Culture as a determinant of the relation between stock markets and economic growth

Culture as a determinant of the relation between stock markets and

economic growth

Evidence for major Asian and developed economies in the period 1987-2008

Master Thesis

University of Groningen

Faculty of Economics and Business

Msc Finance

Author:

Yde Otter

Supervisor:

Prof. dr. R.A.H. van der Meer

Abstract

This paper examines the role of culture in explaining the strength of the relation between lagged stock returns and economic growth. The sample consists of 9 emerging Asian economies and 9 developed economies in the period 1987-2008. Hofstede‟s cultural dimensions are used to proxy for national culture. Evidence is found that countries with a culture of high uncertainty avoidance and femininity have a stronger link between lagged stock returns and economic growth. These two cultural factors can predict the strength of this relation better than the development level of a country can. The estimations have been controlled for the relative size of stock markets and legal origin has been found to be not significant. The results are robust to changes in sampling method, estimation method and fairly robust to changes in sample period.

JEL codes: E44, G15

Table of contents

1. Introduction

...

4

2. Theoretical framework

...

7

2.1 Theoretical connection between growth and return

... 7

2.2 Short term effects of macro economic news ... 8

2.3 Long term relation between the stock market and GDP ... 9

2.4 Emerging economies ...10

2.5 Hofstede‟s cultural dimensions ...11

2.6 Culture in Finance ...13

3. Data and

methodology

...16

3.1 Country selection and data availability

...16

3.2 Data summary ...18

3.3 Model specifications ...22

4. Results

...24

4.1 Regressions across developed and emerging countries ...24

4.2 Panel analysis across cultural dimensions ...26

4.3 Robustness test results ...28

4.4 Interpretation and possible extensions ...31

5. Conclusion

...34

6. References

...36

1. Introduction

The stock market is often seen as an important leading indicator to predict growth of the real economy. Research on this has been extensive in the US and to a lesser extent some other developed countries. It is however not clear if this relation between stock markets and the real economy can be generalized and applies to the same degree to each country. Within developed countries, Continental European countries have compared to Anglo-Saxon countries often a smaller stock market, more financing by banks and a larger public sector. And the fast growing economies in Asia are even more distinct. The relatively new stock markets in emerging countries might represent only a small amount of the total economy, be illiquid or overly speculative. In the last 20 years the economic dominance of Western countries has been decreasing in favor of the upcoming economies in Asia. Some of these countries such as Singapore and Korea have already reached a similar level of welfare as Western countries, while others as Indonesia and China are still on a lower level of development but growing rapidly. At the same time these countries adapted their financial systems to a more Western model. By the end of the 1980‟s almost all countries in South, East and South-East Asia had a stock exchange. Several events impacted both stock markets and the real economies in Asia and across the world in recent times. The Asia crisis in 1997, the economic boom in Asia at the start of the millennium and currently the financial crisis are examples of this. Because of this shift in economic power away from countries with long histories of financing through stock markets, the question whether the forecasting ability of stock markets on economic growth can be generalized becomes more urgent.

Another motive for researching this is that the strength of the relation between the stock market and the real economy is especially interesting for emerging economies. First of all, there are not many leading indicators that can forecast economic growth. In emerging countries these are even scarcer than in developed countries. Frequent changes in financial structure and relatively low liquidity of financial markets make it difficult to find these indicators for a sufficiently large sample period. If a clear link is found, stock markets could be a useful leading indicator in forecasting economic growth (Mauro, 2003). Secondly, it is found that volatility of returns in emerging countries is higher than in mature markets (Richards, 1996). If there is a two way causality in the relation between returns and output, reducing this volatility might be interesting for policy makers of governments in emerging markets to stabilize their economies. Thirdly, it is possible to analyze if countries with certain characteristics have a stronger association between stock returns and growth. This information can help explain cross-country differences in stock market behavior.

across countries. Cultures have grown over many centuries, are deep-rooted and relatively stable over time. Although the rate and pace of change of cultures is open to debate.

The use of culture as an explanatory tool in economic analysis has been pioneered by the famous sociologist Max Weber (1904-1905) in „The Protestant Ethic and the Spirit of Capitalism‟. Weber argues that the protestant focus on hard work, discipline and delayed gratification was instrumental in developing modern capitalistic economies1. It has also been argued that a “Neo-Confucian” culture exists which can be seen as a modern day Asian version of the Protestant ethic (Kahn 1979). Hofstede and Bond (1988) confirm this view in an empirical study, which will be discussed in more detail in section 2.6. Another branch of research is based on the research of McClelland (1961). He proposes that the entrepreneurial ambition of people is instrumental in achieving economic growth. He found a strong positive correlation between measures of „need for achievement imagery‟ in school books and economic growth. Recent empirical analyses of this theory have yield mixed results. Beugelsdijk and Granato, Inglehart and Leblang (1996) find some evidence that an entrepreneurial culture has influence on economic growth, Pryor (2005) and Beugelsdijk and Smeets (2008) find no support for this.

The research in the previous paragraph focused on the role of culture on the actual growth level. This paper analyzes the influence that culture has on the relation between stock markets and the real economy. The strength of this relation shows to what extent stock prices are influenced by the economy at large. A weaker link could signal a few things. For example a higher degree of noise trading or that the listed companies are not representative for a country‟s economy. However it can also mean that there are more possibilities to invest based on company specific information in contrast to macroeconomic information. In recent times countries of tremendous cultural diversity have decided to adapt to an economic system originally developed in the West. Culture can influence finance in two major ways. It can shape the legislative and institutional environment of a country and it can have an influence on the human behavior in this environment. This paper aims to discover if certain cultural characteristics of a country can help to explain the strength of the relation between real stock market returns and real economic growth. The major emerging markets in Asia and major developed countries are analyzed. Most economic activity is concentrated in these area‟s and most countries in other regions have stock markets for a insufficient time period to be able to draw conclusions.

This paper uses the research Hofstede has done between the 1960‟s up till now in trying to quantify culture. He organizes for a large number of countries their culture across 5 dimensions assigning a number to each of these dimensions. Interesting is that between individual countries in both Asia and

1 _{Weber (1921) argued with similar fervor that the spirituality and otherworldliness of Hinduism in India, along}

the Western world large differences can be seen. On a specific cultural dimension a developed country can have more in common with certain emerging countries than it has with other developed countries. Hofstede‟s research has been used extensively in marketing and international management but recently also in finance. However to my knowledge there have not been any previous attempts in trying to explain the relation between economic growth and stock returns using cultural factors. However in a related study, Mauro (2003) tried to explain the strength of the growth-return relation by using several country characteristics. In his study he investigates 8 emerging countries and 17 developed countries in the period 1971 to 1998. Output growth is most strongly correlated with stock returns lagged by one year. He finds a positive and significant correlation between output growth and lagged stock returns for several but not all of the countries in the sample. No large differences are found between emerging and developed economies in the relation between stock markets and economic growth. However the strength of the correlation is related to a number of stock market characteristics, most importantly high market capitalization to GDP and, less robustly, English legal origin and the number of listed companies and initial public offerings. A drawback of Mauro‟s (2003) study is that the sample of emerging countries is quite limited and the countries are often small and atypical (8 countries across the world such as Greece, Zimbabwe and Argentina).

