Could IFRS Replace US GAAP?

A Comparison of Earnings Attributes and Informativeness in the US Market*

Elizabeth A. Gordon**

Bjorn N. Jorgensen***

Cheryl L. Linthicum****

October 22, 2009

* We would like to thank Russell Craig, Scott Taub, and seminar participants at the 2009 American Accounting Association Meeting, 2008 International Accounting Section Midyear Conference, University of British Columbia, University of Colorado at Boulder, Columbia Business School, Cornell, and University of Florida for helpful comments and suggestions. We gratefully acknowledge HakJoon Song who provided research assistance. ** Merves Scholar, Fox School of Business and Management, Temple University, 453 Alter Hall, 1801 Liacouras Walk, Philadelphia, PA 19122. Ph: (215) 204-6422, Fax: (215) 204-5587. E-mail: egordon@temple.edu. ***Asociate Professor, University of Colorado at Boulder, 419 UCB, Boulder, CO 80309-0419. Ph: (303) 735-5027. Email: bjorn.jorgensen@colorado.edu. **** Associate Professor of Accounting, Department of Accounting, College of Business, University of Texas - San Antonio, San Antonio, TX 78249-0632. Ph: (210) 458-6744. Fax: (210) 458-4322. E-mail: cheryl.Linthicum@utsa.edu.

Could IFRS Replace US GAAP?

A Comparison of Earnings Attributes and Informativeness in the US Market

We compare accounting-based and market-based earnings attributes under IFRS and US

GAAP for a sample of US-listed, IFRS reporting firms in fiscal 2004 through 2006. Our sample

and research design provide a powerful setting to compare IFRS and US GAAP. For each

earnings attribute, we calculate two separate cross-sectional analyses, one for US GAAP and

another for IFRS. As each firm appears twice, once in each cross-sectional analysis, our analysis

can therefore be thought of as if it were a comparison of matched pairs where the firm is used as

its own control. As US-listed, the accounting and reporting of these firms was subject to similar

US regulatory scrutiny, which provides a more controlled setting to examine differences in

reporting under IFRS and US GAAP. Our evidence suggests that earnings under US GAAP and

IFRS are of comparable quality. Nonetheless, US GAAP-reconciled earnings is incrementally

relevant and informative in this period, suggesting that discontinuing reconciliation of IFRS to

US GAAP result in less useful financial statements for valuation. When we extend our

comparison to examine the relationships between home country incentives and the two sets of

accounting standards, we find differences between reporting under IFRS and US GAAP are

mainly found in a few specific earnings attributes in code law countries, where auditors are

strong, analysts following is high, and investor protection low.

1

Could IFRS Replace US GAAP? A Comparison of Earnings Attributes and Informativeness in the US Market

1. Introduction

Given that more than 100 countries have adopted, or are in the process of adopting International

Financial Reporting Standards (IFRS), it is natural that U.S. regulators deliberate whether to

permit or require US firms to prepare primary financial statements using IFRS (SEC 2008, SEC

2007a).

We appreciate that the U.S. public capital market has not experienced the co-existence of two sets of accounting standards for use by U.S. issuers. The Commission is issuing this Concept Release to gather input on the potential significance and effect of any such change to investors, issuers and market participants as well as to the accounting profession in general. (SEC 2007a, p. 7.)

With respect to the statement above, we are interested in the “potential significance of

IFRS versus US GAAP reporting to investors, issuers and market participants”. Specifically, we

examine the quality of IFRS reported earnings using metrics adapted from the academic

literature. As explained in the SEC’s quote above, US domiciled firms that are listed in the US

are currently required to report under US GAAP. We effectively employ the set of non-US. firms

that report both IFRS and US GAAP earnings and shareholders’ equity in the same year, in the

same (US) market to assess the two standards.

We address three related research questions using a sample of foreign firms that use IFRS

for their financial statements in their home country and files US GAAP reconciliations with the

SEC. First, are IFRS earnings of higher or lower quality than US GAAP earnings? Second, are

IFRS earnings more or less informative than US GAAP earnings? Third, we explore whether the

comparison of earnings attributes under the two sets of rules are sensitive to home country

incentives, institutions, and regulatory factors in the spirit of Ball et. al., (2000, 2003).

2

Our first research question acknowledges that the relevant measure of earnings quality

depends on the intended (or presumed) use of financial accounting information such as for

valuation, contracting or other reasons. Rather than limiting our assessment of quality to

valuation, we operationalize earnings quality using a host of measures of earnings attributes that

have generally been associated with quality of financial reporting (Francis et al. 2004). We

document that US GAAP and IFRS are not distinguishable using earnings attributes most

commonly evaluated in the accounting literature. This finding is consistent with allowing US

firms to report under IFRS or US GAAP.

In response to our second research question, we find that neither US GAAP nor IFRS

exhibit relative value relevance over the other. For our sample period, however, US GAAP

reconciliation remains incrementally informative. Our second research question should be

interpreted with caution since it does not consider alternative non-market uses of accounting

information that should also be relevant in regulatory deliberations.

Regarding our third research question, we consider whether the financial reporting

practices, under IFRS or US GAAP, are sensitive to the home country incentives of preparers.

While our sample consists only of the population IFRS reporting, non-US, US listed firms, we

consider incentives, institutions, and regulatory factors in the firms’ home country jurisdictions.

With the adoption of IFRS, we find differences in earnings attributes under US GAAP and IFRS

are few and have decreased from prior home GAAP, but continue to exist in code law countries,

where auditors are strong, analysts following is high, and investor protection low.

Our sample of all foreign firms that list in the US and report under IFRS is clearly self-

selected. These firms chose to list in the US, creating a selection problem that might lead to

biased inferences, see Ashbaugh (2000) and Lang, Raedy and Yetman (2006). Nevertheless,

3

these firms are more likely to be similar to US firms in terms of product market interactions and

investor clientele, see Khanna, Palepu, and Srinivasan (2004) and Bradshaw, Bushee and Miller

(2004), suggesting that they might be the most relevant group of firms from which to draw

inferences applicable to the general population of publicly-traded US firms for which the SEC is

currently deliberating whether to permit or require the use of IFRS.

Our paper makes both conceptual and methodological contributions. First, our research

design exploits an international setting where one common regulatory body, the SEC,

announced, well in advance of filing deadlines, that it would carefully review the filings of

financial accounting information under two accounting standards (Nicolaisen 2005). Thus fiscal

year ends 2005 and 2006 offer a one-time setting where U.S.-listed firms provide a matched pair

of IFRS and US GAAP numbers for comparing earnings attributes of IFRS and US GAAP in a

single regulatory environment. Second, unlike other international studies, we control for cash

flows, consistent with prior literature using US data. Third, based on the FASB and IASB

Conceptual Frameworks, we supplement the list of earnings attributes commonly associated with

earnings quality by proposing cash flow predictability and cash flow persistence as additional

earnings attributes. Fourth, as we test for incremental value relevance, we employ the effect of

accounting standards on both the income statement and the balance sheet. We provide a detailed

discussion of these contributions in sequence below.

SEC review process

Prior international accounting research documents that, in addition to accounting

differences, legal institutions, enforcement, capital structure, and market demand affect the use

4

of accounting information.1 This complicates the evaluation of international comparisons of

earnings prepared under different accounting standards. In our setting, this concern that

regulatory regimes are correlated with earnings quality is mitigated because all sample firms are

registered with the SEC and the SEC made it known, well in advance, that 2005 and 2006 fiscal

year-end IFRS filings would be subject to review (Nicolaisen 2005).

If the SEC review is of as high, or higher, quality than the review that foreign regulators

in, say Mexico2 otherwise would have done, then all firms in our sample are subject to the same

level of scrutiny regarding their financial reporting under both US GAAP and IFRS. If instead

some foreign regulator, say Switzerland, applies a more stringent regulatory review, then the

filings subject to the more stringent review could be viewed as more credible. The effect of

either higher or lower level of enforcement from a foreign regulator is further mitigated in our

research setting because we use the firm as its own control when we compare earnings prepared

under IFRS and earnings prepared under US GAAP. Nonetheless, foreign regulators are more

likely to review IFRS earnings than US GAAP reconciliations which could a priori induce a bias

towards detecting higher quality of IFRS earnings than US GAAP earnings, but that runs counter

to our findings. In summary, the SEC review should result in less cross-country variation due to

differences in the level of enforcement.

Since U.S. regulators base decisions on cost-benefit analyses that consider many diverse

financial statement users, our analyses employ a large number of different earnings attributes.

For each of our earnings attributes, we calculate these attributes in two separate cross-sectional

analyses, one for US GAAP and another for IFRS. Each firm appears twice, once in each cross-

sectional analysis. Our analysis can therefore be thought of as if it were a comparison of matched

1 See Bhattacharya et al. (2000), Holthausen and Watts (2001), Leuz, Nanda, Wysocki (2003) and Ball and Shivakumar (2005), among others. 2 See Bhattacharya et al. (2000).

5

pairs where the firm is used as its own control as more than one set of accounting data is

provided for the same firm, in the same year, given the same set of underlying transactions.

Cash flow and value relevance

While the U.S. literature on incremental value relevance separates earnings into cash flow

and accruals, prior international studies have failed to do this. Consistent with prior literature

using US data, we control for the presence of cash flows. Because we control for the information

content of cash flows, we are less likely to find incremental information content in U.S. GAAP

as compared with IFRS, and vice versa. Current IASB and FASB conceptual frameworks cite

cash flows as central to the objectives of financial reporting (FASB 1978, IASB 2001).3

Corresponding to these objectives, we introduce two related earnings attributes: predictability

and persistence of cash flows.

Incremental value relevance tests

Our incremental value relevance tests consider the simultaneous effect on the income

statement and on the balance sheet. Under International Financial Reporting Standard no. 1

(IFRS 1), first-time adopters of IFRS must reconcile transition net income and shareholders’

equity from the previous basis of accounting to IFRS. For example, a U.K. company that adopted

IFRS in 2005 would provide 2004 and 2005 net income reconciled from UK GAAP to IFRS, and

reconcile beginning 2004 shareholders’ equity from UK GAAP to IFRS. The effect of adoption

has a current year effect on the income statement and a cumulative effect on the balance sheet.

3 “Financial reporting should provide information to help present and potential investors and creditors and other users in assessing the amounts, timing, and uncertainty of prospective cash receipts from dividends or interest and the proceeds from the sale, redemption, or maturity of securities or loans. Since investors' and creditors' cash flows are related to enterprise cash flows, financial reporting should provide information to help investors, creditors, and others assess the amounts, timing, and uncertainty of prospective net cash inflows to the related enterprise.” (FASB 1973, p. 25).

6

The effect on the balance sheet is the carryover year effect from previous years and the

cumulative prospective effect of items such as asset revaluations.

Prior value relevance studies regress either stock price or stock returns on unexpected

accounting variables. Following Kothari and Zimmerman (1995), we use returns as the

dependent variable. One framework for value relevance studies is the Edwards-Bell-Ohlson

argument that, under clean surplus accounting, market prices are based on book value and

earnings. It follows immediately that cumulative dividends are based on levels of earnings and

changes in earnings relative to the previous year. This holds true for any single accounting

standard, either US GAAP or IFRS. However, all firms in our sample disclose two sets of

earnings and book value. In this setting, it is reasonable to think of investors employing a two-

stage heuristic. First, investors estimate market value under both US GAAP and IFRS separately.

Second, investors weigh the two value estimates according to their assessment of various metrics

regarding earnings attributes. Consequently, our value relevance tests include levels of cash

flows and levels of accruals and changes of cash flows and changes of accruals. In our

subsequent tests of the incremental value, relevance of US GAAP over IFRS information employ

both levels and changes for cash flows, IFRS accruals, and the difference between US GAAP

and IFRS.

The results of our study should be of interest to regulators, standard-setters, and US

companies. As noted above, the SEC has specifically called for input on the comparability of the

two standards. As standard-setters like the IASB or FASB move toward convergence, better

understanding the similarities and differences in the earnings attributes of IFRS and US GAAP

should be of use in decision-making. Finally, this paper has the potential to provide US

7

companies information regarding the quality of US GAAP versus IFRS earnings should those

firms be given the choice, or someday be required, to adopt IFRS.

The rest of the paper is organized as follows: Section 2 discusses related literature.

Section 3 describes the research design and develops hypotheses. In Section 4, we discuss

sample selection and description. In Section 5, we present results of earnings attributes tests and

sensitivity tests. Section 6 discusses incentives and instructional factors, and presents rules of

this analyses. Section 7 concludes.

