Level of economic development of a firm's country of domicile and the patterns in stock market...

ABSTRACT: Using the level of economic development as a proxy for the degree of capital market segmentation, this study tests the global market segmentation hypothesis, which predicts that the magnitude of the market's reaction to an earnings announcement of an American Depositary Receipt (ADR)

firm will vary systematically with the extent to which the firm's home country capital market is segmented. Supporting this idea, the evidence shows that the earnings announcement of an emerging economy ADR induces a larger market reaction than a developed economy ADR. This finding contributes to the market segmentation literature by providing evidence that the market can remain segmented even after cross-listing in the sense that the information environment of firms from segmented economies remains weaker. It also extends prior literature on the information content of earnings by documenting that the degree of capital market segmentation in a multinational firm's country of domicile is an important determinant of the information content of its U.S. earnings announcement.

INTRODUCTION

Miller (1999) documents that the U.S. stock market's reaction to the cross-listing announcement of a firm from an emerging economy is generally larger than its reaction to the equivalent announcement made by a firm from a developed economy. His evidence is consistent with the view that firms domiciled in segmented economies can circumvent the segmentation by cross-listing securities on foreign capital markets, thus increasing firm value and lowering the cost of capital (Stapleton and Subrahmanyam 1977). In these studies, global market segmentation is commonly assumed to arise from barriers to cross-national capital flows that usually lead to differences in risk premia and information environments among the countries (see for instance, Stulz 1981; Errunza and Losq 1989).

While Miller (1999) suggests that cross-listing by firms from emerging markets improves the imbalance of information environments (i.e., the information environment of firms from emerging economy is not as weak as it was before cross-listing relative to the information environment of firms from developed economies), it may not completely equalize the information environment of firms in the global capital market. It is possible that existing differences in legal, business, cultural, and political environments prevent the complete de-segmentation of capital markets even after cross-listing.

Consistent with this idea, prior studies report evidence that suggests an economy can remain segmented even when foreigners have relatively free access to its capital markets (see, for example, Errunza and Losq 1989; Bekaert and Harvey 1995). These studies have generally tested for segmentation using the degree of correlation of the local market return with the world return, the level of investment restrictions, or a time-series pattern in the covariance between a country's expected return and a world benchmark portfolio. However, these measures of world capital market segmentation have been criticized on the basis of the rather restrictive assumptions they impose. (See Bekaert and Harvey [1995] for a discussion.) Therefore, the conclusions drawn by these studies are better viewed as preliminary rather than conclusive, and the question of whether removal of investment barriers fully integrates world capital markets is still subject to much debate.

An alternative approach for testing world capital market segmentation that is yet to be explored in the literature is to examine the patterns in stock market reaction surrounding the earnings announcements of multinational firms. This approach is based on Atiase (1985), who finds that a difference in pre-disclosure information environment leads to a differential price reaction around the time of announcement. It allows us to test for market segmentation because if the market is segmented and, thus, there is a difference in the firms' pre-disclosure information environment, then the market will react differently (i.e., react more strongly) to the earnings announcements of firms from segmented economies. Such differential reaction would in turn suggest that cross-listing of securities fails to even firms' information environments and, therefore, does not fully integrate the market.

After controlling for firm size and listing location, the evidence indicates that the market's reaction to U.S. earnings announcements of American Depositary Receipt firms (ADRs), measured by the earnings response coefficient over a short window (i.e., three days) surrounding the announcements, is statistically significant in the emerging economy ADR sample but not in the developed economy ADR sample. (1) This finding suggests that market segmentation persists after cross-listing, and supports the view that relaxing restrictions on the flow of capital may not automatically integrate the world capital market. The finding also suggests that the level of economic development of the home country economy of an ADR security is an important determinant of the information content of U.S. earnings reports released by the firm and, thus, future research that examines the stock price behavior of cross-listed firms should control for this variable.

The rest of this paper is organized as follows. In Section II, I summarize institutional background and review related literature. In Section III, I develop the hypothesis. In Section IV, I describe the sample and methodology. In Section V, I discuss the results. I conclude in Section VI.

II. INSTITUTIONAL BACKGROUND AND RELATED LITERATURE

Institutional Background

An American Depositary Receipt (ADR) is a U.S. security of a non-U.S, firm cross-listed under the U.S. Securities Act of 1934. (2) J. P. Morgan first created this type of security in 1972 as a vehicle through which U.S. investors can participate in the London Stock Market (Miller 1999). It is a negotiable security, which is issued by a U.S. commercial bank. ADRs are commonly traded on major stock exchanges (NYSE, ASE, or NASDAQ) or an over-the-counter (OTC) market in U.S. dollars and pay dividends or interest in dollars. They are also settled, cleared, and transferred according to standard U.S. practices. They have gained popularity in recent years as a means of global diversification within an investor's portfolio.

Currently, there are four major levels of ADR programs available--Level I, Level II, Level III, and 144a. Level I ADR programs that are traded on the OTC market are subject to home country accounting standards. Levels II and HI ADR programs are traded on major stock exchanges such as NYSE, AMEX, or NASDAQ. Firms that issue these levels of ADR programs must meet full SEC disclosure requirements and meet the listing requirements of the U.S. stock exchange on which they are listed. Firms using Level II and Level III programs must complete Form 20-F (and a semi-annual Form 6-K) that reconciles their home country GAAP earnings with the corresponding U.S. GAAP figure or alternatively to file a 10-K report. (3)

While these three types of ADR programs are traded publicly, the last type of ADR program, called 144a, is placed privately to qualified institutional investors. As a result, the 144a (also referred to as PORTAL) does not require compliance with U.S. GAAP. Due to these reasons, it has been pointed out that the listing and reporting requirements for ADR programs are, in general, less stringent than those for U.S. firms (Alford et al. 1993).

