Does cross listing in developed capital markets increase liquidity and return in home the market? : evidence from Africa

(1)

1

MASTER THESIS

Does cross listing in developed capital markets increase liquidity and return in home the market? Evidence from Africa.

Micheal Bediako

10639276

Master International Finance MSc

University of Amsterdam, August 2015

Thesis Supervisor: Jens Marten

(2)

(3)

2 Dedication & Acknowledgements

Submitted as partial fulfilment for a master degree requirement to the Amsterdam Business School, University of Amsterdam.

This master thesis is dedicated to my late father, Nathan Bediako; a man who never had much of an education himself but tolled from cradle to his grave to ensure that his children could become anything they dream of.

(4)

3 Abstract

By using a sample of 85 African firms that are crossed listed on LSE, NYSE and OTC between 1995 and 2015, this paper adopts an event study to analyse whether empirical evidence shows that cross listing of African firms on developed capital markets increases firm value via increased liquidity and excess stock price returns. This paper also examines whether listing characteristics, such as type, location of listing and industry, have an impact on cross-listed firm value and liquidity. The study results show that African firms that are cross-listed in developed capital markets generally report an increase in domestic stock liquidity, but not abnormal stock returns. The analysis shows that the increase of liquidity is driven by an increase of the transaction price and the number of trades after cross listing. Furthermore, when comparing the different platforms of trading, NYSE traded stocks report more liquidity and a higher abnormal return than LSE and OTC traded stocks. LSE report a higher abnormal return than OTC, although OTC trade at a lower cost or a higher liquidity than LSE. The difference in the abnormal returns between stocks traded on the stock exchange and stocks traded OTC indicates that a higher level of information symmetry and bonding in cross-listed firms is associated with higher abnormal returns. Lastly, with regard to the different industries, the Financial Services sector reports the highest abnormal increase of returns and the second highest increase in liquidity. The Energy, Industrial, IT Telecom & Materials sectors record the highest increase in liquidity and the second highest increase in post listing abnormal returns. After controlling the changes in volumes, price and the number of trades after cross listing, the liquidity spreads and premiums increased for all stocks. These results do not support the hypothesis that the African Financial Services sector is more integrated into the global financial system than the African Energy, Industrial, IT Telecom & Materials sectors. It is however more integrated than the Consumer and Health sectors.

(5)

4

TABLE OF CONTENT

1. INTRODUCTION 5

2. LITERATURE REVIEW AND HYPOTHESIS 7

3. DATA AND METHODOLOGY 11

4. RESULTS 18

5. DISCUSSION 34

6. CONCLUSION 37

7. APPENDIX 38

(6)

5

1. Introduction

Since the 1980s, a rapid global capital market integration has been supported by the growth in international cross listing1_{. Early research has credited the growth of} international cross listing to reasons such as mitigating market barriers and segmentation, access to developed capital markets, tax benefits, gain in liquidity and global market reputation (Errunza & Etienne, 1985; Alexander, Eun, & Janakiramanan, 1987; Foerster & Karolyi, 1993). Later research has concentrated on the bonding to stronger shareholder protection, agency and other corporate governance issues (La Porta, LopeZ-de-Silanes, Shleifer, & Vishny, 1997; Doidge, Karolyi, & Stulz, 2004).

In the late 1990s, international cross listing had slowed down, as witnessed by a wave of delisting on many global capital markets. By 2002, the number of international cross listing had fallen by 50% of its 1997 value (Karolyi, 2006). In a particular case, about 10% of Chinese firms delisted from the Singapore Stock Exchange between 2011-2013 (Cogman & Orr, 2013). In the US, where a fair share of delisting has taken place, firms sighted the cost of cross-listing due to regulation such as the Sarbanes-Oxley (SOX) act of 2002 as key factors (Doidge C. , Karolyi, Lins, Miller, & Stulz, 2009). However, these delisting sentiments have not stopped the competition of global stock exchanges from vying for pole position in the race for more cross listing.

In 2014, the London Stock Exchange (LSE) launched a campaign to increase its cross listing of African firms. Following its previous strategy of cross London-Johannesburg listings, the LSE has partnered exchanges in Nigeria, Morocco, Egypt, and Kenya to facilitate cross listings (Blas, 2014 ). This paper studies the reasoning underlying cross listing, focusing on liquidity and stock return. It seeks to answer the following central question: Based on available empirical evidence, does cross listing of African firms on developed capital markets increase their firm value in particular, via increased liquidity and excess stock price returns? A closely related question is; does the location of listing and industry characteristics have impact on cross-listed firm value & liquidity? Such an insight is important because, most cross-listing research has concentrated on listing on the US market (You, Parhizgari, & Srivastava, 2012), many

1_{Cross listing as opposed to foreign listing always refers to a dual or multiple listing in the} home country and abroad.

(7)

6 of whom have drawn conflicting conclusion on the benefiting factors underlying cross-listing (Foerster & Karolyi, Multimarket trading and liquidity: a transaction data analysis of Canada–US interlistings, 1998). Moreover, very little research has studied regional cross listing involving Africans firms. For instance Maka, Onyum, & Okumu (2015) provided inconclusive evidence that cross-listing improves domestic turnover or volume trades in 4 East African firms cross listed in Nairobi Stock exchange. On the other hand, Adelegan J. O. (2008); Adelegan J. O. (2009) has shown that there is a positive announcement effect, including an increase in post cross-listing performance in regional cross listing in sub-Saharan Africa. However, this later research was limited to cross listing within sub-Saharan Africa. At the time of writing, very limited research had examined the domestic impact of international cross listing of African firms on global capital markets.

Using domestic stock returns, trading spreads and premiums from 85 sub-Saharan African firms, this study uses an event study methodology to:

1. Examine the effect of the cross listing of African firms on developed capital markets (i.e. LSE, NYSE and OTC) on liquidity in domestic capital markets.

2. Examine the effect of the cross listing of African firms on developed capital markets (i.e. LSE, NYSE and OTC) on return in domestic capital markets.

3. Analyses the importance of choice and type of listing location and industry on the liquidity and return of cross-listed firms.

The answers to these questions have an important implication for African firms’ assessment of the cost and benefit of venturing into global capital markets via international cross listing. For investors, understanding the impact of inter-market competition on order flow is key in developing cost effective investment strategies for trading. For firms, such insight is important when evaluating the value-added in a cross-ling decision.

This paper is organized as follows: Section 2 starts with explaining the background theory relating cross listing to the liquidity and stock return. It ends by developing the expected hypothesis flowing from the theory. Section 3 provides an overview of depositary receipts, the type of data & sample selection and methodology for studying cross listing impact on liquidity and returns. Section 4 presents the results of both data analysis and hypothesis testing. Section 5 discusses the results within the

(8)

7 context of earlier research and theory of cross listing. Section 6 summarizes and concludes the study. In the appendix, which is section 7, one can find further information on the data, the methodology and the literature review.

2. Literature Review and Hypothesis

This section reviews research and empirical literature on the impact of cross listing on liquidity and price reaction to cross-listings. It ends with a hypothesis flowing from the current state of research.

Liquidity is the ease of dealing and transacting an asset, particularly the ease of turning it into cash. It can also imply the ease at which the asset can be traded with minimum price impact (Maka, Onyum, & Okumu, 2015; Roosenboom & van Dijk, 2009). To measure changes in liquidity, quoted bid-ask spread has been traditionally used as a proxy2_{(Berkman & Nguyen, 2010). Alternatively, turnover has also been} used to measure liquidity in other studies (Datar, Naik, & Radcliffe, 1998). For the purpose of this study, we take quoted bid-ask spread as the measure liquidity as used by Foerster & Karolyi (1998).

