The Liberalization of the Brazilian Stock Market,
State-owned Versus Private Companies
How did Brazilian Companies with Different Ownership Structures React Differently on the Liberalization of the Equity Market?
Name: Simon Berger Student Number: 10003442 E-mail: simonberger@live.nl Supervisor: Dr. Rafael Almeida da Matta
Date: 2/7/2015
Presented to the Faculty of Economics and Business University of Amsterdam
In Partial Fulfillment of the Requirements for the Degree of Master of Science in Finance
2
Statement of Originality
This document is written by Student Simon Berger 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
ABSTRACT
This paper examines the difference in effect of the liberalization of the Brazilian equity market on state-owned versus private companies. Several event studies are performed using different lengths of estimation and event windows. In order to explain if the abnormal returns after the liberalization are indeed caused by the announcement of the liberalization regressions are performed using different dummies and variables to control for political reforms, co-movements with foreign equity markets and macro-economic shocks. Finally a different event date is used to check for robustness. Results show that private firms outperform state-owned firms due to the liberalization of the Brazilian capital account. The abnormal returns obtained by the firms is indeed caused by the announcement of the liberalization.
4
ACKNOWLEDGEMENTS
During my bachelor and master studies at the University of Amsterdam I learned a lot about financial markets and stock markets. I learned to think critically and analyze information with caution and question thinks that look to good to be true. The thing that was by far the most influential and shaped me the most during my time as a university student was my exchange to the FGV (Fundação Getulio Vargas) University in Rio de Janeiro, Brazil. During this exchange I learned a lot about the Brazilian economy, its financial markets and complex economic situation. It was the year 2013 and Brazil was still experiencing high economic growth which came together with many financial challenges. One of the things that struck me most was the big influence the government had on the economy, this was a big contrast with the Netherlands were we are used to a more liberal approach by the government when it comes to governmental interference in the economy. When I was back in the Netherlands to finish my master's degree I had no other option than to investigate this subject further for my master thesis.
I would like to thank my supervisor Rafael Almeida da Matta for his enthusiasm and wide knowledge about the working of financial markets. Without his help and assistance I would not have been able to finish this thesis. I would like to thank my parents who always believed in me and supported my decisions and helped me in any way they could, I am forever grateful. Finally I would like to thank two fellow students who helped me a lot. First of all Maarten Tellegen for his wide knowledge of econometrics and his endless patience. Secondly I would like to thank Stefan van den Born who followed the same study course as I did and was always there to give me feedback.
5
Contents
List of Tables ... 5
1. Introduction ... 6
2. Literature Review ... 7
2.1 State-owned versus private ... 7
2.2 Liberalization... 10 2.3 Summary ... 16 3. Methodology ... 17 3.1 Hypothesis ... 17 3.2 Formulas ... 18 3.3 Controls ... 20 4. Data Analysis ... 22
4.1 Liberalization and announcement date ... 22
4.2 Data ... 22
5. Empirical Results ... 25
5.1 Abnormal returns and t-statistics ... 25
5.2 Regression analysis ... 27
6. Robustness Checks ... 33
7. Conclusion ... 39
6
List of Tables
Table 1 Major Political/Economic Events During the Announcement of the
Liberalization of the Capital Account of Brazil (15-5-1991)
Table 2 List of State-owned Companies Traded on the Bovespa Index on May 15th
1991
Table 3 The Different Coefficients and Beta's when Regressing the Return of the
Different Ownership Structures on the Market Return for Brazil for the Announcement of the Liberalization of the Equity Market
Table 4 T-Statistics for the Cumulative Abnormal Returns for the Different Ownership
Structures at the Announcement Date
Table 5 T-Statistics for the Hypothesis that State-Owned Companies Outperform
Private Companies at the Announcement Date
Table 6 Stock Market Reactions to Announcement of Stock Market Liberalization
Table 7 The Different Coefficients and Beta's when Regressing the Return of the
Different Ownership Structures on the Market Return for the Liberalization of the Brazilian Equity Market
Table 8 T-Statistics for the Cumulative Abnormal Returns for the Different Ownership
Structures at the Liberalization Date of the Brazilian Equity Market
Table 9 T-Statistics for the Hypothesis that State-Owned Companies Outperform
Private companies at the Announcement Date
7
1- Introduction
On May 31st 1991 Brazil changed its foreign investments laws and resolution 1832 Annex IV
was introduced (CVM) which gave foreign investors the possibility to buy a majority of shares in a Brazilian company and gave Brazilian investors the possibility to perform transactions in foreign equity securities. The liberalization of a countries equity market is one of its biggest decisions with big consequences (Bekaert, Harvy and Lundblad, 2003) and has been widely investigated. The increasing globalization and liberalization (Stulz, 1999) makes this topic relevant and important. The impact of the liberalization of the equity market is especially influential in economies with a less active stock market like emerging countries (Manova, 2008) because the foreign funds compensate for an underdeveloped domestic equity market. The results of this research could help assess the impact of future liberalizations in other emerging markets like China (IMF, 2004) and Nigeria (Singh, 1998) and thereby contribute to the discussion about liberalization in those countries (Henry*, 2000)1.
Most papers focus on the effect of the liberalization on GDP (Bekaert and Harvey, 2000), this research distinguishes itself by looking at the domestic financial market and dividing the sample into two groups: state-owned and private companies. State-owned companies are defined as companies being either fully or partially state-owned (Boardman and Vining, 1989). This paper could add to the discussion of liberalization the difference in effect of liberalization on the different ownership structures of companies and thereby be used in the political debate that is very relevant in Brazil right now (Carvalho, 2005). The ownership theory of companies argues that private companies perform better than state-owned companies, empirical evidence is indecisive about this subject (Boardman and Vining, 1989). This puzzle has not been widely investigated and leaves room for further deepening.
