The Determinants of Stock Market

Development: Implications of a Dynamic

Panel Data Model
MSc in Finance Thesis
Authors:
Daniel Johansson and Andreas Schou Kongstad
Supervisor:
Ola Bengtsson
School of Economics and Management
Lund University
May 27, 2013
Abstract
The motivation for writing this thesis has been to evaluate which pa-
rameters that aect the stock market development in a country by using a
Dynamic Panel Data model with Panel Data from 98 countries between the
years 1992-2011. A scoring system called Stock Market Indicator (SDI) is
also developed, in which the worlds countries is ranked in order to assess
how the 98 selected countries are performing relative to each other. The
SDI-score will be calculated using the coecients of the variables that show
a signicant relationship with the determinant variable as the weighting
parameters.
The econometrical model is constructed based on 16 dierent Macroe-
conomic and Sociologic parameters, and their elasticity in relation to the
growth in market capitalization is then evaluated using the selected econo-
metrical methodology. Using a Dynamic GMM Panel Data Model, 9 of the
selected parameters was found to have a signicant relationship with the de-
pendent variable, and was then selected for the SDI-ranking system. Seven
of the signicant variables in the tests showed the predicted signs, and two
variables did not. Gross Domestic Product, Mergers and Acquisitions, Gross
Fixed Capital Formation, Domestic Credit to Private Sector, Rule of Law,
Political Stability and Mobile Users had a signicant, positive impact on the
change in market capitalization. Ination and the Ocial Exchange Rate
showed a signicant negative relationship with the determinant, opposite to
what we had expected after reviewing previous research.
The SDI-score mainly favored developing countries, as it benets economies
that have seen the largest growth in the selected parameters. It is generally
easier to obtain a large percentage increase in a parameter if the previous
value is on a very low level. On the other hand, those countries might also
be the ones where the most signicant future growth is, as the same logic
applies to the dependent variable. The top 5 countries in our SDI-score
turned out as follows: Mongolia, Papa New Guinea, El Salvador, Kyrgyz
Republic, United Arab Emirates. At the other end, Egypt, Tunisia and
United Kingdom, turned out as the least attractive countries to invest in.
Keywords: Dynamic Panel Data, Stock Market Development, Developed
Countries, Developing Countries, GMM
1
Contents
1 Introduction 4
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Research Contribution . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Theory 7
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Theoretical Literature . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.1 Theory of Asset Pricing . . . . . . . . . . . . . . . . . . . . 7
2.2.2 The Stock Market and Stock Valuation Methods . . . . . . . 9
2.3 Empirical Literature . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.1 Political Stability and Growth . . . . . . . . . . . . . . . . . 10
2.3.2 Technology and Growth . . . . . . . . . . . . . . . . . . . . 11
2.3.3 Financial Intermediation Variables and Growth . . . . . . . 12
2.4 Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . 14
3 Method 15
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Introduction to the Econometrical Methodology . . . . . . . . . . . 15
3.2.1 Static Panel Data Analysis . . . . . . . . . . . . . . . . . . . 16
3.2.2 Static Fixed Eects Model . . . . . . . . . . . . . . . . . . . 17
3.2.3 Static Random Eects Model . . . . . . . . . . . . . . . . . 18
3.2.4 Dynamic Panel Data Analysis . . . . . . . . . . . . . . . . . 20
3.2.5 Generalized Method of Moments . . . . . . . . . . . . . . . 21
3.2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3 Point System Methodology . . . . . . . . . . . . . . . . . . . . . . . 24
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3.2 Building the Point System . . . . . . . . . . . . . . . . . . . 24
2
4 Measuring Financial Development and Growth; Variable and Data
Issues 25
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Measuring the Performance of the Stock Markets . . . . . . . . . . 25
4.2.1 Market Capitalization of Listed Companies . . . . . . . . . . 25
4.2.2 Ination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.2.3 Foreign Direct Investment . . . . . . . . . . . . . . . . . . . 26
4.2.4 Gross Fixed Capital Formation . . . . . . . . . . . . . . . . 27
4.2.5 GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2.6 Exports of Goods and Services . . . . . . . . . . . . . . . . . 27
4.2.7 Ocial Exchange Rate . . . . . . . . . . . . . . . . . . . . . 28
4.2.8 Domestic Credit to Private Sector . . . . . . . . . . . . . . . 28
4.2.9 Control of Corruption . . . . . . . . . . . . . . . . . . . . . 28
4.2.10 Cash Surplus/Decit . . . . . . . . . . . . . . . . . . . . . . 29
4.2.11 Rule of Law . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2.12 Political Stability/Absence of Violence . . . . . . . . . . . . 29
4.2.13 Merger and Acquisition Transactions . . . . . . . . . . . . . 30
4.2.14 Internet Users . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.2.15 Mobile Cellular Subscriptions . . . . . . . . . . . . . . . . . 30
4.2.16 Compensation of Employees . . . . . . . . . . . . . . . . . . 31
4.2.17 Data on the Financial Variables . . . . . . . . . . . . . . . . 31
4.3 Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . 32
5 Evidence From Dynamic Panel Data Models 33
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.2 Model 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.3 Model 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6 Results From the Point-System 38
7 Conclusion and Discussion 40
7.1 Suggested Further Research . . . . . . . . . . . . . . . . . . . . . . 42
8 References 43
3
1 Introduction
1.1 Background
Africa is considered a continent full of hope according to the Economist in their
special report about the emerging countries in their March 2nd 2013 issue. In
their research, they included several variables such as Foreign Direct Investment,
Mortality Rate, Mobile Phone Users and Imports of Commodities. This made us
wonder, what will be the triggering variables on the global scene? Our research
will therefore be taken one step further and include all countries with available
stock market data. We also want to assess which of the variables that have the
highest impact on future growth, and which countries that are currently improving
these variables the most. Inspiration was also taken from the Growth Environ-
ment Score (GES), mainly developed by Jim ONeill during his time as chairman
of Goldman Sachs Asset Management. The GESs main focus is to see how the
growth environment looks like for all countries. In essence, it shows which vari-
ables that should be developed in a country in order to promote growth. This is
where our research diers from earlier research. We want to know which variables
that cause growth and by how much. This will be done by conducting empirical
research using econometric models. The research will therefore include as many
countries as possible, in our case 98 countries, since sucient stock market data
was not possible to nd for the remainder of the 214 countries we had data on.
After determining the impact each explanatory variable has on market capitaliza-
tion (our dependent variable), a ranking system will be created to compare each
country relative to one another. We are well aware of the fact that those variables
will by no means be a safe indicator of the stock market development, but it will
hopefully provide a snapshot of where to start looking for interesting investment
opportunities. The main goal of this thesis is therefore to look at dierent Macro-
and Socioeconomic variables and their elasticity relative to the market capital-
ization, which will be used as an indicator of nancial development. Due to the
diculties of nding consistent index data from many of the developing countrys
stock markets, we have primarily used stock market capitalization as an indicator.
There are several variations of stock market indicators such as Market Capitaliza-
4
tion in relation to GDP, looking at stock market indices and our selected method,
focusing on growth in total market capitalization. We chose the latter as this was
where we had the most data available and since we are looking for pure growth
and not the relationship to GDP.
We are mainly interested in the emerging markets, but we will use data from
all available countries in our analysis. After the analysis, we will create a ranking