This paper will use a similar method as Mauro (2002) to analyze the relation between stock markets and economic growth. However this paper will look at cultural differences between countries in addition to other country characteristics in order to explain the strength of this relation. Also the sample selection is distinctively different. The countries included in this paper are the major developed countries and major emerging economies in Asia. The main research questions are the following:

1. Can stock markets forecast the real economy in both major developed economies and major emerging economies in Asia?

2. Do cultural factors influence the strength of the relation between stock returns and economic growth?

2. Theoretical framework

In the financing literature the issue of the relation between financial and real variables has been analyzed using some different approaches. This section will start with the theoretical connection between economic growth and stock returns. It then continues with research that focuses on the short term reactions of stock markets to macroeconomic news. Of interest also are two studies that look at the long term relation between economic growth and stock market returns. On emerging economies much research is done on the level of financial development of a country and what the effects are on economic growth. In the last two subsections attention will be given to Hofstede‟s cultural dimensions and research that is done relating these to finance. This section concludes with several hypotheses on the influence of the cultural dimensions on the relation between lagged stock returns and economic growth.

2.1 Theoretical connection between growth and return

The observed correlation between stock returns and output growth has lead to some discussions on the causal link between these two factors. Researchers in the relatively new field of behavioral finance have shown in many papers over the last 20 years that stock markets behave not always in a rational and efficient way. In many cases investor sentiment as well as news about market fundamentals influence stock prices. Important is the extent in which the stock market is merely a reflection of current economic conditions and sentiment or that it actually affects the real economy. If the stock market is merely a sideshow then market inefficiencies would just redistribute wealth between smart investors and noise investors (Morck, Shleifer and Vishny, 1990). However if stock markets affect real activity then also investor sentiment that affects stock markets can affect the real economy.

is more attractive to finance expansion by equity issuance than buying up firms. Under the „stock market pressure on managers‟ hypothesis investment can be affected even if neither stock market information nor financing costs are changed. If investors have negative views on a certain company‟s prospects and the stock price is lowered, managers may cut investment or divest to protect themselves from being fired or taken over.

While these theories differ in the underlining reasons and direction of the causality between the correlation between stock market returns and economic growth, they also have some similarities. All of the theories see the discounted future dividends model as the main connection between stock markets and the real economy. And by implication see the stock market as a leading indictor for predicting economic growth.

A few other considerations have to be made about the connection between stock markets and the real economy. Due to the increasing integration of global markets, many of the companies listed on stock exchanges have extensive operations and sales abroad. Especially in small, open economies the success of certain companies might depend more on the world economy than the local economy. On the other side in many countries smaller, non-listed companies are very important for the local economy. These two cases of disconnections between local stock markets with local economies are difficult to address because they work in opposite directions. One way to take account of these issues is to add market capitalization as a percentage of GDP as a factor in the equation. A higher market capitalization in comparison to the economy can signal a better representation of the country‟s economy in the stock market by having relatively more publicly listed companies. It can however also suggest that many large multinational companies are listed, which would decrease the link between the local stock market and local economy. Mauro (2003) finds that market capitalization as a percentage of GDP is significant and positive in explaining the strength of the relation between stock market return and economic growth. This suggests that the force of more representation of the economy by having relatively more public companies is more important than the effect of listed multinational firms.

2.2 Short term effects of macroeconomic news

macroeconomic variables have explanatory power for explaining variations in stock returns. Kim, McKenzie and Faff (2004) look at the impact of government scheduled announcements on financial markets. They investigate the news effect of six different macroeconomic variables on bonds, stocks and foreign exchange markets. Their results give evidence that the markets do not respond just to the news announcement itself. Rather the news content, if it differs from expectations, causes the markets to react. Fama (1990) seeks to explain the variation in stock returns in the USA. One of the main results is that growth rates of production, used to proxy for shocks to expected cash flows, explain 43% of the return variance. He uses growth rates of production but also mentions that growth rates of real GNP could also have been used. 2

The empirical evidence confirms the notion that changing expectations on macro economic variables directly influence stock markets. This fast response of stock prices to changing economic circumstances also reinforces the belief that stock prices are a leading indicator.

2.3 Long term relation between the stock market and GDP

Ritter (2005) and Siegel (2007) both compare long term averages of stock returns and economic growth. The per capita average real GDP growth is compared with the average stock market return for 16 countries over the period 1900-2005. They find in contrast to the commonly held perceptions a negative correlation. On average countries with higher growth have had a lower stock market return. Ritter (2005) argues that although consumers and workers may benefit from economic growth, the owners of capital do not necessarily benefit. What matters is how much of an economy‟s growth comes from reinvestment into positive NPV investments in existing publicly traded companies, versus how much of it comes from personal savings that are then invested in private companies or in new issues of equity from existing companies. Because historically much economic growth has come from the infusion of capital in new firms, Ritter (2005) argues that future economic growth is largely irrelevant for predicting future equity returns. Siegel (2007) makes a similar analysis for 25 developing countries (for various start dates till 2006). He also finds here a negative correlation. Although these results are quite startling, it should be noted that Ritter (2005) and Siegel (2007) look at a single average figure per country for the total time period. Also the negative cross-country correlation does not reject the possibility the relation between lagged stock returns and economic growth per country is positive. In addition, the negative cross-country correlation might hold for a very long time horizon of more than 100 years, but this does not mean that the same is true for the short and medium term. Ritter (2005) does recognize that cyclical effects should rationally have an effect on equity valuation, but he regards them as largely transitory. These effects on a shorter term are most important for this paper, since economic growth is regressed on stock returns of one year in the

past. In section 3.2 the analysis of Ritter (2005) and Siegel (2007) is repeated for our sample in the period between 1987 and 2008.

2.4 Emerging economies

Studies on links between the stock market and real economy in emerging economies is relatively scarce. One reason for this is the relatively short time span stock markets have existed in many countries. Bilson et al. (2001) find some evidence for a sample of 20 emerging countries in explaining stock returns by both global and local macroeconomic variables. Local macroeconomic variables are significant in their association with equity returns above that explained by the world factor. However most import factor in this is the exchange rate, real activity is only significant for one of the countries in the sample.