In Brazil a lot of companies are state-owned, either fully or partially, or receive aid from state banks like BNDES (Banco Nacional de Desenvolvimento Econômico e Social). In Brazil a lively discussion is going on whether the market should be more liberalized (Carinhato, 2008 and Marquetti, 2003). Proponents of opening domestic equity markets to foreign investors
state that this leads to investment booms and more export (Manova, 2008 and Henry*, 2000)
1 Two papers of Henry are used. The first paper is: “Stock Market Liberalization, Economic Reform, and Emerging Market
Equity Prices”. The second paper is: “Do Stock Market Liberalizations Cause Investment Booms?”. The * refers to the
8
while opponents argue that it does not lead to growth in capital markets in the long run (Levine and Zervos, 1998). The consensus about the effects of liberalized equity markets is thus not unanimous. Leal and Rêgo (1997) for example argue that it did not lead to a change in the cost of capital on the Brazilian equity market while Bekeart and Harvey (2000) do find a reduction. This made me formulate the following research question:
“To what extent did the announcement of the liberalization of the Brazilian equity market impact private and state-owned companies differently?”
2- Literature Review
The related literature can be divided in two parts: literature concerning the differences in profitability between state-owned and private firms, and literature concerning the effect of liberalizing the capital account. Both will be discussed in the following two paragraphs, the third paragraph will be a summary of all the literature together.
2.1– STATE-OWNED VERSUS PRIVATE
Megginson and Netter (2001) argue that privatization has a long history and is still a relevant topic. Already in ancient Greece, China and the Roman empire many industries were state-owned. In western Europe a lot of firms were privatized after World War II like several firms in Great Britain for example under Thatcher and in the Federal Republic of Germany under the Adenauer Government while some key industries like petroleum and electricity were kept under government control. The latest big IPO's of nationalized companies were in the 1990's in France, Germany, Spain and big parts of Latin America. On the other hand China has only slightly privatized companies just like many other countries in Africa. Countries like Nigeria and several Asian countries still have a lot of state-owned companies. In Brazil a lot of big companies were privatized under the Cardosso government with the sale of CVRD and Telebras in 1997 and 1998 although there was a lot of political resistance. Over 1 trillion US dollars has been collected worldwide till 1999 due to the IPO's of privatized firms. The percentage of GDP that can be attributed to state-owned firms has declined from 8.5% to 6% for developed countries from 1984 till 1991. In general, more and more companies are privatized worldwide. For underdeveloped countries the decrease was even more, a reduction from the highest point of 16% to 7% in 1995 of GDP that came from state-owned firms.
9
The theory behind the discussion of privatization mainly focuses on the discussion of efficiency of the state and the efficiency of private ownership according to Megginson and Netter (2001). What should be the role of the government in the economy? Proponents of government intervention state that the government should intervene in the case of public goods, certain externalities like pollution and natural monopolies. Opponents argue that markets are most efficient when there is no intervention at all. Privatization is in essence a response to the failure of state ownership. Nationalization in turn, is a response to the failure of the free market. One of the problems with state-owned firms is to define their goal. A lot of times they serve different goals besides profit maximization. This is one of the reasons that profit maximization theories for corporate companies have grown in popularity in recent years: there goal is clearly defined. Social welfare is a difficult thing to measure and the aim might not be economically efficient but still favorable for the government. Megginson and Netter (2001) argue that one of the difficulties in comparing the two ownership structures lies in the reason why a certain company is either owned or not. A company might be state-owned because of market failure in his industry, this affects its potential profitability in spite of his ownership structure. This is the case when the ownership structure in endogenous. Another problem is that there is more data available for more developed countries than for less developed countries, this could lead to sample bias. None of the papers controls for the effects of monopoly power after the privatization. The increased profitability could thus be less attributable to an increase in efficiency and more to the new status of monopolist. After analyzing extensive past research (38 studies) Megginson and Netter (2001) conclude that the majority of the researchers claim that private firms are more profitable than state-owned firms.
Boardman and Vining (1989) investigate the puzzle that private companies empirically perform worse than state-owned companies according to past research. Which is a contradiction with what Megginson and Netter (2001) have found in their past research and in contradiction with the commonly accepted ownership theory of companies which argues that private companies perform better than state-owned companies. The ownership theory argues that because the ownership is not transferable in state-owned companies, specialization of ownership does not happen what in turn means that owners incentives to monitor management is reduced. Increased competition incentives should make private companies perform better than public companies. On the other hand it is argued that the possibility to be taken over also reduces management's incentives in private firms. They investigate 500 non-American
state-10
owned enterprises, private corporations and mixed enterprises. They give a good distinction between state-owned and private companies by saying that the difference lies in the nontransferable ownership in the case of public firms. Past investigations have argued that mixed enterprises perform better than state-owned, but not as good as private companies. Past research looks a lot at companies that have a natural monopoly or operate in a duopoly and not at firms operating in a competitive environment, research of the difference in performance between the different ownership structures in a competitive environment is scarce. Different profitability measures are used like return on equity, return on assets, return on sales and net income, they are regressed using ordinary least squares. Boardman and Vining (1989) find that private companies outperform mixed and state-owned firms. Mixed and state-owned perform the same when it comes to profitability in their research. Their results agree with the ownership theory and contradicts past research. The results could be biased when bad performing companies were nationalized in order to prevent it from making further loses. The ownership structure would then be correlated with the profitability. Boardman and Vining (1989) state that most of the companies that were nationalized in their sample were not nationalized because of bad performance and therefore their sample does not face this problem. Their paper is one of the only ones that does not focus solely on North America firms. However, their research does not look at the effect of an equity market liberalization on the return on equity. This research will make the connection between ownership structure and liberalizing equity markets in an emerging market.
The general consensus is that state-owned firms perform less than private firms due to the focus on political and social goals which undermine the maximization of profit according to Dewenter and Malatesta (2001). Monitor incentives are less because there is less control by shareholders with state-owned firms. On the other hand the same agency problems exist for private firms as for state-owned firms. Private shareholders own small parts of the firm which makes monitoring costly. Just like Megginson and Netter (2001), Dewenter and Malatesta (2001) claim that so far empirical results, as well as the theory that underpins them, has been mixed about which ownership structure is more profitable. Dewenter and Malatesta (2001) perform a similar investigation as Boardman and Vining (1989) only with a larger sample and longer time period. Their sample is three times as big as the one by Boardman and Vining (1989) and they control for business cycles. They find that state-owned firms underperform compared to private firms when comparing accounting output. An interesting result is that they find a significant increase in profitability in the three years before the privatization when
11
the government is gradually reducing its ownership. They argue that this could be a sign that the government is interfering in the accounting data in order to make the company look more interesting for future investors. One of the results of their investigation is also that labor intensity decreases after the privatization which supports the theory that government companies are less efficient when it comes to the use of labor. Dewenter and Malatesta (2001) also argue that state-owned firms use more debt because the equity market is less accessible for them and a lot of times they can borrow from the government at favorable interest rates. The return on sales of private firms is twice the value as that is for government firms. Also the return on equity is significantly higher for private firms than for state-owned firms. Also the ratios employees to sales and employees to assets are higher for state-owned firms. They control for the business cycle, geographic location, firm size and industry membership while performing a multivariate regression and obtain the same results as when they compared the accounting ratios.
