University of Amsterdam
MSc Business Economics, Finance Track
Master Thesis
July, 2015
Creditor Rights, Stock Market Development and IPO activity:
a cross-country analysis
Hendrik Reimand Student number: 10065121 Supervisor: dr. Rafael Perez Ribas
Abstract: This Master thesis examines the relationship between creditor rights, stock market
development and the number of initial public offerings (IPOs) across 66 countries between 1989 and 2014. Firstly, the thesis finds evidence that strong creditor rights lead to higher stock market development as measured by liquidity. Secondly, the thesis uncovers that a more liquid stock market encourages IPOs, thereby illustrating a channel how market liquidity contributes to economic growth. Thirdly, the paper discovers that creditor rights on their own are also an important driver of the number of IPOs. Finally, the thesis finds no evidence that creditor rights or market liquidity affect when in their life cycle companies decide to go public.
2
Statement of Originality
This document is written by Hendrik Reimand, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Table of Contents
1. Introduction
4
2. Literature review
6
3. Data and Descriptive Statistics
11
4. Empirical Method
14
5. Results
18
5.1 Creditor Rights and Market Liquidity 18
5.2 Determinants of the number of IPOs 20
5.4 The Determinants of the Average Size of IPOs 23
6. Robustness Checks
26
7. Conclusion
29
8. References
32
4
1. Introduction
Several Eastern European stock markets are experiencing historically low trading volumes and have not witnessed an initial public offering (IPO) for many years. At the same time, most stock markets in Western Europe are recovering from the financial crisis and seeing substantial increases in the number of IPOs. This has sparked considerable political debate in these Eastern European countries about the need to liven up the stock markets to boost economic growth. The political theory goes that liquid stock markets, combined with favorable regulation, enhance capital allocation and encourage investment by providing an exit strategy for early investors. Therefore, liquid stock markets should lead to more IPOs earlier in the firm's life, improve capital allocation and boost economic growth. However, it remains to be studied what actually determines the level of IPO activity at a stock market and what type of regulation can be regarded as favorable.
This Master thesis explores a channel through which stock market liquidity could affect economic growth and studies the relationship between the strength of creditor rights, the level of market liquidity and the number and size of IPOs. Firstly, the paper investigates whether creditor rights affect the level of stock market development as measured by liquidity. Secondly, the thesis studies if average market liquidity and the strength of creditor rights impact the level of IPO activity in a stock market. Thirdly, the paper examines whether market liquidity and creditor rights also impact when companies choose to do an IPO as measured by the average size of an IPO.
Several authors, for example Rousseau and Wachtel (2000), Beck and Levine (2004) or Cooray (2010) find that stock market liquidity contributes to economic growth. Additionally, Djankov, McLiesh and Shleifer (2007) show that stronger creditor rights lead to higher aggregate lending in the economy, thereby increasing the level of financial development and potentially also contributing to economic growth. However, the authors do not empirically investigate the channels through which stock market liquidity affects the broader economy or whether the strength of creditor rights also affects stock market development and the number of IPOs. One channel through which stock market liquidity could affect economic growth is IPO activity. Firstly, a public offering is the only way how a stock market can affect capital allocation. Thus, if the political argument is that stock markets foster growth by enhancing capital allocation, this channel must operate through initial and secondary public offerings. Secondly, IPOs themselves cause positive externalities for companies. For example, firms generally need to improve governance mechanisms and
5 reporting standards before doing an IPO. Thirdly, if higher market liquidity would encourage firms to do an IPO earlier in their life cycle, it would further boost growth as firms can more quickly benefit from the positive effects of an IPO. Nevertheless, it is unreasonable to assume that market liquidity is the only determinant of IPO activity. Creditor rights can play an important role here by determining the sources of financing that companies prefer and affecting the number of potential IPO candidates. Furthermore, if creditor rights impact stock market development, they can have an indirect effect on IPOs through market liquidity.
Existing literature provides reasons to believe that market liquidity has an important impact on IPO activity. Ellul and Pagano (2006) find that IPO underpricing increases when investors expect the stock to be less liquid in the after-market. As underpricing represents a significant part of the cost of doing an IPO, higher underpricing, ceteris paribus, should lead to fewer IPOs. If underpricing decreases with the level of market liquidity, a more liquid stock market should positively affect the number of IPOs. Existing literature also supports the view that the strength of creditor rights influences the supply of potential IPO candidates. On the one hand, Acharya and Subramanian (2009) find that low creditor rights lead to more innovation by encouraging firms to employ more leverage and invest more. Thus, lower creditor rights should lead to more potential IPO candidates. On the other hand, Djankov, McLiesh and Shleifer (2007) find that the supply of credit increases with creditor rights. Thus, higher creditor rights make it easier for firms to borrow and develop up to a size suitable for an IPO. This thesis contributes to existing literature by examining which effect prevails and whether creditor rights affect the level of IPO activity and stock market development. An alternative explanation is that creditor rights are simply correlated with investor rights, which Djankov et al (2008) find to be an important determinant of IPO activity. Thus, the paper also investigates this possibility. In addition, this thesis contributes to existing literature by examining a clear channel through which stock market liquidity could lead to economic growth and investigates whether market liquidity impacts the level of IPO activity. Finally, the thesis contributes to sparse literature on the determinants of the average size of IPOs.
To study the aforementioned points, this thesis undertakes numerous regressions using annual panel data on 66 countries. The regressions are estimated with both random and country fixed effects, whereas the regressions using count data on the number of IPOs are estimated both assuming a Poisson and negative binomial distribution. Due to limited data, the paper uses two different sample periods that are based on two different sources of creditor rights. The first period, ranging from 1989 until 2003, is based on data from a paper by
6 Djankov, McLiesh and Shleifer (2007). The second time period ranges from 2004 until 2014 and creditor rights data is obtained from the World Bank Doing Business Report. The thesis uses the Amihud (2002) illiquidity measure to gauge market liquidity, which is calculated by obtaining the daily stock price movements of all stocks in 66 sample countries over a period of 25 years. Finally, all regressions include a number of control variables that are obtained from various sources.
The Master thesis finds evidence that stronger creditor rights increase the level of stock market development, as measured by market liquidity. However, this relationship only holds when creditor rights are very strong. Additionally, the paper finds strong evidence that both market liquidity and creditor rights play an important role in determining the level of IPO activity. Going from low creditor rights to average creditor rights significantly increases the number of IPOs, whereas having very high level of creditor rights has a negative effect on IPO activity. In addition, the average level of market illiquidity has a strong negative effect on the number of IPOs. Thus, a more liquid stock market seems to encourage more IPOs. The thesis finds no consistent and clear evidence that creditor rights or market liquidity influence when companies decide to go public, as measured by the size of IPOs. All the results are robust to including the level of investor rights in the regressions.