system we have named the Stock Development Indicator (SDI) with a weighting
which will be decided upon looking at the elasticity of the explanatory variables
as well as the signicance of the dierent variables.
1.2 Research Contribution
There has been a signicant research made on specic countries and what causes
growth. However, much of this research is focused on smaller samples using be-
tween 1-60 countries. Our research also diers in the way that we measure how
much impact each variable has on growth on a global level and thus the creation
of a comparison system among them when creating our ranking system. Our en-
dogenous as well as exogenous variables are transformed into yearly returns. As
mentioned above our main inspiration source has been the Growth Environment
Score by Jim ONeill. Therefore, some of the variables included in our test have
been taken from ONeills research.
1.3 Purpose
The motivation for writing this thesis is to evaluate which parameters that aect
the stock market development in a country. This is done using a Dynamic Panel
Data Model with Panel Data from 98 countries from 1992-2011. We aim to nd
Socioeconomic and Macroeconomic variables that have a signicant impact on the
growth in the stock market capitalization.
The main questions we want to answer are: (1) which Macroeconomic and So-
cioeconomic variables have a signicant impact on the stock market capitalization?;
(2) which countries show the most signicant increase in those variables?
In order to address those issues, we will start with developing a Panel Data
set with 16 dierent variables. We will then run a Dynamic GMM Panel Data
5
analysis in order to assess which of the variables that prove to have a signicant
impact on the determinant. We will then develop a ranking system, in which the
98 selected countries will be ranked in relation to one another.
1.4 Delimitations
Our Panel Data Model covers 214 countries from 1960 to 2012, and the major part
of the data is found in the World Bank database. There have been diculties to
nd data for 2012 for all variables that we wanted to include in our research. We
have therefore decided to cut our sample at 2011. In addition, a lot of data was
missing before 1980-1990, which made us unable to do a regression for more than
20 years. In addition, many of the less developed countries had no active stock
market, and we were therefore not able to run the analysis on those countries. The
nal regression was made on the 98 countries with an active stock market. We
also excluded variables with little data that might be of interest when more data
is collected in the future.
1.5 Thesis Outline
The thesis is structured as following: Chapter two will go into the underlying the-
ory behind our research. Here we will present a literature review which includes
the theoretical as well as empirical literature. Chapter three will go through the
econometrical methodological considerations taken when conducting the analysis.
The econometrical approach, including the Static Panel Data Analysis, the Static
Fixed and Random Eect Models, the Dynamic Panel Data Analysis and the gen-
eralized method of moments will be covered in this section. The methodology
behind the point-system will also be presented here. In section four, the logic
behind adding the selected variables to the analysis will be explained. In sec-
tion ve the evidence from the Dynamic Panel Data Model will be analyzed and
presented as well as the dierences between our selected models. The following
section will then present the results from the point-system, where a ranking of the
top countries will be displayed. We will then round up the thesis by a conclusion
and discussion of the results and models in section seven. Lastly we will suggest
further research possibilities for the reader.
6
2 Theory
2.1 Introduction
In this chapter we will review the major microeconomic concepts in order to assess
the way Macroeconomic and Socioeconomic variables inuence the stock market.
The rst section (2.2) will review theories about nancial markets such as the e-
cient market hypothesis, the rational expectations theory and the arbitrage pricing
theory. In section 2.3 we will review previous research about how Macroeconomic
and Socioeconomic variables aect growth and stock markets. We review this
research based on their hypotheses, econometric methodologies as well as their
conclusions.
2.2 Theoretical Literature
2.2.1 Theory of Asset Pricing
This thesis will look at how Socioeconomic and Macroeconomic variables aect
the stock markets. In order to understand how those factors inuence the market
value, one rst need to understand the underlying theories and their implications.
Some of the most well-known theories are equilibrium models such as the Capital
Asset Pricing Model (CAPM), which is an extension of the pioneering work of
Markowitz (1959) in which he developed what is today known as modern portfo-
lio theory, and the Arbitrage Pricing Theory (APT), suggested by Ross (1976).
Those models examine the connection between stock market returns and dierent
Macroeconomic variables with the perspective of the Ecient Market Hypothesis
(EMH) suggested by Fama (1965) and the rational expectations hypothesis sug-
gested by Muth (1961). In later work, Fama (1970) have made some signicant
contributions in making the ecient market hypothesis testable. In his work, he
lists three information subsets, indicating how ecient a market is: (1) weak form,
which mean that the investor cannot make a prot from purely relying on historical
prices; (2) semi strong form, where no investor is able to make an abnormal return
by using publicly accessible information; (3) strong form, where no investor is able
to make an abnormal prot even when they have access to inside information.
7
The ecient market hypothesis and rational expectations hypothesis are based
on the idea of a perfect capital market, which is not to be confused with a perfectly
competitive market. According to Grossman and Stiglitz (1980), the stock market
is actually not perfectly competitive. The reasons as to why this is true are,
amongst others, that there are several sources of imperfections such as the inelastic
supply curve, transaction cost, taxes and informational ineciency. Copeland and
Weston (1988) listed the main characteristics of a perfect capital market as follows:
(1) Markets are perfectly competitive, meaning that there are many buyers
and sellers, suppliers have homogenous products, no barriers to entry and exit,
producers supply goods and services priced at their minimum average cost, rms
are price takers and aim to maximize prots;
(2) markets are frictionless as a result of the absence of transaction costs and
taxes, all assets are perfectly separable, all assets are perfectly marketable and
there are no constraining regulations;
(3) markets are informationally ecient, which mean that information is free
and is received by all individuals simultaneously;
(4) all individuals are utility maximizers with rational expectations.
If and when those conditions are met, both the product market and the securi-
ties markets are allocatively, operationally and productively ecient. Fama (1995)
dene an ecient market as a market where there is a large number of rational,
prot maximizing investors. They are actively competing with each other trying to
predict future market values of individual securities, and important current infor-
mation is almost freely available to all participants. In his work, Fama also showed
that security prices quickly adjust to new information. As a result of new technol-
ogy where information is instantaneously available along with a large number of
buyers and sellers, securities markets are more ecient than before. Whether the
stock markets really are ecient is still widely debated. Many from academia still
arguing that the stock markets are indeed absolutely ecient, implying that all
available information is accounted for in stock prices. They argue that investors
might be able to beat the markets over a short period of time due to sheer luck, but
that this should be impossible in the long run. For example, Malkiel (1973) showed
that two-thirds of professional portfolio managers have failed to outperform the
Standard and Poors 500 Index over 30 years. At the other end, value investors ar-
8
gue that an investor who conducts an in-depth analysis of stocks might be able to
nd stocks that have been overlooked by the market. There are several examples
of investors that have been able to achieve superior returns over extended periods.
Price and Kelly (2004) analyzed Warren Buets performance and concluded that