Related Literature

This study mainly relates to two streams of literature in accounting and finance. First, it is related to prior studies on the cross-listing of securities and corporate disclosure practices in a segmented global capital market (see for instance, Stulz 1981; Stapleton and Subramanyan 1977; Errunza and Losq 1989; Cooke and Wallace 1990; Bekaert and Harvey 1995; Miller 1999; Errunza and Miller 2000) and studies on the association between a firm's pre-disclosure information environment and the properties of accounting information. Stulz (1981) constructs a model of an international asset, which suggests that some risky foreign assets can be non-traded (in the sense that they are not held by domestic investors and would not be held if their expected return changed slightly) in the presence of barriers to international investment. Stapleton and Subrahmanyan (1977) suggest that cross-listing on foreign capital markets can circumvent market segmentation, thus increasing firm value and lowering the cost of capital. Errunza and Losq (1989) propose a model that suggests that removal of investment barriers generally leads to an increase in the aggregate market value of the affected securities. Bekaert and Harvey (1995) provide evidence consistent with the view that some markets appear segmented, even though foreigners do have relatively free access to them.

Miller (1999) finds that a market's reaction to the listing of an ADR program is related to the choice of exchange and the geographical location of the home country economy (i.e., whether emerging or developed). Errunza and Miller (2000) document a significant decline in the cost of capital from ADR offerings, suggesting that financial market liberalizations have significant economic benefits. Cooke and Wallace (1990) find the level of corporate financial disclosure regulation in many developed countries is likely determined more by internal factors, whereas that of many developing countries is likely determined more by external factors. Extending this line of literature, I provide evidence that suggests that the market remains segmented even after cross-listing in the sense that the information environment of firms from segmented economies remains weaker.

Second, this study is related to prior accounting literature on the association between the pre-disclosure information environment and the information content of earnings (see for example, Atiase 1985; Collins et al. 1987; Grant 1980). Atiase (1985) finds that firm size, a proxy for differential informational environments, explains the magnitude of the returns-earnings relation. Using firm size as a proxy for the amount of information and for the number of traders and professional analysts processing the available information about an enterprise, Collins et al. (1987) find that price-based earnings forecasts outperform univariate time-series forecasts by a greater margin for larger firms than for smaller firms.

In a similar context, Grant (1980) finds that the amount of interim information available for OTC firms is systematically less than that available on firms that trade on a more organized stock exchange such as NYSE. Specifically, he notes that much of the data contained in the annual reports of NYSE firms are made available to investors on a more timely basis (relative to those of OTC firms) by the many existing interim sources of information such as interim financial reports, trade journals, security analysts' forecasts, industry forecasts, litigation, prospectuses, etc.

Meek (1991) finds that firm size is associated with the magnitude of the market's reaction to the earnings announcements of foreign companies traded in the U.S. Frost and Pownall (1994), who compare the stock price response to annual and interim earnings disclosures released in the U.S. and the U.K. capital markets, document that the response is significantly greater in the U.S. than in the U.K. This study contributes to this stream of literature by documenting that the level of development of the home country economy of an ADR program is an important environmental factor that determines the amount of pre-disclosure information available in the market.

III. HYPOTHESIS DEVELOPMENT

Global capital market segmentation is a subject that has been examined extensively by prior studies (see for instance, Stulz 1981; Errunza and Losq 1989; Bekaert and Harvey 1995; Miller 1999). These studies show that obstacles to the free flow of capital will in general result in an additional risk premium on securities that are not freely accessible. They also note that segmented economies often impose ownership restrictions on foreign investors in order to protect their economy. Reduced investment opportunities will decrease foreign investors' incentives to do an information search, limiting the amount of information available in the market. Along these lines, one can expect that there will be less information available to U.S. investors about firms in segmented economies where their investment opportunities are rather limited compared to firms in nonsegmented economies to which U.S. investors have easier access.

A common proxy for market segmentation in the literature has been the level of development of the home country economy (see for instance, Miller 1999; Foerster and Karolyi 2000). Emerging economies tend to have (1) higher restrictions on capital flows so as to prevent the exchange rate from deteriorating, (4) (2) higher ownership restrictions, (3) higher inflation rate, and (4) higher volatility in exchange rates than more developed economies. Most importantly, restrictions to capital flow can severely segment the global capital market (see for example, Stapleton and Subrahmanyam 1977), making the information events (such as cross-listing announcements and earnings announcements) of firms from segmented economies to have different implications for firm value and cost of capital relative to those of firms from nonsegmented economies.

In support of the market segmentation hypothesis, which predicts that the stock price reaction to the listing will differ across securities in ways that are related to the severity of restrictions if share value is influenced by intemational restrictions to capital flows, Miller (1999) documents that the U.S. market reacts more strongly to the cross-listing announcement of an ADR finn from an emerging economy than from a developed economy. This evidence suggests that cross-listing a security mitigates segmentation by improving the imbalance in the information environments.

