensitive to a variety of factors associated with

firms’ economic environments that are unattributable to the financial reporting system.

Although our matching procedures mitigate these effects, some may remain. Therefore, our

variability metric is the variance of the residuals from the regression of change in net income on

variables identified in prior research as controls for these factors (Ashbaugh, 2001; Pagano,

Röell, and Zehner, 2002; Lang, Raedy, and Yetman, 2003; Lang, Raedy, and Wilson, 2006;

Barth, Landsman, and Lang, 2008), ΔNI*:

ΔNIit = α 0 +α1SIZEit +α 2GROWTHit +α 3EISSUEit +α 4LEVit + α 5DISSUEit +α 6TURNit +α 7CFit + ε it

(1)

SIZE is the natural logarithm of end-of-year market value of equity, GROWTH is percentage

change in sales, EISSUE is percentage change in common stock, LEV is end-of-year total

liabilities divided by end-of-year equity book value, DISSUE is percentage change in total

liabilities, TURN is sales divided by end-of-year total assets, and CF is annual net cash flow

from operating activities divided by end-of-year total assets. Equation (1) also includes country

and industry fixed-effects, as do equations (2) through (4). We estimate equation (1) pooling

observations that are relevant to the particular comparison we test. For example, when

comparing IAS and US (IAS and 20-F) firms, we pool all firm-year observations for IAS and US

11 DataStream provides several definitions of income. The one we use is operating income, which does not include extraordinary items and other non-operating income. However, because the criterion for extraordinary items differs across countries and excluding extraordinary items could result in differences based on the location on the income statement of one-time items, we replicate the analysis including extraordinary and non-operating items. Experimental inferences are unaffected.

21

(IAS and 20-F) firms. The variability of ΔNI* is the cross-sectional variance of the IAS or US

(20-F) firms’ residuals from equation (1).

Our second earnings smoothing metric is based on the ratio of variability of change in net

income, NIΔ , to variability of change in operating cash flows, CFΔ . Firms with more volatile

cash flows typically have more volatile net income, and our second metric controls for this. If

firms use accruals to manage earnings, variability of change in net income should be lower than

that of operating cash flows. As with NIΔ , CFΔ is likely to be sensitive to a variety of factors

unattributable to the financial reporting system. Therefore, we estimate an equation similar to

equation (1), but with CFΔ as the dependent variable:12

ΔCFit = α 0 +α1SIZEit +α 2GROWTHit +α 3EISSUEit +α 4LEVit + α 5DISSUEit +α 6TURNit +α 7CFit + ε it

(2)

As with equation (1), we pool observations appropriate for the particular comparison. The

variability of ΔCF* is the cross-sectional variance of groups of residuals from equation (2),

where the composition of the groups depends on the particular comparison we test. Our resulting

second metric is the ratio of variability of ΔNI* to variability of ΔCF*.

Our third metric of earnings smoothing is based on the Spearman correlation between

accruals, ACC, and cash flows. ACC is NI minus CF. As with the two variability metrics based

on equations (1) and (2), because the accruals and cash flows correlation may be sensitive to

factors unattributable to the financial reporting system, we compare correlations of residuals

from equations (3) and (4), CF* and ACC*, rather than correlations between CF and ACC

directly. As with the equations (1) and (2), both CF and ACC are regressed on the control

variables, but excluding CF:

12 For ease of exposition, we use the same notation for coefficients and error terms in each equation. In all likelihood they differ.

22

CFit = α 0 +α1SIZEit +α 2GROWTHit +α 3EISSUEit +α 4LEVit + α 5DISSUEit +α 6TURNit + ε it

(3)

ACCit = α 0 +α1SIZEit +α 2GROWTHit +α 3EISSUEit +α 4LEVit + α 5DISSUEit +α 6TURNit + ε it

(4)

Our metric for managing towards positive earnings is the coefficient on small positive net

income, SPOS, in equation (5).

IAS(0,1)it = α0 +α1SPOSit +α2SIZEit +α3GROWTHit +α4EISSUEit + α5LEVit +α6DISSUEit +α7TURNit +α8CFit + ε it

(5)

)1,0(IAS is an indicator variable that equals one for IAS firms and zero for US firms or 20-F

firms, depending on the comparison we test, and SPOS is an indicator variable that equals one if

net income divided by total assets is between 0 and 0.01 (Lang, Raedy, and Yetman, 2003).

Equation (5) also includes industry fixed effects for the IAS and US firms comparison, and

country and industry fixed effects for the IAS and 20-F firms comparison.13 A positive

coefficient on SPOS indicates that IAS firms manage earnings toward small positive amounts

more frequently than do US, or 20-F, firms. We use the coefficient on SPOS from equation (5)

rather than comparing the percentages of small positive income for each group of firms to assess

whether IAS firms are more likely to manage earnings. This is because equation (5) includes

controls for firm characteristics unattributable to the financial reporting system.

4.2.2 Timely Loss Recognition

We measure timely loss recognition as the coefficient on large negative net income,

LNEG, in equation (6) (Lang, Raedy, and Yetman, 2003; Lang, Raedy, and Wilson, 2006; Barth,

Landsman, and Lang, 2008).

13 It is not possible to include country fixed effects for the IAS and US firms comparison because all US firms are from the same country.

23

IAS(0,1)it = α 0 +α1LNEGit +α 2SIZEit +α 3GROWTHit +α 4EISSUEit + α 5LEVit +α 6DISSUEit +α 7TURNit +α 8CFit + ε it

(6)

LNEG is an indicator variable that equals one for observations for which annual net income

divided by total assets is less than −0.20, and zero otherwise. As does equation (5), equation (6)

also includes industry fixed effects for the IAS and US firms comparison, and country and

industry fixed effects for the IAS and 20-F firms comparison. A negative coefficient on LNEG

indicates that IAS firms recognize large losses less frequently than US, or 20-F, firms. As with

equation (5), we use the coefficient on LNEG rather than directly comparing the percentages of

large losses across the comparison groups of firms to assess whether IAS firms are more likely to

manage earnings because equation (6) includes control variables.14

4.2.3 Value Relevance

Our primary value relevance metric is based on the explanatory power from a regression

of stock price on earnings and equity book value. To obtain a measure of price that is unaffected

by mean differences across countries and industries, which could affect our comparisons of

explanatory power, we first regress stock price, P, on country and industry fixed effects.15 We

regress the residuals from this regression, P*, on equity book value per share, BVE, and net

income per share, NI, separately for IAS and US firms, and for IAS and 20-F firms. Following

prior research, to ensure accounting information is in the public domain, we measure P six

months after fiscal year-end (Lang, Raedy, and Yetman, 2003; Lang, Raedy, and Wilson, 2006;

14 Following Lang, Raedy, and Wilson (2006) and Barth, Landsman, and Lang (2008), in the analyses of small positive and large negative net income, we report results from OLS estimation rather than from logit estimation because Greene (1993) reports that logit models are extremely sensitive to the effects of heteroscedasticity. 15 As noted in Barth, Landsman, and Lang (2008), we could permit BVE and NI coefficients to reflect cross-industry differences in the relation between price and accounting amounts (Barth, Konchitchki, and Landsman, 2007). However, small sample sizes in many of our industries make this impractical.

