1 Introduction

Differences between US Generally Accepted Accounting Principles (GAAP) and International Accounting Standards (IAS) have received increasing attention over the last decade as the number of international firms accessing US capital markets through cross listing has increased dramatically. In 2005, almost 17% of listed companies on the New York Stock Exchange (NYSE) were from outside the United States. The total valuation of these companies is in excess of $1 Trillion (NYSE Research). Currently, US stock markets all allow non-US firms to use IAS, however, the Securities and Exchange Commission (SEC) requires these companies to provide a 20-F reconciliation of income statement and balance sheet items according to US GAAP. While international and domestic pressure has been put on the SEC to relax this requirement, the SEC has held that the high level of disclosure required by US GAAP is important to shareholder protection. In 2006, many additional companies will follow IAS, putting further pressure on the SEC. Footnote 1 Although the US Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) are working towards the convergence of US GAAP and international standards (the Norwalk Agreement, Release 2002-154) with various short-term and long-term projects, the remaining accounting differences are substantive and many obstacles still exist in the movement towards convergence (FASB website; Quinn 2004; Campbell et al. 2002; Reason 2002). Another complication is enforcement of standards. The SEC maintains investor protection for US exchanges through enforcement that is stronger than in other countries (La Porta et al. 1999; Reese and Weisbach 2002; Hung 2000; Quinn 2004).

Given the increasing number of international firms cross-listing in the United States and the differences in disclosure levels between differing accounting standards, firms may see value in adopting higher disclosure standards through cross-listing. Bailey et al. (2006) provide several reasons that may motivate a company to increase disclosure including reducing information asymmetry, reducing the cost of financing and increasing analyst following. In this study, we use a sample of foreign firms that cross-list on the NYSE to examine the relationship between the cost of equity capital and disclosure quality. Specifically, we develop an international asset-pricing model to measure disclosure risk and systematic risk for international firms in the initial year of listing in the US Additionally, we examine the role of analyst following and shareholder protection though the cultural and legal system.

Prior research has investigated this issue in several ways. Much of the research has focused on reconciliation effects and associated enforcement including examinations of the size and frequency of 20-F reconciliations (e.g., Amir et al. 1993), the relevance of such reconciliations (e.g., Harris and Muller 1999), and differences in reconciliations across different jurisdictions and across different country specific variables (e.g., Ali and Hwang 2000). Ashbaugh and Olsson (2002), Leuz (2003), and Bartov et al. (2005) explore the accounting valuation effects of IAS and US GAAP.

However, effective mechanisms of disclosure other than the 20-F reconciliation may exist. Healy and Palepu (2001) list several aspects of disclosure. Disclosure can be accomplished through mandatory reporting requirements, voluntary reporting and information intermediaries such as financial analysts. For example, in a non-international context, Botosan (1997) finds some evidence supporting a link between disclosure quality and analyst following. As the role of analysts is to uncover information and disseminate conclusions, analysts can be a substitute for disclosure through accounting statements. Indeed, Baker et al. (2002) show that analyst following is an important aspect of visibility in the international context. Lang et al. (2003a) and Bailey et al. (2006) show that analysts also have improved forecast accuracy for non-US companies listing in the United States. Thus, the increases in disclosure associated with cross-listing could be due to regulatory reporting requirements such as the 20-F reconciliation or due to the effectiveness of information intermediaries such as level of analysts following, or other substitutes for disclosure.

The impact of disclosure quality can be measured through the firm’s cost of equity capital. Low disclosure quality entails risk to the shareholders. In return for bearing this risk, shareholders demand a higher expected return (Cheng et al. 2006) . Many of the prior investigations on 20-F reconciliations use a proxy for the cost of equity. A stock market based approach includes annual stock return; where-as an accounting statement approach includes return on equity. However, Botosan (1997) provides some critical comments of empirical investigations into the association between disclosure level and the cost of equity capital when they use only variables correlated with the cost of equity and do not account for systematic risk. Botosan and Plumlee (2002) extend the analysis, including systematic risk (beta) and report timing. More recently, Cheng et al. (2006) simultaneously perform joint tests of disclosure and shareholder rights. They find that a combination of increased disclosure coupled with stronger shareholder rights results in a lower cost of capital.

In this paper, we hypothesize that non-US firms cross-listing on the NYSE will experience a decrease in the cost of equity capital. We find that cross-listing firms experience a decrease in the cost of equity of −0.204% per week. This decrease is not present in a matched group of similar companies, which do not cross-list. The magnitude of the decrease in the cost of equity is hypothesized to be related to the quality of disclosure required in the firm’s home country as well as the amount of disclosure from analysts. That is, if a foreign firm is from a country that already requires a relatively high level of disclosure, that firm may benefit very little from cross-listing in the United States. However, a firm from a low disclosure country could benefit greatly from cross-listing.

We examine three sources of disclosure: accounting standards; analyst following; and exchange/regulatory investor protection. We find that the magnitude of the observed decrease in the cost of equity is related to the level of disclosure firms have prior to cross-listing. Specifically, firms from countries with low levels of accounting disclosure benefit from the added disclosure of cross listing. Also, firms with low analyst following or from countries with relatively low levels of exchange/regulatory disclosure benefit the most from cross-listing.

Our results suggest that increased disclosure is beneficial to both investors and listing firms. However, disclosure can also be accomplished through mechanisms other than through accounting standards, such as analyst following. In addition to academics, this paper should be of interest to the exchanges and regulators (government and private) who set disclosure standards, as well as firms considering cross-listing outside their home country.

