Do non-u.s. firms issue equity on U.S. stock exchanges to relax capital constraints? Karl Lins

Transcription

1 Do non-u.s. firms issue equity on U.S. stock exchanges to relax capital constraints? Karl Lins The Kenan-Flagler Business School The University of North Carolina at Chapel Hill Campus Box 3490, McColl Building Chapel Hill, NC , USA Deon Strickland The Ohio State University Fisher College of Business 1775 College Road Columbus, OH and Marc Zenner The Kenan-Flagler Business School The University of North Carolina at Chapel Hill Campus Box 3490, McColl Building Chapel Hill, NC , USA January 2000 JEL classification: G15; C22 Keywords: ADRS; emerging makets; information asymmetry; liquidity; capital constraints We would like to thank Gregory Mancini and René Vanguestaine (J.P. Morgan) for a helpful discussion, Tim Adam, Geert Bekaert, Oyvind Bohren, John Coffee, Stuart Gillan, Rafael La Porta, Darius Miller, Urs Peyer, Sergei Sarkissian, Henri Servaes, David Scharfstein, Anil Shivdasani, Ignacio Velez, Michael Weisbach, and participants at the NBER conference on Macro Economic Effects of Corporate Finance for invaluable comments, Darius Miller for providing data on ADRs that raised funds, NASDAQ and the NYSE for providing initial data on non-u.s. listings, and Reena Bhatnagar, Umit Ozmel, and Rui Yao for expert help with the data collection. Deon Strickland acknowledges financial support from the Charles A. Dice Center. Marc Zenner acknowledges financial support from the Frank H. Kenan Institute for Private Enterprise and the Jefferson Pilot Fellowship at the University of North Carolina.

2 Do non-u.s. firms issue equity on U.S. stock exchanges to relax capital constraints? An increasing number of non-u.s. firms have listed their equity on U.S. stock exchanges. While non-u.s. firms may seek U.S. listings for a variety of reasons, we focus on one often-mentioned motive, namely that a U.S. listing enhances access to external capital markets. We find that following a U.S. listing, the sensitivity of investment to cash flow does not change for developed markets firms, but decreases significantly for firms from emerging capital markets. We also document that about half of the listing firms from both emerging and developed markets explicitly state the importance of access to external capital markets in their annual reports. Further, we find that firms access international capital markets more frequently following a U.S. listing and that this increase is more pronounced for firms from emerging markets. Our findings are robust to a variety of alternative specifications and suggest that greater access to external capital markets is an important benefit of a U.S. stock market listing, especially for emerging markets firms.

3 Stock markets from Mexico City to Sao Paulo have sunk in importance recently with trading shifting to Latin stocks listed in New York as American Depositary Receipts, or ADRs. The focus abroad means the only Latin American companies that can raise financing nowadays are those that have access to international markets. Thousands of small and mid-sized businesses the bulk of Latin America s productive capacity are left behind. [Latin American Stock Trading Migrates North, Wall Street Journal, October 27, 1999] 1. Introduction The number of non-u.s. firms with American Depositary Receipts (ADRs) listings on U.S. equity markets has surged in recent years. 1 Fanto and Karmel (1997), for example, report 108 listed ADRs in 1985, 176 in 1990, and 426 in Practitioners cite the following reasons for issuing ADRs: (i) increased access to new capital when the local markets cannot provide enough capital; 2 (ii) increased liquidity; (iii) increased visibility and brand recognition; (iv) access to a currency for stock-based takeovers; 3 and (v) use of U.S.-traded stock to compensate and motivate U.S. executives. 4 Recent evidence by Foerster and Karolyi (1999), Miller (1999), and Bruner, Chaplinsky, and Ramchand (1999) suggests that non-u.s. firms benefit from a U.S. listing by decreasing the cost of capital and by enhancing the shareholder base. 1 In general, an ADR is a negotiable certificate sponsored by U.S. depositary banks that represent the company s shares in its home market. Not all non-u.s. listings in the U.S. are ADR listings. Some non-u.s. companies list their shares (rather than depositary receipts) on the U.S. market (these are called New York shares). We follow the convention on by using the term ADR to refer to both depositary receipts and New York shares. 2 Limited access to external capital markets is also an essential element in Stulz (1999). In this paper, Stulz develops two arguments why the cost of equity capital will decline with the globalization of capital markets. First, he shows that the expected returns that investors require to compensate them for the risk they bear fall. Second, he argues that agency costs, which restrict access to external capital and increase the cost of capital, become less important as markets globalize. 3 A recent article in Euromoney Magazine (1999) mentions that ADRs have become very important as a takeover currency. Deals using ADRs include some of the largest recent transactions, including the purchases of Airtouch by Vodafone, Amoco by British Petroleum, Chrysler by Daimler-Benz, and Pacificorp by by Scottish Power. 4 We would like to thank Mr. René Vanguestaine, Managing Director at J.P. Morgan, for discussing these motives with us. 1