24

Barth, Landsman, and Lang, 2008). Our primary value relevance metric is the adjusted R2 from

equation (7).16

itititit NIBVEP εβββ +++= 210* (7)

Our other value relevance metric is based on the explanatory power from regressions of

net income on annual stock return. We estimate the earnings-returns relation separately for

positive and negative return subsamples. Because we partition firms based on the sign of the

return, we estimate two “reverse” regressions with earnings as the dependent variable, where one

is for good news firms and the other is for bad news firms. As with price, to obtain a measure of

net income that controls for mean differences across countries and industries, we first regress net

income per share divided by beginning-of-year price, NI/P, on country and industry fixed effects.

We regress the residuals from this regression, NI/P*, on annual stock return, RETURN.

Following Lang, Raedy, and Wilson (2006) and Barth, Landsman, and Lang (2008), we measure

RETURN as the natural logarithm of the ratio of stock price three months after fiscal year end to

stock price nine months before fiscal year end, adjusted for dividends and stock splits. Our good

and bad news value relevance metrics are the R2s from equation (8) estimated separately for

good news and bad news firms.17

ititit RETURNPNI εββ ++= 10*]/[ (8)

As with equation (7), we estimate equation (8) separately for IAS and US firms and for IAS and

20-F firms.

16 We also estimated versions of equation (7) using undeflated amounts. Inferences from the untabulated findings are the same as from the tabulated findings. 17 Because RETURN is the log of a price ratio and therefore can be negative even when stock prices increase during the twelve month window, when determining whether an observation belongs in the good news or bad news regression, we measure return as the stock price three months after fiscal year end to stock price nine months before fiscal year end, adjusted for dividends and stock splits.

25

5. Data and Sample

We obtain our sample of IAS firms from Worldscope, which identifies the set of

accounting standards a firm uses to prepare its financial statements and its industry. The two

Worldscope standards categories that we code as IAS based on the Worldscope Accounting

Standards Applied data field are “international standards” and “IASC” or “IFRS”.18 There are

two sources of potential error in classifying a firm as applying IAS. The first is that firms do not

always indicate clearly the accounting standards that they apply in their financial statements.

The second is that Daske et al. (2007a) reports that the Worldscope data field has classification

error. If a substantial portion of firms we classify as IAS firms are affected by these sources of

classification error, it is likely that our findings are biased in favor of our prediction that US

firms have higher accounting quality that IAS firms. This is because Barth, Landsman, and Lang

(2008) finds that IAS firms have higher accounting quality than non-US firms applying domestic

standards and it is likely that any non-IAS applied by misclassified IAS firms are domestic

standards.19

We gather data for IAS firms from DataStream and data for US firms from Compustat

and CRSP. We obtain data for 20-F firms from Datastream, except net income and equity book

value based on US GAAP, which we obtain from Forms 20-F.20 We winsorize at the 5% level

all variables used to construct our metrics to mitigate the effects of outliers on our inferences.

The resulting sample of IAS firms comprises 11,443 firm year observations for 2,212 firms that

adopted IAS between 1995 and 2006. Of the 11,443 firm years, 2,804 are post-IAS adoption

observations, and 8,639 are pre-adoption observations.

18 Worldscope category 23 was “IASC” prior to 2004 and “IFRS” in 2005 and 2006. 19 Limiting our comparisons to IAS firms classified by Worldscope as applying IASC or IFRS results in inferences similar to those we obtain from our tabulated findings. 20 We eliminate ten 20-F firm year observations with negative equity book value.

26

Table 1, panel A, provides a breakdown of sample firms by country. Sample firms are

from 24 countries, with the greatest proportion from the United Kingdom, France, and Germany.

Panel B of table 1 provides an industry breakdown. Sample firms are from many industries, with

the greatest proportion from manufacturing, services, finance, insurance and real estate, and

construction. Panel C of table 1 provides a breakdown by IAS adoption year. The number of

firms adopting IAS each year rises until 2000, then declines somewhat through 2004, and jumps

substantially in 2005 when IAS became mandatory for firms in many countries.

Table 2 reports descriptive statistics for IAS and US firms based on data for IAS firms

and their matched US firms after the IAS firms adopted IAS. It reveals that US firms have fewer

incidents of small positive earnings (6% versus 8%) and more incidents of large negative

earnings (13% versus 5%). These statistics suggest that US firms are less likely than IAS firms

to manage earnings towards a target and more likely to recognize losses in a timely manner.

Regarding control variables, the most notable difference in table 2 are that IAS firms are more

highly levered than US firms (median 1.43 versus 0.89). Table 2 also reveals US and IAS firms

are similar in size (median 5.48 versus 5.44), reflecting the effectiveness of our matching

procedure.

6. Results

6.1 COMPARISON OF IAS AND US FIRMS

6.1.1 Comparison of IAS and US Firms After IAS Firms Adopt IAS

Table 3 presents results for earnings management, timely loss recognition, and value

relevance for IAS and US firms after the IAS firms adopt IAS. It reveals consistent evidence

that US firms have higher accounting quality than IAS firms. In particular, six of the eight

27

accounting quality metrics are different in the predicted direction, with five of the differences

being significant.

Results for earnings smoothing provide evidence of less earnings smoothing for US firms

relative to IAS firms. In particular, US firms exhibit significantly higher variability of change in

net income, ΔNI*, 0.0058 versus 0.0045, a significantly higher ratio of variances of change in net

income, ΔNI*, and change in cash flow, ΔCF*, 1.1785 versus 1.0725, and a significantly less

negative correlation between accruals, ACC*, and cash flow, CF*, –0.1993 versus –0.3465.

Results for managing toward positive earnings are inconsistent with US firms managing

earnings less than IAS firms. In particular, the negative coefficient on SPOS from equation (5),

–0.0181, is opposite in sign from our prediction. However, because the coefficient is

insignificantly different from zero, the finding indicates no difference between US and IAS

firms’ tendencies to recognize small profits.

The timely loss recognition results indicate that US firms recognize large losses more

frequently than do IAS firms. In particular, the coefficient on LNEG in equation (6), −0.1905, is

significantly negative. Coupled with the results for earnings smoothing, this finding is consistent

with US firms being less likely to defer and smooth large losses rather than recognizing them in

a timely manner.

The final set of findings in table 3 relates to value relevance of accounting amounts.

First, regressions of price on earnings and equity book value for IAS and US firms reveal that the

R2 for US firms is significantly larger than that for IAS firms, 50.02% versus 28.17%.

Untabulated regression summary statistics indicate that, as expected, the coefficients on earnings

and equity book value are significantly positive for both IAS and US firms, and that both

28

coefficients are smaller for IAS firms. The findings are consistent with accounting amounts

being more value relevant for US firms than for IAS firms.

