Cross-Country Listing and Trading Volume: Evidence from...

28
Abstract This study investigates the effects of listing on the U.S. exchanges on trading volume for stocks listed on the two Canadian stock exchanges: the Toronto Stock Exchange (TSE) and the Vancouver Stock Exchange (VSE). The results show substantial differences between the two samples. When a TSE security is cross-listed, both trading volume and stock turn- over, the number of shares traded as a percentage of number outstanding, almost double their pre-listing levels. In contrast, when a VSE stock is cross-listed, there is only a slight increase in trading volume and a sharp decline in turnover. The TSE is also able to main- tain its pre-listing levels of trading volume in cross-listed securities, whereas the VSE loses about half the trading volume in these stocks to the U.S. exchanges. Even after controlling for the firm-specific factors, the Canadian exchange-specific factors remain the dominant factors in explaining the cross-sectional variation in liquidity effects. Neither the differences in trading costs nor in listing and disclosure requirements between the two exchanges explain these results. 1. Introduction The purpose of this study is to provide insights into the determinants of liquidity gains of cross-country listing. To enhance stock liquidity is one of the major motivations of firms listing their securities abroad. 1 An in- creasing number of firms are listing across borders to attract international investors. For example, 179 foreign companies listed their securities on the New York Stock Exchange between 1991 and 1995. The value of for- eign equity trades on the NYSE, AMEX, and NASDAQ reached $715 billion in 1994, and trading in non-U.S. stocks on the NYSE grew 35% in 1995 compared to a 19% increase in overall trading. 2 A relevant question is whether the liquidity gains from cross-listing are evenly distributed across firms. * I am grateful to Rex Thompson, Richard Levich, two anonymous reviewers and the seminar par- ticipants at the University of British Columbia and the Northern Finance Association meetings for many valuable comments and suggestions. The financial support from the Social Sciences and Humanities Research Council is gratefully acknowledged. All correspondence to: Usha R. Mittoo, Faculty of Management, University of Manitoba, Winnipeg, Manitoba R3T 2N2. Phone: (204) 474- 8969, Fax: (204) 474-7544, EMAIL: [email protected]. © Blackwell Publishers Ltd. 1997, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA. Cross-Country Listing and Trading Volume: Evidence from the Toronto and Vancouver Stock Exchanges Usha R. Mittoo* University of Manitoba Journal of International Financial Management and Accounting 8:3 1997

Transcript of Cross-Country Listing and Trading Volume: Evidence from...

Abstract

This study investigates the effects of listing on the U.S. exchanges on trading volume forstocks listed on the two Canadian stock exchanges: the Toronto Stock Exchange (TSE) andthe Vancouver Stock Exchange (VSE). The results show substantial differences betweenthe two samples. When a TSE security is cross-listed, both trading volume and stock turn-over, the number of shares traded as a percentage of number outstanding, almost doubletheir pre-listing levels. In contrast, when a VSE stock is cross-listed, there is only a slightincrease in trading volume and a sharp decline in turnover. The TSE is also able to main-tain its pre-listing levels of trading volume in cross-listed securities, whereas the VSE losesabout half the trading volume in these stocks to the U.S. exchanges. Even after controllingfor the firm-specific factors, the Canadian exchange-specific factors remain the dominantfactors in explaining the cross-sectional variation in liquidity effects. Neither the differencesin trading costs nor in listing and disclosure requirements between the two exchangesexplain these results.

1. Introduction

The purpose of this study is to provide insights into the determinants ofliquidity gains of cross-country listing. To enhance stock liquidity is oneof the major motivations of firms listing their securities abroad.1 An in-creasing number of firms are listing across borders to attract internationalinvestors. For example, 179 foreign companies listed their securities onthe New York Stock Exchange between 1991 and 1995. The value of for-eign equity trades on the NYSE, AMEX, and NASDAQ reached $715billion in 1994, and trading in non-U.S. stocks on the NYSE grew 35% in1995 compared to a 19% increase in overall trading.2 A relevant questionis whether the liquidity gains from cross-listing are evenly distributedacross firms.

* I am grateful to Rex Thompson, Richard Levich, two anonymous reviewers and the seminar par-ticipants at the University of British Columbia and the Northern Finance Association meetings for many valuable comments and suggestions. The financial support from the Social Sciences andHumanities Research Council is gratefully acknowledged. All correspondence to: Usha R. Mittoo,Faculty of Management, University of Manitoba, Winnipeg, Manitoba R3T 2N2. Phone: (204) 474-8969, Fax: (204) 474-7544, EMAIL: [email protected].

© Blackwell Publishers Ltd. 1997, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

Cross-Country Listing and Trading Volume:Evidence from the Toronto and VancouverStock Exchanges

Usha R. Mittoo*University of Manitoba

Journal of International Financial Management and Accounting 8:3 1997

While many recent studies have examined the stock price behaviouraround foreign listing, only a few have analyzed the liquidity effects ofsuch listing.3 Overall, the evidence reveals that while average trading vol-ume increases after cross-listing, the cross-sectional variation in liquidityeffects is substantial. Foerster and Karolyi (1993, 1994) study these effectsfor Canadian stocks that cross-listed in the United States and find thatwhile average volume more than doubled after listing, the median gain wasonly 62%. Canadian trading volume also exhibited substantial variationacross stocks. Of the 52 sample stocks, only 20 realized a large increase.The remaining 32 stocks experienced a decline in their home marketvolume. The effect on bid-ask spread also differed widely across stocks.Hargis (1996) confirms similar findings in a sample of stocks from fourLatin American emerging markets that listed on the U.S. exchanges. Ofthe 65 sample stocks, only 28 showed a significant increase, while nineexperienced a significant decrease. The remaining 37 stocks had no sig-nificant change in the total trading volume. Domowitz et al. (1995) exam-ine a set of 25 Mexican stocks that listed as ADRs on the NYSE between1989 and 1992 and find that while domestic trading volume increased for 62% of the companies, the differences among shares with differentrestrictions on foreign investments were significant.

Thus, the average liquidity effect conceals a large cross-sectional vari-ation across stocks; a small number of firms gain significantly, while othershave only modest to insignificant gains in liquidity. Very little evidence,however, is available on the determinants of the liquidity gains of foreignlisting. Several factors related to country, exchange, and firm characteristicscould potentially determine the liquidity effects. Isolating these factors in a large sample from different countries is not an easy task, especiallybecause of the difficulty in obtaining quality data for stocks listed fromsmall, and newly emerging markets which form a large percentage of foreignlisting on the U.S. exchanges.

This paper investigates these factors by comparing a sample of firmslisting in the U.S. from two Canadian stock exchanges—the Toronto StockExchange (TSE) and the Vancouver Stock Exchange (VSE)—that representtwo extremes in the type of securities listed and other characteristics. The TSE is the largest Canadian stock exchange with about 75% of thetrading value in Canadian stocks; the VSE is primarily a venture capitalexchange inhabited largely by highly speculative stocks. Both marketsalso differ in their reputation of the institutional and regulatory environ-ment concerning fair trading practices and investor protection. The TSE is perceived to be tightly regulated while the VSE has an unfavourable© Blackwell Publishers Ltd. 1997.

148 Usha R. Mittoo

reputation in protecting investors.4 Trading structures also differ on thetwo markets. The TSE is an agency market with designated market makers,whereas the VSE is a pure auction market with no market makers. Institu-tional trading is also lower on the VSE and comprises about 20% of thetotal VSE trading compared to more than 50% on the TSE.5

This study contributes to the cross-listing literature in several areas. First,the influence of firm- and exchange-specific factors on liquidity gains isanalyzed after controlling for the country-specific factors. Second, the de-terminants of the trading location in cross-listed stocks are also investigated.Whether the domestic or foreign exchange emerges as a dominant marketin cross-listed stocks is of particular interest to the smaller stock exchangessince cross-listed stocks form a major component of their trading. Theanalysis is also done using stock turnover, the number of shares traded as apercentage of the number of shares outstanding, in addition to the marketadjusted measures of trading volume used in previous studies. Since for-eign listings are generally accompanied by new stock issues, stock turn-over controls for the effect of the new shares issued by the firms.