Liquidity is the lifeblood of the capital market. In a liquid market, there are at least one matching bid and ask quotes that makes transactions possible. And most importantly, adequate matching bid and ask quotes are essential to trade a certain amount of stock without price impact. In simple terms, without liquidity, there can be no trading (Wyss, 2004; Maka, Onyum, & Okumu, 2015). For exchanges and listed firms, adequate liquidity is necessary to operate and sustain the smooth trading of stocks, i.e. a primary medium of trade, financing and valuation.

Moreover, the importance of liquidity to firm stock value can be seen in time trading; that is the ability to execute a transaction instantly at the current price, tightness of trading; that is the ability to execute a transaction about the same quoted prices at the same time, depth of trading; that is the ability to execute an amount of transaction without affecting the quoted prices, and resiliency of trading; that is the ability to execute an amount of transaction with little effect the quoted prices (Wyss, 2004).

(9)

8 Lastly, to decrease cost of capital and increase stock value, firms can leverage on liquidity by cross listing in deeper and more liquid equity markets (Roosenboom & van Dijk, 2009). Hamilton (1979) was first to theorize and hypothesize that an increase in stock demand competition from cross listing in deep liquid markets can lower trading cost as measured by bid-ask spread. Theoretically, once cross-listed, a stock will be available to a bigger cumulative equity market; i.e. home plus new foreign equity market. This will increase the potential number of buyers and sellers of the stock. Consequently, all other things being equal, stock trading volumes and turnover will increase, resulting in fall in bid/ask spread (Hamilton, 1979). Admati & Pfleiderer (1988) and Chowdhry & Nanda (1991) built on this foundation to show that, in multimarket trading, one market emerges as the winner, as measured by order volume and turnover. This happens because informed traders strategically maximize their profit by trading when or where the market has little impact on the price of their trade. This is usually done along uninformed liquidity traders where stocks trade at the lowest bid-ask spread (Admati & Pfleiderer, 1988). In essence, cross-listing’s liquidity only benefits trades on the dominant market. In concurrent empirical study, Hargis (1997) investigates liquidity in cross listing Southern American firms in the US. He finds a general rise in domestic increase in trade volumes and liquidity. Furthermore, Mittoo (1992) has noted that one of the principal reasons managers cite for US cross listing is liquidity by increase in trading volumes. More recently, Chan, Hong, & Subrahmanyam (2005) have found that, for emerging market firms, ADR premiums and discounts are associated with higher local stock market liquidity and illiquidity. Conversely, Mendelson (1987) draws scenarios where cross listing increases bid-ask spread and trading cost as a result. According to Mendelson, equally possible is a situation where competition between market makers in multimarket trading leads to a division of stock trading across smaller fragmented markets. This creates the risk of increase bid-ask spread and trading cost in individual fragmented market, consequently, an increase in the weighted average spread. To support this scenario, Noronha, Sarin, & Saudagaran (1996) investigated US firms cross listing in London and Tokyo, and found that the weighted average spread remained unchanged while the depth of bid-ask spread increased significantly.

While the research results reviewed above are non-conclusive, at least in the absolute sense, they do show that, for firms, liquidity is an extra layer on the cost of

(10)

9 capital that can reduce the market value of a firm. Hence, if one should assume that firms are profit maximizers, their decision and timing of cross listing should be taking towards increasing stock and ultimately firm value.

Miller (1999) studied the price reaction of cross listing by examining first time depositary receipt program of 183 stocks from 32 countries between 1983 and 1990. He found a 1.15% abnormal stock return at the home market within a 160 days event window surrounding the listing date. This conclusion supports the hypothesis that cross listing adds value by overcoming market segmentation and barriers to investments. In a concurring study done the same year, Foerster & Andrew (1999) examined ADRs of 153 stocks from 11 countries from 1976 to 1992. They found that, stocks achieve abnormal stock return around the announcement and listing dates. The abnormal return then declines thereafter. They concluded that, strategic decision market timing of cross listing during bullish market is equally important in explaining the abnormal returns achieved during cross listing.

Moreover, before Miller (1999); Foerster & Andrew (1999); McConnell & Sanger (1987), had examined 2482 stocks cross-listed on the NYSE from 1926 to 1982. They also found that stock experienced a positive and negative abnormal return before and after the listing date dates respectively. In a similar study, Dharan & Ikenberry (1995) explains that the persistence negative post listing abnormal return is timing of to avoid bad news. Consequently, one would expect that cross listing will generally lead to an increase in stock liquidity and positive stock return around the listing date. Following these arguments, one can hypothesize;

• Hypothesis 1a: Cross-listed African firms in developed equity markets will report an increased post listing liquidity compared to pre-listing liquidity in their domestic equity market.

• Hypothesis 1b: Cross-listed African firms in developed equity markets will report positive and significant abnormal returns within the cross listing event window. Some studies have attributed post cross-listing liquidity and returns differences to firm specific and listing characteristics such as corporate governance, type of listing, firm’s size, location of listing, industry etc.

(11)

10 First, proponents of the bonding effect (Berkman & Nguyen, 2010), have shown that cross-listed firms from countries of low quality of corporate governance or accounting benefits more from domestic liquidity and returns in the first 2 years of cross-listing, diminishing thereafter. In a concurring result, Doidge, Karolyi, & Stulz (2004) show that controlling shareholders of cross-listed firms in the US have an incentive of limiting their private privilege of control by bonding to a higher level of disclosure, in the hope of raising cheaper capital for future growth.

Second, the results from Silva & Chavez (2008) indicate that, liquidity and returns benefits of cross-listed vary with firm size and countries of origin. In empirical study, Pagano, Röell, & Zechner (2002) have concluded that the European capital markets typically cater for large and mature firms with little need for growth finance. The US market typical attracts the opposite. Moreover, they suggest that the more integrated two markets are, the less benefits cross listing would bring. They also show that small growth oriented firms are more likely to increase liquidity and stock return via cross listing than mature cash oriented firms.

Furthermore, Foerster & Karolyi (1993) tested and confirmed that the risk and price differential between cross-listed firms can be attributed to the type of industry. They found that cross-listed Canadian resource firms in the sectors forestry and mining, show lower abnormal return than non-resource firms. This implies that, Canadian resource firms were less segmented from the US market than non-resource firms. Smith & Sofianos (1997) confirmed this, by examining the global distribution of 254 non-US stocks listed on the NYSE. They found that, financial services show lower increase in liquidity and stock return than non-financial firms, partly because financial markets are more globally integrated.

Lastly, Silva & Chavez (2008) showed that higher levels of information asymmetry in cross-listed firms are associated with higher liquidity cost and lower stock return. Smith & Sofianos (1997) draw a similar conclusion by showing that level three ADRs are more liquid in trading than level two ADRs. Other researches have looked at the list-type dynamics via the bonding effect. For instance, Coffee (2002) argues that the bonding mechanism accounts for the higher liquidity and stock return for level three ADRs than level two ADRs. That is, stocks that agree to confirm to stricter US report and regulations tend to have a higher liquidity and stock return. In addition to the

(12)

11 aforementioned, Miller (1999) found a higher abnormal return for stock traded on exchanges than OTC.

Following these arguments, one can hypothesize;

• Hypothesis 2a: Exchange traded cross-listed stocks (i.e. DRs level 2&3 on NYSE & LSE) will have greater the informational symmetry & investor protection, hence the higher is cross-listing liquidity than OTC stocks.