In contrast to the price regression, results from the good news and bad news value

relevance regressions indicate no difference in value relevance between US and IAS firms. In

particular, R2s for US firms are insignificantly smaller for good news firms (0.35% versus

0.93%) and insignificantly larger for bad news firms (7.36% versus 6.18%). However, the

higher R2s for both US and IAS bad news firms than good news firms is consistent with bad

news being reflected in earnings in a more timely manner than good news (Basu, 1997).21

6.1.2 Comparison of IAS and US Firms Before and After IAS Firms Adopt IAS

The preceding results suggest that accounting quality is higher for US firms than for IAS

firms. A related question that we address in this section is whether IAS application moves

accounting quality for IAS firms closer to that for US firms.

Table 4, panel A, presents a comparison of findings for earnings management, timely loss

recognition, and value relevance for IAS firms and US firms before the IAS firms adopted IAS,

i.e., when they applied domestic standards. The findings in panel A indicate a more striking

difference in accounting quality between US and IAS firms than in table 3. In particular, US

firms exhibit significantly less earnings management than do IAS firms as evidenced by the

variability of net income, the correlation of cash flows and accruals, and a smaller tendency than

IAS firms to recognize small profits. Also as predicted, the variability of net income relative to

that of cash flows is also higher for US firms, although not significantly so. In addition, the

21 An alternative approach to assessing the association between bad news returns and earnings is based on the coefficient on returns as described in Ball, Kothari, and Robin (2000). Untabulated findings show that the coefficient on returns is larger for bad news US firms than bad news IAS firms, which is consistent with US firms recognizing losses in a more timely manner than IAS firms.

29

timely loss recognition difference and all of the value relevance differences are significant in the

predicted direction.

The next question we address is whether the difference in accounting quality between

IAS and US firms is smaller when IAS firms apply IAS than when they applied domestic

standards. Table 4, panel B, addresses this question by presenting differences in each of the

accounting quality metrics between IAS and US firms in both periods, and changes in the

differences. Based on Barth, Landsman, and Lang (2008), which shows that the quality of

accounting amounts of IAS firms improved after applying IAS, we predict that the difference in

accounting quality between IAS and US firms is smaller after the IAS firms adopt IAS. The

evidence in the final column indicates that changes in the differences between IAS and US firms

are significant in the predicted direction for only one of the four earnings management metrics –

managing towards small positive income. In fact, for one earnings management metric, the

correlation of accruals and cash flows, the change is significantly negative, which indicates an

improvement in accounting quality for US firms relative to IAS firms. Panel B also indicates

that the change in the difference in the timely loss recognition metric is negative, but

insignificant. The result in panel B most consistent with our prediction relates to value

relevance. In particular, the relative difference in value relevance between US and IAS firms

decreases significantly after the IAS firms adopt IAS. Thus, although the evidence is mixed

regarding whether application of IAS is associated with an improvement in accounting quality

relative to US firms, the value relevance metrics suggest that there is some relative improvement.

6.1.3 Additional Analyses

Comparison of IAS and US Firms Before and After 2005

30

In 2005 IAS adoption became mandatory in many countries. It is possible that

accounting quality differences between US and IAS firms could differ in the years before and

after IAS became mandatory, although it is difficult to predict the direction. On the one hand,

incentives of early adopters of IAS could result in them having higher accounting quality than

firms that waited until IAS become mandatory. On the other hand, more uniform application of

IAS in the period after IAS became mandatory could result in improvement in accounting

quality.

To address the question of whether accounting quality differences between US and IAS

firms changed in 2005, we repeat our accounting quality comparison between US and IAS firms

separately before 2005 and in 2005 or after. Findings presented in table 5 for the two

comparisons generally are consistent with those based on the comparison using the combined

IAS sample in table 3. In particular, the before and after 2005 columns indicate that accounting

quality is higher for US firms in both periods although there are some minor differences. For

example, the difference in value relevance for the price regression is insignificant after 2005.

The final column in table 5 presents the changes in the differences in accounting quality

metrics for the US and IAS firms after and before 2005. The findings are mixed. For example,

one earnings management metric, the variability of the change in net income, is significantly

smaller by 0.0010 after 2005. In contrast, the gaps between the ratio of variability of change in

net income to variability of change in cash flows and the correlation of accruals and cash flows

widen in the latter period, although insignificantly so. However, findings relating to the value

relevance metrics indicate a significant relative improvement in value relevance after 2005.

Most strikingly, for the price regression, the difference in value relevance declines by 31.76

percentage points. Taken together, the findings in table 5 indicate that the potential effects on

31

accounting quality arising from incentives to voluntarily adopt IAS do not affect inferences

relating to the comparison of quality of IAS-based and US GAAP-based accounting amounts.

Comparison of US Firms and IAS Firms by Legal Origin

As discussed in section 3.1, findings in the literature suggest that application of

accounting standards is not likely to have uniform effects on accounting quality in reporting

regimes with different regulatory and litigation environments. To determine whether cross-

country differences in features of the financial reporting environment result in differences in

accounting quality among firms that apply IAS, we next compare accounting quality metrics for

US and IAS firms separately for IAS firms in countries grouped by legal origin.

La Porta et al. (1998) and subsequent studies indicate that legal origin differences affect

institutional features of financial markets, including the financial reporting environment. Key

groups are English, Germanic/French, and Scandinavian legal origins. We categorize IAS firms

into one of four groups, where the first three groups include countries following the

categorization in La Porta et al. (1998) and the fourth group includes China. The English group

(EL) includes Australia, Canada, Hong Kong, Ireland, Singapore, South Africa, and UK; the

Germanic/French group (GFL) includes Austria, Belgium, France, Germany, Greece, Italy,

Netherlands, Portugal, Spain, and Switzerland; the Scandinavian group (SL) includes Denmark,

Finland, Norway, and Sweden.

Table 6, panel A, presents comparisons of accounting quality metrics for US firms and

IAS firms separately for each of the four groups. Panel B presents comparisons pairwise

between each of the IAS firm groups. The findings in panel A indicate that differences in

accounting quality between US firms and non-US firms for the full sample in table 3 are largely

attributable to the difference in quality between US and German/French legal origin firms. Most

32

notably, for the English legal origin and US firm comparison, for only one accounting quality

metric—the price value relevance regression—is the difference between US and EL firms

significant in the predicted direction. In fact, for five of the metrics the differences indicate

higher accounting quality for EL firms, although the differences are insignificant. The findings

relating to the comparison of US and Scandinavian firms and of US and Chinese firms are

similar to those for the US-EL comparison in that only two of the differences are significant in

the predicted direction, i.e., those relating to the variability of the change in net income, the

frequency of small positive income, and timely loss recognition. In fact, the findings indicate

that SL firms have significantly higher value relevance for good news firms. Only the findings

relating to the US and German/French legal origin comparison show a systematic difference in

quality between US and IAS firms. In particular, for seven of the eight metrics the differences

are in the predicted direction, with significant differences for three of the four earnings

management metrics, the timely loss recognition metric, and the price regression metric. Only

the good news quality metric is significantly higher for GL firms than for US firms.