The remainder of this paper proceeds as follows. Section 2 develops our hypotheses. The research methodology and data description are presented in Sect. 3. Section 4 provides the results. Finally, Sect. 5 presents the discussion and conclusions.

2 Hypothesis development

Disclosure is a very important component of the efficient functioning of a capital market. Healy and Palepu (2001) provide an extensive literature review and discussion of factors influencing corporate disclosure policy. Disclosure can be accomplished through mandatory reporting (e.g., financial statements, regulatory filings with the SEC), voluntary reporting (e.g., press releases) and intermediaries (e.g., analysts). Within the capital market system, information asymmetries can develop, that can potentially be mitigated by regulation that provides fuller disclosure. The SEC is concerned with disclosure in the capital markets as seen in requirements such as Regulation Fair Disclosure (FD) that requires companies who disclose material information to others (e.g., analysts) to publicly disclose that information with a goal of eliminating selective disclosure (Irani 2004). Bailey et al. (2006) provide several reasons for why a company may wish to increase disclosure including: reducing information asymmetry, reducing the cost of financing, making the firm more attractive to investors, and increasing analyst following because of the lower cost of information acquisition for analysts.

The theoretical link between disclosure and the cost of equity capital is motivated by both Merton’s (1987) model of incomplete information and Easley and O’Hara’s (2004) model of asymmetric information. Merton proposes a two-factor model in which, in addition to the market factor, there is an information incompleteness factor. We discuss this model in the next section. Easley and O’Hara present a model in which firms have various levels of private and public information. The existence of private information becomes a risk factor for uninformed investors. The higher the level of private information, the higher the risk premium required of the firm. Thus, a high degree of information asymmetry causes a high cost of equity capital. An outcome of the model is that managers can influence their cost of equity by affecting the precision and quantity of information available to investors. Easley and O’Hara suggest how this can be done:

“This can be accomplished by a firm’s selection of its accounting standards, as well as through its corporate disclosure policies. Attracting an active analyst following for a company can also reduce a company’s cost of capital, at least to the extent that analysts provide credible information about the company. Yet another way to influence its information structure is through the firm’s choice of where to list their securities for trading. Since investors learn from prices, the microstructure of where a firm’s securities trades can influence how well and how quickly new information is reflected in the stock price.”

Thus, the Easley and O’Hara model illustrates the importance of accounting standards, and analyst following for a firm’s cost of capital. These two types of disclosure have received a considerable amount of individual attention in the literature.

While the United States uses standards (GAAP) set by the Financial Accounting Standards Board, non-US firms use either their home country’s standards or international standards (IAS) set by the IASB . If a non-US firm wishes to be listed in the United States and the firm uses IAS (or any other non-US GAAP), currently the SEC requires those companies to provide 20-F reconciliations for key financial statement data between IAS and GAAP. The SEC is under increasing pressure to relax these additional disclosure requirements, which is increasing as IAS gains greater acceptance such as the recent requirement for companies listed on the European Market to follow IAS.

A number of papers have investigated whether there is any value in requiring firms cross-listing in the United States to provide 20-F disclosure reconciliations between IAS and GAAP. Much of the prior research addressing this question has involved examination of accounting earning comparisons across countries and their predictive value. For example, Amir et al. (1993) regress earnings and specific 20-F reconciliation components to returns. They find that material differences emerge between US GAAP and international measures, which they interpret as value relevant. Rees and Elgers (1997) draw upon the work of Amir et al. but focus only on initial registrations. Regressing earnings data on both market to book ratios and annual returns, they find that differences between GAAP and international earnings measures are not significant. They conclude that the data available in SEC disclosures is already incorporated into the market. Harris and Muller (1999) turn their attention specifically to differences between US GAAP and IAS standards reported in 20-F reconciliations. They find that the differences are smaller in magnitude between US GAAP and IAS standards than other international differences reported in previous studies. Regressions of their earning differences on market value, price and returns provide mixed evidence. IAS earnings are associated with price whereas GAAP earnings are associated with returns. They conclude that SEC 20-F reconciliations are still value relevant. However, recent studies comparing US GAAP and IAS for firms trading in Germany fail to find strong support for US GAAP having higher value relevance than IAS (Bartov et al. 2005; Leuz 2003).

In contrast, Ali and Hwang (2000) incorporate companies from many countries and examine country-specific factors including sources of GAAP (private/government), accounting clusters (British-American/Continental), financial-tax alignment, and spending on audit services. They find more value relevance for 20-F reconciliations when countries set their GAAP primarily through the government, are within the Continental accounting cluster, tax alignment is greater, lower amounts are spent on auditing services and when countries are more oriented towards banks rather than financial markets.

In a non-international context, Botosan (1997) and Botosan and Plumlee (2002) find evidence consistent with the hypothesis that higher disclosure levels lower the cost of capital. Footnote 2 Foreign firms cross listing in the United States are required to file 20-F reconciliations. If the home country accounting standards of a firm provide adequate disclosure, then cross-listing and the resulting 20-F reconciliation will not provide improved disclosure. In this case, the firm’s cost of equity will not be affected. However, if its accounting standards are not adequate, the cost of equity will be reduced by increasing disclosure quality through the reconciliation. The above studies suggest that this will benefit firms from countries with lower levels of accounting disclosure than the United States. This then gives rise to our first hypothesis:

H 1

Firms cross-listing in the United States will experience a reduction in their cost of capital if United States accounting disclosure levels are higher than in their home country.