2. Related Literature

2.1 Earnings Attibutes, 20-F Reconciliations and Accounting Numbers Derived under Alternative Accounting Standards

Prior to the wide scale adoption of IFRS, researchers have examined relative information

content of accounting numbers derived under alternative accounting standards. A body of

literature has focused on reconciliations from non-US to US GAAP (SEC Form 20-F, Item 17 or

18). While our study is, in part, designed to provide insight for regulators who may ultimately

permit or mandate the use of either IFRS or US GAAP, the literature provides an historical view

of the value relevance of reconciliations. Pownall and Schipper (1999) discuss the body of

research using Form 20-F reconciliation data, observing that prior research documents

differences between US GAAP and both non-US GAAPs and IAS, and offers some evidence that

the differences are value-relevant. For example, Amir, Harris, and Venuti (1993) investigate the

value relevance of reconciling items between domestic and US GAAP earnings and shareholders'

equity using a sample of 101 cross-listed companies during the period 1981-1991. Their results

suggest that the reconciliations are value relevant, both in aggregate and for some specific

components (property revaluations and capitalized goodwill). Harris and Muller (1997), examine

8

only reconciliations between US GAAP and IAS for 31 companies from 1992 to 1996, provide

inconclusive evidence of the usefulness of the reconciliations. They find that US GAAP earnings

reconciliation is value relevant and US GAAP is associated more highly with market measures

after controlling for IAS amounts in certain models (market value and returns) but not all models

(per-share).

Taking a different approach, studies that are more recent investigate the properties of

accounting measures for US cross-listed companies. Land and Lang (2002) compare US cross-

listed to non-cross-listed companies using a sample from 1990 through 2001, and find that cross-

listed companies exhibit less earnings smoothing, more timely loss recognition, and more value-

relevance than non-cross-listed companies. Complementing this study, Lang, Raedy and Wilson

(2006) compare US cross-listed companies to US companies. They find that US GAAP

accounting measures of cross-listed firms differ from those of US firms in terms of the time-

series properties of earnings and accruals, and the degree of association between accounting data

and share prices. Taken together, these studies suggest that although the properties of US cross-

listed companies differ from those of non-cross-listed companies, differences in the reporting of

US cross-listed companies and US companies exist even with the reconciliations. As an example,

evidence from these studies suggests that cross-listed firms engage in less earnings management

than non-cross-listed firms, even though they manage earnings more than US firms do. While

these studies compare earnings across firms, our study focuses on earnings attributes resulting

from differences in IFRS and US GAAP accounting standards, rather than across firm types.

Next, we consider the results of studies that compare IAS to US GAAP.

Research focusing on the properties of accounting information under IAS generally

suggests that IAS reporting produces accounting measures of higher quality when compared to

9

domestic GAAP, but not US GAAP. Using a sample of 319 IAS reporting companies from 1990

to 2003, Barth, Landsman, and Lang (2006) show that companies using IAS exhibit less earnings

smoothing, more timely loss recognition, and more value-relevance than those applying domestic

(non-US) GAAP. Ashbaugh and Pincus (2001) find that the analysts forecast error of companies

using IAS are smaller than those using domestic GAAP. However, comparing IAS to US GAAP

companies using a sample 428 IAS reporters from 1990 through 2004, Barth, Landsman, Lang,

and Williams (2006) find that IAS firms exhibit more earnings smoothing, more timely loss

recognition, and a lower association between accounting amounts and share price. For the sub

sample of firms that are cross-listed, they observe similar reporting quality for IAS and US

GAAP measures.

In the context of the current policy debate, the evidence these studies offer is somewhat

limited. First, most companies in these studies adopted IAS voluntarily, producing a self-

selection bias. Prior research shows that profitable, growing companies are more likely to adopt

IFRS.4 Inferences from such a sample are likely to be less applicable to a population of

companies that are mandated to follow the standards. Second, several studies investigate the

reconciliations from domestic GAAP but these are not being reconsidered by the SEC and

domestic (non-IFRS) GAAP is no longer the primary focus in those countries now requiring

IFRS. Third, since IAS (now IFRS) were revised substantially in the early 2000’s, results from

pre-2002 studies may or may not be generalizable to the present. Consequently, the results of

past studies cannot sufficiently satisfy the demand for evidence relevant to the current debates of

whether or not to permit or require US companies to use IFRS instead of US GAAP..

4 International Financial Reporting Standards (IFRS) consist of the body of accounting standards issued by the IASB. The term IFRS is used throughout this paper to refer to the body of standards issued by the International Accounting Standards Board, and those in-force International Accounting Standards (IASs) issued by the IASB’s predecessor, the International Accounting Standards Committee.

10

Additionally, the methodology employed in the international studies often ignores cash

flows. This is especially evident when the cross-country literature is compared with US,

“domestic” information content literature. Ignoring cash flows may be the result of lack of data.

Cash flow data have not been required in all home country jurisdictions, as it is under IFRS.

Nevertheless, for the most part, prior studies ask: “Are (US-based) earnings incrementally

informative above and beyond (non-US) earnings?” and, for the most part, ignore cash flows.

Finally, at a conceptual level, we consider that either IFRS or US GAAP provide higher

information content. That is, we do not assume, a priori, that US GAAP will provide incremental

information relative to IFRS as appears to be the case in prior studies of US GAAP

reconciliations. Our research questions are explored in light of the current regulatory questions.

Consequently, we structure our tests to allow for either US GAAP or IFRS to emerge as the more

value-relevant. Absent conditions where one can empirically test the question “How do investors

interpret IFRS reported by US registrants without a US GAAP reconciliation?”, we provide

indirect evidence regarding the relative information content over a brief window where two sets

of earnings are provided for the same firm, in the same year, in the same regulatory environment.

2.2 Incentives and Accounting Standards: Institutional Factors

It is well-established that there is variation in the reported accounting quality across firms

due to incentives, enforcement institutions, and regulatory factors, among others. Ball, Kothari

and Robin (2000), Ball, Robin and Wu (2003) and Burgstahler, Hail and Leuz (2006) document

that this variation persists within environments with the same accounting standards. One reason

is that accounting standards allow managers to exercise financial reporting discretion as they

prepare the financial statements. Our initial analysis does not rule out the possibility of firm level

11

variation in earnings quality but rather presumes that this variation has the same effect on

earnings reported under IFRS and US GAAP for our firms.

This presumption is not completely unreasonable for a number of reasons. First, each

firm is used as its own control. Second, the SEC undertook a formal review of both IFRS and US

GAAP filings of most of our sample period. To the extent that the SEC review generally subjects

firms to more scrutiny and higher level of enforcement than they would have been subject to

from their home country regulator, the SEC review should create a level playing field in terms of

high enforcement level under both sets of accounting standards’ earnings. The higher level of

SEC enforcement would tend to mute the room for a role of other factors, such as, reporting

incentives and enforcement on reported earnings. Nevertheless, it is quite possible that that these

other factors affect IFRS earnings differently than US GAAP earnings. For example, the use of

earnings for compensation is typically based on earnings as reported under IFRS. Consequently,

we proceed to investigate several factors that may differentially affect the reported earnings’

attributes under these two current accounting standards, US GAAP and IFRS. To this end, we

partition our sample based on whether the country has a code or civil law tradition, the role of

auditors in the country, a country’s analysts following, and the investor protections and

enforcement in a country. These factors are discussed in detail in Section 6.

3. Research Design

In our first set of tests, we examine earnings attributes under IFRS and US GAAP. Our

second set of tests further explores the incremental and relative value relevance of the two

standards. Our third set of tests consider the relative incentive effects for the same firm and time

period, as evaluated in a period of dual accounting standards.

3.1 Earnings Attributes

12

We first discuss the earnings attributes that do not refer to market returns and next

discuss those that do.

3.1.1 Accounting-based Earnings Attributes

While there are multiple earnings attributes in the literature, we first follow Francis,

LaFond, Olsson and Schipper (2004) and describe our implementation of their measures of the

attributes below. We adapt our implementations because we are interested in comparing the

earnings attributes of a given set of firms under two accounting standards, denoted by s below,

whereas Francis et. al. (2004) are interested in constructing firm-specific measures of earnings

attributes. In addition, we face different data limitations.5

Accrual Quality

Francis et. al. (2004) use the accrual quality measure proposed in Dechow and Dichev

(2002). Discretionary current accruals in a period are expected to relate to lagged,

contemporaneous and leading cash receipts and disbursements. The discretionary part of current

accruals is estimated as the residual (ε) from the following cross-sectional regression:

tjstjsstjsstjsstjs vCFOCFOCFOACC ,,1,,3,,,2,1,,1,0,, ++++= −+ ϕϕϕϕ

where ACCs,j,t = firm j’s total accrual calculated as net income less cash flows from operations in

year t under standard s, where s is IFRS or US GAAP, CFOj,t = cash flows from operations of

company j in year t and subscripts j, t, and s denote firm, year, and accounting standard,

respectively. All variables are divided by the firm’s market value at time t.

5 Due to limitations of US GAAP accounting information required in the 20-F and availability, we are not able to obtain the same financial statement information under both IFRS and US GAAP as Francis et. al. (2004) use to calculate some of their attributes. For instance, total current assets are not required to be reported under US GAAP nor is depreciation.

13

Based on the above regressions, we define accrual quality as the standard deviation of the

estimated residuals for each accounting standard, IFRS and US GAAP. We differ from Francis

et. al. (2004) in two respects. First, we note that the dependent variable employed in Dechow

and Dichev (2002) is Working Capital or the firm’s total current accruals measured as (ΔCA –

ΔCL – ΔCash + ΔSTDEBT), where CA is current assets, CL is current liabilities and STDEBT is

debt in current liabilities. In contrast, we use total accruals because measures of working capital

are not uniformly reported under both IFRS and US GAAP. This difference is driven, in part by

the difference in accounting standards, since IFRS defines, but under IAS 1 does not require

separate listing of current assets (IASB 2008b). Second, we scale by total market capitalization

while Francis et. al. scale by the book value of total assets. Two different asset measures (IFRS

and US GAAP) exist for each of our sample firms. Third, we estimate cross-sectional regression

for each accounting standard, s=IFRS, US GAAP, whereas Francis et. al. estimate a time-series

regression for each firm. Parallel to Dechow and Dichev (2002), the standard deviation of the

residuals ( ( )stjs vlityAccrualQua ,,2 ˆσ= ) from equation (1) is our measure of accrual quality for

accounting standard s= IFRS, USGAAP.

Earnings Persistence and Predictability

Persistence is associated with earnings quality because transitory earnings components

are supposed to have been smoothed out, see Penman and Zhang (2002), among others. Our

measures of earnings persistence and earnings predictability are based on the relation between

current and past earnings as follows:

tjstjsssotjs vNINI ,,1,,,1,,, ++= −φφ

14

Where NIs,j,t is firm j’s net income in year t under standard s, where s is IFRS or US GAAP.

Variables are divided by the firm’s market value at time t-1.

The estimated coefficient, φ 1,s, measures earnings persistence. The larger (smaller)

values of persistence relate to more (less) persistent earnings. The standard deviation of the

residuals from the equation is interpreted as earnings predictability. Large (small) values of

predictability suggest less (more) predictable earnings.

Francis et. al. (2004) estimate (2) as a time-series regression for each firm and find firm-

specific measures of earnings persistence that vary across firms in their sample. Our earnings

persistence captures only average under any accounting standard. Despite the firm-level

variation in earnings persistence documented in FLOS, our cross-sectional measure is

comparable because our firms reported under both accounting standards and therefore any

difference in average earnings persistence should be due to accounting standards.

Cash Persistence and Predictability

Given that both standards purport to create measures of income that better predict future

cash flows, we generate measures of cash flow persistence and predictability similar to earnings

persistence and predictability. Our measures of cash flow persistence and cash flow

predictability are also based on the relation between current cash flows, past earnings where past

earnings is separated into past cash flows and past accruals as follows:

stjstjsstjssstj vACCCFOCFO ,,,1,,2,1,,1,0,, +++= −− φφφ

where CFOs,j,t is firm j’s net income in year t under standard s, where s is IFRS or US GAAP.

All variables are divided by the firm’s market value at time t -1.

15

The regression assesses the ability of accruals (under the different standards) to aid in the

prediction of current cash flows, controlling for past cash flows. Similar to the earnings-based

measures, the estimated coefficient, φ 1,s, measures cash flow persistence and the standard

deviation of the residuals from the equation is interpreted as cash flow predictability.

