1
Effects of US companies cross-listing on Börse Frankfurt
by Davide Lodigiani
S2759284
3
Effects of US companies cross-listing on Börse Frankfurt
THESIS
MSC FINANCE
FACULTY OF ECONOMICS AND BUSINESS
UNIVERSITY OF GRONINGEN
by Davide Lodigiani
(Student number S-2759284)
Abstract
I study the financial effect of the US companies which cross-list on the Börse
Frankfurt. The sample consists in monthly observations of bonds issued by US
companies on the New York Stock Exchange and on the Börse Frankfurt between
2000 and 2015.
I find a significant cost of capital reduction of 22 basis points when the bonds are
cross-listed on the Börse Frankfurt.
I provide novel evidence of the financial effect of US companies cross-listing bonds
on a particular European exchange market.
Keywords: Bond spreads, corporate bonds, cross-listing
Supervisor: Dr. Bert Kramer
4
Table of contents
1. Introduction………...5
2. Theoretical Research Framework...………....6
2.1 Literature Review...6
2.2 Hypothesis Development...8
3. Data and Descriptive Statistics………..9
3.1 Sample selection……….9
3.2 Bond yield data………..11
3.3 Firm specific variables and macroeconomic variables...11
3.4 Bond specific variables...12
5
1. Introduction
This study analyzes the impact of cross-listing on the firm’s cost of capital measured by the bond yield, applied to a list of US companies belonging to the S&P500 index which choose to list different bonds on the local market (NYSE) and on a foreign market (Börse Frankfurt , FF). The evaluation of the bond yield during the lifetime regards coupon, maturity, market price, liquidity spread and the issuer: these characteristics can be different for each bond issued. The coupon is the interest rate paid as a percentage of the par value paid at fixed intervals; the maturity date is the date when the investor will receive the principal; the market price is the bond price at the end of the month; the liquidity spread is the cost that the investor face when trading the security; the issuer is related to its ability to refund the obligations and it is measured by the credit rating of the company.
I refer to yield to maturity, or redemption yield as the internal rate of return of the investment if all payments are made as scheduled and the bond is held till maturity, as bond yield because it smooths the variables of financial instruments. This return allows to make a clear comparison across different bond characteristics.
The purpose of this study is to determine the financing effect of US companies cross-listing bonds in the German market.
The research analyzes the bond yield spreads between US companies issuing bonds on the local market, the NYSE, and bonds on the Börse Frankfurt.
I use the bond yield spread, which is the bond yield to maturity less the prevailing risk free interest rate with the same currency of the bond’s cash flows, to measure yield differences in the
securities.
I focus on the New York Stock Exchange (NYSE) and the Deutsche Börse, which are the most appealing exchange markets for worldwide investors and investment funds, because these markets are well known for high regulatory regimes and require an increase of information disclosure to foster small shareholder rights.
The choice of Deutsche Börse stresses a reverted trend which fosters US companies to favor cross-listing in a specific European market, with an increased number of bonds after 2006.
I find strong evidence that US companies cross-listing bonds on the foreign market have a reduction of cost of capital, measured with bond yield spreads, of 22 basis points on average during the time period 2000-2015.
The structure of the paper is the following: I introduce the previous studies about the choices for cross-listing. Section III presents the hypothesis, Section IV describes the data set and the
6
2. Theoretical Research Framework
2.1 Literature Review
Since the last part of the 20th century, several studies about cross-listing have been done to understand the reasons why companies want to list in a foreign market.
The theoretical and empirical contributions explain the targets to be achieved by this action: the range of possibilities varies from corporate governance motivations, the overcome of market segmentation and geographical barriers, to lowering the cost of capital.
The theoretical and empirical contributions concern the issue of stock and bonds in foreign markets, denoting a positive increase in firm value and a decrease in cost of capital, with a magnitude related to firm’s specific variables.
Errunza and Losque (1985) make use of the capital asset pricing model to test the return of two different stocks to explain the issue of market segmentation.
The eligible securities, which belong to a worldwide portfolio and are part of completely
integrated markets, are priced depending on the security market line while the ineligible securities have a restricted access to all possible investors and they require a super risk-premium due to the impossibility to diversify the risk globally.
The effect on the firm is the increasing cost of capital deriving from market segmentation because it’s not possible to apply the standard CAPM for ineligible securities.