his success could not be solely explained by chance (for a more extensive discus-
sion on this topic, see for example, Nicholson (1968), Basu (1977), Rosenberg et
(1985)). Despite the criticism, the ecient market hypothesis is generally still seen
as a good place to start when it comes to attempting to understand the nancial
markets according to Beechey (2000).
2.2.2 The Stock Market and Stock Valuation Methods
The stock market consists of several individual stocks. Hence, eects that inuence
individual companies will also aect the stock market return as a whole. Dierent
economic variables will aect stock prices directly or indirectly, either through the
nominator or denominator eects, which they called this as they are inuencing the
stock prices by eecting the nominator or denominator in the valuation formula.
Microeconomic variables such as the price of a commodity inuence the stock
price of the individual rm, and are seen as variables aecting the idiosyncratic
risk. This risk can however be eliminated through diversication. Macroeconomic
variables such as interest rates, on the other hand, aect the market as a whole
and are the systematic risk of the market. This risk cannot be eliminated through
diversication.
There are three major methods to value the stock market: (1) Fundamental
Analysis; (2) Technical Analysis and (3) Macroeconomic Analysis. Fundamental
Analysis is related to the semi strong, microeconomic version of EHM. It uses
publicly available information to value a company based on nancial reports and
microeconomic information. Common valuation methods are discounted cash ow
analysis (DCF) and multiple analyses such as P/E, EV/EBITDA. Technical Anal-
ysis, related to the weak eciency of the EHM, focus on historical stock prices
and investor psychology. Technical analysts (also known as a chartists), believe
that past prices can be used to achieve a superior return. Macroeconomic Analy-
sis focuses on the semi strong, Macroeconomic version of the EHM. It focuses on
9
the return of a portfolio instead of analyzing individual stocks, and use publicly
available Macroeconomic information to predict future prices of the stock market.
2.3 Empirical Literature
2.3.1 Political Stability and Growth
Looking at how nancial liberalization correlates with economic growth, numerous
empirical work have found a possible negative correlation between liberalization
and growth. Seck (1993) found little eect of the nancial liberalization in Africa
and its growth rates. Earwell (1996) found that nancial liberalization may actu-
ally hinder growth and development. Bandiera (2000) concluded that there is a
possible relationship between nancial liberalization and falling savings rates.
Contrasting previously mentioned studies, several studies have found the oppo-
site to be true. Pill (1997) conducted a cross-sectional analysis of 87 countries over
the years 1971-95. He found that nancial liberalization might actually stimulate
economic growth. Bittlingmayer (1998) conducted empirical research on Germany
in the interwar period and found that stock market volatility did not have an ex-
plicit negative impact on the stock market return, but that political uncertainty
and events had a large impact on the stock market return as well as industry
output. Henry (2000) showed that the stock market reacted positively to stock
market liberalizations. On average, the market had an abnormal return of 3.3 per-
cent during an eight-month window leading up to the implementation of the initial
stock market liberalization. This was conrmed by Bekaert (2004), who concluded
that equity market liberalizations led to a one percent increase in growth, meaning
that market liberalizations both boost the stock market and the countrys growth
at large. Barro (1991) found that political unrest in the form of the amount of as-
sassinations and occurrence of revolutions and military coups signicantly reduce
a countrys average growth. Also, McGuire (1985) and Barro (1989) were able
to show a connection between the level of political rights and economic growth.
Alesina and Tabellini (1989) examined the eect of political uncertainty on invest-
ment and capital ight. They found that capital ight and excessive government
borrowing is more likely to occur in countries with high political uncertainty.
A possible answer to the somewhat diverged result when it comes to the impact
10
of liberalization and political stability was given by Grossman (1991), who found
that the inhabitants of the country was no better o with high political stability
and high production if the technology of resurrection also induces the government
or ruler to set a lower tax rate. He also found that in countries where the rule
of law is weak, there is a higher risk of revolutions as the population has higher
incentives to participate in such activities.
Related to the issue of political uncertainty is whether dictatorships are harm-
ful or benecial to economic growth. Looking at countries such as China, which
have had an incredible growth over the recent centuries, one might come to the
conclusion that a country might benet from having a non-democratic govern-
ment, especially when tough decisions have to be made. The idea behind this
theory is that dictatorships are less pressured by interest groups, and are therefore
less likely to adhere an opportunistic policy in order to maximize chances of re-
election. This would imply that they are expected to be more short-sighted. The
empirical evidence of this theory is, however, quite varied. Adelman and Morris
(1967), Huntington and Dominguez (1975) and Marsh (1979) have showed a mixed
relationship between economic growth and dictatorship. The general conclusion in
their work was that a technocratic dictatorship resulted in faster growth compared
to democracies, but the opposite was true with a kleptocratic dictatorship.
2.3.2 Technology and Growth
Development in technology is seen as a driving force in the economy. The theory
behind typically stem from the neoclassical models, which treat technology as an
exogenous given factor, which mean that the rate of the long-run technological
growth is given by the exogenous rate of technological change. Important con-
tributors to the neoclassical model are Solow (1956), Cass (1965) and Koopmans
(1965). In the empirical literature, it is suggested that the industrial revolution
(1760 1830) triggered a signicant growth in the average per capita output. Ac-
cording to Maddison (1995), per capita output increased thirteen-fold during the
years 1820-1992.
There has not been a general way to measure technological development in the
literature, but in their article, ONeill (2005) suggest using penetration of mobile
11
phones, penetration of PCs and internet usage as indicators of technological de-
velopment. This is based on the fact that this technology helps transfer broader
technology that aid growth as well as measuring the adaption of new technology
and access to the internet. In Solows model, the only way to grow per capita out-
put is through technological progress. The neoclassical models come in very handy
when attempting to explain the income convergence, but they are limited when
it comes to the ability to explain dierences in per capita output levels between
countries (Ratnasiri, 2009).
Romer (1986) oers an alternative to the neoclassical model. This model takes
a dierent approach by ruling out technology as an exogenous variable. His work
has later been followed by Rebelo (1991) and Grossman and Helpman (1991).
Here, technological growth is instead explained as the accumulation of knowledge
through learning by doing (Romer, 1986) or technological development as a result
of research and development (Rebelo, 1991, Grossman and Helpman, 1991). These
endogenous growth models seem to perform better than the neoclassical models
when it comes to explaining modern growth experiences as they better account for
the non-convergence in incomes across countries according to Ratnasiri (2009).
Another model which is increasingly used after Barro (1990) published his work
is one that attempted to explain divergence of incomes. This kind of models are
called AK models, and are assuming a constant return on capital, instead of the
assumed increased return of scale, which is assumed in Romers (1986) work. The
reason as to why a constant return on capital is more plausible in this model is
because that capital is dened as comprising both human and physical capital.
2.3.3 Financial Intermediation Variables and Growth
A lot of the empirical research that has been done on nancial growth focuses