However, effective mechanisms of disclosure other than the mandatory reporting requirements of the 20-F reconciliation may exist. Healy and Palepu (2001) list several aspects of disclosure. One type of disclosure is through information intermediaries such as financial analysts. Botosan (1997) suggests that financial analysts following the firm provide an additional source of disclosure that may substitute for accounting disclosure. Footnote 3 Specifically, firms with a high level of disclosure experience a lower cost of equity. But firms with lower levels of disclosure can still experience a lower cost of equity by having a high analyst following. Botosan finds empirical support for her hypothesis in a non-international environment. However analyst following is very relevant in an international context as well. Baker et al. (2002) show that foreign firms increase analyst following by an average of 6 analysts the year after cross-listing on the NYSE. They argue that a lack of adequate information about a firm imposes a “shadow cost” on the firm’s cost of equity. They report that the change in analyst following associated with cross-listing is related to the change in a firm’s shadow cost.

In a related paper, Lang et al. (2003a) find a relationship between analyst coverage and forecast accuracy for a sample of firms that cross-list in the United States. They also find that changes in analyst coverage and forecast accuracy are correlated with changes in firm value following cross-listing. Lang et al. (2004) show that these findings are even greater for countries that have the least protection for minority shareholders. Since previous studies support the notion that analyst following is another form of disclosure, our second hypothesis is:

H 2

Firms with a low number of analysts before cross-listing in the United States will experience a reduction in their cost of capital.

Huddart et al. (1999) suggest that firm management may have strong incentives to list on high-quality disclosure exchanges. The primary concern the SEC has in requiring cross-listing firms to provide earning reconciliations is to protect investors. The SEC maintains investor protection through enforcement that is stronger than in other countries (La Porta et al. 1999; Reese and Weisbach 2002; Hung 2000; Quinn 2004).

One argument that critics have made towards the SEC is that such disclosures are costly and punitive. However, Huddart et al. argue that a higher disclosure exchange results in lower trading costs and greater liquidity. Information asymmetries develop at lower disclosure exchanges where insiders have an opportunity to make larger abnormal returns on incomplete information. This results in wider spreads due to higher adverse selection costs. Amihud and Mendelson (1986) argue that the cost of capital is directly related to transaction costs, since shareholders want to be compensated for the costs of holding a firm’s securities. Thus, wider spreads require listing firms to pay a higher cost of capital to reward shareholders for these risks. Therefore, firms have an incentive to move to higher disclosure markets to reduce risks and the cost of capital.

The final approach examines the disclosure provided through regulatory enforcement of laws. La Porta et al. (1999) argue that countries such as the US, which have stronger investor protection through high-quality enforcement of standards, have more developed capital markets. They argue this is because the increased protection gives the investor greater confidence to provide capital. Reese and Weisbach (2002) argue that cross-listings on an SEC enforced exchange can provide an opportunity for non-US companies to voluntarily bond themselves to provide shareholder protection and increase equity. Hung (2000) hypothesizes that investor protection can impact the relevance of accrual accounting numbers. Hung finds that accrual accounting adversely affects accounting numbers for countries with weak shareholder protection but not countries with strong shareholder protection. Lang et al. (2003b) find that firms that choose to cross-list on a United States exchange have a higher level of disclosure than similar firms in their home country, prior to cross-listing. They also find that cross-listing firms improve the quality of their accounting disclosure around the time of cross-listing. This suggests that firms that exceed their home country accounting disclosure level have an incentive to cross-list to bond themselves to higher disclosure standards as suggested in Coffee (2002). Therefore, some firms may adopt higher accounting disclosure as a result of cross-listing, while others may use the cross-listing as a signal that they have already adopted a higher disclosure standard. Effectively, these are equivalent paths in that both produce a higher level of disclosure.

In the United States, exchanges are self regulating organizations—charged by the SEC with enforcing shareholder protection standards established by law. In this study we examine international firms cross-listing on one US exchange—the NYSE. Therefore, exchange disclosure quality and exchange regulatory effects are intertwined. This study does not attempt to disentangle the impact of this type of exchange based disclosure. Instead we examine the combined effect of primary listing exchange and home country quality in our last hypothesis:

H 3

Firms cross-listing in the United States will experience a reduction in their cost of capital inversely related to their level of disclosure before cross-listing.

Since several of the variables that could be used to test the hypotheses described in this section may be correlated, it is important to sort out the incremental impact of each hypothesis. In the next section we describe the methods we use to test our individual hypotheses. Our method for sorting out the incremental impact of each hypothesis is discussed in a later section.

3 Methods and data

3.1 Measuring changes in the cost of equity

Botosan (1997) provides some criticism of previous empirical studies of the association between disclosure level and the cost of equity capital in that they examine the impact of disclosure on variables correlated with the cost of equity and not on the cost of equity itself. Specifically, she states that “...traditional methods of estimating cost of equity capital either provide no role for information (this is the case with the traditional Capital Asset Pricing Model) or yield measures of this cost that demonstrate little or no relation to market beta (the most well accepted measure of firm risk) and are difficult to defend as a result” (1997, p 324). We employ an empirical model to examine the cost of equity around cross-listing that includes both a role for information and market beta.