However, if the market remains segmented even after cross-listing, as prior studies suggest (see Bekaert and Harvey 1995), then the market's reaction to subsequent information events that trigger its revision of future firm prospects and thus firm value (such as earnings announcements) will also be systematically different. Based on Atiase (1985), who finds that the market's reaction at the time of earnings announcement is inversely related to the amount of pre-disclosure information available, one can predict that the U.S. market's reaction to the earnings announcements by ADR-issuing firms from an emerging economy will be, on average, larger than the reaction to the earnings announcements of their counterparts from a developed economy if the market remains segmented even after cross-listing (i.e., emerging economy ADRs have less pre-disclosure information available in the market than developed economy ADRs). In this vein, I develop the following hypothesis:

[H.sub.0]: The market's reaction to the U.S. GAAP earnings announcements of non-U.S. firms cross-listed in the U.S., whose home country economy is better characterized as emerging, will be larger than its reaction to the earnings announcements of firms whose home country economy is developed.

IV. SAMPLE AND METHODOLOGY

To be included in the sample, a firm has to meet the following requirements: (1) be listed in the Bank of New York (2000) Global Equity Investment Guide; (5) (2) have quarterly earnings announcement date available on Compustat; and (3) have other required accounting and stock return data available on Compustat and CRSP. The sample period is 1992 to 1999. Abnormal returns are market-adjusted returns available on the CRSP excess returns tape. To control for the effects of seasonality in earnings, unexpected earnings are computed as the current quarter's earnings minus earnings from four quarters before. 1,008 firm-quarter observations that meet these conditions are initially identified. (6) After eliminating observations that fall in the top and bottom 0.5 percent of each variable (see for instance, Fama and French 2000), 1,001 firm-quarter observations are used for the empirical analysis.

Shown in Table 1 are countries represented in the sample, along with a classification of each country into the developed/emerging category. There are 34 countries represented in the sample, and the U.K. is the most represented with 213 firm-year observations, followed by The Netherlands (n = 133) and Mexico (n = 118). Table 2 shows that firms are fairly well distributed across different industries, with telecommunication representing the largest number of firm-quarter observations (n = 119), followed by oil and gas (n = 108) and pharmaceutical (n = 100). There are a total of 31 industries represented in the sample, based on the industry classification scheme provided by the Bank of New York Global Equity Investment Guide. (7)

The last column of Table 1 shows the internationalization rank, which measures the extent to which a country participates in international trade and international flows. The more internationalized the country, the higher the ranking. The mean internationalization rank is 13.78 in the developed economy sample, while it is 30.64 in the emerging economy sample; the difference is statistically significant with a t-statistic of -8.50 (p<0.001). This suggests that the degree of internationalization varies significantly with the level of economic development, lending support to the use of level of economic development as a proxy for market segmentation.

Panel A of Table 3 displays descriptive statistics of selected variables. The average cumulative abnormal return (hereafter CAR) over the three-day period around earnings announcements in the emerging (developed) economy sample is -0.1 percent (-0.4 percent), and it ranges from -21.7 percent to 20.3 percent (from -24.8 percent to 20.1 percent). Firm size, measured as the log of total assets at the end of each quarter, has an average value of 7.678 and ranges between 4.528 and 12.902 in the emerging economy sample. In contrast, it has an average of 9.237, with a range of 4.692 and 13.130 in the developed economy sample. The difference in average firm size is statistically significant (p < 0.01 ), suggesting that developed economy firms tend to be larger than emerging economy firms.

Consistent with this result, Panel B of Table 3 shows a strong correlation between firm size and the level of development of home country economy (both Pearson and Spearman correlations are statistically significant at p < 0.01). This result generally supports the view that emerging economy firms have less pre-disclosure information available to investors relative to their developed economy counterparts when firm size and listing location are used as proxies for information environment. (8) Panel C of Table 3 shows that the majority of firms included in the sample trade on major stock exchanges, i.e., NYSE and AMEX.

To test for the market's reaction to earnings announcements, I start from the following conventional returns-earnings regression model, where three-day CAR around the quarterly earnings announcement date (day -1 to day +1) appear as the dependent variable, and unexpected earnings ([DELTA][E.sub.i,t]), measured as the change in earnings from four quarters before the announcement, deflated by beginning-of-the-period price, appear as the test variable.

(1) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Next, in order to test for a statistical difference in the magnitude of the earnings response coefficients in the two groups of firms, I introduce a dummy variable that represents the level of economic development (i.e., CDUMMY), to the previous specification. The dummy variable, CDUMMY, takes the value of 0 (1) if the home country economy of a cross-listed firm is categorized as emerging (developed). Under this specification, coefficient [[PSI].sub.2] measures the difference (i.e., the developed economy increment) in the magnitude of earnings response coefficients (both levels and changes) between the two groups of firms. A negative and significant [[PSI].sub.2] would mean that the market reacts more to the earnings announcement of firms from emerging economies.