The evidence reveals substantial differences in the two samples. Whena TSE stock is cross-listed, there is a large increase in total trading volumeand stock turnover. In contrast, the VSE stocks experience a sharp declinein stock turnover. The TSE is also able to maintain its pre-listing levels oftrading in the cross-listed securities. The VSE, on the other hand, losesabout half its trading volume in these securities to the U.S. exchanges.Even after controlling for the firm-specific factors, the exchange-specificfactors appear to be the dominant factors in explaining the cross-sectionalvariation in liquidity effects.

The rest of the paper is organized as follows. Section II discusses themain characteristics of the U.S. and Canadian stock exchanges. Section IIIdiscusses the data and section IV presents the empirical results. Section Vsummarizes the findings and presents some directions for future research.

II. Main Characteristics of the U.S. and Canadian Stock Markets

The main characteristics of the Canadian and U.S. stock market structuresand regulation are summarized in Table 1. The U.S. over-the-countermarket, NASDAQ, is a dealer system in which one or more dealers makea market in each stock. These dealers compete with each other but notwith the public order flows. The major U.S. stock exchanges, NYSE andAMEX, are match-making systems where intermediaries act as agents for

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 149

Table 1. Main Characteristics of the Canadian and U.S. Stock Markets1

NYSE/AMEX NASDAQ TSE VSE

Market structure Agency/Auction Dealer market Agency/Auction Auction

Market makers Monopoly Multiple dealers Monopoly No market makersAffirmative obligation No obligation Limited affirmative –

to make the market obligationCompetition with floor Competition with other Competition with floor –

traders and public orders dealers traders and public orders –

Order book Not public No Order Book Electronic Order Book No Order Book

Listing requirements High/Medium Low Medium Minimal

Minimum net worth2 US$18,000,000/ US$2,000,000 C$1,000,000 C$275,000 in seedUS$4,000,000 capital and C$200,000

in expenditures

Minimum trading tick size US$1/32 US$1/64 C$1/200 C$1/100

Surveillance system yes no yes yes

Short selling rule Not allowed on an uptick No restrictions Not allowed on an uptick Not allowed on an uptick

1 Based on Schwartz (1988), Tinic and West (1974), and information collected from various publications of the stock exchanges.2 Minimum net worth is one of the main listing requirements. Other requirements include requirements on the minimum number of shareholders, minimum netincome and working capital and these also vary on different stock exchanges.

investors but allow for dealers who are official market makers, referred toas the specialists. The specialists have an affirmative obligation in per-forming a continuing market-making function in the assigned stock. Theexchanges hold them responsible if transaction-to-transaction pricechanges or price changes involving trades of over one thousand sharesexceed certain limits.

The Canadian stock exchanges, TSE and VSE, are also match-makingsystems but differ from the U.S. stock exchanges in the way the dealershipfunction is organized. On both exchanges, the method of market-makingfocuses primarily on the development of an agency marketplace in whichpublic buyers and sellers trade with each other. The TSE has officialmarket-makers, referred to as the Registered Traders (RTs). The RT hasan obligation to make a market in the assigned stocks on the TSE similarto that of the specialist on the NYSE and AMEX, but the primary role ofthe RT is to be a passive facilitator of trading. The RTs have the privilegeof trading on the floor for firm accounts in which they participate in the profits but their primary role is to make a fair and orderly market in the assigned stocks. The RT is not supposed to contribute to swings inthe price of a stock and for this reason the exchange requires that no morethan 30% of the RT’s trades can be destabilizing trades.6 The VSE, on the other hand, has no designated official market-maker. It is a pure auctionmarket where a dealer’s orders compete with orders from other dealersand customers.

The TSE has higher transparency than the NYSE and AMEX since theorder book on the TSE is electronic and the TSE member firms have ac-cess to the exchange’s electronic book.7 The NASDAQ relies on competitionamong dealers to regulate the market. The TSE also has the minimumallowable price variation of C$1/200, for stocks trading at less than C$0.50 compared to the minimum price variation of US$1/32 on theNYSE and AMEX, and US$1/64 on the NASDAQ.8

Table 1 also shows that the listing requirements on the VSE are muchlower relative to those of other exchanges, consistent with the VSE’s man-date to encourage listings by small, developmental-stage companies. TheVSE, largely inhabited by venture capital stocks, has an unfavourable re-putation in regulating trading manipulation and disclosure of information.In recent years, the VSE has introduced many regulatory changes to pro-tect investors including the establishment of a new securities commissionwith increased power and staff resources, and upgrading of the minimumlisting requirements. In 1990, the VSE replaced its manual trading systemwith a fully-automated trading system and introduced a more powerful

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 151

surveillance system. In contrast, the disclosure and reporting requirementsare more stringent on the TSE which is much more tightly regulated inproviding investor protection. In summary, while there are some differ-ences in trading structures between the TSE and the VSE, major differ-ences lie in the type of securities listed and in listing and disclosurerequirements.

III. Sample and Data

The initial sample consists of those TSE or VSE listed Canadian Stocksthat also listed on the NYSE, AMEX or NASDAQ during January 1980–December 1991. The primary data base consists of 156 Canadian securitiesthat were identified from the lists provided by the NYSE, AMEX, andNASDAQ. For each stock, observations on the bid and ask on the lasttrading day of each month and the monthly trading volume for one yearbefore and one year after the cross-country listing were collected from theReuters Data Service. The price for each stock was proxied as the averageof its bid and ask prices. The price variability was proxied as the variance ofthe monthly rates of return in the pre- and post-listing periods respectively.The data on the number of outstanding shares were collected from the TSEand VSE Reviews. The average bid-ask spreads and the trading volume inthe pre-listing and post-listing periods were calculated by averaging overthe 12 observations in each of these periods respectively. The observationsin the listing month were not included in the analysis. Wherever possible,the data were cross-checked from other sources for accuracy.9 A firm wasincluded in the final sample if it had a minimum of four observations ineach of the pre-listing and post-listing periods.10 The firms that first listedon the NASDAQ and then listed on other U.S. exchanges were not in-cluded in the sample.

The TSE and VSE Reviews were also examined to check for namechanges and major firm-specific events surrounding cross-listing. Firmswith concurrent major events, such as merger and amalgamation, aroundthe U.S. listing were excluded from the sample.

As shown in Table 2, the final sample consists of 92 stocks. Fifty-fivefirms are listed on the TSE and 37 firms on the VSE prior to listing in theU.S.11 A majority of the stocks were listed in the U.S. after 1986, a trendsimilar to that observed for international listings. Twenty firms, all TSElisted, are listed on the NYSE and AMEX. The remaining 72 firms arelisted on the NASDAQ.

© Blackwell Publishers Ltd. 1997.

152 Usha R. Mittoo

IV. Empirical Results

A. Sample Characteristics

Table 3 provides the descriptive statistics for sample firms in the pre-listing period. The sample is partitioned into three subsamples based on the domestic and foreign exchanges of listing. The TSE stocks listedon the NYSE or AMEX are denoted by TSE(EX) and those listed on theNASDAQ are denoted by TSE(NQ). All VSE sample stocks are listed onthe NASDAQ and are denoted by VSE(NQ).