4 We provide additional evidence on the benefits of an ADR listing by examining whether a U.S. listing by a non-u.s. firm reduces the firm s dependence on internally generated cash flows and enhances access to external capital markets. 5 We also test whether these benefits are more pronounced for firms that are domiciled in emerging markets, where access to external capital is likely to be more restricted. 6 Recent media reports see our opening statement indicate that this is a particularly important issue for emerging markets firms. We use three methods to investigate whether a U.S. equity listing enhances access to capital for non-u.s. firms. First, we examine whether the investment to cash flow sensitivity of listing firms declines following the ADR listing using the method described by Fazzari, Hubbard, and Petersen (henceforth FHP) (1988). If internal and external markets for capital are not perfect substitutes in the presence of asymmetric information about firm value, FHP argue, along the lines of Myers and Majluf (1984), that this information asymmetry can make both debt and equity sufficiently expensive that firms are effectively rationed in their access to the external capital markets. As a result, internally generated cash flow will influence a firm s investment policy. We argue that, compared to their home market, greater disclosure requirements, shareholders rights protection, liquidity, and analyst following reduce the information asymmetry for non-u.s. firms listed on U.S. equity markets, especially firms from developing economies. Thus, if a firm is rationed in its access to capital markets and becomes eligible to list on an U.S. exchange, it 5 An enhanced i.e. less costly - access to capital markets should also lead to a reduction in the cost of capital and an increase in firm value, as documented in the studies cited above. These studies do not, however, examine whether, how and when access to external capital markets is enhanced following a U.S. listing. 6 In the first part of our tests, we use the International Finance Corporation classification of emerging markets to identify the non-u.s. firms that are most likely to benefit from a U.S. listing. We assume that the emerging markets identification proxies for a variety of restrictions or limitations characterizing these markets. La Porta, Lopez-de- Silanes, Shleifer, and Vishny (1997, 1998) discuss these issues for various international markets. In later tests, we specifically examine some of the variables developed by La Porta et al., such as the rule of law, the degree of financial development, and the historical background of the legal system. 2

5 may list to reduce the level of information asymmetry and enhance its access to the capital markets. 7 We posit that if listing firms gain access to a market with improved information production and a larger investor base, such as the New York Stock Exchange (NYSE) or NASDAQ, then investment liquidity will improve. As such, the external capital markets constraints described by FHP should decline following a U.S. listing, especially for emergingmarkets firms. To examine our hypothesis, we use a sample of ADR listings on the NYSE and the NASDAQ over the period from 1986 to 1996 for firms that are publicly traded on the stock market in their country of domicile. 8 Using the FHP methodology, we find a significant decline in the investment to cash flow sensitivity following the U.S. market listing, but only for listing firms from emerging markets; that is, firms from markets that are likely to be characterized by more limited access to external capital markets. In additional tests, we examine how the decline in cash flow to investment sensitivity is related to individual market characteristics. These tests confirm that firms from countries with less-developed external capital markets and with more limited rule of law (as defined by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998)) benefit most from a U.S. listing, at least with respect to their investment to cash flow sensitivity. These results are also robust to various alternative specifications. 7 Two studies contemporaneous to ours provide evidence consistent with our hypotheses. Pagano, Roëll, and Zechner (1999) find that European firms that list in the U.S. or cross-list in Europe tend to be large firms, firms with high foreign sales, and firms with high R&D spending. If a higher R&D intensity is associated with higher information asymmetry, then their results are at least partially consistent with the notion that firms with larger information asymmetry benefit more from a foreign listing. Reese and Weisbach (1999) argue that improving shareholder protection is an important motivation for ADR listings. Consistent with their hypothesis, they find that firms, especially those from countries with a French legal system, increase their use of external capital markets after listings on U.S. exchanges, but not after other cross-market listings. 8 We only use listings on NASDAQ and the NYSE because Miller (1999) does not document listing benefits for firms with ADR programs on PORTAL or the OTC market. Additionally, most of these firms would be eliminated because they typically do not have accounting data on Worldscope, our major source of accounting information. 3

6 For our second test, we examine annual reports, 10K, and 20F filings by ADR firms and read firm-issued statements on the importance of external capital market access to accomplish the firms growth and investment objectives. We find that a large proportion of firms in both developed and developing markets explicitly discuss a need to enhance their access to external capital markets. Finally, in our third set of tests we collect and analyze the frequency and magnitude of the access of external capital by listing firms surrounding the ADR listing. We gather security issue data from Securities Data Corporation (SDC) and find, using a variety of measures, that ADR firms increasingly access external capital markets following the ADR listing, especially those from emerging markets. Overall, our evidence suggests that access to external capital markets is important for non- U.S. firms listing in the U.S., and especially so for emerging markets firms. Our evidence provides additional support for the notion that capital markets are segmented and that the depth of U.S. capital markets and the reduction in information asymmetry following a U.S. listing reduces the listing firm s cost of capital and enhances its access to external capital markets. Various studies have analyzed the costs and benefits of ADR listings. A large fraction of these studies have focused on stock returns at the time of the listing announcements and have emphasized the market segmentation hypothesis, which posits that securities are priced differently in different markets because of regulatory and information barriers. If an ADR listing reduces market segmentation and increases the number of potential investors, then, following Merton (1987), equity values should increase at the announcement of an ADR listing. Recent evidence supports this hypothesis. Foerster and Karolyi (1999) and Miller (1999) document a 4