(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

However, this specification imposes a rather restrictive assumption about the time-series property in earnings (i.e., random walk). Ali and Zarowin (1992) show that using both levels and changes in earnings, rather than just changes in earnings, better measures the relation between abnormal/ unexpected stock returns and unexpected earnings. Specifically, they note that this specification is robust to the presence of transitory components in annual earnings because earnings levels act as an additional proxy for unexpected earnings. In this vein, I also estimate another model, where I augment the right-hand side of Equation (1) with earnings level deflated by the beginning-of-the-period price (i.e., [E.SUB.i,t]) as in the following:

(3) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Under this specification, the response coefficient for unexpected earnings would be [[PHI].sub.1] + [[PHI].sub.2]. To test for the significance of these coefficients jointly, I transform Equation (3) as in the following:

(4) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

These equations (i.e., Equations (3) through (6)) are mathematically equivalent, and coefficient [[PSI].sub.1] in Equation (6) is equivalent to the summation of coefficients on levels and changes in earnings in Equation (3). Estimating Equation (6) is preferable to estimating Equation (3) for the purpose of testing my hypothesis given the dummy variable I will introduce. To test for a statistical difference in the magnitude of earnings response coefficients in the two samples using Equation (6), I use the model specification in Equation (7). A negative and significant [[PSI].sub.2] would mean that the market reacts more to the earnings announcements of emerging economy firms.

(7) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Next, in order to examine whether the association between the level of economic development and the returns-earnings relation may be affected by other factors of a firm's information environment, I re-estimate Equation (7) after controlling for firm size (see for instance, Atiase 1985; Meek 1991) and listing location (see for instance, Grant 1980). To achieve this objective, I form dummy variables for firm size and listing location. First, 1 use the median of firm size to demarcate the sample, and assign the value of 0 (1) for firms that fall below (above) the median.

Next, I divide firms into two groups depending on whether their ADRs are traded on a major stock exchange (i.e., AMEX or NYSE), and assign the value of 1 if the firm's ADR is traded on a major stock exchange, and 0 otherwise. Both of these dummies are designed to take a higher value (i.e., 1) as the firm's information environment gets richer. I also introduce additional interaction terms to examine the incremental effects of firm size and listing location on explaining the returns-earnings association beyond the level of economic development (i.e., coefficients [[PSI].sub.4] and [[PSI].sub.5]). If firm size and listing location do not fully explain the association between the level of economic development and the information content of earnings, then [[PSI].sub.1] will remain significant.

(8) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

V. RESULTS AND DISCUSSIONS

The results from estimating Equations (2) and (7) are shown in Table 4. In model 1, where the random walk process in earnings is assumed, [[PSI].sub.1] is positive and significant (p < 0.01) while [[PSI].sub.2] is negative and significant (p < 0.05). These results are consistent with the prediction that the market will react more (less) to the earnings announcements of emerging (developed) economy firms. In model 2, [[PSI].sub.2], the coefficient of interest, which tests for a difference in the information content of earnings announcements between the two samples, has a negative sign as predicted, and it is statistically significantly different from zero. (9)

However, [[theta].sub.1] is insignificant, suggesting that the market participants do not assume that the earnings follow a random walk process in setting their expectations. (10) Overall, these results suggest that the market reacts systematically less (more) to the earnings announcement of developed (emerging) economy firms even when the strict assumption about the time-series properties of earnings is relaxed. (11) It also suggests that the market remains segmented even after cross-listing in the sense that the imbalance of information environment between the two groups of firms persists.

Next, Table 5 reports results from estimating Equation (8), which augments Equation (7) with additional interaction terms for firm size and listing location. To the extent that the documented association between the level of economic development and the information content of earnings is explained by the two proxies for pre-disclosure information environment, [[PSI].sub.1] will no longer be significant when the additional interaction terms are included in the model. The result shows that [[PSI].sub.1] stays significant even after the interaction terms are included, suggesting that the main finding of this study is robust to the inclusion of other common proxies for a firm's pre-disclosure information environment. (12) It also shows that the returns-earnings association decreases with firm size and listing location, but only listing location is statistically significant. (13)

Prior literature also finds that earnings persistence is an important determinant of the returns-earnings relation (see for instance, Kormendi and Lipe 1985). In light of this finding, it is possible that the association I document in this study can be explained by the fact that the earnings of ADR firms from emerging economies have higher persistence than those of ADR firms from developed economies. To test for this possibility, I estimate the model specified in Equation (9), which tests for a difference in the persistence of earnings in the two groups of firms. As in the preceding models, [E.sub.i,t] refers to earnings level for firm i during period t, deflated by the beginning-of-the-period price. If the earnings of emerging economy firms have higher persistence, then [[PHI].sub.3] will be negative and significant.

(9) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

The results reported in Table 6 show that although [[PHI].sub.3] has a negative sign, it is not statistically different from zero in a significant manner. This suggests that there is no statistical difference in the persistence of earnings between the two groups of firms. Therefore, I conclude that a difference in earnings persistence is unlikely to be driving the main findings of the study. (14)

VI. CONCLUSIONS

In this study, I test the global market segmentation hypothesis, which predicts a larger market reaction to the U.S. earnings announcement of firms from segmented economies. Supporting my hypothesis, the results indicate that the market's reaction to the earnings announcement of emerging economy firms is on average larger than its reaction to developed economy firms. The evidence I present in this study is consistent with the view that cross-listing of security itself does not necessarily fully integrate otherwise segmented markets (see for instance, Bekaert and Harvey 1995).

The result also suggests that the level of economic development of the home country of an ADR program is an important determinant of the information content of U.S. earnings reports prepared by cross-listed firms. In this regard, my study extends prior literature on the determinants of the information content of earnings announcements, and suggests that future research that examines the information content of earnings announcements of multinational firms should control for the level of economic development of the home country, in addition to other known proxies for pre-disclosure information environment such as firm size and listing location, in the research design.