We begin with Merton’s (1987) model of capital market equilibrium with incomplete information. Merton proposes a two-factor model of expected returns in which one factor is market risk and the other is an information completeness factor. Specifically, the expected return for firm k, E(R k ), is defined by:

$$ E(R_k)=r_{\rm f} +\beta_k(R_{\rm M} -r_{\rm f})+\lambda_k -\beta_k \lambda_{\rm M} $$
(1)

where r f is the riskless rate, R M is the expected return on the market portfolio, and β k is the firm’s systematic risk. The shadow cost to the firm of providing incomplete information is denoted as λ k , where λM is the average shadow cost for all securities. Firms with lower information disclosure than average will have a higher shadow cost, λ k , and thus a greater expected cost of equity capital, E(R k ).

The discrepancy between the security’s expected return in (1) and the traditional Security Market Line represents the incomplete information premium and is

$$ \alpha_k =\lambda_k -\beta_k \lambda_{\rm M}. $$
(2)

Substituting (2) into (1) and rearranging, it follows that

$$ E(R_k)-r_{\rm f} =\alpha_k+\beta_k (R_{\rm M} -r_{\rm f}). $$
(3)

This equation provides two theoretical measures of the cost of equity capital. Foreign firms cross-listing on the NYSE and reconciling differences in accounting standards may experience a change in the cost of equity due to greater disclosure (a reduction in α k ) or a decrease in market risk, β k . A decrease in our estimate of market risk may occur when the local country’s market is segmented from the world market. Foerster and Karolyi (1999) suggest that cross-listing could allow the firm to span the segmentation. In our tests, we focus on the change in the cost of equity from before cross-listing to afterwards. That is, the parameters α k and β k . should decrease after the reconciliation and cross-listing if the greater disclosure is relevant to investors. Footnote 4

We begin building our empirical model by proposing the following regression of weekly returns:

$$ R_{it}-r_{{\rm f}t} =\alpha_i+\beta_i(R_{{\rm m}t}-r_{{\rm f}t})+\varepsilon_{it} $$
(4)

where R it and R m t are the returns for firm i and the market portfolio (based on the firm’s home country) during week t, respectively, and r f t is the risk free rate. The parameters, α i and β i , are the estimates for firm i’s incomplete information premium and market risk. However, we are interested in the change in α i and β i . Therefore, we estimate a model using 52 weeks of returns before the cross-listing and 52 weeks after the cross-listing. Using a dummy variable, D POST t , that indicates a one when the week is after the cross-listing and a zero otherwise, we specify

$$ R_{it}-r_{{\rm f}t}=\alpha_i+\beta_i(R_{{\rm m}t}-r_{{\rm f}t})+\alpha_i^{\rm Post} D_t^{\rm POST} +\beta_i^{\rm POST}(R_{{\rm m}t}-r_{{\rm f}t})D_t^{\rm POST} +\varepsilon_{it}. $$
(5)

The estimates α POST i and β POST i represent the change in the incomplete information premium and market risk after the reconciliation and cross-listing. Footnote 5 We use these measures as direct tests of the change in the cost of equity associated with cross-listing from countries with different accounting standards.

3.2 Data

This study examines firms cross-listing on the NYSE, which due to relatively high NYSE listing standards, limits our sample to larger firms. Footnote 6 The largest firms are more likely to have higher levels of disclosure than small firms. So we believe this sample is biased against finding differences in cost of capital changes related to disclosure. Therefore, our results should be considered conservative estimates. We obtain from the NYSE a list of non-US securities listed on the NYSE and their listing date. We limit our sample by only including common stocks that are listed in their home market at the time of listing on the NYSE, and were not previously listed in the US. To avoid a survivorship bias, our sample includes securities that were later delisted.

We also construct a matching sample of similar firms that do not cross-list in the US before or during the event period. Our matching algorithm is similar to that of Errunza and Miller (2000), except we add an industry component. To find a matching firm for each of the sample firms, we use Disclosure’s Global Access database to locate other firms in the same home country and with the same four-digit SIC code. In the event of multiple potential matching firms, we select the firm whose total assets (1 year before the event date) were closest to the listing firm’s total assets (1 year before the listing). In many cases (e.g., Telefonos de Mexico), a suitable matching firm is unavailable, and hence is excluded from any matching firm analysis.

We next obtain daily stock price data for each firm in our sample listing for the 52 weeks before and 52 weeks after the listing week from the Datastream International database. Stock prices originate from the firm’s domestic stock exchange. This allows us to measure the change in the cost of equity from before to after the cross-listing. Studies investigating the efficacy of 20-F reconciliations use prices from the US market and thus can only compute the cost of equity after the cross-listing and not the change. Friday-to-Friday excess returns are then calculated from these prices. The domestic risk free rate is proxied using a local Eurodollar deposit rate obtained from Datastream. Footnote 7 To estimate Eq. 5, a local domestic market model is also required. We use the local market index to compute the market return.

A complication of our analysis is that the sample stock prices are denoted in local currencies and thus changes in exchange rates may represent a risk factor. If systematic exchange rate risk is priced, then this could impact the firm’s cost of equity. Jorion (1991) finds that the unconditional exchange rate risk premium is small and not significant. He concludes that there is no constant impact on the firm’s cost of equity. However, Patro et al. (2002) does find a significant time-varying (or conditional) risk premium. In addition, Carrieri et al. (2006) show that emerging market currency risk is strong during crises episodes. Converting all returns to a single currency (US dollars in this case) only masks the exchange rate risk, but does not eliminate it.

Using historical exchange rates obtained from Datastream, we convert all returns to US dollar returns. Only firms from our initial sample that have daily stock prices and exchange rates available on DataStream International for the 2-year period surrounding the cross-listing are examined in this study.