7 positive stock price reaction around the announcement of an ADR listing. 9 Specifically, Foerster and Karolyi show that the price gains at the announcement of the listing are related to the enhanced liquidity and shareholder base, supporting Merton s (1987) investor recognition hypothesis. Miller documents that the positive announcement returns are higher for issuers from emerging markets than from developed markets and for ADRs on NASDAQ and NYSE rather than on other smaller markets. He also argues that these findings are consistent with the market segmentation hypothesis. Errunza and Miller (1998) provide additional evidence on the sources of gains from ADR listings by examining post-listing returns and documenting that a firm s cost of capital declines following an ADR. Corroborating evidence is provided by Bruner, Chaplinsky, and Ramchand (1999) who interview investment bankers handling ADR-IPOs and document that managers of foreign firms believe that their shares would be undervalued in their original market and that they will obtain greater financial flexibility by listing their shares in the U.S. 10 The remainder of the paper is organized as follows. In section 2, we describe the data and sample selection procedure. In section 3, we examine the cash flow to investment sensitivity of ADR firms surrounding the ADR listing and examine whether the cross-sectional differences are explained by stock market and legal characteristics. We also examine the discussion of the firms need for external capital in their annual reports as well as various robustness tests. In section 4, we investigate whether ADR firms actually increase their access to external capital markets following the ADR listing. We conclude in section 5. 9 Early studies such as Alexander, Eun, and Janakiramanan (1988) and Jayaraman, Shastri, and Tandon (1993) do not find significant stock price reactions at the time of the listing. 10 Although the studies above indicate gains at the time of a listing in the U.S., Stulz (1999) reviews the evidence on ADR listings and argues that the wealth gains are substantially smaller than what one would expect under reasonable assumptions about the transition from a segmented to an integrated capital market. Stulz argues that part of this small wealth gain may be explained by the fact that the capital markets have anticipated the listing. Another 5

8 2. Data collection, sample construction, preliminary descriptive statistics Our data collection and sample construction procedure is illustrated in table 1. The NYSE and NASDAQ provided our initial ADR listing data. The NASDAQ list covers a period from December 1970 to July The NYSE list covers a period from December 1928 to September These data sets contain listing dates, the country of origin of the listing firm, and the type of listing. Because all the non-u.s. firms we examine list on the NASDAQ or the NYSE, our sample consists only of level II and level III ADRs (see Miller (1999), Table 1, for a definition of ADR programs). Next, we use three criteria to construct a sample that is best suited for our tests. First, we eliminate financial firms because financial firms are highly regulated in most countries and because the FHP method we use to evaluate the cash flow to investment sensitivity cannot easily be applied to financial firms. 11 Second, we also eliminate observations if a firm is not already a publicly traded company on its home-country stock exchange prior to its ADR listing on the NYSE or NASDAQ. We define ADR to comprise a listing of either: 1) depository receipts that represent ownership or 2) actual shares of common voting stock which already trade on a stock exchange in the firm s home country prior to the U.S. listing (New York shares). Thus, our sample excludes non-u.s. firms that simultaneously issue stock and/or depository receipts for the first time, but includes listings plausible reason is that markets do not instantaneously become fully integrated, but that in fact integration is a gradual process [see Bekaert and Harvey (1995)]. 11 Financial firms consist of commercial banks, insurance companies, diversified financial services, and brokerage houses, following the definitions provided by the stock exchanges. 6

9 by existing publicly-traded firms that raise new equity as well as those that re-deploy existing shares to the U.S. market. 12 Finally, we collect accounting and market value data from Worldscope. We require that listing firms have sufficient coverage on Worldscope to be included in our sample. Sufficient coverage means that Worldscope provides financial data for the firm for at least two years before and after the U.S. listing date. To complement Worldscope data for several observations, we use Datastream, a Mexican Bolsa database, company web sites, and Global Data Direct. Because we require two years of accounting data following the listing, our analysis does not include listings beyond Additionally, because Worldscope coverage is extremely sparse prior to the early 1980s, our analysis includes only ADRs listed after In table 1, we present the number of ADRs listed on the NYSE or NASDAQ based on the lists provided by both exchanges. As expected, we document a significant increase over time in the number of U.S. listings by non-u.s. companies. Before 1986, 104 non-u.s. companies were listed on the NYSE or NASDAQ. Between 1986 and 1996, another 540 non-u.s. companies listed on these two markets. There is an acceleration of the listing trend toward the end of the period. Consistent with one s expectation about the rise of emerging markets, companies from developed markets constitute an overwhelming majority of the listings through We eliminate 67 observations (12.4% of the sample) because they are financial firms. We eliminate 142 observations (26.3% of the sample) because they are ADR/IPOs. Finally, we lose 226 firms (41.9% of the sample) because insufficient data are available from Worldscope. While one would expect that the quality and quantity of data will be inferior for emerging 12 This definition is consistent with Foerster and Karolyi (1999) and Miller (1999) who require the listing firm s returns prior to listing to compute ADR announcement returns. We require prior listing because (1) we would like to test the benefits of a U.S. listing compared to the existing home country listing and (2) our methodology requires market values to compute market to book value ratios. In contrast, Bruner, Chaplinsky, and Ramchand (1999) 7