• Hypothesis 2b: Exchange traded cross-listed stocks (i.e. DR level 2&3 on NYSE & LSE) will have greater the informational symmetry & investor protection, hence the higher is abnormal stock return within the listing event window than OTC stocks.

• Hypothesis 3a: NYSE traded cross-listed stocks will have greater investor protection; hence the higher liquidity than LSE traded cross-listed stocks.

• Hypothesis 3b: NYSE traded cross-listed stocks will have greater investor protection, hence the higher is abnormal stock return within the listing event window than LSE traded cross-listed stocks.

• Hypothesis 4a: All other things being equal, more globally integrated financial services sector firms will experience lower cross-listing liquidity than firms in the Energy, Industrial, IT Telecom & Materials and Consumer & Health Care sectors. • Hypothesis 4b: All other things being equal, more globally integrated financial

services sector firms will experience lower abnormal stock return within the listing event window than firms in the Energy, Industrial, IT Telecom & Materials and Consumer & Health Care sectors.

3. Data and Methodology

The section starts with an overview of depositary receipts. Explaining and summarising the type of data and the sample selection follow this. It ends with the methodology used for the study cross listing impact on liquidity and stock returns. Cross listing can be done with ordinary shares or with Depository receipt (DRs). Once a stock is cross-listed as an ordinary share, the company has to fulfil all the requirement of a full IPO in the exchange of cross listing. Alternatively, with an depositary, the company can choose different levels of bonding with the foreign exchange or whether it prefer any association an exchange at all. In this study, all

(13)

12 stocks used are depositary receipts with various levels of bonding. Hence, further discussion will be concentrated on Depositary receipts.

Depositary receipts (DRs)3_{are negotiable U.S. securities by a U.S. bank, known as a} “depositary bank” that usually symbolizes a non-U.S. company’s equity. Depositary receipts are freely traded in many global markets. They can also be listed on stock exchanges such as the NYSE, NASDAQ, LSE and the Luxembourg Stock Exchange (BNY Mellon, 2015). DRs are grouped into two classifications: sponsored and unsponsored DRs. This classification is further categorized into four facilities. One or more depositary banks can issue unsponsored DRs in response to market demand for a specific foreign security without a formal agreement with the company (BNY Mellon, 2015). Alternatively, the issuing company under a deposit agreement can choose a depositary bank to issue sponsored DRs. This gives the issuing company the right to input and control the DR program, particularly the right to list the DR on a stock exchange and to raise capital. Sponsored DRs can be issued in four facilities: Level I, II, III and Rule 144A programs.

Under Level I, DRs are traded in OTC Markets. Level I DRs are not required to fully register with and report under the U.S. Securities and Exchange Commission (SEC). Level I DRs are the simplest way for African companies to access global capital markets. Given their relative simplicity, 89% of our sample has cross-listed under this program. Level II & III make it possible for companies to trade on global stock exchanges and raise capital respectively. Under these levels, companies must meet applicable requirements for listing on the stock exchanges and are required to register, disclose and report to the SEC. The last option is private placement (SEC RULE 144A / regulation S). SEC Rule 144A allows to raise capital via private placement with qualified institutional buyers (QIBs) without the requirement to register, disclose and report to the SEC (BNY Mellon, 2015).

The sample for this study consists of 85 African firms that are crossed listed on LSE, NYSE and OTC between 1995 and 2015. The sample firms represent the total number of listed firms with complete and usable data. To identify African firms cross-listed on NYSE, LSE and OTC, a security search was done in the Bank of New York

3_{Negotiable securities can also be issued in other international non-U.S. depositary banks in} other major currencies.

(14)

13 Depository Receipt Directory, which contains a global database of all cross-listed DRs. A total of 127 African stocks are cross-listed globally. Using their ISIN numbers, the stock data for 85 companies was available in DataStream for 200 days before and after the listing dates. 76 stocks are cross-listed under Level I, six under Level II, three under Level III and one under Rule 144A. Table 1 below shows the summary composition of stock according to sector, industry and exchanges.

Table 1: This table gives a summary overview of all African stocks cross-listed on NYSE, LSE and OTC between 1995 and 2015 and their respective industry group.

Table 2 shows some key data highlights; 89% stocks are OTC traded, 76 total cross-listed firms representing 89% of stocks are South African. Each company can have only one DR in the sample. If a company had issued multiple DRs during the period covered, the first issuance was used in this study.

Table 2: This table shows a cross-sectional overview of all African stocks cross-listed on by country and exchange.

Country NYSE LSE OTC Total

Egypt 2 3 5

Nigeria 1 1

South Africa 5 1 70 76

Zambia 3 3

Total 5 4 76 85

(15)

14 Data about transaction prices, bid and ask quotes, number of trades and volumes for a 60-day period surrounding the cross-listing date is used to test the hypothesis regarding liquidity. The 60-day period includes day t= -30 relative to cross listing day t=0 until day t=+30. This methodology is similar to that of the event study as used by Fama et al (1969). Christie and Huang (1994) also used a similar approach to study market liquidity of firms cross listing from the NASDAQ to AMEX and NYSE. Forester et. al. (1998) used the same model to examine market liquidity of cross listing between the Toronto Stock Exchange and NYSE, AMEX and NASDAQ. Like Forester et. al. (1998), this study uses a portfolio approach and averages equal weighted across firms to diversify firm specific and industry noise. Moreover, a narrow window is used to control for non-listing related events. Unlike Forester et. al.(1998), the number of trades is used in place of trade size as an alternative measure of liquidity because of data availability.

The sample data is filtered on a number of criteria. First, all quotes used occur during the regular trading hours, that is between 09.30-16.00. Secondly, to facilitate proper comparison, all quotes are converted into one currency, the US Dollar, using the day close exchange rate. For each firm, the daily transaction prices, bid and ask quotes, number of trades and volumes are computed from pre-listing (day t= -30 to t=-1) to post-listing (day t=0 to day t=20). This is then average across all firms on an equal weighted basis.

Two proxies are used to measure liquidity (Forester 1998). The first is the percentage bid-ask spread:

𝑆𝑆𝑆𝑆% = (𝐴𝐴𝑆𝑆𝐴𝐴 − 𝐵𝐵𝐵𝐵𝐵𝐵)/𝑀𝑀𝐵𝐵𝐵𝐵𝑆𝑆𝑀𝑀𝐵𝐵𝑀𝑀𝑀𝑀 (1)

Where

Midpoint=(Ask+Bid)/2. This is the traditional measure of liquidity. The second proxy is the Percentage Liquidity Premium:

𝐿𝐿𝑆𝑆% = (𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 − 𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀)/𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀

(2)

Where PRICE is the actual transaction price.

To test the difference in liquidity between the pre-listing (day t= -30 to t=-1) and post-listing (day t=0 to day t=30) using the percentage bid-ask spread and percentage liquidity premium, the following regressions are estimated:

(16)

15 𝐿𝐿𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖+ 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖

(3)

𝑆𝑆𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖+ 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖

(4)

Where 𝐵𝐵𝑖𝑖 is a dummy, 0 for -t30 to –t1 and 1 for t0 to t30.

A robust t-statistics is estimated using Newey and West (1987) process to correct for autocorrelation and heteroscedasticity with a paired F – test at 5% level. R2 _{is tested} to check the goodness of fit of the data to the model adjusted for degrees of freedom.