2.2- LIBERALIZATION
The most common way to compare what the effect of liberalizing the equity market has on the economy is to look at the cost of capital and the return on equity. The cost of capital might decrease due to diversification benefits between foreign investors and the investors in the particular emerging market according to international asset pricing models says Henry (2000). When the cost of capital decreases, holding the future cash flow constant, this will lead to an increase in the equity price index in the country of concern. Because of this decrease in the cost of capital an increase in the amount of investments should be observed because the reduced cost of capital makes some investment opportunities that before had a negative NPV, now have a positive NPV. He finds in his paper that the equity index of an emerging country increases by 3.3 percent per month in a window of eight months before the equity is liberalized due to risk sharing between local and international investors. This implies a revaluation of the equity of 26 percent in total and 6.6 percent in the month leading up to the liberalization. As methodology an event study is used while controlling for co-movements with the world economy and economic policy reforms to see what the effect of the liberalization of an equity market is on the cost of capital for that country. A new data set is created containing all the policy reforms relevant for the emerging markets that will be investigated. Not controlling for these reforms would result in biased results. This is because the liberalization of an equity market most of the times coincides with other important political reforms like macroeconomic stabilization, trade liberalization, privatization and
12
changes in exchange controls. Henry looks at 12 emerging countries and uses an eight month event window which is considered long and may overstate the effect of the liberalization if the government chooses to liberalize the equity market in times of exceptional high returns. To control for this Henry also uses five-month, two-month and one-month (implementation only) all of which result in significant abnormal returns. Another liberalization date is used than the one in this paper. Henry uses the first country fund introduction which was March 1988. In this paper the announcement of the official liberalization date is used. Henry is one of the only choosing the first fund introduction as starting date. It is argued that the revaluation of the equity starts as soon as the news spreads that the equity market will be liberalized. Because this information spreads gradually and there is probably some news leakage leading up to the event a long event window is chosen. The paper unfortunately does not distinguish in ownership structure like this paper will do.
If the correlation of the emerging economy with the world market increases after liberalizing the equity market a decrease in the cost of capital is achieved through investors bidding up local prices in the search for diversification. Bekaert and Harvey (2000) use a cross-sectional time series to assess the impact of the liberalization of the equity market on the cost of equity, betas and volatility with the world market for 20 different emerging markets. They argue that the cost of capital decreases by 5 to 75 basis points. They control for macro-economic influences like the capitalization of the countries stock market relative to its GDP and the size of the trade sector. They find an increase in volatility due to the liberalization as well as an increase in the beta and correlation with the world market leading to a lower cost of capital. The paper uses a different methodology and does not focus on the difference between state-owned and private companies in contradiction to this paper. It compares emerging markets with the world market instead of looking at the difference in effect of the liberalization between sample groups with a different organizational structure within one country. Although they find an effect, it is small (less than 1 percent) and might be biased because they tread the liberalization as an exogenous effect. The question is whether this is justified because governments will only liberalize the stock market when it is in the best interest of the country (economic wise).
The cost of capital consists of two parts, the equity premium and the risk free rate which results in three factors that can cause the cost of capital to decrease after a market
13
inflow and thus might reduce the risk free rate. Second, it may lead to diversification because of risk sharing between foreign and domestic investors which reduces the equity premium. Thirdly, increases in capital inflows might increase liquidity and thus the equity premium. Henry* (2000) argues that liberalizing an equity market almost always reduces the equity premium, but it may also increase the risk free rate and therefore causing an overall increase in the cost of capital. The effect on the risk free rate depends on two factors. One of the them being what the liberalization says about capital outflow of domestic investors and the second one is how high the countries domestic risk free rate is compared to the world risk free rate. If the domestic risk free rate is lower than the world risk free rate and capital is allowed to flow out of the country, capital will flow out of the country till the two rates are equal. This may cause the risk free in the emerging country to increase and when this effect is bigger than the decrease of the equity premium, the total cost of capital increases. Henry* (2000) gives a theoretical explanation on why past research has found only temporarily growth changes in investments. He uses the “closed economy Solow model (1956) in a steady state” which looks at the question from a macro-economic point of view instead of a financial view. This model states that the labor force and the capital market experience the same growth rate. After foreign capital flows into the country, the cost of capital is reduced. The response of agents will be that they reduce the marginal product of capital till it is equal to the cost of capital. The only way this can happen is when the capital inflow experiences a higher growth rate than the labor force. After this has happened the growth rate of the capital will decrease again to its old value. Another reason why the cost of capital might not decrease after the capital account liberalization is due to one of the assumptions of the model. Namely that the output of the emerging country does not change after the liberalization. When you consider investments
will increase after this event, this assumption might not really be realistic. Henry* (2000) tries
to test this theory in his paper by performing an event study and comparing growth rates of non liberalization periods with the growth rates during capital account liberalizations. He finds that opening up equity markets leads to private investment booms. Out of the 11 emerging countries investigated, 9 experience significant temporarily growth when it comes to external investment within 1 year after the liberalizing. The investment growth rate increases by 22 percent within 3 years. For Brazil he found a growth in investment of respectively 31,3%, 14.5%, 40.2% and -16.1% in the four years following the capital account liberalization. He argues that a reduction in the cost of capital leads to higher growth rates in investments because of higher stock prices. He controls for world business cycles, economic reforms and domestic characteristics. Because these booms could be caused by exogenous
14
decreases in the cost of capital or a change in the future marginal productivity of foreign capital, it cannot be concluded that there is causality between these investment booms and market liberalizations. History show that politicians like to liberalize the equity market when good news is expected which results is high stock prices, making the liberalization endogenous.
