The impact of U.S. and U.K. Cross-‐Listings on
performance of Industry Competitors
Jeroen Davidson1
University of Groningen, June 2013
Abstract
This study examines the competitive effects of U.K. and U.S. cross-‐listings on industry competitors from the period of 2002 until 2012. The findings of an event study show that the industry competitors of U.S. cross-‐listers experience significant negative short-‐term abnormal returns and significant negative effects on the long-‐term performance. However, the competitors of U.K. cross-‐listers experience significant positive short-‐term abnormal returns, and no significant long-‐term performance effects. Furthermore, a cross-‐ sectional analysis shows that the first mover effect is a significant predictor of the competitive effects.
Key words: cross-‐listing, competitive effects, industry competitor, and event study.
JEL-‐classification: G14, F31
WHEN A LISTED COMPANY decides to list its common shares on a foreign stock
companies that choose to list their common stock on stock exchanges based in the United States. However, firms may also choose to cross-‐list on European or Asian exchanges. A company can cross-‐list directly on a foreign stock exchange, or can do this by using a depositary receipt. A depositary receipt represents ownership of common stock in a foreign firm and is issued against common stock, held by a depositary bank in the home market of the issuing company2. For example, a German company can use depositary receipts to have its shares traded on the NYSE. Depositary receipts that trade on a U.S. stock exchange are known as American depositary receipts (ADRs) or American depositary shares (ADSs).
A company’s decision to choose for a cross-‐listing on an international stock exchange has received substantial attention from the corporate finance literature in the last two decades. This research focuses on the reasons for cross-‐listing and the cost-‐benefit analysis of such a transaction (see e.g. Karolyi, 2006; Benos and Weisbach 2004; Foerster and Karolyi 1999). Most studies focus on companies that cross-‐list on U.S. exchanges and find significant benefits for the issuing company. However, next to the impact a cross-‐listing may have on the performance of the cross-‐listing company, a secondary effect may be that this will impact the performance of industry competitors on the home-‐market of the company. There are two reasons why cross-‐listings may impact the performance of industry competitors. First, a cross-‐listing may change the competitive position of the cross-‐ listing company relative to its competitors, which results in valuation effects for the competitor. Second, industry competitors respond to an event if this event embeds
information that has industry wide implications3. In other words, if a cross-‐listing
conveys information that has industry wide ramifications, industry competitors should be affected by the cross-‐listing.
subsample and positive abnormal returns for the U.K. subsample. Finally, this study examines the effect on other performance indicators, next to the effect on the stock return. The empirical results show a negative effect on the stock return and operating income growth, but no significant impact on sales growth or capital expenditure growth.
Analysing the impact of a cross-‐listing on industry competitors is a fascinating subject in itself. However, the competitive effects of cross-‐listings are also important factors to take into consideration for several parties there being; (institutional) investors, industry competitors, and issuing firms. In a well-‐diversified portfolio, an issuing company comprises a relatively small portion of the total portfolio value. It is therefore valuable information for investors to know how the performance of companies is affected by a cross-‐listing in their industry, because this could severely impact the portfolio allocation decision. Likewise, companies that face a cross-‐listing in their industry find it valuable to understand the impact a new issuance will have on the performance and how to strategically anticipate on this information. Finally, also issuing firms will find it valuable to know the effect their issuance will have on their competitiveness relative to their industry competitors.
I. Literature Review
In this section, I will provide information on cross-‐listings and the competitive effects this listing may have on industry competitors. First, I will discuss cross-‐listings in general and provide reasons why firms cross-‐list. Secondly, I will examine the effect a cross-‐listing may have on industry competitors. The third part will analyse the effect cross-‐listings may have on longer-‐term performance of industry competitors.
A. Cross-‐listing
hypothesis), and improved protection (the bonding hypothesis). I focus on these three hypotheses to find explanations for the short-‐ and long-‐term performance of cross-‐listed companies. Next to that, I examine if possible differences exist between listing in the U.S. and the U.K.
A.1. Investor Recognition Hypothesis
In the capital asset pricing model (CAPM) with imperfect information (Merton, 1987), an extra factor is included in the Sharpe-‐Lintner Capital Asset Pricing Model (CAPM), the “shadow cost of information”, which models for the imperfect information about available investments. The model assumes that some investments are known to only part of the investors, and investing in these relatively unknown investments requires a return premium for taking this unsystematic risk. This assumption has two important implications. First, the number of investors who are aware of the investment positively affects the valuation of an investment. Second, the expected return on an investment is a decreasing function of the number of investors who are aware of the investment. Using this model in a cross-‐listing setting, companies have an incentive to cross-‐list if this will increase their number of stockholders, which will result in a higher market value of the company.