To test hypothesis 1a, regressions (3) and (4) are computed using the cumulative data from the domestic market.

Ho: 𝛼𝛼₁= 𝛽𝛽₁= 𝛾𝛾₁= 𝛿𝛿₁≠ 0,

To test hypothesis 2a, companies will be divided into two subgroups based on their listing type. Exchange trades DRs in one group and OTC in the other. Then, regressions (3) and (4) are computed, using the cumulative data in the domestic market.

Ho: 𝛼𝛼₁= 𝛽𝛽₁= 𝛾𝛾₁= 𝛿𝛿₁≠ 0,

To test hypothesis 3a, companies will be divided into two subgroups based on their listing location. NYSE trades DRs in one group and LSE trades DRs in the other. Then, regressions (3) and (4) are computed, using the cumulative data in the domestic market.

Ho: 𝛼𝛼₁= 𝛽𝛽₁= 𝛾𝛾₁= 𝛿𝛿₁≠ 0,

To test hypothesis 4a, firms with be divided into three subgroups; Financial services sector, Energy, Industrial, IT Telecom & Materials sector and Consumer & Health Care sector. Then, regressions (3) and (4) are computed, using the cumulative data from the domestic market.

Ho: 𝛼𝛼₁= 𝛽𝛽₁= 𝛾𝛾₁= 𝛿𝛿₁≠ 0

Robustness Test of Stock Returns

To test the hypothesis on stock return, the event study adopts the market model to gauge price reaction for a 40-day period surrounding the cross-listing date. Since

(17)

16 cross listing is a firm level decision, the price reaction assessment also starts on a firm level, then aggregates up into portfolio. This method is based on the assumption that markets are efficient and the effect of a cross listing event will be immediately reflected in the stock price. At its core, it argues that, if cross listing will add value for shareholder’s wealth, the announcement of such an event will change investor valuation of the future value of the firm, with price increase as the immediate impact. The process has the following steps:

1. Choose the event date and event window. 2. Select the stocks and exchanges into samples.

3. Select the non-event window to estimate normal returns. 4. Estimate abnormal returns within the event window. 5. Test for the statistical significance of the abnormal returns.

Beaver (1966); Pattel (1976); Fama (1991) and others have used similar methods. This study does however follow the market model approach by Kothari & Warner, 2008. The market model is a univariate linear model of a return of s stock to the return of the market index. This model is chosen because it decreases the variance of abnormal return by eliminating the section of the stock return which is due to the change in the market return. To test if stock returns are abnormal during a cross-listing window, a 20-day event window is selected around the listing date, that is -20 days to +20 days. For each stock, the 180 days before the event window (i.e. 180 days to -20 days) are used as the estimation windows to determine the parameter of the market model. Since the market model will be used to determine the estimated normal returns, using the data prior to the event window ensures that the model is not affected by the cross-listing event. It is however important to note that the best time to estimate the market model for estimated normal returns will be prior to the announcement date of the cross-listing (Foerster & Andrew, 1999). However, limited data is available on the announcement date of the DRs in the sample. Hence, the listing date is used as the effective data. This carries with it the potential for some contamination of the market model by cross-listing announcements made prior to t-20 days.

The market model is stated as follows:

(18)

17 Where 𝑅𝑅𝑖𝑖𝑖𝑖 is the return of stock i defined as (𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖− 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖−1) ÷ 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖−1, 𝑅𝑅𝑚𝑚𝑖𝑖 is the return in excess of risk free rate of market portfolio of

stock i. The All Share Indexes in the Egyptian, Nigerian, South African and Zambian stock exchanges are used as proxies for the market portfolio. The daily return of 10-year government bond is used as the risk free rate.

The abnormal return is derived as follows:

𝐴𝐴𝑅𝑅𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑖𝑖𝑖𝑖− ( 𝛼𝛼�𝑖𝑖+ 𝛽𝛽̂𝑖𝑖𝑅𝑅𝑚𝑚𝑖𝑖 ) (6)

The abnormal return then becomes the error term that is calculated out of sample. The abnormal return is then averaged across all stock events N during the event window t:

𝐴𝐴𝐴𝐴𝑅𝑅 ��_{𝑖𝑖𝑖𝑖} ₌1

𝑁𝑁∑𝑁𝑁𝑖𝑖=1.𝐴𝐴𝑅𝑅𝑖𝑖𝑖𝑖 (7)

The test statistic for significance on daily abnormal returns is as follows: 𝑀𝑀𝐴𝐴𝐴𝐴𝑅𝑅��. = 𝐴𝐴𝑅𝑅𝑖𝑖�_{𝑆𝑆𝑃𝑃}_{𝑖𝑖𝑖𝑖} (8)

𝑆𝑆𝑃𝑃𝑖𝑖𝑖𝑖 is the standard error of the normal market model, estimated during 180days

before the event window. 𝑆𝑆𝑃𝑃𝑖𝑖𝑖𝑖 is defined as var (𝐴𝐴𝑅𝑅𝑖𝑖). Under the null hypothesis, the

test follows the normal distribution, defined as AARit ∼ N (0, σ2 (AARit)).

To test if the abnormal return within the event window is not equal to zero, the cumulative average abnormal return method is used to sum the daily average abnormal return as follows:

𝐶𝐶𝐴𝐴𝐴𝐴𝑅𝑅 (𝑀𝑀1, 𝑀𝑀2) = ∑𝑇𝑇𝑖𝑖=𝑇𝑇2 1.𝐴𝐴𝐴𝐴𝑅𝑅�� 𝑖𝑖 9

Under the null hypothesis, the CAAR follows a normal distribution, defined as: 𝐶𝐶𝐴𝐴𝐴𝐴𝑅𝑅(𝑇𝑇1,𝑇𝑇2)~𝑀𝑀 (0, 𝜎𝜎2 (𝑀𝑀1, 𝑀𝑀2)) 10

If CAAR is skewed or subject to kurtosis, then abnormal return within the event window is not equal to zero.

The test statistic for CAAR significance is as follows: 𝑀𝑀𝐶𝐶𝐴𝐴𝐴𝐴𝑅𝑅 =𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶(𝑇𝑇1,𝑇𝑇2)

�𝜎𝜎2_{𝐶𝐶𝐶𝐶𝐶𝐶}_𝑡𝑡_∗𝐿𝐿 11

Where 𝜎𝜎2_{𝐴𝐴𝐴𝐴𝑅𝑅}

𝑖𝑖 is the variance of an average abnormal return over one period, L is

the period time between 𝑀𝑀1 𝐴𝐴𝑀𝑀𝑀𝑀 𝑀𝑀2. This t-statistics is checked against critical t-value

of a normal distribution to determine if the t-statistics is significantly different from zero. To test hypothesis 1b, the cumulative data from all stock returns in the domestic

(19)

18 market are used. To test hypothesis 2b, companies will be divided into two subgroups based on their listing type. Exchange trades DR in one group and OTC in the other. To test hypothesis 3b, companies will be divided into two subgroups based on their listing location. NYSE trades DRs in one group and LSE trades DRs in the other.

To test hypothesis 4b, firms will be divided into 3 subgroups: Financial services, Energy, Industrial, IT Telecom & Materials and Consumer & Health Care industries. Their CAAR t-statistics is then tested to see if it is significantly different from zero.

4. Results

Preliminary Results

The summary statistics of the data is presented in table 3. Across all stocks, average liquidity spread and premium decreased by 12% and 54% respectively. Between individual exchange platforms, NYSE recorded the highest decrease in both liquidity spread and premium of 35% and 89%; LSE recording a decrease of 10% and 64% follows this respectively. As expected, OTC trade had the least decrease in liquidity spread and premium of 9.19% and 45% respectively.