These subsamples show significant differences on several dimensions.TSE(EX) stocks have the highest average price and trading volume andthe lowest average spread and variance among the three subsamples. Theaverage price of TSE(EX) stocks is more that twice the average price ofthe TSE(NQ) stocks and more that four times that of the VSE(NQ) stocks.The TSE(EX) stocks are also the most heavily traded stocks with almostfour times the trading volume than that of the TSE(NQ) stocks and fivetimes that of the VSE(NQ) stocks. Similar differences are observed in thevariance of stocks; average variance of the TSE(EX) stocks is less thanone half the variance of the TSE(NQ) stocks and less than one sixth that of the VSE(NQ) stocks. Average relative bid-ask spread, measured asthe bid-ask spread divided by the stock price, for a typical TSE(EX) stock

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 153

Table 2. Frequency Distribution of Sample Firms by Year of Listing: CanadianStocks Cross-listed on the U.S. Exchanges (1980–1991)

Canadian Exchange U.S. Exchange

Year Number TSE VSE NYSE/AMEX NASDAQ

1980 1 1 0 1 01981 1 1 0 1 01982 3 2 1 0 31983 8 4 4 1 71984 6 2 4 1 51985 8 5 3 2 61986 7 6 1 3 41987 13 7 6 3 101988 13 8 5 2 111989 12 8 4 3 91990 12 6 6 2 101991 8 5 3 1 7

Total 92 55 37 20 72

is almost one half that of a TSE(NQ) stock and less than one third that of a VSE(NQ) stock. The three subsamples represent stocks with verydifferent characteristics. However, all sample firms have positive averagerates of return in the pre-listing period; this indicates that firms generallydecide to cross-list their stocks after a period of strong performance in themarket.12

B. Effect of Cross-Country Listing on Stock Liquidity

B.1. Average Trading Volume. To control for the influence of market trad-ing, most studies use market adjusted measures of trading volume. Themarket adjusted values are computed by dividing the monthly tradingvolume and traded value for each company by the market trading volumeand traded value respectively for that month. While this study uses bothunadjusted and market adjusted measures, it also uses stock turnover as © Blackwell Publishers Ltd. 1997.

154 Usha R. Mittoo

Table 3. Descriptive Statistics for the TSE and VSE Sample Firms in the Pre-Listing Period

PRICE Trading TradedSample RELSP (C$) RETURN VAR volume value (C$)

TSE(ALL) Mean 0.029 7.74 0.028 0.030 866,097 8,577,412(S.D.) (0.02) (6.90) (0.06) (0.04) (1,471,758) (15,541,960)

TSE(EX) Mean 0.019 12.03 0.027 0.018 1,579,828 17,391,700(S.D.) (0.02) (7.61) (0.05) (0.021) (2,150,415) (21,287,580)

TSE(NQ) Mean 0.035 5.30 0.028 0.037 458,250 3,540,679(S.D.) (0.02) (5.12) (0.07) (0.04) (620,632) (7,654,375)

VSE(NQ) Mean 0.067 2.82 0.069 0.13 358,129 1,001,177(S.D.) (0.07) (2.02) (0.14) (0.38) (297,672) (973,087)

Mann-Whitney Test StatisticsTSE(EX)-TSE(NQ) –3.45*** 4.27*** –0.19 –3.11*** 3.38*** 4.12***TSE(NQ)-VSE(NQ) –2.50** 2.79*** –0.93 –2.32** 0.37 0.96

TSE(ALL) is the sample of all TSE listed stocks. TSE(EX) is the subsample of the TSE stocks listedon the U.S. stock exchanges, TSE(NQ) is the subsample of the TSE stocks listed on the NASDAQ andthe VSE(NQ) is the sample of VSE stocks listed on the NASDAQ. All VSE sample stocks are listedon the NASDAQ.RELSP is the average relative bid-ask spread, RETURN is the average monthly rate of return, VARis the monthly variance of the rate of return, Trading Volume is the monthly average number of sharestraded and Trading Value is the monthly average dollar value of shares traded based on twelvemonthly observations in the pre-listing period. S.D. is the standard deviation of variables.Mann-Whitney test is a non-parametric test of the equality of medians in the two subsamples. Thesignificance of the test statistics at the 0.1, 0.05, and 0.01 levels is denoted by *, **, and ***respectively.

a proxy for the trading volume. Stock turnover, measured as the numberof shares traded as a percentage of the number of shares outstanding, con-trols for the influence of new stock issued by firms concurrent with orsubsequent to cross-listing. The new stock issue can be significant as con-firmed in our sample; the TSE as well as the VSE sample firms experiencean increase of 13% and 38% respectively in the number of shares outstand-ing. This increase is statistically significant at the 1% level in all sub-samples (Table 4), implying that the change in the number of sharesoutstanding needs to be controlled.

Table 4 provides comparisons of the trading volume and stock turnoverand Table 5 those of the traded value, in the pre- and post-listing periods.Average unadjusted total trading volume for the TSE stocks more thandoubles, from over 860,000 shares in the pre-listing period to more than1.8 million shares, one year after listing. Even after adjusting for themarket volume, the growth is 112%. Table V confirms similar results forthe traded value. For a typical TSE stock, unadjusted average traded valuesurges from less than C$9 million to over C$20 million; the market adjustedtraded value escalates by 152% from 0.18% to over 0.47%. The results aresimilar for the TSE(EX) and TSE(NQ) stocks. The TSE trading volumeand traded value, both unadjusted and market adjusted, also exhibit anincrease of about 20%. These results are similar to the findings of Foersterand Karolyi (1993) and confirm that the gain in liquidity is not a temporaryphenomenon. The gains are, however, much less pronounced when stockturnover is used as a proxy variable. While the total turnover for a typicalTSE stock increases from 4.3% to 7.8%, a growth of 81%, the effect onthe TSE turnover for a cross-listed stock is negligible, increasing from4.3% to 4.5%.

In sharp contrast to the TSE stocks, the VSE stocks show very littlegains in liquidity. Even the unadjusted trading volume for a typical VSEstock increases only by about 20% and is reduced to less than 5% after con-trolling the market volume. The small liquidity gain, however, turns intoa loss when stock turnover is used as a proxy variable. For a typical VSEstock, the total stock turnover declines from 7.5% in the pre-listing periodto 5.8% in the post-listing period, a decline of 22%. The effect is evenmore dramatic for the VSE trading volume which shows a sharp decline,irrespective of the measure of volume used. Even the unadjusted monthlytrading volume on the VSE declines from an average of more than 350,000shares in the pre-listing period to less than 245,000 shares after listing, adecline of about 32%. The decline is even more pronounced after control-ling for the changes in market volume and in the number of shares

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 155

Table 4. Trading Volume in the Pre- and Post-Listing Periods: Canadian Stocks Cross-Listed in the U.S. (1980–1991)

Unadjusted Market adjusted Stock turnover

(CDVOL) (TOTVOL) (CDVOL) (TOTVOL) SHARESOS (CDVOL) (TOTVOL)

Panel A: TSE(ALL)Pre-listing 866,097 866,097 0.00239 0.00239 22,728,960 0.043 0.043

(1,471,758) (1,471,758) (0.00344) (0.00344) (32,824,830) (0.05) (0.05)Post-listing 1,112,462 1,875,661 0.00292 0.00507 25,745,240 0.045 0.078

(2,042,000) (3,610,714) (0.00473) (0.0087) (36,193,840) (0.05) (0.09)Wilcoxon 2.14** 4.69*** 0.838 4.70*** 5.18*** 0.43 3.94***

Panel B: TSE(EX)Pre-listing 1,579,828 1,579,828 0.00454 0.00454 43,717,180 0.048 0.048

(2,150,415) (2,150,415) (0.0046) (0.0046) (47,617,640) (0.05) (0.05)Post-listing 2,058,768 3,649,678 0.00524 0.00983 47,669,000 0.049 0.095

(3,067,933) (5,492,427) (0.0068) (0.0126) (52,682,570) (0.06) (0.11)Wilcoxon 1.49 3.58*** 0.635 3.29*** 2.84*** 0.19 2.84***

Panel C: TSE(NQ)Pre-listing 458,250 458,250 0.00117 0.00117 10,735,690 0.041 0.041

(620,632) (620,632) (0.0016) (0.0016) (6,081,664) (0.05) (0.05)Post-listing 571,715 861,937 0.00160 0.0024 13,217,380 0.043 0.07

(731,461) (969,133) (0.0027) (0.0033) (9,235,276) (0.05) (0.07)Wilcoxon 1.46 3.04*** 0.60 3.28*** 4.41*** 0.67 2.70***

