Is there a cross listing permium for non-exchange traded depositary receipts?

1 ECONOMICS DEPARTMENT WORKING PAPERS SERIES N185/12/07 Is there a cross listing premium for nonexchange traded depositary receipts? by Thomas G. O’Connor Department Of Economics National University of Ireland, Maynooth NATIONAL UNIVERSITY OF IRELAND, MAYNOOTH MAYNOOTH CO. KILDARE A. IRELAND http://www.nuim.ie/academic/economics/ http://ideas.repec.org/s/nuim/mayecw.html http://eprints.nuim.ie/ 2 Is there a cross listing premium for non-exchange traded depositary receipts? Thomas G. O’Connor♣ September 9th, 2007. Abstract In this paper, I examine the valuation effects of trading in the U.S. as non-exchange issues i.e. Level 1 and 144 firms for non-U.S. firms. The study is motivated by two facts; first, while the number of new Level 2/3 issues has fallen 2001, Level 1 issues have remained an attractive listing option for non-U.S. firms. Second, while on theoretical grounds, firms from low-disclosure regimes have most to gain from exchange listing; these firms tend to list in the U.S. as non-exchange issues. Here, I examine whether the continuing attractiveness of, and the tendency of firms to choose a Level 1/144a listing is value enhancing. My results suggest that the tendency on the part of firms from low-disclosure regimes to choose non-exchange issues is justified. Relative to their highdisclosure peers, these firms tend to gain most from trading in the U.S. However, for Rule 144a issues, the valuation gains are short-lived. JEL Classification: G15, G34, G35. Key Words: Cross listing, Level 1, Rule 144a, Tobin’s q. Acknowledgements: The author would like to acknowledge the receipt of financial support from the Maynooth Finance Research Group (MFRG). The paper has benefited from helpful comments from Thomas Flavin, and from two anonymous referees. Any remaining errors are entirely my own. ♣ Department of Economics, NUI Maynooth, Maynooth, Co. Kildare, Ireland. This paper is a revised version of Chapter 6 of my PhD thesis at NUI Maynooth. E-mail Address: [email protected] 1 1. Introduction During the 1990’s, United States capital markets became the most attractive location for a secondary listing on International markets for non-U.S. firms. For example, at its peak, the number of depositary receipt programs numbered almost 2,200. At the same time, the share of foreign firms listed on European exchanges fell (See Pagano, Roell, Zechner (2002), and International Federation of Stock Exchanges (www.fibv.com)). For example, the share of foreign firms listed on the NYSE and the Nasdaq rose from 10.97% and 7.04% in 1995 to 19.95% and 10.44% in 2002, respectively. Foreign lists on the London Stock Exchange fell from 531 (21.22%) to 383 (16.81%) over the same period. Furthermore, non-U.S. firms have demonstrated a marked preference towards trading as Level 1 over-the-counter non-exchange issues (as opposed to Level 2/3 exchange issues) in the U.S. For example, in 2001 the number of Level 1 issues stood at 759, compared to 563 Level 2/3 exchange issues. In addition, the number of new Level 1 issues has outstripped new Level 2/3 issues in every year since 2001 (See Bank of New York (2006)). Grounded in what is commonly referred to as the ‘legal finance’ literature, an exchange-cross-listing (as opposed to a Level 1/Rule 144a issue) in the U.S. provides a remedy for firms to overcome their financing constraints at home, and thus facilitate their hitherto stagnated growth. Financing constraints tend to be greatest for firms domiciled in countries where investors are poorly protected (See La Porta et al., 1998 and Demerguc-Kunt and Maksimovic, 1998 for the legal finance view, and Coffee, 1999, 2002 and Lins et al., 2005 for arguments specific to cross-listing). Consistent with this line of reasoning, Doidge, Karolyi, and Stulz (2004, DKS Hereafter) outline a theoretical model whereby the valuation gains from exchange cross-listing, what they term a ‘cross listing premium’, is increasing in growth opportunities, and decreasing in domestic investor protection. This suggests that firms from low-disclosure regimes should demonstrate a marked preference for Level 2/3 exchange issues. However, Hope, Kang, and Zang (2007, HKZ Hereafter), document evidence to the contrary. Using logit analysis, they show that while firms from low-disclosure regimes are more likely cross-list; they are less likely to exchange cross-list (i.e. Level 2/3 issue). In this paper, I examine whether the decision on the part of these firms to list as Level 1/144a issues is justified, at least on the grounds of value1. Using a panel of Level 1 and Rule 144a issues over the period from 1990 to 2003, I begin by estimating the cross listing premium for Level 1 and 144a firms in calendar time (1997) as DKS (2004) do. From here, I motivate the use of a longitudinal approach, which is then outlined. My results suggest that the tendency on the part of lowdisclosure firms to trade in the U.S. as non-exchange issues is justified. Specifically, these firms tend to experience the greatest gains from trading, relative to their high-disclosure counterparts. However, unlike Level 1 issues, the valuation gains for 144a firms tend not to be long lasting. The paper is organised as follows. In the next section I outline the data. Then I present estimates of the cross listing premium in calendar and event time. Section 4 concludes. 