The literature finds empirical evidence of the “investor recognition hypothesis”4.
For example, Baker, Nofsinger, and Weaver (2002) show that companies that cross-‐ list on the NYSE or the London Stock Exchange (LSE) benefit from a significant increase in investor recognition. Their results, however, display larger benefits for listings on the NYSE compared to listings on the LSE. A study by King and Segal (2009)
Table I
Overview of papers on cross-‐listings
Authors Period Hypotheses examined Findings Foerster and Karolyi (1999) 1976-‐ 1992 Investor recognition and liquidityPositive effect of both investor recognition and liquidity
Baker et al. (2002)
1976-‐ 1996
Investor recognition
Positive effects of investor recognition, but stronger effects for NYSE
Mittoo (2003) 1976-‐ 1998
Liquidity Positive liquidity effects in the short run Bris et al.
(2006) 1987-‐ 1996 Investor recognition, bonding and liquidity
Positive effect of investor recognition is more than double that of bonding and significant increase in liquidity
King and Segal (2009) 1988-‐ 2005 Investor recognition and bonding
Positive investor recognition and bonding effects. Investor recognition effect is
permanent when increased shareholder base is maintained Berkman and Nguyen (2010) 1996-‐ 2005 Bonding and liquidity
No support for bonding or for liquidity hypothesis Cetorelli and Peristiani (2010) 1990-‐ 2006
Bonding Positive effects of bonding over the five-‐year period following the listing
show that the impact of enhanced investor recognition is a durable increase in market value for companies that attract and are able to sustain a wider investor base. However, companies that fail to sustain a wider investor base experience a post listing decrease, with market values equal to levels prior to the cross-‐listing within two years after the listing.
A.2. Liquidity hypothesis
in a lower cost of capital (see, e.g. Amihud and Mendelson, 1986). The literature assumes that prior to cross-‐listing the company’s ability to raise capital is restricted by the liquidity accessible in its home market, which may not be sufficient for a company’s need for capital. The competition between exchanges and the extra revenues generated when the stockholder base is widened can entice exchanges to decrease spreads on the home market and increase trading activity, as shown by Foerster and Karolyi (2000). Pagano, Roëll, and Zechner (2002) argue that since U.S. markets are larger in size compared to European markets, they offer cross-‐listing companies more liquidity. Empirical studies find that the liquidity gain is the major ground on which the short-‐run abnormal returns of cross-‐listing companies are based, but it does not impact the long-‐term performance5. However, these benefits also arise for domestic non-‐cross-‐listed companies.
A.3. Bonding Hypothesis
If a firm selects a foreign exchange with tighter regulations than its home exchange, it commits to conform to the higher disclosure and corporate governance standards. Benos and Weisbach (2004) and Karolyi (2006) find that companies can show quality by listing on an exchange with a strict regulatory environment, which results in a lower cost of capital. This means there will be a positive share price reaction to such an announcement and may result in firms actively choosing their listing location because of differences in regulation between exchanges.
The channels through which a cross-‐listing on an exchange with a stricter regulatory environment may result in positive valuation effect for the listing
company are formulated in the “bonding hypothesis” of Coffee (1999; 2002). According to this hypothesis the bonding may occur either through legal bonding or through reputational bonding. Legal bonding refers to investor protection laws that allow minority stockholders to pursue lawsuits against foreign managers. Reputational bonding refers to financial agents who supervise the cross-‐listed company and increase the information supply, thereby mitigating the information gap between controlling and minority stockholders.
This expectation is in line with a number of studies included in Karolyi (1998; 2006) that find a signification valuation effect for firms cross-‐listing in the U.S., which has the strictest regulatory environment, and not significant valuation effects otherwise. Abdallah, Abdallah, and Saad (2011) find that all cross-‐listed companies will benefit from a higher level of trading volume after cross-‐listing. However, companies that cross-‐list on highly regulated exchanges experience the largest increase in their trading volumes after their cross-‐listing. They further find a larger increase in trading activity for companies from low-‐investor-‐protection exchanges compared to those from high-‐investor-‐protection exchanges.