Table 4. Continued

Unadjusted Market adjusted Stock turnover

(CDVOL) (TOTVOL) (CDVOL) (TOTVOL) SHARESOS (CDVOL) (TOTVOL)

Panel D: VSE(NQ)Pre-listing 358,129 358,129 0.001437 0.001437 6,126,255 0.075 0.075

(297,672) (297,672) (0.001656) (0.00166) (3,272,881) (0.075) (0.075)Post-listing 242,335 436,782 0.000841 0.001502 8,448,896 0.034 0.058

(264,113) (413,840) (0.000875) (0.00133) (4,159,278) (0.034) (0.05)Wilcoxon –2.96** 1.53 –2.93*** 1.09 5.83*** –3.99*** –1.56

Unadjusted trading volume is the average number of monthly shares traded and market adjusted trading volume is the unadjusted trading volume divided by theCanadian market trading volume for that month. CDVOL is the trading volume on the Canadian stock exchanges, TOTVOL is the total trading volume on boththe Canadian and U.S. exchanges, SHARESOS is the number of shares outstanding. Stock Turnover is the number of shares traded as a percentage of the numberof shares outstanding. All variables are the averages of twelve monthly observations in the pre-listing and post-listing periods respectively. Standard deviation isin parentheses. TSE(ALL) is the sample of all TSE listed stocks. TSE(EX) is the subsample of the TSE stocks listed on the U.S. stock exchanges, TSE(NQ) isthe subsample of the TSE stocks listed on the NASDAQ and the VSE(NQ) is the sample of VSE stocks listed on the NASDAQ. All VSE sample stocks are listedon the NASDAQ. Wilcoxon test is a nonparametric matched pair test of the equality of medians in the pre-listing and post-listing periods. The significance at the0.1, 0.05, and 0.01 significance levels is denoted by *, **, and *** respectively.

© Blackwell Publishers Ltd. 1997.

158 Usha R. Mittoo

Table 5. Traded Value in the U.S. in the Pre- and Post-Listing PeriodsCanadian Stocks Cross-Listed in the U.S. (1980–1991)

Unadjusted traded Market adjustedvalue (C$) traded value (C$)

(CDVAL) (TOTVAL) (CDVAL) (TOTVAL)

Panel A: TSE(ALL)Pre-listing 8,577,412 8,577,412 0.00177 0.00177

(15,541,963) (15,541,963) (0.00288) (0.00288)Post-listing 12,175,980 20,893,454 0.00244 0.00446

(25,053,120) (44,442,571) (0.0044) (0.0089)Wilcoxon 1.77* 2.69*** 1.68* 4.11***

Panel B: TSE(EX)Pre-listing 17,391,700 17,391,700 0.00381 0.00381

(21,287,580) (21,287,580) (0.00385) (0.00385)Post-listing 26,105,030 45,020,420 0.00512 0.00957

(37,079,480) (66,551,760) (0.00619) (0.0131)Wilcoxon 1.67* 2.37** 1.53 3.211***

Panel C: TSE(NQ)Pre-listing 3,540,679 3,540,679 0.0006 0.0006

(7,654,375) (7,654,375) (0.00103) (0.00103)Post-listing 4,216,522 7,106,620 0.00091 0.000154

(7,014,684) (1,112,653) (0.00154) (0.0027)Wilcoxon 0.71 2.18** 0.754 2.54**

Panel D: VSE(NQ)Pre-listing 1,001,177 1,001,177 0.00326 0.00326

(973,087) (973,087) (0.00305) (0.00305)Post-listing 685,922 1,321,119 0.00232 0.00432

(1,191,499) (1,815,879) (0.00403) (0.0055)Wilcoxon –2.51** 0.656 –2.25** 5.83***

Unadjusted traded value is the average monthly dollar value traded. Market adjusted traded value isthe unadjusted traded value divided by the Canadian market traded value for that month. CDVAL isthe trading value on the Canadian stock exchanges, TOTVAL is the total trading value on both theCanadian and U.S. exchanges, All variables are the averages of twelve monthly observations in thepre-listing and post-listing periods respectively. Standard deviation is in parentheses. TSE(ALL) isthe sample of all TSE listed stocks. TSE(EX) is the subsample of the TSE stocks listed on the U.S. stockexchanges, TSE(NQ) is the subsample of the TSE stocks listed on the NASDAQ and the VSE(NQ) isthe sample of VSE stocks listed on the NASDAQ. All VSE sample stocks are listed on the NASDAQ.Wilcoxon test is a nonparametric matched pair test of the equality of medians in the pre-listing andpost-listing periods. The significance at the 0.1, 0.05, and 0.01 significance levels is denoted by *, **,and *** respectively.

outstanding. The stock turnover on the VSE reduces to less than 50% ofits pre-listing levels.

Overall, there are significant differences in liquidity gains between theTSE and VSE stocks, especially when stock turnover is used as a proxyfor the trading volume.

B.2. Cross-Sectional Variation in Liquidity Effect. Figure 1 displays the percentage change in the total turnover across sample stocks whichconfirms that average effect is influenced heavily by a small subset ofthese stocks. Of the 55 TSE stocks, only 20 experience an increase of more than 100%. The stock turnover actually declines for 14 stocks. The decline is more pervasive among the VSE stocks. Of the 35 VSEstocks, 23 experience a decline in turnover. There are a few large winnerseven among the VSE stocks, six stocks more than double their turnoverafter cross-listing. Similar patterns are observed for the stock turnover on the domestic exchanges, except that there are fewer outliers now(Figure 2). Only six TSE stocks gain more than 100% in stock turnover on the TSE. Again, the decline is more pervasive across the VSE stocks.Only five VSE stocks show an increase of more than 50% in the VSEturnover.

Figure 3 reveals a similar scenario for the trading volume on the U.S.exchanges. The TSE remains the dominant market in more than 75% ofthe cross-listed stocks, trading more than 50% of their trading volume.Only three TSE stocks trade more than 70% of their volume on the U.S.markets. In contrast, NASDAQ becomes a dominant market for over35% of the VSE stocks. The average U.S. trading volume for a typicalVSE(NQ) stock is about 44% compared to 31% for a typical TSE(NQ)stock; the difference is statistically significant at the 5% level. Theevidence suggests that the differences are linked to the Canadian and notto the U.S. location of listing.

Overall, the evidence shows a higher degree of cross-sectional variationamong the TSE stocks compared to a more pervasive effect among theVSE stocks.

C. Determinants of the Cross-Sectional Variation

C.1. Exchange Versus Firm-Specific Factors. Since the TSE and VSEdiffer substantially in the type of securities listed, it is plausible that the observed differences between the two samples reflect primarily in-vestors’ preferences for different securities. In particular, the tradingvolume may be influenced by the demand of U.S. investors who may

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 159

Figure 1. Effect of U.S. Listing on Stock Turnover: U.S. and Canadian Exchanges (TSE versus VSE Firms)

35%

30%

25%

20%

15%

10%

5%

0%

Per

cent

age

of F

irms

>–5

00%

–100

% to

–50

0%

–75%

to –

100%

–50%

to –

75%

–25%

to –

50%

0 to

–25

%

0 to

25%

25%

to 5

0%

50%

to 7

5%

75%

to 1

00%

100%

to 5

00%

>50

0%

Percentage Change in Stock Turnover(Post-Listing to Pre-Listening)

TSE

VSE

TSE

VSE

Figure 2. Effect of U.S. Listing on Stock Turnover: Canadian Exchanges (TSE versus VSE Firms)

45%

40%

35%

30%

25%

20%

15%

10%

5%

0%

Per

cent

age

of F

irms

TSE

VSE

>–5

00%

–100

% to

–50

0%

–75%

to –

100%

–50%

to –

75%

–25%

to –

50%

0 to

–25

%

0 to

25%

25%

to 5

0%

50%

to 7

5%

75%

to 1

00%

100%

to 5

00%

>50

0%

JIFM8 3

TSE

VSE

Percentage Change in Stock Turnover(Post-Listing to Pre-Listing)

Figure 3. Percentage of Total Volume Traded on U.S. Exchanges (TSE versus VSE Firms)

25%

20%

15%

10%

5%

0%

0-10

%

10-2

0%

20-3

0%

30-4

0%

40-5

0%

50-6

0%

60-7

0%

70-8

0%

80-9

0%

90-1

00%

Percentage of Total Volume on U.S. Exchanges

Per

cent

age

of F

irms

TSE

VSE

TSE

VSE

prefer Canadian stocks with some unique characteristics. This section ex-plores the contribution of stock characteristics in explaining the observeddifferences between the two samples.