Earnings smoothness

Several prior papers – including Trueman and Titman (1988) and Tucker and Zarowin

(2006) – suggest that smoothness is a desirable earnings attribute. Following Leuz, Nanda, and

Wysocki (2003) and Francis et al. (2004), we measure earnings smoothness relative to that of

cash flow from operations. Smoothness is defined as the ratio of the standard deviation of

income before extraordinary items to the standard deviation of cash flows from operations. We

measure smoothness as the cross-sectional measure of earnings smoothness for each accounting

standard. The measure of smoothness is computed as the ratio of the standard deviation of all

company’s net income under either IFRS or US GAAP divided by beginning market values over

the standard deviation of all company’s cash flows from operations divided by beginning market

values. Smoothnesss = σ(NIs,t)/σ(CFOt).

3.1.2 Market-based Earnings Attributes

Timeliness and Conservatism

Watts (2003a, 2003b) argues that conservatism in earnings is a desirable property. Kim

and Kross (2005) suggest that increasing accounting conservatism plays a role in the greater

ability of earnings to predict future cash flows. The FASB/IASB conceptual framework project

16

report lists one of the objectives of financial reporting as meeting capital providers’ interest in

assessing “the entities ability to generate net cash inflows” (FASB-IASB 2008).

Following Basu (1997), Pope and Walker (1999), Givoly and Hayn (2000) and Francis et.

al. (2004), among others, we use the following regression to obtain measures of timeliness and

conservatism.

tjstjtjsstjstjssstjs RETNEGRETNEGNI ,,,,,,2,,1,,,1,0,, ωββαα +•+++=

where NEG is an indicator variable equally one is RET is negative and zero otherwise. NI is

divided by the firm’s market value at time t -1.

Similar to other earnings attributes, the above equation is estimated separately for each standard,

s. Our measure of timeliness is the negative adjusted R2 from the above regression. Our

measure of Conservatism is the ratio of the coefficient on bad news to the coefficient on good

news = ( ) sss ,1,2,1 / βββ + . Larger values of timeliness (conservatism) imply more timely

(conservative) earnings.

Relevance

The final earnings attribute we examine is value Relevance. Following FLOS, we

measure value relevance as the negative adjusted R2 from the following regression:

jtstjsstjssstj eNINIRET ,,,,2,,,1,0, +Δ++= γγγ .

Large (small) values of Relevance suggest less (more) value relevant earnings. We further

explore and discuss value Relevance in the next section. NI and ΔNI are divided by the firm’s

market value at time t -1.

3.2 Incremental Value Relevance

17

Our next set of tests further explores the Relevance of each standard. Incremental value

relevance tests consider the simultaneous effect on the income statement and on the balance

sheet. As is well-known, adoption of a new standard has a current year effect on the income

statement and, possibly, a cumulative effect on the balance sheet. The effect on the balance

sheet is the cumulative year effect from previous years and the cumulative prospective effect of

say asset revaluations.

One analytical starting point for value relevance studies is the Edwards-Bell-Ohlson

argument that, under clean surplus accounting, market prices are based on book value and

earnings. It follows immediately that cumulative dividends are based on levels of earnings and

changes in earnings relative to the previous year. This holds true for any single accounting

standard, either US GAAP or IFRS. However, all firms in our sample disclose two sets of

earnings and book value of equity. In this setting, it is reasonable to think of investors

employing a two stage heuristic. First, investors estimating market value under both US GAAP

and IFRS separately. Second, investors weigh the two value estimates according to their

assessment of various metrics regarding earnings attributes. Prior studies on value relevance

report regressions of stock returns on unexpected cash flows and unexpected accruals. Because

of the argument in the previous paragraph, our initial value relevance tests include levels of cash

flows and levels of accruals and changes of cash flows and changes of accruals. Our subsequent

tests of the incremental value relevance of US GAAP over IFRS information employs both levels

and changes for cash flows, IFRS accruals, and the difference between US GAAP and IFRS.

Prior accounting studies, including Rayburn (1986), report that accounting accruals have

incremental information content above and beyond cash flows from operations in the US. An

accrual model provides insights beyond an (earnings) model, in that each accrual measure is

18

incrementally informative to operating cash flows. While the earnings relevance regression can

show that each earnings measure is significantly related to returns, one could not rule out the

possibility that operating cash flows could be driving results. To investigate this hypothesis in

our sample, we regress stock returns on both cash flows and accruals under each accounting

standard:

tjstjsstjsstjstjsstj eACCACCCFOCFORET ,,,,,4,,,3,,2,,1,0, +Δ++Δ++= γγγγγ (1)

where RETj,t is firm j’s 12-month return ending the month the current year’s, t, Form 20-F is

filed with the SEC.

Large (small) values of Relevance suggest more (less) value relevant earnings. As is

common, earnings under accounting standard s have incremental information content over cash

flows from operations when the null hypothesis that 0,4,3 == ss γγ .

The reconciliation difference between IFRS earnings and US GAAP earnings can also be

defined as the differences between IFRS accruals and US GAAP accruals:

tjIFRStjGAAPUStjIFRStjGAAPUStj ACCACCNINIDiff ,,,,,,,,, −=−= (2)

because cash flows from operations are assumed independent of the accounting standard.

Estimating the following regression allows a test the null hypothesis that 0,6,5 == ss γγ

which would imply that earnings prepared under accounting standard s have no incremental

information content above and beyond earnings prepared under the other standard. These tests

provide evidence of whether differences between IFRS and US GAAP are informative.

tjstjsstjss

tjsstjsstjstjsstj

eDiffDiff

ACCACCCFOCFORET

,,,,,6,,,5

,,,4,,,3,,2,,1,0,

+Δ++

Δ++Δ++=

γγ

γγγγγ (3)

3.3 Relative Value Relevance

19

In our final tests, we examine the relative informativeness of difference in accruals

between US GAAP and IFRS earnings. We compare the explanatory power of estimating

regression (1) twice-once under IFRS and once under US GAAP. We follow Dechow (1994)

and compare the explanatory power of the goodness of fits associated with these two regressions,

2IFRSR and 2

GAAPUSR , using a Vuong test.

4. Sample Selection and Descriptive Statistics

Our sample includes firms who filed with the U.S. Securities and Exchange Commission

using IFRS, with 20-F reporting dates from 1/1/06 to 8/1/08. Since 20-F filers have until six

months after their fiscal year end to file with the SEC6, and some file for an extension to file,

12/31/05 year end filers are expected to file no later than 7/30/06. In determining whether or not

a firm is an IFRS filer, we examined the audit report and accounting policy footnote for each 20-

F filed with the SEC during this time.

We collect fiscal year 2005 IFRS earnings, US GAAP earnings and operating cash flows

from fiscal year 2005 20-F filings. Operating cash flows and IFRS earnings for 2004 are also

collected from fiscal 2005 20-F filings. Market data for the US stock market are collected from

Compustat.

Insert Table 1

Table 1 provides descriptive statistics firms in this study. Panel A of Table 1 describes

the sample for which we have on market values and stock returns of 161 firms from 26 countries.

For comparison, we converted the data into a common currency, U.S. Dollars. Firms in the 13

countries represented in the European Union and in Australia and Norway were required to adopt

6 Beginning in 2009, 20-F filers must file 20-Fs within 4 months of year end. Release Nos. 33-8900; 34-57409; International Series Release No. 1308; File No. S7-05-08. A copy of the release is available on the SEC's website at www.sec.gov/rules/final/2008/33-8959.pdf.

20

IFRS no later than fiscal years starting January 1, 2005.

Panel B of Table 1 reports the means, standard deviations, and medians for our

accounting variables under IFRS and US GAAP scaled by the beginning of year market values

(the scalar is the same under both accounting standards for each firm-year pair of observations).

The average (median) net income is 0.121 (0.0869) under IFRS and 0.105 (0.0760) under US

GAAP. As expected, cash flows from operations exhibit statistically significantly higher

variance than net income calculated under either IFRS or US GAAP with p-values of 0.00

(untabulated). This arises because cash flows from operations and accruals as determined under

the two sets of accounting standards are highly negatively correlated as reported in Table 2

below, which is consistent with accruals inducing smoothing of earnings prepared on an accrual

basis relative to the benchmark of cash basis of accounting.

Insert Table 2

The mean (median) net income under IFRS exceeds US GAAP by 0.016 (0.006). This

difference represents the reconciliation difference reported in 20-F.7 We can reject the null

hypothesis that the difference between IFRS and US GAAP net income is significantly different

from zero with a p-value of 0.048 (untabulated). IFRS is less conservative for our sample of

firms. The difference in IFRS and US GAAP earnings represents the difference in reported

accruals defined as the difference between IFRS earnings and (IFRS) cash flows. To the extent

that cash flows are independent of the accounting standard for a given reporting entity, the

reconciliation difference between earnings reported under US GAAP and IFRS mainly

represents the difference between current accruals under these two accounting standards.

Nevertheless, the difference in accruals that would have been reported under IFRS and US

7 The Appendix describes accounting differences by category between IFRS and US GAAP net income reported by companies in our sample. We identify 19 main categories of differences, showing that most of these can be positive (increasing IFRS net income relative to US GAAP) or negative.

21

GAAP may also represent differences between the accounting standards’ definition of the

reporting entity and consolidation rules. Specifically, IFRS allows proportional consolidation,

while US GAAP does not. To avoid capturing differences in the accounting for minority-

interests, we carefully analyzed minority interests for each company and took care to measure

earnings as net income excluding minority interests. While not reported, we also tabulated

income under Home GAAP used prior to IFRS adoption and compared their means. Consistent

with the much discussed convergence of accounting standards as a consequence of IFRS

adoption, we observe a statistically significantly larger difference between home GAAP and US

GAAP net income relative to the difference between IFRS and US GAAP income,

Table 2 reports correlations between our variables. As expected, we observe high

correlations between net incomes reported under US GAAP and IFRS. However, there is low

correlation between operating cash flows and annual market returns and the remaining variables.

Barth, Landsman, and Lang (2006), among others, interpret the correlation between cash flows

from operations and accruals (both scaled) as an alternative indicator of smoothness. In our

sample, the correlation between cash flows from operations and accruals are -0.714 and -0.703

under IFRS and US GAAP, respectively.

5. Empirical Analysis and Results

5.1 Earnings Attributes

5.1.1 Accounting-based Earnings Attributes

Table 3 compares six accounting-based earnings attributes under IFRS and US GAAP.

The four earnings attributes of Earnings Persistence, Cash Persistence, Cash Predictability, and

Smoothness appear more favorable for US GAAP, while the results on, Accrual Quality, and

22

Earnings Predictability imply that IFRS is more favorable. Our statistical tests show that none of

these differs significantly between IFRS and US GAAP.8

Insert Table 3

5.1.2 Market-based Earnings Attributes

Table 4 compares three market-based earnings attributes under IFRS and US GAAP.

Timeliness and conservatism appear greater under US GAAP, statistical tests show they are not

significantly different.9The relevance of earnings is significantly greater under US GAAP (p-

value = 0.005).

Insert Table 4

5.2 Incremental and Relative Value Relevance

Table 5 reports the results of our incremental value relevance tests. As a benchmark, we

initially report the results of a returns regression on the levels and changes in cash flows from

operations denoted by CFO and ΔCFO, respectively. This regression specification follows from

Easton and Harris (1991) under the cash basis of accounting. As expected, we find that cash

flows from operations are significantly related to returns. The adjusted R2 of the cash basis

regression is 3.5%. Given this benchmark, we next address the incremental value relevance from

investors observing – in addition to cash flows – earnings under each accounting standard.10

First, IFRS accruals levels and changes significantly contribute to the explanatory power with an

increase in the adjusted R2 to 4.9%. Similarly, US GAAP accruals levels and changes add the

explanatory power of the regression beyond cash flows with an increase in the adjusted R2 to

8 We use standard t-tests to test the equality of the variances following attributes: Accrual Quality, Earnings Predictability, and Cash Predictability. We use bootstrapping techniques to test the equality of the others: Earnings Persistence, Cash Persistence, and Smoothness. 9 We use the Vuong test to examine the equality of Relevance, and bootstrapping techniques to test Conservatism and Timeliness. 10 The number of observations in each regression varies slightly due to the deletion of regression outliers, identified as those observations with the absolute value of the residual exceeding three.