1 In a recent paper, like HKZ (2007), I find that the greatest valuation gains to exchange cross listing accrue to firms from high-disclosure regimes (See O’Connor (2007)). HKZ (2007) hypothesize that this result is primarily driven by the greater costs faced by low-disclosure firms in their efforts to comply with U.S. GAAP. This paper differs from HKZ (2007) in two respects. First, I pay special attention towards examining nonexchange firms (i.e. Level 1 and Rule 144a issues). In their analysis, HKZ (2007) do not examine the valuation effects for these firms separately. They create two dummy variables: ‘XLIST’ that takes the value of 1 if the firms cross-lists in the U.S., and ‘ORG_EXC_XLIST’ which is 1 if the firm exchange cross-lists. Second, I estimate both cross-sectional and panel regressions. 2 2. Data I begin by sourcing a full list of firms with a cross listing in the U.S. All information on cross-listed firms is sourced from the Bank of New York, and cross-referenced with information sourced from Deutsche Bank, JP Morgan, the New York Stock Exchange, and Nasdaq. The final sample, outlined in Appendix 1 is comprised of 4,310 firms from 36 different countries: 3,624 domestic firms, 471 Level 1 firms, and 215 Rule 144a firms. From my original cross-listed sample of firms, I classify firms according to their first depositary receipt level, and classify simultaneous Level 1/Portal ‘listings’ as Level 1 issues. I outline in Appendix 1, the number of non-cross-listed firms, and the number of cross-listed firms listed in the United States. I exclude from my final sample firms domiciled in Russia, the Czech Republic and Indonesia because of a lack of data. I provide the percentage that each country contributes to each depositary receipt level and adopt an identical approach for my non-cross-listed sample. The majority of our non-cross-listed sample is domiciled in the U.K. There also exists a sizable difference across countries in their contribution to each depositary receipt level. For example, Hong Kong, Australia, U.K., and South Africa provide the majority of Level 1 issues, with 97 (20.59%), 61 (12.95%), 51 (10.83%), and 37 (7.86%) programs, respectively. Together, they supply 52.23% of the entire sample of Level 1 firms. In contrast, Argentina and Taiwan provide none. Similar trends are observed for private placement issues. The majority of these firms originate in India (50), Taiwan (42), and South Korea (21). Belgium, Denmark, Israel, Malaysia, and New Zealand provide no firm. I follow DKS (2004), and HKZ (2007) and employ Tobin’s q to measure firm value, where Tobin’s q is defined book value of debt + market capitalization book value of assets ⎛⎞ ⎜⎟ ⎝⎠ where book value of debt is calculated as book value total assets less the book value of equity. All variables are expressed in local currency, sourced from Worldscope and are collected on the 31st of December in each year from 1990 to 2003. I employ the following firm-level control variables in my empirical specifications: I use the average sales growth over the last two years (geometric average) and Global Industry q to account for firm and industry growth, respectively. Based upon primary standard industry classifications, the (yearly) mean Global Industry q is calculated as the average q of all global firms within each classification. I employ over 15,000 international firms from the Worldscope database to calculate the mean Global Industry q for each year. To remove the influence of outliers, I remove the top 1% of observations for Tobin’s q, two-year average sales growth, and total assets. Finally, I include La Porta, Lopez-de-Silanes, Shleifler, and Vishny (1998) country-level governance variables in order to examine the valuation effects of listing across different governance regimes. I employ legal origin (English Common, French, Scandinavian and German Civil Law), and anti-director rights index, an equally weighted index of 6 different shareholder rights, which ranges from a low of 0 to a high of 5. The country-level governance variables are outlined by country in Appendix 2. 3 3. Empirical Results This section presents the main results on cross listing and firm value. I begin by estimating the cross listing premium/discount for Level 1/Rule 144a firms in 1997 (as DKS (2004) do). In unreported results, I also estimate this cross-sectional relation in 2000 as HKZ (2007) do. The results are outlined in Table 1. I then proceed to analyse the relationship over time. To do so, I present univariate comparisons of cross-listed to non-cross-listed firms in calendar and event time. The results are presented in Table 2. Next, I estimate multivariate/panel data regressions that span the period from 1990 to 2003. Finally, in Table 4, I examine the dynamics of firm value around the time of listing. 