The evidence I present in this manuscript should be interpreted subject to the following caveats. One concern is that I do not control for the effects of home country earnings announcements. While I believe that this not a serious concern in view of Amir et al. (1993) that in most cases in their sample the two earnings are simultaneously known to U.S. investors, one cannot completely rule out the possibility that the results I document have been affected by the preemption of earnings information for the developed economy ADRs. (15)

Another limitation of my study is the potential for sample selection bias. Since Compustat collects and reports only the earnings announcement dates of ADR firms that are released through various newswires, the sample may not be representative of all ADR firms currently trading in the U.S. Finally, readers should bear in mind that interpretation of the results presented in this study is subject to assumptions about the fundamental asset-pricing model, inefficiency in the market, and rejection of market integration (see for instance, Bekaert and Harvey 1995). (16)

TABLE 1
Home Country Distribution of Sample Firms

                      Number of      Level of
                        Firm-        Economic     Internationalization
Home Country         Quarter Obs.   Development         Rank

Argentina                24             E                21
Australia                36             D                27
Brazil                   13             E                39
Chile                    44             E                23
China                    15             E                20
Denmark                  12             D                12
Dominican Republic        4             E                NA
Finland                   4             D                11
France                   13             D                10
Germany                  18             D                 8
Ghana                    20             E                NA
Greece                    3             D                35
Hong Kong                 3             D                 3
Hungary                   2             E                26
Indonesia                 2             E                36
Ireland                  61             D                 7
Israel                   18             E                33
Italy                    41             D                19
Japan                    82             D                34
Korea                     3             E                NA
Luxemburg                 3             D                 4
Mexico                  118             E                40
The Netherlands         133             D                 6
New Zealand               4             D                16
Norway                   36             D                15
Philippines              38             E                28
Portugal                  4             D                13
Russia                    2             E                30
Singapore                 4             D                 2
South Africa              2             E                NA
Spain                    14             D                18
Switzerland               4             D                17
United Kingdom          213             D                 5
Venezuela                 8             E                41

Total                 1,001

The classification of the level of development of the home country
economy is based on the International Finance Corporation's (IFC)
scheme. IFC is under the umbrella of the United Nations. The
internationalization rank is taken from Pocket World in Figures
(Economist 1999). This measure is based on the extent to which a
country participates in international trade and international flows.
The more internationalized the country, the higher the ranking.

TABLE 2
Industry Distribution of Sample Firms

                          Number of      Number of
                        Firm-Quarter    Firm-Quarte
                        Observations    Observations
                          (Emerging      (Developed
Industry                Economy ADRs)   Economy ADRs)

Airlines                     20              28
Automobile                    0              29
Banking                       5              33
Beverage                     16               7
Chemicals                     3              32
Consumer Electronics          0              20
Construction                 14               4
Electrical Equipment          0              31
Engineering/Machinery         0              14
Financial Services            0               3
Food/Tobacco                  7              37
Household Products            1              20
Insurance                     0              35
Luxury Goods                  8               0
Media/Entertainment          31              29
Mining/Minerals              21              17
Multi-Industry               33              30
Oil/Gas                      21              87
Packaging                     0               6
Paper                        12               8
Pharmaceutical                0             100
Printing/Publishing           0               3
Real Estate                   1               0
Retailing                    17              14
Steel                        19               3
Technology/Hardware           0              15
Technology/Software           0               2
Telecommunication            63              56
Textiles                      0               9
Transportation               16               0
Utilities                    18               3

Total                       326             675

The industry classification is taken from the Global Equity Investment
Guide issued by Bank of New York (2000).

TABLE 3
Other Sample Characteristics

Panel A: Descriptive Statistics

Panel A-1: Emerging Economy Sample

         Mean   Median   Std. Dev.   Maximum   Minimum

CAR    -0.001   -0.002     0.066       0.203   -0.217
EP     -0.001   -0.013     0.086       0.125   -0.882
SIZE    7.678    7.786     1.214      12.902    4.528

Panel A-2: Developed Economy Sample

CAR    -0.004    0.005     0.049       0.201   -0.248
EP      0.014    0.015     0.051       0.151   -0.809
SIZE    9.237    9.529     1.824      13.130    4.692

Panel B. Correlation Matrix

          DEVELOP       SIZE       LIST

DEVELOP               0.405 ***   0.014
SIZE      0.431 ***               0.132 ***
LIST      0.014       0.129 ***

Panel C: Listing Location

             Number of Firm-Quarter     Number of Firm-Quarter
                 Observations                Observations
Location   (Emerging Economy Sample)   (Developed Economy Sample)

AMEX                 17                            4
NYSE                295                          646
OTC                   6                           24
PORTAL                8                            1

Total               326                          675

*, **, and *** indicate significance at 10, 5, and 1 percent levels
(two-tailed), respectively.

The classification of the level of development of the home country
economy is based on the International Finance Corporation's (IFC)
classification scheme. The IFC is under the umbrella of the United
Nations. Cumulative abnormal returns are three-day (day  -1 to day +1)
market-adjusted returns available in the CRSP excess returns tape.
Earnings are both level and change in earnings from four quarters
before the announcement, scaled by price. SIZE is measured as the log
of total asset at the end of the quarter. In Panel B, Pearson
(Spearman) correlations are above (below) the diagonal.
The classification of listing location and region is based on the
International Finance Corporation (IFC)'s classification scheme.