The results across different industrial sectors are mixed, Financial services had the highest decrease in liquidity spread of 27%, this is followed by the group of Energy, Industrial, IT Telecom & Materials sectors, recording a decrease of 19%, while the Consumer & Health Care sectors all actually saw an average increase in liquidity spread by 15%. On Liquidity premium, the Energy, Industrial, IT Telecom & Materials sectors lead the pack with a decrease of 71%, followed by Financial services with a decrease of 30%. Again, the Consumer & Health Care sectors went opposite with an increase in liquidity premium by 73%.

Price, volume and number of trade reaction appear much more complex. Across all stocks, the average price fell by 0.03% while the volume increased by 0.51% with a net effect of a decrease of 1.04% in market capital. Across exchanges, the results are mixed. On NYSE, the average price fell by 4.15% while the volume increased by 3.62% with a net effect of an increase of 8.47% in market capital. On LSE, the average price increased by 3.31%; the volume increased by 43% with a net effect of a fall of 2.28% in market capital. Regarding OTC trading, the average price increased by 0.29% while the volume decreased by 8.53% with a net effect of a fall of 1.59% in market capital. Along industry sectorial lines, the results are also mixed.

(20)

19 The group of Consumer & Health Care sectors sees an average price fall of 0.91%, the volume decreased by 2.71% with a net effect of a fall of 5.50% in market capital. The group of Energy, Industrial, IT Telecom & Materials sectors follows a similar pattern; the average price fell by 1.21% and the volume decreased by 11% with a net effect of a fall of 0.75% in market capital. The financial services sector is the only industry that does better, with an average price increase of 4.72%, a volume increase of 10% and a net effect of an increase of 3.80% in market capital.

This preliminary results suggests a complex relationship between average liquidity spread and premium on one side and volume, transaction price, return, market capital and number of trades on the other. It also points to the dynamics of platforms where cross-listed stock is traded and which industrial sectors the company belongs. From a domestic market perspective, it appears that apart from the broad lines increase in liquidity, the industrial sector and platform of exchange does determine whether cross listing adds value to domestic shareholders or otherwise. These relationships and interactions will be further explored in the following regression analysis.

Test of Significant of Liquidity Spread and Premium for All Stocks

Table 4 presents the test of significance for all stocks used in the study. The table contains two panels of regressions: Panel A is for liquidity premium and Panel B is for Liquidity spread. Each panel is further split into two regressions. In the first regression in each panel, the first independent variable 𝛼𝛼0 is a constant; the second 𝛼𝛼1 is the

post-listing dummy D multiplied by a constant. The first coefficient, 𝛼𝛼0 represents the

mean pre-listing liquidity premium and spread, that is, the conditional mean cost of trading before cross listing. The second coefficient 𝛼𝛼1, represents the mean

post-listing liquidity premium and spread, which is the conditional mean cost of trading after cross listing. Following the hypothesis above, one will expect liquidity premium and spread to fall after cross listing. Hence, the positive pre-listing and the negative post-listing coefficients would be consistent with hypothesis 1a. However, it should be noted that this overall average increase in liquidity does not account or control for changes in volumes, price and number of trades. This is what the second regression in each panel seeks to do. Given that many studies have shown the dependence of post-listing changes liquidity premium and spread on changes in volumes, price and number of trades, ignoring these independent variables can distort the inferences

(21)

20 one can make from 𝛼𝛼0 and 𝛼𝛼1 (Foerster & Karolyi, 1998). Hence, the question is, do

post-listing liquidity premium and spread still remain significantly negative after controlling the change in volumes, price and the number of trades?

In Panel A of Table 4, the first mean post-listing liquidity premium shows a decrease from 0,00084 to -0,00045. The decrease is significant on the 5% level. However, in the second regression, after controlling for average changes in volumes, price and number of trades, post-listing liquidity premium increased from -0,01057 to 0,00153, which is not significant. In Panel B of Table 4, the first mean post-listing liquidity spread shows a decrease from 0,01118 to -0,00133. The decrease is also significant at the 5% level. In the second regression, after controlling for average changes in volumes, price and number of trades, postlisting liquidity spread increased from -0,00336 to 0,01149; the increase is not significant. Lastly, the f-statistics of both panels A and B are significant at 5% level. This shows that post-listing coefficients 𝛼𝛼1 = 𝛽𝛽1=

𝛾𝛾1= 𝛿𝛿1≠ 0 are significantly different from prelisting coefficients. The null hypothesis of

1a is thus not rejected.

Generally, this result indicates that the average domestic stock liquidity of African firms as measured by premium and bid-ask spread increases significantly after cross listing in LSE, NYSE and OTC cumulatively. However, this increase is dependent on changes in volumes, price and the number of trades after cross listing. This is confirmed by the increase in liquidity premium and spread after controlling these variables. The drawback regarding such an overall average analysis is that, within the general trend, there can be more dramatic changes in domestic liquidity, depending on which exchange the firms choose to cross-list as well as the level of segmentation of the industrial sector from the global market. The test significance will be further refined by reviewing the listing location or type and the industrial sector of the company.

(22)

Table 3: This table shows the cross-sectional summary statistics comparing pre-listing t-30 liquidity premium, liquidity spreads, average transaction price, average return, average volume, average quotes and average numbers of trades to its post-listing equivalent t+30.

(23)

Table 4: This table shows the regression test of significant of Liquidity Spread and Premium of Cross listing on All Stock Exchanges between 1995 and 2015

Model Pre-listing _Intercept Pre-listing _Volume Pre-listing _Price Pre-listing No. of Trades

Post-listing

Intercept Post-listing Volume Post-listing Price

Post-listing Number of

Trades R²

F-stats (P-Value)

A. Liquidity premium regression (LP%)

Co-ef 0.00084 -0.00045 0.07272 4.64185 SE 0.00014 0.00021 P-value 0.00000 0.03530 0.03530 Co-ef -0.01057 0.00000 0.00193 0.00000 0.00153 0.00000 -0.00024 0.00000 0.23835 3.86659 SE 0.01951 0.00000 0.00307 0.00000 0.02831 0.00000 0.00429 0.00000 P-value 0.59032 0.30482 0.53137 0.58295 0.95704 0.93819 0.95590 0.31751 0.00181

All Stock

B. Liquidity Spread regression (SP%)

Co-ef 0.01118 -0.00133 0.21506 15.90579 SE 0.00029 0.00033 P-value 0.00000 0.00019 0.00019 Co-ef -0.00336 0.00000 0.00163 -0.00002 0.01149 0.00000 0.00001 0.00002 0.36934 3.64907 SE 0.02760 0.00000 0.00565 0.00001 0.04759 0.00000 0.00703 0.00001 P-value 0.90344 0.33401 0.77414 0.09465 0.81009 0.42482 0.99831 0.04733 0.00277 𝐿𝐿𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖 + 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖 + 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖 + 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖 𝑆𝑆𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖+ 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖

Where 𝐵𝐵𝑖𝑖 is a dummy, 0 for -t30 to –t1 and 1 for t0 to t30

(24)

Table 5: This table shows the regression test of significant of Liquidity Spread and Premium of Cross listing on NYSE, LSE and OTC between 1995 and 2015

