Speed of convergence to market efficiency for NYSE-listed ...

[Pages:32]Speed of convergence to market efficiency for NYSE-listed foreign stocks

Nuttawat Visaltanachoti a, Ting Yang b,* a Department of Commerce, Massey University, Private Bag 102904, Auckland, New Zealand b Department of Finance, Auckland University of Technology, Private Bag 92006, Auckland 1142, New Zealand

Abstract

This paper contributes to the cross-listing literature by documenting the speed of convergence to market efficiency for foreign stocks listed on the NYSE. We find that on average it takes between 30 and 60 minutes for an ADR to achieve market efficiency. For a comparable U.S. stock listed on the same exchange, it takes only 10 to 15 minutes. The significant difference between foreign and U.S. stocks remains robust when the speed is measured by the number of transactions instead of in calendar time. Among various trading, firm, and country characteristics, factors associated with information asymmetry and investor participation significantly affect the speed to market efficiency for foreign stocks.

JEL classification: G14; G15 Keywords: Cross-listing; Speed of convergence; Market efficiency; Intraday evidence

*Corresponding author. Tel. +64-9-9219999 ext. 5397; fax: +64-9-9219940. E-mail addresses: n.visaltanachoti@massey.ac.nz (N.Visaltanachoti); ting.yang@aut.ac.nz (T. Yang)

1. Introduction In addition to listing on a domestic exchange, a firm sometimes chooses to cross-

list its shares on a foreign stock exchange. The NYSE is one of the most important listing destinations for foreign firms. At the end of 2005, the number of foreign stocks listed on the NYSE reached 453, a 472% increase from 96 in 1990. During the same period, the number of domestic listings on the NYSE only increased by 34%. Foreign firms account for about 17% of all NYSE-listed companies and their market capitalization represents approximately 37% of the total market capitalization of all NYSE companies at the end of 2005. 1

The growth motivates a vast literature on cross-listings (see Karolyi 2006 for a survey). Among all the studies, to the best of our knowledge, only two papers study the efficiency of the ADR market. 2 Rosenthal (1983) conducts serial correlation and runs tests on weekly, biweekly, and monthly returns for 54 ADRs over the period of 1974 through 1978. The results are consistent with weak form efficiency. Webster (1998) studies the market efficiency of three ADRs using Dickey-Fuller unit-root test and daily stock prices. The results show that the market for these ADRs is efficient over the daily horizon. Given the finding that ADR market is efficient over the daily horizon, a natural question to ask is how fast the ADR market becomes efficient within a day. The answer to this important question requires intraday analysis using high-frequency data. Rosenthal (1983) and Webster (1998) use daily or lower frequency data and therefore are silent on this issue. In this study, we try to

1 Data are collected from the NYSE website. 2 ADR refers to American Depository Receipts. Most foreign firms list their stocks in the U.S. as ADRs. For the basics of ADRs, please refer to , a website maintained by JP Morgan.

1

contribute to the cross-listing literature by providing evidence on the speed of convergence to market efficiency for ADRs. We use intraday data on a sample of 320 ADRs listed on the NYSE and find that, on average, it takes more than 30, but less than 60, minutes for ADRs to reach efficiency.

Our ADR sample also gives us a unique opportunity to explore several other important issues. On one hand, ADRs share the same trading venue as U.S. domestic stocks listed on the NYSE. The same market mechanism allows a sensible comparison of the speed to market efficiency for ADRs versus domestic stocks. We find that it takes between 10 to 15 minutes for comparable domestic stocks to reach efficiency, which is significantly faster than ADRs. On the other hand, ADRs are different from U.S. domestic firms in that they are from foreign countries with possibly very different legal, judicial, political, accounting, or corporate governance institutions. When we explore the determinants of the speed of convergence to market efficiency for ADRs, these differences enable us to examine whether such institution variables, in addition to trading and firm characteristics, are correlated with the speed to market efficiency.