B. Competitive Effects of Cross-‐listings
industry competitors: information effects and competitive effects. Information effects originate from the theory that companies respond to an event in their industry if this event embeds information that has industry wide implications. In the cross-‐listing literature this theory is translated in the risk-‐sharing hypothesis. The competitive effects stem from the theory that an event may change the competitive position of a company relative to its competitors, which is translated in growth opportunities hypothesis in the cross-‐listing literature. Table II shows an overview of the literature on the competitive effects of cross-‐listings. I identify variables related to both risk-‐ sharing and growth opportunities effects to determine the importance of these variables in explaining the reaction of industry competitors to cross-‐listings.
B.1. Risk-‐Sharing Hypothesis
causes for a market integration effect where not cross-‐listed industry competitors are, as a result, valued in an international context rather than in an isolated market. To the extent that there is an effect on competitors, the largest impact should be expected from the “first mover” from a country or an industry, since this listing is likely to cause for a larger additional exposure than following cross-‐listings.
B.2. Growth Opportunities Hypothesis
According to the growth opportunities hypothesis, a company with highly valued growth opportunities demands additional possibilities to raise capital and therefore decides to cross-‐list, which reflects positively on the valuation of the cross-‐listed company. This hypothesis implicates that not cross-‐listed industry competitors are
Table II
Overview of papers on competitive effects of cross-‐listings
Authors Period Hypotheses examined Findings Bradford et al. (2002) 1970-‐ 1999 Growth opportunitiesU.S. competitors experience positive growth opportunity effects, whereas domestic competitors do not
Lee (2002) up to 2001
Growth opportunities and risk-‐ sharing
Negative growth opportunity effects, but no risk-‐sharing effects
Karolyi (2004) 1976-‐ 2000
Risk-‐sharing Negative risk-‐sharing effects because domestic markets become less integrated and decrease in size
Fernandes (2009)
1983-‐ 2001
Risk-‐haring Positive risk-‐sharing effects on domestic competitors, which are larger than negative effects on liquidity
Melvin and Valero (2009) 1974-‐ 2002 Growth opportunities and risk-‐ sharing
perceived as having relatively less growth opportunities, and therefore a negative effect on the market value of these companies is expected. The valuation effects on industry competitors may be negatively associated with the relative size of the industry competitor because in the typical case less information is available about relatively small companies (Slovin, Sushka, and Bendeck 1991). Accordingly, investors are more likely to reassess the value of relatively small competitors because the enclosed information may already be contained in the value of relatively large competitors. This leads to the hypothesis that the growth opportunity effects are greater for relatively small competitors.
B.3. Empirical Evidence of Competitive Effects
The implication of the risk-‐sharing hypothesis is different from the growth opportunities hypothesis. The impact of a cross-‐listing on the market value of industry competitors therefore depends on whether the effect of a decreased cost of capital is larger than the impact of the industry competitors being perceived as having less growth opportunities relative to the cross-‐listed company. If the effect of lower growth opportunities is weaker than the liberalization effect of the increased risk-‐sharing, then the market value of industry competitors should rise. Because of these offsetting effects, a significant effect of a cross-‐listing on industry competitors could be hard to detect.
around a cross-‐listing. A more recent paper by Melvin and Valero (2009) analyses the effect of a cross-‐listing on the primary industry competitor of the listing company. They find negative abnormal returns for industry competitors around both listing and announcement dates. Their analysis suggests that investors perceive industry competitors as less transparent and with less growth opportunities compared to the cross-‐listed company. Another study that examines the impact of cross-‐listings on industry competitors is that by Bradford, Martin and Whyte (2002). They examine the effect of cross-‐listings on both U.S. competitors and home-‐market competitors, by analysing listing dates. They create a portfolio of all industry competitors for which data are available, and find significant evidence of a positive effect on U.S. competitors. On the other hand, they do not find a significant effect on home-‐market competitors. Karolyi (2004) shows that cross-‐listings create negative spill over effects on the home-‐market, using a sample of emerging market exchanges. Furthermore, applying a sample of 55 countries, Levine and Schmukler (2006) show significant negative spill over effects of internationalization on the liquidity of companies in the home-‐market of the listing company.