Cross-listing in the foreign market should help the firms to overcome market segmentation and to achieve a larger investor base, which will include the cross-listed corporate bonds in their portfolio because of the availability of the security on the Börse Frankfurt; the result is the inclusion of the corporate bond in the security market line, which for definition does not require a super risk-premium and it should lower the bond yields.
Merton (1987) develops a model of capital market equilibrium with incomplete information to explain why an investor should require an abnormal return to include a not known instrument in its portfolio.
The model outlines a discrepancy in the mean-variance efficiency related to a lack of information which allocates such a security out of the Capital Asset Pricing Model; this bias is exactly a
confirmation that the security doesn’t belong to the Security Market Line.
The more the information about a firm and its security, the less is the risk-premium required by the investors because of less “shadow costs” in gathering information.
This issue is connected with a small size of the investor base, which also increases cost of capital: a security held by few investors means low liquidity and therefore it requires a higher price to reflect the trading difficulty, and also the investors experience a lack of full diversification which demands in return a higher yield associated to the idiosyncratic risk.
Coffee (2002) finds that technology and globalization are essential in gaining both access to information and global capital flows.
These are the reasons that in the late 20th century firms can choose where to issue securities and raise capital.
Cross-listing abroad is related to transparency and liquidity/trading volumes, which are
7
security can be eligible to new investors, achieving market integration, who consider a high level of disclosure as a form of protection towards small investors.
This increased investor base, connected with a decrease of asymmetric information, shows that the firm’s risk is spread across more investors it creates a diversification effect which entails a reduction in cost of capital for the cross-listed company.
Specifically, the corporate bond which is cross-listed in the foreign market releases information about the issuing company and it gets the attention of the investors which previously did not consider the chance to invest on the US exchange market. This outcome should allow the companies to reduce their cost of capital by cross-listing.
Pagano et al. (2002) report how foreign listing is useful to reduce cost of capital and enhance the company’s reputation by accessing foreign capital markets.
They find different geographical trends in cross-listing. During 1990-1998 European companies choosing the US were basically high-tech focused, the reason was connected to raise capital to undertake investment programs.
The contribution of the paper is the geographical pattern according to two variables: common language and distance foster the cross-listing activity.
The US market attracted foreign companies over a decade while the European markets had their peak in 1991, then they declined till the end of the century: the attractiveness of NYSE gained popularity.
US companies cross-listing during the first part of the 21th century show the willingness to
continue the approach of listing abroad their securities to increase the company’s reputation and having a reduction of bond yields offered.
Ball et al. (2013) control for changes in cost of debt after a foreign firm cross-lists stocks in US. The result is a cut in expenses of 45 basis points on bonds yield offered to the investors,
determined by the bonding hypothesis due to the migration towards a stricter regulation imposed by US exchange markets, implying the decreasing effect of high asymmetric information, which stands for agency costs.
Massa and Zaldokas (2011) link the bond yield reduction of international bond issues with a positive firm’s recognition in the cross-listed exchange market.
In contrast to previous studies, cross-listing within the period from 1998 to 2006 is deepened from the side of American companies, where the aim is not related to the bonding hypothesis.
Indeed the authors of the paper find that “investors care about the international diversification of their portfolio”, which “are less sensitive to US shocks, but on the other hand, are less effective monitors than more proximate US investors”; when this tradeoff is positive the bond yield required is lower.
It’s noteworthy that the fundamental issue of the firm is a high reputation which represents a warranty for the European investor in this specific example.
The results confirm that US companies with a high worldwide recognition can lower cost of capital through an international issues of bonds rather than an issue in the domestic market.
Chambers et al. (2015) find a reduction of the cost of capital related to bonds for US companies listed on the London Stock Exchange equal to 15-22 basis points during the period 1870 – 1913. This evidence is attributable to a reduction in market fragmentation and an increase in
8
Indeed the bond yield spreads are inconstant over the sample period due to the level of market segmentation, which changes as time progresses in line with technological improvements; and the cost of information is significant in determining the magnitude of cost of capital reduction, the proof is tested by comparing different geographical areas of the US with the telegraph costs borne by English investors in order to gather information about US companies.
Levine and Schmukler (2006) focus on companies that internationalize and study the effect of the liquidity on the securities on the local market.
The finding is a reduction of liquidity on the local market due to migration of the investor base on the international market, which allows the investor base to lower the country-specific risk.
The effect on bonds on the local market is a larger bid-ask spread, which is a direct cost increase for investors; the outcome is an increase of the bond yield in the local market.