Stulz (1999) finds that the cost of capital decreases because of globalization after analyzing past research. First of all, investors demand lower equity premium for their market portfolio to compensate them for the risk they face because of diversification and secondly, agency costs decrease. Stulz (1999) finds significant evidence for this although the results are lower than expected according to the theory. When the cost of capital decreases, the equity index in that particular country increases and it will experience more investment from foreign investors. Foreign investors introduce value to a company by bringing their own expertise and monitor possibilities which reduces the cost of capital by lowering agency cost problems. Globalization also reduces the costs for the firm to raise money because there are now more investors willing to invest. Globalization does not only introduce positive effects. A firm that is well monitored and governed in its local market might be able to undertake unfortunate investment decisions in a foreign market. The decisions the firm makes outside the local market might not be beneficial for the investors in the local market. One of the reasons why the cost of capital does not decrease as much as theory would suggest according to Stulz (1999) is because investors are still biased to their local market when it comes to selecting their portfolio and they thus do not use all the benefits that can be obtained from diversification. When capital markets are closed investors face all the risk the local economy is facing and investors want to be compensated for this. When the capital market opens up investors are able to possess a portfolio that is more like the world portfolio and thus they face less risk according to asset pricing theories. Stulz (1999) discusses different methodologies to investigate the effect on the cost of capital. He argues that times series methodology only works well when you assume that the risk premium will be relatively stable over long period of time. Empirical evidence contradicts this assumption. Because the value of equity is calculated using the discount rate (which contains the equity premium), a reduction in the equity premium will result in a higher valuation of the equity. It then looks that the equity premium is high because of the high returns of the firms securities. Stulz (1999) discusses also the fundamental problem faced by event studies: as soon as the news of the liberalization is know, this information will already be incorporated into the values of the equity and the
15
expectations of the investors. This is why recent research has only found a small decrease in the cost of capital instead of a large one like the theory would suggest. It is argued that when the market expects a liberalization with 90% and the decrease in the cost of capital is expected to be 1000 basis points. The noticeable effect on the liberalization will only be 0.1 * 1000 = 100 basis points. This is one of the reasons we take the announcement of the liberalization of the equity market as the main event in this research. The theory also assumes that the emerging market is immediately integrated and correlated with the world portfolio, this is not a realistic assumption. The actual impact of the capital account liberalization might only be noticeable later. Even after a countries changed its laws and opened up this equity market, there are still restrictions in place in most of the cases. This limits the effects of the liberalization of the equity market on the cost of capital.
Countries with a less active stock market benefit from liberalizing equity markets because the new capital that flows into the country compensates for an underdeveloped domestic equity market and makes the domestic investors demand a lower equity premium for their market portfolio because they can diversify as argued by Manova (2008). Credit constraints are the big determinant when it comes to international trade. Opening a countries financial market to the outside world should increase the amount of money available for industries in the emerging country, causing the cost of capital to decrease. Because the timing of a liberalization in an emerging country is the result of politics, this event should be endogenous for producers and exporters Manova argues. She thereby disagrees with Bekaert and Harvey (2000) who say that a liberalization might not be an endogenous event because it's more likely for a country to liberalize when it's stock market is priced high. The question should be asked if this assumption does not lead to biased results. Although Manova her paper builds on past results about reductions in the cost of capital after opening a countries equity market, her paper mainly focuses on the effect on trade instead of the effect on returns. To look at the change in exports an events study in the 1980-1997 period is conducted in 91 emerging countries and 27 different industries while controlling for country-sector fixed effects. An event window is chosen after the specific event as well as before the event like Henry (2000) did ranging from 1 to 5 years. The exports increase significantly within three years after removing the countries barriers around international equity ownership. Unfortunately she does not look at the difference in performance between state-owned companies and private companies and the event windows are long compared to other research.
16
Kim and Singal (2000) argue about the importance of free capital markets although exceptions exist were economic growth is accompanied with tighter state ownership like in China and Chile. Malaysia temporarily closed its capital markets in 1998, a trend that is going on in more Asian countries. It is argued that capital account liberalization is positively correlated with economic growth in the long run. Kim and Singal (2000) warn for the dangers of so called “hot money”. Money that flows into emerging countries because of favorable investment opportunities, but leaves the country with the same speed when there are changes in interest rates and economic outlooks. By opening your equity market you become vulnerable to influences from outside, which are not always favorable. Foreign equity might introduce more volatility to the local market. The domestic currency might appreciate because of capital inflow which is bad for emerging countries focused on export. Kim and Singal (2000) find a temporarily increase in the returns of the stock market directly after the capital account liberalization, volatility stays constant on average in the sample of 20 emerging countries. Just like Stulz (1999) they state that the effect of the liberalization will not show the real difference of the liberalization because of leaked news and the anticipation of the capital account liberalization. They investigate a rather long period of 10 years, 5 before the opening of the market and 5 afterwards. Kim and Singal (2000) do not use event studies but simply compare the returns in real dollar terms between the different countries and they do not control for economic reforms or macroeconomic influences. For Brazil they find an excess return of 1.52% combined with a volatility of 16.48% in a period starting from 1976 till 1996. Petrobras common stock is not available for purchase as is the voting stock of banks in Brazil which decrease the possible benefits that can be obtained from diversification. Returns on general increase for 12 months and then start to fall again and reach their original value after about 24 months on average. The average total return difference is 0.21%. Kim and Singal (2000) state that the integration in the world economy should result in lower volatility and also lower equity premiums in the market portfolio due to equilibrating effects with the world economy.
The effects of resolution 1832 Annex IV on the Brazilian economy is analyzed by Leal and Rêgo (1997). They find no effect on the return after the liberalization on the IBovespa exchange which is in contradiction with my expectation and the findings of other authors like Bekaert and Harvey (2000). They also did not notice a change in volatility, which is in
contrast with what they expected. They show that after the 31st of May 1991 till 1995 a total
17
significant increase and confirms the results of Henry* (2000). The only change they found was in the increased liquidity of the equity market in two investigation periods, one before the liberalization (1988-1991) and one afterward (1992-1993). Their methodology is rather simple, they only look at total volume and change in average returns.