Fama and French (1992) find that size and book-to-market valuecapture much of the cross-sectional variation in stock returns. The VSEfirms are much smaller than the TSE firms; the average market value ofequity for a VSE firm, in the year prior to listing, is C$18 million com-pared to C$75 million for a TSE sample firm. Approximately 15% of theVSE stocks are penny stocks (less than C$1.00 per share) and about 70%are priced less than C$3.00. In contrast, none of the TSE stocks is a pennystock and about 75% of these stocks are priced more than C$3.00. Aver-age book-to-market value, available for a subset of 68 firms (47 TSE and21 VSE), is 4.83 for the TSE firms compared to 5.62 for the VSE firms.The Canadian market has a preponderance of resource stocks which maybe another factor in investors’ decisions.13 While the proportion of resourcestocks is similar in the two samples, 65% and 58% in the VSE and TSEsamples respectively, the junior VSE resource firms are likely to be riskierthan most of the TSE sample firms. Investor recognition, hypothesized as a determinant of investors’ demand in a model by Merton (1987), alsodiffers across the two samples.14 Investors may also prefer the TSE stocksbecause of their higher liquidity.

To examine the effect of these characteristics, the following regressionis estimated:

%DIFFVOLi = α + β1CDEXi + β2USEXi + β3SIZEi + β4RESOURCEi + β5TURNOVERi + β6PRICE i/BVi + εi (1)

where %DIFFVOLi is the percentage of change in the stock turnover forsecurity i after cross-listing, CDEXi is a dummy variable equal to one forthe VSE stocks and zero for the TSE stocks, USEXi is a dummy variableequal to one for the NYSE and AMEX listed and zero for the NASDAQlisted stocks. SIZEi, TURNOVERi and PRICEi /BVi are measured by theaverage of the natural logarithm of the market value of equity, stock turn-over, and price-to-book value of equity for security i, in one year prior tothe cross-listing, respectively. RESOURCE is a dummy variable with onefor the resource and zero for the non-resource stocks.

Regression estimates of equation (1) are shown in Table 6. The first regres-sion is run with only two exchange dummy variables, CDEX and USEX.Regressions (2)–(5) are run with additional firm-specific variables. Thevalue of R-square increases from 8% in regression (1) to 27% in regression

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 163

(5), implying that most of the variation in trading volume can be attributedto the stock-specific characteristics.

Both firm size (SIZE) and pre-listing stock turnover (TURNOVER) havesignificant coefficients at the 1% level in all regressions. Gains in tradingvolume are larger for smaller firms, indicating higher demand for suchstocks among investors. Listing in the U.S. enhances the visibility of afirm and serves as a positive signal about its future prospects. The valueof this signal is likely to be higher for smaller firms relative to the largerfirms, which may generate higher demand for smaller stocks after cross-listing. The negative coefficient of TURNOVER is somewhat puzzlingsince turnover proxies for stock liquidity which is preferred by investors.One possibility is that high turnover in the pre-listing period may alsoreflect a high degree of differences in the asymmetric information about astock.15 Results are essentially the same when trading volume is used asan alternative measure of liquidity. PRICE/BV is positively related to thechange in stock turnover, indicating higher demand for firms with largerprice-to-book ratio.© Blackwell Publishers Ltd. 1997.

164 Usha R. Mittoo

Table 6. Cross-Sectional Regressions of the Percentage Change in StockTurnover from Pre- to Post-Listing Periods on Firm and Exchange SpecificFactors (Canadian and U.S. Stock Exchanges)

TURN- PRICE/CDEX USEX SIZE OVER RESOURCE BV R2 N

1. Coeff –1.18*** –0.22 0.083 92T-stat (–2.72) (–0.43)

2. Coeff –1.72*** 0.74 –0.488*** 0.17 92T-stat (–3.79) (1.26) (–3.04)

3. Coeff –1.49*** 0.90 –0.54*** –8.23*** 0.24 92T-stat (–3.34) (1.57) (–3.43) (–2.74)

4. Coeff –1.45*** 1.05* –0.56*** –7.99*** 0.55 0.25 92T-stat (–3.28) (1.82) (–3.60) (–2.67) (–1.47)

5. Coeff –1.61*** 1.03 –0.56*** –10.06** –0.247 0.0375* 0.27 68T-stat (–2.83) (1.54) (–2.82) (2.04) (0.5) (1.96)

Stock Turnover is the number of shares traded as a percentage of the number outstanding. CDEX is adummy variable for the Canadian exchange of listing with 1 for the VSE and 0 for the TSE. USEX isa dummy variable for the U.S. exchange of listing with 1 for the NYSE and AMEX and 0 for theNASDAQ. SIZE is the natural logarithm of the market value of the equity and TURNOVER is thestock turnover in the pre-listing period, respectively. RESOURCE is a dummy variable with 1 for the resource firms and 0 for the non-resource firms. PRICE/BV is the price-to-book ratio of equity inthe pre-listing period. N is the number of sample firms used in the regression analysis. Coeff is theestimated coefficient and T-stat is the t-statistics. Significance of the test statistics at the 0.1, 0.05 and0.01 levels is denoted by *, **, and *** respectively.

While the liquidity gains are not sensitive to the U.S. location of listing(USEX), the effect of Canadian location of listing (CDEX) is significantat the 1% level in all regressions. The negative coefficient on CDEXimplies that the liquidity gains for the VSE stocks are much lower than fortheir TSE counterparts. Surprisingly, the coefficient of CDEX becomeseven more negative as firm-specific variables are added in regressions;thus, the differences between the TSE and VSE stocks become even morepronounced after controlling for the firm-specific variables.

The results remain essentially the same when accounting variables,debt to equity and Tobin’s Q ratio (market-to-book value of assets), areadded to the regression.16 A dummy variable measuring investor recognitionbased on the press reports in one year prior to listing also yields insignific-ant coefficients. Similar conclusions are suggested when the percentagechange in stock turnover on the Canadian exchanges is used as a depend-ent variable. As seen in Table 7, the Canadian exchange of listing remainsthe dominant factor in all regressions while SIZE and TURNOVER havesignificant coefficients in some regressions.

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 165

Table 7. Cross-Sectional Regressions of the Percentage Change in StockTurnover from Pre- to Post-Listing Periods on Firm and Exchange SpecificFactors (Canadian Stock Exchanges)

TURN- PRICE/CDEX USEX SIZE OVER RESOURCE BV R2 N

1. Coeff –0.579*** –0.237 0.09 92T-stat (–2.97) (–1.02)

2. Coeff –0.76*** 0.0951 –0.169** 0.143 92T-stat (–3.70) (0.36) (–2.304)

3. Coeff –0.648*** 0.17 –0.193*** –4.13*** 0.22 92T-stat (–3.22) (0.68) (–2.73) (–3.04)

4. Coeff –0.65*** 0.15 –0.19*** –4.16*** 0.070 0.23 92T-stat (–3.22) (0.59) (–2.65) (–3.04) (0.41)

5. Coeff –0.65*** 0.035 –0.135*** –4.02* 0.26 0.0051 0.18 68T-stat (–2.69) (0.122) (–1.60) (–1.92) (1.25) (0.63)

Stock Turnover is the number of shares traded as a percentage of the number outstanding. CDEX is adummy variable for the Canadian exchange of listing with 1 for the VSE and 0 for the TSE. USEX is a dummy variable for the U.S. exchange of listing with 1 for the NYSE and AMEX and 0 for theNASDAQ. SIZE is the natural logarithm of the market value of the equity and TURNOVER is thestock turnover in the pre-listing period, respectively. RESOURCE is a dummy variable with 1 for the resource firms and 0 for the non-resource firms. PRICE/BV is the price-to-book ratio of equity inthe pre-listing period. N is the number of sample firms used in the regression analysis. Coeff is theestimated coefficient and T-stat is the t-statistics. Significance of the test statistics at the 0.1, 0.05 and0.01 levels is denoted by *, **, and *** respectively.