2. Literature review
Most of the literature on the finance growth nexus finds that both financial development in general and stock market development in particular provide valuable services to the economy and thereby contribute to GDP growth. Levine (1997) provides an overview of the early literature on the subject. He also distinguishes and analyzes five different functions of finance that contribute to economic growth: 1) facilitating trading, hedging, pooling and diversifying of risk; 2) allocating resources; 3) monitoring managers and exerting corporate control; 4) mobilizing savings; and 5) facilitating the exchange of goods and services. By doing an IPO, firms could benefit from several of Levine's functions of finance. For example, firms generally have to improve governance mechanisms when going public. Thus, a higher number of IPOs would mean that more firms employ better corporate governance. Another example is that an IPO would enable the initial investors of the firm to benefit from diversifying their risk, while the company would benefit from outside investors allocating their resources to the firm. Furthermore, a higher number of IPOs would help to mobilize a larger amount of savings in the society, thereby improving aggregate capital
7 allocation. Thus, there are several theoretical channels how the level of IPO activity would affect economic growth.
There is also strong empirical evidence that stock market liquidity contributes to economic growth. Levine and Zervos (1998) argue that both banking development and stock market development have their own separate effects on economic growth. Other studies, such as Rousseau and Wachtel (2000), Beck and Levine (2004), and Cooray (2010), find evidence that a higher turnover ratio enhances economic growth. However, these studies do not focus on particular channels through which stock market development might affect economic growth. A more developed and liquid stock market could lead to more IPOs, which in turn could contribute to economic growth, but this remains to be tested.
There are other papers that examine some specific channels through which stock market liquidity and developments can affect firm decisions, implying that the stock markets are not simply a sideshow. Campello, Ribas and Wang (2014) examine a change in Chinese regulation that had a positive effect on market liquidity. The authors find that the reforms gave firms greater access to equity financing and increased firm performance, investments and productivity. Hau and Lai (2013) study fire sales by distressed equity funds and find that exogenous stock underpricing has a significant effect on firm's investment and employment decisions. Thus, there is substantial evidence that stock markets can affect corporate decisions. Moreover, if stock markets affect the investment decisions or productivity of firms, they also directly affect economic growth. This thesis contributes to existing literature by studying whether the level of IPO activity could be another channel through which market liquidity contributes to economic growth.
Existing literature provides a potential link how market liquidity could affect IPO activity by influencing IPO underpricing. Ellul and Pagano (2006) develop a model where IPO underpricing depends on both the amount of asymmetric information and after-market liquidity, which may stem from lack of information. They argue that a less liquid after market and the uncertainty regarding the liquidity profoundly increase IPO underpricing. In contrast, Hahn, Ligon and Rhodes (2013) argue that the effect is reversed and that higher underpricing leads to more secondary market liquidity, which persists for a relatively long time. Similarly, Zheng and Li (2008) find that IPO underpricing reduces the amount of institutional block-holders, which in turn increases after-market liquidity. In addition, they find that IPO underpricing has also a direct effect on after-market liquidity, even when controlling for ownership concentration. Furthermore, Pham, Kalev and Steen (2003) use data on Australian IPOs and also reach the conclusion that IPO underpricing increases after-market liquidity.
8 Thus, there is strong evidence that IPO underpricing and secondary market liquidity are connected. The issuer of shares may want to ensure a certain level of secondary market liquidity and choose the underpricing accordingly. However, if the stock market on average is less liquid, the underpricing that results in the desired level of liquidity is larger. IPO underpricing is a significant part of the cost of doing an IPO. If the theory is correct, a less liquid stock market will lead to higher underpricing and a higher cost of an IPO, thereby reducing IPO activity. Therefore, this thesis expects to find that stock market liquidity increases the number of IPOs and, through that, enhances economic growth.
In addition to market liquidity, there are most probably other determinants of IPO activity and the existing literature on the topic has used rather different approaches. Pastor and Veronesi (2005) study IPO waves and conclude that IPO waves are usually preceded by high market returns and followed by low market returns. Lowry and Schwert (2002) also study IPO waves and argue that the average market returns around filing for the IPO contain no information for IPO underpricing, which usually constitutes a substantial amount of the total cost of doing an IPO. Thus, IPO waves tend to occur after periods of high market returns, but the authors do not investigate whether also market liquidity might play a role in determining the number of IPOs. Other studies use cross-country data and look at country specific determinants of the number of IPOs. La Porta et al (1997) show that countries with lower investor protections have narrower capital markets, as for example measured by the number of IPOs. Djankov et al (2008) use a newer, more robust measure of investor protection and again show that better investor protection leads to more stock market development, where one measure of stock market development is the number of IPOs per GDP. Therefore, the authors show that better investor protection makes it easier for firms to raise equity capital through public offerings and increases the number of IPOs. Additionally, Haidar (2009) argues that stronger investor protection leads to higher GDP growth, making a potential link between increasing IPO activity and GDP growth. Finally, Doidge, Karolyi and Stulz (2013) show that stronger domestic institutions lead to a higher likelihood of a firm conducting a domestic IPO. Thus, there is strong evidence that country specific regulation, such as the level of investor protection, affects the level of IPO activity. Nevertheless, the papers do not address whether market liquidity plays any role in determining the number of IPOs. Moreover, existing literature has not studied whether the level of creditor rights might also affect IPO activity.
At first, it may be difficult to see how creditor rights could potentially affect the level of IPOs, as shareholders are protected by investor rights, not creditor rights. Nevertheless,
9 papers addressing the level of creditor rights provide a basis for a theoretical channel how the latter could impact the number of IPOs. Acharya and Subramanian (2009) argue that debtor friendly bankruptcy laws, which in essence imply lower creditor rights, foster innovation by encouraging firms to invest more. Similarly, Lee et al. (2011) find that entrepreneur-friendly and lenient bankruptcy laws, again corresponding to lower creditor rights, encourage entrepreneurship. Therefore, on the one hand, lower creditor rights should lead to more small, quickly developing companies that are potential IPO candidates. On the other hand, lower creditor rights naturally imply that creditors will demand a higher interest on loans or will refrain from lending altogether. This is confirmed by Djankov, McLiesh and Shleifer (2007), who find that stronger creditor rights increase the amount of private credit in an economy. Thus, very low creditor rights could decrease firms' access to finance, limit their growth and reduce the amount of potential IPO candidates. Moreover, as capital markets are usually integrated, a higher interest rate caused by lower creditor rights could also translate into a higher cost of equity capital, making an IPO more expensive for the firm. Independently from which theory is correct, there are reasons to believe that creditor rights may play a role in determining the level of IPO activity. Moreover, if creditor rights affect the level of financial development as measured by aggregate lending, they can also potentially affect stock market development.