Much of the research on emerging economies focuses on a related but different subject as addressed in this paper, namely the link between financial development and economic growth. Institutional development and the sophistication of the financial structure is used to explain growth of an economy. Rousseau and Wachtel (2000) identify four reasons why the stock market is important in promoting economic activity even if equity issuance is a relatively minor source of funds. First of all an equity market provides an exit mechanism for investors and entrepreneurs. Which makes starting businesses more attractive and can boost entrepreneurial activity. Secondly it offers liquidity to international investors which encourages investment in the country. Thirdly it provides companies with permanent capital that can be invested in large, indivisible projects. Finally it can generate information about the quality of potential investments. While these theoretical advantages of a developed financial system seem quite striking, the empirical results are conflictive. Most of the cross-country studies find that there is a positive relation between financial development and economic growth, time-series studies mostly give contradicting results (Demetriades and Hussein, 1996; Bloch, Tang and Hak, 2003; Enisan and Olufisayo 2009). It is also observed that the case of the East Asian countries contradicts the proposition that financial development must precede economic growth. None of these countries had an strong financial infrastructure before their economies started growing in the 1960‟s and 1970‟s. Bloch, Tang and Hak (2003) suggest a bi-directional relation between financial development and economic growth. Rioja and Valev (2004) find that the effect of financial development on economic growth is contingent on the a priori level of financial development. In countries with an intermediate level, financial development has a large, positive effect on growth. In countries with a high level of financial development the effect is smaller and in countries with a low level of financial development the effect on growth is uncertain.3 Market capitalization as a percentage of GDP can indicate the level of financial development somewhat and will be used as a variable in the equations in this paper.

Roll (1992) and Morck (2000) both write about the observation that stock markets across the world behave differently. In developing countries individual stock prices move much more in line with the market index than in developed countries. Morck (2000) finds that in emerging countries like China, Malaysia, and Poland over 80% of stocks often move in the same direction in a given week. In contrast, Denmark, Ireland, and the United States lack any instances of more than 57% of the stocks moving in the same direction during any week. It is found that these results are not due to market size, fundamentals volatility, country size, economy diversification, or the co-movement of firm-level fundamentals. The main reason according to Morck (2000) is differences in property right protection. Strong property rights promote informed arbitrage, which capitalizes detailed firm-specific information. Less emphasis on firm-specific information in emerging countries can suggest a stronger bond between macro economic factors such as economic growth in explaining stock prices.

2.5 Hofstede‟s cultural dimensions

Culture is a concept that is notoriously hard to define, it can be used in a number of ways and can have different meanings to people. Although most people can agree to the large impact culture has on peoples daily lives, generalizations and comparative formalized analyses are by nature controversial.4 This paper will use the research that Hofstede did between 1965 up to the present in quantifying national cultures across cultural dimensions.5 The definition of culture that Hofstede (2001) uses is the following:

“Culture is defined as collective programming of the mind; it manifests itself not only in values, but in more superficial ways: in symbols, heroes and rituals.”

One other useful definition from the Britannica Encyclopedia defines culture as:

“The integrated pattern of human knowledge, belief, and behavior that is both a result of and integral to the human capacity for learning and transmitting knowledge to succeeding generations.”

In Hofstede‟s studies of national cultures, he recognizes that culture is not homogeneous across a national society and that the level of homogeneity varies between societies. The base data for his research was collected through an international employee attitude survey program Hofstede conducted for IBM between 1967 and 1973. In subsequent years additional countries were added and studies were done to validate the findings. At first 4 and later 5 cultural dimensions were identified by

4 _{See McSweeney (2002) for a critical analysis of the methodology of Hofstede and the assumptions he makes.} 5 _{Hofstede (2001) is used as the main source on the research of Hofstede in this paper. As his most recent book,}

Hofstede in comparing differences and similarities between countries. Each country can be positioned on the scale represented by each of the dimensions.

The cultural dimensions are the following:

1. Power distance index (PDI), which is related to the problem of how to deal with human inequality. It is the extent to which less powerful members of organizations and institutions (such as the family) accept and expect an unequal distribution of power. It suggests that the degree of inequality in a society is endorsed by both the leaders as well as the followers. 2. Uncertainty avoidance index (UAI), which is related to the level of stress in societies in facing

an uncertain future. It indicates to what extent a culture programs the people to tolerate uncertainty and ambiguity. Uncertainty avoiding cultures try to minimize unstructured situations by strict laws and rules and more focus on security. Uncertainty accepting cultures are more tolerant of different opinions and are less focused on rules.

3. Individuality (IND), as opposed to collectivism describes the relationship of the individual with collective units in a society. It defines to what degree in a society people are integrated in groups. In other words, the extent in which they are expected to take care of themselves and the immediate family or in which they belong to a larger group of often extended family which protects each other in exchange for loyalty.

4. Masculinity (MAS), versus femininity refers to the division of roles between men and women. Typical masculine values are assertive and competitive, more feminine values are modest and caring. Men‟s values have been found to vary more across countries than women‟s. In feminine countries men‟s and women‟s values are more similar to each other, in masculine countries there is a gap.

5. Long-term orientation (LTO), versus short-term orientation is the fifth cultural dimension that Hofstede added after a study with a questionnaire designed by Chinese scholars conducted in 1985. Both poles of this dimension refer to Confusion values, but also apply to other countries. Values of long-term orientation are thrift and perseverance. Values of short-term orientation are respect for tradition and personal stability.

Section 3.2 describes the scores of the countries in our sample on each of the cultural dimensions. A controversy of Hofstede‟s scores on these 5 dimensions is the level and pace of cultural change. The scores on the cultural dimensions are for the most part from the IBM studies (which are conducted almost 40 years ago). Hofstede sees national culture as remarkably stable over time and he refutes the hypothesis that cultures are more and more converging in the world.6 An explanation for this is the relative nature of the scores on the dimensions. The scores are relative to other countries, therefore

global shifts by technological innovation for example may have a small impact. Hofstede (2002) also believes that the effects of migration are relatively small in most countries due to the assimilation of peoples in the host countries within a few generations.

2.6 Culture in Finance

Hofstede‟s cultural dimensions have been extensively used in marketing and management, but their use in macroeconomics and finance research has been limited up to recent times. In the last several years however this has changed dramatically. Breuer and Quinten (2008)even suggest to proclaim a new, autonomous disciple of „Cultural Finance‟ which would be like behavioral finance at the interface between finance and the social sciences. Breuer and Quinten (2008) summarize in their article the main findings of 30 studies which relate culture to finance, 20 of which use Hofstede‟s cultural dimensions.7 Most of these studies which relate culture to finance have been conducted in the last decade. The topics range from diverse subjects such as the effect of culture on national stock market characteristics, on cash management or on capital structure. In this section the studies relevant for this paper are discussed.