The number of analyst following a firm is obtained from I/B/E/S. Due to data availability on I/B/E/S and Datastream, the firms cross listing on the NYSE prior to 1985 need to be excluded (along with their matches). Our final sample consists of 94 foreign firms cross listing on the NYSE and 24 matching firms.

3.3 Descriptive statistics

Table 1 breaks down our cross-listing and matching samples by country. The samples consist of firms from 17 countries. Five of these countries have emerging capital markets while the rest have developed capital markets. The country with the highest representation in the sample is the United Kingdom with 20 firms. Canada has the second largest representation at 15 firms.

Table 1 Descriptive statistics
Full size table

Table 1 also contains descriptive statistics for the measures of disclosure quality used in this study. The first measure we examine is the number of analysts following each firm in our samples (cross listing and matching). The data are obtained from the I/B/E/S international and domestic data files. Following Baker et al. (2002), we obtain the number of analysts estimating the firm’s annual earnings 1 year before the exchange listing. Table 1 reports the mean number of analysts following cross-listing firms. The mean number of analysts following the sample firms is 5.6 analysts and 13.0 analysts before and after the cross-listing, respectively. The minimum is 0 and the maximum is 37. Firms from Argentina, Hong Kong, and Israel have very low analyst following. Firms cross-listing from France and Norway have high analyst following.

The next disclosure measure examined is accounting disclosure. We use two alternative measures: the source of GAAP and an index based on the level of accounting disclosure required in the home country. The source of GAAP refers to whether the government sets GAAP alone or whether the private sector is influential. Ali and Hwang (2000) propose that the source of the GAAP is an important factor in 20-F reconciliation (a requirement for cross-listing). In the US, the private sector sets the GAAP. Countries where GAAP is set by the government alone include Australia, France, Italy, Japan, and Norway. Countries where the private sector is involved include Canada, Denmark, Hong Kong, the Netherlands, and the UK.

The second measure of home country accounting disclosure quality used in this study is the accounting quality index obtained from Table 5 of La Porta et al. (1998). They state that the measure is computed by examining the inclusion or omission of 90 items in the annual reports. The country in our sample with the lowest reporting standard is Argentina (index level of 45) and the highest is the United Kingdom (index level of 78). As a comparison, the United States scored a 71. Therefore, for some firms, cross-listing in the US will result in an increase in accounting disclosure, while for others it results in a decrease. Accordingly, for our empirical tests, we construct a relative accounting disclosure measure defined as the US index value minus the firm’s home country accounting disclosure index. Thus the relative measure for Argentina is 71−45 = 26.

Huddart et al. (1999) suggest that exchange disclosure quality is inversely related to trading costs. Thus, we use trading costs as a proxy for the exchange disclosure quality for each stock’s home country. Jain (2003) contains information on trading costs for exchanges in 51 different countries. Jain provides several measures of trading costs. While quoted spreads (ask-bid) are sometimes used as a measure of trading costs, they do not take into account the fact that trades may occur inside the spread. Therefore, effective spreads, which take the location of trades into account, are generally viewed as a superior measure of trading costs. Effective spread is expressed as a percentage and is defined as \({\frac{2\left| {P-\frac{(bid+ask)}{2}}\right|}{\left[{\frac{(bid+ask)}{2}} \right]}}\), where P is the trade price; \({\frac{(bid+ask)}{2}}\) is the midpoint of the quoted spread at the time of the trade, and |·| is the absolute value operator. We obtain the average effective spreads for the 25 largest firms on each home country exchange from Jain’s Table 1.2.Footnote 8 Examining the last column of our Table 1 reveals that Danish firms have the highest effective spread (1.71%) and thus are assumed to have the lowest exchange disclosure quality. Canada, France, Korea, and Spain appear to have the highest exchange disclosure quality as evidenced by its relatively low average effective spread. The NYSE has a smaller average effective spread of 0.10%, which suggests it has the highest exchange disclosure of all the markets examined. Since all of the firms in our sample come from countries with higher effective spreads than the NYSE, no relative adjustment is necessary as it was for the accounting quality index.

In the next section we estimate the cost of equity, which will then be compared to the measures of disclosure quality examined in this section.

4 Empirical tests

4.1 Cross-listing and the cost of equity

We begin by estimating Eq. 5 for each of the 94 firms. The mean parameter estimates of these regressions are reported in Table 2. We postulate that these firms suffer from incomplete information disclosure and therefore the parameter, α i , is predicted to be positive. Since these firms tend to be the larger firms in their local stock markets, their measure of systematic risk, β i , should be positive. We argue that firms can reduce the incomplete information premium by improving their disclosure through cross-listing on the NYSE. A negative incomplete information premium coefficient in the post listing period, α POST i , is consistent with our hypothesis. Lastly, Foerster and Karolyi (1999) and Errunza and Miller (2000) suggest that market segmentation can be partially overcome through cross-listing. Cross-listing firms are able to orient toward a more diversified market portfolio than their home equity market can provide. Therefore, we predict that firms will experience a reduction in systematic risk, as measured on a local market portfolio. A negative β POST i coefficient is consistent with their hypothesis.