Another possibility is that creditor rights affect IPO activity and stock market development simply because they are correlated with other regulatory variables. Tian (2011) finds that the high average underpricing in Chinese IPOs is mainly caused by government intervention in IPO pricing regulation and government control of IPO share supplies. Thus, country specific regulation can significantly impact the cost of doing an IPO. A different study by Banerjee, Dai and Shrestha (2011) uses cross-country data on 36 countries to investigate the effect of country-specific regulation on IPO underpricing. The authors find that country-level information asymmetry, investors' home-country bias, effectiveness of contract enforcement mechanisms, and accessibility of legal recourse, all have important effects on IPO underpricing. Furthermore, Boulton, Smart and Zutter (2010) and Hopp and Dreher (2013) use panel data covering between 20 to 30 countries and conclude that stronger investor rights lead to higher IPO underpricing. In contrast, Engelen and Essen (2010) find that a higher quality of investor protection leads to lower IPO underpricing. Although the results are contradictory, there seems to be strong evidence that country specific regulation, in particular the level of investor protection, can affect IPO underpricing. Thus, it is also plausible that the strength of investor protection or other regulatory variables affect IPO
10 activity directly. If creditor rights are correlated with the level of investor protection, any regression of IPO activity on creditor rights could be biased. This thesis takes the possibility into account in the robustness checks and expects to find that creditor rights affect the number of IPOs even when controlling for country fixed-effects and the strength of investor protection.
Existing literature does not investigate the cross-country determinants of the average size of IPOs. This may well be because no relationship exists at all and there are no variations in the average size of IPOs across countries, at least when controlling for variables such as GDP size and the development stage of a country. However, if market liquidity somehow affects when companies decide to go public, it would prove to be another channel how market liquidity affects economic growth. As stated before, IPO activity would generally be beneficial for the economy by improving capital allocation or the level of corporate governance. If a more liquid stock market encourages firms to do an IPO earlier in their life cycle, it would mean that firms benefit more quickly from the positive externalities of an IPO. Thus, it would constitute another channel how market liquidity contributes to economic growth. In addition, creditor rights could also impact the average size of IPOs. Both Acharya and Subramanian (2009) and Lee et al. (2011) find that lower creditor rights increase innovation and entrepreneurship. Thus, lower creditor rights should lead to more small, quickly developing companies. However, if lower creditor rights lead to lower aggregate lending, as witnessed by Djankov, McLiesh and Shleifer (2007), these companies would experience lending constraints. As a result, these companies could choose an IPO to fund their expansion. This argument provides a theoretical reason why lower creditor rights should lead to a smaller average size of IPOs, but this remains to be studied. This thesis contributes to existing literature by examining whether market liquidity and creditor rights affect the average size of IPOs.
All in all, this Master thesis contributes to the existing literature in several ways. Firstly, it investigates whether the strength of creditor rights affects stock market development directly, apart from affecting general financial development as measured by lending. Secondly, this thesis adds to the literature on the finance growth nexus by examining whether influencing the number of IPOs is one channel how market liquidity affects economic growth. Finally, this thesis contributes to current literature by examining whether market liquidity and creditor rights affect the average size of IPOs.
11
3. Data and Descriptive Statistics
This section describes the data sources and the construction of variables used in this paper. It also presents descriptive statistics on these variables. Table A1 in the Appendix presents a short summary of the variables used and their sources.
The full sample of this paper consists of 66 countries and the sample period is divided into two, because of limited availability of creditor rights data, which is obtained from two different sources. Firstly, Djankov, McLiesh and Shleifer (2007) provide an extensive dataset on creditor rights for 129 countries from 1978-2003 (DMS). Their creditor rights index ranges from 0 (very weak creditor rights) to 4 (very strong creditor rights) and is constructed by adding one, whenever one of the following conditions is met: 1) there are restrictions when a borrower wants to file for reorganization; 2) secured creditors can seize the collateral securing their loans when a reorganization petition is approved; 3) secured creditors are paid before other creditors and the government or workers when a company is liquidated; 4) management has to waive the right to administer the property when there is a pending reorganization resolution. Secondly, The World Bank World Development Indicators (WDI) database provides data on the strength of legal rights for almost all World Bank member countries from 2004 to 2014. This index is similar to DMS as it measures the strength of legal rights that protect creditors and enhance lending. Nevertheless, the two indices come from different sources and use different scales. Therefore, this thesis uses two different sample periods: 1978-2003 using DMS data and 2004-2012 using WDI. For robustness checks, this paper also uses data on the strength of investor rights, which is obtained from the World Bank Doing Business Report. The investor rights index is available from 2004 to 2014 and thus robustness checks are only estimated for the second sample period. The investor rights index ranges from 0 to 10 and measures the level of minority shareholder protection. The construction of the variable is based on Djankov et al (2008).
Data on the number of IPOs, their size and underpricing is obtained from Thomson One database, which is preferred to Zephyr, because the latter only starts in 1997. The size of an IPO is defined as the total proceeds from share issuance in millions of US Dollars and underpricing is defined as the percentage change in stock price during the first trading day. The full sample consists of 41,184 IPOs observed between 1989 and 2014. However, not all data fields are available for some IPOs, with more missing values for developing countries in the early 1990s. The full sample is still used to assess the number of IPOs per country, but smaller samples are used for other purposes. When excluding all IPOs with missing
12 underpricing, the sample reduces to 10,963 IPOs. Table A2 in the Appendix presents country-level values for the number of IPOs, IPO average size and the average underpricing. The average underpricing is the weighted average underpricing of all IPOs during one year, where the weight is the relative size of the IPO.
This paper differs from earlier authors, such as Beck and Levine (2004), and does not use turnover ratio to measure market liquidity, as turnover ratio measures trading volume, rather than liquidity. For example, turnover ratios were high during the financial crisis although it was difficult to sell shares. Instead, this paper adopts Amihud's (2002) Illiquidity Ratio that is adjusted for currency differences. This ratio has been widely used in the literature and Goyenko, Holden and Trzcinka (2009) conclude that the Amihud ratio does a good job at measuring the price impact of trading. The Illiquidity Ratio is based on equation 1 below.
!""!#!,! =!"#$! !"#$%&' !"#$%&!"#$%& ∗ !"#ℎ!"#$ !"#$ (1) The Amihud (2002) illiquidity ratio is a monthly average of daily absolute returns divided by the daily trading volume, computed separately for all stocks at an exchange. This paper discards penny stocks, which is a standard practice in financial literature. The measure for the stock market is simply the average of all individual illiquidity ratios, weighted by their market capitalization. However, the return is measured in percentages whereas the trading volume is measured in the currency units, introducing a currency bias. To eliminate this bias, all trading volume values are converted to US Dollars, based on annual exchange rates. These annual exchange rates come from the WDI database. Finally, the annual value for the stock market is obtained by taking the average of all monthly values. Data on daily stock prices and trading volumes for all stocks in all 66 sample countries is obtained from the Compustat Global Daily Stock Prices database.