Post-listing

Trades

R² _(P-Value)F-stats

Co-ef 0.00048 -0.00043 0.00099 0.05722 SE 0.00166 0.00178 P-value 0.77529 0.81178 0.81178 Co-ef -0.00787 0.00001 -0.00146 -0.00002 0.02739 -0.00001 0.00268 0.00002 0.08307 1.26108 SE 0.01169 0.00001 0.00583 0.00002 0.05642 0.00001 0.00595 0.00002 P-value 0.50365 0.36538 0.80290 0.28585 0.62935 0.20551 0.65428 0.37855 0.28738

NYSE

Co-ef 0.01876 -0.00658 0.17680 12.48100 SE 0.00160 0.00186 P-value 0.00000 0.00081 0.00081 Co-ef -0.01381 -0.00001 -0.00843 0.00002 0.10891 0.00001 0.01099 -0.00001 0.25703 3.40123 SE 0.01626 0.00001 0.00500 0.00001 0.04812 0.00001 0.00532 0.00002 P-value 0.39933 0.22144 0.09777 0.24672 0.02773 0.37108 0.04387 0.71219 0.00450

Co-ef -0.00138 -0.00089 0.00726 0.42762 SE 0.00108 0.00136 P-value 0.20589 0.51570 0.51570 Co-ef -0.07518 0.00000 0.00075 0.00015 0.06931 0.00000 0.02251 -0.00016 0.06879 1.80018 SE 0.10612 0.00000 0.01680 0.00014 0.11791 0.00000 0.03752 0.00023 P-value 0.48178 0.96257 0.96438 0.28948 0.55915 0.93479 0.55110 0.49441 0.10658

LSE

Co-ef 0.01122 -0.00113 0.00556 0.33155 SE 0.00130 0.00197 P-value 0.00000 0.56694 0.56694 Co-ef 0.08124 0.00000 -0.02350 0.00031 0.00116 0.00000 0.00088 -0.00024 0.09845 2.61557 SE 0.25067 0.00000 0.01612 0.00009 0.25545 0.00000 0.08065 0.00035 P-value 0.74714 0.00040 0.15087 0.00108 0.99641 0.05532 0.99135 0.49580 0.02139

(25)

Post-listing

Trades R²

Co-ef 0.00095 -0.00043 0.05406 3.38208 SE 0.00016 0.00024 P-value 0.00000 0.07094 0.07094 Co-ef -0.01615 0.00000 -0.00073 0.00000 0.02417 0.00000 0.00338 0.00000 0.19809 1.89715 SE 0.01771 0.00000 0.00248 0.00000 0.02384 0.00000 0.00365 0.00000 P-value 0.36602 0.06669 0.76875 0.76997 0.31522 0.97305 0.35872 0.45084 0.08839

OTC

Co-ef 0.01065 -0.00098 0.14244 9.65115 SE 0.00027 0.00032 P-value 0.00000 0.00291 0.00291 Co-ef -0.00614 0.00000 0.00430 -0.00001 -0.00250 0.00000 -0.00215 0.00001 0.36794 3.42415 SE 0.01833 0.00000 0.00426 0.00001 0.03333 0.00000 0.00506 0.00001 P-value 0.73876 0.50315 0.31780 0.07104 0.94049 0.74161 0.67233 0.01992 0.00430 𝐿𝐿𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖 + 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖 + 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖 + 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖 𝑆𝑆𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖+ 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖

(26)

Table 6: This table shows the regression test of significant of Liquidity Spread and Premium of Cross listing by industry between 1995 and 2015

Post-listing

Trades R²

F-stats (P-Value) A. Liquidity premium regression (LP%)

Co-ef 0.00006 0.00005 0.00030 0.01789 SE 0.00021 0.00035 P-value 0.75735 0.89405 0.89405 Co-ef -0.02483 0.00000 -0.00104 0.00000 0.02763 0.00000 0.00576 -0.00001 0.20659 2.61467 SE 0.02823 0.00000 0.00219 0.00000 0.03118 0.00000 0.00515 0.00000 P-value 0.38306 0.70684 0.63672 0.05132 0.37941 0.54937 0.26796 0.08046 0.02142

Consumer &

Health Care

Co-ef 0.00893 0.00135 0.14282 9.95132 SE 0.00024 0.00043 P-value 0.00000 0.00253 0.00253 Co-ef 0.01617 0.00000 -0.00169 0.00000 0.00520 0.00000 0.00009 0.00001 0.22299 3.23633 SE 0.03824 0.00000 0.00185 0.00000 0.03990 0.00000 0.00669 0.00000 P-value 0.67418 0.21312 0.36488 0.14743 0.89677 0.48476 0.98899 0.04672 0.00622

Co-ef 0.00123 -0.00088 0.07066 4.43877 SE 0.00034 0.00042 P-value 0.00063 0.03939 0.03939 Co-ef -0.06688 0.00000 0.00436 -0.00001 0.03978 0.00000 0.00466 0.00001 0.28952 2.44802 SE 0.03278 0.00000 0.00239 0.00001 0.03793 0.00000 0.00510 0.00001 P-value 0.04631 0.49201 0.07406 0.05720 0.29904 0.12557 0.36482 0.05437 0.02986 Energy, Industrial, IT

Telecom & Materials B. Liquidity Spread regression (SP%)

Co-ef 0.01355 -0.00253 0.20749 15.18519 SE 0.00057 0.00065 P-value 0.00000 0.00025 0.00025 Co-ef -0.01354 0.00000 0.00302 -0.00001 0.01414 0.00000 0.00062 0.00000 0.41481 3.76625 SE 0.04427 0.00000 0.00548 0.00001 0.06196 0.00001 0.00805 0.00001 P-value 0.76097 0.36501 0.58372 0.19852 0.82034 0.42548 0.93906 0.66450 _0.00220

(27)

Post-listing

Trades R²

Co-ef 0.00118 -0.00041 0.01398 0.81897 SE 0.00041 0.00045 P-value 0.00542 0.36916 0.36916 Co-ef 0.01466 0.00000 -0.00364 0.00000 0.00371 0.00000 0.00092 0.00000 0.06144 0.79789 SE 0.01200 0.00000 0.00396 0.00000 0.02166 0.00000 0.00463 0.00000 P-value 0.22726 0.54870 0.36197 0.44253 0.86447 0.35453 0.84240 0.74126 0.59254

Financials

Services

Co-ef 0.01014 -0.00281 0.26785 21.13408 SE 0.00056 0.00061 P-value 0.00000 0.00002 0.00002 Co-ef 0.04359 0.00000 -0.01775 0.00000 0.04416 0.00000 0.01045 0.00000 0.44157 5.97908 SE 0.01988 0.00000 0.00609 0.00000 0.03384 0.00000 0.00720 0.00001 P-value 0.03274 0.37617 0.00523 0.42291 0.19757 0.21399 0.15271 0.41766 0.00004 𝐿𝐿𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖 + 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖 + 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖 + 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖 𝑆𝑆𝑆𝑆%𝑖𝑖𝑖𝑖 = 𝛼𝛼0+ 𝛽𝛽0𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾0𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿0𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖+ 𝐵𝐵𝑖𝑖(𝛼𝛼1+ 𝛽𝛽1𝑉𝑉𝑀𝑀𝑉𝑉𝑉𝑉𝑉𝑉𝑃𝑃 𝑖𝑖𝑖𝑖+ 𝛾𝛾1𝐴𝐴𝐴𝐴𝑃𝑃𝑃𝑃𝐴𝐴𝐴𝐴𝑃𝑃 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑇𝑇𝐴𝐴𝑃𝑃𝑀𝑀𝑃𝑃𝑀𝑀𝑀𝑀 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖+ 𝛿𝛿1𝑀𝑀𝑉𝑉𝑉𝑉𝑁𝑁𝑃𝑃𝑃𝑃 𝑀𝑀𝑜𝑜 𝑀𝑀𝑃𝑃𝐴𝐴𝑀𝑀𝑃𝑃𝑇𝑇𝑖𝑖𝑖𝑖) + 𝜀𝜀𝑖𝑖𝑖𝑖