23

10.0%. Thus, the accrual basis of accounting is incrementally informative relative to the

benchmark cash basis of accounting under both IFRS and US GAAP. To ensure that our

comparisons are not driven by differences in cross-sectional variation in the dependent variables,

we follow a suggestion by Lo and Lys (2000) and reran all analyses reported in Table 5 for a

fixed common sample.11

Table 5 also reports two results regarding the incremental value relevance of

reconciliation between US GAAP earnings and IFRS earnings. First, we take the US perspective

and questions whether reconciliation from IFRS to US GAAP is incrementally informative to

IFRS earnings and cash flows. Following Easton and Harris (1991), we again include all

accounting variables in both levels and changes. We find that reconciliations from IFRS to US

GAAP – the differences between IFRS Accruals and US GAAP Accruals – are incrementally

informative to IFRS. Second, we take the non-US perspective that questions whether

reconciliation from US GAAP to IFRS is incrementally informative to US GAAP earnings and

cash flow from operations. This perspective is relevant for whether non-US regulators should

allow home firms that are cross-listed in the US to continue to exclusively report under US

GAAP, which has been the practice on the German exchange, Neue Markt (Leuz 2003). In

addition, this perspective might be relevant for the European Union considering whether to

require that US firms listed in the EU be required to reconcile from US GAAP to IFRS. We find

no evidence that reconciliations from US GAAP to IFRS – the differences between IFRS

Accruals and US GAAP Accruals – are incrementally informative to IFRS. From a capital

market perspective alone, this result provides no basis for concern for foreign regulators

regarding US GAAP reporting companies relative to IFRS reporting companies.

11 Rayburn (1986) conducts similar analysis, excluding changes in earnings and changes in cash flows from operations. We conduct similar analysis and find similar (untabulated) results.

24

Insert Table 5

The above results from Table 5 concerned incremental information content of IFRS and

US GAAP and tested whether additional information is “better” for investors. In Table 5, panel

C, we report the results on the relative information content of IFRS and US GAAP which are

designed to address the question as to which of two mutually exclusive accounting standards is

preferable. Following Dechow (1994), we test relative information content by comparing the

adjusted R2s of the two non-nested regressions in Table 5. Using Vuong tests, the adjusted R2s

are significantly different. For non-U.S. firms reporting both IFRS and US GAAP net income,

U.S. GAAP net income provides relatively more explanatory power than IFRS. (p value =

0.015).

5.3 Sensitivity Analyses

5.3.1 Analyses of Sample and Alternative Research Design Choices

We also applied a battery of sensitivity analyses to the pooled regression analysis

reported in Table 3, 4, and 5. Our sample only includes firms for the fiscal years 2004 through

2006. Based on the findings of Daske et al. (2007, 2008) we recognize that early IFRS (pre-

2005) adopters might be different from last minute adopters (2005). We conduct three additional

(untabulated) analyses. First, we exclude the pre-2005 adopters, second we exclude the first time

adoption year from our sample. A second analysis excludes the transition year, 2004.12 The third

specification provides evidence as to whether our results are influenced by adoption-year effects.

For example, one-time adjustments under IFRS 1 may result in accounting adjustments to

income statements and to a lesser extent, balance sheets in the first year. Furthermore, the

preparers might not have much experience with IFRS in the adoption year as in future years. In 12 In this sensitivity check, we estimate Accruals Quality using two rather than three years of data.

25

addition, US regulators might spend more, or less, resources on monitoring the accounting

numbers in the adoption year. Finally, financial statement users might gain experience with IFRS

over time leading to lesser value relevance in the adoption year. Our results are robust to these

sensitivity tests.

We also collect fiscal 2007 IFRS data for the sample companies. Because the 20-F

reconciliation between IFRS and US GAAP earnings and shareholders’ equity was effectively

discontinued for fiscal 2007, we do not have US earnings. Overall, our results are robust to

inclusion of the additional year of IFRS data.

Because observations (firms) from the UK compose 28.3% (29.8%) of our sample, we

perform all analyses excluding the UK. Our results are robust to its exclusion.

We conduct a sensitivity check for the smoothing measure reported in Table 3.

Smoothness is reported in Table 3 based on the aggregate US and IFRS samples. In addition to

this test, we calculated smoothness using a firm-specific measure. For companies in our sample

for three years, we divide the standard deviation of net income over the three years by the

standard deviation of cash flows from operations over the three years to obtain a smoothness

measure by company. Tests indicate that the firm-specific smoothing measure under US GAAP

is not significantly different than that under IFRS.

For the pooled regression analysis reported in Table 5, sensitivity tests provide

information about the possibility of clustering in our data when we use the same firm in more

than one year. The presence of clustering may result in overstated t-statistics that, in turn, would

more likely provide evidence to reject the null hypotheses. We conducted our analyses using

each firm only once and found qualitatively similar results.

26

5.3.2 Pre-IFRS Benchmark

Given that home-country GAAP has been replaced by IFRS in our sample firms’ home

jurisdictions, home-country GAAP is not of particular interest in our study. Nevertheless, we

consider the period 2002-2003 as a benchmark period, and evaluate home-country versus US

GAAP earnings properties for our set of US cross-listed firms reported under home-country

GAAP (pre-IFRS) and US GAAP in the same period. In particular, we observe whether or not

the variation in earnings qualities that exists post-IFRS may simply represent a carry over of

variation that existed between pre-IFRS home country and US GAAP earnings.

Our untabulated results suggest that the variation observed between IFRS and US GAAP

is not identical to the variation observed between home-country GAAP and IFRS in the pre-IFRS

period. Specifically, although we find no significant differences in accounting-based earnings

attributes when we compare IFRS to US GAAP, we find that significant differences between

home-GAAP and US GAAP in three of the accounting-based earnings attributes: Accrual

Quality, Earnings Predictability and Earnings Predictability (Similar to Tables 3 and 6). When

we compare market-based earnings attributes under home GAAP to US GAAP (similar to Tables

4 and 7) in the IFRS pre-adoption period (fiscal years 2002-2004), we find no significant

differences. After IFRS adoption, we find only Relevance significantly differs between under

IFRS and US GAAP.

When we compare the incremental explanatory power of home GAAP to US GAAP

(similar to Table 5) in the IFRS pre-adoption period (fiscal years 2002-2004), find the difference

in earnings is not incrementally relevant to either GAAP. After IFRS adoption, difference in

earnings is incrementally relevant to IFRS but not US GAAP.

27

When we compare the relative power of home GAAP to US GAAP in the IFRS pre-

adoption period (fiscal years 2002-2004), find the no difference. After IFRS adoption, US GAAP

had great explanatory power.

Overall, we find that some accounting-based earnings attributes differ before adopt of

IFRS but market-based earnings attributes are similar. Under IFRS accounting-based earnings

attributes are similar. Even though prior research suggests that companies will select home

accounting choices similar to those allowed US GAAP, perhaps after adopting IFRS companies

are able to make accounting choices even closer to US GAAP then they were under their home

GAAPs. Differences, though, still exist between IFRS and US GAAP that are informative to the

equity markets.

6. Incentives and Institutional Factors

In this section, we investigate whether incentives and institutional factors differentially

affect the reported earnings’ attributes under these two current accounting standards, US GAAP

and IFRS. We partition our sample based on whether the country has a code or civil law

tradition, the role of auditors in the country, a country’s analysts following, and the investor

protections and enforcement in a country. These factors and results of analyses are discussed

below.

For each incentive examined, we separate countries into terciles. We then compare the

accounting-based and market-based earnings attributes of the highest and lowest terciles of each incentive

variable. The incentives and results of analyses are discussed below. Tables 6 and 7 display

results of our tests designed to assess whether accounting-based and market-based attributes.

Additionally, similar to the sensivity performed on overall results, we investigate whether our

results would have been the same if we had compared the earnings attributes of U.S. GAAP

28

relative to Prior Home GAAP (Results are untabulated). If reporting incentives were the primary

driver of differences between U.S. GAAP and IFRS, then we would expect to find the same

differences when we compare U.S. GAAP to IFRS as we find when we compare U.S. GAAP to

Home GAAP. For parsimony, we also compare and contrast the findings on Prior Home GAAP

with those of IFRS and U.S. GAAP in each section below.

Insert Tables 6 and 7

6.1 System for Determining and Enforcing Accounting Standards: Code versus Civil Law

La Porta et al. (1998) document that accounting disclosure practices and corporate

governance mechanisms vary with countries’ legal origins. Specifically, they find that investor

protection tends to be higher in countries with civil law origin (such as most English speaking

countries) relative to countries with civil law origin (i.e., Roman law countries).13 These

differences suggest that legal institutions, for one, influence the demand for, and supply of,

accounting information. By defining acceptable legal remedies in times of conflict between

contractual parties (e.g., in case of disclosure fraud), the legal institutions and legal environment

provide the framework within which accounting can effectively serve its many roles (see for

example, Bushman and Smith 2001). We would therefore expect that a country’s legal origin

affect the incentives for reporting high quality earnings. We use a dichotomous variable equaling

1 if the country's legal system is civil law and zero if it is code law.

Panel A of Table 6 indicates that only earnings predictability in code law countries is

significantly different under US GAAP and IFRS. Untabulated results comparing Prior Home

GAAP to US GAAP also show statistically significant differences in earnings predictability in

both code and civil law countries. Therefore, the adoption and reporting under IFRS appears to

13 See La Porta et al. (2008) for a recent survey.

29

have reduced the difference in earnings predictability in civil law countries. The differences in

accounting-based earnings attributes that exist between code and civil law countries also existed

under prior Home GAAP, but were stronger and Cash Persistence under prior Home GAAP was

also significantly difference. From panel A of Table 7, we find that the Relevance of US GAAP

is greater than IFRS in code law countries. Untabuluted results indicate no differences between

market-based earnings attributes between prior Home GAAP and US GAAP.

6.2 Auditor

Becker, DeFond, Jiambalvo, and Subramanyam (1988) study the effect of audit quality

on earnings management which affects the quality of reported earnings. The prior literature

documents earnings management due to explicit contracts (such as, executive compensation

based on earnings), implicit contracts (between say the firm and its suppliers), and industry-

specific factors (including firms’ incentives for import relief, see Jones). Managers’ reporting

incentives for earnings management vary across firms with many factors, including internal

corporate governance, prior year accounting decisions, and the costs of earnings management if

revealed. One proxy for the cost of being revealed doing earnings management is the quality of

auditor. Becker et al. (1988) use Big-6 auditor as a proxy for high quality auditors and find that

clients with non Big-6 auditors report larger discretionary accruals.

The evidence on the link between auditor quality and earnings quality is contentious for a

number of reasons. First, it is difficult to appropriately compare clients with big versus small

auditors because clients’ size tends to vary with auditor size. Clients voluntary self-selection of a

larger auditor with deeper pockets is associated with the demand for accounting information and

affects the accounting preparers’ incentives for high quality reporting. Second, Choi, Kim, Liu,

30

and Simunic (2008) provide analytical arguments and empirical evidence that the choice of large

auditor varies with countries’ legal origin.

Nevertheless, we investigate the effect of the quality of auditor moderates the differences

in earnings attributes often associated with earnings quality between IFRS and US GAAP. In

similar vein to Becker (1988), we use Big-4 auditors as a proxy for high quality auditors. Since

auditors face higher legal penalties in the US where they are subject to higher legal liability and

higher regulation, including Sarbanes-Oxley Act of 2002, they are likely to take more care in the

audit of US GAAP earnings. Prior to the initiation of SEC review of foreign filings of IFRS,

financial statement preparers and auditors were likely subject to less regulatory scrutiny. On the

one hand, if the perceived likelihood of getting reprimanded for reporting lower quality non-US

GAAP earnings increased with IFRS reporting, it is likely that any prior differences disappeared

were lowered surrounding the adoption of IFRS. On the other hand, to the extent that experience

matters, financial statement users and auditors may have reported lower quality earnings in our

sample period which includes only the years immediately following IFRS adoption. This could

exacerbate the differences between US GAAP and IFRS relative to the difference between US

GAAP and prior Home GAAP. Our variable for Big4 is the percentage of companies in a

country audited by a Big4 auditor from Hope et. al (2008).