The findings in panel B confirm the findings in panel A. In particular, with the exception

of comparisons of GL firms and firms in other legal groups, there is no clear pattern of

accounting quality differences. For example, EL firms have exhibit somewhat less earnings

management than SL firms, but somewhat lower value relevance. In addition, many of the

accounting quality metric differences for all six comparisons are insignificant, including those

involving GL firms.

Taken together, the table 6 findings suggest that accounting quality differences between

US and IAS firms arise from differences between US and German/French legal origin firms.

However, although the lack of significance of differences for the US and Chinese comparison

33

could be affected by low power arising from the relatively small number of observations, cross-

country differences in features of the financial reporting system other than accounting standards

among firms that apply IAS in the other three legal origin groups do appear to result in

differences in accounting quality.

6.2 COMPARISON OF IAS AND 20-F FIRMS

Table 7 presents descriptive statistics for 20-F firms paralleling those in table 2 for IAS

and US firms. Together with table 2, table 7 reveals that 20-F and IAS firms exhibit similar

growth, extent of external financing, and asset turnover. However, the 20-F firms are larger than

the IAS firms.

Table 8 presents results for earnings management, timely loss recognition, and value

relevance for IAS and 20-F firms. Overall, the results reveal no clear pattern of differences in

accounting quality between IAS and 20-F firms. In particular, there are three significant

differences, two of which indicate higher accounting quality for 20-F firms. These are the

correlation between accruals, ACC*, and cash flow, CF*, –0.1286 for 20-F firms versus –0.3315

for IAS firms, and timely loss recognition (LNEG coefficient = –0.0729). The significant

difference consistent with higher accounting quality for IAS firms is the ratio of the variability of

change in net income to the variability of change in cash flow (1.7049 for IAS firms versus

1.1890 for 20-F firms). Of the remaining five comparisons, three (two) differences are

consistent with higher accounting quality for IAS (20-F) firms.

The table 8 findings have two important implications for our study. The first is that

although the quality of accounting amounts of IAS firms is lower than that of US firms, it is not

different from that of US GAAP reconciled amounts for non-US firms. This result is consistent

with findings in Lang, Raedy, and Wilson (2006), and may reflect the fact that cross-listed firms

34

likely face less rigorous legal and regulatory oversight than do US firms (Bradshaw and Miller,

2007). This suggests that by removing the reconciliation requirement for firms applying IAS, the

SEC might not sacrifice accounting quality relative to that associated with reconciled US GAAP

amounts, and might remove an impediment to non-US firms trading securities in the US. The

second is that these findings enhance the internal validity of the tests on which we base

inferences about accounting quality differences between IAS and US firms using the consistent

pattern of significant differences. The findings in table 8 indicate that finding such a pattern is

unlikely in the absence of an underlying difference in accounting quality.

7. Summary and Concluding Remarks

This study compares the quality of accounting amounts of firms that apply IAS and a

matched sample of US firms that apply US GAAP. Our results indicate that US firms have

higher accounting quality than IAS firms. In particular, US firms have a significantly higher

variance of change in net income, a significantly higher ratio of the variances of change in net

income and change in cash flows, a significantly less negative correlation between accruals and

cash flows, a significantly higher frequency of large negative net income, and a significantly

higher value relevance of earnings and equity book value for share prices.

Comparisons of accounting amounts of IAS and US firms before and after the IAS firms

adopt IAS suggest that application of IAS has no systematic impact on the differences in

accounting quality between the two sets of firms, although differences in value relevance

decrease after the IAS firms adopt IAS. Additional findings indicate that US firms have higher

accounting quality than IAS firms before and after 2005, when application of IAS became

mandatory in many countries. However, US firms have higher accounting quality than IAS

firms only in the German/French legal origin groups of countries comprising our IAS sample.

35

Although our findings suggest that the quality of accounting amounts of firms applying

IAS is generally lower than that of firms applying US GAAP, it is premature to conclude that the

quality of accounting amounts of US firms applying IAS would be lower than that of US firms

applying US GAAP. This is because we cannot subject our IAS-based accounting amounts to all

features of the US financial reporting system. Also, if a substantial portion of firms that we

classify as IAS firms are affected by known potential sources of classification error, it is likely

that our findings are biased in favor of our prediction that US firms have higher accounting

quality than IAS firms.

We also compare the quality of accounting amounts of firms that apply IAS and that of

accounting amounts based on US GAAP of a sample of non-US firms that cross-list on US

exchanges and reconcile accounting amounts from domestic standards to US GAAP on Form 20-

F. Results from this comparison reveal no clear pattern of differences in quality between IAS-

based and reconciled US GAAP-based accounting amounts. Of our eight accounting quality

metrics, there are three significant differences, two of which indicate higher accounting quality

for 20-F firms and one of which indicates higher quality for IAS firms. These findings suggest

that by removing the reconciliation requirement for firms applying IAS, the SEC might remove

an impediment to non-US firms trading securities in the US, while not sacrificing accounting

quality relative to that associated with reconciled US GAAP amounts.

36

REFERENCES

Alford, A., J. Jones, R. Leftwich, and M. Zmijewski, 1993. The Relative Information Content of

Accounting Disclosures in Different Countries. Journal of Accounting Research 31

(Supplement): 183-223.

Ashbaugh, H. 2001. Non-U.S. Firms’ Accounting Standard Choices. Journal of Accounting and

Public Policy 20: 129-153.

Ashbaugh, H., and M. Pincus, 2001. Domestic Accounting Standards, International Accounting

Standards, and the Predictability of Earnings, Journal of Accounting Research 39: 417-434.

Ball, R.; S.P. Kothari; and A. Robin. 2000. The Effect of International Institutional Factors on

Properties of Accounting Earnings.” Journal of Accounting and Economics 29: 1-51.

Ball, R.; A. Robin, and J.S. Wu. 2003. Incentives versus Standards: Properties of Accounting

Income in Four East Asian Countries.” Journal of Accounting and Economics 36: 235-270.

Barth, M.E., W.H. Beaver and W.R. Landsman. 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., Y. Konchitchki, and W.R. Landsman. 2007. Cost of Capital and Financial

Statement Transparency. Working paper, Stanford University and University of North

Carolina.

Barth, M.E., W.R. Landsman, and M. Lang. 2008. International Accounting Standards and

Accounting Quality. Journal of Accounting Research, forthcoming.

Bartov, E., S. Goldberg, and M. Kim. 2004. Comparative Value Relevance among German, U.S.

and International Accounting Standards: A German Stock Market Perspective. Working

paper, New York University.

37

Basu, S., 1997, The Conservatism Principle and the Asymmetric Timeliness of Earnings, Journal

of Accounting and Economics 24, 3-37.

Bhattacharya, U., H. Daouk, and M. Welker. 2003. The World Price of Earnings Opacity. The

Accounting Review 78: 641-678.

Bradshaw, M.T., and G.S. Miller. 2007. Will Harmonizing Accounting Standards Really

Harmonize Accounting? Evidence from Non-U.S. Firms Adopting US GAAP. Working

paper, Harvard Business School.

Breeden, R. 1994. Foreign Companies and U.S. Markets in a Time of Economic Transformation.

Fordham International Law Journal 17.