To examine the effect of firm-specific factors on the trading location, a regression analogous to (1) is run with the trading volume on the U.S.exchanges as a percentage of total volume as a dependent variable. Table 8shows that the influence of firm-specific variables is much weaker in this regression. The value of R-square increases from about 7% to 16% asfirm-specific variables are added. TURNOVER and RESOURCE havenegative coefficients but they are significant only in some regressions.Canadian exchange-specific factors dominate the explained part of thevariation; CDEX is the only significant factor in all regressions.

In summary, while the stock-specific characteristics explain some cross-sectional variation, Canadian exchange-specific factors emerge as signific-ant determinants of both total trading volume and the U.S. trading volumein a cross-listed security.

C.2. Exchange-Specific Factors. Exchange-specific factors play a majorrole in a theoretical model of multimarket trading developed by Chowdaryand Nanda (1991) in which large liquidity traders can split their tradesacross markets, but small traders are constrained to trade on a singlemarket. When a security is cross-listed, the informed traders trade more© Blackwell Publishers Ltd. 1997.

166 Usha R. Mittoo

Table 8. Cross-Sectional Regressions of the Percentage of Total Trading Volumeon the U.S. Exchanges on Firm and Exchange Specific Factors

TURN- PRICE/CDEX USEX SIZE OVER RESOURCE BV R2 N

1. Coeff 0.137** 0.054 0.068 92T-stat (2.54) (0.83)

2. Coeff 0.148** 0.034 0.0100 0.071 92T-stat (2.52) (0.44) (0.48)

3. Coeff 0.172*** 0.0497 0.005 –0.83** 0.12 92T-stat (2.91) (0.66) (0.26) (–2.08)

4. Coeff 0.178*** 0.078 0.004 –0.79** –0.099** 0.16 92T-stat (3.07) (1.03) (0.02) (–2.01) (–2.06)

5. Coeff 0.193*** 0.101 0.0088 –0.649 –0.111* 0.014 0.162 68T-stat (2.74) (1.22) (0.36) (–1.06) (–1.83) (0.58)

Total trading volume is the monthly average number of shares traded on Canadian and U.S. exchangesbased on the 12 monthly observations in the post-listing period. CDEX is a dummy variable for theCanadian exchange of listing with 1 for the VSE and 0 for the TSE. USEX is a dummy variable forthe U.S. exchange of listing with 1 for the NYSE and AMEX and 0 for the NASDAQ. SIZE is thenatural logarithm of the market value of the equity and TURNOVER is the stock turnover in the pre-listing period, respectively. RESOURCE is a dummy variable with 1 for the resource firms and 0 forthe non-resource firms. PRICE/BV is the price-to-book ratio of equity in the pre-listing period. N isthe number of sample firms used in the regression analysis. Coeff is the estimated coefficient and T-stat is the t-statistics. Significance of the test statistics at the 0.1, 0.05 and 0.01 levels is denoted by*, **, and *** respectively.

aggressively since they can camouflage their trades more easily on severalmarkets which increases the total trading volume in that security.17 Theirmodel predicts that one of the markets emerges as a dominant market for trading in that security. The concentration of trading occurs in a marketlocation in which market makers “crack down” on insider trading. Thismarket also provides the lowest cost of trading by attracting the largestproportion of liquidity traders and informed traders. Thus, both tradingcosts and regulatory environment play a role in determining the locationof trading in a cross-listed security. It is plausible that the cost of tradingsimilar securities is higher on the VSE relative to the TSE, which may drivemost of the trading in the VSE stocks to the U.S. markets. Alternatively,the differences in the regulatory environment of the two exchanges mayplay a key role in explaining the higher U.S. volumes in the VSE stocks.

To distinguish between the two explanations, I first examine the tradingcost hypothesis. Trading costs are proxied by the bid-ask spread of a se-curity which represents a portion of the transaction costs. Standard marketmicrostructure models specify price, volume, and variance as the primarydeterminants of the bid-ask spread which is hypothesized to be positivelyrelated to the price and trading volume and negatively to the variance of astock.18 The relative spread, bid-ask spread as a percentage of price, for theTSE and VSE stocks is compared by estimating the following regressionsin the pre- and post-listing periods:

Pre-Listing : CDSPRi = α + β1CDEXi + β2CDPRICEi + β3CDVOLi + β4CDVARi εi (2)

Post-Listing : %∆CDSPRi = α + β1CDEXi + β2USEXi + β3%∆CDPRICEi + β4%∆CDVOLi + β5%∆CDVARi εi (3)

where CDSPRi is the relative bid-ask spread of security i, CDVOLi is thenatural logarithm of the average monthly trading volume, CDVARi is themonthly variance and CDPRICEi is the inverse of the price of security i,in the pre-listing period.19 In the post-listing equation (3), each of thesevariables is replaced with the percentage change in its post-listing valuerelative to its pre-listing value. CDEXi and USEXi are the exchange dummyvariables for the Canadian and U.S. exchanges respectively as defined inequation (1).

The results of these regressions are presented in Table 9. About 69% of the cross-sectional variation in the relative spreads is explained by a stock’s price, variance, and trading volume. The relative spread is

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 167

positively associated with price and variance, and negatively with thetrading volume, as predicted. The coefficient of CDEX is positive but isnot significant at any reasonable significance level. The results suggestthat the transaction costs as measured by the relative bid-ask spreads arenot significantly different between the two exchanges after accounting forthe stock-specific determinants.© Blackwell Publishers Ltd. 1997.

168 Usha R. Mittoo

Table 9. Cross-sectional Regressions of the Relative Bid-Ask Spread and thePercentage Change in the Relative Bid-Ask Spread on the Firm and ExchangeSpecific Variables

Pre-listing period Post-listing period

TSE(NQ) TSE(EX) Dependent All Dependent All and andvariable: sample variable sample VSE(NQ) TSE(NQ)CDSPR firms %CDSPR firms sample sample

Coefficient Coefficient Coefficient Coefficient

INTERCEPT 0.26*** INTERCEPT 0.035 0.032 0.09(7.02) (0.37) (0.30) (0.86)

CDEX 0.006 CDEX 0.17 0.17(0.82) (0.136) (1.22)

USEX USEX –0.09 –0.11(–0.55) (–0.85)

CDPRICE 0.037*** %∆CDPRICE 0.431*** 0.43*** 0.40***(3.83) (7.61) (6.67) (4.85)

CDVAR 0.11*** %∆CDVAR 0.077** 0.08** 0.06(8.52) (2.28) (2.17) (1.14)

CDVOL –0.02*** %∆CDVOL –0.16*** –0.16** –0.21**(–6.77) (–3.53) (–3.11) (–2.53)

R-square 0.69 0.57 0.56 0.51

Relative bid-ask spread is the bid-ask spread as a percentage of price. CDSPR is relative bid-ask spreadin the pre-listing period and %CDSPR is the percentage change in the relative bid-ask spread in thepost-listing period relative to the pre-listing period. CDEX is a dummy variable for the Canadian ex-change of listing with 0 for the TSE listed and 1 for the VSE listed firms. USEX is a dummy variablefor the U.S. exchange of listing with 1 for the NYSE and AMEX listed firms, and 0 for the NASDAQlisted firms. CDPRICE is the inverse of average price, CDVAR is the average variance and CDVOLis the natural logarithm of the number of the shares traded in the pre-listing period. %∆CDPRICE isthe percentage change in the inverse of the price, %∆CDVAR is the percentage change in the variance,and %∆CDVOL is the percentage change in the trading volume on the Canadian stock exchanges inthe post-listing period relative to the pre-listing period. TSE(ALL) is the sample of all TSE listedstocks. TSE(EX) is the subsample of the TSE stocks listed on the U.S. stock exchanges, TSE(NQ) isthe subsample of the TSE stocks listed on the NASDAQ and the VSE(NQ) is the sample of VSEstocks listed on the NASDAQ. All VSE sample stocks are listed on the NASDAQ. T-statistics are inparentheses. Significance of the test statistics at the 0.1, 0.05 and 0.01 levels is denoted by *, **, and*** respectively.