Our results, presented in panel B of Table 6, indicate that the only effect of accounting-

based earnings attributes often associated with audit quality on US GAAP relative to IFRS is

statistically significantly different only for earnings predictability in countries with strong

auditors. This difference, however, was already present when we compared the earnings

predictability under US GAAP relative to prior Home GAAP for both countries with strong and

weak auditors (prior Home GAAP results are untabulated). Hence, IFRS adoption and reporting

31

per se does not appear responsible for the differences in earnings predictability in countries

where audit quality is strong. Rather, our results are consistent with auditor strength affecting the

relative earnings quality of US GAAP for foreign registrants no matter their financial reporting

standard in their home country.

We also find evidence consistent with IFRS adoption and reporting lowering the

differences between earnings attributes under US GAAP relative to home country GAAP. First,

earnings predictability under US GAAP relative to IFRS was not statistically significantly

different while the earnings predictability under US GAAP was statistically significantly higher

than under prior Home GAAP in countries with weak auditors. In addition, we find that US

GAAP leads to higher accrual quality than prior Home GAAP only in countries where auditors

are strong, while this difference did not persist when we compare accrual quality under US

GAAP to IFRS.

6.3 Analyst Following

A large literature investigates the relation between analysts following and firms

disclosure quantity and quality. For example, Lang and Lundholm (1993, 1996) rely of study

causality through lead and lag relationship between analyst following and analysts’ rankings of

firms’ disclosures using (the FAF/AIMR disclosure index). Chang, Khanna, and Palepu (2001)

study how the number of analysts, their forecast accuracy and forecast dispersion relate to the

CIFAR disclosure measure first used in La Porta et al. (1998), yet are unrelated to legal regime.

There are two possible views on the role of analysts that can lead to a circular

relationship between analyst following and earnings quality. Under one view, analysts critical

monitoring of firms’ reported earnings and other disclosures lead to higher quality earnings.

32

Under an alternative view, analyst following creates additional demand which motivates

financial statement preparers to report higher quality. It is difficult to a priori ascertain which

view dominates.

Again, we are interested in the effect of analyst following on the quality of US GAAP

earnings relative to IFRS earnings reported by the same firm in the same fiscal year. Since we

hold fixed the underlying economic conditions, we focus on the interaction between analyst

following and relative earnings quality. Our analysts following variable is the number of analysts

providing an annual earnings forecast per company, averaged across the 30 largest companies in

a country from Chang, Khanna, and Palepu (2000).

Our results, presented in panel C of Table 6, only find a statistically significant difference

in accounting-based earnings attributes between US GAAP and IFRS for two attributes. First,

Earnings Predictability is higher for firms in countries with high analysts following. However,

we also find statistically significant and stronger differences in earnings predictability for

previous years between US GAAP and prior Home GAAP. Consequently, our results do not lend

support to the thesis that the adoption and reporting under IFRS explain the differences in

earnings predictability under US GAAP relative to IFRS, but may diminishes them. Second,

value relevance is higher for US GAAP earnings relative to IFRS earnings only with high

analysts following based on results presented in panel C of Table 7. Since statistically significant

difference does not appear when comparing US GAAP to IFRS for neither high nor low analysts

following, we would infer that IFRS adoption and reporting has lowered the value relevance of

non-US GAAP earnings reported in the home country around IFRS adoption. This finding – and

our other findings regarding value relevance after partitioning on our incentives variables – is

33

overall consistent with the financial statement users in the capital market relying less on IFRS

earnings because they do not have experience.

6.4 Investor Protection and Enforcement

The level of enforcement affects financial statement preparers’ incentives for producing

high quality earnings because the regulators impose a cost from their ability to initiate costly

investigations, impose direct fines and indirect costs through delays of security offerings, and

create bad publicity. US typically ranks highly in most orderings of countries based on the level

of investor protection. Consequently, we expect the SEC review of foreign registrants to increase

the costs of non-compliance. However, we are interested in how the level of investor protection

in the foreign registrant’s home country may affect our earnings quality measures under US

GAAP relative to IFRS.

On the one hand, if SEC previously monitored only US GAAP reconciliations without

attention to prior Home GAAP, the we might expect that the initiation of the SEC review of

foreign registrants increased the likelihood and severity of these regulatory costs more for

earnings reported at home under IFRS relative to prior Home GAAP. Under this hypothesis, any

differences present before IFRS adoption should disappear, not necessarily because of IFRS

adoption but rather because of the SEC review and increased enforcement. Alternatively, SEC

may have generally ignored enforcement of both prior Home GAAP and US GAAP

reconciliation filings by foreign registrants, perhaps due to limited resources. Under this view,

initiation of the SEC review would lead to a more level playing field by increasing the earnings

quality of both US GAAP and IFRS to the same, higher US level. We expect that this effect is

more pronounced for firms located in countries with lower investor protection. However, firms

34

in countries with lower investor protection are also more likely to have benefited from ‘bonding’

with the US regulatory system through cross listing in the US.

Our variable for investor protectionis the mean of four standardized measures of the legal

protection of investors reported in La Porta et al. (1998): rule of law, antidirector rights, one-

share one vote, and creditor rights. Our variable for enforcement is the monitoring and

enforcement component of disclosure from Frost et al. (2006).

Panels D of Table 6 indicate that, consistent with the latter hypothesis, earnings

predictability under US GAAP relative to IFRS is statistically significantly and the different is

less than it was under US GAAP relative to Home GAAP only in countries with low investor

protection. Additionally, Cash Predictability was statistically significantly different between

prior Home GAAP and US GAAP. From panel D of Table 7, we find that the Relevance of US

GAAP is greater than IFRS in countries with low investor protection. Untabuluted results

indicate no differences between market-based earnings attributes between prior Home GAAP

and US GAAP.

Panel E of Table 6 indicate that earnings predictability in countries with both with weak

and strong enforcement is significantly different under US GAAP and IFRS. Untabulated results

comparing prior Home GAAP to US GAAP also show a statistically significant difference only

in countries with weak enforecement. Additionally, Accruals Predictability was statistically

significantly different between prior Home GAAP and US GAAP. From panel E of Table 7, we

find statistcailly significant differences in the Relevance and Timelinesss of US GAAP and IFRS

in countries with weak enforcement. Untabuluted results indicate no differences between market-

based earnings attributes between prior Home GAAP and US GAAP.

35

To summarize, similar to our main analyses, we find few significant differences in our

tests of whether accounting–based earnings attributes and market-based earnings attributes are

affected by accounting standards (IFRS and U.S. GAAP) and various incentive variables. For

accounting based earnings attributes, the differences we find also usually existed under prior

Home GAAP. However, under prior Home GAAP the statistical significance of the differences

was generally greater and more differences between prior Home GAAP and US GAAP were

found. The differences between reporting under IFRS and US GAAP are mainly found for

Earnings Predictability in code law countries, where auditors are strong, analysts following is

high, and investor protection low. So, IFRS does appear to reduce differences in accounting-

based earnings attributes but differences continue to exist in under certain incentives and factors.

For market-based earnings attributes, we find consistent differences in Relevance, which is

generally greeter under US GAAP in the same institutional factors and incentives: in code law

countries, where auditors are strong, analysts following is high, and investor protection low.

These differences did not appear under prior Home GAAP.

We reconcile findings of the accounting-based and market-based earnings attributes. Our

results indicate that accounting-based earnings attributes under IFRS and US GAAP are more

similar than they were under prior Home GAAP and US GAAPs. Research has found (Bradshaw

et al. 2004) that companies (active in US markets) make accounting choices to reduce

differences between reported local GAAP and US GAAP. Perhaps under IFRS, companies can

make reporting choices more similar to US GAAP than they were able to under their prior Home

GAAPs. That is, IFRS is more similar to US GAAP than prior Home GAAP. (For certain

incentives and institutional factors though differences are greater.) The differences between IFRS

and US GAAP are, however, relevant to the market. Market-based earnings attributes displayed

36

little difference under US GAAP and prior Home GAAP. Perhaps Home GAAP reporting

choices were made to minimize differences between Home GAAP and US GAAP but neither

were informative to the equity market. Finally, the market-based earnings attributes findings

regarding Relevance are also consistent with the financial statement users in the capital market

relying less on IFRS earnings because they do not have experience.

37

6. Conclusion

This paper is set at the backdrop of three recent and related regulatory developments.

First, many countries have recently adopted a new set of accounting standards, IFRS. Second,

the SEC no longer requires foreign firms that report under IFRS to submit reconciliation to US

GAAP. Third, the SEC is considering whether to permit, or perhaps someday require, some or all

US-registered firms to prepare filings using IFRS. Leading up to this proposal, the SEC

announced that it would review, and subsequently reviewed IFRS-based 20-F filings for fiscal

years ended 2005 and 2006. Accordingly, US-listed, IFRS reporting firms were subject to US

regulatory scrutiny.

We exploit this setting to compare the accounting-based and market-based earnings

attributes of US-listed, IFRS reporting firms. Overall, our evidence suggests that earnings under

US GAAP and IFRS are of comparable quality. Nonetheless, US GAAP-reconciled earnings is

incrementally value relevant and informative in this period, suggesting that discontinuing

reconciliation of IFRS to US GAAP may result in less useful financial statements for valuation.

Interestingly, when we compare results to differences in prior Home GAAP and US GAAP, it

appears that companies can make reporting choices under IFRS that are more similar to US

GAAP. The differences between IFRS and US GAAP are, however, relevant to the market.

We extend our comparison of IFRS and US GAAP to examine the relationships between

incentives and the two sets of accounting standards. We find few significant differences in our

tests of whether accounting–based earnings attributes and market-based earnings attributes are

affected by accounting standards (IFRS and U.S. GAAP) or various incentive variables. We find

differences between reporting under IFRS and US GAAP are mainly reflected in a few specific

38

earnings attributes (i.e., earnings predictability and relevance) in code law countries, where

auditors are strong, analysts following is high, and investor protection low.

Three caveats are in order when interpreting our results. First, our sample consists of

firms voluntarily subjected to U.S regulation and securities markets. Bradshaw, Bushee and

Miller (2004) show that these firms tend to choose accounting methods when available under

their home GAAP that are closer to US GAAP. This would bias us against finding evidence of

incremental informativeness of US GAAP reconciliations.

Second, because we are limited to examining certain time periods when US GAAP –

IFRS reconciliations are available, there is time clustering in our data. We cannot rule out that

the time differences in the reporting incentives, the overall economy enforcement and regulatory

institutions, could influence our results.

Third, to the extent that a subset of our tests utilizes value-relevance measure, we ignore

the notion that financial statements users rely on accounting for other purposes, including

contracting and performance evaluation. For example, Bharath, Sunder, and Sunder (2008)

document how earnings attributes vary with firms’ choice of public or private debt. Further, Wu

and Zhang (2008) find that firms CEO turnover and employee layoffs become more sensitive to

earnings after firms voluntarily adopt IFRS. Finally, Dye and Sunder (2001) argue allowing two

accounting standards to coexist facilitates competition among standard setters, leading to higher

quality standards in the long run. Such non-market trade-offs should also be considered as

regulators contemplate allowing US firms to report under IFRS.

38

References

Amir, E., Harris, T., and Venuti, E., 1993. A comparison of the value-relevance of U.S. vs. non-U.S. GAAP accounting measures using Form 20-F reconciliations. Journal of Accounting Research 31 (Supplement), 230-264. Ashbaugh, H., 2001. Non-U.S. firms’ accounting standard choices. Journal of Accounting and Public Policy 20 (Summer), 129-153. Ashbaugh, H., and Pincus, M., 2001. Domestic accounting standards, International accounting standards, and the predictability of earnings. Journal of Accounting Research 39 (December), 417-434. Ball, R., Kothari, S. P., and Robin, A., 2000. The effect of international institutional factors on properties of accounting earnings, Journal of Accounting and Economics 29, 1-51. Ball, R. Kothari, S.P., Robin, A., and Wu, J., 2003. Incentives versus standards: properties of accounting earnings in four East Asian countries. Journal of Accounting and Economics 36, 235-270. Ball, R., and Shivakumar, L., 2005. Earnings quality in U.K. private firms. Journal of Accounting and Economics 39 (1), 83-128. Barth, M., Landsman, W., and Lang, M.H., 2008. International accounting standards and accounting quality. Journal of Accounting Research 46 (June), 467-498. Barth, M., Landsman, W., Lang, M.H., and Williams, C., 2006. Accounting quality: International Accounting Standards and U.S. GAAP. Working paper, University of North Carolina. Basu, S., 1997. The conservatism principle and asymmetric timeliness of earnings. Journal of Accounting and Economics 24, 3-37. Becker, C.M., DeFond, M., Jiambalvo, J., and Subramanyam, K.R. 1998. The effect of audit quality on earnings management. Contemporary Accounting Research 2 Bharath, S. T., Sunder, J., and Sunder. S.V., 2008. Accounting quality and debt contracting. The Accounting Review 83 (January), 1-28. Bhattacharya, U., Daouk, H., Jorgenson, B., and Kehr, H., 2000. When an event is not an event: The curious case of an emerging market. Journal of Financial Economics 55 (1), 69-101. Bradshaw, M., Bushee, B., and Miller, G., 2004. Accounting choice, home bias, and U.S. investment in non-U.S. firms. Journal of Accounting Research 42 (5), 795-841.