Control variables include annual GDP, population size, inflation, the average lending rate, the number of listed companies, and the share of GDP produced in the services and agriculture sector. The average lending rate is the average bank interest rate that meets the short and medium term financing needs of the private sector, calculated by World Bank. These variables are mostly from WDI or the United Nations Statistics Office, expect for Taiwan, for which the data is obtained from the National Statistics Agency’s website.
Table 1 presents descriptive statistics on all of the aforementioned variables for both sample periods used in this study.
13 Table 1: Descriptive Statistics
This table presents the descriptive statistics for the variables used in this paper. Panel A reports the descriptive statistics for the first time period from 1989 to 2003. Panel B reports the same descriptive statistics for the second sample period from 2004 to 2014. The first column presents the number of annual observations of the variable. The second column presents the number of countries for which this variable is available in at least one year. The third, fourth and fifth column present the mean, median and standard deviation of the variable respectively.
Panel A: 1989-2003
Nr of Obs Nr of Panels Mean Median
Standard Deviation
Nr of IPOs 990 66 21.255 2.000 77.389 Average IPO Size 596 64 137.889 53.033 322.013 Underpricing 188 40 33.928 7.430 188.344 Log of Illiquidity Ratio 497 57 1.133 1.102 2.474 Creditor Rights 930 62 1.955 2.000 1.145 Log of GDP 966 66 25.477 25.580 2.170 Log of Population Size 987 66 16.748 16.798 1.586 Services 970 65 58.886 61.089 12.667 Agriculture 970 65 6.974 3.744 7.446 Inflation 876 61 48.062 3.985 364.893 Lending Rate 805 62 29.796 11.476 191.522 Nr of Companies Listed 867 64 461.514 210.000 777.767 Panel B: 2004-2014
Nr of Obs Nr of Panels Mean Median
Standard Deviation
Nr of IPOs 726 66 27.737 6.000 67.372 Average IPO Size 609 66 229.417 127.000 380.881 Underpricing 454 64 21.070 7.146 127.879 Log of Illiquidity Ratio 629 59 1.875 2.054 2.221 Legal Rights 715 62 5.903 6.000 2.488 Log of GDP 659 66 25.883 26.014 2.166 Log of Population Size 661 66 16.903 16.968 1.540 Services 657 66 61.639 62.015 11.767 Agriculture 657 66 5.641 3.023 6.077 Inflation 688 64 4.238 2.961 4.426 Lending Rate 525 55 9.500 7.750 6.931 Nr of Companies Listed 576 64 596.792 236.000 986.688 Investor rights 521 66 5.794 5.700 1.593
In both sample periods the mean number of IPOs in a year is over 20, whereas the median number of IPOs in a year ranges from 2 to 6. This suggests that there are a few countries with a large number of IPOs per year, while many other countries only witness a few (or none) IPOs per year. Similarly, the mean of the average IPO size is roughly two times its median in both sample periods. In addition, both variables almost doubled from the first sample period to the second. This is expected as large outliers in the average IPO size are
14 probable and the average size of IPOs has increased with the growing world economy over the past 25 years.
The mean underpricing was 33.93% in 1989-2003 and 21.07% in 2004-2014. Due to a different source and measurement scale, it is impossible to compare the two creditor right indices. Finally, the average illiquidity ratio increased from the first sample period to the second, indicating that, on average, the liquidity of the stock markets decreased. This is actually quite a probable result as more and more trading has concentrated on a few large stock exchanges around the world, leaving the smaller national exchanges less liquid and active.
The average GDP and population size increased slightly from the first sample period to the next, which was a general trend across all countries. Together with the overall economic growth, the importance of the services sector increased from 58.89% to 61.64% of GDP, while the share of the agricultural sector dropped from 6.97% to 5.64% of GDP. There was a remarkable drop in the mean inflation rate, down from 48.06% in the first period to just 4.24% in the second period, whereas the drop in the median inflation rate was much less pronounced. This is due to the fact that some countries experienced hyperinflation during the 1980s and 1990s, but none of the countries experienced extremely high inflation after 2003. The average lending rate decreased along with inflation, whereas the average number of companies listed in the domestic stock market has grown substantially.
4. Empirical Method
To answer the research questions presented in the introduction and investigate whether creditor rights impact market liquidity and whether both influence the number of IPOs and their size, this paper undertakes a number of regressions using two complementary sample periods. This section presents the empirical methods used in the order of the research questions that were outlined in the introduction.
To test the first hypothesis that creditor rights affect stock market development as measured by liquidity, the modified Amihud (2002) Illiquidity Ratio is regressed on dummy variables, measuring the strength of creditor rights, and on several control variables. As stated earlier, this paper uses two different measures of creditor rights during two different time periods. The creditor rights index is a categorical variable and is therefore represented by dummy variables. The construction of Dummy variables is outlined in Table 2.
15 Table 2: Specification of Dummy Variables
This table presents the specification of Dummy Variables in both sample periods. The first column presents the values of the creditor or legal rights index, while the other columns present what values the respective dummy variables take in each case.
Panel A: 1989-2003
Creditor Rights Index CR Dummy 1 CR Dummy 2 CR Dummy 3 CR Dummy 4
0 0 0 0 0 1 1 0 0 0 2 1 1 0 0 3 1 1 1 0 4 1 1 1 1 Panel B: 2004-2014
Legal Rights Index LR Dummy 1 LR Dummy 2 LR Dummy 3
< 4 0 0 0
4 - 6 1 0 0
7 - 9 1 1 0
10 - 12 1 1 1
The regressions are run separately by using the fixed and random effects panel data models. On the one hand, it is reasonable to assume that country specific unobserved effects play an important role in determining the level of market liquidity, economic development, regulation, business environment and a range of other measures. On the other hand, there is very limited variability in the creditor rights index within countries and, in the fixed effects model, all time-invariant variables are absorbed by the intercept. Thus, the fixed effects model could result in a loss of information as it assumes all cross country variance that does not change over time to be country-fixed effect. When the country specific effect is uncorrelated with the independent variable, the random effects estimator is more efficient than the fixed effects estimator. However, when the country specific effect and independent variables are correlated, the random effects estimator becomes inconsistent. To test whether to employ the random or fixed effects estimator, this paper conducts the Hausman test, which tests whether the error terms are correlated with the country fixed effect. The test statistics and p-values are reported under regression results and whenever the p-value of the test is under 0.05, one should use fixed effects. However, when the p-value is larger than 0.05, the random effects estimator is more efficient. The regression specification for the fixed effects model is outlined in equation 2,
!"#_!"!,! = !!+ !!!"#!,!+ !!!"#!,!+. . + !!!"#$%"&!,!+ !!+ !!+ !!,! (2) where IL stands for the logarithm of the modified Amihud Illiquidity ratio, CR1, CR2 and so forth stand for the creditor rights dummy variables and control stands for the various control
16 variables. The latter enable the author to control for several observable factors that plausibly affect market liquidity, such as the size and development stage of a country. The unobservable country specific factors are controlled for in the fixed effects model. In addition, the regression includes year specific effects to account for any year-specific shocks across all countries. The use of various control variables and fixed effects should reduce any potential omitted variable bias. Furthermore, reverse causality seems unlikely as it is hard to see how market liquidity could potentially affect the country level regulation on creditor protection. The interpretation of results is fairly simple. If creditor rights dummies enter the regression significantly with a negative sign, it provides proof that higher creditor rights lead to lower market illiquidity and thus higher stock market development. If the dummies are found to have a significantly positive impact, it means that the opposite is true.