Similar results are revealed in the post-listing regression. About 57% ofthe variation in the relative spreads is explained by the changes in the threeindependent variables. None of the exchange dummy variables have sig-nificant coefficients. Results for the NYSE and AMEX listed or NASDAQlisted subsamples are also similar to those for the overall sample. Thus,the trading cost differences do not appear to be large enough to explain thehigher U.S. volumes for the VSE stocks.

The second explanation relates to the differences in the regulatoryenvironment of exchanges. There is a considerable disparity between thetwo exchanges in two aspects of the market regulation. First, the listingand disclosure requirements are more stringent on the TSE relative to theVSE; this implies that disparity in quality and reliability of accounting andmarket information could be a plausible explanation for the observeddifferences. However, until July 1991, all Canadian firms listed on the U.S.exchanges were required to meet the U.S. Security and Exchange Com-mission’s tougher requirements for disclosure and financial reporting, in-cluding the extensive disclosures of proxy requests, executive compensation,relationship between management and directors, and broader liabilityprovisions, not required under the Canadian regulation.20 Thus, this aspectof regulatory environment became more equal for both the TSE and VSEstocks after cross-listing. Since our sample consists of firms that cross-listed prior to 1991, information and disclosure differences cannot be amajor factor in explaining these results.

Second, disparity in the market regulation concerning investor protec-tion between the two exchanges could be another plausible explanation.One can argue that the degree of investor protection accounts for theinsignificant liquidity gains for the VSE stocks. When a stock is traded ondual markets, investors have two avenues in which they can exploit theirinformation and trade large numbers of shares which is a valuable option,particularly for large institutional investors. However, when one of the twomarkets is not tightly regulated, investors face the risk of trading againstwell informed or insider traders on that market. In this case, the benefitsof multimarket trading are reduced considerably, which may lead to a lowerdemand and consequently, to an insignificant effect on trading volume forstocks listed from exchanges with lower degree of investor protection. Also,the higher U.S. trading volumes for the VSE cross-listed stocks relative to their TSE counterparts may reflect investors’ preference to trade on amore tightly regulated market, all else being equal.

Previous research has shown that foreign investors tend to concen-trate their investments in securities in which there is less information

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 169

asymmetry between foreign and domestic investors.21 A similar argumentcan be made for the concentration of foreign investments in well-regulateddomestic markets in which they have more confidence. A testable implica-tion of this argument, which is beyond the scope of this paper, is that weshould observe positive correlation between the degree of foreign invest-ment and investor protection on a domestic exchange.

It is possible that some important firm-specific effects have not beenaccounted for in the analysis. However, since the differences between the two samples become even more pronounced after controlling for thestock-specific variables, it appears unlikely that exchange-specific factorswill lose significance with additional variables. Additional factors, such asdifferences in market and trading structures, could also potentially explainthese results. While the evidence presented here reveals large exchange-specific effects of cross-listing, the factors driving these effects remain tobe identified.

VI. Summary and Future Research

This study investigates the effects of listing on the U.S. exchanges onstock liquidity for stocks listed on the two Canadian stock exchanges—theToronto and Vancouver Stock Exchanges—that represent two extremes inthe type of securities listed and other characteristics.

The results show substantial differences between the two samples. Thetrading volume in the TSE stocks more than doubles, and the stock turn-over increases by over 80% after cross-listing. The VSE stocks, on the otherhand, experience an insignificant increase in trading volume and a sharpdecline in stock turnover. The TSE is also able to maintain its pre-listinglevels of trading volume in cross-listed securities, whereas the VSE losesabout half the trading volume in these stocks to the U.S. exchanges. Thecross-sectional analysis reveals that the differences between the twosamples become even more pronounced after controlling for firm-specificfactors.

The analysis suggests that exchange-specific factors are significantdeterminants of the trading volume and trading location in cross-listedsecurities. The trading cost differences between the two exchanges do notappear to be significant enough to explain these results. Disparity in list-ing and disclosure requirements is also not consistent with the observedresults since this disparity is reduced after cross-listing, because both the TSE and VSE stocks became subject to U.S. Security and ExchangeCommission’s disclosure and reporting requirements, prior to 1991. It can© Blackwell Publishers Ltd. 1997.

170 Usha R. Mittoo

be argued that market regulation concerning investor protection may belinked to the insignificant liquidity effects observed for the VSE stocks.Since most investors like to trade on a well-regulated market, the benefitsof dual market trading are reduced when one of the two markets is nottightly regulated because investors face a higher risk of trading againstinsider traders on that market. This could result in a lower demand andhence, insignificant liquidity effect for the VSE stocks relative to theirTSE counterparts. Alternatively, differences in market and trading structurescould also potentially explain the observed differences. The analysis ofthese factors is left for future research.

Absence of adequate investor protection on many emerging stockmarkets has been cited as a major impediment in foreign portfolio invest-ments. Previous studies have documented that comprehensive institutionalreforms that curtail share price manipulation and insider trading are linkedto the increased demand for emerging market equities since well function-ing institutions reduce the risk of loss associated with trading securities.22

It can be argued that such reforms may be a significant determinant of the trading flow and trading volumes in cross-listed securities as well.Approximately half the new listings on the NYSE in recent years belongto the new emerging market countries with small and immature stockmarkets.23 These markets are likely to benefit significantly by adoptingeffective market surveillance system, stock watch, and other measures forinvestor protection that increase investor confidence in trading on theirmarkets, because the evidence presented here suggests that order flow likesa well-regulated environment.

The ability of the TSE to maintain its market share in Canadian cross-listed securities may partly be attributed to its aggressive response to thecompetitive pressures of the U.S. exchanges.24 It appears that factors suchas cost, transparency and market structure play a crucial role in the com-petition for order flows. Market regulators and policy markets around theworld are grappling with the issue of finding a right mix of regulation thatprovides adequate protection to investors as well as enhances the com-petitiveness of their markets. Future research should explore the role ofthese additional factors that are likely to become important with the in-creasing competition among the stock exchanges in the global marketplace.

Notes

1. To increase stock liquidity and to access larger capital markets are cited as the twomajor reasons for listing abroad in a survey of Canadian firms (Mittoo, 1992). Low tradingvolume on the Tokyo Stock Exchange prompted many foreign firms to delist their stocks

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 171

from that exchange. Between 1991 and 1995, the Tokyo Stock Exchange lost 48 foreignlistings including may blue chip U.S. stocks such as Ford Motor, General Electric and WaltDisney, while the NYSE added 179 non-U.S. companies during the same period (Mertens,1996)

2. See Cochrane, Shapiro and Tobin (1995) and New York Stock Exchange, TheExchange, January 1006, Vol. 3, No. 1, p. 6.

3. See Karolyi (1996) for a survey of literature on foreign listing.4. A survey of 300 VSE investors found that almost two-thirds felt that the VSE

operates for the benefit of its members rather than investors and over 50% believe that the only way to make money is to have insider information. The current investors wereleast critical and only a small number had a complaint about any aspect of the VSE trading.(The Globe and Mail, October 15, 1993, p. B3).

