کنترل موثر برای تحقیقات در حسابداری بین المللی

Numerous studies have examined the value relevance and other characteristics of international and country-specific accounting standards. This paper evaluates the experimental designs of selected studies with respect to the controls essential for effective research. Both cross-country and within-country research designs share a common need for controls. Perhaps the most distinctive element of effective design in these studies is the control for institutional, cultural, and structural differences between countries. Previous research shows that the key outputs of various country-specific accounting standards are functions of a variety of factors including accounting-related issues, legal origin, shareholder protection, the information environment, financial markets, and enforcement of these standards. In addition, effective studies of country-specific or international accounting principles require control for firm and industry-specific effects and for self-selection bias, as in conventional within-country examinations. When controls are not sufficient, observed differences in the outputs of alternative accounting standards may result from differences between countries or firms rather than from different accounting principles. We review a sample of recent research with attention to these control considerations.

The continuing evolution toward global economic integration has been paralleled by the growth of world-wide capital markets. These trends, in turn, have influenced the move toward common international accounting standards. Numerous countries have joined this trend and in 2005, the European Union required all member countries to adopt International Financial Reporting Standards (IFRS). Within the US, both the Securities and Exchange Commission (SEC) and the Financial Accounting Standards Board (FASB) have cooperated in the trend toward common worldwide standards. The FASB, for example, has issued joint exposure drafts with the International Accounting Standards Board that propose uniform accounting principles for selected accounting and disclosure areas. International accounting standards are generally more principles-based and less detailed than US Generally Accepted Accounting Principles (GAAP). This leads some observers to question the relative informativeness or decision-usefulness of alternative accounting standards. Are US standards more informative? Or is the informativeness of financial reports prepared under international standards comparable to or superior to that of reports prepared under US GAAP? If international standards do appear to be potentially more useful, should the US subscribe to these standards? These questions clearly have important implications for investors, auditors, and regulators.1 Consequently, a considerable body of research seeks to compare the empirical properties of accounting information produced under “home country”, international, and US accounting standards.2 While these studies typically focus on the value relevance of earnings (e.g., Barth et al., 2005), some also examine other properties, such as the extent of earnings conservatism and asymmetric timeliness (e.g., Ball et al., 2003). A second set of issues relates to the access of foreign companies to US capital markets and the listing of securities on US exchanges. Foreign firms that wish to trade their securities in the US may file financial statements with the SEC that are prepared under either home country or international GAAP. The SEC requires, however, that these firms also file reconciliations of these home country (or international) financial statements to US GAAP. Such requirements are made to increase the comparability of the foreign financial statements to those of domestic firms and to increase the transparency of such statements (assuming that US GAAP produces more transparent financial statements). One problem is that the required reconciliation is costly. In addition, some argue that reconciliation is unnecessary and may be potentially misleading to investors (Edwards, 1993). In response to this position, the SEC is considering permitting foreign companies to file financial statements prepared in accord with IFRS without reconciliation. Unless IFRS is identical to US GAAP, relaxing the US-GAAP reconciliation constraint, of course, sacrifices comparability. This paper evaluates the experimental designs of recent empirical research in international accounting. Numerous studies compare and evaluate differences in the relative information provided by financial statements prepared under alternative GAAP – namely home country GAAP, international financial standards, and US GAAP. One subset of the literature seeks to identify differences in the properties of accounting earnings calculated under alternative GAAP (e.g., US GAAP vs. home country GAAP). Other literature examines the benefits of dual reporting as in the reconciliation of home country GAAP to US GAAP. Although the various studies incorporate a variety of experimental designs, they share a common need for controls, particularly controls for differences between countries. Consequently, much of the focus in this paper relates to the comparison and contrast of controls, which we believe to be essential to the effectiveness of research in international accounting. In doing so, we develop background for the assessment of research that investigates the decision-usefulness of alternative reporting standards and disclosures. Our analysis directs particular attention to a recent paper by Ndubizu and Sanchez (2006) that examines the value relevance of accounting earnings and book values in Peru and Chile. Peru uses accounting which is similar to international accounting standards while GAAP in Chile closely resembles US GAAP. The literature commonly evaluates differences in accounting standards by comparing the value relevance, conservatism of earnings, and book values of alternative GAAP. The (sometimes implicit) goal of these comparative studies is to evaluate how differences in GAAP affect managers’ ability to effectively communicate their private information to outside market participants, and in turn, whether this information is decision-useful to these outside parties. Our analysis of the literature suggests that extant empirical studies can be classified into two broad classifications with respect to experimental design: (1) between-country designs, and (2) within-country designs. Moreover, within-country experiments can be further divided into between-firm (i.e., the typical cross-sectional study) and within-firm studies. One group of within-firm studies examines the firm over time (i.e., the typical time-series study). Another focuses upon firms that report simultaneously under two sets of accounting standards. With respect to cross-country comparisons, one of the most important empirical design issues is adequate control for differences