39

Bushman, R.M., and Smith, A.J. 2001. Financial accounting information and corporate governance. Journal of Accounting and Economics 32 (1-3): 237-333. Chang, J., Khanna, T., and Palepu, K.J., 2001. Analyst activity around the world. Working Paper, Harvard Business School. Choi, J., Kim, J.G, Liu, D., and Simunic, D. 2008. Audit pricing, legal liability regimes, and big 4 premiums: Theory and cross-country evidence. Contemporary Accounting Research 25 (1), 55-99. Daske, H., Hail, L., Leuz, C., and Verdi, R., 2007. Adopting a label: Heterogeneity in the economic consequences of IFRS adoptions. Working paper, University of Chicago. Daske, H., Hail, L., Leuz, C., and Verdi, R., 2008. Mandatory IFRS reporting around the world: Early evidence on the economic consequences. Journal of Accounting Research 46 (5), 1085-1142. 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 (July), 3-42. Dechow, P. M., and Dichev, I.D., 2002. The quality of accruals and earnings: The role of accrual estimation errors. The Accounting Review 77 (Supplement), 35-59. Dye, R., and Sunder, S., 2001. Why not allow FASB and IASB standards to compete in the U.S.? Accounting Horizons 15 (September), 257-271. Easton, P., and Harris, T., 1991. Earnings as an explanatory variable for returns. Journal of Accounting Research 29 (1), 19-36. Europa, 2006. The use of non-EU accounting standards on EU stock markets is extended for two more years. December: Brussels. Financial Accounting Standards Board (FASB). 1978. Statement of financial accounting concepts no. 1. FASB. Norwalk, CT. FASB – IASB., 2006. A roadmap for convergence between IFRS and U.S. GAAP – 2006-2008 Memorandum of Understanding between the FASB and the IASB”, February 27: (the 2006 Memorandum of Understanding) available at http://72.3.243.42/intl/mou_02-27-06.pdf. Financial Accounting Standards Board – International Accounting Standards Board (FASB-IASB), 2008. Project update conceptual framework—Phase A: Objective and qualitative characteristics. (4 February). London, UK. Available at: http://www.fasb.org/project/cf_phase-a.shtml. Francis, J., LaFond, R., Olsson, P. M., and Schipper, K., 2004. The costs of equity and earnings attributes. The Accounting Review 79 (October), 967-1010.

40

Frost, C., and Pownall, G., 1994. A comparison of the stock-price response to earnings disclosure in the United States and the United Kingdom. Contemporary Accounting Research 11(1), 59-83. Givoly, D., and Hayn, C., 2000. The changing time-series properties of earnings, cash flows and accruals: Has financial reporting become more conservative? Journal of Accounting and Economics 29 (June), 287-320. Hail, L., Leuz, C., and Wysocki, P.M. 2009. Global Accounting Convergence and the Potential Adoption of IFRS by the United States: An Analysis of Economic and Policy Factors. February. Working paper, University of Chicago. Harris, M., Muller, K., 1999. The market valuation of IAS versus U.S. GAAP accounting measures using form 20-F reconciliations. Journal of Accounting and Economics 26 (January), 285-312. Hewitt, C., 2007. Remarks before the 2007 foundation of accounting education SEC/FASB conference. http://www.sec.gov/news/speech/2007/spch012207cwh.htm. January: Washington, DC. Holthausen, R., and Watts, R., 2001. The relevance of the value-relevance literature for financial accounting standard setting. Journal of Accounting and Economics 31 (1-3, September), 3-75. Hung, M. 2000. Accounting standards and value relevance of earnings: An international analysis. Journal of Accounting and Economics 30 (December): 401-420. International Accounting Standards Board (IASB), 2001. Framework for the Preparation and Presentation of Financial Statements. IASB. London: UK. International Accounting Standards Board (IASB), 2008a. www.iasb.org International Financial Reporting Standards (IFRS), 2008b. International Accounting Standards Board. London: UK. Khanna, T., Palepu, G., and Srinivasan, S., 2004. Disclosure practices of foreign companies interacting with U.S. markets. Journal of Accounting Research 42 (May), 475-508. Kim, M., and Kross, W., 2005. The ability of earnings to predict future cash flows has been increasing- not decreasing. Journal of Accounting Research 43 (December), 753-780. Kothari, S.P., Ramanna, K., and Skinner, D. 2009. What Should GAAP Look Like? Working paper, University of Chicago.

41

Kothari, S. P., Zimmerman, J. L., 1995. Price and return models. Journal of Accounting and Economics 20 (September), 155-192. Land, J., and Lang, M.H., 2002. Empirical evidence on the evolution of international earnings. The Accounting Review 77 (Supplement), 115-133. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and Vishny, R.W., 1998. Law and finance. Journal of Political Economy 106 (6): 1113–1155. La Porta, R., Lopez-de-Silanes, F., and Shleifer, A., 2008. The economic consequences of legal origins. Journal of Economic Literature 46 (2), 285–332. Lang, M.H., and Lundholm, R.J. 1993. Cross-sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research 31 (2): 246-271. Lang, M. H., and Lundholm, R.J. 1996. Corporate disclosure policy and analyst behavior. The Accounting Review 71 (4): 467-492. Lang, M.H., Raedy, J., and Wilson, W., 2006. Earnings management and cross listing: Are reconciled earnings comparable to U.S. earnings? Journal of Accounting and Economics 42 (October), 244-283. Lang, M.H., Raedy, J., and Yetman, M., 2006. How representative are firms that are cross-listed in the United States? An Analysis of Accounting Quality. Journal of Accounting Research 41 (2), 363–386. Leuz, C., 2003. IAS versus U.S. GAAP: Information asymmetry-based evidence from Germany's new market. Journal of Accounting Research 41 (3), 445-463. Leuz, C., Nanda, D., and Wysocki, P., 2003. Earnings management and investor protection: An international comparison. Journal of Financial Economics 69, 505-527. Lo, K., and Lys, T., 2000. The Ohlson model: Contribution to valuation theory, limitations, and empirical applications. Journal of Accounting, Auditing and Finance 15, 337-367. Nicolaisen, D., 2005. A securities regulator looks at convergence. Northwestern Journal of International Law and Business 25 (3), 661-686. Penman, S. and Zhang, X-J., 2002. Accounting conservatism, the quality of earnings and stock returns. The Accounting Review 77, 237-264. Pope, P.F., and Walker, M.. 1999. International differences in the timeliness, conservatism, and classification of earnings. Journal of Accounting Research 37 (Supplement), 53-87. Pownall, G., and Schipper, K., 1999. Implications of accounting research for the SEC's consideration of International Accounting Standards for U.S. securities offerings. Accounting

42

Horizons 13 (September), 259-280. Rayburn, J., 1986. The association of operating cash flows and accruals with security returns. Journal of Accounting Research, Studies on Alternative Measures of Accounting Income 24, 112-133. Trueman, B., and Titman., S., 1988. An explanation for accounting income smoothing. Journal of Accounting Research 26 (Supplement), 127-139. Tucker, J., and Zarowin, P., 2006. Does income smoothing improve earnings informativeness? The Accounting Review 81 (January), 251-270. U.S. Securities and Exchange Commission (SEC). 2008. Roadmap for the Potential Use of Financial Statements Prepared in Accordancew tih International Financial Reporting Standards by U.S. Issuers. Release NoS. 33-8982; 34-58960; http://www.sec.gov/rules/proposed/2008/33-8982.pdf U.S. Securities and Exchange Commission (SEC), 2007a. Concept release on allowing U.S. issuers to prepare financial statements in accordance with International Financial Reporting Standards (Corrected) Other Release Nos.: 33-8831; http://www.sec.gov/rules/concept/2007/33-8831.pdf U.S. Securities and Exchange Commission (SEC), 2007b. Acceptance from foreign private issuers of financial statements prepared in accordance with international financial reporting standards without reconciliation to U.S. GAAP. Other Release Nos.: 33-8879; International Series Release No. 1306. http://www.sec.gov/rules/final/2007/33-8879.pdf. Watts, R., 2003a. Conservatism in accounting part I: Theory and implications. Accounting Horizons 17 (September), 207-221. Watts, R., 2003b. Conservatism in accounting part II: Evidence and research opportunities. Accounting Horizons 17 (December), 287-301. Wu, J., Zhang, I., 2008. The adoption of International Accounting Standards and firm internal performance evaluation. Working paper, University of Minnesota. Wu, J.S., and Zhang, I., 2009. The Adoption of Internationally Recognized Accounting Standards: Implications for the Credit Markets. Working paper, University of Minnesota.

43

Table 1: Descriptive Statistics Panel A: Market Capitalization and Annual Returns

Number of:

Market Capitalization (USD mil) RET

Country Companies Obs. Mean Median Mean Median Australia 8 17 4,239 17 0.123 0.162 Austria 1 2 10,149 10,149 0.201 0.201 Belgium 1 3 5,989 6,152 0.273 0.189 Bermuda 1 3 1,380 1,252 0.723 0.601 China 8 24 7,645 2,652 0.282 0.192 Denmark 3 6 8,873 9,892 0.253 0.216 Finland 4 12 24,711 9,879 0.248 0.188 France 19 48 29,976 15,239 0.305 0.317 Germany 9 22 27,103 17,756 0.316 0.404 Ireland 6 14 8,138 5,666 0.245 -0.159 Italy 8 18 27,370 14,177 0.194 0.295 Luxembourg 3 10 3,312 1,807 0.456 0.179 Mexico 1 3 179 205 0.418 0.360 Netherlands 13 31 20,046 13,266 0.344 0.357 Norway 1 2 13,643 13,643 0.376 0.348 Papua New Guinea 1 3 1,539 1,398 0.147 0.376 Portugal 2 5 11,261 11,295 0.168 0.147 Russia 1 1 1,334 1,334 1.000 0.194 South Africa 3 7 5,849 3,238 0.321 1.000 Spain 6 14 54,373 57,098 0.277 0.334 Sweden 7 16 15,056 6,621 0.486 0.241 Switzerland 5 14 52,756 23,903 0.185 0.258 Turkey 1 2 16,671 16,671 0.228 0.157 United Kingdom 48 110 32,866 11,772 0.192 0.228 Venezuela 1 2 1,432 1,432 0.154 0.186 All 161 389 24,337 9,939 0.262 0.232

Market Capitalization is the year-end market value and presented in U.S. dollars for comparability. RET is the 12-month return ending the month of the 20-F filing.

44

Table 1: Descriptive Statistics (continued)

Panel B: Descriptive Statistics for Variables Used in Regression Models

Variables by Standard na IFRS n USGAAP

Tests of Equality of Means and

Medians (p-values)

NI mean 389 0.121 389 0.105 0.776 std 0.869 0.760 med 0.079 0.073 0.081 ΔNI mean 386 -0.411 383 0.026 0.288 std 6.556 0.130 med -0.060 0.011 0.201 ACC mean 386 -0.411 386 -0.427 0.978 std 6.556 6.619 med -0.060 -0.069 0.400 ΔACC mean 295 -0.557 371 -0.219 0.593 std 9.397 9.430 med 0.000 0.000 0.844 Other Variables n Diff mean 389 0.016 std 0.162 med 0.006 ΔDiff mean 304 -0.006 std 0.119 med 0.001 CFO mean 386 0.533 std 7.099 med 0.141 ΔCFO mean 375 0.240 std 9.404 med 0.008

NI is the annual net income of the company, CFO is the annual cash flows from operations of the company, ACC is the difference between NI and CFO, Diff is the difference between ACC IFRS and ACC US GAAP, the symbol, Δ, represents the annual change in the variable, RET is the 12-month return ending the month of the 20-F filing. All financial statement variables are scaled by beginning year market price.

45

a The number of observations varies due to the data required to estimate each variable. For instance, the change variables require two years of data, so there are fewer observations.