Table 2 Mean regression estimates for each firm, the following equation is estimated using 52 weeks of returns before and after cross-listing on the NYSE
Full size table

The first row of results shows the mean parameter estimates for all 94 firms. Reported t-statistics are used to determine one-tailed probability levels. The mean incomplete information premium estimate is a significant 0.223% per week. This indicates that these firms have a higher shadow cost of information that the average firm. However, after submitting 20-F reconciliations and cross-listing on the NYSE, the incomplete information premium experiences a significant decrease of −0.204% per week (p-value = 0.005). This is equivalent to a reduction of over 10% per year in the cost of equity. On average, these firms have an estimated market beta of 0.870. However, the market beta decreases by −0.055 after cross listing. The estimate is significant at 5% level. The observed reduction in beta is consistent with the segmentation spanning hypothesis of Foerster and Karolyi (1999).

Our estimate of a greater than 10% annualized decrease in the cost of equity is economically large and should be put into perspective. Using a different sample and methods, Foerster and Karolyi (1999) in their Table VI report that ADRs exhibit an annualized mean abnormal return of 7.8% in the year prior to listing. They also find that for their sample the annualized abnormal return decreases to 0.52% in the year after listing in the US Given that our sample contains more emerging market firms than Foerster and Karolyi (1999, 2000), our findings appear reasonable.

Also, Botosan (1997) estimates the difference that the quality of disclosure has on the cost of equity for US firms. She states (p. 323) that... “The magnitude of the effect is such that a one-unit difference in the disclosure measure is associated with a difference of approximately twenty-eight basis points in the cost of equity capital, after controlling for market beta and firm size.” Her disclosure measure is an index of the quality of disclosure for each firm. She shows that firms in the first percentile have an index level of 12 while the median firm has a measure of 29.6. Therefore, if the lowest firm increases disclosure to that of the median firm, her findings imply a decrease in the cost of equity of almost 5%. This estimate is half our estimate and the sample includes only US firms that already enjoy a relatively high degree of disclosure compared to foreign firms.

The mean parameter estimates are also reported by country of origin. As the sample from each country is so small, t-statistics are not meaningful and therefore not reported. Note that firms from 13 of the 17 countries in our sample experience a mean reduction in the incomplete information premium. Only four countries have a mean increase in the premium. The mean market risk is decreased in 12 of the 17 countries. Only one country, France, does not experience either a reduced incomplete information premium or a decreased market beta.

4.2 Robustness tests

In this section we explore three potential sources of bias in our estimation reported in Table 2. First, it is possible that when the cross-listing event is announced, the stock price will increase. The increase in the price may cause our pre-listing alpha and beta estimates to be too high. This might bias downward our estimates of the change in the incomplete information premium. Foerster and Karolyi (1999) find that, for those firms for which they could find cross-listing announcement dates, the average firm announced their cross-listing decision 10 weeks before listing. To determine the impact of this potential bias, we estimate Eq. 5 again but omit the 13 weeks (3 months) of returns before the cross listing. The results are reported in Panel A of Table 3. The estimate of the change in the incomplete information premium is −0.176% per week. This estimate is statistically significant and large in economic magnitude, though slightly smaller than the −0.204% estimate reported in Table 2.

Table 3 Robustness tests
Full size table

Throughout this study we compare the change in the cost of equity between firms by creating a common basis—the US dollar return. However, it can be argued that the firm experiences capital consumption and cost relative to its domicile country. Therefore, we report the results of estimating Eq. 5 using local returns instead of US dollar returns. The estimates, reported in Panel B, show that the change in the incomplete information premium is still economically large, −0.134% per week, as stated in local returns.

Another potentially complicating factor is that the firms’ domestic governments liberalizing their stock markets may influence our results. Henry (2000) finds that firms experience positive abnormal returns surrounding the decision by a government to liberalize their domestic equity markets. The observed return pattern documented in Henry is consistent with firms obtaining a lower cost of equity capital through a liberalization of markets. When examining international cross-listing, Stulz (1999) suggests using a control sample of firms that remained domestic to distinguish between decreases in the cost of equity from the cross-listing versus domestic market liberalization. Therefore, we use a matched sample of same-country firms that remain on the domestic market.

Only 24 firms in our sample have matching firms with complete data.Footnote 9 Panel C of Table 3 reports the mean parameter estimates of Eq. 5 for the 24 cross-listing firms and the 24 matching firms. The listing firms with matches experience a mean decrease in the incomplete information premium of −0.127% per week, using US dollar returns. The low t-statistic of −0.91 may be the result of a small sample causing low power to the test. However, the matching firms that do not cross-list experience an increase in the incomplete information premium of 0.332% (p-value = 0.09). The difference between the decrease in the premium of the cross-listing firms and the increase in premiums of the matching firms is significant at 5% level. It may appear surprising that our matching firms experience an increase in the incomplete information premium after the sample firm cross lists. However, Valero-Tonone (2002) provides a theoretical model to motivate the issue of whether rivals of firms which cross-list should benefit or be harmed from the listing. When one firm lists in the US, its rivals become viewed as less informative and of lower quality. Our evidence characterizes the harm done to rivals by suggesting that non cross-listing rivals are viewed as having higher incomplete information. While we do not believe that their information quality has actually changed, it clearly has changed in terms relative to the cross-listing firm.

Similar results are reported for the change in market beta. The cross-listing sample experiences a decrease while the matching firms sample experiences an increase during the same period. The difference is significant at 5% level. These results suggest that the decrease in the cost of equity experienced by foreign firms listing of the NYSE is associated with the cross listing, rather than market liberalization in the home domestic markets.