To analyse the competitive effects of cross-‐listings, I form hypotheses to structure the empirical tests. The primary question in this thesis is whether cross-‐listings have an impact on the performance of industry competitors. The effect on the performance can be estimated in several ways. The first hypothesis applies to how the market value of an industry competitor reacts, in the short term, to a cross-‐ listing in its industry. A cross-‐listing is expected to allow the issuing company to compete more successfully against industry competitors. Based on the discussion in the previous section, I expect a negative effect on the stock returns of industry competitors. More formally:
Hypothesis 1: Stock returns of publicly traded companies react negatively to
cross-‐listings in their industry.
C. Performance
A study by Hawawini, Subramanian, and Verdin (2003) examines the question whether company performance is driven primarily by industry or company factors. They find that industry effects are more important for the operating performance of a company than company-‐specific factors. This result shows that the competitive balance within an industry is an important factor in explaining the operating performance of a company. For example, Otchere (2009) studies the effect of a bank privatization on rival banks. He finds that investors see a privatization as negative news for industry competitors, which results in a negative impact on the operating performance. In line with these results, Kennedy (2000) finds that industry competitors of financially distressed companies experience declines in operating performance around a bankruptcy filing. Most related to this study, Hsu, Reed and Rocholl (2010) examine the operating performance of companies after an IPO in their industry. They show that sales growth, operating income, capital expenditure and stock return are significantly negatively affected in the four years after an IPO. This study also analyses the short-‐term competitive effects of IPOs and finds significant negative abnormal stock returns of industry competitors. Since these results are similar to the short-‐term competitive effects of cross-‐listings, I anticipate that there may also be longer-‐term performance effects of cross-‐listings. In other words, the completion of a cross-‐listing is expected to give the cross-‐listed firm a competitive advantage over industry competitors, which results in a negative impact on their operating performance. More formally:
Hypothesis 2: The operating performance of existing industry competitors will
II. Data & Methodology
A. Data
A.1. Sample Description
I measure the effect of a cross-‐listing on the performance of listed industry competitors. To analyse this effect I use a sample of 55 foreign companies that listed on organized stock exchanges in the United States (U.S.) and the United Kingdom (U.K.) between 2002 and 2012. The sample includes listings on the New York Stock Exchange (NYSE), the American Stock Exchange (AMEX), NASDAQ, and the London Stock Exchange (LSE). Listings must meet the following criteria to qualify for inclusion in the sample: (a) a company must have an identifiable listing date for a cross-‐listing; (b) a prior listing on an exchange in the home country; (c) the listing firm has at least one industry competitor with the same four-‐digit Standard Industrial Classification (SIC) code at the time of the listing6, which has share price data available from
Datastream; (d) the listing is a first-‐time listing in the US or the UK and is not a switch from another exchange in the respective country; and (e) the listing does not occur on the same date as other listings in the industry. The sample consists of cross-‐ listings identified from data compiled by the Bank of New York7, from stock
exchange websites, and from data obtained from M&A database Zephyr8. To avoid survivorship bias, I also include firms that were cross-‐listed at some point during the sample period but that are no longer listed today9. I exclude investment funds.
6 For more information see: http://www.sec.gov/info/edgar/siccodes.htm 7 http://adrbny.com
8 zephyr2.bvdep.com/
Within a group of companies with the same four-‐digit SIC code there are large differences between the companies in terms of size and actual trade description. Furthermore, as pointed out by Melvin and Valero (2009), the effects of companies being perceived with relatively lesser growth opportunities compared to the cross-‐ listing company should apply for the closest industry competitors. This is the reason that I do not form an equally weighted portfolio consisting of all industry competitors that share the same four-‐digit SIC code, but follow Melvin and Valero (2009) and hand-‐pick the closest competitor in terms of size and trade description, and use this company in the analysis. This results in a unique sample of cross-‐listed companies and closest industry competitors.
Miller (1999) and Doidge (2004) emphasize that in efficient markets, investors’ expectations regarding the change in the valuation of the firm as a result of the cross-‐listing are incorporated into stock prices immediately. This means that studying stock price reactions around the announcement date enhances the assessment of the market’s reaction to cross-‐listings. However, Foerster and Karolyi (1999) and Sarkissian and Schill (2009) point at potential problems in the identification of announcement dates. When attempting to collect accurate announcement dates, I discovered that identifying these dates is indeed very hard and prone to large errors. Therefore I decided not to include an analysis of stock returns around the announcement date in this study, but to focus on listing dates10.
banks update the effective date of listing whenever a cross-‐listed company changes its listing type, listing exchange, or depositary bank; as a result, the listing date may not reflect the original cross-‐listing date. For that reason I crosscheck listing dates with the available sources.