on nding evidence of the relationship between the nancial intermediation vari-
ables and the countrys growth. For example, Wallich (1969) found that the degree
of intermediation display a positive relationship with growth. In his work, the
rate of growth in money-supply had a non-linear relationship with the growth
in income. King and Levine (1992) found that many nancial indicators had a
signicant relationship with growth. Schwert (1989) conducted research on how
12
Macroeconomic variables aected stock market volatility. In his work, he found
that most of the uctuations in stock market prices cannot be explained by the
volatility in Macroeconomic variables, even though there were indeed some corre-
lations between the two. He found that output volatility, interest rates and bond
volatility appeared to have a slight, positive correlation to each other. He found a
more clear relationship between high volatility on the markets and a high nancial
leverage in the corporate sector. Roubini and Sala-i-Martin (1992) showed that
nancial repressions have a negative impact on ination and growth in reserve
ratios. Looking at how the development of a country aects the nancial markets
using a cross-sectional analysis of 40 countries in the years 1960-85, Atje and Jo-
vanovic (1993) found that the general development of a country had large eects
on the nancial markets. Looking at the opposite, Levine and Zervos (1996) found
that a growing stock market had a signicant, positive, correlation with a coun-
trys economic growth. Looking at nancial indicators and economic eciency,
Odedokun (1996) found that indicators of nancial development have a signicant
relationship with economic eciency. Using Panel Data with a GMM-estimation,
Benhabib and Spiegel (2000) found that indicators of nancial development have a
correlation with total factor productivity growth and investment. Shallu and Su-
parn (2012) mention that their research indicates a possible long-run equilibrium
between nancial progress and economic growth. After running Granger Causality
tests they found that especially banks have a big impact in promoting economic
growth. Another study that indicates that nancial sector development causes
economic growth is Khadraoui (2012). One of his ndings was that credits to the
private sector had a positive relation to GDP. Mehrara (2010) found evidence of
exports and FDI Granger Cause GDP in the long-run. The opposite holds also
for GDP which Granger cause FDI inows. When considering governance indica-
tors in dierent countries, Aisen (2013) found that political instability adversely
aects growth. The stability in a country also aects the willingness of privati-
zation and foreign investments. Zhao et al (2013) found evidence of statistically
signicance to support their hypothesis that privatization and FDI inow pro-
motes economic growth. A further prediction is that if the country experiences
privatization and opening up, then this denationalization causes more economic
growth. Zhao et al also describes the large economic growth of China from 1978
13
until now as a miracle. What caused this miracle seems to be the change from
public ownership and closed-door policy to more private ownership and a policy
that promotes openness. Bittencourt (2012) have conducted a thoroughly research
about what impacts economic growth in Latin America. He highlights the impor-
tance of an open, competitive and active nancial sector. Bittencourt also mention
the importance of not having hyperination at any point, which reduces growth
signicantly. Schumpeter (1912) highlights the importance of nancing when de-
veloping a capitalistic economy. This means that credits given to an entrepreneur
promotes innovations and increases the possibilities of economic growth.
2.4 Summary and Conclusion
The empirical literature showed that improving technology and growth in nan-
cial intermediation variables do indeed have a signicant eect on growth, which
is likely to also aect the market capitalization. Looking at the eect of political
stability and liberalization, the result was somewhat diverged. Some empirical
literature suggested that one should expect little or no eect on growth as a result
of nancial liberalization, with some research suggesting the opposite. Research
that looked at dictatorships as the liberalization variable suggested that a tech-
nocratic dictatorship resulted in faster growth compared to democracies, but that
the opposite was true with a kleptocratic dictatorship. Higher political unrest did
however show a signicant negative relationship with growth, which is why we
chose to focus on indicators looking at political unrest and similar instead of the
level of nancial liberalization.
14
3 Method
3.1 Introduction
In this chapter we will outline the econometrical methodology and theory behind
our tests. The theory is from Baltagi (2005) unless stated otherwise. We will start
with reviewing the basic concepts behind the Panel Data Methodology in section
3.2. The Static Panel Data methodology will then be reviewed in section 3.2.1, and
the two major models (xed and random eects) will be explained in more detail
in section 3.2.2 and 3.2.3. Lastly, the Dynamic Panel Data model will be explained
in section 3.2.4, where we will also review the Dynamic GMM Panel Data Model
in section 3.2.5 which is the model selected in our analysis. We will then explain
how our point system is set up and how the ranking system is calculated in section
3.3.
3.2 Introduction to the Econometrical Methodology
In this thesis, the Cross Sectional Time Series Data Model (Panel Data) is used
as the main econometric method. Using the denition of Panel Data from Hsiao
(2003), Panel Data provides the possibility of learning an individual parameters
behavior by observing others. Panel data is a form of longitudinal data analysis in
which a regression is combined with Time Series Analysis. The Panel is a group of
entities which are observed over time, which makes it a cross section of entities. In
the Panel Data, yearly time increments were used as most of the used parameters
are observed on a yearly basis. By using Panel Data, it is possible to assess what
changes occurred during the selected time period. If there are no missing values in
the data, meaning that all entities have the same amount of observations during the
selected time period, the Panel is considered to be balanced. If there are missing
values in the Panel Data, the Panel is considered to be unbalanced Frees (2004).
Our study mainly entail unbalanced data, which is due to some countries (mainly
less developed countries) do not have the same transparency and data-availability.
In order to correct this, unbalanced mathematical adjustments are made to the
models automatically by the selected software.
The Panel Data have numerous advantages over conventional cross-sectional
15
and time-series data. Some of the advantages listed by Hsiao (2003), Baltagi
(1995) and Klevmarken (1989) are as follows: (1) Panel Data tests control for het-
erogeneity, which is one of its most signicant advantages. The normal time-series
and cross-section tests result in biased results when not controlling for heterogene-
ity; (2) Panel Data provide more informative data as you are able to add large
amounts of data points in the analysis, and also to increase the degrees of freedom
and reduce the collinearity amongst the explanatory variables, thereby increasing
the eciency of the estimates; (3) when using longitudinal data, it is possible to
analyze eects that is not detectable when using a pure cross sectional or time
series data set; (4) Panel Data Models make it possible to conduct more complex
behavioral models than when using cross sectional or time series data; (5) Panel
Data reduce the diculties of nding eects in the presence of omitted variables
that are correlated with the explanatory variables; (6) Panel Data provide a more
accurate prediction of individual outcomes compared only relying on time series
data. There are, however, some limitations when using the Panel Data Model: (1)
the main problem with the Panel Data is the design and collection of data; (2) the
occurrence of cross section dependence, meaning that Macro Panels on countries
or entire regions with long time series that do not account for cross country de-
pendence may lead to ambiguous inference; (3) short time series dimension as the
general Micro Panel involve annual data over a short period of time.
3.2.1 Static Panel Data Analysis
Static Panel Data can be analyzed using xed eects (FE) models and random
eects (RE) models. The general formula for the Panel Linear Regression is as
follows,
Y
it
= a
i
+X