46

Table 2: Correlations between Operating Cash Flows, Net Income and Accruals under IFRS, US GAAP, and Annual Returns

(Pearson correlations above the diagonal and Spearman correlations below)a

CFO

ΔCFO NI

IFRS ΔNI IFRS

NI US GAAP

ΔNI US GAAP

ACC IFRS

ΔACC IFRS

ACC US GAAP

ΔACC US GAAP Diff ΔDiff RET

CFO 0.885 0.661 0.151 -0.995 0.662 0.115 -0.996 0.439 0.063 0.216 0.885 0.000 0.000 0.008 0.000 0.000 0.025 0.000 0.000 0.271 0.000 0.000 375 386 305 386 386 381 386 386 304 330 375 ΔCFO 0.342 0.260 0.146 -0.924 0.258 0.194 -0.920 0.181 0.054 0.062 0.342 0.000 0.000 0.012 0.000 0.000 0.000 0.000 0.000 0.355 0.263 0.000 375 375 295 375 375 371 375 375 295 325 375 NI IFRS 0.471 0.154 0.377 -0.583 0.989 -0.022 -0.595 0.722 0.295 0.221 0.471 0.154 0.000 0.003 0.000 0.000 0.000 0.675 0.000 0.000 0.000 0.000 0.000 0.003 386 375 305 386 389 383 386 389 304 333 386 375 ΔNI IFRS 0.166 0.161 0.446 -0.128 0.269 0.687 -0.140 0.629 0.686 0.127 0.166 0.161 0.004 0.006 0.000 0.026 0.000 0.000 0.014 0.000 0.000 0.041 0.004 0.006 305 295 305 305 305 304 305 305 304 260 305 295

NI US GAAP -0.703 -

0.317 0.098 0.146 -0.586 -0.127 1.000 -0.380 -0.041 -

0.059 -0.703 -0.317 0.000 0.000 0.055 0.011 0.000 0.013 0.000 0.000 0.477 0.288 0.000 0.000 386 375 386 305 386 381 386 386 304 330 386 375 ΔNI US GAAP 0.446 0.122 0.859 0.355 0.070 -0.022 -0.595 0.612 0.131 0.285 0.446 0.122 0.000 0.018 0.000 0.000 0.170 0.665 0.000 0.000 0.022 0.000 0.000 0.018 386 375 389 305 386 383 386 389 304 333 386 375

a Correlations are in the first row, p-values in the second and number of observations in the third

47

Table 2: Correlations between Operating Cash Flows, Net Income and Accruals under IFRS, US GAAP, and Annual Returns (continued)

(Pearson correlations above the diagonal and Spearman correlations below)a

CFO

ΔCF

O NI

IFRS ΔNI IFRS

NI US GAAP

ΔNI US GAAP

ACC IFRS

ΔACC IFRS

ACC US GAAP

ΔACC US GAAP Diff ΔDiff RET

ACC -0.703 -

0.317 0.098 0.146 -0.586 -0.127 1.000 -0.380 -0.041 -0.059 -0.703 -0.317 IFRS 0.000 0.000 0.055 0.011 0.000 0.013 0.000 0.000 0.477 0.288 0.000 0.000 386 375 386 305 386 381 386 386 304 330 386 375 ΔACC 0.446 0.122 0.859 0.355 0.070 -0.022 -0.595 0.612 0.131 0.285 0.446 0.122 IFRS 0.000 0.018 0.000 0.000 0.170 0.665 0.000 0.000 0.022 0.000 0.000 0.018 386 375 389 305 386 383 386 389 304 333 386 375 ACC 0.082 0.262 0.321 0.719 0.050 0.343 -0.126 -0.011 -0.058 0.241 0.082 0.262 US GAAP 0.108 0.000 0.000 0.000 0.333 0.000 0.014 0.824 0.316 0.000 0.108 0.000 381 371 383 304 381 383 381 383 304 327 381 371

ΔACC -0.714 -

0.305 0.012 0.020 0.912 0.118 0.072 -0.401 -0.057 -0.019 -0.714 -0.305 US GAAP 0.000 0.000 0.809 0.733 0.000 0.021 0.164 0.000 0.318 0.729 0.000 0.000 386 375 386 305 386 386 381 386 304 330 386 375 Diff 0.034 -0.067 0.200 0.177 0.077 -0.155 -0.106 -0.187 0.784 -0.124 0.034 -0.067 0.508 0.196 0.000 0.002 0.129 0.002 0.038 0.000 0.000 0.024 0.508 0.196 386 375 389 305 386 389 383 386 304 333 386 375

ΔDiff 0.072 -

0.028 0.141 0.283 0.028 -0.062 -0.246 -0.145 0.505 -0.148 0.072 -0.028 0.209 0.633 0.014 0.000 0.627 0.285 0.000 0.011 0.000 0.018 0.209 0.633 304 295 304 304 304 304 304 304 304 259 304 295 RET 0.244 0.152 0.298 0.182 -0.078 0.337 0.294 -0.033 -0.068 -0.169 0.244 0.152 0.000 0.006 0.000 0.003 0.160 0.000 0.000 0.554 0.215 0.007 0.000 0.006 330 325 333 260 330 333 327 330 333 259 330 325

48

a Correlations are in the first row, p-values in the second and number of observations in the third.

49

Table 3: Accounting-based Earnings Attributes under IFRS and US GAAP Standard

Accounting-based Attributes na IFRS US GAAP Accrual Quality 116 0.214 0.217 Earnings Persistence 304 0.843 0.895 Earnings Predictability 304 0.062 0.065

Cash Persistence 294 1.206 1.208

Cash Predictability 294 0.163 0.162

Smoothness 396 2.055 1.822

Accrual Quality is the standard deviation of the residual from the regression of accruals on future year, current year, and previous year’s cash flows from operations. Earnings Persistence is the estimated coefficient on previous year’s net income from a regression of current net income on previous year’s net income. Earnings Predictability is the standard deviation of the residual from the Earnings Persistence regression. Cash Persistence is the estimated coefficient on previous year’s operating cash flows from a regression of current operating cash flows on previous year’s cash flows from operations and accruals. Cash Predictability is the standard deviation of the residual from the Cash Persistence regression. Smoothness if the standard deviation of net income divided by the standard deviation of operating cash flows by standard.

a The number of observations varies due to the data required to estimate each attribute. For instance, Accrual Quality requires three years of data for the same company, Earnings Persistence requires two years, and Smoothness requires only one year. Therefore, we have fewer observations for the Accrual Quality attribute compared to the Smoothness attribute.

50

Table 4: Market-based Earnings Attributes under IFRS and US GAAP Standard Market-based Attributes na IFRS US GAAP Relevance 259 0.040 * 0.148 Timeliness 332 0.114 0.157 Conservatism 332 1.041 1.082 Timeliness is the adjusted R2 from a regression of net income on an indicator variable equaling one if the company’s annual return is negative and zero otherwise, the company’s annual return, and the interaction of the annual return and the indicator variable. Conservatism is the ratio of the sum of the estimated coefficients on the annual return and the interaction variable divided by the estimated coefficient on the annual return from the Timeliness regression. Relevance is the adjusted R2 from a regression of annual returns on net income and changes in net income. a The number of observations varies due to the data required to estimate each attribute. For instance, Relevance requires two year of data for the same company and Timeliness and Conservatism only one year. Therefore, we have fewer observations for the Relevance attribute compared to the Timeliness and Conservatism attributes. * Vuong test p-value 0.005.

51

Table 5: OLS Regressions of Operating Cash Flows and Accruals under IFRS and US GAAP and Changes in Operating Cash Flows and Accruals under

IFRS and US GAAP on Annual Returns The dependent variable is RET, the 12-month return ending the month of the 20-F filing. Panel A: Regression Models

(1) (2) (3) (4) (5) INT 0.249

(0.000) 0.242 (0.000)

0.246 (0.000)

0.248 (0.000)

0.248 (0.000)

CFO 0.220

(0.002) 0.231 (0.010)

0.293 (0.002)

0.318 (0.001)

0.218 (0.007)

ΔCFO -0.030

(0.833) 0.257 (0.178)

0.478 (0.012)

0.414 (0.027)

0.414 (0.027)

ACC IFRS 0.0498

(0.657) 0.172

(0.138)

ΔACC IFRS 0.185

(0.108) 0.338

(0.005)

ACC US GAAP 0.166

(0.158) 0.172

(0.138) ΔACC US GAAP 0.333

(0.006) 0.338

(0.005) Diff -0.417

(0.263) -0.245 (0.483)

ΔDiff -0.722

(0.056) -0.384 (0.283)

Adj. R2 3.5% 4.9% 10.0% 12.8% 12.8% Number of obs. 255 255 255 255 255

52

Table 5: OLS Regressions of Operating Cash Flows and Accruals under IFRS and US GAAP and Changes in Operating Cash Flows and Accruals under

IFRS and US GAAP on Annual Returns (continued) Panel B: Tests of Incremental Explanatory Power Explanatory power of: Incremental to: p-value ACC IFRS and ACC ΔIFRS

> CFO and ΔCFO 0.053

ACC US GAAP and ΔACC US GAAP

> CFO and ΔCFO 0.000

Diff and ΔDiff

> CFO and ΔCFO and ACC IFRS and ΔACC IFRS

0.008

Diff and ΔDiff > CFO and ΔCFO and

ACC US GAAP and ΔACC US GAAP

0.001

Panel C: Test of Relative Explanatory Power (Using Vuong Test) ACC IFRS and ACC ΔIFRS

< ACC US GAAP and ΔACC US GAAP

0.015

NI is the annual net income of the company, CFO is the annual cash flows from operations of the company, ACC is the difference between NI and CFO, Diff is the difference between IFRS Accruals and US GAAP Accruals, the symbol, Δ, represents the annual change in the variable, ANNUAL RETURN is the 12-month return ending the month of the 20-F filing. All financial statement variables are divided by beginning year market price. The number of observations in each regression varies slightly due to the deletion of regression outliers, identified as those observations with the absolute value of the residual exceeding three.

53

Table 6: Accounting-based Earnings Attributes under IFRS and US GAAP By Incentivea

Attribute Incentive n IFRS USGAAP Panel A: Legal System Accrual Quality Code 62 0.119 0.129 * Civil 51 0.153 0.159

Earnings Persistence Code 162 0.539 0.460 Civil 134 1.074 0.919

Earnings Predictability Code 162 0.136 *** 0.108 *** *** Civil 134 0.095 0.086

Cash Persistence Code 155 0.837 0.767 Civil 130 1.059 1.091 Cash Predictability Code 155 0.227 0.229 *** *** Civil 130 0.751 0.752

Smoothness Code 205 0.557 0.536 Civil 169 0.770 0.630 Panel B: Auditors Accrual Quality Weak Auditors 62 0.186 0.194 Strong Auditors 32 0.082 0.077 Earnings Persistence Weak Auditors 154 0.765 0.823 Strong Auditors 82 0.820 0.779 Earnings Predictability Weak Auditors 154 0.080 0.089 Strong Auditors 82 0.136 *** 0.096 Cash Persistence Weak Auditors 150 0.923 0.920 Strong Auditors 78 1.402 1.377 Cash Predictability Weak Auditors 150 0.659 0.659 *** *** Strong Auditors 78 0.404 0.404 Smoothness Weak Auditors 199 0.673 0.704 Strong Auditors 103 0.530 0.466

54

Table 6: Accounting-based Earnings Attributes under IFRS and US GAAP By Incentivea (continued)

Attribute Incentive n IFRS USGAAP Panel C: Analysts Following Accrual Quality Low Analysts Following 13 0.129 * 0.073 High Analysts Following 51 0.123 0.135 Earnings Persistence Low Analysts Following 38 1.173 0.873 High Analysts Following 125 0.793 0.747 Earnings Predictability Low Analysts Following 38 0.078 0.076 *** *** High Analysts Following 125 0.122 ** 0.103 Cash Persistence Low Analysts Following 38 1.027 1.018 High Analysts Following 119 1.262 1.236 Cash Predictability Low Analysts Following 38 0.087 0.086 *** *** High Analysts Following 119 0.344 0.344 Smoothness Low Analysts Following 51 1.506 1.459 High Analysts Following 157 0.542 0.053 Panel D: Investor Protection Accrual Quality Low Investor Protection 33 0.118 0.115 *** *** High Investor Protection 23 0.066 0.058 Earnings Persistence Low Investor Protection 76 0.408 0.648 High Investor Protection 66 1.124 0.828 Earnings Predictability Low Investor Protection 76 0.114 * 0.089 *** *** High Investor Protection 66 0.046 0.043 Cash Persistence Low Investor Protection 75 1.055 1.035 High Investor Protection 63 1.117 1.025 Cash Predictability Low Investor Protection 75 0.098 0.095 ** *** High Investor Protection 63 0.139 0.141 Smoothness Low Investor Protection 106 0.724 0.761 High Investor Protection 81 0.333 0.329