4.3 Time-series analysis

It may also be useful to examine the cross-listing of firms over time. The NYSE listing standards and accounting standards of US GAAP have increased over the years. So have the standards and accounting quality in countries around the world. Therefore, we would expect that the relative disclosure increase (reduction in the incomplete information premium) from cross-listing to be smaller in more recent years. Figure 1 shows the number of firms in our sample cross-listing each year and the mean change in the incomplete information premium for those firms. The first and last years (1985 and 1996) are omitted from the graph because they have only one firm cross-listing.Footnote 10

Fig. 1
figure 1

Results by year. The mean change in the incomplete information premium is shown as bars (left-hand scale). The number of cross-listing firms each year is shown in the line (right-hand scale)

Full size image

Figure 1 suggests that, consistent with our hypothesis, the mean decrease in the incomplete information premium has declined over the years. Specifically, firms cross-listing after 1991 seem to have achieved a smaller decline in the premium, if not an increase.

4.4 Correlation analysis

Before examining the relationship between our disclosure measures and the cost of equity, we report the correlations coefficient between these measures. Recall that we use two alternative measures of accounting disclosure quality Relative Accounting Disclosure is defined as a country’s accounting quality index (from La Porta et al. (1998)) minus 71 (the United States index value). Our second measure of accounting disclosure quality is a dummy variable Private Sector GAAP given the value of 1 if the private sector assists in setting accounting standards for a country; otherwise its value is zero. Determination of the involvement of the private sector in determining accounting standards is based on Ali and Hwang (2000).

To measure the disclosure attributed to analysts we use the number of analysts following a stock. Finally, as mentioned earlier, exchange disclosure quality has been shown to be inversely related to trading costs. Accordingly, our measure of exchange disclosure quality is the average effective spread for the 25 largest capitalization firms in a cross-listing firm’s home country. These data are taken from Jain (2003).

Table 4 reports these correlations and associated p-values. The Relative Accounting Disclosurevariable is highly positively correlated with all three of the other variables examined. Private Sector GAAP is negatively correlated with Average Effective Spread indicating that the private sector tends to participate in setting accounting standards in countries with relatively high exchange disclosure quality. The Number of Analysts is also negatively correlated with Average Effective Spread suggesting that analysts are more likely to follow firms from countries with high exchange disclosure.Footnote 11 In the next section we examine these relationships further through the use of regression models.

Table 4 Correlation coefficients of independent variables
Full size table

4.5 Regression analysis

In this section we further examine disclosure variables as they relate to the cost of capital within a multivariate analysis. Our general methodology is to regress changes in cross-listing firms’ information premium against our disclosure variables. We also include a dummy variable that takes a value of one when the cross listing firm raises new equity capital with (or up to a year following) NYSE listing. This variable controls for the possibility that a cross-listing firm has good growth opportunities and is cross listing to access new capital. Doidge et al. (2004) argue that firms cross-listing in the US have higher value because of these growth opportunities. Of our 94 cross-listing firms, 68 (or 72%) raise equity capital at, or following, the NYSE listing.

We conduct an OLS regression of the change in the information premium on combinations of various independent control and disclosure variables. The generalized regression equation is

$$\hbox{Reduction in information premium} = \alpha+\gamma\times \hbox{Offering} + \delta\times\hbox{Disclosure Proxy} + \varepsilon$$
(6)

where the Disclosure Proxy is the accounting, analyst, or spread variables. We report the parameter estimates and White (1980)-corrected t-statistics for the various models we examine.Footnote 12 Recall that the dependent variable is the change in the incomplete information premium (summarized in Table 2) and that it is generally negative.

The relationship between the change in the incomplete information premium and firms’ home country accounting disclosure quality is examined first. Panel A of Table 5 lists the results for our two alternative measures of accounting disclosure quality. As explained earlier, firms from countries with lower accounting disclosure quality countries than the US have positive relative accounting disclosure, while those with higher accounting standards have a negative value. Therefore, our hypothesis predicts that the parameter estimate will be negative. Examining Model 1 of Panel A reveals that indeed the parameter estimate for this variable is negative, which is consistent with our hypothesis that firms from countries with low accounting disclosure quality will benefit most from cross-listing on the NYSE. However, the parameter estimate is not statistically significant at acceptable levels.

Table 5 Regression analysis
Full size table

Our alternative measure of home country accounting disclosure quality is a dummy variable, Private Sector GAAP,which has a value of one if the firm’s home country accounting standards are set in part by the private sector—an indication of higher accounting disclosure quality. Model 2 of Panel A provides this parameter’s estimate, which is positive and statistically significant. This suggests that firms from countries with high accounting disclosure quality will benefit less from cross-listing and is consistent with our hypothesis. In both Model 1 and 2, the Equity Offering parameter estimate is positive, suggesting that firms seeking expanded capital access for growth opportunities will have a smaller reduction in their incomplete information premium following cross listing on the NYSE. However, the estimate is not statistically significant at usual levels of acceptance.

The next measure of disclosure that we examine is analyst following. Our hypothesis is that firms with little or no analyst following benefit most from international cross-listing to an exchange in a country that provides other forms of disclosure. We would therefore expect the parameter estimate for Analyst Following to be positive, indicating a lower reduction in the incomplete information premium following cross-listing for firms with large analyst followings.Footnote 13 Examining the estimate in Panel B of Table 5 reveals that indeed the estimate is positive and statistically significant. This is consistent with our hypothesis and Botosan’s (1997) belief that analysts may substitute for firm disclosure.