3.1. Cross-sectional estimates of the cross-listing premium In Table 1, I present regression estimates of the impact of listing on the value of Level 1/Rule 144a firms. I present three sets of estimates: in columns (1-6), I present ordinary least squares estimates. In the remaining columns, I explicitly control for self-selection bias, and estimate treatment effects and two-stage least squares estimates, respectively. To conserve space, I have outlined the treatment effect methodology in greater detail in Appendix 3. In effect, I correct the ordinary least squares estimates for selection-bias, and estimate the following: ii ii qXCL λ =α+ β+δ +δ λ +ε (1) Where q is Tobin’s q, i Xis a vector of firm and country level controls, CL is a standard 0/1 dummy, corresponding to either a Level 1 over-the-counter issue, or a privately placed Rule 144a issue (the treatment effects and two-stage least squares are estimated separately for each cross-listed sub-set of firms), i λ is the inverse-mills ratio, generated from a first-stage probit model2 (i.e. proxy for unobserved/private information), and i ε is a standard error term. The results from Table 1 suggest the following. First, the ordinary least squares estimates presented in columns (1-2) suggest that Level 1 firms, unlike Rule 144a firms are worth more than non-cross-listed firms. In both columns, the coefficient estimate on the Level 1 dummy is large, and statistically different from zero. However, absent a self-selection correction, it is not clear whether this valuation premium is a cross-listing premium. I examine this issue in the remaining columns of Table 1. In contrast, the corresponding coefficient estimate for Rule 144a firms is small, and indifferent from zero. For Level 1 firms, the valuation premium is robust to the inclusion of firm, industry, and country-level control variables. In column 2, with all firm, industry, and control variables included, the coefficient on the Level 1 dummy is a statistically significant 0.23. The firm, industry, and control controls are all of the correct sign, and statistically different from zero. Firm value is increasing in firm and industry growth, and level of investor protection (anti-director rights). Larger firms tend to worth less. However the ordinary least squares estimates are biased (See Appendix 3). In the remaining columns of Table 1, I control for the endogeneity of the cross-listing decision, and estimate treatment effects and two-stage least squares estimates. In the treatment effects regressions, I include, along with the firm (excluding size), industry, and 2 The inverse-mills ratios i.e. for both Level 1 and Rule 144a firms are generated from a first-stage probit, whereby I model the decision to crosslist as a function of firm size (log of total assets (US$)), and legal origin. The results from the first-stage probit are available from the author upon request. To satisfy the exclusion restrictions, size is excluded from the second-stage selection-corrected valuation regression. The two-stage least squares estimates are generated using the same first-stage criteria. 4 country controls, a proxy for unobservable/private information i.e. the inverse mills ratio. The inclusion of the inverse-mills ratio dramatically affects the coefficient estimates on both cross-listing dummies. For both Level 1 and Rule 144a firms, the cross-listing dummy variables are both negative, and statistically different from zero. For Level 1 firms, the inclusion of the inverse mills ratio reverses the sign on the cross-listing dummy from positive to negative. In short, the positive sign seen in the ordinary least squares estimates are soaked up by the coefficient for the inverse mills ratio. However, of concern here is the flip in sign of the cross-listing dummy variables. For robustness sake, I also present two-stage least squares estimates. Consistent with the treatment effects estimates, when I control for the endogeneity of the listing decision, the coefficient estimates on both cross-listing dummies are now negative, and statistically so3. Interestingly, the coefficient estimate on the inverse mills ratio is positive and statistically different from zero for both sub-sets of firms. This implies that those unobservable factors that influence the decision to list, impact positively on firm value. However, once we account for this, there is no crosslisting premium. In fact, the results suggest the opposite, a cross listing discount. Next, I examine whether the quality of firms that list abroad has any impact on post-listing value. For example, one might