10 markets, we actually lose slightly fewer emerging markets firms (39.4% of the sample) than of the developed market firms (43.0% of the sample) because Worldscope coverage is insufficient. With the Worldscope data requirement we are more likely to lose small than large firms. Hence, the composition of our sample is likely to bias against finding results that are consistent with our hypothesis to the extent that small firms which may have been lost because of data requirements are more likely to suffer from restricted access to external capital markets. Because of differences in international accounting practices, we are concerned about the intertemporal comparability of the accounting data we use. 13 As a general rule, Worldscope reports accounting data using local standards, and does not change reporting practices surrounding a U.S. listing. We find two exceptions to this rule, Pechiney and Daimler Benz, for which the data are in local standards before and in U.S. GAAP following the listing. We reexamine all our results by excluding these two firms and find that this exclusion does not materially change our results. In table 2, we present key descriptive statistics for the main variables (the ratios) that we use in our regressions, which are described further in the next section. All financial variables are in U.S. Dollars and are then standardized by dividing them by the firm s total assets. Mean (median) investment as a percentage of total assets is 12% (9%) and the mean (median) free cash flow as a percentage of total assets is 16% (14%). The median of the market to book ratios is 1.55, and the median sales to total assets, debt to total assets, and cash to total assets ratios are 0.64, 0.24, and 0.06, respectively. Seventy-nine percent of the observations are from developed markets. examine a sample of non-u.s. IPOs on a U.S. equity market because they study the IPO underwriter s fees. For the reasons mentioned above, we do not include these observations in our sample. 13 We provide some cross-sectional comparisons in our descriptive statistics table, but our main analysis relies on a time-series comparison within firms, and not on a comparison across firms. 8

11 3. Investment to cash flow sensitivity 3.1. Regression methodology To examine the change in the sensitivity of investment to free cash flow, we use the FHP (1988) methodology with some modifications to take into account the characteristics of our data and tests. The regression specification takes the following form: 14 TA t t 1 I = FreeCashF TA + t t 1 5 Sales TA + 2 t 1 t 1 PostList Cash TA 3 t 1 t 1 FCF + t TA t *PostList t M/B t 1 The dependent variable is investment (I t ) scaled by TA t-1, the total assets in the preceding period. In FHP and most of the related literature, the scalar is K, the initial capital stock. We use the firm s initial total assets instead because we believe that international firms are likely to be more consistent over time in reporting total assets than in reporting book values of capital employed. The independent variables and expected relations with the dependent variable are: Free cash flow (FreeCashF t ) scaled by TA t-1 ; FHP argue that if one controls for investment opportunities and if there is costly access to external capital markets, then there will be a positive relation between internally generated cash flow and investment. Free cash flow is the sum of income before extraordinary items and depreciation net of cash dividends. Post-List, is a dummy variable equal to one after the listing date and equal to zero otherwise. We include this variable to control for changes in the investment pattern following the listing that are not related to the investment to internal free cash flow sensitivity. 14 FHP, Hoshi, Kashyap, and Scharfstein (1991), Whited (1992), and Kaplan and Zingales (1997), among others, provide a detailed discussion of the methodology. 9

12 Free Cash Flow multiplied by Post-List; if a U.S. listing enhances access to external capital markets, then there should be a negative relation between this variable and investment, indicating that the reliance on internally generated free cash flow is lower following the listing. Market to book ratio (M/B) of debt plus equity for the preceding period; classical investment theory predicts a positive relation between Tobin s q and investment if q correctly measures the firm s investment opportunities and if the firm invests according to these investment opportunities. The definition and hence construction of Tobin s q uses the replacement value of assets in the denominator. We use the market to book ratio of assets as an approximation for Tobin s q and estimate the market value of debt plus equity as the book value of debt plus the market value of equity. This approach is consistent with many other papers because, like these papers, we are not able to obtain reasonable estimates of replacement values for international firms. 15 Sales t-1 relative to Total Assets; scaled lagged sales, as a proxy for production, are included in the regression to control for a possible accelerator effect. Hoshi, Kashyap, and Scharfstein (1991) argue that production should be included because it is correlated with the liquidity variables. Thus, if one excludes production, the liquidity variables might proxy for production effects that are empirically important but not well understood in the investment theory literature. Cash t-1 relative to Total Assets; if access to external capital markets is costly and there is a positive investment to cash flow sensitivity, then this sensitivity is likely to be lower when 15 Perfect and Wiles (1994) and Lewellen and Badrinath (1997), among others, show that the improvement in q estimations from using more complicated algorithms is limited. 10