A paper written by Bekaert, Harvey and Lundblad (2003) commissioned by the Federal bank of St. Louis investigates the effect of liberalizing an equity market on the overall GDP level of 30 emerging countries including the country of Brazil. They calculate the growth in GDP using a pooled time-series cross-sectional regression. They distinguish between the concepts of liberalization and integration. Liberalization meaning dropping any laws that prohibit the movement of capital flows that go to international investors. Integration meaning that the local capital market gets integrated into the world economy, in other words, that the correlation with the world economy increases. That is why they use a special measure to calculate if the date of the liberalization of the equity market really led to the integration of the equity market. The exact date of the integration of the Brazilian equity market is calculated through a ratio using the indexes of the IFC2 created by Edison and Warnock (2003). When the IFC investible index divided by the IFC Global index reaches one, liberalization caused integration. The IFC investible index represents a portfolio of domestic securities that are accessible by foreign investors, the IFC Global index represents the overall market portfolio for Brazil. This method will be used in this paper as well. They find a larger growth when it comes to GDP of 0.84 percent on average after the liberalization of the equity market. The paper however does not distinguish by ownership structure and does not focus on the effect on the rate of return.
2.3- SUMMARY
As a summary of the above research we can state that the majority agrees with the statement that private firms, on average, outperform state-owned firms although the support is not unanimous (Boardman and Harvey, 1989). This research will make a connection between the two subjects addressed above: the liberalization of an equity market, and the difference in performance between state-owned and private companies. This has not been widely
2 IFC (International Finance Corporation, part of the World Bank Group) provides financing to the private sector of
developing countries by international financial institutions through the means of funds and indexes. IFC investible index: represents a portfolio of domestic securities that are accessible by foreign investors, IFC Global index: represents the overall market portfolio for Brazil.
18
investigated and will therefore be meaningful and add to the discussion about privatizing companies and liberalizing capital accounts. Brazil is chosen as country of interest to investigate the effects of the liberalization on the different organizational structures because the debate in Brazil about privatizing companies is still very relevant and has supporters on both sides of the discussion. Most of the past research finds a decrease in the cost of capital and an increase in the return of the equity index of the emerging countries investigated. Theory though, does not let us believe that opening the equity market to foreign investors will always lead to a decrease in the cost of capital. As Henry* states the cost of capital might increase when the risk free rate of the emerging country is lower than the world risk free rate and capital is allowed to flow out of the country, the domestic risk free increases due to the capital outflow. What comes out in most of the papers is that the effect of the liberalization might be underestimated due to leaked news about the liberalization which already gets incorporated in the expectations of the investors and therefore will reduce the reduction in the cost of capital as observed close to the date of liberalization (Henry 2000, Kim and Singal 2000 and Stulz 1999). Another problem is the concept of endogeneity. Is the liberalization of the capital account an exogenous event like Bekaert and Harvey (2000) and state, or is it in fact endogenous because politicians will only sell state-owned firms when the market is favorable like Manova (2008) states. Another reason why the cost of capital does not decrease as much as expected by the theory, is that even after the liberalization most investors have a bias to their home country portfolio (Stulz, 1999). They therefore forego part of the diversification possibilities.
3- Methodology
3.1- HYPOTHESIS
Liberalization can lead to market integration, it is commonly accepted that this is a good development for real GDP (Bekaert and Harvey, 2000) and gives foreign investors the possibility of diversification but what is the effect on the domestic financial sector? This research will try to analyze what the effect of this important decision is on the Brazilian Bovespa exchange located in São Paulo. Especially it will look at the difference in effect between state-owned and private companies.
The hypothesis will state that private companies will outperform state-owned companies after the announcement of the liberalization of the equity market, because this is advocated by the
19
theory of ownership (Boardman and Vining, 1989). Due to more efficiency and a higher level of competition private companies are expected to gain more from the increased international control after Brazil opens his capital account for foreign investors. State-owned companies will gain less from the possible international investments because the gains these investors can achieve will stay limited due to the control these companies will experience by the Brazilian government even after the liberalization (Megginson and Netter, 2001).
Hypothesis: Private firms yield a higher rate of return than state-owned firms due to the announcement of the liberalization of the equity market in Brazil.
3.2- FORMULAS
As methodology an event study will be used to look at the effect of the liberalization of the equity market. An event study will best capture the effect of the liberalization on the equity price index (Stulz, 1999). An increase in the price equity index will lead to a decrease in the cost of capital according to asset pricing models holding future cash flows constant. Taking in mind that we consider the market to be efficient, the announcement of the new information (opening up the equity market for foreign investors) should show the valuation effect of this event (Kothari and Warner, 2006). After the integration with the world portfolio, the price of risk will decrease in the emerging country due to diversification benefits with the world portfolio who has a lower price of risk. This in turn will lead to a decrease in the equity premium holding future cash flows constant and therefore a decrease in the cost of capital. Because of this fall in the cost of capital of the emerging country its equity price index will be re-valuated to a higher level (Henry, 2000).
The goal is to identify the abnormal return that was realized due to the event, the abnormal
return is represented by . The event window will be centered around the 15thof May 1991,
which was the date the liberalization was announced (Wang, 2006). Because the companies traded at the Bovespa index did not all have the same liquidity, not every stock was traded on
the 15th of May 1991 when the liberalization was announced. Because the state-owned group
20
of 10 days) was therefore selected when there was no trading taking place on the 15th of May
1991 for this group in order to keep this group substantial and representative. Because of this illiquidity we will use the amount of trades conducted instead of days in the event and estimation windows. In reality this will approximate the amount of days because only sometimes there was not a recorded trade for a specific stock. Different event and estimation windows will be investigated in order to check for robustness. We will start with an event window ranging from 10 trades before the event, till 10 trades after the event as is generally accepted with event studies (Campbell, Lo and Mackinlay, 1997) and an estimation window ranging from 220 trades before the event till 11 trades before the event. Then a shorter estimation window (11 trades before the event till 90 trades before the event) will be used to see if this affects the calculation of the beta's of the two sub groups. Then the length of the event windows will be gradually increased from 30 to 60 and finally 90 trades to see the difference in effect in the short run and long run (Kothari and Warner, 2006). We will look at the period leading up to the announcement and the period following the announcement (Cowan, 1992). In the estimation period before the event window the beta and the constant for
the CAPM model for the Bovespa index will be defined by regressing on (the
Bovespa index) for the different stocks in the two different sample groups. Different estimation periods ranging from 30 to 90 trades will be used to control for business cycles in the returns.