Hofstede (2001) analyzes correlations between cultural dimensions and economic growth. Some evidence for a correlation between a high score on long-term orientation and higher economic growth is found between 1965 and 1995.8 Also for developed countries, some evidence suggests that countries with higher scores on individuality had higher growth rates. It is difficult to draw conclusions based on these correlations, because it does not prove that a causal link exists in which a certain cultural makeup of a country leads to higher growth levels. Ryh-Song and Lawrence (1995) critique the research of Hofstede and Bond (1988) on two main points. Firstly Long-Term Orientation (LTO) and Individuality (IND) seem to be highly interrelated.9 And secondly they claim that culture is not a sufficient condition to explain economic growth but is related to other factors that play a role in determining economic growth. Dieckmann (1996) also uses Hofstede‟s cultural dimensions to explain different growth rates across 38 countries. He suggests that cultural characteristics determine the maximum rate of growth of an economy. The empirical results are mixed, but a significant negative relation of uncertainty avoidance with economic growth is found.

Stulz and Williamson (2003) attempt to explain differences in investor protection between countries. They find that a country‟s principal religion predicts the cross-country variation in creditor rights better than its openness to international trade, its language, its income per capita or the origin of the legal system. They find that countries which have Protestantism as the dominant religion have better investor protection than countries in which Catholicism is dominant. More openness of a country mitigates the influence of religion on creditor rights. De Jong and Semenov (2002) use uncertainty

7 _{A competing model to compare national cultures from Schwartz (1999) is also used in several studies.} 8

Hofstede and Bond (1988) found earlier similar results for a shorter period.

avoidance (UAI) and masculinity (MAS) from Hofstede to explain cross-country differences in stock market capitalization. They find evidence that countries with high UAI and low MAS have weaker legislative protection of minority shareholders, a higher ratio of public provided pensions to the national income and a lower market capitalization relative to GDP. Kwok and Tadesse (2006) also look at the influence of culture on the financial system of a country. The hypothesis is that countries with a high UAI are more likely to be associated with a bank-based financial system. This argument is based on the fact that bank-based systems have higher return predictability than market-based systems. On a cross-section of 41 countries, after controlling for the significant effect of legal origin, the hypothesis is confirmed. Countries with national cultures with a high UAI tend to have bank-based financial systems.

Beckmann, Menkhoff and Suto (2008) use a comparative survey to find evidence for the influence of culture on asset managers views and behavior. Hofstede‟s cultural dimensions are used to proxy for culture. The main results are that a higher individuality (IND) predicts less herding behavior, more power distance (PDI) leads to older and relatively less experienced managers, more masculinity (MAS) leads to more men in high positions and higher assets under their personal responsibility, more uncertainty avoidance (UAI) leads to higher safety margins against the tracking error allowed and relatively more research effort.

To my knowledge this is the first research that uses cultural factors in order to explain the strength of the relation between lagged stock returns and economic growth. The research discussed above shows that culture can shape the legislative and institutional environment in a country. Also Beckmann, Menkhoff and Suto (2008) give some evidence of the influence of culture on human behavior. Using Hofstede‟s cultural dimensions, five hypotheses are formed on the influence of culture on the relation between stock returns and economic growth. The relatively novel nature of this research makes it desirable to test for both positive and negative effects of each cultural dimension on the growth-return link. The alternative hypotheses are that the cultural dimensions have no influence on this relation.

H1: PDI

“The level of power distance in a country influences the strength of the relation between lagged stock returns and economic growth”

H2: UAI

H3: IND

“The level of individuality in a country influences the strength of the relation between lagged stock returns and economic growth”

H4: MAS

“The level of masculinity in a country influences the strength of the relation between lagged stock returns and economic growth”

H5: LTO

3. Data and methodology

3.1 Country Selection and data availability

This study focuses on the emerging economies in Asia and major developed countries. One reason for limiting the study to these countries is that they amount to more than 80 percent of the market capitalization of the stock exchanges in the world.10 The sample is also an interesting mix of the traditional developed countries (Western countries and Japan) and the rapidly developing countries in South, East and South-East Asia. In the media much attention has been given to the economic power shift from West to East. This has been met with some anxiety in Western countries for perceived and real loss of power, stereotypes about Asian cultures and the overwhelming size of the populations. When Western people talk about culture in Asia often no distinctions are made between specific countries. Common stereotypes of Asian cultures are a focus on the community instead of the individual, long term orientation and authoritarianism. While there might be some truth in these stereotypes, it is by no means applicable to the same degree to each of the countries in Asia. In the second part of the investigation using the cultural indicators of Hofstede this paper will analyze the influence of culture. Of interest will be whether certain cultural characteristics will facilitate a stronger bond between stock markets and the real economy. And whether there will be a clear divide between traditional developed countries and the emerging Asian countries.

The sample consists of 18 countries of which 9 are considered developed and 9 emerging. There is some controversy whether to label certain countries as emerging or developed. In this paper the list of emerging countries tracked by The Economist is followed. Included as emerging countries are Singapore and Hong Kong, which have by now similar GDP per capita levels as developed countries. However because this paper examines the relation between stock markets and the economic growth over the last 20 years my belief is that the qualification of The Economist is the most accurate to use in this case. Of all the developed countries Japan is the only non-Western country. Since economically it developed more or less in parallel with the West but retained in many ways its own distinct culture, Japan is an especially interesting case for this study. Taiwan is not included in the sample because due to political reasons, it has no longer country statistics available in the databases of the IMF and Worldbank. With the exception of Taiwan all of the countries with major stock markets and economies in the South, East and South-East Asia region are included.11

Differences in the political development of the countries in Asia also has some influence on the data that is available. Many of the countries have relatively recently opened stock exchanges. For example data on stock markets from China is only available from 1992 while Hong Kong, due to it‟s

10 _{Source: World Federation of Exchanges (WFE)} 11

connection with the UK, has data available from 1964. This investigation studies the period from 1987 to 2008. For China the first years are not available and for India the first year is missing. The developed countries have had stock markets for much longer, however the major indices in some of these countries have existed for a short time. For France and Spain the first year is missing and for Switzerland the first two years are missing. Table 3.1 shows which indices are used to proxy for the stock market of each country.

Table 3.1

Stock Indices used for each country to measure stock returns

Country Stock Index Available from

Canada S&P/TSX Index 1987

China Shanghai SE A Share Index 1992

France CAC 40 1988

Germany DAX 30 1987

Hong Kong Hang Seng Index (HSI) 1987

India BSE 100 1988

Indonesia Jakarta SE Composite (JSX) 1987

Japan Nikkei 225 1987

Malaysia KLCI 1987

Netherlands AEX Index 1987

Philippines PSE Composite Index (PSEi) 1987 Singapore Straits Times Index 1987 South Korea Korea SE Composite (COSPI) 1987

Spain IBEX 35 1988

Switzerland Swiss Market Index (SMI) 1989

Thailand SET Index 1987

United Kingdom FTSE 100 1987

United States S&P 500 1987

To measure stock market performance annual real stock returns are used.12 Real stock returns are used to filter out the effects of inflation, which can be considerable for many emerging economies. Failing to adjust for inflation would overstate stock growth in high inflation economies. The data on stock returns is from the Datastream database and the data on inflation is from the International Financial Statistics of the International Monetary Fund (IMF). Stock returns are a leading indicator. Existing literature on advanced countries find that economic growth is most strongly correlated with stock returns lagged by one year (Mauro 2003). This specification is followed in this research. Real economic growth is represented by GDP growth adjusted for inflation. The GDP growth is measured in local currency to avoid distorting effects of exchange rate movements. This data is from the International Financial Statistics of the IMF as well. To take into account different sizes of stock markets relatively to their economy, market capitalization as a percentage of GDP is added as variable in the regression. This data comes from the World Development Indicators (WDI) by the Worldbank.