Burgstahler, D., and I. Dichev. 1997. Earnings Management to Avoid Earnings Decreases and

Losses.” Journal of Accounting and Economics 24: 99-126.

Burgstahler, D., L. Hail, and C. Leuz. 2006. The Importance of Reporting Incentives: Earnings

Management in European Private and Public Firms. The Accounting Revew 81, 983-1016.

Cairns, D. 1999. Degrees of compliance. Accountancy International (September): 68-69.

Cramer, J. S. 1987. Mean and Variance of R2 in Small and Moderate Samples.” Journal of

Econometrics 35: 253-66.

Daske, H., L. Hail, C. Leuz, and R. Verdi. 2007a. Adopting a Label: Heterogeneity in the

Economic Consequences of IFRS Adoptions. Working paper, Wharton School.

Daske, H., L. Hail, C. Leuz, and R. Verdi. 2007b. Mandatory IFRS Reporting Around the World:

Early Evidence on the Economic Consequences. Working paper, University of Chicago.

Dechow, P.M. 1994. Accounting Earnings and Cash Flows as Measures of Firm Performance:

The Role of Accounting Accruals. Journal of Accounting and Economics 18: 3-42.

38

Eccher, E. and P. Healy. 2003. The Role of International Accounting Standards in Transitional

Economies: A Study of the People's Republic of China. Working paper, Massachusetts

Institute of Technology.

Ewert, R., and A. Wagenhofer. 2005. Economic Effects of Tightening Accounting Standards to

Restrict Earnings Management. The Accounting Review 80: 1101-1124.

Greene, W., 1993. Econometric Analysis. MacMillan, New York.

Harris, M., and K. Muller. 1999. The Market Valuation of IAS versus US GAAP Accounting

Measures Using Form 20-F Reconciliations. Journal of Accounting and Economics 26: 285-

312.

Hung, M. and K.R. Subramanyam. 2007. Financial Statement Effects of Adopting International

Accounting Standards: The Case of Germany. Review of Accounting Studies, forthcoming.

Jenkins, E. 1999. Financial Reporting in a Global Capital Market World. Financial Accounting

Series, No. 198-A (June 29): 2-6.

Land, J., and M. Lang. 2002. Empirical Evidence on the Evolution of International Earnings. The

Accounting Review 77: 115-134.

Lang, M., J. Raedy, and M. Yetman. 2003. How Representative are Firms that are Cross Listed

in the United States? An Analysis of Accounting Quality. Journal of Accounting Research

41: 363-386.

Lang, M., J. Raedy, and W. Wilson. 2006. Earnings Management and Cross Listing: Are

Reconciled Earnings Comparable to US Earnings? Journal of Accounting and Economics 42:

255-283.

La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R.W. Vishny. 1998. Law and Finance.

Journal of Political Economy 106: 1113-1155.

39

Leuz, C. 2003. IAS Versus U.S. GAAP: Information Asymmetry-based Evidence from

Germany’s New Market. Journal of Accounting Research 41: 445-472.

Leuz, C.; D. Nanda; and P. Wysocki. 2003. Earnings Management and Investor Protection: An

International Comparison. Journal of Financial Economics 69: 505-527.

Myers, L.A., and D.J. Skinner. 2007. Earnings Momentum and Earnings Management. Journal

of Accounting, Auditing, and Finance 22: 249-284.

Nicolaisen, D.T. 2005. A Securities Regulator Looks at Convergence. Northwestern Journal of

International Law and Business Vol. 25, No. 3: 661-686.

Pagano, M.; A. Röell; and J. Zehner. 2002. The Geography of Equity Listings: Why do

Companies List Abroad? Journal of Finance 57: 2651-2694.

Pownall, G., and K. Schipper. 1999. Implications of Accounting Research for the SEC’s

Consideration of International Accounting Standards for U.S. Securities. Accounting

Horizons 13 (September): 259–280.

Schipper, K. 2003. Principles-Based Accounting Standards. Accounting Horizons, March: 61-72.

Securities and Exchange Commission. 2007a. Acceptance from Foreign Private Issuers of

Financial Statements Prepared in Accordance with International Financial Reporting

Standards without Reconciliation to U.S. GAAP; Proposed Rule.

Securities and Exchange Commission. 2007b. Concept Release on Allowing U.S. Issuers to

Perpare Financial Statements in Accordance with International Financial Reporting

Standards; Proposed Rule.

Street, D. and S. Gray. 2001. Observance of International Accounting Standards: Factors

Explaining Non-compliance. ACCA Research Report No. 74.

40

Tokar, M. 2005. Convergence and the Implementation of a Single Set of Global Standards: The

Real-Life Challenge. Northwestern Journal of International Law and Business Vol. 25, No.

3: 687-710.

Van Tendeloo, B. and A. Vanstraelen. 2005. Earnings Management under German GAAP versus

IFRS. European Accounting Review 14:1, 155-180.

Wysocki, P. 2005. Assessing Earnings and Accruals Quality: US and International Evidence.

Working Paper, Massachusetts Institute of Technology.

41

TABLE 1 IAS Firm Sample Composition by Country, Industry, and Year of Adoption

Panel A: Composition by Country

Number of Firm-Year

Observations

Percentage of Firm-Year

Observations Number of

Firms Percentage of

Firms Australia 526 4.60 108 4.88 Austria 178 1.56 31 1.40 Belgium 290 2.53 59 2.67 Canada 2 0.02 1 0.05 China 111 0.97 41 1.85 Czech Republic 57 0.50 11 0.50 Denmark 488 4.26 80 3.62 Finland 497 4.34 97 4.39 France 1,496 13.07 326 14.74 Germany 1,384 12.09 294 13.29 Greece 99 0.87 47 2.12 Hong Kong 58 0.51 11 0.50 Hungary 27 0.24 8 0.36 Ireland 184 1.61 26 1.18 Italy 741 6.48 146 6.60 Netherlands 587 5.13 81 3.66 Norway 531 4.64 107 4.84 Portugal 165 1.44 32 1.45 Singapore 36 0.31 6 0.27 Spain 12 0.10 7 0.32 Sweden 878 7.67 201 9.09 Switzerland 381 3.33 70 3.16 Turkey 73 0.64 28 1.27 United Kingdom 2,642 23.09 394 17.81 Totals 11,443 100.00 2,212 100.00

42

TABLE 1 IAS Firm Sample Composition by Country, Industry, and Year of Adoption

Panel B: Composition by Industry

Number of Firm-Year

Observations

Percentage of Firm-Year

Observations Number of

Firms Percentage of

Firms Agriculture, Forestry and Fishing 75 0.66 17 0.77 Mining 642 5.61 114 5.15 Construction 1,098 9.60 175 7.91 Manufacturing 4,671 40.82 858 38.79 Utilities 419 3.66 74 3.35 Gas Distribution 41 0.36 9 0.41 Retail Trade 319 2.79 65 2.94 Finance, Insurance and Real Estate 1,370 11.97 296 13.38 Services 2,629 22.97 571 25.81 Public Administration 179 1.56 33 1.49 Totals 11,443 100.00 2,212 100.00