TABLE 4
Difference in the Magnitude of ERCs in the Two Samples
(n = 1,001)

(2) [CAR.sub.i,t] = [[PHI].sub.0] + [[PHI].sub.1] CDUMMY + [[PSI].sub.1]
[DELTA][E.sub.i,t] * CDUMMY + [[epsilon].sub.i,t]

(7) [CAR.sub.i,t] = [[PHI].sub.0] + [[PHI].sub.1] CDUMMY +
[[PSI].sub.1][E.sub.i,t] + [[PSI].sub.2][E.sub.i,t] * CDUMMY +
[[theta].sub.1][E.sub.i,t 1] + [[theta].sub.2][E.sub.i,t--1] *
CDUMMY + [[epsilon].sub.i,t]

Equation (2)
Regression      Equation (7)     Coefficient     Coefficient
Variable       Predicted Sign   (t-statistic)   (t-statistic)

Intercept                          0.001           -0.001
                                  (0.13)          (-0.34)
CDUMMY              (-)           -0.003           -0.005
                                 (-0.96)          (-1.09)
[DELTA]
  [E.sub.t]         (+)            0.145
                                  (2.66) ***
[DELTA]
  [E.sub.t]
  * CDUMMY          (-)           -0.146
                                 (-1.85) **
[E.sub.t]           (+)                             0.101
                                                   (2.01) **
[E.sub.t]
  * CDUMMY          (-)                            -0.144
                                                  (-2.37) **
[E.sub.t-1]         (-)                             0.037
                                                   (0.73)
[E.sub.t-1]
  * CDUMMY          (?)                             0.127
                                                   (1.26)
Adjusted
  [R.sup.2]                        0.45%            2.30%
F-statistic                        2.47 *           2.44 *

*, **, and *** indicate significance at 10, 5, and 1 percent levels
(two-tailed), respectively.

The classification of the level of development of home country economy
is based on the International Finance Corporation's (IFC)
classification scheme. IFC is under the umbrella of the United Nations.
The listing location (exchange) and country classifications are also
based on the Global Equity Investment Guide issued by Bank of New York
(2000). Cumulative abnormal returns are three-day (day -1 to day +1)
market adjusted returns available in the CRSP excess returns tape.
[DELTA][E.sub.t] (unexpected earnings) is the change in earnings
relative to four quarters before, deflated by the
beginning-of-the-period price. [E.sub.t] is level earnings deflated by
the beginning-of-the-period price. CDUMMY is a dummy variable that
takes the value of 1 (0) if the home country economy of the firm is
categorized as developed (emerging).

White's (1980) heteroscedasticity-consistent standard error estimates
are used to compute the t-statistics.

TABLE 5
Difference in the Magnitude of ERCs in the Two Samples
After Controlling for Size and Listing Location
(n=1001)
                Predic-   Coeffi-                               F-
Regression       ted       cient                 Adjusted    statistic
Coefficient      Sign     Estimate   t-stat      [R.sup.2]   (p-value)

                                                   9.20%     14.28 ***
[[PHI].sub.0]    (?)       0.007      0.71
[[PHI].sub.1]    (?)       0.003      0.39
[[PHI].sub.2]    (?)      -0.004     -0.67
[[PHI].sub.3]    (?)      -0.011     -1.35
[[PSI].sub.0]    (+)       1.282      6.29 ***
[[PSI].sub.1]    (-)       1.337     -6.65 ***
[[PSI].sub.2]    (-)      -0.058     -0.14
[[PSI].sub.3]    (-)       1.247     -5.96 ***
[[PSI].sub.4]    (?)       0.489      1.11
[[PSI].sub.5]    (?)       1.109      4.48 ***
[[theta].
   sub.0]        (-)       1.312     -5.83 ***
[[theta].
   sub.1]        (+)       1.359      5.92 ***
[[theta].
   sub.2]        (+)       0.743      1.78 *
[[theta].
  sub.3]         (+)       1.320      5.75 ***
[[theta].
  sub.4]         (?)      -1.261     -2.87 ***
[[theta].
  sub.5]         (?)      -0.946     -3.56 ***

*, **, and *** indicate significance at 10, 5, and 1 percent levels
(two-tailed), respectively.

The classification of the level of development of home country economy
is based on the International Finance Corporation's (IFC)
classification scheme. IFC is under the umbrella of the United Nations.
The listing location (ex-change) and country classifications are also
based on the Global Equity Investment Guide issued by Bank of New York
(2000). Cumulative abnormal returns are three-day (day -1 to day +1)
market adjusted returns available in the CRSP excess returns tape.
Unexpected eamings are current quarter's earnings minus earnings from
four quarters before. SIZE is measured as the log of total asset at the
end of the quarter, deflated by beginning of the period stock price.
Industry fixed effects are included, but not reported.

TABLE 6
Differences in Earnings Persistence between the Two Samples
(n = 1,001)

(9) [E.sub.i,t] = [[PHI].sub.0] + [[PHI].sub.1] CDUMMY -
[[PHI].sub.2][E.sub.i,t-1] + [[PHI].sub.3][E.sub.i,t-1 *] CDUMMy
+ [[epsilon].sub.i,t]

Regression     Predicted    Coefficient
Variable         Sign      (t-statistic)

Intercept        (?)          -0.006
                             (-1.17)
CDUMMY           (?)           0.014
                              (2.06) **
[E.sub.t 1]      (+)           0.401
                              (2.31) **
[E.sub.t--1]
  * CDUMMY       (-)          -0.007
                             (-0.03)
Adjusted
  [R.sub.2]                   15.03%
F-statistic                    7.30 ***

*, **, and *** indicate significance at 10, 5, and 1 percent levels
(two-tailed), respectively.