The data for 2006 to 2008 were not available yet in the WDI database but have been obtained with the information published on the website of the World Federation of Exchanges (WFE). Data on market capitalization is not available for the first few years for China and the first year for South-Korea. In section 4.1 is also tested whether legal origin is relevant in explaining the strength of the growth-return relation in a country. A dummy for English legal origin is added to the regression. The paper of Djankov, La Porta, Lopez-de-Silanes, Shleifer (2001) is used to identify per country English or non-English legal origin. The data on the cultural dimensions are from Hofstede (2001).13 On the dimension of long-term orientation data on Malaysia and Indonesia is missing.

3.2 Data Summary

Table 3.2 presents the countries included in the sample with some of the main variables used in this study. Large differences exist in the Gross Domestic Products per head of the population of the countries included in the sample. In 1988 the highest GDP per capita of Switzerland is 79 times higher than the lowest of India. In 2008 this statistic decreased somewhat to 63. Also GDP per capita within the group of emerging countries varies widely. This alludes to the problem discussed in the previous sector in classifying countries as emerging or developed. When looking at the average GDP growth in the period 1988-2008, the picture is different. A clear divide can be seen between the group of emerging and developed countries. Even after adjusting for inflation which is often higher in the emerging countries, all of the emerging countries have growth levels that are higher than 3 percent, with many having much higher figures. In the developed group, only Spain has a real GDP growth which is higher than 3 percent.

The average annual stock returns in the period 1987-2007 are somewhat higher in the emerging group than in the developed group. After adjusting for inflation the differences become a bit smaller. Japan is a notable exception, it has a small negative real return on the stock index in the period. This can be attributed to the economic bubble in Japan at the end of the 1980‟s. The stock market hit an all time high in 1989 which up to the present has not been reached again. The real stock returns of the rest of the developing countries hold up remarkably well with the emerging countries. The average real stock returns range from 3 to 8 percent in those developed countries comparing to 4 to 10 percent in the emerging countries. Although the figures are lower for the developed group, the difference is much smaller compared to economic growth. Market Capitalization as a percentage of GDP shows the size of the stock market compared to the economy. Interesting to see is that for almost all countries in the sample this figures has risen in the period 1988-2008. The figure for Japan however shows a large

13

Table 3.2

Indicators on the economy and stock market per country GDP per Capita (US$) Average GDP Growth Average Return on Stock index Average Inflation Average Real GDP Growth

Average Real Return on Stock Index Market Capitalization as a percentage of GDP 1988 2008 1988-2008 1987-2007 1988-2008 1988-2008 1987-2007a 1988 2008 Developed Canada 18,489 45,154 5.02% 7.95% 2.43% 2.54% 5.42% 49% 78% France 17,883 46,047 3.86% 7.08% 1.99% 1.94% 5.13% 24% 54% Germany 15,929 44,303 4.04% 8.23% 2.11% 1.89% 6.24% 19% 33% Japan 24,333 38,563 1.69% 0.58% 0.65% 1.02% -0.01% 133% 60% Netherlands 16,324 52,698 5.01% 6.59% 2.18% 2.77% 4.56% 45% 47% Spain 9,242 36,061 7.54% 9.36% 3.90% 3.51% 5.46% 25% 62% Switzerland 29,254 65,154 3.30% 9.52% 1.80% 1.48% 7.75% 74% 173% United Kingdom 14,973 43,672 5.78% 6.97% 3.55% 2.17% 3.41% 93% 89% United States 20,507 46,336 5.34% 8.58% 3.04% 2.24% 5.55% 55% 80% Emerging China 373 3,237 15.23% 15.07% 6.36% 8.50% 9.88% n.a. 40% Hong Kong 10,477 30,845 6.99% 12.03% 3.83% 3.02% 8.13% 125% 614% India 369 1,042 13.19% 16.66% 7.12% 5.70% 9.58% 8% 114% Indonesia 517 2,263 17.39% 16.06% 10.41% 6.55% 5.68% 0% 22% Malaysia 2,063 8,236 10.53% 7.58% 2.96% 7.37% 4.85% 66% 89% Philippines 637 1,874 11.41% 14.86% 7.23% 3.95% 7.88% 11% 33% Singapore 8,746 39,469 8.52% 8.65% 1.76% 6.64% 7.17% 94% 149% South Korea 4,297 19,667 10.57% 9.24% 2.96% 7.51% 6.65% n.a. 61%

Thailand 1,119 4,060 9.27% 8.03% 3.91% 5.17% 4.25% 14% 39%

a

drop, which can also be attributed to the market bubble described earlier. United Kingdom shows a small drop from an already high figure. Many of the stock markets in Asia were in 1988 still in their start phases as can be seen by their low market capitalizations. In 2008, market capitalization as a percentage of GDP is comparable between emerging countries and developed countries, but large differences between individual countries exist.

As discussed in sector 3.3, Siegel (2007) analyzes the correlation between real stock market returns and real economic growth for a long time period. Surprisingly, it is found that this correlation is negative for both developed and emerging countries. This analysis is replicated in figure 3.1 for the sample of countries in this study in the period 1987-2008. The results from this sample directly conflict with the results from Ritter (2005) and Siegel (2007). On average countries with a higher average stock return also have higher average economic growth. A positive correlation of 0.47 is found which is however only significant on a 10% level. R² is 0.21. It has to be noted however that this result does not necessarily contradict the theory of Ritter and Siegel, because the time period in our study is too short for the effects in their theory to take effect.

China India Hong Kong Philippines Singapore Korea Indonesia Malaysia Thailand Japan United Kingdom Switzerland Germany United States France Canada Netherlands Spain Figure 3.1

Table 3.3 shows the cultural dimensions of Hofstede per country. The score on the Power Distance Index (PDI) is on average higher in the emerging Asian countries, but it is far from homogeneous within the group. The score on the PDI is relatively similar in the group of developed countries, with France on a bit higher end. Scores on the Uncertainty Avoidance Index (UAI) are difficult to categorize. This cultural dimension seems to cut through the division of countries by developed versus emerging or Western versus Asian. Singapore has a remarkably low score, while Japan, France, Spain and South Korea have very high scores. The index of Individuality (IND) is higher for Western developed countries and remarkably low and uniform for the Asian emerging economies. Japan, India and Spain are somewhat in between. Masculinity (MAS) is a dimension as the UAI which is harder to categorize. For MAS the differences between developed countries are especially large. Japan and Switzerland have high rates on masculinity, while Netherlands is a more feminine country. On Long-Term Orientation (LTO) especially the East Asian countries score very high (China, Hong Kong, Japan and South Korea). Also other Asian countries have in general higher scores than the Western countries, with the Philippines being a remarkable exception. Unfortunately the scores for LTO are not available for Indonesia and Malaysia.