expect that higher quality firms would reap greater benefits from listing in the U.S. I classify cross-listed firms as high quality firms, if they are worth more than their corresponding (domestic) non-cross-listed firm on the year of listing. To dos o, I use Relative Tobin’s q. Relative q is calculated as q of cross-listed firm mean q of domestic firms ⎛⎞ ⎜⎟ ⎝⎠ , where a Relative q of greater than 1 suggests that the cross-listed firm is worth more than their average counterpart noncross-listed domestic firm. A Relative q of less than 1 suggests the opposite. To shed some light on the relationship, I begin by outlining the unconditional relationship between Relative q on the list year, and post-listing Relative q. The data is calculated on a country-by-country basis, and is presented in a series of scatter plots (See Figure 2). The country-by-country data is outlined in Appendix 2. For both sets of firms, the relationship is positive: it appears that firms that are worth more on the list year continue to be worth more after listing. Next, I examine whether this relationship is robust to the inclusion of firm, industry, and country controls. To examine this, I create a simple dummy variable that is 1 if the firm is worth more than their counterpart domestic firms on the list year, and interact these with the cross-listing dummy variables. The results are presented in column 4 of Table 1. First, I find that the coefficient estimates on the interaction terms (i.e. Level 1 * High Rq, Rule 144a * High Rq) are positive and statistically significant form zero. However, with the inclusion of the interaction terms, the coefficient estimates on the Level 1 and Rule 144a dummies are now negative, and statistically different from zero. Taken together, the results suggest that it is only those firms that are more highly valued that gain from listing in the U.S. Finally, before I proceed to the longitudinal analysis, I use the cross-sectional estimates to make an initial attempt to examine the dynamics of corporate value, post-listing. In order to do so, I calculate for each firm, the number of years that each firm is listed in the U.S. in 1997. I denote this variable as ‘Years’. To examine the behaviour of firm value in the post-listing period, I interact this variable with each cross-listing dummy variable 3 Notice also that the constant is statistically different from zero in all of the ordinary least squares regression estimates, suggesting that missing variables may have an important influence on the results. However, when I include the IMR, the intercept terms are smaller, and no longer statistically different from zero. 5 creating, Level 1 * Years, and Rule 144a * Years, respectively. The results are presented in column 3 of Table 1, and suggest the following. First, the coefficient estimates suggest a slightly upward (downward) trend in value postlisting for Level 1 (Rule 144a) firms, although both sets of estimates are statistically different from zero. In contrast, I find that for firms that are worth more, the post-listing period is synonymous with an upward trend for both sets of firms. The coefficient estimates on the interaction terms ‘Level 1 * High Rq * Years’ and ‘Rule 144a * High Rq * Years’ are positive and different from zero. The results thus far suggest the following. First, after controlling for self-selection bias, Level 1 and Rule 144a firms are worth less after trading in the U.S. Once we control for ‘positive’ unobservable/private information, the coefficient estimates on both the Level 1 and Rule 144a dummies, are negative, and statistically so. Next, I find that only those firms that are worth more on the list year gain from trading in the U.S. as non-exchange issues. Furthermore, their value continues to increase after trading in the U.S. 3.2. Year-by-year and event time valuation comparisons In Table 2 (Panel A) I compare the value of cross-listed firms to non-cross-listed firms in each year from 1990 to 2003. For each subset of cross-listed firms, I outline the value of the mean and median firm in each year. In the remaining columns, I outline the mean and median adjusted Relative q measure. The mean/median Relative q is calculated as q of cross-listed firm mean or median q of domestic firms ⎛⎞ ⎜⎟ ⎝⎠ . The conclusions drawn from Panel A (and B) are largely dependent on how the Relative q measure is calculated. In general, when I compare cross-listed firms to the median non-cross-listed firms, cross-listed firms (both Level 1 and Rule 144a) are worth more than non-cross-listed firms. For example, using median-adjusted q, Level 1 firms are worth statistically more than domestic firms in every year, with the largest valuation difference originating in 1993. Rule 144a firms are worth more than the median domestic firm in all but one calendar year (1990). I compare in Table 2 (Panel B), the value of cross-listed firms to non-cross-listed firms in event time. I