13 the firm has a lot of financial slack. We also control for the firm s available balance of cash and marketable securities. We estimate a fixed-effects model that includes a dummy variable for each firm (i.e. a firm fixed-effects model), but does not include a dummy variable for each year. Time dummies are not included because the sample is aligned in event time rather than calendar time. For each firm we have the same number of pre- and post-listing observations, where we use two years on each side of the listing event as a minimum and three years as a maximum. We also estimate the models with only two years of data on each side of the listing. This approach yields qualitatively similar results and we do not tabulate these results for brevity. An essential part of our analysis is to compare the changes in investment to cash flow sensitivity of emerging markets firms to those of developed markets firms. If access to external capital markets is more constrained for emerging markets firms, then the decrease in reliance on internally generated cash flow following the U.S. listing should be more pronounced for these firms. Emerging markets include all markets classified as such by the International Finance Corporation (henceforth IFC) Univariate comparisons First, we examine our key investment regression variables to gauge whether there are substantial changes in these variables following the listing. In panel A, table 3, we compare key variables from one year before to one year following the listing. For firms from emerging markets, there is an increase in the investment level even though there is a decrease in the free cash flow level. There is also an increase in the market to book value ratio, which perhaps causes the increase in investment. None of these changes are significant, however, possibly 11

14 because of the small sample size for emerging market firms. For developed markets firms, there is no significant difference between year 1 and +1 around the listing. There are, however, some significant differences in key variables between emerging and developed markets. Specifically, emerging markets firms invest more and have higher free cash flow the year preceding the listing, but their sales to total assets levels is lower than for developed market firms. It is possible that the higher investment levels in emerging markets can be explained by higher growth in these markets, but the fact that the market-to-book value ratios are not different between the two groups reduces the strength of this argument. Because accounting standards are vastly different between various countries, we are cautious about our interpretation of these results and put more weight on our later time-series multivariate results. Nevertheless, we still believe it is interesting to examine these univariate comparisons to better understand the data. In panel B, we compare variables from two years before to two years following the listing. The results for the two-year comparison show a decrease in the free cash flow level for both emerging markets and developed markets firms. We also confirm the higher investment level and lower sales to total assets level for emerging markets firms compared to developed markets firms, but the same caveat about cross-country differences in financial reporting also applies in panel B. While the results in panel A are consistent with our expectation that investment levels should increase for emerging markets firms following a listing, our results do not follow the same pattern in panel B, and none of these differences are significant. Additionally, the investment literature shows that other factors, such as the investment opportunity set, affect investment levels, and one therefore needs to interpret these univariate comparisons with caution. To 12

15 control for these other factors, we estimate and discuss multivariate regression results in the next section Regression evidence We present the main results of our investment to cash flow sensitivity tests in table 4. In the first column, we estimate the model using only emerging market firms and in the second column, we estimate the model for developed market firms. There is a positive relation between the firm s free cash flow and the firm s investments for both subsamples, consistent with the prior literature. 16 Whereas the coefficients on sales and the market to book value ratio are positive, consistent with the prior literature, they are not significant. This latter result may be due to the difficulty of using international accounting data to proxy for Tobin s q. 17 We find a positive relation between the level of cash plus marketable securities and investment, but this relation is only significant for developed market firms. More importantly for our analysis, however, is the coefficient on the interaction between the free cash flow and the post-listing variable. The negative and significant sign for the interaction suggests that the investment to cash flow sensitivity declines significantly following the listing for emerging market firms. This interaction variable is not significant for developed markets firms. The results in table 4 suggest that there is a decrease in the investment to cash flow sensitivity following a listing on a U.S. equity market for emerging markets firms and not for firms from developed capital markets. The increase in the investment level for emerging 16 The coefficient between the internally generated cash flow and the investment level is 0.95 for emerging markets firms preceding the listing. We verify the nature of this relation by examining the simple correlation between cash flow and investments. The correlation is 0.98 preceding the listing and only 0.39 following the listing. 13