between countries. For instance, Country A may share both the same border and the same language as Country B. These adjacent countries, however, may well differ with respect to economic, legal, historical, cultural, political, social, and religious characteristics. As a result, value relevance measures will reflect differences in the underlying constructs. In the absence of adequate controls, the apparent differences in the value relevance of alternative GAAP may actually reflect differences in the fundamental, underlying characteristics of the countries themselves rather than differences in accounting standards. For example, accounting standards in a particular country may be highly detailed, but subject to little effective enforcement (e.g., Ball et al., 2003). One possibility, in this case, is that accounting information will appear to rank low on the scale of value relevance. But, is the low apparent value relevance attributable to deficiencies in the standards? Or is value relevance low because “high quality” accounting standards are offset by weak enforcement? Thus, controls for enforcement and other cross-country differences are necessary to alleviate concerns with respect to correlated omitted variables. Significant, and sometimes subtle, differences among countries compound the researcher’s difficulties. As one example, Meek and Thomas (2004) cite research indicating that in the Finnish stock market, we observe post-earnings-announcement drift for bad news firms, but not for good news firms. In comparison, the post-earnings-announcement-drift literature for US firms shows that drift applies to both good and bad news (Bernard et al., 1989 and Bernard and Thomas, 1990). One possibility is that Finnish investors may be naı¨ve or slow to respond to bad earnings news, but not to good news. It turns out, however, that the differential post-announcement drift observed for good and bad news disclosures in the Finnish market is explained not by investors’ irrationality, but by the unique characteristics of the securities market in Finland. The Finnish market does not permit short selling, thus limiting the ability of investors to immediately take advantage of bad news. The inability to sell short, of course, should not affect the reaction to good news, thus explaining the differential reaction. An understanding of unique national phenomena, such as this one, is necessary for the development of meaningful implications concerning the relative usefulness of financial information prepared in different countries. Another challenge relates to the limited availability of data, particularly computer-readable data to facilitate development of control variables. Faced with this limitation, one could embark upon a long and costly project to hand-collect the necessary data. Alternatively, the researcher can compromise. For example, rather than attempt to control for all aspects of cross-country differences, he or she can concentrate upon relatively simple, coarse characteristics, such as the code vs. common-law dichotomous variable discussed later in this paper. Such an approach, however, requires that any simplification adequately capture the pertinent differences between countries. Within-country studies avoid the concerns with respect to differences between countries, but introduce other problems. Specifically, within-country studies necessitate reduced, and sometimes very small sample sizes, and because they apply only to a single country, the results of such studies are seriously limited with respect to external validity. Another concern is that while the entire sample is taken from the same country, the sample will still be characterized by firms with different characteristics (i.e., cross-sectional differences that arise from typical panel data). Thus, as with the more usual analysis of reporting alternatives within a particular country, it is still necessary to control for systematic differences across firms such as firm size and industry membership. An even simpler experimental design focuses upon the same firm that reports under different accounting standards over time. Firms in these studies typically change accounting standards from home country to international GAAP. As we shall see later, this research design also requires control for factors other than GAAP. Self-selection bias is one potentially serious problem. Another is the need to consider temporal changes in the underlying characteristics of the firms that change accounting standards (i.e., time-series differences). Finally, some studies examine samples of firms that contemporaneously report under two sets of accounting standards. For example, foreign firms that trade on US stock exchanges may use home country or international GAAP for SEC reporting purposes. If they do so, however, they must disclose a reconciliation to US GAAP, thus effectively disclosing financial information under two sets of accounting standards. While research that uses reconciliations eliminates many control needs, samples tend to be small and self-selection bias and external validity remain as significant issues. Consideration of the preceding overview leads to questions concerning the potential success of investigations of alternative accounting standards. Our review of selected research indicates that compromises can be made to sufficiently alleviate concerns about controls while still creating research designs that should permit the development of appropriate inferences. Much, but not all, of the work examined in this paper, appears to control quite effectively for the items cited above. The next section of this paper reports on factors associated with differences in value relevance and other properties of accounting information associated with alternative GAAP. Following this, we summarize some of the more recent empirical cross-country examinations. We then consider selected studies that incorporate same-country designs. Our focus is upon relatively recent research and as a consequence, we limit our discussion to studies published after 2000. One observation is a generally sharp increase in the quality of experimental controls for the more recent studies in this group compared to those published just a few years earlier.

In this paper we evaluate the experimental designs of recent empirical research in international accounting. These experimental designs include both cross-country and within-country studies. Given the extensive differences between countries that bear upon the decision-usefulness of accounting information, we observe that the vast majority of the recent studies adopt within-country designs to minimize control considerations. We then consider a paper recently published in this journal in light of the experimental design and control needs discussed previously. Our conclusion is that further work needs to be done in this area in order to develop a basis for meaningful policy implications.