(28)

27

Test of Significant Difference in Liquidity Spread and Premium for NYSE, LSE and OTC Traded Stocks

Table 5 presents the test of significance between the different platforms of trading. Each trading platform table contains two panels of regressions: Panel A for liquidity premium and Panel B for Liquidity spread. As set out in the analysis above, each panel is further split into 2 regressions. In the first regression in each panel, the first independent variable 𝛼𝛼0 is a constant; the second 𝛼𝛼1 is the post-listing dummy D

multiplied by a constant. The first coefficient, 𝛼𝛼0 represents the mean pre-listing

liquidity premium and spread, that is, the conditional mean cost of trading before cross listing. The second coefficient 𝛼𝛼1 represents the mean post-listing liquidity

premium and spread, which is the conditional mean cost of trading after cross listing.

Following the hypothesis 2a, it is expected that exchange traded cross-listed stocks (i.e. DRs Level ll and lll on NYSE & LSE), which have a greater informational symmetry and investor protection than OTC, will report higher and significant cross-listing liquidity than OTC stocks. Moreover, with a domestic equity market capitalization of USD 19.4 trillion compared to USD 4 trillion on LSE (World Federation of Exchanges members, 2015), one will expect NYSE traded cross-listed stocks to have a higher liquidity than LSE traded cross-listed stocks (hypothesis 3a). These assumptions are consistent with the summary statistics presented in Table 3, which reported that NYSE had the highest decrease in both liquidity spread and premium of 35% and 89%; LSE recorded a decrease of 10% and 64% respectively. As expected, OTC trade had the least average decrease in liquidity spread and premium of 9.19% and 45% respectively. However, are these decreases significant? On liquidity premium, NYSE‘s post-listing liquidity premium shows a decrease from 0,00048 to -0,00043 (not significant), LSE marginally increases from -0,00138 to -0,00089 (not significant) and OTC decreases from 0,00095 to -0,00043 (significant at 10%). After controlling the changes in volumes, price and the number of trades after cross listing, liquidity premium of NYSE traded stocks increased from -0,00787 to 0,02739 (not significant), LSE traded stocks increased from -0,07518 to 0,06931 (not significant) and OTC traded stocks increased from -0,01615 to 0,02417 (not significant). Three key observations are essential here: First, exchange traded stocks did not perform better than OTC. NYSE and OTC traded stocks both reported a decrease in premium, while LSE’s premium increased. Second, contrary to expectation, only the post-listing

(29)

28 liquidity premium of OTC traded firms were significant. Lastly, only the f-statistics of OTC post-listing liquidity premium is significant at 5% level.

On liquidity spread, NYSE average post-listing liquidity spread shows a decrease from 0,01876 to -0,00658 (significant at 5%), while LSE marginally decreased from 0,01122 to - 0,00113 (not significant) and OTC decreased from 0,01065 to -0,00098 (significant at 5%). After controlling for the changes in volumes, price and number of trades after cross listing, liquidity spread of NYSE traded stocks increased from -0,01381 to 0,10891(significant at 10%), LSE traded stocks decreased from 0,08124 to 0,00116 (not significant) and OTC traded stocks increased from -0,00614 to -0,00250 (not significant). Three key observations are essential here: First, exchange-traded stocks did not perform better than OTC. NYSE and LSE traded stocks both reported a higher decrease in spread than OTC. However, the coefficient of LSE was not significant, while the coefficient of OTC was significant at 5%. Second, after controlling for the changes in volumes, price and number of trades, only the post-listing liquidity spread of NYSE traded firms increased significantly at 10%. Lastly, the f-statistics of NYSE, LSE and OTC on Panel B’s post-listing liquidity spreads were all significant at 5% level. This shows that post-listing coefficients 𝛼𝛼1= 𝛽𝛽1= 𝛾𝛾1= 𝛿𝛿1≠ 0 are significantly different from

prelisting coefficients.

Overall, this result indicates that the average domestic stock liquidity of African firms as measured by premium and bid-ask spread trading on an exchange, such LSE and NYSE, will not necessary translate into higher liquidity in domestic cost trading than OTC trades stock. However, it does suggest that for exchange traded stocks, NYSE traded stocks do report lower domestic liquidity premiums and spreads than LSE traded stocks. It follows that the null hypothesis 2a is rejected and the null hypothesis 3a is not rejected.

Test of Significant Difference in Liquidity Spread and Premium by Industry

Table 6 presents the test of significance between the group of the Energy, Industrial, IT Telecom & Materials sectors, the group Consumer & Health Care and the Financials services sector. Each sector’s table contains two panels of regressions: Panel A is for liquidity premium and Panel B is or Liquidity spread. Each panel is further split into two regressions. In the first regression of each panel, the first independent variable 𝛼𝛼0 is a constant; the second variable 𝛼𝛼1 is the post-listing dummy D multiplied

(30)

29 by a constant. The first coefficient 𝛼𝛼0 represents the mean pre-listing liquidity

premium and spread, that is, the conditional mean cost of trading before cross listing. The second coefficient 𝛼𝛼1, represents the mean post-listing liquidity premium

and spread, which is the conditional mean cost of trading after cross listing.

Following hypothesis 4a, more globally integrated Financial Services firms will experience lower and less significant cross-listing liquidity than firms in the Energy, Industrial, IT Telecom & Materials sectors and firms in the Consumer & Health Care industries. The summary statistics presented in Table 3 are not completely consistent with the assumption. As reported in Table 3, the Financial services had the highest decrease in liquidity spread of 27%; this is followed by the Energy, Industrial, IT Telecom & Materials sectors recording a decrease of 19%, while the Consumer & Health Care sectors saw an increase in liquidity spread by 15%. On Liquidity premium, Energy, Industrial, IT Telecom & Materials sectors led the pack with a decrease of 71%, followed by the Financials Services sector with a decrease of 30%. The Consumer & Health Care sectors goes opposite with an increase in liquidity premium by 73%.

On liquidity premium, Consumer & Health Care sectors’ post-listing liquidity premium decreased from 0,00006 to 0,00005 (not significant); The group of the Energy, Industrial, IT Telecom & Materials sectors decreased from 0,00123 to -0,00088 (significant at 5%) and the Financials Services sector decreased from 0,00118 to -0,00041 (not significant). After controlling the changes in volumes, price and the number of trades after cross listing, the liquidity premium of the stock of the Consumer & Health Care sectors increased from -0,02483 to 0,02763 (not significant); Stocks of the Energy, Industrial, IT Telecom & Materials sectors increased from -0,06688 to 0,03978 (not significant), while stocks of the Financials Services sector decreased again from 0,01466 to 0,00371 (not significant).

Two key observations are essential here: First, all industrial sectors reported a decreasing premium, with Financials services performing better than the Consumer & Health Care even after controlling the changes in volumes, price and number of trades. This is inconsistent with the summary statistics, which showed an increase in spread for the Consumer & Health Care sectors. Second, the f-statistics of all sectors

(31)

30 with exception of the Financials Services in Panel A’s post-listing liquidity premium are significant. This shows their post-listing coefficients 𝛼𝛼1= 𝛽𝛽1= 𝛾𝛾1 = 𝛿𝛿1 ≠ 0 are

significantly different from pre-listing coefficients.