43

TABLE 1 IAS Firm Sample Composition by Country, Industry, and Year of Adoption

Panel C: Composition by Year

Number of Firm-Year

Observations

Percentage of Firm-Year Observations

Number of Firms Adopting IAS

Percentage of Firms Adopting IAS

1995 10 0.09 3 0.14 1996 30 0.26 6 0.27 1997 47 0.41 8 0.36 1998 109 0.95 17 0.77 1999 253 2.21 53 2.40 2000 198 1.73 40 1.81 2001 177 1.55 33 1.49 2002 228 1.99 50 2.26 2003 175 1.53 40 1.81 2004 358 3.13 83 3.75 2005 8,539 74.62 1,682 76.04 2006 1,319 11.53 197 8.91 Totals 11,443 100.00 2,212 100.00

44

TABLE 2 Descriptive Statistics: IAS Firms and US Firms

IAS Firms (N=2,804) US Firms (N=2,804)

Mean Median Standard Deviation Mean Median

Standard Deviation

Test Variables ΔNI 0.01 0.01 0.04 0.01 0.01 0.06 ΔCF 0.01 0.01 0.05 0.01 0.01 0.07 ACC −0.04 −0.03 0.05 −0.05 −0.04 0.06 CF 0.07 0.07 0.06 0.04 0.06 0.11 SPOS 0.08 0.00 0.27 0.06 0.00 0.24 LNEG 0.05 0.00 0.21 0.13 0.00 0.33 RETURN 0.22 0.22 0.29 0.07 0.08 0.35 Stock Return 0.30 0.25 0.38 0.14 0.09 0.39 NI/P 0.06 0.06 0.11 0.01 0.03 0.09 P 37.37 10.37 81.76 19.32 12.81 19.02 BVE 0.77 0.62 0.67 0.53 0.46 0.31 NI 0.06 0.06 0.10 −0.01 0.03 0.11

Control Variables

LEV 1.99 1.43 1.87 1.25 0.89 1.15 GROWTH 0.02 −0.01 0.19 0.14 0.10 0.23 EISSUE −0.03 −0.08 0.11 0.09 0.06 0.24 DISSUE 0.06 0.02 0.25 0.14 0.06 0.29 TURN 0.91 0.91 0.50 0.93 0.84 0.60 SIZE 5.47 5.44 1.76 5.59 5.48 1.66 CF 0.07 0.07 0.06 0.04 0.06 0.11

All statistics are for IAS and US firms after IAS firms adopted IAS. We define ΔNI as change in annual earnings, where earnings is scaled by end-of-year total assets; ΔCF as change in annual net cash flow from operating activities, where cash flow is scaled by end-of-year total assets; ACC as earnings less cash flow from operating activities, scaled by end-of-year total assets; CF as annual net cash flow from operating activities, scaled by end-of-year total assets; SPOS as an indicator equal to 1 for observations for which annual earnings scaled by total assets is between 0 and 0.01; LNEG as an indicator equal to 1 for observations for which annual earnings scaled by total assets < −0.20; RETURN as natural logarithm of the ratio of stock price three months after fiscal year end to stock price nine months before fiscal year end, adjusted for dividends and stock splits; Stock Return cumulative percentage change in stock price beginning nine months before fiscal year end and ending three months after fiscal year end, adjusted for dividends and stock splits; P as price six months after the fiscal year-end; NI/P as earnings per share scaled by price per share at the beginning of the year; LEV as end-of-year total liabilities divided by end-of-year total equity, GROWTH as percentage change in sales; EISSUE as percentage change in common stock; DISSUE as percentage change in total liabilities; TURN as sales divided by end-of-year total assets; SIZE as natural logarithm of market value of equity in thousands of dollars as of the end of the year; BVE as book value of shareholders’ equity per share; NI as net income per share.

45

TABLE 3 Comparison of IAS and US Firms’ Accounting Quality after IAS Firms Adopt IAS

Earnings Management IAS US Metric Prediction (N = 2,804) (N = 2,804)

Variability of NI* US > IAS 0.0045 0.0058*

Variability of NI* over CF* US > IAS 1.0725 1.1785*

Correlation of ACC* and CF* US > IAS 0.3465 0.1993*

Small Positive NI (SPOS) + 0.0181

Timely Loss Recognition

Metric

Large Negative NI (LNEG) 0.1905#

Value Relevance

Regression Adjusted R2

Price US > IAS 0.2817 0.5002*

Good News US > IAS 0.0093 0.0035

Bad News US > IAS 0.0618 0.0736

*Significantly different between IAS and US firms at the 0.05 level.

#Significantly different from zero at the 0.05 level.

All variables in each of the regressions are winsorized at the 1% level.

We define variability of NI* ( CF*) as the variance of residuals from a regression of NI

( CF) on control variables, and the variability of NI* over CF* as the ratio of the variability

of NI* to the variability of CF*. Correlation of ACC* and CF* is the Spearman correlation

between the residuals from the ACC and CF regressions; both sets of residuals are from a

regression of each variable on control variables. NI, CF, ACC, and CF are defined in table 2.

We regress an indicator variable that equals 1 for IAS firms and 0 for US firms on SPOS (LNEG)

and control variables. SPOS (LNEG) is an indicator variable that equals 1 when annual net

income scaled by total assets is between 0 and 0.01 (< –0.20) and 0 otherwise; the coefficient on

the indicator variable is tabulated.

The price regression is based on two-stage estimation. In the first stage, P is regressed on

country and industry fixed effects, where P is price as of six months after fiscal year-end. The

second stage regression is P* = 0 + 1 BVEPS + 2 NIPS + , where P* is the residual from the

first-stage regression, BVEPS is book value of equity per share, and NIPS is net income per

share. The good/bad news regression is based on a two-stage regression. In the first stage, NI/P

is regressed on country and industry fixed effects. The second stage regression is NI/P* = 1

46

RETURN + , where NI/P* is the residual from the first stage regression, and RETURN is the

natural logarithm of the ratio of stock price three months after fiscal year end to stock price nine

months before fiscal year end, adjusted for dividends and stock splits. Good (bad) news

observations are those for which RETURN is nonnegative (negative). Adjusted R2 is from the

second-stage regressions.

47

TABLE 4 Accounting Quality Analysis of IAS and US Firms Before IAS Firms Adopt IAS and Differences

Panel A: Pre Adoption Period

Earnings Management IAS US

Metric Prediction (N = 8,639) (N = 8,639)

Variability of NI* US > IAS 0.0042 0.0054*

Variability of NI* over CF* US > IAS 1.0395 1.0622

Correlation of ACC* and CF* US > IAS 0.3345 0.2563*

Small Positive NI (SPOS) + 0.0381#

Timely Loss Recognition

Metric

Large Negative NI (LNEG) 0.1138#

Value Relevance

Regression Adjusted R2

Price US > IAS 0.1159 0.4513*

Good News US > IAS 0.0018 0.0110*

Bad News US > IAS 0.0004 0.0779*

Panel B: Differences

Earnings Management Diff Pre Diff Post Change

Metric IAS US IAS US Prediction Post Pre

Variability of NI* 0.0012 0.0012 + 0.0000

Variability of NI* over CF* 0.0227 0.1055 + 0.0828

Correlation of ACC* and CF* 0.0782 0.2272 + 0.1490*

Small Positive NI (SPOS) 0.0381 0.0181 0.0562*

Timely Loss Recognition

Metric

Large Negative NI (LNEG) 0.1138 0.1905 + 0.0767

Value Relevance Diff Pre Diff Post Change

Regression Adjusted R2 IAS – US IAS – US Prediction Post – Pre

Price 0.3354 0.2185 + 0.1169*

Good News 0.0092 0.0058 + 0.0150*

Bad News 0.0775 0.0118 + 0.0637*

*Significantly different between IAS and US firms at the 0.05 level.