This study shows the liquidity spread as an important variable to determine the changes in the bond yields, I expect to find a lower bid-ask spread in the sample of bonds listed on the Börse Frankfurt compared to the bonds issued by US companies on the New York Stock Exchange.
Chen et al. (2009) test the liquidity to explain the bond yield spreads.
The finding is that the liquidity spread is included in the corporate bond price by investors, which require a higher return for more illiquid securities.
The liquidity spread affects the market price of the security and therefore the company issuing bonds should take this variable into account.
I choose to include the liquidity spread as an independent variable in the regression model to directly test its significance on the bond yield spreads.
2.2 Hypotheses Development
Given the previous empirical findings, I verify the following hypothesis:
(H1): The effect of cross-listing reduces bond yield spread for the security issued on the foreign
market, allowing US companies to decrease their cost of capital.
(H2): Dual listed US companies should experience a worse liquidity on the local market than not
9
3. Data and Descriptive Statistics
3.1 Sample Selection
I manually check US companies listed on S&P500 which choose to issue at least both one bond on NYSE and one bond on Börse Frankfurt; I require that these bonds are issued in only one specific market, not counting bonds traded simultaneously on two different markets.
I exclude bonds listed on other markets than Börse Frankfurt because the sample provided would not be quantitatively significant.
I include also bonds issued before January 2000, if they are still alive at that date, to increase the number of observations and have a clear pattern of bond yields.
The sample is an unbalanced cross-section panel data with monthly bond yield spreads of 130 bonds listed on New York Stock Exchange and 86 bonds listed on Börse Frankfurt issued by 52 US companies.
The data set contains monthly data from January 2000 to December 2015 obtained from Thomson Reuters Datastream.
I use the zero coupon and straight, both fix and variable, bond type but I exclude redeemable as well as convertible bonds in order to have a linear payoff of the securities in order to better evaluate the sample.
In line with Ball et al. (2013), if during the same year a company issues multiple bonds, I select only the one with the largest amount.
I collect 16 years of monthly returns instead of daily to reduce the random fluctuations and estimation errors, which is consistent with trading with less noise.
I analyze the companies’ issuing bonds trend over time (from 1989 to 2015, according to data availability).
US firms decrease the amount of bonds issued on the foreign market during the crisis periods, the
Dot Com Bubble (2001), the Global Financial Crisis (2007-2008), and the European Sovereign Debt Crisis (2010); a natural explanation is the presence of negative market conditions during crises and
the required bond yield is higher as gauged in this study.
Cross-listing on Börse Frankfurt gets relevant after 2012, with a peak in the year 2014 due to the choice to list more securities in the foreign market than in the local one, but besides this there is not a clear pattern in the distribution between markets.
10
Figure 1: Bond issue frequency distribution on New York Stock Exchange and Börse Frankfurt (FF)
Figure 2 shows the yield spreads of bonds issued by US companies on the New York Stock Exchange and on Börse Frankfurt; the spreads are lower on the foreign market but for 7 months during the sample period (192 monthly observations), which denotes cross-listing as a persistent source of cost of capital reduction over time.
Figure 2: New York and Frankfurt spreads for bonds traded on NYSE and FF
11
I collect from Datastream the bid and ask quote of every bond to calculate the market liquidity percentage as the difference between the bid and ask price over the ask price. The dataset available on Datastream is restricted from July 2009 to December 2015.
In line with the finding of Chen et al. (2009) about the importance of the liquidity to explain bond yield spreads, I calculate the average bid-ask spread during the sample period on NYSE, equal to 0.589%, and on FF, which is 0.547%; I choose to include this variable in the regression because the liquidity premium is higher in the local market than in the foreign market and it could be the reason of different bond yields.
The subsample to verify the second hypothesis counts 130 bonds listed on NYSE by 52 US dual-listed companies and I manually check US companies dual-listed on S&P500 which do not issue bonds in the German markets, in this way I match 272 bonds issued on the local market by 120 US
companies not cross-listing on Börse Frankfurt.
The trading migration study is from January 2000 to December 2015 despite the lack of the liquidity spread observations; the overall aim is to test the pattern of the liquidity spread in the same exchange market for bonds with a precise characteristic (cross-listed vs. not cross-listed) while I can observe differences in both bond and firm attributes.
Finally, I convert the bond spreads non US Dollars denominated in US Dollars with the respective exchange conversion rate.
3.2 Bond Yield Data
The implied cost of capital for the companies issuing bonds is the Bond Yield Spread, which I calculate as the yield to maturity of the bond minus the risk free interest rate matching the same maturity: this is the dependent variable.