16 markets firms following the listing is further evidence of the benefits of a U.S. listing for firms from less-developed capital markets. Overall, our results are consistent with our prediction that access to external capital markets is particularly costly for firms from emerging markets where information asymmetries are likely to be higher. These results suggest that the benefits of a U.S. listing are more important for emerging markets firms. Given that a majority of the firms in our sample are from developed markets, it is likely that these firms reap other benefits from listing their stock in the U.S., such as enhanced commercial visibility, or the use of stock as a currency for takeovers or to compensate employees. One interesting feature of the regressions presented in table 4 is the magnitude of the R 2 for the developing markets model. For the developing countries model the R 2 is approximately ninety four percent. We investigate the source of the high R 2 and determine that prior to listing in the US the correlation between investment and free cash flow for the developing sample is ninety nine percent. After listing the correlation falls to approximately sixty percent. The suggests that prior to listing companies in the developing sample invest all of their free cash flow. The presence of this one to one relation is the source of the high R 2 for the developing markets model What is unique about emerging markets? In the tests discussed so far, we separate emerging markets firms from developed markets firms based on an IFC classification of emerging markets. The IFC classification results in two unequal subsets with 24 emerging markets firms (with nine firms from Mexico and seven firms from Chile) and 81 developed markets firms. How robust is our result to this classification? 17 Some other recent papers do not find a consistent positive relation between q and investments. Gertner, Powers, and Scharfstein (1999), for example, do not find a positive relation between q and capital expenditures for various 14

17 Further, does the IFC classification capture the richness of the substantial variations in the degree of development of the markets that are classified as developed or emerging markets? La Porta, Lopez-de-Silanes, Shleifer, and Vishny (henceforth LLSV) (1997) categorize the degree of development of various markets using variables based on, among others, a country s judicial system, shareholders and creditors rights, and external stock market capitalization relative to the size of the economy. If these variables are also related to the degree of information asymmetry in the market, and thus to the costs of accessing external capital, then these variables may constitute a more refined way of identifying countries where access to outside capital is costly and hence a U.S. listing is more beneficial. We focus on two LLSV (1997) variables from their Tables I and II (pages and 1138) and on the judicial system. The first variable, which we call financial development, is the ratio of external capital to GNP, which measures how important the external equity capital market is in relation to the GNP. Countries with a high external capital to GNP ratio have more developed external equity markets and we assume that access to external capital is less restricted in those countries. 18 The second variable, rule of law, measures the country s quality of law enforcement. It is collected by LLSV from the International Country Risk Guide and represents an investor assessment of the quality of law and order environment. This variable has values from 1 to 10 and a high score represents a high quality of law enforcement. We focus on the financial development variable because it directly relates to the relative importance of the local capital market, the object of our study. We focus on the rule of law variable because La Porta, Lopez-de-Silanes, Shleifer, and Vishny (2000) show that the common element explaining the subsets of related and unrelated spinoffs in the U.S. 18 LLSV also examine other variables measuring the size of the external capital market such as IPOs per population or domestic firms per population. We also estimate our tests with these variables instead of external equity capital over GNP and our results are qualitatively similar in nature. For brevity these results are not tabulated. 15

18 large differences between the access to external finance and the development of capital markets is how well investors are protected against expropriation. We also focus on the origin of the legal system, namely the English-style common law system versus the civil law system. Recent evidence by LLSV (1997, 1998, and 2000), Demirguc-Kunt and Maksimovic (1998), and Reese and Weisbach (1999) suggests that this is an important classification. They find that English Common Law offers better protection to minority shareholders, and as a result firms in English Common Law country have easier access to external financing. We estimate and present regression estimates for two models for three subsamples: In columns (1) and (3) and (5) of table 5, we estimate the models for firms with financial development and rule of law values below the median value and with a non-english legal system. In columns (2) and (4) and (6) we estimate the models for firms with higher than median values for these two variables and with an English legal system. Although these variables are likely to be highly correlated with the emerging markets dummy, there are some important differences. South Africa, and Taiwan, for example, are classified as emerging markets by the IFC but have financial development values that are higher than almost all developed markets. For listing firms from less-developed and less-protected capital markets (i.e., where the values for the two variables are less than the median) we expect a larger benefit from listing on a U.S. market and more of a decline in the investment to cash flow sensitivity following the listing. We use the median to separate the sample rather than a continuous variable. These variables have been collected for a specific year. While the values and rank order are likely to change frequently depending on fluctuations in the stock market, it is 16

19 less likely that a country will switch from the high development group to the low development group during our sample period. Overall, we find a positive relation between investment and free cash flow for all subsets. Similar results are obtained for the interaction between the post-listing dummy and the free cash flow variable. The cash flow to investment sensitivity declines following the listing for the lessdeveloped and less-protected capital markets and for countries that do not have the English Common Law legal system. This result indicates that firms from developed markets with a legal system that offers less protection to the minority shareholders also benefit from a reduced linkage between internal cash flow and investment following an ADR listing. To examine whether the effect we find is due to the more limited minority protection from the non-english legal system countries or to the fact that the country is an emerging market, we examine separately the twenty developed market firms that are not from non-common Law countries. We find a negative but insignificant sign on the post-listing cash flow interaction variable. Overall, these results (not tabulated) are consistent with our results in table 4 that specifically firms from emerging capital markets benefit more from a U.S. listing through a reduction in the investment to cash flow sensitivity. The emerging markets dummy is likely to capture a variety of factors that limit a firm s access to external capital markets, and the legal system background is only one of these relevant variables Discussion of the methodology We use the FHP methodology to estimate the cash flow to investment sensitivity. Hoshi, Kashyap, and Scharfstein (1991) point out, however, that there are a number of important weaknesses in attempting to measure liquidity constraints with this regression model. First, 17