#Significantly different from zero at the 0.05 level.

48

All variables in each of the regressions are winsorized at the 1% level.

We define variability of NI* ( CF*) as the variance of residuals from a regression of NI

( CF) on control variables, and the variability of NI* over CF* as the ratio of the variability

of NI* to the variability of CF*. Correlation of ACC* and CF* is the Spearman correlation

between the residuals from the ACC and CF regressions; both sets of residuals are from a

regression of each variable on control variables. NI, CF, ACC, and CF are defined in table 2.

We regress an indicator variable that equals 1 for IAS firms and 0 for US firms on SPOS (LNEG)

and control variables. SPOS (LNEG) is an indicator variable that equals 1 when annual net

income scaled by total assets is between 0 and 0.01 (< –0.20) and 0 otherwise; the coefficient on

the indicator variable is tabulated.

The price regression is based on two-stage estimation. In the first stage, P is regressed on

country and industry fixed effects, where P is price as of six months after fiscal year-end. The

second stage regression is P* = 0 + 1 BVEPS + 2 NIPS + , where P* is the residual from the

first-stage regression, BVEPS is book value of equity per share, and NIPS is net income per

share. The good/bad news regression is based on a two-stage regression. In the first stage, NI/P

is regressed on country and industry fixed effects. The second stage regression is NI/P* = 1

RETURN + , where NI/P* is the residual from the first stage regression, and RETURN is the

natural logarithm of the ratio of stock price three months after fiscal year end to stock price nine

months before fiscal year end, adjusted for dividends and stock splits. Good (bad) news

observations are those for which RETURN is nonnegative (negative). Adjusted R2 is from the

second-stage regressions.

49

TABLE 5 Differences in Accounting Quality Between IAS and US Firms after IAS Firms Adopt IAS, Partitioned on IAS Adoption Before and

After 2005

Before 2005 After 2005

Earnings Management IAS US IAS US

Metric Prediction (N = 924) Prediction (N = 1,880) Prediction After � Before

Variability of NI* 0.0020* 0.0010* ? 0.0010*

Variability of NI* over CF* 0.0490 0.1191* ? �0.0701

Correlation of ACC* and CF* 0.1181* 0.1736* ? �0.0555

Small Positive NI (SPOS) + 0.0714 + 0.0627 ? �0.1341#

Timely Loss Recognition

Metric

Large Negative NI (LNEG) 0.2013# 0.1822# ? 0.0191#

Value Relevance

Regression Adjusted R2

Price 0.3242* 0.0066 ? 0.3176*

Good News 0.0069 0.0159* ? 0.0228*

Bad News 0.0463 0.0367 ? 0.0830*

*Significantly different between IAS and US firms at the 0.05 level.

#Significantly different from zero at the 0.05 level.

All variables in each of the regressions are winsorized at the 1% level.

We define variability of NI* ( CF*) as the variance of residuals from a regression of NI ( CF) on control variables, and the

variability of NI* over CF* as the ratio of the variability of NI* to the variability of CF*. Correlation of ACC* and CF* is the

Spearman correlation between the residuals from the ACC and CF regressions; both sets of residuals are from a regression of each

variable on control variables. NI, CF, ACC, and CF are defined in table 2.

50

We regress an indicator variable that equals 1 for IAS firms and 0 for US firms on SPOS (LNEG) and control variables. SPOS

(LNEG) is an indicator variable that equals 1 when annual net income scaled by total assets is between 0 and 0.01 (< –0.20) and 0

otherwise; the coefficient on the indicator variable is tabulated.

The price regression is based on two-stage estimation. In the first stage, P is regressed on country and industry fixed effects, where P

is price as of six months after fiscal year-end. The second stage regression is P* = 0 + 1 BVEPS + 2 NIPS + , where P* is the

residual from the first-stage regression, BVEPS is book value of equity per share, and NIPS is net income per share. The good/bad

news regression is based on a two-stage regression. In the first stage, NI/P is regressed on country and industry fixed effects. The

second stage regression is NI/P* = 1 RETURN + , where NI/P* is the residual from the first stage regression, and RETURN is the

natural logarithm of the ratio of stock price three months after fiscal year end to stock price nine months before fiscal year end,

adjusted for dividends and stock splits. Good (bad) news observations are those for which RETURN is nonnegative (negative).

Adjusted R2 is from the second-stage regressions.

51

TABLE 6 Differences in Accounting Quality Between US Firms and IAS Firms by Legal Origin Grouping After IAS Firms Adopt IAS

Panel A: Legal Origin Comparisons

Earnings Management EL† US GFL

† US SL

† US Chinese

† US

Metric Prediction (N = 593) (N = 1,547) (N = 532) (N = 97)

Variability of NI* 0.0001 0.0017* 0.0012* 0.0075*

Variability of NI* over CF* 0.1099 0.2407* 0.0022 0.4537

Correlation of ACC* and CF* 0.0745 0.2120* 0.0984 0.2167

Small Positive NI (SPOS) + 0.0814 0.0118 0.0864 0.2694#

Timely Loss Recognition

Metric

Large Negative NI (LNEG) 0.0598 0.2629# 0.1620# 0.1624

Value Relevance

Regression Adjusted R2

Price 0.3494* 0.1854* 0.1045 0.0568

Good News 0.0008 0.0126* 0.0210* 0.0066

Bad News 0.0150 0.0225 0.0042 0.0162

52

TABLE 6 Differences in Accounting Quality Between US Firms and IAS Firms by Legal Origin Grouping After IAS Firms Adopt IAS

Panel B: Cross Legal Origin Comparisons

Earnings Management

Metric EL†

GFL† EL

†SL

† EL

†Chinese

† SL

†GFL

† SL

†Chinese

† GFL

†Chinese

†

Variability of NI* 0.0018* 0.0013* 0.0076* 0.0005 0.0063* 0.0058*

Variability of NI* over CF* 0.3506* 0.1077* 0.3438 0.2429* 0.4515* 0.6944*

Correlation of ACC* and CF* 0.1375* 0.0239 0.1422 0.1136* 0.1183* 0.0047

Small Positive NI (SPOS) 0.0932 0.0050 0.1880 0.0982 0.3558 0.2576

Timely Loss Recognition

Metric

Large Negative NI (LNEG) 0.2031# 0.1022 0.1026# 0.1009# 0.0004 0.1005

Value Relevance

Regression Adjusted R2

Price 0.1640 0.2449* 0.2926 0.0809 0.0477 0.1286

Good News 0.0118 0.0202* 0.0058 0.0084 0.0296 0.0060

Bad News 0.0375 0.0108 0.0012 0.0267 0.0120 0.0387

*Significantly different between IAS and US firms at the 0.05 level.