To test the second hypothesis and investigate whether both market liquidity and creditor rights affect the level of IPO activity, this thesis runs a number of regressions that are based on count panel data models. The latter are used because the number of IPOs in a country is by definition count data. The number of IPOs is regressed on the logarithm of the Amihud Illiquidity Ratio, the creditor right dummy variables, the average IPO underpricing and control variables using two different models. Firstly, it is assumed that the dependent variable is Poisson distributed, which implies a restriction that, for a given country, the mean of each count must equal its variance (Allison and Waterman, 2002). To test whether this restriction is satisfied, the paper tests for overdispersion and the test statistic is reported under regression results. Secondly, the regressions are run by assuming that the dependent variable is negative binomially distributed. The negative binomial regression is less efficient than the Poisson regression, but it is robust to overdispersion. Equation 3 below summarizes the regression specification that assumes Poisson distribution.
!"(!!,!) = !!+ !!+ !!!"#!,!+. . . +!!!"#_!"!,!+ !!!!!"!,!+ !!!!!"#$%"&!,! (3)
In the Poisson regression model for panel data, it is assumed that the dependent variable, the number of IPOs, varies across countries and across time and is Poisson distributed with parameter !!,!. The regression includes both country- and year-fixed effects to account for any country- or year-specific unobservable factors. The independent variables have the same interpretation as before. Additionally, UP corresponds to the annual size-weighted average underpricing of all IPOs in a country and is included to account for the cost of doing an IPO. The negative binomial specification is similar, but simply assumes a different distribution for the dependent variable. In addition, both the Poisson model and the negative binomial model
17 use the number of companies listed on the domestic stock exchange as the exposure variable. As both models use the logarithm of the dependent variable in estimation, the coefficients of the independent variables should be interpreted as semi-elasticities.
To test the third and final hypothesis of this paper and investigate whether market liquidity and the strength of creditor rights impacts the average size of IPOs, the logarithm of the latter is regressed on the Amihud Illiquidity ratio, creditor rights dummies and control variables. Similarly to testing the first hypothesis, the panel data regressions are run both by assuming fixed and random effects and the Hausman test is used to determine which model yields better estimates. Equation 4 presents the regression specification for the fixed effects model,
!"#_!"#$!,!= !!+ !!!"#!,!+. . . +!!!"#_!"!,!+ !!!!!"#$%"&!,!+ !!+ !!+ !!,! (4) where log_Size corresponds to the logarithm of the average size of all domestic IPOs within a year. Other variables are defined as before and the regressions include year-fixed effects. As the dependent variable is once again a logarithm, the coefficients should be interpreted as elasticities or semi-elasticities, depending on whether the independent variable has also been log-transformed.
As outlined in the introduction and literature review, this paper estimates several robustness checks to examine the strength of the results. Firstly, all the regressions outlined before are estimated again with the strength of investor rights is included as an explanatory variable. Secondly, it seems plausible that the effect of creditor rights and market liquidity on the average size of IPOs is different for various IPO size quantiles. For example, a more liquid and developed stock market could reduce the average size of the smallest IPOs while increasing the size of the largest IPOs. This could be the case when creditor rights and market liquidity encourage IPOs across the size spectrum, so that small companies conduct IPOs earlier but also large private companies are more likely to conduct an IPO. To investigate this possibility, the thesis estimates quantile regressions. These regressions are run in essentially the same way as in equation four, but are based on IPO level data. Moreover, the regressions are run separately for different quantiles of the average size of IPOs. The results of the robustness checks are reported after the main results.
18
5. Results
This section is divided into three subsections, based on the specific relationship studied. The first subsection presents the results from the regressions that investigate whether creditor rights affect market liquidity. The second subsection analyzes the empirical evidence on whether creditor rights and market liquidity impact the number of IPOs. The third subsection outlines the results of regressions that examine the determinants of the average size of IPOs.
5.1 Creditor Rights and Market Liquidity
The first aim of this paper is to investigate whether creditor rights affect stock market development, when the latter is measured by market liquidity. Several authors, for example Beck and Levine (2004) and Cooray (2010) use measures of market liquidity as proxies for market development and argue that they lead to higher economic growth. Djankov, McLiesh and Shleifer (2007) however find that creditor rights have a substantial effect on the financial development of a country by affecting the amount of aggregate lending. If creditor rights are detrimental for financial development, they can also directly influence stock market development. To investigate this, the thesis estimates panel data regressions of market liquidity on creditor rights. The results of these regressions are presented in Table 3 below.
Firstly, it is important to note that the Hausman test does not provide any reasons to believe that the error terms are correlated with the regressors. Therefore, the random effects model, with results presented in columns 1 and 3, provides more efficient estimates of the actual coefficient. Secondly, CR Dummy 1 and CR Dummy 4 are omitted in column 2 because the variables do not vary across time in the first sample period. Thus, in the fixed-effects regression, these variables are perfectly multicollinear with the country-fixed fixed-effects and cannot be estimated. There is no such problem in the second sample period.
19 Table 3: Creditor Rights and Market Liquidity
Panel A of the table presents regression results for the first time period and panel B for the second time period. For both panels, the results in the first column correspond to the random effects model and the results in the second column to the fixed effects model. Standard errors are clustered over country and reported in parentheses. The year-specific effects are not reported. The Hausman test below the results can be used to determine whether to use the random or fixed effects results. When the p-value of the Hausman test statistic is below 0.05, there is reason to believe that the unique errors are correlated with regressors and one should use the fixed effects results. Otherwise, the random effects model provides more efficient results. *, ** and *** indicate significance at 10%, 5% and 1% level respectively.