I choose to match the risk free interest rate with the currency of the cash flows provided by the bonds: I use the US treasury rate as a proxy for the risk free rate when the bonds are denominated in US$ but I use specific government bond yields of the same currency of the bond to find the yield spreads for the bond with a different currency.
I convert the obtained yield spreads in US$ to make a clear comparison.
There are neither defaults nor companies in financial distress during the period observed. The dataset of bond and firm characteristics is obtained from Datastream.
3.3 Firm Specific Variables and Macroeconomic Variables
Consistent with the study of Ball et al. (2013), I use the following firm-specific independent variables, all expressed in US$:
- the Market to Book Value is the ratio of market value of equity to book value of equity.
It is connected to growth options, the probability of high future cash flows; a high market to book value could be perceived as risky by the investors.
12
- the Total Assets, expressed US$ million, reflects the tangible resources of the firms and the investor base recognizes it as proxy for good reputation.
- the Leverage is the ratio of long-term debt divided by total assets, which relates the firm to the possibility of default.
The effect should be an increase in bond yield with higher leverage due to more risk of bankruptcy.
- the Inflation, accounted as the monthly percentage change in the consumer price index, to include country specific macroeconomic factors not depending on firm characteristics, which should reflect differences between US and Germany.
- the Gross Domestic Product growth rate (GDP), measures how fast an economy is growing and it controls for differences in the financial development between USA and Germany.
I draw a dummy variable for every variable listed above, which assumes the regular value if the bonds are listed on the Börse Frankfurt but equal to zero if the bonds are listed on the New York Stock Exchange: this procedure allows me to control the significance of the firm-specific and macroeconomic variables when a company cross-lists.
I do not include the Tangibility variable because, as mentioned by Ball et al. (2013): “it is highly correlated with firm size” and this variable is statistically insignificant in the full model.
3.4 Bond Specific Variables
I draw a cross-listing variable, the Cross-listing independent variable, which assumes value of the unity if the bond is listed on the Börse Frankfurt, and zero otherwise.
The aim of the Cross-listing variable is to test the significance of the cross-listing factor related to the bond yield spread.
I introduce the dummy variable Liquidity factor, which is equal to the unity if the bonds are issued on the New York Stock Exchange by US companies not cross-listing on the Börse Frankfurt, and zero if the US companies cross-list bonds.
The Liquidity factor variable allows to determine the effect of the liquidity spreads on the bond yield spreads.
In line with the study of Ball et al. (2013), I control for bond specific features, which should help in determining the bond yield:
- the Bond Size is equal to the principal amount at issuance denominated in US$ million.
The larger the amount the higher is the probability of default of the firm; on the other hand this characteristic allows for a better liquidity which lowers the bond yield.
- the Bond Maturity calculated as the remaining months to maturity; it is proportional to risk, which increases the investor required return.
- the Investment Grade dummy, which assumes the value of 1 if the credit rating from Standard &
Poor’s is defined as BBB- or higher, otherwise it is 0, is related to the probability of default of the
issuer.
13
- the Previous Bond Issue dummy is an indicator to gauge firm reputation in the market, it is equal to 1 if the company has issued other bonds in the same market.
This concept is related to reputation; investors which in the past held the security should require a lower spread due to less information asymmetry about a specific company.
- the Liquidity Spread is the bid price minus the ask price, divided by the ask price; when an asset is illiquid the investor pays more because there is a spread in the trading book.
I adopt the procedure to create dummy variables used for firm-specific and macroeconomic variables, to measure the significance of the bond specific variables in the changes of the bond yield spreads paid on the Börse Frankfurt.
I do not control for the currency of the securities because only 8 bonds from the dataset are not denominated in American Dollars, this is a suggestion that US companies do not cross-list for hedging their currency exposure.
3.5 Descriptive Statistics
I present the summary statistics of the dependent and independent variables.
Panel A of Table 1 shows the mean, median, maximum, minimum and the standard deviation
values of the bond variables, specified in the previous section, about the sample of cross-listing US firms.
Table 1: Descriptive Statistics
Table 1 depicts the descriptive statistics for the monthly dataset between January 2000 and December 2015.
Panel A shows the summary statistics of the bonds issued on the New York Stock Exchange and on the Börse Frankfurt by the same US companies. Panel B represents the summary statistics of firm-specific variables and the
macroeconomic factors.