it
+u
it
, i {1, ..., N} and t {1, ..., T} (1)
where i is denoting individuals, countries, regions etc. and t is denoting time. The
interpretation of those two variables is that i represents the cross-section dimen-
sion and t represents the time-series dimension. is K 1, y
it
is the dependent
variable, X
it
is an independent variable representing the itth observation of K and
16
a
i
is the intercept for each individual. a
i
will absorb the impacts of the time-
invariant variables in the equation as well as any heterogeneity in the data. u
it
is the idiosyncratic errors associated with the model. Most Panel Data Models
utilize a one-way error component model for the disturbances,
u
it
=
i
+v
it
(2)
where
i
represents the unobservable, individual-specic, eect. v
it
represents the
disturbance not accounted for in
i
.
3.2.2 Static Fixed Eects Model
The FE model is used when analyzing the impact of variables that vary over time.
It is called a FE regression as it assumes that the factor loads (coecients) are
constant over time, which makes the model somewhat restrictive. In the model,
you explore the relationship between the determinant and the explanatory vari-
ables within an entity (in our case, each country). In the FE model, it is assumed
that something within the country may have a biasing eect on the outcome vari-
ables, which we need to control for. This is the reason to why we assume that
there is a relationship between the countrys error term and the explanatory vari-
ables. Hence, the FE model controls the eects of those time-invariant features
from the explanatory variables in order to better evaluate the explanatory vari-
ables net eect on the determinant. The second key assumption in the FE model
is that the time-invariant eects are unique to each country and thusly should not
be correlated with other individual characteristics. If this assumption should not
be true, the FE model will not be a suitable choice and another model such as
the RE model should be used instead. In order to assess this, a Hausman test is
generally used. In the xed model, the following matrix form equation is used,
y
it
= X