The classification of the level of development of home country economy
is based on the International Finance Corporation's (IFC)
classification scheme. IFC is under the umbrella of the United Nations.
The listing location (ex-change) and country classifications are also
based on the Global Equity Investment Guide issued by Bank of New York
(2000). Cumulative abnormal returns are three-day (day -1 to day +1)
market adjusted returns available in the CRSP excess returns tape.
[E.sub.1] ([E.sub.t-1]) is (last year's) level earnings deflated by the
beginning-of-the-period price. CDUMMY is a dummy variable that takes
the value of 1 (0) if the home country economy of the firm is
categorized as developed (emerging).

White's (1980) heteroscedasticity-consistent standard error
estimates (1980) are used to compute the t-statistics.

I thank two anonymous reviewers and an associate editor for providing highly constructive comments. I also express a special thanks to R. S. Olusegun Wallace (the editor) for his insightful comments and encouragements.

(1) Overseas economies were classified into "developed" and "emerging" based on International Finance Corporation's (2000) Investment Regulations Summary (available at http://www.ifc.org). The International Finance Corporation (IFC), a member of the World Bank Group, shares the primary objective of all Bank Group institutions: to improve the quality of the lives of people in its developing member countries. IFC classifies a country into developed/emerging based on whether the country is/is not a donor to IFC. I have adopted the economic development dichotomy to capture a rich array of differences in the information environments of emerging versus developed economies including barriers to capital flows, share ownership restrictions, and multiple share classes that can be found in some emerging countries such as China and Mexico.

(2) As of February 2000, they represented approximately one-tenth of total U.S. shares, and their size has been growing rapidly at the rate of 30 to 40 percent annually (Global Equity Investment Guide, Bank of New York 2000).

(3) The 20-F reconciliation is analogous to a 10-K filing for a domestic (U.S.) company, but allows the cross-listed company to retain its local (non-U.S.) GAAP reporting for the primary financial statements provided to U.S. shareholders. The SEC mandates that the Form 20-F be filed within six months of the fiscal year-end.

(4) In fact, markets can be segmented not only because of restrictions on capital flows, but also because of the desire to protect national sovereignty, to limit the spread of ownership to foreigners and unique local taxes, inflation, and liquidity risk, among others. Adverse effects of market segmentation include high steady state returns due to high-segmentation risk premia. Higher costs of capital can lead to decreased investment and lower growth in the economy.

(5) I use this data source because information on the country of origin, industry membership, and listing location of cross-listed firms can all be obtained from this directory.

(6) A conversation with a Compustat representative has revealed that, prior to the year 2000, Compustat collected and reported the earnings announcement dates of ADRs that released earnings through various newswires. The sample size in this study is comparable to that of other studies that examine ADRs. For instance, Miller 0999) had 181 firm-year observations in his sample, and other studies often have smaller sample sizes.

(7) It appears that there is some degree of industry clustering between the three groups of finns (e.g., pharmaceutical, insurance, and automobile, among others).

(8) As a sensitivity check for a difference in the pre-disclosure information environment. I randomly select 20 firms from each sample, and count the number of Wall Street Journal appearances during a one-month period (i.e., from one month before the announcement to the date of the actual earnings announcement). Confirming the idea that developed (emerging) economy firms have relatively richer (poorer) information environment, I find that emerging economy firms tend to have fewer WSJ appearances than developed economy firms; while the average number of appearance is .47 times for an emerging economy firm, it is .88 for a developed economy firm.

(9) The results do not change when industry dummies are included in the model.

(10) If the market perceived earnings to follow a random walk process, then the coefficient on [E.sub.t] would be the same as the coefficient on [E.sub.t-1] but with the opposite sign. Since this does not appear to be the case, the use of both level and change earnings in the model seems warranted. It is also noteworthy that the [R.sup.2]s have a five-fold increase when the level earnings are included in the model.

(11) As another sensitivity check, I re-estimate Equation (5) with year dummies and industry dummies to assess the effects of year-specific macroeconomie events and industry-wide factors on the inference. The results remain qualitatively similar when the dummies are included, suggesting that the main finding of this study is robust to the choice of sample period and industry composition of the sample firms.

(12) One potential explanation is that the level of economic development captures a unique dimension that affects the information content of earnings. That is, it is possible that the "quality" of home country accounting standards between the two groups of firms may differ systematically. For instance, Lee (1987) contends that accounting infrastructure is related to the level of economic development. Thus, if most firms in the sample pre-announced their home country earnings and there was a systematic difference in their information content, then the documented results would be expected.

(13) A potential explanation for the insignificance of firm size, among others, is that it might not be a reliable proxy for a cross-listed firm's information environment. For example, the fact that the underlying economic conditions can vary widely by country and that the total asset figure (expressed in U.S. dollars) is affected by foreign exchange rate, which has little to do with a firm's information environment, might offer some insights for this result.