denote the list year as ‘0’, and compare cross-listed to non-cross-listed firms for the five years before to five years after listing. As before, I outline the mean and median adjusted Relative q measure. To complement these numbers, I present in Figures 3-4, the mean and median value of cross-listed firms, and the mean/median adjusted Relative q. As before, the conclusions drawn are largely contingent on the valuation adjustment employed. Using medianadjusted q, both sets of firms are worth more in every period around the list year. However, irrespective of the adjustment method employed, the trends in value around the time of listing remain the same. Prior to trading in the U.S., the value of Level 1 firms falls in almost every period leading up to the list year. Value continues to fall post-listing, although the magnitude of the decline is much smaller. On a median-adjusted basis, Level 1 firms, do, nevertheless continue to be worth more than domestic firms (although not on a mean-adjusted basis). Finally, the data presented in Table 2 (Panel B) suggest that Rule 144a firms ‘time’ their listing in the U.S. Rule 144a firms experience a sizable run-up in value prior to listing in the U.S. This is followed by a corresponding fall-off, postlisting. On a median-adjusted basis, around the time of listing, Rule 144a firms continue to be worth more than domestic firms, and the greatest valuation difference occurs on the year of listing. 6 The last three rows of Panel B summarize the value of firms subsequent to cross listing. Level 1 firms are worth less after listing in the U.S., both on an absolute and relative basis. The average Rule 144a firm is worth less, but shows no change on a median-adjusted basis. In summary, the results from Table 2 suggest the following. First, on a median-adjusted basis, Level 1 and Rule 144a firms are worth more than domestic firms in almost every calendar period. Around the time of listing, both sets of firms continue to be worth more than domestic firms. The value of Level 1 firms falls around the time of listing, but the greatest fall in value is experienced pre-listing. Rule 144a firms ‘time’ their listing in the U.S. However, these univariate comparisons do not control for other factors that influence firm value. In the next section, I outline multivariate panel regressions. 3.3. Panel regression estimates of the cross listing premium In this section I examine the effect of cross listing on firm value. I begin with the following specification, whereby I model firm value as a function of firm characteristics: it it 1 it 2 it it q X Level 1 Rule 144a u=α+ β+δ +δ + (2) Where it X is a set of exogenous observable characteristics of the firm, i t Level 1 , and i t Rule 144a are standard dummy variables that take the value of 1 if the firm trades in the United States as a Level 1, or as a privately placed Rule 144a issue on Portal, respectively. it u is a standard idiosyncratic disturbance term, and 12 {,, , }αβδ δ is a vector of parameters to be estimated. I explicitly acknowledge the non-randomness of the cross-listed sample, and model their decision to cross list as follows: *** it it it it it it it CL Z ,CL 1 if CL 0,CL 0 if CL 0=γ +η = > = < (3) Where * it CL ( it it it Level 1 ,Rule 144a CL∈) is an unobserved latent variable, i t Z is a set of observable firm-level characteristics that determine the decision to cross-list in the United States, and i t η is a disturbance term. Selection bias arises because of the correlation between i tit Level 1 ,Rule 144a and i t u. This correlation can arise in two instances i.e. (1) selection on observables which arises through correlation between it Z and it u, or (2) through selection on unobservables i.e. correlation between i t η and i t u. Both instances render ordinary least squares estimates of the effect of cross listing on value, biased. In my analysis, I estimate the effect of listing on firm value using two approaches. First, I estimate a variant of the standard firm-fixed effect regression. Specifically, I estimate a pooled ordinary least squares regression, with unobserved heterogeneity specified as Mundlak (1978) correction terms i.e. time averages of time-variant 7 explanatory variables over time ( T iii i it s1 1 Xa, where X X T= α= ζ+ = ∑ )4. I do not estimate a firm-fixed effects model as I find that the assumption that strict exogeneity holds is violated5. Consequently, I estimate the following: it it 1 it 2 it 3 it i it q X Level 1 Level 2/3 Rule 144a X=α+ β+δ +δ +δ + ζ+μ (4) Next, I explicitly model for unobservables by proxying for them. To do so I estimate a treatment effects model, whereby I augment the second stage equation with a selection correction term namely the inverse mills ratio, from a first-stage probit model. The inverse mills ratios are generated on a year-by-year basis (using yearly probit models), thus resulting in a series of time-variant unobservables in the second stage