55

Table 6: Accounting-based Earnings Attributes under IFRS and US GAAP By Incentivea (continued)

Attribute Incentive n IFRS USGAAP Panel E: Enforcement Accrual Quality Weak Enforcement 37 0.134 0.139 *** *** Strong Enforcement 12 0.057 0.069 Earnings Persistence Weak Enforcement 93 0.886 0.687 Strong Enforcement 40 0.443 0.113 Earnings Predictability Weak Enforcement 93 0.142 ** 0.115 ** Strong Enforcement 40 0.194 *** 0.113 Cash Persistence Weak Enforcement 91 0.956 0.937 Strong Enforcement 40 1.325 1.200 Cash Predictability Weak Enforcement 91 0.129 0.128 *** *** Strong Enforcement 40 0.303 0.307 Smoothness Weak Enforcement 120 1.214 1.256 Strong Enforcement 48 0.674 0.573

a Asteriks between two numbers indicate the following significance levels:

*** significant at the 0.01 level or better ** significant at the 0.05 level or better

* significant at the 0.10 level or better

56

Table 7: Market-based Earnings Attributes under IFRS and US GAAP By Incentivesa

Attribute Incentive n IFRS USGAAP Panel A: Legal System Relevance Code 143 0.015 *** 0.126 Civil 108 0.084 0.169 Conservatism Code 182 1.028 1.085 Civil 140 1.086 1.129 Timeliness Code 182 0.014 0.066 Civil 140 0.199 0.209 Panel B: Auditors Relevance Weak Auditors 134 0.101 0.194 Strong Auditors 76 -0.001 ** 0.106 Conservatism Weak Auditors 176 1.010 1.135 Strong Auditors 96 1.052 1.061 Timeliness Weak Auditors 176 0.111 0.096 Strong Auditors 96 0.131 0.175 Panel C: Analysts Following Relevance Low Analysts Following 29 0.170 0.167 High Analysts Following 115 0.007 ** 0.096 Conservatism Low Analysts Following 41 1.043 1.046 High Analysts Following 147 1.029 1.037 Timeliness Low Analysts Following 41 0.273 0.256 High Analysts Following 147 0.029 0.044

57

Table 7: Market-based Earnings Attributes under IFRS and US GAAP

By Incentivesa (continued) Attribute Incentive n IFRS USGAAP Panel D: Investor Protection Relevance Low Investor Protection 73 0.000 * 0.105 High Investor Protection 53 0.078 0.164 Conservatism Low Investor Protection 102 1.013 1.006 ** High Investor Protection 64 -0.142 0.571 Timeliness Low Investor Protection 102 0.073 0.146 High Investor Protection 64 0.015 0.035 Panel E: Enforcement Relevance Weak Enforcement 83 0.054 ** 0.155 Strong Enforcement 29 0.124 0.143 Conservatism Weak Enforcement 110 1.004 0.994 Strong Enforcement 32 0.853 1.022 Timeliness Weak Enforcement 110 0.040 * 0.272 Strong Enforcement 32 0.017 0.038

a Asteriks between two numbers indicate the following significance levels:

*** significant at the 0.01 level or better ** significant at the 0.05 level or better

* significant at the 0.10 level or better

58

Appendix: Reported Differences between IFRS and US GAAP Net Income

This appendix provides descriptive statistics for topical accounting differences by category between

IFRS and US GAAP net income. Differences in taxes, compensation (pensions, share-based payments),

business combinations, property, plant and equipment and investments were the most often noted as

presented in Table 1A, panel a. Positive (negative) mean and median differences correspond to US GAAP

net incomes that are higher (lower) than net incomes reported under IFRS, reported in Table 1A, panels b

and c.

Tax differences are the most commonly reported IFRS/US difference. This is not surprising since

taxes are driven, in part, by the other IFRS/US differences in income and shareholders’ equity. Nevertheless,

the standards for accounting for income taxes under IFRS (IAS 12) and US GAAP (FAS 109) are similar.

Both standards require firms to account for current and expected future tax consequences of book to tax

differences. Nevertheless, differences between IFRS and US GAAP are evident. For example, IFRS does

not require recognition of deferred tax effects on the initial recognition of asset or liability whereas US

GAAP does. As a result of differences between IFRS and US GAAP, 133 (41%) of sample companies

reported differences due to deferred taxes separately. Sample companies such as Ducati (2005) reported

differences in deferred tax.

Compensation differences are reconciled in 111 (33%) of the sample firm years. Under IFRS, under

US GAAP (Appendix A, Panel a). Of those 111 firm-years for which differences were reported, 42 were

associated with a higher US GAAP net income, whereas 69 were associated with a lower US GAAP net

income as compared with IFRS.

Compensation differences between IFRS/US GAAP income are noted mostly where IFRS and US

GAAP yields different current period expenses for pensions and share-based compensation. IAS 19,

59

Employee Benefits, is the principal source of guidance for non- share based employee benefits and

compensation under IFRS, whereas IFRS 2 provides guidance for share-based payments. Multiple standards

apply under US GAAP, including FAS 87, FAS 132, and FAS 112 for employee benefits, and FAS 123R for

share-based payments. With respect to share-based payment, in 2005, IFRS 2 required measurement of

compensation under fair value, while US GAAP maintained provisions for calculating awards using the

intrinsic value method. As a result, in 2005, companies (e.g., Thomson (2006)) reported differences in their

share-based compensation expense. Others, such as Royal & Sun Alliance Insurance Group plc (2005)

recorded pension compensation differences under US GAAP and IFRS.

Forty-one percent of the sample firm-years’ included an IFRS to US GAAP reconciling item for

business combinations. While revised standards IFRS 3 and SFAS 141R converge business combination

standards, the new standards are required for fiscal years beginning on or after January 1, 2009. During the

time of our study, the standards for business combinations remained diverged under IFRS and US GAAP.

Differences between IFRS and US GAAP pre-convergence include (i) difference in partial consolidation

(allowed under IFRS, but not under US GAAP), (ii) the recognition of negative goodwill as an increase to

current period income in the year of acquisition (allowed under IFRS, not under US GAAP, e.g., Air-France

KLM (2006)), as well as (iii) accounting differences related to the date of an acquisition, and accounting for

business combinations achieved in stages.

Panels b and c of Table 1A provides a summary of reconciling items by positive and negative

differences. Of the reconciling items reported by sample firms, compensation differences were associated

with the highest mean increase from IFRS to US GAAP. Differences in the accounting for business

combinations were associated negative difference between IFRS and US GAAP, driving US income lower

than income reported under IFRS.

60

Table 1A Summary of Differences by Category in Net Income under IFRS and Net Income under

US GAAP Panel a: Overall summary All a ,b

Category na %

Reporting Rank of %

Rank Absolute Value of

Mean Mean Std Dev Median

Tax 133 41% 1 3 -0.106 1.113 0.000 Compensation (including share based) 111 34% 2 2 0.173 1.078 -0.001 Business combinations 88 27% 3 1 -1.806 16.670 -0.002 Otherc 83 25% 4 15 0.002 0.031 0.000 Property, plant and equipment 75 23% 5 11 -0.005 0.026 0.000 Investments 68 21% 6 14 -0.002 0.018 0.000 Intangibles 66 20% 7 4 -0.088 0.619 -0.002 Derivative 54 16% 8 18 -0.001 0.027 -0.001 Debt 36 11% 9 5 0.023 0.115 0.001 Discontinued operations 35 11% 10 9 -0.007 0.009 -0.003 Leases 33 10% 11 17 0.001 0.023 0.000 Revenue recognition 26 8% 12 12 0.004 0.014 0.000 Foreign currency 25 8% 13 19 -0.001 0.007 0.000 Provisions 24 7% 14 10 -0.005 0.016 0.000 Inventory 18 5% 15 16 -0.002 0.004 0.000 Restructuring 17 5% 16 7 0.011 0.141 0.000 Deferred charges 11 3% 17 6 -0.011 0.019 -0.003 Cumulative effect of accounting change 10 3% 18 8 -0.010 0.020 -0.001 Loans (primarily financial institutions) 10 3% 19 13 -0.002 0.003 -0.001 Total Number of Observations 328

61

Table 1A Summary of Differences by Category in Net Income under IFRS and Net Income under

US GAAP (continued)

Panel b: Summary by positive items

Positive Items a, d

na %

Reporting Rank of %

Rank Mean Mean

Std Dev Median

Tax 73 22.3% 1 10 0.007 0.012 0.003 Compensation (including share based) 42 12.8% 2 1 0.473 1.722 0.002 Business combinations 30 9.1% 5 8 0.008 0.011 0.002 Otherc 33 10.1% 4 7 0.011 0.046 0.002 Property, plant and equipment 39 11.9% 3 12 (tie) 0.003 0.006 0.001 Investments 28 8.5% 6 11 0.006 0.014 0.001 Intangibles 17 5.2% 10 2 0.065 0.252 0.001 Derivative 18 5.5% 8 5 0.016 0.03 0.004 Debt 24 7.3% 7 4 0.037 0.139 0.001 Discontinued operations 4 1.2% 16 15 (tie) 0.001 0.001 0.001 Leases 17 5.2% 9 9 0.008 0.026 0.001 Revenue recognition 14 4.3% 11 6 0.012 0.013 0.005 Foreign currency 12 3.7% 13 12 (tie) 0.003 0.005 0 Provisions 12 3.7% 12 14 0.002 0.002 0.001 Inventory 8 2.4% 15 15 (tie) 0.001 0.001 0.001 Restructuring 11 3.4% 14 3 0.054 0.116 0.004 Deferred charges 3 0.9% 19 18 (tie) 0 0.001 0 Cumulative effect of accounting change 3 0.9% 18 15 (tie) 0.001 0.002 0 Loans (primarily financial institutions) 3 0.9% 17 18 (tie) 0 0 0

62

Table 1A Summary of Differences by Category in Net Income under IFRS and Net Income under

US GAAP (continued)

Panel c: Summary by negative items Negative Items a, d

na %

Reporting Rank of %

Rank Mean Mean

Std Dev Median

Tax 60 18.3% 2 2 -0.244 1.654 -0.004 Compensation (including share based) 69 21.0% 1 9 -0.011 0.022 -0.003 Business combinations 58 17.7% 3 1 -2.744 20.531 -0.006 Otherc 50 15.2% 4 16 (tie) -0.004 0.011 -0.001 Property, plant and equipment 36 11.0% 7 (tie) 7 -0.014 0.035 -0.002 Investments 40 12.2% 6 12 (tie) -0.007 0.019 -0.002 Intangibles 49 14.9% 5 3 -0.142 0.698 -0.003 Derivative 36 11.0% 7 (tie) 10 -0.009 0.022 -0.002

Debt 12 3.7% 12

(tie) 15 -0.005 0.012 -0.001 Discontinued operations 31 9.5% 9 11 -0.008 0.01 -0.004 Leases 16 4.9% 10 13 -0.007 0.016 -0.003

Revenue recognition 12 3.7% 12

(tie) 14 -0.006 0.007 -0.003 Foreign currency 13 4.0% 11 16 (tie) -0.004 0.007 -0.001

Provisions 12 3.7% 12

(tie) 8 -0.013 0.021 -0.001 Inventory 10 3.0% 15 16 (tie) -0.004 0.005 -0.002 Restructuring 6 1.8% 19 4 -0.067 0.16 -0.001 Deferred charges 8 2.4% 16 5 -0.016 0.021 -0.003 Cumulative effect of accounting change 7 2.1%

17 (tie) 6 -0.015 0.023 -0.003

Loans (primarily financial institutions) 7 2.1% 17

(tie) 19 -0.003 0.003 -0.002 Notes: a Mean and median amounts are divided by beginning market value. b Reconciling data items are combined for years 2004, 2005, and 2006 c The category “Other” includes reconciling items that company's identified as "other" plus other categories of reconciling items collected and combined due to few number of companies reporting (2% or less) or small magnitude.

d Companies reconcile from IFRS net income to US GAAP net income. So, positive (negative) reconciling

items increase (decrease) IFRS net income.