20 researchers generally must employ average q rather than the more appropriate marginal q. While Hayashi (1982) derives conditions under which average q is sufficient to assess how much the firm should invest, it is not clear that these conditions are always met. Second, it is likely that q is measured with error. In our case, we do not attempt to estimate the replacement value of assets and use the market to book value ratio instead of Tobin s q because of the difficulties of working with international financial data. The measurement issue of q or the failure to meet the Hayashi (1982) conditions may lead to significance for both the liquidity variables and q in the investment to cash flow sensitivity regressions. Erickson and Whited (1999) employ a measurement consistent test and show that the cash flow coefficient becomes insignificant and that only 40% of the variation in investment is explained by marginal q. Third, it is also possible that liquidity proxies for an important omitted variable, possibly the profitability of investment. The time series nature of our tests is robust to measurement bias because the time series difference in the slope of the free cash flow coefficient is unbiased. Hoshi, Kashyap, and Scharfstein (1991) suggest alternative specifications to examine the robustness of the results. One concern is that cash flow during period t may contain investment opportunity information not contained within beginning of period Tobin s q. To address this issue, we consider two alternative specifications. The first alternative consists of estimating the model with the addition of the end of period market to book ratio. The end of period market to book ratio includes all the additional information known at time t, including the information from the cash flow during period t. The second alternative consists of estimating the model with lagged values of free cash flow. Hoshi, Kashyap, and Scharfstein (1991) argue that the inclusion of lagged free cash flow eliminates the component of free cash flow which cannot be predicted 18

21 given beginning of period Tobin s q. With both tests we lose observations but the results are qualitatively unchanged. Kaplan and Zingales (1997) criticize the FHP methodology. They argue that most papers in the investment-cash-flow sensitivity literature are only able to identify constrained firms and not firm years. They conclude that such an exercise is valuable only if the investment-cash-flow sensitivity is monotonically increasing with respect to the difference in the cost of external and internal capital. Because we focus on the times series component by comparing the investment sensitivities of ADR-listing firms in the pre- versus the post-listing we are less concerned about this issue. Further, the time series nature of our tests also alleviates Kaplan and Zingales concern about precautionary savings and overly risk-averse managers because it is conducted within the sample of listing firms. Kaplan and Zingales buttress their arguments by examining whether the firms identified by FHP as cash constrained (i.e. non-dividend paying firms) are truly cash constrained according to the managerial statements in the annual reports. We follow Kaplan and Zingales approach and examine annual reports as well as 20F and 10K statements for a random subsample of 116 listing firms. In reading these reports, we attempt to identify references to external capital markets constraints that are consistent with our hypothesis and our regression results. We could not find such reports on Lexis-Nexis for 40 of the ADR firms. For 33 of the firms, we found the report but no mention of external capital needs. Finally, for 43 of the ADR firms, we find information suggesting that the firms need external capital, and, to a certain degree, are concerned about their access to such capital. Some of the mentions are relatively explicit in terms of the need to raise external financing, such as the following September 30, 1994, 20F filing from Telefonica de Argentina S.A.: 19

22 The Company anticipates making capital expenditures well in excess of the amounts required to satisfy the List of Conditions. For fiscal years 1995 through 1998, total budgeted capital expenditures are expected to be in excess of P$5.0 billion. As a result, the Company anticipates that its capital requirements for the next several years will be such that they will not be able to be funded entirely by cash flow from the Company's operations. The Company expects that during the early part of such period it will have to raise additional funds in the private or public capital markets. No assurances can be given as to the availability of such financing on terms attractive to the Company. From a developed capital market, the Canadian company Biomira Inc. states in its 20F of 12/31/1993: Since the incorporation of Biomira in 1985, the Company's research programs, capital expenditures and investments have been financed from several sources. These sources have included research collaboration agreements with both governmental and industrial partners, up-front licensing fees of the Company's technology, interest income, and to a much greater extent, public and private placements of the Company's common shares. The Company has not produced an operating cash surplus since its inception nor is an operating cash surplus expected until the approval of the Company's products by the regulatory authorities and their subsequent commercialization. Overall, the interpretation of these firm-issued documents is that while not all firms are necessarily capital constrained, about half of the non-u.s. firms listing in the U.S. for which we find relevant information mention that they need external capital and are, to a certain extent, capital constrained. This fraction is likely to understate the true number of firms that are capital constrained for the following reasons: (1) managers may not mention or prefer not to discuss their external capital needs lest this provides important information to competitors; (2) our search may have missed the documents or the place in the filed documents where these issues are discussed; (3) it is more difficult to access the relevant information for small firms, but they are more likely to be capital constrained; (4) managers may not have been as concerned about capital constraints in their domestic market because they were anticipating the ADR issue. 20