equation. Consequently, I estimate: it it 1 1 it 1 2 i it qXC c=α+ β +δ +λβ + +υ (5) I outline this method in greater detail in Appendix 3. The coefficient estimates corresponding to Eqs (4-5) are presented in Table 3. In Table 3, I replicate much of the analysis that I originally examined in a cross-sectional setting in Table 1. First, in both the pooled ordinary least squares and treatment effect regressions (See Columns 1 and 6), the coefficient estimate on the Level 1 dummy is positive and statistically different from zero. In contrast, Rule 144a firms are not worth more. Like before, I find that the coefficient estimate (Lambda) on the IMR is positive, and statistically different for both sets of firms. Next, I examine the dynamics of firm value in the post-listing period. As in Table 1, I interact each cross-listing dummy variable with ‘Years’, which denotes the number of years each firm is listed in the U.S., in each cross-sectional period. The coefficient estimates are outlined in Columns 2, 7, and 10. Unlike the results presented in Table 1, the results now suggest that a Level 1 issue is associated with a fall in value, post-listing, albeit insignificantly so. Rule 144a firms experience an even greater downward trend in value, which is statistically different to zero. In Table 4, I return to this issue, and examine the short and long-term valuation gains from listing. Next, I examine whether better quality firms perform better. As before, I classify better firms, as those that are worth more than their counterpart domestic firms on the year of listing. The results are outlined in Columns 3, 8, and 11. In line with the coefficient estimates presented in Table 1, I find that better quality Level 1 and Rule 144a firms gain the most from trading in the U.S. As before, the inclusion of the interaction terms dramatically reduces the coefficient estimates on the Level 1 and Rule 144a dummy variables. The coefficient estimate on the Level 1 dummy is smaller (0.01), and no longer statistically different from zero. The corresponding coefficient estimate on the Rule 144a dummy is negative, and statistically different from zero. Using the interaction terms ‘Level 1 * Rel q * Years, Rule 144a * Rel q * Years’, I find that the value of these firms continues to increase post-listing. However, the gains are greatest for Level 1 firms. 4 The Mundlak (1978) correction terms are included in all pooled ordinary least squares regressions, but to conserve space, they are not reported. They are available form the author upon request. However, I do report the p-value from a standard F-stat that tests whether they are jointly different from zero. 5 Violations of strict exogeneity are likely in this case because of feedback effects i.e. from Tobin’s q to future values of the cross-listing dummy variables. I formally test for this possibility, following Wooldridge (2002), by inserting the one-year forwarded cross-listing variables as independent variables and testing whether their coefficients are jointly equal to zero. The results suggest that the assumption of strict exogeneity holds is violated. 14 i iii qXCL λ = α+ β+δ +δ λ +ε (A.6) The panel version of the treatment effects model is as follows. First, the valuation equation is given by: i tit1ittit qXCL = α+ β +δ +α +υ (A.7) And the selection equation: *** it it it it it it it CL Z ,CL 1 if CL 0,CL 0 if CL 0=γ +η = > = < (A.8) Under the assumption that the error terms are bivariate normal, the generalized residual from the first-stage probit is: i tit 1 it E(q |C 1) ( Z ) υ = =ρσ λ β (A.9) Where 1it (Z)λβ is computed as it i t (Z) (Z) ϕβ φβ , which is a series of time-specific ‘inverse mills ratios’. Substituting into Eq. (A.7) yields: it it 1 it 1 2 i it qXCL c = α+ β +δ +λβ + +υ (A.10) 15 References 1. Bank of New York, 2006. The Depositary Receipts Markets 2005 Yearbook. 2. Campa, J.M., and Kedia, S., 2002. Explaining the diversification discount. Journal of Finance, 4, 1731-1762. 3. Doidge, C., Karolyi, G.A., Stulz, R.M., 2004. Why are foreign firms listed in the U.S. worth more? Journal of Financial Economics, 71, 205-238. 4. Hope, O., T. Kang, and Zang, Y., 2007. Bonding to the improved disclosure environment in the United States: Firms’ listing choices and their capital market consequences. University of Toronto working paper. 5. La Porta, R., F. Lopez-de-Silanes, Shleifer, A., and Vishny, R., 1998. Law and finance. Journal of Political Economy, 106, 1113-1155. 6. Marosi, A., and Massoud, N., 2006. You can enter but you cannot leave – U.S. securities markets and foreign firms. University of Alberta working paper. 7. O’Connor, T.G., 2007. Does cross listing in the U.S. really enhance the value of emerging market firms? Department of Economics, National University of Ireland working paper. 8. Pagano, M., Roell, A.A, and J. Zechner, 2002. The geography of equity listing: Why do European companies list abroad? Journal of Finance, 57, 2651-2694. 