(1)
The CAPM model will be used (formula 1) to predict what the cross sectional normal return should be during the event window and then this will be compared to the actual returns to calculate the abnormal returns of the equally-weighted portfolio. An equally weighted portfolio is used because the state-owned firms in Brazil are bigger in market capitalization than their private counterparts and using an equally weighted market model should avoid biased results. Aggregate abnormal returns will be calculated according to aggregation across time for each sample group and the average across firms with formulas 2 and 3 (Kothari and Warner, 2006).
21
(3)
Finally a test of significance will be performed according to formula 4 and then we will test if the cumulative abnormal returns of the private firms group did indeed outperform the state-owned firms group using formula 5 (Welch t-test):
(4) (5) 3.3- CONTROLS
An event study is the most general accepted way to calculate the effect of the liberalization of an equity market on the stock market. In contrast to other event studies, the liberalization of an equity market coincides with other political and economic events that influence the stock market. It is very difficult to isolate the effect of the liberalization of an equity market using an event study (Henry, 2000). The liberalization of the capital account coincides with other political and macro-economic events that can bias the results found in the event study when not controlling for these events. That is why the abnormal returns as specified in the paragraph above will be regressed on dummies and extra variables to control for these events. The following regression formula is specified:
(6)
During the announcement of the liberalization other important political events were taking place that are likely to have influenced the Brazilian stock market. The different “Collor plans” were economic reforms designed to fight inflation and generate more economic growth executed by the Brazilian president Fernando Collor de Mello (Collor 3 was named the “Marcilio plan” after the minister of finance at that time). Mercosur (Mercado Comum do Sul) is an economic alliance between Brazil, Argentina, Paraguay, Uruguay and Venezuela to increase free trade and economic connections between its members resulting in great
22 Event Date Announcement of lib. 15-5-1991 Collor 1 16-3-1990 Collor 2 31-1-1991 Marcilio 10-5-1991 Impeachment of Collor 29-9-1991 Mercosur 26-3-1991
Country Fund Opening 5-5-1992
Table 1. Major Political/Economic Events During the Announcement of the Liberalization of the Capital Account of Brazil (15-5-1991)
economic benefits for its members. Impeached is dummy variable for the impeachment of president Fernando Collor de Mello which led to political unrest and economic uncertainty.
"Collor1", "Collor2", "Mercosur", "Marcilio" and "Mercosur" are dummy variables that take on 1 during the event window of these political/economic events. "Announcement" is a dummy variable taking on the value 1 during the event window of the announcement of the liberalization of the capital account and will be the variable that is most important in this paper. If "Announcement" is significant this means that we can say that the announcement of the liberalization of the equity market in Brazil really contributed to the abnormal return obtained in the event study. "Fund" is a dummy variable representing the event window of the first country fund opening. Country funds are investments funds accessible by foreign investors who want to invest in an emerging country who cannot do this directly (Bekaert and Harvey, 2000). " "is the return of the S&P500 index, which is widely used as an index
that represents the American economy. " " is the return rate of the Templeton
Emerging Markets Investment Trust (TEMIT), the largest global emerging investment fund in the UK. Including these returns in the regression controls for co-movements with the world economy and other emerging economies. Including these returns means we are also controlling for macro-economic growth that might be the reason for politicians to consider a liberalization of the equity market. A country is more likely to open his capital account during good economic times because than it gets a higher price. In this way we address the endogeneity issue which was discussed previously.
23
4- Data Analysis
4.1- LIBERALIZATION AND ANNOUNCEMENT DATE
First the date that the Brazilian equity market was liberalized needs to be precisely defined. Resolution 1832 Annex IV stated that on the 31st of May 1991:
“Foreign institutions can own up to 49% of voting stock and 100% of nonvoting stock. Economic ministers approve rules allowing direct foreign investments, 15% tax on distributed earnings and dividends but no tax on capital gains. Foreign investment capital must remain in the country for 6 years (was 12 years).” (Bekaert, Harvey and Lundblad, 2003).
In order to check whether this date really was the beginning of the liberalization of the equity market of Brazil Edison and Warnock (2003) have created a measure that reflects the investibility of the country. This is the amount of domestic firms that are tradable divided by the amount of domestic companies that are tradable for foreign investors and is developed by the IFC ¹. An IFC investible index divided by the IFC Global index that reaches 1 means total foreign investment, a measure of 0 means no foreign investment. This measure shows a big increase and approaches 1 at the liberalization date chosen by Bekaert and Harvey for Brazil.
That's why the 31st of May 1991 will be used as the date that the Brazilian equity market was
liberalized.
Because of spillover effects it is logical to assume that the real effect of the liberalization of the capital account was already incorporated in the stock price before the official liberalization date, namely during the announcement of the official liberalization date. As soon as it is announced that Brazil will liberalize its stock market and this information is credible, investors will start to trade on this information and we will notice a change in the Bovespa index (Stulz, 1999). The announcement date was 15-5-1991 (Wang, 2006) and this date will be used as the date from which on we expect to notice a difference in the main equity index in Brazil.
4.2- DATA
All the companies that were traded on the Bovespa exchange on the 15th of May 1991 will be
divided into two groups: private companies and state-owned companies in order to make a comparison between the two sample groups. State-owned companies are companies that are
24
either fully state-owned or mixed (partly stated-owned) (Boardman and Vining, 1989). The list of companies that were under state-control on the 15th of May 1991 was found on the website of DEST (Department of Co-ordination and Corporate Governance of State Enterprises: http://www.planejamento.gov.br/ministerio.asp?index=4&ler=c638) within the Ministry of Planning, Budget and Management. After cleaning the data we were left with 38 state-owned companies and 441 private companies. Some companies only went public after the announcement date of the liberalization or were not liquid enough to provide representative information, those companies were excluded.