Bond spread is the percentage bond yield less the prevailing risk free rate, Bond size is the principal amount in US$, Bond maturity refers to the remaining life of the bond calculated in months, Investment grade is a dummy variable which assumes the value of the unity if the bond’s credit rating is BBB- or higher by Standard & Poor’s, Previous bond
Panel A: Bond Specific Variables
NYSE FF
Mean Median MaximumMinimumStd. Dev. Mean Median MaximumMinimumStd. Dev. Bond spread (%) 1.61 1.29 10.53 -0.49 1.25 1.37 1.25 14.30 -0.96 1.13 Bond size (US$ million) 0.57 0.35 100.00 0.01 3.56 0.60 0.50 4.00 0.00 0.54 Bond maturity (months) 146.53 107.97 360.93 4.33 100.30 141.93 114.37 360.43 3.87 95.18 Investment grade (indicator) 0.92 1.00 1.00 0.00 0.27 0.98 1.00 1.00 0.00 0.13 Previous bond issue (indicator) 0.63 1.00 1.00 0.00 0.48 0.38 0.00 1.00 0.00 0.48 Liquidity spread (percent) 0.62 0.53 14.06 0.00 0.59 0.58 0.48 10.00 0.00 0.63 Panel B: Firm-Specific and Macroeconomic Control Variables
NYSE FF
14
issue is an indicator with the value of the unity if the company has already issued a bond in a specific exchange market and zero otherwise, Liquidity spread is the percentage bid-ask spread of the bond; Market to book value is the ratio of market value of equity to book value of equity, Return on assets is the ratio of the operative income to total assets of the company, Total assets is the tangible resources of the firms, Leverage is the ratio of long-term debt divided by total assets, Inflation is the monthly percentage change in the consumer price index, Gdp Growth is the percent rate of increase in real gross domestic product.
The 52 US companies have a propensity to issue slightly larger bonds on the foreign country than on the domestic exchange market, while the maturity of the bonds are similar; the indicator
Previous bond issue indicates that a double number of bonds are issued on the NYSE than on FF,
this fact denotes that US companies which choose to cross-list bonds do not completely leave the domestic market in spite of a bond yield spread reduction on average (1.61 % paid on NYSE vs. 1.37% on FF).
The difference of the mean value regarding the credit rating of the company, explained by the
Investment grade indicator, shows the decision of US firms with a better recognition to issue more
bonds on the foreign market than the firms with a lower credit rating.
The bonds listed in Europe have a smaller liquidity spread than the bonds in USA, this is an advantage for the investor base which have less costs in trading the security.
The values in Table 1 about firm-specific variables show differences because companies issue a different number of bonds on the market.
Panel B depicts information at the firm-specific and macroeconomic level for the same dataset.
The mean of the Market to book variable suggests that the issue of bonds in the foreign market is proportionally driven by this ratio, while the median value is roughly the same in both the
exchange markets.
The Return on assets is the variable that most differs across markets; the mean and median values are higher on Börse Frankfurt, denoting that firms which can create higher profit choose to issue bonds on the foreign market.
The Total assets presents a large standard deviation and an inverse pattern for mean and median values.
The Leverage does not show high deviations among the markets, which should not imply differences in the bond yield spreads.
During the period that the bonds are issued, the average inflation is lower and the economy experiences a higher GDP growth in the USA than in the German market.
Table 2, following the same characteristics of the previous table, depicts the values of bonds
15
Table 2: Descriptive Statistics
Table 2 shows the same information of Table 1. The samples are the bonds issued on NYSE by US companies which choose to cross-list on the Börse Frankfurt and the bonds listed on the NYSE by US companies which do not cross-list.
Panel A shows a bond yield reduction for firms that choose not to cross-list (1.38% average spread vs. 1.61%).
The Bond size and Investment grade indicator are similar for both the groups of the US firms. Cross-listing firms issue bonds with a longer life, around 32 months, and the Previous bond issue indicates that each of these firms choose to issue more bonds on NYSE than the sample of US firms not cross-listing on FF.
The Liquidity spread highlights higher trading costs for US cross-listing companies: the choice to cross-list on an important exchange market as Börse Frankfurt implies a trading migration of the investor base which negatively affects the liquidity spread of the bonds traded on the local market.
The mean, standard deviation and median values show that US firms not cross-listing on FF have a lower liquidity spread: if the investor base wants to add the security in the portfolio, the only strong exchange market is the NYSE and the majority of the trades will be in the domestic market. Panel B depicts the firm-specific and macroeconomic variables of the second sample.