5. See, Halliday (1989)6. A destabilizing trade is a purchase which is made at a price above or a sale which

is made at a price below the preceding different priced trades.7. The TSE handles orders in several ways according to the nature of the order and

the stock. Stocks are listed either as floor stocks or as Computer-Assisted Trading System(CATS) stocks which has been in use for the less active stocks since 1979. Orders for floor-traded stocks are routed to the TSE’s Electronic Order Book (The 1993 Handbook of WorldStock and Commodity Exchanges, 1993, Blackwell Publishers).

8. In 1996, all Canadian exchanges moved to decimal trading to attract more tradingin the U.S. listed Canadian stocks.

9. Standard and Poor’s Daily Stock Price Records were examined to cross-check thedata on the bid, ask and trading volume in cross-listed securities on the NASDAQ.

10. Only three firms had less than six observations in the pre-listing and only two firmshad less than six observations in the post-listing period.

11. A sample stock is defined as VSE listed stock if it was listed on the VSE for thetwelve months prior to the U.S. listing and was not listed on the TSE. The sample stockslisted on the TSE and on other Canadian stock exchanges (including the VSE) in the pre-listing period are defined as the TSE sample stocks. Only five VSE sample stocks werealso listed on the TSE and most of these were listed on the TSE prior to listing on the U.S.

12. See Alexander, Eun and Janakiramanan (1988) and Foerster and Karolyi (1993).13. Roll (1992) shows that industry is a primary determinant of the correlations among

national stock market index returns.14. About 60% of the TSE sample firms had at least some reference in a Canadian busi-

ness newspaper during one year prior to cross-listing. In contrast, only 20% of the VSEsample firms received such coverage in the press.

15. Kim and Verrechia (1991) discuss the relationship between private information andthe trading volume.

16. The results are not reported for brevity.17. Kyle (1985) and Admati and Pfleiderer (1988) have developed models of market

equilibrium in the presence of informed and liquidity traders. Freedman (1992) studies thecase when the foreign and domestic markets operate non-simultaneously and the informa-tion of the traders is long-lived.

18. See Schwartz (1988), Demsetz (1968), Tinic and West (1974), and Hamilton (1979).19. The inverse of the price is used as an independent variable to reflect that the effect

of price on spread is increasing but at a decreasing rate.20. On July 1991, Canada and U.S. implemented the MultiJurisdictional Disclosure

System (MJDS) under which the Canadian companies will be deemed to have met the U.S.requirements by simply filing Canadian documents with the SEC. Accounting disclosureand legal costs are considered as the major costs of listing abroad by majority of firms inter-listed on the U.S. exchanges (see Biddle and Saudagaran (1989) and Mittoo (1992)).

© Blackwell Publishers Ltd. 1997.

172 Usha R. Mittoo

21. For example, Kang and Stulz (1994) find that foreign investment in Japanese firmsis concentrated in larger firms which provide more information relative to the smaller firms.

22. Hartman and Khambata (1993) argue that comprehensive institutional reforms con-cerning quality, of market and accounting information, the operational efficiency of the ex-change, and the legal regulatory environment that protects investors, are substantially capableof increasing the demand for equities, since well functioning institutions reduce the risk ofloss associated with emerging stock market equity investments, primarily those caused by problems of an institutional nature. They argue that the market reforms introduced byMexican Securities commission in 1980s increased the flow of information, helped reducethe share price manipulation and insider trading, and lowered transactions costs for investors.These and structural market reforms helped Mexico’s stock market grow from $4 billionin 1985 to $33 billion in 1991.

23. See Ogden and Wipperfürth (1996).24. For example, in 1991, when the NYSE announced extended hours of trading, the

TSE followed suit to protect its claim as the premier exchange for Canadian-based equities.

References

Alexander, G., C. Eun and S. Janakiramanan, “International Listings and Stock Returns:Some Empirical Evidence,” Journal of Financial and Quantitative Analysis (1988) 23,pp. 135–149.

Admati, A. R. and P. Pfleiderer, “A Theory of Intraday Patterns: Volume and PriceVariability,” Review of Financial Studies (1988) 1, pp. 3–40.

Biddle, G. C. and S. Saudagaran, “The Effects of Financial Disclosure Levels on Firms’Choices among Alternative Foreign Stock Exchange Listings,” Journal of InternationalFinancial Management and Accounting (1989) 1(1), pp. 55–87.

Chowdhary, B. and V. Nanada, “Multimarket Trading and Market Liquidity,” The Review of Financial Studies (1991) 4(3), pp. 483–511.

Cochrane, J., J. Shapiro and J. Tobin, Foreign Equities and U.S. Investors: Breaking DownBarriers Separating Supply and Demand (NYSE Working Paper, #95-04, 1995).

Demsetz, H., “The Cost of Transacting,” The Quarterly Journal of Economics (1968) 82,pp. 33–53.

Domowitz, I., J. Glen and A. Madhavan, International Cross-Listing, Ownership Rightsand Order Flow Migration: Evidence from Mexico (University of Southern CaliforniaWorking Paper, 1995).

Fama, E. F. and K. R. French, “The Cross-section of Expected Stock Returns,” Journal ofFinance (1992) 46, pp. 427–466.

Foerster, S. and G. A. Karolyi, “International Listings of Stocks: The Case of Canada andthe U.S.,” Journal of International Business Studies (1993) 24, pp. 763–784.

Foerster, S. and G. A. Karolyi, Multimarket Trading and Liquidity: A Transactions DataAnalysis of Canada-U.S. Interlistings (Ohio State University Working Paper, 1996).

Freedman, R., A Theory of the Impact of International Cross-Listing (Working Paper, The University of British Columbia, Vancouver, 1992).

Halliday, L., “The International Stock Exchange Directory,” Institutional Investor (1989)pp. 197–204.

Hamilton, J., “Marketplace Fragmentation, Competition, and the Efficiency of the StockExchange,” Journal of Finance (1979) 34, pp. 171–187.

Hargis, Kent, ADRs in Emerging Markets: Market Competition or Fragmentation(Working Paper, University of Southern California, 1996).

Hartman, M. and D. Khambata, “Emerging Stock Markets: Investment Strategies of theFuture,” Columbia Journal of World Business (1993) 28, pp. 82–104.

© Blackwell Publishers Ltd. 1997.

Cross-Country Listing and Trading Volume 173

Kang, J. and R. Stulz, Why is there a Home Bias? An Analysis of Foreign OwnershipPortfolio Equity Ownership in Japan (NBER Working Paper series #5166, 1995).

Karolyi, A., What Happens to Stocks that List Shares Abroad? A Survey of the Evidence andits Managerial Implications (Working Paper, Richard Ivy School of Business, Universityof Western Ontario, 1996).

Kim, O. and R. Verrechia, “Trading Volume and Price Reactions to Public Announcements,”Journal of Accounting Research (1991) 29, pp. 302–321.

Kyle, A., “Continuous Auction and Insider Trading,” Econometrica (1985) 53, pp. 1315–1335.

Merton, R. C., “A Simple Model of Capital Market Equilibrium with IncompleteInformation,” Journal of Finance (1987) 42, 483–510.

Mertons, B., “Exchange Eyes Asian Listings,” Asian Business (September 1996) pp. 36–39.

Mittoo, U. R. “Managerial Perceptions of the Net Benefits of Foreign Listing: CanadianEvidence,” Journal of International Financial Management and Accounting (1992)4(1), pp. 40–62.

Ogden, J. and H. Wupperfürth, “Should all those Foreign Companies be Listing on theNYSE,” Global Finance (1996) pp. 54–58.

Roll, R., “Industrial Structure and the Comparative Behavior of International Stock MarketIndices,” Journal of Finance (1992) 47, pp. 3–41.

Schwartz, Robert A., Equity Markets, (New York, Harper and Row, 1988).Tinic, S. M. and R. West, “Marketability of Common Stocks in Canada and the U.S.A.:

A Comparison of Agent versus Dealer Dominated Markets,” Journal of Finance (1974)32, pp. 729–746.

© Blackwell Publishers Ltd. 1997.

174 Usha R. Mittoo