it
+D
N

i
+v
it
, i {1, ..., N} and t {1, ..., T} (3)
where y is the NT 1 vector observation of the dependent variable, X is the
NT K matrix observation of the explanatory variables. D contains the matrix
17
of N individual dummies. Lastly, it is assumed that v
it
is an independent error
term, which is homoscedastic and has zero mean. In addition, the explanatory
variables are assumed to be non-stochastic and independent of the error-terms.
With regards to the assumptions above, the OLS estimators in the xed eects
model are the best linear unbiased estimators BLUE. The OLS estimator can be
written as,

= (X

QX)
1
X

Q
y
(4)
with var(

) =
2
v
(X

QX)
1
and where Q is an idempotent matrix of order NT
and rank NT N.
There are some concerns when using the FE model. One of the issues is that
those models cannot be used to study time-invariant causes of the determinants.
The reason for this is because the time-invariant characteristics of individuals are
perfectly collinear with the country. Stated another way, the FE model is designed
to study causes of changes within countries when the time-invariant individual-
specic characteristics is not relevant to the relationship in question.
3.2.3 Static Random Eects Model
The main dierence between the RE model and the FE model is that the dier-
ences across entities (i.e. countries) are assumed to be random and uncorrelated
with the explanatory variables rather than xed. This allows the time-invariant
variables to be analyzed rather than being absorbed by the intercept as in the FE
model, meaning that if dierences between the entities are believed to inuence
the determinant variable, a RE model should be used. An advantage of the RE
model is that time-invariant variables such as gender can be added to the model.
Those variables are, as mentioned above, absorbed in the FE models. In addi-
tion, the loss of degrees of freedom can be avoided using the random model if
i
is assumed to be random, so that
i
IID(0,
2

), v
it
IID(0,
2
v
) and
i
is in-
dependent of v
it
. The standard RE model with individual eects looks as follows,
y
it
= X

it
+
i
+v
it
, i {1, ..., N} and t {1, ..., T} (5)
18
The equations key dierence is that an error term, v
it
is included in the model. This
error term is associated with the variables within each entity. Another distinction
is that the error term u
it
only represents the between-entity errors, which will
change with time and do not depend on the individuals characteristics.
In the model, we assume the following,
E[
i
] = E[v
it
] = 0 (6)
E[

] =
2

I
N
(7)
E[vv

] =
2
v
I
NT
(8)
E[
j
v
it
] = 0 for all i, t and j (9)
E[v
js
v
it
] = 0 if t = s or i = j (10)
E[
i

j
] = 0 if i = j (11)
When using a Random Eects Model, it is possible to use the OLS as the esti-
mator. However, using OLS is not ecient relative to the GLS estimator. This
ineciency stem from an ineciency weighting of the two LS estimators, as the
OLS place too much weighting on the between-unit variation. Therefore, the GLS
estimator is generally used when using a random eects model.
In order to compute the GLS estimator, several calculations have to be made.
19
First, the variance-covariance matrix for the individual i is to be computed,
= E(uu

) =
2

(I
N
J

) +
2
v
(I
N
I

) (12)
We then need to nd the inverse of the covariance matrix
1
, which is a matrix
of the dimensions NT NT.

GLS
= (X

1
n
X)
1
X

1
n
y (13)
The inverse is computed by,

1
=
1

2
1
W +
1

2
v
+T
2

B (14)
where W = I
N
(I
T

J
T
T
) is the within-units projection and B =
J
T
T
(I
N

J
N
N
)
is the between-units projection.
Inserting the inverse covariance matrix in the GLS estimator will then take
the following form,

GLS
= (X

WX + X

BX)
1
(WX + X

By) (15)
3.2.4 Dynamic Panel Data Analysis
As many economic relationships are Dynamic, the Panel Data Model gives an
advantage relative to other models as the Panel Data allow for a better under-
standing of the Dynamics of adjustment and economic behaviors at the individual
level. The Dynamic relationships are distinguished from the Static Models by the
inclusion of a lagged dependent variable amongst the explanatory variables,
y
it
= y
i,t1
+X

it
+u
it
, i {1, ..., N} and t {1, ..., T} (16)
where is a scalar X

it
is 1 K and is 1 K. It is also assumed that u
it
is
following a one-way error-component model.
20
u
it
=
i
+v
it
(17)
where
i
IID(0,
2

) and v
it
IID(0,
2
V
) independent of one another, and
amongst themselves.
There are two general problems when using a Dynamic Panel Data Model:
(1) the dependent lagged variable y
i,t1
is correlated with the error term u
it
; (2)
the individual estimations may lead to poor nite sample eciency bias and bias
originating from heterogeneity of the cross-section units.
The usual approach to adjust for the lagged dependent variables correlation
with the error term is to remove
i
. This gives the equation,
y
it
y
i,t1
= (y
it
y
i,t1
) +(X

it
X

i,t1
) + (u
it
u
i,t1
) (18)
Another common method to correct for the correlation with the error term is to
use y
i,t2
as the lagged dependent variable. This was suggested by Andersen and
Hsiao (1981) as y
i,t2
is not correlated with the error term u
it
. The use of y
i,t2
have however been criticized by Arellano and Bond (1991) who noted that if T > 3,
there are better instruments available. Also, since y
i,t2
= y
i,t2
y
i,t3
is a linear
combination of Z, the estimator will be inecient. Instead, they suggest using
a Generalized Method of Moments (GMM) version of the Dynamic Panel Data
Estimator.
3.2.5 Generalized Method of Moments
The Dynamic GMM is able to optionally exploit the linear moment restrictions
which is a result of the assumption that there is no serial correlation in the er-
ror terms. The GMM equation is estimated using the level or rst dierences of
the variables. It includes individual eects, lagged dependent variables and do not
have any strictly exogenous variables. The Dynamic GMM estimator is given by,