On liquidity spread, the Consumer & Health Care sectors’ post-listing liquidity spread decreased from 0,00893 to 0,00135 (significant at 5%); the Energy, Industrial, IT Telecom & Materials sectors decreased from 0,01355 to -0,00253 (significant at 5%), while the Financials services decreased from 0,01014 to -0,00281 (significant at 5%). After controlling the changes in volumes, price and the number of trades after cross listing, the liquidity spread of the Consumer & Health Care sectors still decreased from 0,01617 to 0,00520 (not significant), while the Energy, Industrial, IT Telecom & Materials sectors increased from -0,01354 to 0,01414 (not significant) and the Financials services sector increased from 0,04359 to 0,04416 (not significant).

Two key observations are essential: First, all industrial sectors reported a decrease in spread, with the Financial Services sector reported the narrowest increase in spread, after controlling the changes in volumes, price and number of trades. This is inconsistent with the summary statistics, which show an increase in spread for the Consumer & Health Care sectors. Second, the f-statistics of all sectors in Panel B post-listing liquidity spread are all significant at 5% level. This shows that the post-post-listing coefficients 𝛼𝛼1= 𝛽𝛽1= 𝛾𝛾1 = 𝛿𝛿1≠ 0 are significantly different from pre-listing coefficients.

The results suggests that African firms in the Financial Services sector may not be well integrated into the global financial system, as one would have expected. That is, the financial services industry would be equally segmented like other industries from the global equity market. This implies that the type of industry is not an important indicator for domestic liquidity premiums and spread in cross listing. Hence, the null hypothesis of 4a is rejected.

Test of Significance of Abnormal Returns

Test of Significance of All Stocks

Table 7 and figure 1 show the cross-sectional cumulative abnormal returns (CAAR) for the various sub-groups in the event window t-20 to t+20 and the significance test for all stocks. They show a pre-listing run-down of prices and consequently CAAR from t-20 to t+1. This is followed by a post-listing price and CAAR run-up between t+2

(32)

31 to t+9 then another decline thereafter. Furthermore, the distribution of CAAR shows a negative mean of -0,0288, skewness of -0,1373 and a platykurtic Kurtosis of -0,8452. This suggests that CAAR exhibits negative abnormal performance and a non-normal distribution. More significantly, there is an indication that domestic investors underreacted to the prospect of prospective gains before the listing date. This underreaction is corrected a few days after listing but not large enough to make a positive CAAR. As Table 7 shows, CAAR is positive and significant only in 1 of 5 windows and negative in all others. Hence, the bigger picture within the event window suggests that cross listing generally does not lead to positive and significant CAAR. Hence, hypothesis 1b is rejected.

Table 7: This table shows the cross-sectional cumulative abnormal returns (CAAR) in the event window t-20 to t+t-20 and significant test for All Stocks, NYSE, LSE and OTC.

(-20 to -10) (-9 to -2) (-1 to 1) (+2 to +9) to 20) All Stock -0.0159 -0.0277 -0.0226 0.0431 -0.0192 NYSE 0.0304 0.0264 -0.0109 -0.0575 -0.0309 LSE 0.0336 -0.0204 -0.0465 0.0493 -0.1563 OTC -0.0204 -0.0316 -0.0227 0.0498 -0.0147 CAAR Exchange/Event Window

Figure 1: This figure compares cumulative abnormal returns (CAAR) in the event window t-20 to t+20 for All Stocks, NYSE, LSE and OTC

-0.2000 -0.1500 -0.1000 -0.0500 0.0000 0.0500 0.1000 (-20 to -10) (-9 to -2) (-1 to 1) (+2 to +9) (+10 to 20)

CAAR t-20 to t+20 Per Exchange

All Stock NYSE LSE hTC

(33)

32

Test of Significance for Trading Platforms

Table 7 and figure 1 show the cross-sectional cumulative abnormal returns (CAAR) for the various sub-groups in the event window t-20 to t+20 and the significant test across platforms of exchange. With exception of NYSE, all trading platforms show a general pre-listing run-down of prices and consequently CAAR from t-20 to t+1. This is followed by a post-listing price and CAAR run-up between t+2 to t+9 and another decline thereafter. On the contrary, for NYSE traded stocks, there is an initial investor pre-listing overreaction from t-20 to t-2, with accompanied price increase and positive though insignificant CAAR, this followed by a decline in prices and CAAR from t-1 to t+20. NYSE has the highest mean CAAR, followed by LSE and OTC. Moreover, OTC showed the most negative skewness of CAAR distribution of -0,3532, followed by NYSE of -0,1857 and LSE of 0,0389. All trade platforms exhibit a platykurtic distribution4_.

The results suggest the following: First, the skewness and platykurtic distribution suggests that CAAR exhibits a non-zero abnormal performance and are not normally distributed. Secondly, it indicates that exchanges traded stocks perform better than OTC stocks as shown by their relatively higher mean CAAR (though insignificant) and positive skewness. The result is consistent with the bonding hypothesis, that suggested that bonding to higher levels of reporting and minority shareholder protection in an exchange traded cross-listing would lead to higher stock price responds. Hence, hypothesis 2b is not rejected. Lastly, the difference between LSE and NYSE is not clear. While NYSE shows the highest mean CAAR, it is negatively skewed in distribution, while LSE is highly positively skewed in distribution. Based on these considerations and the limited data from these exchanges used in the study, hypothesis 3a will not be rejected.

(34)

33 Table 8: This table shows the cross-sectional cumulative abnormal returns (CAAR) in the event window t-20 to t+t-20 and significant test for All Stocks, Consumer & Health Care sector, Energy, Industrial, IT Telecom & Materials sector and Financial Services sector.

(-20 to -10) (-9 to -2) (-1 to 1) (+2 to +9)

to 20)

All Stock -0.0159 -0.0277 -0.0226 0.0431 -0.0192

Consumer & Health Care -0.0484 -0.0456 -0.0256 0.0576 -0.0201 Energy, Industrial, IT Telecom

& Materials -0.0217 -0.0294 -0.0271 0.0388 -0.0304

Financials Services 0.0250 0.0028 -0.0095 0.0355 -0.0019

CAAR Industry/Event Window

Table 2: This figure compares cumulative abnormal returns (CAAR) in the event window t-20 to t+20 for All Stocks,Consumer & Health Care sector, Energy, Industrial, IT Telecom & Materials sectors and

Financial Services sector.

-0.0600 -0.0400 -0.02000.0000 0.0200 0.0400 0.0600 0.0800 (-20 to -10) (-9 to -2) (-1 to 1) (+2 to +9) (+10 to 20)

CAAR t-20 to t+20 Per Industry

All Stock

Consumer & Iealth Care

Energy, Industrial, IT Telecom & Materials Cinancials Services

Test of Significance for Industrial Sectors

Table 8 and figure 2 show the cross-sectional cumulative abnormal returns (CAAR) for the various sub-groups in the event window t-20 to t+20 and significant test across the industrial sectors. The Consumer & Health Care sectors and the Energy, Industrial, IT Telecom & Materials sectors show a general pre-listing run-down of prices and consequently CAAR from t-20 to t+1. This is followed by a post-listing price and CAAR run-up between t+2 to t+9 (significant at 5% level) then another decline thereafter.