Stock prices are collected from Datastream. I require stock return data for 250 trading days before the earliest public announcement by the firm of its plan to cross-‐ list. The final sample therefore consists of 55 cross-‐listings, of which 38 are cross-‐ listed in the US, and 17 are cross-‐listed in the UK.
B. Methodology
B.1. Event study
This thesis uses the event-‐study methodology to measure the reaction of investors to news of a cross-‐listing event.11 The methodology, firstly introduced by Fama and French (1969), is based on the assumption that capital markets are sufficiently efficient to examine the effect of new information (events) on the anticipated future profitability of a company. The objective is to test whether abnormal returns (ARs) can be found in the dataset around the event. It involves the following steps: (1) prediction of the “normal” return within the event window without the occurrence of the event;12 (2) estimation of the AR during the event
window; and (3) analysing whether the AR is statistically different from zero.
11 For more details, see Brown and Warner (1980, 1985)
12 The event window comprises of the day where the event occurred (day 0) and some days prior to
The calculation of the actual returns (R) is the first step in calculating the estimate “normal” returns. The actual returns are computed applying the continuously compounded method:
𝑅!" = ln ( !!"
!!(!!!)) (1)
where 𝐼!" equals the total return index of security i at day t.
Next, the OLS market model is employed to predict the “normal” returns. The model assumes that the return of all securities is linearly related to the market portfolio return. By removing the portion of the return that is related to variation in the market’s return, the variance of the abnormal return is reduced (MacKinlay, 1997). This in turn can lead to increased ability to detect event effects. The OLS market model is given by:
𝑅!" = 𝛼!+ 𝛽!𝑅!" + 𝑒!" (1) (2)
with 𝐸 𝑒!" = 0 and 𝑉𝑎𝑟 𝑒!" = 𝜎!!!
where 𝑅!" and 𝑅!" are the returns on security i and the company’s home stock market’s index respectively during period t. 13
Equation (2) is estimated over a 200 trading-‐day period which runs between 230 trading days prior to the event up to 31 trading days prior to the event14. With the
estimates of 𝛼! 𝑎𝑛𝑑 𝛽! from equation (2), the “normal” return in the event window is estimated. The estimation error (difference between the actual and the estimated normal return), generally referred to as the abnormal return (AR), is calculated as:
𝐴𝑅!" = 𝑅!" − 𝛼!− 𝛽!𝑅!" (3)
13 Data on the home stock market index is collected from Datastream. Returns are then calculated
where 𝐴𝑅!" is the abnormal return for security i at day t. 𝑎! 𝑎𝑛𝑑 𝛽! are the OLS estimates of the parameters of the market model.
Under the null hypothesis, the ARs are jointly normally distributed with a zero conditional mean and conditional variance 𝜎!(𝐴
!"): 𝜎! 𝐴
!" ~ 𝜎!!!. (4)
To analyse the perseverance of the effect of the event during the event window, the abnormal return can be added to get the cumulative abnormal returns (𝐶𝐴𝑅!) for security i in the event window period:
𝐶𝐴𝑅! 𝑇!, 𝑇! = !! 𝐴𝑅!"
!!!! (5)
where 𝑡 − 𝑎 ≤ 𝑇! < 𝑡 < 𝑇! ≤ 𝑡 + 𝑏 ∈ event window, and t-‐a and t+b are the lower and upper limits of the event window, respectively. The variance of the CAR for security i is:
𝜎!! 𝑇
!, 𝑇! = 𝑇! − 𝑇! + 1 𝜎!!! (6)
An aggregation of interest can also be performed across both time and events. In that scenario, the average cumulative abnormal return is defined as:
𝐶𝐴𝐴𝑅 𝑇!, 𝑇! =!! ! 𝐶𝐴𝑅!(𝑇!, 𝑇!)
!!! (7)
To examine if the abnormal returns differ significantly from the zero conditional mean, statistical test are required. Applying the parametric student t-‐test tests the statistical significance. The parametric test is based on standardized abnormal returns (Brown and Warner, 1985). The test takes into account cross sectional dependence in the company’s specific excess return. Under the null hypotheses that the abnormal returns are zero:
𝑡 = !""# !!,!!
!"# !""# !!,!! !/!