Dependent variable: log of Amihud Illiquidity Ratio
Panel A: 1989-2003 (1) (2) Panel B: 2004-2014 (3) (4)
CR Dummy 1 0.1589 Omitted LR Dummy 1 -0.3936 -0.2972
(0.9059) - (0.3293) (0.5161) CR Dummy 2 -1.1468 -1.4162 LR Dummy 2 0.0277 0.1420 (1.3763) (2.4485) (0.3808) (0.5032) CR Dummy 3 0.7320 0.5174 LR Dummy 3 -0.6404* -0.0282 (0.7228) (0.7515) (0.3547) (0.4707) CR Dummy 4 -1.8390*** Omitted (0.6472) - Log GDP -0.5675* 1.3653 Log GDP -0.4891 -1.0376 (0.2935) (3.7899) (0.2991) (2.0090)
Log Population size 0.4183 -10.5312** Log Population size 0.1174 1.4843
(0.3088) (5.1563) (0.3637) (2.9520) Services 0.0606* 0.0437 Services 0.0624** 0.0025 (0.0354) (0.1097) (0.0265) (0.0852) Agriculture 0.0353 -0.0398 Agriculture 0.0762 -0.0227 (0.0553) (0.1597) (0.0773) (0.2014) Inflation 0.0411*** 0.0489*** Inflation 0.0218 0.0159 (0.0159) (0.0177) (0.0277) (0.0292)
Lending Rate -0.0220 -0.0167 Lending Rate 0.0345** 0.0460*
(0.0259) (0.0354) (0.0167) (0.0247)
Constant 0 139.2921** Constant 6.5521* 1.8918
. (68.0813) (3.6447) (39.6569)
Country-fixed effects No Yes Country-fixed effects No Yes
Year-fixed effects Yes Yes Year-fixed effects Yes Yes
N 426 426 N 378 378
Hausman test statistic 20.00 Hausman test statistic 18.29
p-value 0.5212 p-value 0.4368
The coefficients of the low and medium creditor right dummy variables are not significantly different from zero in either time period. However, the coefficient of CR Dummy 4 in column 1 and that of LR Dummy 3 in column 3 are both negative and significantly different from zero. Thus, there is clear evidence that very strong creditor rights lead to lower levels of illiquidity and higher stock market development, when the latter is measured by market liquidity. Moreover, the coefficients are also economically highly significant. Based on these estimates, very high creditor rights decreased the average illiquidity by 64% in the second sample period and by a remarkable 184% in the first sample period. The results are in line with those of Djankov, McLiesh and Shleifer (2007), who
20 argue that higher creditor rights lead to more financial development. It seems that the higher financial development can also lead to more stock market development and higher liquidity. Alternatively, creditor rights may be correlated with investor rights and have no effect on market liquidity on their own. This possibility is addressed in the robustness checks.
5.2 Determinants of the number of IPOs
The second aim of this paper is to examine whether market liquidity and creditor rights influence the level of IPO activity, when the latter is measured by the number of IPOs. Existing literature provides reasons to believe that market liquidity should have a positive effect on the number of IPOs, whereas the effect of creditor rights is unclear. On the one hand, lower creditor rights lead to more innovation and entrepreneurship (Acharya and Subramanian, 2009; Lee et al, 2011), which should increase the number of potential IPO candidates. On the other hand, lower creditor rights lead to lower aggregate lending (Djankov, McLiesh and Shleifer, 2007), which reduces financing and restricts companies growing into a size suitable for an IPO. Table 4 presents the results of the regressions that examined which of the two effects is stronger and whether market liquidity also impacts the number of IPOs.
In both panels of Table 4 it is clear that the number of IPOs exhibits strong signs of overdispersion. Thus, the analysis for both periods concentrates on columns 5 to 8 that are based on negative binomial regression. In the regression results, it is visible that both market liquidity and creditor rights are important determinants of the number of IPOs, even when controlling for a range of variables and country- and year-fixed effects. In the first sample period, the coefficients of creditor rights and market liquidity are significantly different from zero both when included separately and together in the regressions. In the second sample period, the significance of the results becomes stronger when creditor rights and market liquidity are both included in the regressions. In both sample periods, there is clear evidence that market illiquidity has a negative effect on the number of IPOs – a more liquid market results in more IPOs. These results therefore illustrate a channel how market liquidity can contribute to economic growth, as is argued by Beck and Levine (2004) and other literature on the finance growth nexus. Moreover, existing literature shows that market liquidity can influence IPO underpricing and thus the cost of doing an IPO (Ellul and Pagano, 2006). These results show that market liquidity increases the number of IPOs, even when controlling for the average underpricing in the country. Therefore, a liquid stock market must make an IPO attractive for companies for further reasons than just the cost of an IPO.
21 Table 4: Creditor Rights, Market Liquidity and the Number of IPOs
Panel A of this table presents the results for the first time period and panel B for the second time period. All regressions include both country- and year-fixed effects, which are not reported. Standard errors are clustered over country and are reported in parentheses. The overdispersion statistic tests whether, for a given country, the mean of each count is less than or equal to its variance. The null hypothesis expects this to be true and in that case the Poisson regression provides more efficient estimates. If the p-value of the overdispersion test is smaller than 0.05, the null hypothesis is rejected and the results of the negative binomial regression should be used. *, ** and *** indicate significance at 10%, 5% and 1% level respectively.