Table 3.3

Scores on the cultural dimensions of Hofstede per country Power Distance (PDI) Uncertainty Avoidance (UAI) Individuality (IND) Masculinity (MAS) Long-Term Orientation (LTO) Developed Canada 39 48 80 52 23 France 68 86 71 43 39 Germany 35 65 67 66 31 Japan 54 92 46 95 80 Netherlands 38 53 80 14 44 Spain 57 86 51 42 19a Switzerland 34 58 68 70 40a United Kingdom 35 35 89 66 25 United States 40 46 91 62 29 Emerging China 80 30 20 66 118 Hong Kong 68 29 25 57 96 India 77 40 48 56 61 Indonesia 78 48 14 46 n.a. Malaysia 104 36 26 50 n.a. Philippines 94 44 32 64 19 Singapore 74 8 20 48 48 South Korea 60 85 18 39 75 Thailand 64 64 20 34 56 a

The short descriptions on each of the separated dimensions, reveal the complexity in describing cultures. Although there are similarities between countries on a certain dimension on another dimension the differences can be just as apparent. Interesting to note is that is seems easier to group some Western countries together than Asian countries. For examples the United States, Canada and the United Kingdom are relatively similar, for Germany and Switzerland14 the same can be said. In the Asian group the only two countries which are culturally similar are China and Hong Kong. Hong Kong is since 1999 a Special Administrative Region within China and they have ethnical and historical roots together. Japan is interesting because it is culturally distinct from any other country. The scores on the dimensions will be used in this paper to group the countries into a low, medium and high group for each of the dimensions.

3.3 Model specifications

This section describes the methodology used in this paper to identify what factors can explain the strength of the relation between stock returns and economic growth. In section 4.1 this relation is analyzed across the division of developed and emerging economies. At first an OLS linear regression is used to estimate for each individual country in the sample the correlation between lagged stock returns and economic growth. Following the existing literature this paper focuses on this relationship by regressing real GDP growth in year t on real stock returns in year t-1. In this setup the stock market is assumed to predict the real economy of one year in the future.15 To this „basic regression‟ is subsequently a factor added for GDP growth t-1. This „lagged dependent variable regression‟ may have a reduced autocorrelation with this setup. This is the same methodology used by Mauro (2003) in his analysis on individual country regressions. 16

Basic regression: GDPt = αSRt-1 + c

Lagged dependent variable regression: GDPt = αSRt-1 + βGDPt-1 + c

Where GDPt is real GDP growth for time t, GDPt-1 for time t-1. SRt-1 is real stock return for time t-1

and c is a constant.

14 _{Hofstede (2002) finds that in Switzerland the scores on the cultural dimensions are divided within the country}

by language, with the scores being more similar to Germany, France and Italy in the parts speaking the same language. With German as the largest language group, Switzerland most closely resembles Germany in it‟s total scores.

15 _{Previous studies have shown that output is most strongly correlated with stock returns lagged by one year}

(Mauro, 2003). The additional of a regressor for lagged stock returns of more than one year is tested and leads to virtually no significant results.

16 _{An econometric issue with this setup is that both real economic growth and real stock returns have unit roots.}

From these individual country regressions the average coefficient will be taken for the total sample, the developing countries and the emerging countries.

In the second part of section 4.1 panel regressions are estimated for the developed, emerging and total group of countries. One advantage of a panel regression is that with a relatively small time period for a number of countries the amount of observations can be increased considerably. The total group has 366 observations and the developed and emerging group respectively 185 and 181. The panel analyses allows for individual country intercepts but constrains the slope coefficient to be the same for the group of countries. This makes the assumption that within the group the strength of the relation between lagged stock returns and economic growth is the same. It allows however for different levels of stock returns and economic growth between countries. The variables in the equations are the same as for the individual regressions, but market capitalization as a percentage of GDP is added to take different sizes of stock markets of the individual countries within the group into account. Mauro (2003) finds that English legal origin is significant in explaining the strength of the relation between stock returns and economic growth. At the end of this section a slope dummy is used to test this hypothesis. The dummy is lagged stock returns times English legal origin. This dummy tests directly whether English or non-English legal origin is of influence to the strength of the growth-return relationship.

Section 4.2 will also make use of panel regressions, but to investigate the influence of culture on the growth-returns relation. It is tested whether cultural differences can predict the strength of this relation. Hofstede‟s cultural dimensions as described in the previous sections are used to proxy for culture. For each of the 5 cultural dimensions from Hofstede the countries are divided into 3 groups based on the scores. Subsequently a panel regression is estimated for the high, medium and low group for each of the cultural dimensions. The estimations are done with the same specifications as the panel analyses in the previous section. Since the total sample is split up in 3 groups, high, medium and low per cultural dimension, the amount of observations per regression is lower than in the panel regressions in the previous section (typically around 120 observations).

4. Results

4.1 Regressions across developed and emerging countries

The results of the regressions of lagged stock returns on economic growth for individual countries are presented in Table 4.1. The correlations for the basic regressions are positive in 14 out of 18 countries, but only significantly so in Thailand and the United States. The slope coefficients range between -0.02 and 0.06, with Thailand as an exception with 0.11. The average coefficient for the developed countries is with 0.020 similar to the emerging countries which have 0.024 on average. However the coefficients are much more stable across countries in the developed group. The R² coefficient is low, varying from 0.00 to 0.16, with Thailand again being an exception with 0.55. The correlations are much weaker than the results of Mauro (2003) in his sample between 1971 and 1998. He finds an average slope coefficient of 0.035 and a R² coefficient of 0.17. And his sample 15 out of 25 countries have positive and significant correlations.

The right hand side of Table 4.1 shows the regressions when including a variable which controls for lagged economic growth. This added variable is the real GDP growth of one year in the past. The coefficient of lagged stock return is positive in 13 out of 18 countries. The coefficient is positive and significant in Indonesia and negative and significant in China. The average coefficients with this setup are with 0.009 for developed countries and 0.014 for emerging countries much smaller than with the basic regressions. R² is 0.23 across all countries and much higher than with the basic setup. Mauro (2003) finds that adding lagged economic growth does not change the coefficients for lagged stock returns very much. In the sample used in this study however, lagged economic growth seems to take away some of the explanatory power of lagged stock returns.