The mean value of Market to book and Total assets are higher for US companies cross-listing on FF than for companies that do not but the standard deviation in enormous, which does not allow to make a clear comparison.
The Roa and Leverage ratio present a better financial statement for FF cross-listed US firms, which should influence the required yields on the bonds issued.
Panel A: Bond Specific Variables
NYSE CROSS-LISTED NYSE NOT CROSS-LISTED Mean Median MaximumMinimumStd. Dev. Mean Median MaximumMinimumStd. Dev. Bond spread (%) 1.61 1.29 10.53 -0.49 1.25 1.38 1.16 16.30 -0.48 1.04 Bond size (US$ million) 0.57 0.35 100.00 0.01 3.56 0.55 0.40 12.50 0.01 1.04 Bond maturity (months) 146.53 107.97 360.93 4.33 100.30 109.90 87.37 362.43 3.80 87.77 Investment grade (indicator) 0.92 1.00 1.00 0.00 0.27 0.82 1.00 1.00 0.00 0.38 Previous bond issue (indicator) 0.63 1.00 1.00 0.00 0.48 0.53 1.00 1.00 0.00 0.50 Liquidity spread (percent) 0.62 0.53 14.06 0.00 0.59 0.48 0.37 6.92 0.00 0.39
Panel B: Firm-Specific and Macroeconomic Control Variables
16
4. Empirical Analysis
4.1 Regression model
This section illustrates the model and the methodology I use to verify the hypothesis.
As suggested by Ball et al. (2013) I estimate the following panel data regression model to test the impact both of cross-listing, the bond-specific and firm-specific and macroeconomic variables on the bond yield spreads:
Bond Yield Spreadi,t = γ0 + β1Cross-listing i,t + β2 Bond-specific controls i,t + β3 Dummy
Bond-specific controls i,t + β4 Firm-specific controls and macroeconomic controls i,t + β5
Dummy Firm-specific controls and macroeconomic controls i,t + i,t
The Bond Yield Spreadi,t is the dependent variable, defined as the return offered by the bond less the equivalent risk free interest rate for time t and bond i, which determines the cost of capital reduction.
β1 tests the significance of the cross-listing in changes of the bond yield spreads.
β2 is the coefficient of the first set of independent variable, which includes the Size, Life, Investment Grade, Previous Bond Issue and Liquidity Premium to control for the bond characteristics,
β3 measures the bond-specific variables effect in changes of the bond yields when listed on the Börse Frankfurt.
β4 gauges the effects on the cost of capital by the firm’s variables, defined as Market to Book Value, Return on Assets, Total Assets and Leverage for each company i for time t, and for the Macroeconomic changes defined by Inflation and GDP growth of USA and Germany.
β5 is the coefficient of the impact of the firm-specific controls and macroeconomic variables on the cross-listed bonds, while i,t denotes the error term.
I test the second hypothesis with the following panel data regression model:
Bond Yield Spreadi,t = γ0 + β1FLiquidity i,t + β2 Liquidity spread i,t + β3 Investment grade i,t + β4 Leverage i,t + i,t
I use the FLiquidity variable, which is equal to the unity if the bonds are listed on the New York Stock Exchange by US companies not cross-listing, and zero if the bonds are issued by US
companies which cross-list on the Börse Frankfurt to gauge the significance of the Liquidity spread in the sample of US firms cross-listing and US firms which do not cross-list on the Börse Frankfurt. β1 tests the influence of the Liquidity spread on the changes of the bond yield spreads issued by not cross-listing companies.
β2 is the coefficient of the Liquidity spread.
β3 measures the credit rating in the changes of the bond yield spreads.
17
4.2 Methodology
The panel data models described in the previous section measure the impact of the cross-listing, the bond-specific and firm-specific factors, including the macroeconomic scenario, on the
differences of the bond yield spreads for every month over the period 2000-2015.
The cross-sectional panel data approach is relevant in capturing the changes of the variables over time, with three specifications in the analysis of the dataset: pooled ordinary least square (OLS), entity-time fixed and random effects.