III. Empirical Results
As discussed in the previous sections, this thesis measures the impact of a cross-‐ listing on industry competitors in the home country. This section presents the empirical results of the short-‐term return reaction of industry competitors around cross-‐listings, the impact on the industry competitors’ operating performance, and the cross-‐sectional analysis.
A. Short-‐Term Return Reaction
Hypothesis 1 states that a part of the evidence on the impact of a cross-‐listing on
while the U.K. subsample experiences an increase of its stock return. The next section will analyse these results more formally by calculating the significance of the CARs over several event windows. The results from figure 1 are used to motivate the choice for the different event windows.
To find formal evidence for abnormal returns of companies around completed cross-‐listings in their industry, I use an event study to analyse the significance of the CARs. Panel A of table III shows the mean cumulative abnormal returns (CARs) of companies around the completion date of a cross-‐listing in their industry. The results of the total sample show significant negative CARs around the completion date of a cross-‐listing. The CAR in the period between 1 day prior to and 10 days after the cross-‐listing adds up to -‐2.78% and is significant at the 5% level. These negative
-‐5% -‐4% -‐3% -‐2% -‐1% 0% 1% 2% 3% 4% 5% -‐30 -‐20 -‐10 0 10 20 30 Mean CAR
Total US UK
Event Date
Figure 1. Cumulative Abnormal Returns of companies around cross-‐listings in their industry.
Table III
Abnormal Returns of Industry Competitors around Completion Dates
This table reports the cumulative abnormal return (CAR) of industry competitors around cross-‐listing events both for the total sample of 55 cross-‐listings and for 38 U.S. and 17 U.K. cross-‐listings separately. Cross-‐listing events are selected as events if the cross-‐listing company has a prior stock exchange listing and an industry competitor with the same 4-‐digit SIC industry code in the "home country". Abnormal returns are calculated as the difference between the realized stock returns and the predicted return using the market model over each event window. The market model uses an estimation window of 200 days of daily stock returns ending 30 days before the cross-‐listing event. t-‐ statistics are reported in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
Days Total U.S. U.K.
(-‐10,1) 0.63% -‐0.86% 3.97%* (0.58) (-‐0.74) (1.79) (-‐10,5) -‐1.37% -‐3.23%* 3.55% (-‐0.77) (-‐1.97) (1.26) (-‐10,10) -‐1.71% -‐3.80%* 2.94% (-‐0.96) (-‐1.78) (0.96) (-‐5,1) -‐0.74% -‐2.16%** 2.45% (-‐0.8) (-‐2.05) (1.56) (-‐5,5) -‐2.51%* -‐4.53%*** 2.03% (-‐1.95) (-‐3.09) (0.9) (-‐1,5) -‐2.21%** -‐3.30%*** 0.25% (-‐2.36) (-‐3.02) (0.15) (-‐1,10) -‐2.78%** -‐3.87%*** -‐0.35% (-‐2.2) (-‐2.81) (-‐0.13)
results are consistent with findings from Melvin and Valero (2009), who also find significant negative abnormal returns around the listing date. Lee (2004) also provides proof for short-‐term negative abnormal returns for industry competitors. Both studies use companies that cross-‐list on U.S. stock exchanges.
The U.K. subsample, on the other hand, shows mostly positive abnormal returns over the tested event windows. The event window, which starts 10 days prior to the listing and ends 1 day after the listing is the only event window which generates a significant CAR. The abnormal return of this window amounts to 3.97% and is significant at the 10% level. Table A.2 of the appendix shows a sensitivity analysis of the above-‐discussed results. The CARs over the different event windows are similar to the results from table III. Interesting to note is that the abnormal returns of the U.S. subsample become even more significant. This shows that the empirical results are robust and provides even more evidence of the presence of competitive effects. Overall, the empirical tests find mixed evidence for hypothesis 1. The U.S. subsample shows a negative abnormal return reaction, which is proof for hypothesis 1. However, the U.K. subsample shows a positive abnormal return reaction. This is further proof of competitive effect, however in an opposite direction from the expectation of hypothesis 1.
B. Performance effects
hypothesis 2, there should be a similar negative effect on the operating performance of competitors. I analyse the operating performance of industry competitors through time to examine whether the performance of a company deteriorates after a cross-‐ listing. Since there are opposite effects of U.S. and U.K. cross-‐listings on the short-‐ term stock return, I will continue to analyse these subsamples apart from each other.