= (

X

ZA
N
Z

X)
1

X

ZA
N
Z

y (19)
21
where

is the vector of coecient approximations of the endogenous and exoge-
nous regressors,

X

and y are the vectors of the dierences of all the explanatory

variables, Z is the vector of instruments and A
N
is a vector used to weight the
instruments.
The Panel GMM has become very popular recently, for two main reasons. (1)
GMM is less hypothesis consuming than the other Panel Data Methods. Other
methods have for a long time been based on strong assumptions about the un-
derlying data. These assumptions have proven to sometimes be very unrealistic.
In his paper, Hansen (1982) set up a exible tool that is applicable to a large
number of models, which are only relying on mild and quite likely assumptions
dened by a function of moments; (2) GMM also provide a framework for infer-
ence. Some of todays best-known estimation methods such as GLS, 2SLS and
Maximum Likelihood are seen as special cases of the GMM estimator (Ahn and
Schmidt, 1992).
Arellano and Bond (1991) suggested the use of two estimators when conducting
the GMM-analysis, the one-step and two-step estimators. Using this setup, the
two-step estimator oer the best estimates of the models coecients. The two-
step model does however have a tendency to underestimate the standard errors of
the estimations, and therefore oer a false sense of accuracy in certain settings.
The general practice is therefore to base the hypothesis on the one-step estimators
statistics while using the two-step model to estimate the coecients (Yartey, 2008).
This is the method used in this thesis.
The one-step model is set up as following,
A
N
=

i=1
Z

i
HZ
i

1
(20)
where H is a T 2 matrix.
The two-way model swaps the H-term with an estimated variance-covariance
matrix formed on the residuals of a preliminary consistent estimate of . The
optimal A
N
is given by,
22
A
N
=
N

i=1
Z

i
(v
i
)(v
i
)Z
i
(21)
where v
i
are the residuals from the initial approximation of .
When using GMM, it needs to be assumed that no second order serial corre-
lation exists in the rst dierences of the error term. This is because the specic
construction of the instruments, which are lagged, will be rendered invalid should
this assumption not be true. Due to the y
i,t1
term, rst order autocorrelation is
expected and is therefore not a sign of miss-specication. The specication test
generally used in the GMM-estimation is the Sargan test, based on his instru-
mental estimation test developed in Sargan (1958). The Sargan test evaluates
the validity of the set of instruments and is able to determine the validity of the
GMM-assumptions of predeterminacy, endogeneity, and exogeneity. When test-
ing for rst- and second order autocorrelation, the Arellano-Bond test for zero
autocorrelation in rst-dierenced errors is commonly used. In this test, the null-
hypothesis is that there is no autocorrelation. Hence, should the null-hypothesis
be true for the second-order test, there is serial correlation in the model which
will result in the GMM-assumption of no second-order autocorrelation not to be
fullled which will render the GMM-model useless Arellano and Bond (1988).
Another common violation is heteroscedasticity, which occurs when the stan-
dard deviations of a variable vary over time. As the error terms are assumed to
be consistent over time in the Panel Data setting, this need to be accounted for in
the model. The selected method to correct for any eventual heteroscedasticity in
the model is to use Whites heteroscedasticity-consistent estimator.
3.2.6 Conclusion
This section has dealt with the econometric theory used in the thesis, where the
Panel Data Model was selected as the main econometric tool. The models outlined
in this section were the Static Panel Data Models and the Dynamic Panel Data
Models. The main advantage of the Panel Data Models is that they enable you to
model the heterogeneity across countries and individual groups. The Static Panel
Data Models outlined was the Random Eects (RE) and Fixed Eects (FE) mod-
23
els. When discussing the Dynamic Panel Data, the GMM-estimator was outlined
as the preferred method, as it enables the analyst to obtain more ecient and
optimal estimators with fewer assumptions than when using the Static Panel Data
models.
3.3 Point System Methodology
3.3.1 Introduction
In order to get a clear view of the dierent countries prospects we have created a
point system based on the ndings in the econometrical analysis. The aim of this
system is to get a clear view of the countries with the best prospects of growth
with a one year perspective.
3.3.2 Building the Point System
The point system will include the variables that we have found to be signicant
in the analysis. The variables will be weighted based on the elasticity of the ex-
planatory variables in relation to the determinant market size. We will set up a
ranking of all countries and the weightings of the variables will be made using the
following formula:

vc
=
v

(actual observation sample minimum)

(sample maximum sample minimum)

(22)
where
vc
is the point given to the country on a specic variable (i.e. the sub-index),

v
= coefficient significance level is the weighting given to the variable based
on the econometrical analysis and is the multiplier used.
In order to get the total score, the simple formula
c
tot
=