(14) Prior studies also report growth as another important determinant of the information content of earnings (Collins and Kothari 1989). Therefore, it is possible that the higher ERC observed in the emerging economy sample is due to the fact that investors have a higher growth expectation about the emerging economy firms. However, the data show that the developed economy firms in the sample in fact have a higher growth expectation (i.e., lower earnings-to-price ratio) than the emerging economy firms. Thus, it appears unlikely that differences in growth, like difference in earnings persistence, are driving the main finding of the study.

(15) In the Amir et al. (1993) sample, the two earnings were released simultaneously in approximately over three-quarters of the cases. Given that their sample firms are mostly from developed economies, I randomly select 15 emerging economy firms from my sample and examine whether these firms pre-announce their home country earnings before the release of U.S. earnings. I find no clear evidence of emerging economy firms pre-announcing their home country earnings--only two out of the 15 firms I examined announced their home country earnings separately from their U.S. earnings. To err on the side of caution, I also re-estimate Equation (6) using returns from an extended window (from day -4 to day +4). The main finding of the study may not hold over such a longer event window if firms announcing their home country earnings separately from their U.S. earnings have systematically affected the inference. However, the results remain qualitatively similar (i.e., the [[PSI].sub.2] coefficient remains significant at p < 0.05) even when the return window is extended, suggesting that the inferences are unlikely to have been confounded by the home country earnings announcements of the sample firms.

(16) Assumptions about asset pricing models, however, are unlikely to be a major concern in light of Fama (1998), who contends that risk differences among investors have negligible effects on the inference in an event study that employs a short time window.

REFERENCES

Alford, A., J. Jones, R. Leftwich, and M. Zmijewski. 1993. The relative informativeness of accounting disclosures in different countries. Journal of Accounting Research Supplement 31:183-223.

Ali, A., and P. Zarowin. 1992. Permanent versus transitory components of annual earnings and estimation error in earnings response coefficients. Journal of Accounting and Economics 15: 249-264.

Amir, E., T. Harris, and E. Venuti. 1993. A comparison of the value-relevance of U.S. versus non-U.S. GAAP accounting measures using Form 20-F reconciliations. Journal of Accounting Research 31 (Supplement): 230-264.

Atiase, R. 1985. Pre-disclosure information, firm capitalization, and security price behavior around earnings announcements. Journal of Accounting Research 23:21-36.

Bank of New York. 2000. The global equity investment guide. New York, NY: Bank of New York. Also available at: http://www.adrbny.com/dr_pub_equity.jsp.

Bekaert, G., and C. Harvey. 1995. Time-varying world market integration. The Journal of Finance 50:403-444.

Collins, D., and S. P. Kothari. 1989. An analysis of intertemporal and cross-sectional determinants of earnings response coefficients. Journal of Accounting and Economics 11: 143-181.

--, --, and J. D. Rayburn. 1987. Firm size and the information content of prices with respect to earnings. Journal of Accounting and Economics 9:111-138.

Cooke, T. E., and R. S. O. Wallace. 1990. Financial disclosure regulation and its environment: A review and further analysis. Journal of Accounting and Public Policy 9 (2): 79-110.

Economist, The. 1999. Pocket World in Figures. London, U.K.: The Economist Books Ltd.

Errunza, V. R., and E. Losq. 1989. Capital flow controls, international asset pricing, and investors' welfare: A multi-country framework. The Journal of Finance 44: 1025-1037.

--, and D. P. Miller. 2000. Market segmentation and the cost of capital in international equity markets. Journal of Financial and Quantitative Analysis 35: 577-600.

Fama, E. F. 1998. Market efficiency, long-term returns, and behavioral finance. Journal of Financial Economics 49: 283-306.

--, and K. R., French. 2000. Forecasting profitability and earnings. The Journal of Business 73 (2): 161-175.

Foerster, S., and G. Karolyi. 2000. The long-run performance of global equity offerings. The Journal of Financial and Quantitative Analysis (December): 499-527.

Frost, C. A., and G. Pownall. 1994. A comparison of the stock price response to earnings disclosures in the United States and United Kingdom. Contemporary Accounting Research 11 (1): 59-83.

Grant, E. B. 1980. Market implications of differential amounts of interim information. Journal of Accounting Research 18 (1): 255-268.

International Finance Corporation. 2000. Investment regulations summary. Available at: http://www.ifc.org.

Kormendi, R., and R. Lipe. 1985. Earnings innovations, earnings persistence, and stock returns. The Journal of Business 60 (3): 323-345.

Lee, C. W. J. 1987. Accounting infrastructure and economic development. Journal of Accounting and Public Policy 6: 75-85.

Meek, G. K. 1991. Capital market reactions to earnings announcements in an international context. Journal of International Financial Management and Accounting (Summer): 93-109.

Miller, P. D. 1999. The market reaction to international cross-listings: Evidence from depository receipts. Journal of Financial Economics 51:103-123.

Stapleton, R., and M. Subrahmanyan. 1977. Market imperfections, capital market equilibrium, and corporate finance. The Journal of Finance 32:307-319.

Stulz, R. M. 1981. On the effects of barriers to international investment. The Journal of Finance 36:923-934.

White, H. 1980. A heteroscedasticity-consistent covariance matrix estimator and a direct test for heteroscedasticity. Econometrica 48 (4):817-838.

Tony Kang is an Assistant Professor at the Singapore Management University.

Submitted: April 19, 2002

Accepted: March 28, 2003