9. Witmer, J., 2006. Why do firms cross-delist? An examination of the determinants and effects of cross delisting. Bank of Canada working paper. 10. Wooldridge, J.M. 2002. Econometric analysis of cross-section and panel data. Cambridge, Massachusetts: MIT Press. 16 Table 1: Cross-sectional estimates of the cross-listing premium in 1997. Level 1 & Rule 144a Level 1 Rule 144a OLS TE & 2SLS (1) (2) (3) (4) (5) TE 2SLS TE 2SLS Constant 2.85 [5.81]*** 2.60 [4.93]*** 2.60 [4.88]*** 2.58 [3.85]*** 2.63 [4.92]*** 0.02 [0.11] 0.07 [0.34] 0.20 [0.92] 0.15 [0.52] Level 1 0.28 [3.16]*** 0.23 [2.65]** 0.21 [1.85]* -0.33 [3.85]*** -0.01 [0.11] -1.28 [4.61]*** -2.91 [6.56]*** Rule 144a 0.01 [0.09] 0.03 [0.14] 0.19 [0.75] -0.43 [3.06]*** -0.26 [1.40] -5.29 [6.17]*** -7.66 [5.41]*** Level 1 * Years 0.01 [0.15] Rule 144a * Years -0.06 [1.08] Level 1 * High Rq 1.20 [6.69]*** Rule 144a * High Rq 1.31 [7.92]*** Level 1 * High Rq * Years 0.22 [4.87]*** Rule144a * High Rq * Years 0.34 [4.91]*** Global q 0.75 [5.44]*** 0.70 [5.07]*** 0.69 [5.04]*** 0.65 [4.77]*** 0.67 [4.91]*** 0.72 [8.70]*** 0.70 [6.96]*** 0.71 [6.82]*** 0.79 [5.66]*** Sales Growth 1.22 [3.67]*** 1.24 [3.51]*** 1.24 [3.51]*** 1.24 [3.48]*** 1.23 [3.47]*** 1.48 [9.35]*** 1.50 [7.87]*** 1.39 [6.91]*** 1.72 [6.38]*** Log (Total Assets) -0.13 [6.56]*** -0.13 [6.61]*** -0.13 [6.46]*** -0.12 [6.37]*** -0.13 [6.52]*** Anti-Director 0.11 [3.29]*** 0.11 [3.34]*** 0.12 [3.48]*** 0.11 [3.42]*** 0.12 [8.12]*** 0.15 [7.93]*** 0.11 [5.88]*** 0.10 [3.61]*** Lambda 0.71 [5.18]*** 2.34 [6.17]*** #Obs (Firms) 2,467 2,467 2,467 2,467 2,467 2,467 2,467 2,467 2,467 R-Squared 0.11 0.13 0.13 0.16 0.14 - - - - Prob>F 0.000 0.000 0.000 0.000 0.000 - 0.000 - 0.000 Prob>Chi - - - - - 0.000 - 0.000 - In this table, I report regression (cross-section) estimates of the impact of listing on firm value in 1997. In columns 1-5, I estimate ordinary least squares estimates with standard errors clustered at the country level. In the remaining columns I estimate treatment effects (two-stage) and two-stage least squares estimates. I proxy for value using Tobin’s q. All variables are defined in the text. ***, **, and * denotes significance at the 10%, 5%, and 1% level, respectively. 17 Table 2: Comparison of cross-listed to non-cross-listed firms Level 1 Rule 144a Mean Median Mean-Adj Relative q Median-Adj Relative q Mean Median Mean-Adj Relative q Median-Adj Relative q Panel A Calendar Time 1990 1.62 1.53 0.96 1.08* 1.03 1.03 0.86* 0.87** 1991 1.66 1.49 0.96 1.06* 1.22 1.17 0.92* 1.00 1992 1.79 1.52 1.05 1.15*** 1.72 1.50 0.96 1.10** 1993 1.98 1.82 1.04 1.19*** 1.73 1.62 0.89 1.04* 1994 2.02 1.72 1.04 1.16*** 2.52 2.21 1.01 1.21*** 1995 1.85 1.62 1.02 1.15*** 1.93 1.83 0.96 1.14*** 1996 1.85 1.59 0.99 1.16*** 1.76 1.53 0.94** 1.13*** 1997 1.89 1.60 0.95* 1.14*** 1.85 1.58 0.94* 1.14*** 1998 1.60 1.36 0.96*** 1.15*** 1.56 1.30 0.88*** 1.09** 1999 1.70 1.45 0.91*** 1.14*** 1.73 1.38 0.89*** 1.17*** 2000 1.68 1.40 0.89*** 1.17*** 1.76 1.32 0.96*** 1.26*** 2001 1.53 1.32 0.91** 1.15*** 1.42 1.23 0.92*** 1.12*** 2002 1.53 1.36 0.92* 1.13*** 1.46 1.27 0.91*** 1.09** 2003 1.63 1.45 0.89* 1.12*** 1.58 1.38 0.92*** 1.11*** Panel B Event Time Mean Median Mean-Adj Relative q Median-Adj Relative q Mean Median Mean-Adj Relative q Median-Adj Relative q -5 2.17 1.71 1.15 1.33*** 1.65 1.33 0.90 1.02 -4 2.09 1.67 1.06 1.30*** 1.59 1.33 0.89 1.02 -3 1.99 1.61 1.07 1.26*** 1.86 1.53 0.98 1.14*** -2 1.97 1.61 1.06 1.28*** 2.17 1.72 1.09 1.25*** -1 1.98 1.60 1.05 1.27*** 2.14 1.64 1.03 1.24*** 0 1.89 1.58 1.03 1.23*** 2.18 1.86 1.07 1.27*** 1 1.78 1.53 0.98 1.16*** 1.96 1.70 1.05 1.22*** 2 1.78 1.53 0.95 1.14*** 1.82 1.48 0.97 1.18*** 3 1.73 1.46 0.95 1.14*** 1.68 1.43 0.92 1.13*** 4 1.70 1.45 0.96 1.15*** 1.58 1.30 0.90 1.13*** 5 1.62 1.38 0.92 1.11*** 1.58 1.28 0.87 1.14*** Event Time (Before-After) Mean Median Mean-adj Relative q Median-adj Relative q Mean Median Mean-adj Relative q Median-adj Relative q Before 1.94 1.62 1.06 1.25 1.86 1.37 0.98 1.13 After 1.70 1.47 0.94 1.14 1.67 1.37 0.93 1.14 Difference (0.24)*** (0.15)*** (0.12)*** (0.11)*** (0.19)*** 0.00 (0.05)** 0.01 In this table, I compare the value of cross-listed to non-cross-listed firms in calendar and event time. For cross-listed firms, I present the mean and median value. To compare cross-listed to non-cross-listed firms, I calculate the mean and median adjusted relative q measure. Both measures are calculated as the value of each cross-listed firm divided by the mean (median) value of non-cross-listed firms. I proxy for value using Tobin’s q. In the remaining rows, I calculate value pre and post-listing. ***, **, and * denotes significance at the 10%, 5%, and 1% level, respectively. 