Daily return data will be used in order to capture accurate the effect of the liberalization in the short term (Kothari and Warner, 2006). The Brazilian Bovespa index will be used because this is the main index in Brazil and it a good representation of the Brazilian economy (Leal and Rêgo, 1997). The data of the returns of the Brazilian companies traded on the BM&F
Bovespa exchange can be found on the website of BM&F Bovespa:
www.bmfbovespa.com.br. Data about the S&P500 index can be found at the CRSP database available at the library of the University of Amsterdam and the data about the returns of the Bovespa exchange can be found on the website of the Brazilian National Bank. As a benchmark for the returns of emerging markets the Templeton Emerging Markets Investment Trust (TEMIT) was used, the largest global emerging investment fund in the UK.
26
5- Empirical Results
5.1- ABNORMAL RETURNS AND T-STATISTICS
First the beta's were calculated for the two sample groups using different estimation and event windows. As can be seen in table 3 none of the beta's were significant, although the t-values were close to significance. Changing the estimation or event windows from short to long periods did not have an effect. There is also no big difference noticeable between the period before and after the announcement of the liberalization of the stock market in Brazil on the 15th of May 1991. What can be noticed is that the state-owned firms have bigger beta's than the private firms (although not significantly) and all beta's are very small and negative. In table 4 we can see that the mean of the cumulative abnormal returns for the state-owned firms were positive while those same cumulative abnormal returns were negative for the private firms during the event windows leading up to the event of the liberalization of the Brazilian capital market. This could be a sign that the private firms did not outperform the state-owned firms in the time period before announcement although we cannot prove this because the results are not significant. Also Dewenter and Malatesta 2001 found an increase in the performance of state-owned companies leading up to the liberalization. They claim that it could be a sign that the government is interfering in the accounting results in order to make the company look more attractive for future buyers. In the periods after the 15th of May 1991 all cumulative abnormal returns were negative and significant at the 1 percent level and the mean CAR's were bigger for the state-owned firms than for the private firms, this could be a sign that the private firms performed better in the period following the announcement than state-owned firms. The same is true for the estimation window ranging from 10 trades before the announcement till 10 trades afterwards. This would mean we could confirm the alternative hypothesis stating that private firms outperformed state-owned firms in the period after the announcement. To be able to conclude if indeed state-owned firms were outperformed by private firms following the announcement of the liberalization of the Brazilian equity market we will have to compare the cumulative abnormal returns obtained by the two sample groups as we did in table 5 using the Welch formula specified in section 3.2.
There we can see that state-owned underperformed compared to private companies in almost all cases at the 1 percent level in the period after the announcement. The longer the event window following the announcement the stronger this effect is, resulting in higher t-statistics. Which is a direct conformation of the hypotheses stated at the beginning of this research and
27
confirms the suspicion we had from looking at table 4. Only with the event windows before the announcement of the liberalization the difference is not significant and the private firms even underperformed compared to the state-owned firms, although not significant. The results confirm the research done by Megginson and Netter 2001, Boardman and Vining 1989 and Dewenter and Malatesta 2001 who all found that private firms outperform state-owned firms after the liberalization of a countries equity market (Megginson and Netter 2001 did a literature research containing of 38 studies and Boardman and Vining 1989 investigated 500 companies while looking at profitability measures). The results are a confirmation of the ownership theory stating that private companies provide more incentives for managers to be efficient due to higher monitoring activities. Higher stakeholder pressure and a transferable ownership make sure that there is specialization of ownership (Boardman and Vining, 1989) and high competition incentives make workers more efficient in private firms. Another reason for the outperformance of the private companies could be that state-owned companies focus more on political and social goals which are difficult to measure. It must therefore also be said that by looking at profitability measures only, such as stock market returns, we might miss results generated by state-owned companies that are difficult to measure such as job security and other social externalities (Megginson and Netter, 2001). The results might also be biased due to the fact that a lot of the time companies that were making loses were nationalized. This means that these companies had a bad starting position and that it might not be fair to compare them with profitable private companies (Boardman and Vining, 1989). The opposite is also true: state-owned firms acquire a monopoly position at times due to support from the government, giving them competitive advantages to private firms which could result in higher profits (Megginson and Netter, 2001).
With these results we can reject the null hypothesis stating that the abnormal return is the same for private and state-owned companies and we can accept the alternative hypothesis stating that private companies outperform state-owned companies in the period after the announcement of the liberalization. We can thus confirm the hypothesis stated at the beginning of this research that private firms outperform state-owned firms. It must be said that this effect is only present after the announcement of liberalizing the Brazilian capital account which is in line with the results of Henry 2000 although he looked at the market in general and not at the different ownership structures. It makes sense that the effect is only visible after the announcement because only than investors are able to trade on this information and influence the stock price. That the effect is only noticeable after the announcement can be a
28
sign that after the liberalization of the capital account of Brazil investors were less enthusiastic to buy shares of state-owned firms because they feared that these stocks still would have limitations because these companies would experience interference from the government which could be a disadvantage for the investors (Dewenter and Malatesta 2001). The fact that we found no effect in performance due to the announcement in advance to the announcement, could be a signal that the announcement date of the 15th of May 1991 was indeed the date that the information of liberalizing the Brazilian capital account was known to the public and investors started to trade on it only after the announcement was made which resulted in a revaluation of the equity market afterwards. The information about the liberalization was therefore not credible before the 15th of May 1991.