Panel A: 1989-2003 Dependent Variable: Number of IPOs
Poisson Regression Negative Binomial Regression
(1) (2) (3) (4) (5) (6) (7) (8) CR Dummy 2 -‐0.0603 -‐0.5776*** -‐0.8579 (0.1737) (0.1621) (0.6022) CR Dummy 3 -‐0.1418 1.0965*** 0.5460*** 1.3216*** (0.3357) (0.4211) (0.1606) (0.4741) Log Illiquidity -‐0.0478 -‐0.1242*** -‐0.0557** -‐0.0879** (0.0407) (0.0398) (0.0245) (0.0422) Underpricing -‐0.0001 -‐0.0002*** 0.0002 0.0001 (0.0001) (0.0001) (0.0002) (0.0002) Log GDP 2.3260*** 2.4357*** 4.3463*** 3.2509* 0.5384*** 0.4083** 1.0942*** 0.7008 (0.8148) (0.7683) (1.1831) (1.8926) (0.1297) (0.1692) (0.2533) (0.4821)
Log Population size -‐10.9597*** -‐7.7734*** -‐17.0501*** -‐14.7125*** -‐0.5088*** -‐0.3142* -‐2.0618*** -‐1.5689**
(3.3806) (2.0877) (2.5392) (3.2465) (0.1405) (0.1687) (0.2806) (0.6208) Services -‐0.1409** -‐0.0066 -‐0.0808** -‐0.0850 -‐0.0371*** 0.0067 -‐0.0234 -‐0.0658* (0.0712) (0.0462) (0.0324) (0.0723) (0.0117) (0.0130) (0.0251) (0.0358) Agriculture 0.2560*** 0.2547*** 0.5367*** 0.4386*** 0.0168 0.0596 0.4593*** 0.2341** (0.0464) (0.0683) (0.0590) (0.1658) (0.0249) (0.0362) (0.0459) (0.0966) Inflation -‐0.0264* -‐0.0165 -‐0.0444*** -‐0.0089 -‐0.0028 0.0022 -‐0.0309* -‐0.0030 (0.0135) (0.0141) (0.0152) (0.0181) (0.0031) (0.0083) (0.0184) (0.0195) Lending Rate 0.0086 -‐0.0434 -‐0.0518 -‐0.0542 0.0015 -‐0.0253** 0.0218* -‐0.0571** (0.0120) (0.0266) (0.0415) (0.0421) (0.0027) (0.0111) (0.0120) (0.0279) Constant -‐2.8619 -‐5.5549** 0.3637 14.2928* (1.9581) (2.6012) (5.1838) (7.8105)
Country fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
N 715 418 159 124 715 418 159 124
Overdispersion test
Alpha value 1.30E+04 6042.20 5612.56 1060.02
22
Panel B: 2004-2014 Dependent Variable: Number of IPOs
Poisson Regression Negative Binomial Regression
(1) (2) (3) (4) (5) (6) (7) (8) LR Dummy 1 0.3022** 0.1091 0.4512* 0.4479* (0.1330) (0.0935) (0.2387) (0.2495) LR Dummy 2 0.6225** 0.7458*** 0.2654 0.6664*** (0.2434) (0.1364) (0.1926) (0.2511) LR Dummy 3 -‐0.2326 -‐0.1923 -‐0.3122 -‐0.7311** (0.1617) (0.1242) (0.2460) (0.3094) Log Illiquidity -‐0.0487* -‐0.0439*** -‐0.0858*** -‐0.0614** (0.0294) (0.0132) (0.0294) (0.0307) Underpricing 0.0010*** 0.0006* -‐0.0010* -‐0.0011* (0.0003) (0.0004) (0.0006) (0.0006) Log GDP 2.4950** 2.2645*** 2.0744*** 2.7664*** -‐0.3174* -‐0.3142 -‐0.8864*** -‐0.8615*** (1.1326) (0.8111) (0.2598) (0.3238) (0.1891) (0.2412) (0.2465) (0.3258)
Log Population size -‐1.3760 -‐0.5398 -‐0.4263 -‐0.7728 -‐0.2307 -‐0.2765 -‐0.0234 0.0970
(1.8952) (1.5638) (0.5865) (0.7116) (0.1968) (0.2374) (0.2389) (0.2887) Services 0.0463 -‐0.0077 0.0466*** 0.0002 -‐0.0253** -‐0.0390*** -‐0.0219 -‐0.0194 (0.0668) (0.0556) (0.0111) (0.0123) (0.0129) (0.0147) (0.0147) (0.0193) Agriculture 0.0552 -‐0.1384 -‐0.0837** -‐0.0177 -‐0.0819** -‐0.0914** -‐0.1359*** -‐0.1303** (0.1358) (0.1055) (0.0367) (0.0445) (0.0365) (0.0419) (0.0487) (0.0525) Inflation -‐0.0583*** -‐0.0452* -‐0.0528*** -‐0.0545*** -‐0.0063 0.0065 -‐0.0105 -‐0.0082 (0.0196) (0.0235) (0.0075) (0.0086) (0.0142) (0.0156) (0.0151) (0.0171) Lending Rate -‐0.0595** -‐0.0635*** -‐0.0412*** -‐0.0610*** 0.0003 0.0000 0.0013 0.0038 (0.0231) (0.0206) (0.0083) (0.0089) (0.0123) (0.0112) (0.0102) (0.0113) Constant 9.0167*** 11.2892*** 21.5023*** 18.2166*** (2.7909) (3.4158) (3.8477) (4.7457)
Country fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
N 400 359 262 233 400 359 262 233
Overdispersion test
Alpha value 3645.83 2768.77 3475.60 2035.95
23
In addition, there seems to be evidence that moving from very low to low and from low to average creditor rights increases the number of IPOs. However, moving from already high to very high creditor rights decreases the number of IPOs, at least in the second sample period. This result is actually in line with both theories on creditor rights that were discussed before. Moving from very low to average creditor rights increases the amount of lending in an economy (Djankov, McLiesh and Shleifer, 2007) and enables firms to grow more quickly. This quicker growth results in more plausible IPO candidates. However, very high creditor rights start to reduce innovation and entrepreneurship, by discouraging leverage and investment (Acharya and Subramanian, 2009; Lee et al, 2011). Therefore, very high creditor rights start to reduce the number of IPOs.
Furthermore, the effect of creditor rights on the number of IPOs is also economically significant. Moving from low to average or from average to above average creditor rights increases the number of IPOs by a range of 44% to 130%, depending on the sample period. This huge jump in IPO activity potentially constitutes an important policy implication for governments that wish to liven up their stock markets. Furthermore, the effect of market liquidity on the number of IPOs is also economically significant, although much less pronounced. Based on the estimates in Table 4, a one percent increase in the relative liquidity results in a 0.05%-0.09% increase in the number of IPOs. All in all, the results provide strong reasons to believe that both the strength of creditor rights and the level of market liquidity are important determinants of IPO activity.
5.4 The Determinants of the Average Size of IPOs
The third and final aim of this paper is to investigate whether market liquidity and creditor rights can further impact economic growth by influencing when companies decide to go public, as measured by the average size of IPOs. Therefore, logarithm of the latter is regressed on measures of market liquidity, creditor rights and several control variables and the results are presented in Table 4.
The average size of IPOs may not be the perfect measure of when companies decide to go public, as some countries are more likely to have very large IPOs, which substantially increase the average, than other countries. Nevertheless, when controlling for the GDP, population size and development stage of a country, the results should provide an indication of whether the variables of interest have any impact on the timing when companies choose to do an IPO. For most of the regressions in Table 5, the Hausman test provides assurance that
24 the error terms are not correlated with the country-fixed effects. Thus, the analysis concentrates on results in columns 1 to 3, which are based on the random effects model. The results themselves provide little reason to believe that market liquidity impacts the average size of IPOs, as the coefficients of the illiquidity ratio are not significantly different from zero in any columns.
The effect of creditor rights on the average IPO size is less clear. Results from the first sample period provide evidence that moving from very low creditor rights to low creditor rights significantly increases the average IPO size. Further improvements in creditor rights however do not seem to significantly impact the average IPO size. However, results from the second sample period provide evidence of exactly the opposite. Improving creditor rights when they are very low seems to decrease the average IPO size, while further improvements in creditor rights again have no significant impact on IPO size. There are at least three possibilities that can explain this puzzling result. Firstly, creditor rights may be correlated with some omitted variables that influence the average size of IPOs and have no meaningful effect on their own. Secondly, the effect of creditor rights on the average IPO size could differ for different size quantiles of IPOs. For example, stronger creditor rights could decrease the average size of the smallest IPOs while increasing the average size of the largest IPOs. If the two effects balance, then creditor rights will have no impact on the average size
25 Table 5: Creditor Rights, Liquidity and Average IPO Size
Panel A and B present the regression results for the first and second sample period respectively. Columns 1-3 are based on the random effects model and columns 4-6 are based on the fixed effects model. When the p-value of the Hausman test is below 0.05 one should look at the fixed effects results. Otherwise, random effects model provides more efficient estimates. All regressions include year-fixed effects and a constant term, which are not reported. Standard errors are clustered across countries and reported in parentheses; *, ** and *** indicate significance at 10%, 5% and 1% level respectively.