The first regression is the pooled regression, determined as Yi,t = + β * Xi,t + εi,t, which is the
most common method used but the underlying assumption is that there is no heterogeneity. The entity-fixed effects model (Yi,t = + β * Xi,t + μi,t + εi,t) adds the μi,t term called entity-fixed effect
which allows the intercept to differ cross-sectionally but not over time, solving for the
heterogeneity among individual companies. In the opposite way, the time-fixed effects model (Yi,t
= + β * Xi,t + λi,t + εi,t) controls for changes over time but not among entities by adding the
constant term λi,t. The final step involves both the entity and time fixed effects combined. The evaluation of the correct model is estimated with the use of the redundant fixed-effects test. On the other side, there is the random effects equation (Yi,t = + β * Xi,t + εi,t), where εi,t = υi,t + ωi,t. This model assumes a constant with a random disturbance term υi,t and the idiosyncratic
component-noise term ωi,t; the component estimates the deviation of the individual constant from the common component to distinguish the unobserved individual effects. The Generalized Least Squares (GLS) method is used instead of the OLS to measure the appropriate coefficients and this model uses the assumption that the random effect on the error term is uncorrelated with the independent variables Corr (υi,t, Xi,t ) = 0; I run the Hausman test, where the null hypothesis is that the constant term is uncorrelated with the independent variable, to verify this assumption. If the result implies to reject the null hypothesis, the fixed-effects model is appropriate and it is performed with the Maximum Likelihood (LM) method.
I can not use the fixed-effects model because the aim of this study is to measure the impact of cross-listing on the bond yields and of the liquidity spread, which are defined as a dummy variable; the drawback of this model is the impossibility to perform variables which do not vary over time, therefore I use the random effect model.
4.3 Regression Results
I use the pooled OLS panel data model to run the regression described in Regression Model section and I add random.
I introduce an indicator variable, Cross-listing, which is equal to unity if the bond is cross-listed on Börse Frankfurt and zero if the bond is issued on the New York Stock Exchange, to test the
significance of the cross-listing on the differences of the bond yield spreads.
I can not use the fixed effect model because the aim of this paper is to find the coefficient of the dummy variable Cross-listing, which will drop as I run a fixed-effect model.
Table 3 depicts the variables of the regression specified with the random-effects panel data
18
Table 3: Random-Effects Regression Analysis of Bond Yield Spreads.
The bond yield spread regression is determined by C, the constant term, CROSS LISTING is a variable which explains the significance of cross-listing in the changes of the bond yields, the variables are explained in Section III and the “CROSS” variables define the impact of cross-listing. ***, ** and * indicate statistical significance at the 1%, 5% and 10% respectively.
The Crosslisting coefficient has the expected negative sign and it is significant at the 1% level, this implies a bond yield spread reduction around 22 basis points for bonds cross-listed on the Börse Frankfurt.
The Size factor is positive and significant for cross-listed bonds but the magnitude is almost zero, the bond yields required should be higher for a huge bond principal while for the aggregate sample of US companies the Size is not statistically significant.
The Maturity coefficient is close to zero for cross-listed bonds and bonds issued on the New York Stock Exchange: this value does not affect changes in the bond yield spreads.
The Investment grade drives down the offered bond yield across the sample with the same magnitude and it is significant at the 1% level.
The Previous bond issue is not statistically significant and it assumes a positive value for bonds cross-listed and negative for the whole sample of US companies.
NYSE VS. FF
C 0.0232***
CROSS LISTING -0.2192***
CROSS BOND SIZE 0.0042***
BOND SIZE -0.0001
CROSS BOND MATURITY -0.0000***
BOND MATURITY 0.0000***
CROSS INVESTMENT GRADE -0.0114***
INVESTMENT GRADE -0.0175***
CROSS PREVIOU BOND ISSUE 0.0007 PREVIOUS BOND ISSUE -0.0003 CROSS LIQUIDITY SPREAD 0.9146***
LIQUIDITY SPREAD 0.0796***
CROSS MTB 0.0000***
MTB -0.0000***
CROSS ROA 0.0004***
ROA -0.0002***
CROSS TOTAL ASSETS 0.0000**
19
The Liquidity spread is an important factor in determine the dependent variable, it is significant at the 1% level for cross-listed bonds and increases the required yield of 92 basis points, while the coefficient related to the whole sample is lower in magnitude (8 basis points).
The firm-specific variables are significant but the values are around zero, these coefficients can not explain the changes in the bond yield spreads.
The Inflation variable is significant at the 1% level and positive for the bonds listed on the Börse Frankfurt while the sign is negative across the sample.