Chordia, Roll, and Subrahmanyam (2005) is the first study on the speed of convergence to market efficiency. They study a sample of 150 U.S. domestic stocks and focus on documenting the speed to market efficiency. Our study complements Chordia et al. (2005) by investigating ADRs of foreign firms and using more recent data. We also try to extend Chordia et al. (2005) by exploring the factors affecting the speed to market efficiency. Among the trading and firm characteristics, we find

2

that stock price, volatility, and trading volume are significantly negatively related to the time needed to reach efficiency. In addition, country-level institutions make a difference. ADRs of firms from countries of common-law legal origin, or with better judicial efficiency, political stability, accounting standards, and anti-director rights are faster to reach market efficiency. Moreover, these institution variables seem to be able to explain some of the difference in the speed of convergence to market efficiency between the ADRs and U.S. domestic stocks.

The remainder of the paper is organized as follows: We describe the sample and data in Section 2. Tests and results on the speed of convergence to market efficiency are presented in Section 3. We explore the factors affecting the speed in Section 4. Section 5 gives concluding remarks. 2. Sample and data

On the NYSE's list of non-U.S. issuers, there were 489 listings from 460 issuers as of December 28, 2004. Some listings are from firms incorporated in "flag-ofconvenience" countries, which are not their real place of operations. Following Pulatkonak and Sofianos (1999) and Bacidore and Sofianos (2002), we delete them from our sample. 3 In addition to common stocks, preferred stocks, exchange traded funds, and global (depository) shares are listed on the NYSE. 4 They are excluded from the sample. Among the remaining stocks, 320 from 39 countries have data for the analyses and comprise our final sample.

3 The following are the countries and the number of listings deleted: Bahamas, 2; Bermuda 29; Cayman Islands, 4; Guernsey, 1; Liberia, 1; Netherlands Antilles, 1; Panama, 2; Puerto Rico, 6. 4 Global shares are designed to raise capital in multiple international markets and are very different from ADRs. For details, see Karolyi (2003).

3

Our sample period covers the year of 2005. Country-level data are from La Porta et al. (1999, 1998) and Eleswarapu and Venkataraman (2006). Firm-level data are from Compustat, CRSP, and Datastream. Intraday transactions data, including trade prices and bid and ask quotes, are collected from the exclusive Reuters dataset maintained by SIRCA.5

Table 1 presents the geographical distribution, firm and countries statistics for sample firms. Canada and the United Kingdom top the list with 56 and 35 stock listings, respectively. This accords with the argument that cultural, economic, and geographical proximity significantly influences the choice of foreign listing locations (see Bruner et al., 2000 and Sarkissian and Schill, 2004). Following Bruner et al. (2000), we use familiarity, a dummy variable equal to one if a listing is from a country sharing a common border, language, or culture with the U.S., to capture this effect. Across the sample firms, there is salient variation in the fraction of trading taking place on the NYSE, from the lowest of less than 1%6 (the U.K.) to the highest of 97% (Peru). On average, the NYSE retains about 27% of the combined trading at home and the NYSE. Benchmarked with an average stock on the NYSE, sample firms have a lower stock price ($29 versus $36), very similar daily return volatility (1.7% versus 1.6%), larger size in terms of market value of equity (about $20 billion

5 SIRCA stands for Securities Industry Research Centre of Asia-Pacific. For details, see .au 6 There is still a substantial amount of trading on the NYSE. The average daily volume is about $16 million for a U.K. stock.