B.1. Univariate Statistics
Table IV Univariate Statistics
This table shows univariate statistics for 5 performance indicators. All indicators are measured as the annual growth rate in 2002 dollars. Panel A shows the average ratios for the total sample of industry competitors. Panel B subdivides the sample in U.K. and U.S. companies and shows their average ratios separately. The table further shows the number of observations, the z-‐score of the Wilcoxon signed rank test, and the number of negative differences.
Period Net Income Growth Operating Income Growth Asset Growth Sales growth Capital Expenditure Growth Panel A: Performance indicators of the total sample
Four-‐year average before the cross-‐listing
21.66% 11.34% 13.56% 11.98% 2.10%
Four-‐year average after the cross-‐listing
15.89% 7.17% 10.82% 9.60% 6.51%
N 41 40 45 47 43
Wilcoxon significance -‐1.37 -‐0.89 -‐0.91 -‐0.42 0.66
% N Negative 60.98% 57.50% 55.56% 63.83% 48.84%
Panel B: Performance indicators of U.K. and U.S. companies compared
U.S. Cross-‐listings
Four-‐year average before the cross-‐listing
22.42% 10.63% 11.18% 9.79% -‐3.62%
Four-‐year average after the cross-‐listing 15.28% 5.50% 11.37% 8.92% 5.32% N 32 30 33 34 31 Wilcoxon significance -‐1.27 -‐0.81 -‐0.26 -‐0.35 1.12 % N Negative 62.50% 60.00% 54.55% 64.71% 41.94% U.K. Cross-‐listings
Four-‐year average before the cross-‐listing
18.94% 13.48% 20.11% 17.71% 16.86%
Four-‐year average after the cross-‐listing 18.07% 12.20% 9.29% 11.38% 9.58% N 9 10 12 13 12 Wilcoxon significance -‐0.30 -‐0.26 -‐1.33 -‐0.18 -‐0.71 % N Negative 55.56% 50.00% 58.33% 61.54% 66.67%
change in the performance indicators of companies after a cross-‐listing in their industry.
B.2. Panel Regression Results
To find significant proof that a cross-‐listing affects the performance of industry competitors, a panel regression will be used. The regression includes other variables that are known predictors of performance. The variables are taken from Hsu, Reed, and Rocholl (2010) and are; company age since listing, company assets, and past performance. This setting allows for testing of hypothesis 2 by estimating several performance indicators, while adjusting for several variables that are established forecasters of performance.
𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒!" = 𝛼 + 𝛽 ∗ 𝐶𝐿!"+ 𝛾 ∗ 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠!"+ 𝜀!" (9) Performance is measured as capital expenditure growth, sales growth, abnormal stock return, and operating income growth in each year t for every company i. The dummy variable CL indicates whether year t is between one year and three years post a cross-‐listing in the industry of company i. The regression uses all years between 1991 and 2012 for which data is available and thus gives a panel regression in which each company has data from both cross-‐listing and non-‐cross-‐listing years15. For every performance indicator, two models are estimated. One model includes the cross-‐listing dummy variable and the other excludes this variable. This setting allows
Table V
The Impact of Cross-‐Listing Events on Industry Competitors
This table reports the findings of a panel regression of industry competitors’ operating performance on a cross-‐listing indicator and control variables from 1991 until 2012. The used performance measures are: capital expenditure growth, sales growth, abnormal stock return, and operating income growth. The first three performance indicators are measured as the difference between the log of the indicator in year y minus the log of the indicator in year y-‐1. The abnormal stock return is the difference between the yearly stock returns and the yearly return on the stock market index in the “home country”. The cross-‐listing dummy (CL) is equal to one in the cross-‐listing event year and the three years after the event. Log (age) is the log of the numbers of years the company is publicly listed. Log (assets) is the log of the level of assets in the previous year. t-‐statistics are stated in the parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Dependent variable Sales Growth
Capital Expenditure Growth Operating Income Growth Abnormal Stock Return Panel A: Performance indicators of the U.S. subsample
CL -‐0.027 0.205 -‐0.165* -‐0.090*
(-‐0.83) (1.42) (-‐1.92) (-‐1.85)
Lag dependent variable -‐0.008 -‐0.233*** -‐0.224*** -‐0.155*** (-‐0.17) (-‐4.38) (-‐3.92) (-‐3.44) Log (Assets) -‐0.054*** -‐0.368*** -‐0.098* -‐0.104*** (-‐2.62) (-‐4.56) (-‐1.86) (-‐3.24) Log (Age) 0.060** 0.357*** 0.113 0.133** (2.13) (2.87) (1.51) (2.12) Intercept 0.725*** 4.352*** 1.278* 1.234*** (2.82) (4.35) (1.96) (3.23)