The last disclosure measure we examine is exchange disclosure quality. Recall that the results of Huddart et al. (1999) suggest that exchange disclosure quality can be proxied by the exchange’s trading cost with trading cost being an inverse measure of exchange disclosure quality. Firms cross-listing from countries where their home exchange has a high effective spread should have greater reductions in the information premium from listing on the NYSE, thus a negative coefficient is expected. Examining Panel C reveals that the parameter estimate for Effective Spread is of the expected sign and statistically significant. This suggests that firms from countries with low exchange disclosure quality will benefit most from cross-listing on the NYSE.

In summary, all three forms of disclosure examined appear to be related to changes in the incomplete information premium after firms cross-list on the NYSE. The findings suggest that firms with low disclosure quality will benefit from cross-listing on an exchange that provides higher disclosure quality. Panel D of Table 5 provides parameter estimates from regressions including all three measures of disclosure quality. Model 1 employs Relative Accounting Disclosure as the measure of accounting disclosure quality, while Model 2 includes Private Sector GAAP.In both models the parameter estimate for Analyst Following is of the expected sign and statistically significant. The exchange disclosure quality proxy, Effective Spread, is of the expected sign in both models, but only statistically significant in one. Similarly, in Model 2 the Private Sector GAAP parameter estimate is of the expected sign, but not significant. In Model 1, the parameter estimate of Relative Accounting Disclosure is now the opposite sign.

These findings suggest that firms that have low analyst following will experience a larger reduction in the incomplete information premium following cross-listing on an exchange with higher disclosure quality. Weaker evidence exists that firms from countries with low accounting and exchange disclosure quality will also benefit from cross-listing on an exchange with higher disclosure quality.Footnote 14

4.6 Discrete analysis

In the previous section, a number of the variables were continuous. The regression analysis is useful for identifying the average relationship between disclosure and the reduction of the incomplete information premium. However, regulators, exchanges, and investors may be more concerned about the marginal firm. Therefore, in this section we partition firms by low and high values of the disclosure quality variables and compare the average change in incomplete information premium between partitioned samples. We first partition firms by whether they have any analyst following and by whether the government alone or the private sector sets accounting standards for a firm’s home country. The mean incomplete information premium is reported for each of the four groups in Panel A of Table 6. The top left quadrant includes all firms that have no analysts and are from countries where the government set GAAP. The top right quadrant contains firms that have no analysts but come from countries where the private sector sets GAAP. The firms that have analyst following before cross-listing are sorted into the bottom two quadrants by the source of GAAP in their home country.

The firms with the lowest total disclosure (no analysts and government set GAAP) experience a decrease in the cost of equity of −0.788% per week. This estimate is statistically significant at 1% level. None of the other three groups experience a statistically significant decrease in the cost of equity. The table also reports F-values resulting from the tests where the null hypothesis is that the mean change in the cost of equity is equal between two sub-samples. The tests show that for firms with no analyst following, those from countries where GAAP is set by the government experience a significantly greater decrease in the cost of equity than firms from private sector set GAAP countries. Also, from all firms from government set GAAP, those without analyst following have a greater decrease in the cost of equity than firms with analyst following. These results show that both accounting disclosure and analyst following are very important for marginal firms.

Panel B repeats the analysis in Panel A, but substitutes exchange disclosure for accounting disclosure. Firms are deemed to come from low exchange disclosure quality countries if the average effective spread, for the 25 largest capitalization firms in that country, is above the median of our sample. Otherwise they are deemed to come from a country with high exchange quality disclosure. As with Panel A, firms with the lowest level of pre-existing disclosure (the top left quadrant) experience the largest decline in their incomplete information premium following cross-listing. Regardless of whether a firm has any analyst following, the average reduction in the incomplete information premium is greater if the firm came from a country with low exchange disclosure quality. This suggests that both variables are important in determining the size of the reduction a firm will obtain by cross-listing on an exchange with higher disclosure quality.

The last panel of Table 6 partitions by exchange and accounting disclosure quality. Examining the panel reveals that only firms from countries with low accounting and exchange disclosure quality have statistically significant reductions in their incomplete information premium following cross-listing, again suggesting a role for both variables in determining the reduction. The results of Table 6 taken with those from Table 5 suggest that all three forms of disclosure are important in determining the benefits of cross-listing.

5 Conclusions

In this study, we use a sample of foreign firms that cross-list on the NYSE to examine the relationship between the cost of equity capital and disclosure quality. Following Botosan’s (1997) comments on the importance of accounting for systematic risk when examining the relationship between disclosure and cost of equity, we develop an international asset pricing model to measure both disclosure risk and systematic risk. We find that cross-listing firms experience a decrease in their cost of equity averaging over 0.2% per week. This decrease is not present in a matched group of similar companies, which do not cross-list.

The magnitude of the decrease in the cost of equity is found to be inversely related to the quality of disclosure prior to cross listing. We examine three sources of disclosure: accounting standards; analyst following; exchange/regulatory investor protection. Specifically, firms from countries with low levels of accounting disclosure benefit from the added disclosure of cross listing. Also, firms with low analyst following or from countries with relatively low levels of exchange/regulatory disclosure benefit the most from cross-listing.

The findings of this study provide additional evidence in the current debate surrounding the SEC’s requirement for 20-F reconciliations for international firms cross-listing in the United States. Our results suggest that US accounting standards result in increased disclosure that is beneficial to investors. Further these results should encourage exchanges, accounting standard setting bodies and regulatory bodies to strive to adopt high disclosure standards.