#Significantly different from zero at the 0.05 level.

All variables in each of the regressions are winsorized at the 1% level.

†Legal origin groupings [countries within the group in brackets] are; English legal origin (EL) [Australia, Canada, Hong Kong,

Ireland, Israel, Singapore, South Africa, and UK] (La Porta et al., 1998); Germanic/French legal origin (GFL) [Austria, Belgium,

France, Germany, Greece, Italy, Netherlands, Peru, Philippines, Portugal, Spain, and Switzerland] (La Porta et al., 1998);

Scandinavian legal origin (SL) [Denmark, Finland, Norway, and Sweden] (La Porta et al., 1998); China; and the US.

We define variability of NI* ( CF*) as the variance of residuals from a regression of NI ( CF) on control variables, and the

variability of NI* over CF* as the ratio of the variability of NI* to the variability of CF*. Correlation of ACC* and CF* is the

53

Spearman correlation between the residuals from the ACC and CF regressions; both sets of residuals are from a regression of each

variable on control variables. NI, CF, ACC, and CF are defined in table 2.

We regress an indicator variable that equals 1 for IAS firms and 0 for US firms on SPOS (LNEG) and control variables. SPOS

(LNEG) is an indicator variable that equals 1 when annual net income scaled by total assets is between 0 and 0.01 (< –0.20) and 0

otherwise; the coefficient on the indicator variable is tabulated.

The price regression is based on two-stage estimation. In the first stage, P is regressed on country and industry fixed effects, where P

is price as of six months after fiscal year-end. The second stage regression is P* = 0 + 1 BVEPS + 2 NIPS + , where P* is the

residual from the first-stage regression, BVEPS is book value of equity per share, and NIPS is net income per share. The good/bad

news regression is based on a two-stage regression. In the first stage, NI/P is regressed on country and industry fixed effects. The

second stage regression is NI/P* = 1 RETURN + , where NI/P* is the residual from the first stage regression, and RETURN is the

natural logarithm of the ratio of stock price three months after fiscal year end to stock price nine months before fiscal year end,

adjusted for dividends and stock splits. Good (bad) news observations are those for which RETURN is nonnegative (negative).

Adjusted R2 is from the second-stage regressions.

54

TABLE 7

Descriptive Statistics for 20-F Firms

20-F Firms (N = 637)

Variablea Mean Median

Standard

Deviation

Test Variables:

�NI 0.01 0.01 0.10

�CF 0.01 0.01 0.09

ACC 0.07 0.06 0.10

CF 0.09 0.09 0.11

SPOS 0.07 0.00 0.25

LNEG 0.07 0.00 0.26

RETURN 0.03 0.02 0.20

NI/ P 0.03 0.05 0.12

P 24.05 17.19 23.89

BVE 7.80 6.77 7.12

NI 0.65 0.30 1.01

Control Variables:

LEV 1.20 0.99 0.83

GROWTH 0.14 0.09 0.36

EISSUE 0.09 0.04 0.27

DISSUE 0.14 0.06 0.42

TURN 0.74 0.67 0.46

SIZE 15.30 15.40 1.74

CF 0.09 0.09 0.11

We define �NI as change in annual earnings, where earnings is scaled by end-of-year total

assets; �CF as change in annual net cash flow from operating activities, where cash flow is

scaled by end-of-year total assets; ACC as earnings less cash flow from operating activities,

scaled by end-of-year total assets; CF as annual net cash flow from operating activities, scaled by

end-of-year total assets; SPOS as an indicator equal to 1 for observations for which annual

earnings scaled by total assets is between 0 and 0.01; LNEG as an indicator equal to 1 for

observations for which annual earnings scaled by total assets is < 0.20, RETURN is natural

logarithm of the ratio of stock price three months after fiscal year end to stock price nine months

before fiscal year end, adjusted for dividends and stock splits; P as price six months after fiscal

year-end; NI/P as earnings per share scaled by price per share at the beginning of the year; LEV

as end-of-year total liabilities divided by end-of-year total equity, GROWTH as percentage

change in sales; EISSUE as percentage change in common stock; DISSUE as percentage change

in total liabilities; TURN as sales divided by end-of-year total assets; SIZE as the natural

logarithm of market value of equity in thousands of dollars as of the end of the year; BVE as

book value of shareholders’ equity per share; NI as net income per share.

55

TABLE 8 Comparison of IAS and 20-F Firms’ Accounting Quality After IAS Firms Adopt IAS

Earnings Management IAS 20-F

Metric Prediction (N = 2,262) (N = 637)

Variability of NI* 20-F > IAS 0.0099 0.0079

Variability of NI* over CF* 20-F > IAS 1.7049 1.1890*

Correlation of ACC* and CF* 20-F > IAS 0.3315 0.1286*

Small Positive NI (SPOS) + 0.0026

Timely Loss Recognition

Metric

Large Negative NI (LNEG) 0.0729#

Value Relevance

Regression Adjusted R2

Price 20-F > IAS 0.2711 0.2006

Good News 20-F > IAS 0.0081 0.0131

Bad News 20-F > IAS 0.0501 0.0145

*Significantly different between IAS and US firms at the 0.05 level.

#Significantly different from zero at the 0.05 level.

All variables in each of the regressions are winsorized at the 1% level.

We define variability of NI* ( CF*) as the variance of residuals from a regression of NI

( CF) on control variables, and the variability of NI* over CF* as the ratio of the variability

of NI* to the variability of CF*. Correlation of ACC* and CF* is the Spearman correlation

between the residuals from the ACC and CF regressions; both sets of residuals are from a

regression of each variable on control variables. NI, CF, ACC, and CF are defined in table 2.

We regress an indicator variable that equals 1 for IAS firms and 0 for US firms on SPOS (LNEG)

and control variables. SPOS (LNEG) is an indicator variable that equals 1 when annual net

income scaled by total assets is between 0 and 0.01 (< –0.20) and 0 otherwise; the coefficient on

the indicator variable is tabulated.

The price regression is based on two-stage estimation. In the first stage, P is regressed on

country and industry fixed effects, where P is price as of six months after fiscal year-end. The

second stage regression is P* = 0 + 1 BVEPS + 2 NIPS + , where P* is the residual from the

first-stage regression, BVEPS is book value of equity per share, and NIPS is net income per

share. The good/bad news regression is based on a two-stage regression. In the first stage, NI/P

is regressed on country and industry fixed effects. The second stage regression is NI/P* = 1

RETURN + , where NI/P* is the residual from the first stage regression, and RETURN is the

natural logarithm of the ratio of stock price three months after fiscal year end to stock price nine

56

months before fiscal year end, adjusted for dividends and stock splits. Good (bad) news

observations are those for which RETURN is nonnegative (negative). Adjusted R2 is from the

second-stage regressions.