Table 4.1

Economic growth and lagged stock returns, individual country regressions, 1988-2008

Basic Regressions Lagged Dependent Variable Regressions

R² DW Constant Lagged Stock Returns R² DW Constant Lagged Growth Lagged Stock Returns Developed Canada 0.01 1.24 0.02 0.012 0.18 1.78 0.01 0.446 -0.015 (3.57) (0.60) (1.60) (2.24) (-0.65) France 0.06 1.78 0.02 0.016 0.11 2.08 0.01 0.231 0.010 (6.30) (1.49) (3.42) (1.39) (0.75) Germany 0.01 1.12 0.02 -0.012 0.18 1.79 0.01 0.414 -0.008 (2.10) (-0.46) (1.74) (1.64) (-0.37) Japan 0.15 0.89 0.01 0.034 0.44 1.56 0.00 0.609 0.014 (1.75) (1.70) (0.60) (4.17) (1.08) Netherlands 0.16 1.27 0.03 0.024 0.30 1.43 0.01 0.458 0.008 (6.94) (1.55) (2.90) (2.38) (0.48) Spain 0.09 1.29 0.03 0.024 0.19 1.82 0.02 0.334 0.023 (6.97) (1.43) (2.16) (1.70) (1.33) Switzerland 0.08 1.43 0.01 0.027 0.21 1.95 0.01 0.347 0.016 (1.75) (1.37) (1.21) (3.54) (0.85) United Kingdom 0.01 0.98 0.02 0.011 0.29 1.60 0.01 0.529 0.006 (4.70) (0.48) (1.48) (3.65) (0.33) United States 0.15 0.94 0.02 0.042 0.28 1.28 0.01 0.430 0.031 (4.61) (2.33)* (1.77) (3.11) (1.87) Average 0.08 0.02 0.020 0.24 0.01 0.422 0.009 Emerging China 0.09 0.42 0.09 -0.022 0.62 1.62 0.02 0.755 -0.045 (7.32) (-1.30) (1.22) (4.18) (-3.05)* Hong Kong 0.02 2.04 0.03 0.017 0.03 2.13 0.03 0.104 0.008 (4.97) (0.75) (3.11) (0.49) (0.23) India 0.06 1.17 0.05 0.021 0.23 1.63 0.03 0.333 0.019 (7.55) (1.54) (4.14) (3.59) (1.41) Indonesia 0.06 2.18 0.06 0.027 0.10 2.02 0.08 -0.248 0.041 (7.72) (1.85) (3.95) (-0.97) (2.28)* Malaysia 0.00 2.17 0.07 0.000 0.01 2.06 0.08 -0.104 0.010 (9.22) (-0.01) (3.51) (-0.32) (0.31) Philippines 0.01 1.66 0.04 0.005 0.06 1.99 0.03 0.271 -0.004 (6.05) (0.45) (1.94) (1.05) (-0.41) Singapore 0.00 1.55 0.07 -0.002 0.02 1.78 0.05 0.191 -0.023 (5.59) (-0.04) (3.54) (1.34) (-0.40) South Korea 0.10 1.14 0.07 0.056 0.25 2.02 0.04 0.400 0.059 (3.75) (1.17) (1.40) (1.49) (1.30) Thailand 0.55 1.59 0.05 0.110 0.64 1.64 0.02 0.430 0.062 (5.15) (3.98)** (1.76) (1.76) (1.54) Average 0.10 0.06 0.024 0.22 0.04 0.237 0.014 All Countries Average 0.09 0.04 0.022 0.23 0.03 0.329 0.012

Newey-West corrected t-statistics in parentheses. DW is the Durbin-Watson statistic. * Indicates that the coefficient is significant at the 5% level.

** Indicates that the coefficient is significant at the 1% level.

The lagged dependent variable regressions in Table 4.2 confirm the previous results. After including lagged economic growth, the coefficients for lagged stock returns are much smaller and not significant. Market capitalization as a percentage of GDP seems mostly relevant for the developed countries, but is also significant for the total sample. The coefficient for lagged growth is large and highly significant for both developed and emerging countries.

Mauro (2002) finds some evidence that English legal origin is a factor that is positive and significant in predicting the strength of the correlation between stock returns and economic growth. To test this hypothesis a slope dummy variable is added in the equation. The dummy used is lagged stock returns times English legal origin. In Table A2 in the Appendix the results are summarized. Although the coefficient is very similar to what Mauro (2002) finds, for the sample used in this study it is not significant in explaining the strength of the growth-return relationship.

4.2 Panel analysis across cultural dimensions

In this section the 18 countries in the sample are divided in groups according to their scores on Hofstede‟s cultural dimensions. The high, medium and low group for each of the 5 cultural dimensions is subsequently analyzed in a panel regression. Table A3 in the Appendix shows the division of countries per cultural dimension. In Table 4.3, the results of the regressions are summarized (Table A4 in the Appendix shows the full equations).

For the Power Distance Index (PDI) only the medium group has a significant and positive coefficient for lagged stock returns. When lagged growth is added in the equation the coefficient is smaller but

Panel estimation of relationship between economic growth and lagged stock returns, 1988-2008 Number of

Observations

Lagged Stock Returns

Basic Regressions Total 366 0.021

(3.42)**

Developed 185 0.012

(1.71)

Emerging 181 0.023

(2.46)* Lagged Dependent Variable

Regressions Total 366 0.010 (1.54) Developed 185 0.005 (0.76) Emerging 181 0.012 (1.17)

T-statistics in parentheses. Estimation with country fixed effects (individual country intercepts). * Indicates that the coefficient is significant at the 5% level.

** Indicates that the coefficient is significant at the 1% level.

still significant on a 1% level. The groups with high and low PDI show small and insignificant coefficients, therefore there is no evidence to suggest that Power Distance matters in explaining the strength of the relation between stock markets and economic growth.