Dependent Variable: Log of Average IPO Size
Panel A: 1989-2003 (1) (2) (3) (4) (5) (6)
CR Dummy 1 0.5070** 0.7360*** Omitted Omitted
0.2562 0.2266 -‐ -‐
CR Dummy 2 -‐0.3086 -‐0.3191 0.6288** 0.2722
(0.3282) (0.2817) (0.3050) (0.3514)
CR Dummy 3 0.0948 -‐0.0704 -‐0.1872 -‐0.2685
(0.3130) (0.3007) (0.6460) (0.8387)
CR Dummy 4 -‐(0.1881) -‐(0.2900) Omitted Omitted
(0.3402) (0.3105) -‐ -‐
Log Illiquidity -‐0.0321 -‐0.0391 -‐0.0516 -‐0.0482
(0.0397) (0.0383) (0.0496) (0.0512)
Log GDP 0.4527** 0.1706 0.1583 2.579* 1.9128 2.1267
(0.2176) (0.1736) (0.1775) (1.3008) (1.5502) (1.8242)
Log Population size -‐0.1816 0.0824 0.145 -‐1.8636 -‐3.1846 -‐3.7184
(0.2053) (0.1696) (0.1833) (3.3747) (3.5849) (3.8683) Services -‐0.013 0.0115 0.0137 0.0559 0.0837 0.0897 (0.0129) (0.0130) (0.0128) (0.0744) (0.0629) (0.0782) Agriculture -‐0.0539 -‐0.0442 -‐0.0548 0.1445 0.0012 0.0064 (0.0504) (0.0372) (0.0436) (0.1391) (0.0878) (0.0932) Inflation -‐0.0210*** -‐0.0189 -‐0.0184* -‐0.0159* -‐0.0153 -‐0.0152 (0.0072) (0.0121) (0.0107) (0.0087) (0.0161) (0.0167) Lending Rate 0.0148** 0.0018 -‐0.0023 0.0096 -‐0.0286 -‐0.0301 (0.0066) (0.0082) (0.0078) (0.0077) (0.0359) (0.0361) Country fixed effects No No No Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes Yes
N 486 333 333 486 333 333
Hausman test statistic 5.09 15.58 30.74 p-value 0.9547 0.7424 0.1017 Panel B: 2004-2014 (1) (2) (3) (4) (5) (6) LR Dummy 1 -‐0.7802** -‐0.9495*** -‐0.9086* -‐1.0066* (0.3223) (0.3331) (0.5179) (0.5142) LR Dummy 2 -‐0.1358 0.1557 0.0812 0.2928 (0.2929) (0.2625) (0.3635) (0.3670) LR Dummy 3 0.2709 0.0121 0.2861 0.2996 (0.4412) (0.4805) (0.7776) (0.7946) Log Illiquidity -‐0.0934 -‐0.0904 -‐0.0144 -‐0.0314 (0.0605) (0.0661) (0.0723) (0.0723) Log GDP 0.6241** 0.1692*** 0.6036** 4.2282** 3.8477** 5.1650*** (0.2648) (0.0420) (0.2697) (1.6880) (1.6655) (1.7852)
Log Population size -‐0.2589 0.0799 -‐0.2395 -‐0.0701 -‐1.0212 -‐1.8073
(0.2327) (0.1030) (0.2280) (1.8440) (1.7113) (1.8792) Services -‐0.0065 -‐0.0061 -‐0.0045 0.1412* 0.1733** 0.1654** (0.0140) (0.0143) (0.0140) (0.0762) (0.0751) (0.0755) Agriculture 0.0247 -‐0.0523* 0.0234 0.6157*** 0.6364*** 0.7216*** (0.0497) (0.0275) (0.0465) (0.1959) (0.2283) (0.2243) Inflation 0.001 -‐0.0053 -‐0.0032 0.0568 0.0239 0.0196 (0.0389) (0.0427) (0.0444) (0.0383) (0.0424) (0.0419) Lending Rate -‐0.0165 -‐0.0066 -‐0.0116 -‐0.1384** -‐0.1293** -‐0.1203** (0.0239) (0.0263) (0.0209) (0.0566) (0.0610) (0.0573) Country fixed effects No No No Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes Yes
N 369 342 329 369 342 329
Hausman test statistic 41.74 42.1 22.42 p-value 0.0007 0.0004 0.2138
26 of all IPOs. Thirdly, the effect of creditor rights on the average size of IPOs could simply have changed between the two sample periods. While the latter seems an easy explanation, it is hard to find economic justification why this may be the case. The other two potential explanations are explored further in the robustness checks. Nevertheless, the results presented in Table 4 do not provide an indication that market liquidity or the strength of creditor rights has a clear and significant impact on the average size of IPOs.
6. Robustness Checks
Existing literature provides some reasons to doubt whether the results presented earlier actually represent a causal relationship and are not simply driven by omitted variable bias. La Porta et al (1997) and Djankov et al (2008) both find that stronger investor protection is associated with higher stock market development in general and more IPOs in particular. Thus, there is strong evidence that better investor protection leads to more developed financial markets. Moreover, La Porta et al (1997) and Djankov, McLiesh and Shleifer (2007) find that also creditor rights are associated with more developed financial markets, at least when measured by aggregate private lending. Therefore, both investor rights and creditor rights seem to affect broad financial development. Furthermore, it is also clear that investor rights have a direct impact on the number of IPOs and on stock market development. However, if investor rights and creditor rights are correlated, which seems plausible, any regression of liquidity or number of IPOs on creditor rights would be biased when investor rights are not included in the regression. To investigate this possibility, all the regressions reported earlier are rerun with investor rights included as a control variable. As data on investor rights is only available from 2004 onwards, these regressions are only estimated for the second sample period. To save space, the results of these regressions are presented in the Appendix in Tables A3, A4 and A5.
The conclusion from the regression of market liquidity on creditor rights does not change when investor rights are included as an explanatory variable. If anything, the effect actually becomes both economically and statistically more significant. Thus, there seems to be strong evidence that very high creditor rights have a profoundly positive effect on market liquidity. At lower levels however, creditor rights do not seem to influence the level of stock market liquidity. Also the results concerning the determinants of the number of IPOs are robust to including investor rights as an explanatory variable. There seems to be still evidence that improving creditor rights at low levels leads to more IPOs and that the effect is reversed for very high creditor rights that tend to reduce the number of IPOs. Nevertheless, it has to be