5.2- REGRESSION ANALYSIS
Table 5 shows the results of regressing the abnormal returns of the state-owned and private firms on a set of dummies and variables. The coefficient on "Announcement" of 2.405 is highly significant when looking at an event window ranging from 10 trades before the event till 10 trades after the event for the state-owned sample group with a 1 percent significance level. An even higher coefficient of 7.196, which is significant, is obtained when looking at the same event window but a shorter estimation window of 90 trades instead of 200 trades. The coefficient of 1.589 is highly significant when looking at the period of 30 trades after the announcement. Also when looking at an event window of 60 or 90 trades after the announcement the effect is noticeable with a significant coefficient of 0.974 and 1.089 respectively, which are significant at the 1 percent level. This is an indication that we indeed can say that the abnormal return of the state-owned firms is explained by the announcement of the liberalization of the equity market of Brazil as we stated in the hypothesis. This is in line with the results of Henry 2000 who did an investigation that is most similar to ours. He found an increase of 3.3% of the equity index per month in an eight month period leading up to the event. In contrast to our research Henry 2000 looked at the liberalization date and not at the announcement date but he found the effect before the liberalization just as we did in our research. Our investigations periods are not the same but they overlap each other. Our results are a confirmation of the result found by Bekaert and Harvey 2000, Stulz 1999, Kim and Singal 2000 and Bekaert, Harvey and Lundblad 2003 who all found a significant effect in the period after the liberalization of the equity market of several emerging markets. Their investigation period was after the liberalization and this is later than our investigation period but has a great overlay with our investigation period. It has to be said that they looked at
29
different performance measures like change in cost of capital, accounting measures and increase in investment rate. Leal and Rêgo did not found a significant effect after the liberalization when looking at returns. The effect we found is noticeable after the announcement, when we look at the period before the event (30, 60 and 90 trades before the announcement) the effect is not significant at the 1 percent level (coefficients of 28.62, 8.183 and 18.23 respectively). We can also see this reflected in the adjusted R squared that reaches 0 in the period leading up to the announcement of the liberalization, and reaching 0.6 when looking at the event windows after the announcement. The change in the abnormal return is less for the private firms than for the state-owned firms. The coefficients of the private firms in the period after the announcement are only significant at the 10 percent level in the 30, 60 and 90 trades after the event with coefficients of 0.089, 0.074 and 0.069 respectively. Just as with the state-owned firms the announcement of the liberalization of the capital account has no effect in the period leading up to the event which is in line with past research.
Also the co-movements with the Templeton Emerging Markets Investment Trust (TEMIT) turns out to be of great importance in the explanation of the abnormal return of the state-owned firms during the announcement in Brazil resulting in significant and very large
negative coefficients on " : 90.13 for 10 trades before till 10 trades afterwards,
-227.6 for a shorter estimation window of 90 trades, -164.2 for an event window of 30 trades after the event and -92.82 for an event window of 60 trades afterwards and finally -97.97 using the event window of 90 trades after the announcement. This explains the negative abnormal return obtained by the two sample groups. We would expect that the announcement of the liberalization causes positive abnormal returns but as we can see the co movement with foreign markets overrides the positive effect caused by the announcement resulting in a total abnormal return that is negative. The significant coefficients are a sign that macroeconomic forces and/or foreign equity markets are also to blame for the abnormal returns obtained by the two sample groups. All these coefficients are significant at the 1 percent level, the
coefficients on are significant at the 5 percent level. If we further look at the coefficients
on "Collor1" and " Collor2" we also see that economic reforms play a great roll in explaining the abnormal return especially in the period after the announcement of Brazil liberalizing its equity market. Just as with the coefficient "Announcement" the effects of the foreign markets and economic reforms seem to play a greater role in explaining the abnormal return obtained by the state-owned firms than the private firms. The coefficients on "Marcilio", "Impeached" and "Mercosur" did not turn out to be of great importance in explaining the abnormal returns
30
obtained by the state-owned and private firms and they were even deleted because of multicollinearity in a lot of the regressions. The answer for their lack in significance could probably also be found in their multicollinearity: those events all followed each other up shortly after one and other and most of the explanatory value is probably in the "Collor" coefficients who turned out to be significant.
31 (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) Le n gth o f Ev e n t W in d o w 10 b ef o re 1 0 a fte r Sh o rt Es ti m a ti o n o f 90 30 b ef o re 30 a fte r 60 b ef o re 60 a fte r 90 b ef o re 90 a fte r Sta te -o w n ed Co e f. -3 3.6 7 -1 02 -3 3.6 7 -3 3.6 8 -3 3.6 6 -3 3.6 8 -3 3.6 8 -3 3.6 8 T-sta ti sti c (-1.3 87 ) (-1.4 07 ) (-1.3 87 ) (-1.3 87 ) (-1.3 86 ) (-1.3 87 ) (-1.3 87 ) (-1.3 87 ) B e ta -0 .0 11 -0 .0 18 4 -0 .0 10 9 -0 .0 11 -0 .0 10 9 -0 .0 11 -0 .0 11 -0 .0 11 N 2,8 64 1,2 77 2,8 64 2,8 64 2,8 64 2,8 64 2,8 64 2,8 64 P ri va te Co e f. -0 .3 12 -0 .2 64 -0 .3 07 -0 .3 11 -0 .3 12 -0 .3 11 -0 .3 07 -0 .3 09 T-sta ti sti c (-0.7 89 ) (-0.3 66 ) (-0.7 77 ) (-0.7 88 ) (-0.7 90 ) (-0.7 86 ) (-0.7 78 ) (-0.7 83 ) B e ta -0 .0 05 -0 .0 03 62 -0 .0 04 9 -0 .0 05 -0 .0 05 -0 .0 05 -0 .0 05 -0 .0 05 N 14 ,8 06 7,3 02 14 ,8 06 14 ,8 05 14 ,8 05 14 ,8 05 14 ,8 05 14 ,8 05 Ta b le 3 . T h e D if fe re n t C o e ff ic ie n ts an d B e ta 's w h e n R e gr e ssi n g th e R e tu rn o f th e D if fe re n t O w n e rsh ip S tr u ct u re s o n t h e M ar ke t R e tu rn f o r B ra zi l f o r th e A n n o u n ce m e n t o f th e L ib e ra li za ti o n o f th e E q u it y M ar ke t D ai ly d ata w as u se d . D if fe re n t e ve n t an d e sti ma ti o n w in d o w s ar e r e p o rte d . A ll e ve n t w in d o w s w e re ar o u n d th e a n n o u n ce me n t d ate o f th e li b e ra li za ti o n . E ve n t w in d o w s ar e in tr ad e s p e rf o rme d . W h e n n o th in g is s ai d a b o u t th e e sti ma ti o n w in d o w , a n e sti ma iton w in d o w o f 20 0 tr ad e s w as u se d . R o b u st t-sta ti sti cs a re r e p o rte d in p ar e n th e se s. Co n sta n ts w e re in cl u d e d in th e r e gr e ss io n s b u t n o t re p o rte d . *, * * an d * ** in d ic ate s ig n if ic an t d if fe re n ce a t th e 1 0, 5 , a n d 1 p e rc e n t le ve ls , r e sp e cti ve ly .