The Uncertainty Avoidance Index (UAI) shows some interesting results. The high group has a highly significant coefficient for lagged stock returns of 0.046. The medium group has a significant

Panel estimation of growth-return relationship for countries grouped by their scores on Hofstede’s cultural dimensions

Basic Regressions

Lagged Dependent Variable Regressions

Number of

Observations Lagged Stock Returns Lagged Stock Returns

PDI High 119 0.008 0.004 (0.78) (0.30) Medium 123 0.049 0.033 (4.13)** (2.79)** Low 124 0.007 0.000 (0.71) (-0.04) UAI High 123 0.046 0.032 (3.81)** (2.88)** Medium 124 0.017 0.015 (2.13)* (1.67) Low 119 0.002 -0.010 (0.14) (-0.76) IND High 123 0.014 0.005 (1.77) (0.66) Medium 124 0.008 0.002 (0.99) (0.21) Low 119 0.032 0.018 (2.45)* (1.27) MAS High 118 0.000 -0.011 (-0.04) (-1.72) Medium 125 0.011 0.001 (0.83) (0.05) Low 123 0.046 0.033 (4.04)** (2.87)** LTO High 97 0.017 0.010 (1.37) (0.78) Medium 102 0.047 0.020 (3.49)** (1.25) Low 125 0.004 -0.003 (0.55) (-0.38)

T-statistics in parentheses. DW is the Durbin-Watson statistic. Estimation with country fixed effects (individual country intercepts). Countries are grouped based on the highest 6, the medium 6 and the lowest 6 scores per cultural indicator. Because 2 countries in the sample have no LTO score, for LTO the division is 5-5-6 from high to low.

coefficient of 0.017 and the low group an insignificant 0.002. When lagged growth is added into the regression the coefficient for the high group drops to 0.033 but it is still highly significant. Both the low and medium group have insignificant coefficients. R² is above 0.3 for all regressions. These results suggest that a culture of high uncertainty avoidance is related to a stronger link between stock market returns and growth of the real economy. This cultural dimension is not easily divided into more traditional groups as developed versus emerging countries, or Western versus Asian countries.

The regressions on the Individuality (IND) dimension do not give strong evidence that it can help to explain the strength of the growth-return relationship. Although the coefficient for lagged stock returns for the low group is 0.032 and significant at the 5% level, when lagged growth is added as a variable the coefficient drops to an insignificant 0.018. The coefficients for the high and medium group are small and insignificant.

The regressions on the Masculinity (MAS) dimension give evidence that more feminine countries have a stronger connection between stock market returns and economic growth. The group with low MAS has a highly significant coefficient of 0.046. It is 0.033 when lagged growth is added into the regression, but it remains highly significant. The medium and high group have coefficients which are small and not significant. R² is above 0.25 for all regressions. Interesting is that MAS, like UAI, is not bound to traditional groupings of countries.

Long-Term Orientation (LTO) has only a significant coefficient for the medium group of 0.047. This is no longer significant when lagged growth is added. The regressions on the high and low group yield no significant results. Therefore I see no evidence to suggest that LTO can help to explain the strength of the growth-return relation.

The panel regressions across the cultural dimensions give evidence that two of the cultural dimensions, namely uncertainty avoidance and masculinity, are important factors in explaining the strength of the relation between stock returns and economic growth. High UAI and low MAS predict a strong relation. It also can be noted that these cultural factors have more power in explaining the strength of this relation than the development level of a country (developed versus emerging).

4.3 Robustness test results

This section will test if the results obtained in the previous section are robust to changes in the sample, sample period and estimation method. The main results from the last section were that according to the panel regressions countries with a high score on UAI and a low score on MAS have a stronger growth-return relation. This section will therefore focus on these two cultural dimensions.

In the sample period there are a few major crises. To test whether the results are robust in the periods without crises in Table A6 in the Appendix the regressions are replicated when a crisis year is taken out of the sample. In the first part is 1998 taken out because of the Asia crisis (1997 for stock returns since these are one year lagged). The coefficients for high UAI and low MAS are with 0.023 and 0.028 respectively much lower than for the whole period. They are however still significant. When adding lagged GDP growth in the equation however the results become insignificant. The Asia crisis does have quite a large impact on the results of the regressions. This is in contrast to what Mauro (2002) finds. The sample in this study has a much larger representation of Asian countries however. The second part of Table A6 in the Appendix shows the results of excluding 2008. This results in slightly higher coefficients for high UAI and low MAS. The effect is much smaller and opposite to excluding 1998. The effects of the credit crisis might be one year later, with lower GDP growth in 2009 instead of 2008.

Market capitalization as a percentage of GDP is representing the size of the stock market of a country. Previous results showed that this factor was positive and significant for the developed group but for the emerging group it was small and insignificant. For high UAI and low MAS it was not significant. However for medium UAI and high MAS it was positive and significant. Table A7 in the Appendix shows the results when market capitalization is excluded from the equation. The coefficients are almost exactly the same for each of the regressions. The results are robust to removing market capitalization from the equations.

One final way to test the robustness of the results obtained in this paper changes the estimation method from a panel analysis with fixed cross-country effects to a system of seemingly unrelated regressions (SUR). Table A8 in the Appendix shows the results. Although the coefficients are a bit smaller, the coefficients remain significant.

Table 4.4 summarizes the results of the robustness analyses conducted in this section. The lagged stock return coefficient for the group with a high Uncertainty Avoidance Index is robust to changing the sample from 3 groups to 2 groups, a change in estimation method to seemingly unrelated regressions (SUR) and removing the proxy for the stock market size relative to the economy. A change in the sample period is more problematic. When 1998, the main year of the Asia crisis, is excluded the coefficient is much smaller and only significant on a 5% level. When lagged growth is added into the equation the coefficient is even smaller and then also insignificant. The coefficient is robust to excluding 2008.

Table 4.4

Lagged Stock Returns for different robustness tests

Regular estimation Sample split in 2 Sample without 1998 Sample without 2008 Estimation without MC Estimation with SUR Basic Regressions UAI High 0.046 0.041 0.023 0.051 0.050 0.029

(3.81)** (4.19)** (2.00)* (4.33)** (4.31)** (3.44)** Medium 0.017 0.016 0.016 0.018 0.010 (2.13)* (2.13)* (1.98) (2.26)* (1.62) Low 0.002 0.004 0.007 0.005 0.002 0.008 (0.14) (0.42) (0.60) (0.40) (0.18) (0.83) MAS High 0.000 0.006 0.000 0.000 0.002 0.007 (-0.04) (0.97) (0.05) (0.02) (0.25) (0.94) Medium 0.011 0.015 0.014 0.012 0.019 (0.83) (1.29) (1.12) (0.89) (2.16)* Low 0.046 0.035 0.028 0.048 0.048 0.029 (4.04)** (3.38)** (2.70)** (4.31)** (4.38)** (3.51)** Lagged Dependent Variable

Regressions UAI High 0.032 0.028 0.008 0.037 0.036 0.022 (2.88)** (3.13)** (0.91) (3.44)** (3.40)** (3.05)** Medium 0.015 0.018 0.015 0.016 0.002 (1.67) (2.22)* (1.64) (1.69) (0.23) Low -0.010 -0.006 0.004 -0.009 -0.009 -0.016 (-0.76) (-0.64) (0.30) (-0.64) (-0.68) (-1.81) MAS High -0.011 -0.004 -0.010 -0.010 -0.010 -0.012 (-1.72) (-0.66) (-1.53) (-1.46) (-1.44) (-1.67) Medium 0.001 0.014 0.002 0.002 0.001 (0.05) (0.01) (0.16) (0.15) (0.09) Low 0.033 0.023 0.018 0.037 0.036 0.017 (2.87)** (2.10)* (1.78) (3.20)** (3.17)** (2.33)* T-statistics in parentheses.