18 Table 3: Panel regression estimates of the cross-listing premium. Level 1 & Rule 144a Treatment Effects POLS Level 1 Rule 144a (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Level 1 0.16 [2.76]*** 0.20 [2.89]*** 0.01 [0.12] 0.09 [1.51] 0.51 [3.59]*** 0.17 [2.97]*** 0.22 [3.11]*** 0.02 [0.28] -0.06 [0.83] 0.10 [1.00] -0.33 [6.46]*** Rule 144a 0.02 [0.24] 0.18 [1.90]* -0.25 [4.76]*** -0.04 [0.65] 0.46 [2.42]*** Level 1 * Years -0.02 [1.08] -0.02 [1.15] Level 1 * Rel q 0.42 [3.68]*** 0.41 [3.70]*** Level 1 * Relq * Years 0.04 [2.63]*** Level 1 * ADR -0.10 [2.54]*** Rule 144a * Years -0.05 [2.83]*** -0.04 [2.6]*** Rule 144a * Rel q 0.51 [4.78]*** 0.53 [5.01]*** Rule 144a * Relq * Years 0.04 [1.80]* Rule 144a * ADR -0.14 [2.47]*** Sales Growth 0.60 [7.36]*** 0.59 [7.29]*** 0.60 [7.43]*** 0.54 [6.89]*** 0.54 [6.86]*** 0.74 [9.4]*** 0.74 [9.3]*** 0.74 [9.3]*** 0.69 [8.2]*** 0.68 [8.2]*** 0.69 [8.25]*** Log (Total Assets) -0.23 [10.4]*** -0.23 [10.4]*** -0.23 [10.26]*** -0.14 [9.81]*** -0.14 [9.84]*** Global Industry q 0.73 [11.5]*** 0.73 [11.4]*** 0.72 [11.4]*** 0.72 [11.37]*** 0.72 [11.29]*** 0.70 [11.0]*** 0.70 [11.0]*** 0.70 [11.0]*** 0.75 [11.6]*** 0.74 [11.6]*** 0.74 [11.6]*** Anti-Director Rights 0.11 [11.4]*** 0.11 [11.5]*** 0.11 [11.5]*** 0.11 [11.47]*** 0.12 [12.07]*** 0.14 [13.3]*** 0.14 [13.3]*** 0.14 [13.3]*** 0.10 [10.0]*** 0.10 [10.1]*** 0.10 [10.2]*** Lambda - - - - - 0.27 [10.5]*** 0.27 [10.5]*** 0.27 [10.3]*** 0.15 [8.7]*** 0.15 [8.6]*** 0.15 [8.61]*** Time Dummies No No No No No Yes Yes Yes Yes Yes Yes Time Ave (Mundlak) Yes Yes Yes Yes Yes No No No No No No Prob>F (Mundlak) 0.000 0.000 0.000 0.000 0.000 - - - - - - # Obs (Firms) 4,310 4,310 4,310 4,310 4,310 4,310 4,310 4,310 4,310 4,310 4,310 R-Squared 0.12 0.12 0.12 0.12 0.12 0.111 0.111 0.114 0.102 0.103 0.105 19 Table 4: Evolution of Tobin’s q by legal characteristics. Level 1 All All: Relative q Anti-Director Rights Legal Origin Mean-Adj Median-Adj Above Median Below Median English Common Civil Law List year 0.27 [3.01]*** 0.16 [3.61]*** 0.19 [3.37]*** 0.21 [1.94]* 0.30 [1.98]** 0.11 [1.01] 0.33 [2.31]** 1 year after list 0.19 [2.19]** 0.13 [3.36]*** 0.12 [2.51]** 0.24 [1.97]** 0.05 [0.48] 0.26 [1.91]* -0.03 [0.42] 2 years after list 0.20 [2.19]** 0.12 [2.90]*** 0.13 [2.61]*** 0.11 [0.94] 0.27 [1.78]* 0.15 [1.13] 0.14 [1.38] 3 years after list 0.08 [0.96] 0.08 [1.77]* 0.08 [1.42] 0.00 [0.01] 0.16 [1.19] 0.00 [0.00] 0.09 [0.80] 4 years after list 0.14 [1.53] 0.11 [2.47]** 0.12 [2.13]** 0.08 [0.67] 0.17 [1.42] 0.07 [0.53] 0.11 [0.85] 5 years after list 0.07 [0.79] 0.07 [1.70]* 0.07 [1.25] 0.03 [0.26] 0.06 [0.63] 0.00 [0.06] 0.08 [0.71] > 5 years after list 0.18 [2.15]** 0.10 [2.00]** 0.13 [2.04]** 0.08 [0.78] 0.31 [2.44]** 0.05 [0.41] 0.28 [2.25]** Global Industry q 0.77 [11.80]*** 0.34 [11.15]*** 0.45 [10.97]*** 0.77 [8.46]*** 0.68 [7.65]*** 0.71 [5.23]*** 0.79 [10.82]*** Sales growth 0.49 [6.18]*** 0.28 [6.67]*** 0.30 [5.86]*** 0.39 [3.32]*** 0.79 [7.34]*** 0.35 [2.55]** 0.48 [4.91]*** Total Assets -0.13 [8.86]*** -0.15 [14.61]*** -0.14 [11.05]*** -0.19 [8.24]*** -0.12 [6.97]*** -0.18 [5.80]*** -0.10 [6.26]*** Time Dummies No No No No No No No # Obs (Firms) 4,310 4,310 4,310 2,172 2,138 1,490 2,820 2 R 0.09 0.07 0.08 0.09 0.11 0.08 0.10 Pr F> 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Pr F(Mundlak)> 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Rule 144a All All: Relative q Anti-Director Rights Legal Origin Mean-Adj Median-Adj Above Median Below Median English Common Civil Law List year 0.18 [1.42] 0.04 [0.84] 0.01 [0.12] 0.47 [1.36] 0.20 [1.80]* 0.54 [1.49] 0.07 [0.64] 1 year after list 0.13 [1.22] 0.11 [1.87]* 0.09 [1.28] -0.27 [1.48] 0.37 [2.91]*** -0.07 [0.32] 0.20 [1.60] 2 years after list 0.10 [0.94] 0.05 [1.00] 0.06 [1.00] -0.34 [2.52]*** 0.35 [2.75]*** -0.29 [1.94]* 0.21 [1.68]* 3 years after list -0.04 [0.53] 0.02 [0.42] 0.02 [0.48] -0.39 [3.76]*** 0.21 [2.31]** -0.37 [3.13]*** 0.08 [0.86] 4 years after list 0.00 [0.01] 0.04 [0.68] 0.03 [0.53] -0.21 [1.46] 0.17 [1.33] -0.14 [0.85] 0.04 [0.33] 5 years after list 0.07 [0.47] 0.02 [0.35] 0.11 [1.01] 0.04 [0.11] 0.13 [1.02] 0.16 [0.47] 0.00 [0.04] > 5 years after list -0.16 [2.31]** -0.07 [2.08]** -0.05 [1.05] -0.25 [1.95]* -0.07 [1.02] -0.14 [1.03] -0.19 [2.91]*** Global Industry q 0.76 [11.75]*** 0.34 [11.10]*** 0.44 [10.94]*** 0.76 [8.34]*** 0.67 [7.58]*** 0.70 [5.20]*** 0.78 [10.75]*** Sales growth 0.49 [6.19]*** 0.28 [6.66]*** 0.30 [5.85]*** 0.40 [3.44]*** 0.76 [7.10]*** 0.34 [2.51]*** 0.47 [4.82]*** Total Assets -0.13 [8.61]*** -0.15 [14.71]*** -0.14 [10.95]*** -0.19 [8.00]*** -0.13 [7.16]*** -0.18 [5.55]*** -0.10 [6.23]*** Time Dummies No No No No No No No # Obs (Firms) 4,310 4,310 4,310 2,172 2,138 1,490 2,820 2 R 0.09 0.07 0.08 0.10 0.11 0.08 0.09 Pr F> 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Pr F(Mundlak)> 0.000 0.000 0.000 0.000 0.000 0.000 0.000 20 Figure 1 Cross-Listing Premium/Discount and Average Years of Listing in 1997 21 Figure 2 Relative q on list year and average post-listing Relative q 22 Figure 3 Absolute and Relative value of Level 1 firms in event time 23 Figure 4 Absolute and Relative value of Rule 144a firms in event time