23 3.6. Other robustness tests Apart from the robustness tests outlined in section 3.5, which address potential criticism of the FHP methodology, we perform a variety of other tests to assess the robustness of our results. Because many of the firms in our sample raise equity capital at the time of their U.S. listing, it is possible that the decline in investment to cash flow sensitivity we observe is due not to the listing, but to the infusion of capital. As we show in the next section, emerging markets firms also tend to access external capital markets more often than developed markets firms, consistent with our expectation as well as with this alternative interpretation of our results. To examine this issue, we estimate the same FHP regressions as in table 4, but separate the developed markets firms by those that raise external financing and those that do not. 19 For both regression estimates, we find that there is no decline in the investment to cash flow sensitivity for developed market firms, suggesting that external financing does not cause the decline in the sensitivity of investment to cash flow. Next, we examine whether outliers affect our results. Specifically, when we plot investment with respect to cash flow, we observe some outliers for the developed market firms. We use a regression estimate that underweights outlier observations and find similar results. We also examine whether the relation between the market to book value ratio and investment changes following the listing. Specifically, it is possible that before the U.S. listing, the market to book value ratio may not have served as a signal of the investment opportunity set (consistent with the insignificant relation with the market to book value ratio), but this may have changed 19 We cannot perform this separation of emerging markets firms because this sample is small and a majority of these firms access external capital markets following the listing. 21

24 following the listing. To test this proposition, we interact the market to book value ratio with the post-listing dummy, but it is insignificant. Twenty-eight firms in our developed markets sample are from Canada. However, Canadian firms can list on a U.S. exchange without modifying their disclosure substantially, hence it is not clear how much informational benefit a listing on a U.S. exchange would generate. We reexamine the results for developed markets firms excluding the Canadian firms, and find similar results. Nine firms in our emerging markets sample are from Mexico, seven from Chile, and a majority of these firms listed in 1993 and 1994, just preceding Mexico s financial crisis. Is our post-listing dummy capturing the effect of the December 1994 Mexico crisis, which also affected other Latin American firms? We re-examine our results for emerging markets firms and non- English Law firms by including a dummy variable equal to one for post-1994 observations and equal to zero for pre-1995 observations. We also interact the pre/post 1994 dummy with the free cash flow variable. The results obtained from the inclusion of the pre/post 1994 dummy and the pre/post 1994 dummy free cash flow interaction in the regression model are quantitatively and qualitatively similar to those presented in table 4. We also examine whether the effect we have documented is due to the enhanced disclosure from listing on an organized exchange, or whether it is due to a firm s ability to access a supply of U.S. capital. To examine this issue, we separate ADR firms according to whether the NYSE or NASDAQ listing was their first listing in the U.S. or whether it follows a prior listing on the pink sheets or a prior 144a issue (an upgrade ). 20 We find that the listing we analyze is the first one for almost all emerging markets firms, so we cannot distinguish between these two hypotheses for emerging markets firms. For twenty developed markets firms, however, the 22

25 listing we examine is an upgrade. We find that there is no difference in the change in investment to cash flow sensitivity for these two developed markets subgroups. Bekaert and Harvey (2000) find that emerging markets which liberalize access to their capital markets experience a five to seventy five basis point decline in the cost of capital. Thus it is possible that countries where investors have had substantial time to avail themselves of access to the capital markets of a developing country may reduce the benefit of a US listing. To examine this possibility we construct a time since liberalization variable. We employ the liberalization dates presented in table Table 1 of Bekaert and Harvey (2000) to construct the time since liberalization variable. The variable is the number of months since liberalization. We include this variable and interact the variable with the cash flow post listing dummy. The results obtained from this analysis are qualitatively and quantitatively similar to those presented before. 4. Issuance The results from the previous section suggest that non-u.s. firms that list on a U.S. exchange experience a decrease in the sensitivity of investment to cash flow. If this decline in sensitivity stems from greater access to the capital markets, firms may increase their access to the capital markets. 21 To investigate this hypothesis, we examine the debt and equity issuance patterns of listing firms in the two years before and after listing on a U.S. exchange. We gather the offer date, the dollar amount raised in seasoned equity offerings and public debt offerings for listing firms from Securities Data Corporation. We classify all convertible bond issues as debt issues, but our results are not materially affected by this classification. 20 We would like to thank Darius Miller for providing these data. 21 Although the opposite argument can be made that firms may now be less concerned about having a cash reserve because they feel that capital markets can now be accessed when needed. In this case, firms would actually not necessarily increase their access to capital markets following the listing. 23