4

versus about $8 billion), and lower daily trading volume (about $11 million versus about $21 million).7

Our sample of international issuers shows substantial divergence in legal, judicial, political, accounting, and corporate governance institutions. Forty percent of the firms are from a country of common-law legal origin, which has stronger legal protection of investors, while the rest are from civil-law countries. Judicial efficiency assesses "the efficiency and integrity of the legal environment as it affects business" (La Porta et al., 1998). In our sample, Indonesia is rated the least efficient (2.5 out of the full rating of 10). However, many countries have a score of 10, the most efficient enforcement of law. The rating on the political stability ranges from 48 (Indonesia) to 95 (Luxembourg). The average score of political stability is approximately 81 (out of 100). Accounting standards assess the accounting quality by examining and rating companies' inclusion or omission of 90 accounting items across 7 categories on their annual reports (see La Porta et al., 1998). The highest rating is 78 (Singapore and the U.K.), more than twice as large as the lowest score of 36 (Portugal). Anti-director rights measure the presence of 6 important corporate governance mechanisms. The rating ranges from 0 (Belgium) to 5 (6 countries). In comparison with the U.S., sample firms' home countries, on average, have lower institutional quality. 8

7 The average stock price and daily trading volume for NYSE firms are estimated from NYSE statistics. The average daily return volatility is calculated using CRSP data. The market value of equity at the end of 2005 is from Compustat. 8 To put it in perspective, the U.S. is a common-law country with high ratings: judicial efficiency (10), accounting standards (71), and anti-director rights (5).

5

3. The speed of convergence to market efficiency

3.1. The speed for foreign stocks on the NYSE

In an efficient market where stock prices fully and readily reflect relevant

information, past trading information cannot be used to predict future returns.

Therefore, the short-horizon return predictability of past trading information is an

inverse indicator of the market efficiency. In this line of thinking, Chordia et al.

(2005) originate an ingenious way to measure how long it takes to achieve weak-

form efficiency. They regress intraday short-horizon returns on lagged returns and

lagged order imbalances over intervals of the same length for intervals of 5, 10, 15,

30, and 60 minutes. If the past returns and order imbalances cannot predict returns

over a particular time interval (e.g., 30 minutes), then trading achieves efficiency

within the length of the interval (that is, the speed of convergence to market

efficiency is within 30 minutes). To measure the speed of convergence to market

efficiency for foreign stocks, we follow the method in Chordia et al. (2005).

Specifically, we run the following regression for each stock:

Re tt = 1 Re tt-1 + 2OIBDt-1(or OIBNt-1) + t

(1)

For every stock, a separate regression is run for each of the following time intervals

over 2005: 5, 10, 15, 30, 60, 90, and 120 minutes. The return over a time

interval, Rett , is calculated from the midpoint of the bid and ask quotes closest to the

end of the time interval. The midpoint returns are free from the serial correlations

induced by the bid-ask bounce. Rett-1 is the lagged return. To measure order imbalance, Lee and Ready (1991) algorithm is used to determine whether a trade is

6

buyer or seller initiated. Two measures of order imbalance are calculated. OIBDt-1 , order imbalance in dollars, is the total dollar amount paid by buyer-initiators minus the total dollar amount received by seller-initiators during the lagged time interval. OIBNt-1 , order imbalance in numbers, is the number of buyer-initiated trades minus the number of seller-initiated trades during the lagged time interval. The first interval of each day is excluded from regressions because lagged variables are needed.

Regression results for NYSE foreign stocks are presented in Panels A and D of Table 2. Results suggest that, on average, it takes more than 30 minutes but less than 60 minutes for a foreign stock to achieve efficiency. Whether order imbalance in dollars (Panel A) or in numbers (Panel D) is used, the past return and order imbalance lose their predictive power after 30 minutes. 3.2. The difference in speed between foreign and U.S. stocks

To better understand the results, we need to put this speed in perspective. A natural benchmark is domestic U.S. firms. We expect that it takes more time for foreign stocks traded on the NYSE to achieve efficiency than comparable domestic U.S. stocks traded on the same exchange for the following reasons. First, Chordia et al. (2005) find that the process to market efficiency is closely related to trading activities of the NYSE specialists. Specialists may behave very differently in the trading of foreign stocks on the NYSE. Indeed, Bacidore and Sofianos (2002) document that specialists' closing inventory positions for foreign stocks are closer to zero than comparable U.S. stocks, and specialists are less willing to participate in or

7

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download