The GDP is significant and it reduces the bond yield spreads when the bonds are issued abroad while the sign is positive across the sample
The second hypothesis is confirmed by the result of the FLiquidity, which is negative for bonds issued by the not cross-listing US companies than for the bonds traded on NYSE issued by the US companies cross-listing: this suggestion implies that cross-listing fosters trading migration to the foreign market.
Table 4 shows the regression results for the sample of US companies cross-listing and not which issue bonds on the New York Stock Exchange.
Table 4: Random-Effects Regression Analysis of Bond Yield Spreads to test for the liquidity.
The Bond yield spreads is the dependent variable of the analysis, C is the constant term, FLIQUIDITY is a dummy variable which tests the impact of the Liquidity spread on the changes of the bond yields offered by US companies not cross-listing on the Börse Frankfurt, Liquidity spread is the bid price minus the ask price, divided by the ask price for the sample of US companies, Investment grade is the credit rating of the companies, Leverage is the ratio of long-term debt divided by total assets. ***, ** and * indicate statistical significance at the 1%, 5% and 10% respectively.
The results of the analysis show that not cross-listing bonds on the Börse Frankfurt reduces by 3.8 basis points the offered bond yield spreads listed on the New York Stock Exchange, while the liquidity spread across the sample implies an increase of the required return of the bonds listed on the NYSE.
The Investment grade and the Leverage have the expected signs and are statistically significant at the 1% level: a better credit ratings lower the bond yield and an increase in the leverage is
proportional to the bond yields. NYSE CROSS vs. NYSE NOT CROSS
C 0.0299***
FLIQUIDITY -0.038***
LIQUIDITY_SPREAD 0.3534*** INVESTMENT_GRADE -0.0159***
20
5. Conclusion
In this paper, I examine the bond yield spreads of US companies which cross-list to identify a source of decrease of cost of capital.
I find a positive and persistent reduction in the yield of bonds issued on the Börse Frankfurt equal to 22 basis points: this result drives a total reduction of cost of capital of 430120$ if I measure the cross-listing effect to the amount of capital issued on the Börse Frankfurt.
Although this result is positive, the variables which explain the bond yield do not have a strong explanatory power.
The Investment grade and Liquidity are in line with the previous studies and they confirm the explanatory power in affecting the bond yield spreads: this outcome is the most significant in explaining the drivers of the cost of capital reduction.
The firm-specific variables do not play a consistent role in determining if a US company should cross-lists or not.
21
6. References
Baker H. K., Nofsinger J. R. and Weaver D. G. (2002). International Cross-Listing and Visibility. The Journal of Financial and Quantitative Analysis, Vol. 37, No. 3 (Sep., 2002), pp. 495- 521.
Ball R. T., Hail Luzi and Vasvari Florin P. (2013). Equity Cross-Listings in the U.S. and the Price of Debt. ECGI - Finance Working Paper No. 274/2010.
Chambers D., Sarkissian S. and Schill M.J. (2015). Market and Regional Segmentation and Risk Premia in the First Era of Financial Globalization. Cambridge University Working Paper.
Chen Long, Lesmond David A. and Wei Jason (2007). Corporate Yield Spreads and Bond Liquidity. The Journal of Finance, Vol. 62, No. 1 (Feb., 2007), pp. 119-149.
Coffee, John C. Jr. (2002).Racing towards the Top? : The Impact of Cross-Listings and Stock Market Competition on International Corporate Governance. Columbia Law Review, Vol. 102, No. 7, pp. 1757-1831.
Errunza V.R. and Losq E. (1985). International asset pricing under mild segmentation: Theory and test. Journal of Finance, Vol. 40, pp. 105-124.
Karolyi G.A. (2006). The World of Cross-Listings and Cross-Listings of the World: Challenging Conventional Wisdom. Review of Finance Vol. 10, No. 1, pp. 99-152.
Levine R. and Schmukler Sergio L. (2005). Internationalization and Stock Market Liquidity. Review of Finance (2006) 10: 1–35.
Litvak K. (2008). The Correlation Between Cross-Listing Premia, US Stock Prices, and Volume of US Trading: A Challenge to Law-Based Theories of Cross-Listing. University of Texas Law School partial and preliminary draft, January 2008.
Massa M. and Zaldokas A.(2011). Investor Base and Corporate Borrowing Policy Evidence from International Bonds. Journal of International Economics, 92(1), 95-110.
Merton R. C. (1986). A Simple Model of Capital Market Equilibrium with Incomplete Information. The Journal of Finance, Vol. 42, No. 3, Papers and Proceedings of the Forty-Fifth.
22