Fixed effects? Yes Yes Yes Yes
N 380 343 317 439
R² 0.007 0.007 0.009 0.009
R² (without CL) 0.006 0.007 0.006 0.007
Panel B: Performance indicators of the U.K. subsample
CL 0.016 -‐0.200 0.125 -‐0.099
0.26 -‐1.04 1.17 -‐1.14
Lag dependent variable -‐0.273*** -‐0.121 -‐0.265*** -‐0.124 (-‐3.96) (-‐1.28) (-‐3.18) -‐1.52 Log (Assets) -‐0.055 0.061 -‐0.125 -‐0.2070** (-‐1.14) (0.37) (-‐1.40) (-‐2.07) Log (Age) 0.058 -‐0.068 0.153 0.377** (0.96) (-‐0.36) (1.40) (2.32) Intercept 0.861 -‐0.441 1.744 2.008* (1.54) (-‐0.23) (1.64) (1.96)
Fixed effects? Yes Yes Yes Yes
N 112 101 97 147
R² 0.060 0.001 0.091 0.000
for testing of the explanatory power of the cross-‐listing dummy in the changes in the performance indicators. The model is estimated using fixed effects.
Furthermore, none of the performance indicators are significantly affected by the occurrence of a cross-‐listing.
C. Cross-‐sectional analysis
The findings of the empirical tests establish that there are competitive effects of cross-‐ listings. This section will further analyse the competitive effects of cross-‐listings by examining the differences in magnitude of the performance impact across industry competitors. In other words, I use a cross-‐sectional analysis to test if differences in performance impact can be explained and maybe even predicted by a numbers of factors. As discussed in the literature review, I expect first of industry and size of the industry competitor to be significant factors in the cross-‐sectional analysis16.
Furthermore, I include several other factors that I expect to predict cross-‐sectional differences. Table VI gives an overview of the variables included in the cross-‐sectional analysis.
Age. This variable measures the number of years the company is publicly listed.
Hsu, Reed, and Rocholl (2010) find that firms have life cycles and that performance and growth rates depend on the stage in the life cycle. This can affect how a change in the competitive balance of an industry affects a company. I therefore use this variable as a control variable in the cross-‐sectional analysis.
Developing. Companies from less developed countries are expected to benefit
more from the liberalizing effect of a cross-‐listing. Melvin and Valero (2009) reason
16 I do not include “first of country” as a variable in the cross-‐sectional analysis since only one of the
Table VI
Variables cross-‐sectional analysis
Variable Definition Source
Age The number of years from the IPO date to the date of the cross-‐
listing. Datastream
Developing An indicator variable equal to one if the home country of the firm is a
developing country according to the IMF www.imf.org
First An indicator variable equal to one if the cross-‐listing company is the
first company of the industry to cross-‐list in the U.S. or the U.K. Datastream
Size Book value of assets Datastream
Legal An indicator variable equal to one if the cross-‐listing company originates from a country with a common law tradition and zero if the legal tradition is civil law.
Crisis A dummy variable equal to one if the cross-‐listing was completed in 2007 or later years. This variable intends to capture the possible effects of the financial crisis on the competitive effects of cross-‐ listings.
Prestige The stock market prestige of the home country of the cross-‐listing company. Stock market prestige is a measure taken from Cetorelli and Peristiani (2010), who define stock market prestige as a measure of the international importance of a stock exchange. This is reflected in its ability to provide capital for foreign companies and its capability to generate information.
Cetorelli and Peristiani (2010)
that the risk-‐sharing effects for industry competitors from less developed countries is larger. I therefore expect a positive sign.
First mover of industry (First). As discussed in the literature review, the first cross-‐
listing of an industry is expected to generate larger additional exposure than consecutive cross-‐listings. To the extent that competitive effects arise because of information effects, these effects will be most powerful for the first cross-‐listing in an industry. I therefore expect a positive risk-‐sharing effect of a first cross-‐listing of an industry.
Size. As discussed in the literature review, in the typical case less information is