p
v=1

vc
n
have been
used.
24
4 Measuring Financial Development and Growth;
Variable and Data Issues
4.1 Introduction
Measuring nancial development is a key concern for central banks, investment
management rms and other nancial institutions interested in emerging markets.
To get an overview of the nancial development worldwide, we have covered 98
countries, emerging as well as developed countries. Some of the countries are not
that well exploited and sucient relevant data have not been possible to nd. To
measure nancial development in these countries we have specied 16 variables of
which includes macroeconomic, technology development, governance, and human
development indicators as well as inow and outow of capital, as we nd relevant.
4.2 Measuring the Performance of the Stock Markets
To measure the performance of the stock market, we have used a total of 16
variables. All of the variables are listed below and have shown evidence of having
an impact on economic growth from a number of dierent research publications.
In all cases we have used market capitalization as dependent variable. The full
Panel consists of 98 countries and the countries we are interested in are those with
a stock exchange with possibilities for international investments.
4.2.1 Market Capitalization of Listed Companies
Market Capitalization is dened as the share price times the number of outstand-
ing shares. The variable used includes all listed domestic companies at the end
of the year in the countrys stock exchanges. Some companies are however ex-
cluded, such as investment companies, mutual funds and collective investment
vehicles. According to Zafar (2013), there are many researchers suggesting that
the performance of the stock market is an accurate reection of a countrys eco-
nomic performance. He also mentions numerous factors that have an impact on
the stock market performance. Some of the factors are the expansion of the coun-
trys economic activities, strength in the exchange rate, decrease in lending interest
25
rates, improvement in recovery of outstanding loans, rescheduling and payment of
foreign debts, mergers and acquisitions, the relationship with neighboring coun-
tries, investor friendly policies and a strong regulatory framework. In emerging
economies, legal, economic and political factors also play a signicant role. Garcia
(1999) has scrutinized the determinants of stock market performance and he has
found that real income, the savings rate, nancial intermediary development and
stock market liquidity are important determinants. Similar research has been done
by Yartey (2008), who suggests that income level, gross domestic investment, the
banking sector development, private capital ows, stock market liquidity, political
risk, law and order, and the quality of the bureaucracy are important determinants.
4.2.2 Ination
The ination is generally measured by the consumer price index, which reects
the annual percentage change in the cost to the average consumer who acquires
a basket of selected goods and services. According to Mohanasundaram (2012),
high ination erodes the benet of higher return from the equity market. There is
evidence of that ination has a strong positive relationship with the equity market,
which also applies to the market capitalization. In his research, Mohanasundaram
is receiving a correlation between ination and the Indian stock exchange Sensex
of 0.80 and between ination and market capitalization of 0.84.
4.2.3 Foreign Direct Investment
The denition of foreign direct investment is the net inows of investment to
acquire a lasting management interest in an enterprise. These investments need to
be 10 percent or more of voting stock. The investment must also take place outside
the investors own economy. Foreign direct investments are the sum of equity
capital, reinvestment of earnings, other long-term capital and short-term capital
as shown in the balance of payments. For a developing country, a major source of
investment inow is foreign direct investments. According to research from Raza
et al (2012), FDI is showing a mixed behavior in economic development. There
is, however, a positive impact of stock market performance on economic growth.
Those countries with a well-established nancial market can achieve more from FDI
26
than those without. When comparing developed countries with developing, FDI
inow can have a signicant positive impact over economic growth in developing
economies. There are also mixed opinions in the impact of the banking sector to the
stock market performance. For example, Zafar (2013) claim that the development
of the banking sector has no signicant impact on stock market performance.
4.2.4 Gross Fixed Capital Formation
Gross xed capital formation, as dened by the World Bank, includes various
land improvements like fences, ditches and drains, but also plant, machinery and
purchases of equipment. The index was previously called gross domestic xed
investment. It also includes the construction of roads, railways, schools, oces,
hospitals etc. and is an indicator of economic improvements. The contribution
capital formation bring to the overall growth may be more limited than was initially
thought according to Maddison (1991). He is, however, still suggesting that there
is a good reason to estimate it since economies devote a growing proportion of GDP
to investment in capital goods through the process of economic advancement.
4.2.5 GDP
Gross Domestic Product is the sum of all residents production within the countrys
economy, plus taxes and minus subsidies that are not included in the value of the
products. Deduction for depreciation of fabricated assets have not been made nor
depletion and degradation of natural resources. Franses and Mees (2011) have
tested if Chinese and US news of national accounts gures aect the stock market.
The research indicates that four of eight Asian stock indexes react to US news,
while only two of eight Asian stock indexes react on Chinese news. They also
found that US stock market returns do not react signicantly to either US news
or to Chinese news.
4.2.6 Exports of Goods and Services
All goods and market services provided to the rest of the world is included in the
Exports of Goods and Services Data. Excluded are Compensation of Employees,
Investment Income and Transfer Payments. Research made by Usman et al (2012)
27
shows that Export has a signicant positive relation to economic growth. They
conclude that enhanced economic growth is dependent of internationalization of
trade and nancial development in the country.
4.2.7 Ocial Exchange Rate
The national authority determines an exchange rate which is made the ocial ex-
change rate. This rate could also be determined in the legally sanctioned exchange
market. Kogid et al (2011) showed that growth in exchange rate causes economic
growth. They say that stable long-term economic growth requires stable trade and
stable foreign exchange markets. The results show that if the real exchange rate
increased by 1 percent, then the real GDP increased by 7.6 percent. They also
tested how the nominal exchange rate aected the real GDP and got an 8.8 percent
increase. However, only the coecient of the real exchange rate was signicant.
The real exchange rate has therefore a positive signicant impact on the real GDP
in the long run.
4.2.8 Domestic Credit to Private Sector
Financial resources provided to the private sector is included in the domestic credit
to the private sector. These could be loans, purchases of non-equity securities, and
trade credits and other accounts receivables. Some countries could include credit to
public enterprises claims. Quartey (2007) test the long-run Dynamic Relationship
between a group of variables including domestic credit to the private sector, which
shows a positive and signicant eect on the stock market capitalization as a share
of GDP. The short-run eect of domestic credit to the private sector was shown
to be positive, but highly insignicant.
4.2.9 Control of Corruption
There are cultural dierences in how to approach corruption. In for example Asia,
corruption is part of the xed costs while in Africa it is a variable cost. When
the corruption is part of the xed costs, the governance structure is based on
relationships. Corruption and violent criminality are often correlated, but many
involved in a corrupt governance structure are not evil. According to Reja (2012),
28
corruption is an emotional disorder and dicult to cure. Bolgorian (2011) has
tested the Corruption Perception Index and used market capitalization data for
46 countries. A positive correlation has been detected between the Corruption
Perception Index and stock market development.
4.2.10 Cash Surplus/Decit
The cash surplus or decit is measured as revenue and grants minus expense and
the net acquisition of nonnancial assets. The former overall budget balance is
closely correlated with the cash surplus and decit data. According to Roley
(1988), the potential eects of the federal decit that stem from discretionary
scal policy on the stock market depends on several factors. One of the factors
that has been proven to be amongst the most signicant factors is the current
condition of the economy. During a recession, raised output and corporate cash
ows from stimulative scal actions are likely to boost stock prices.
4.2.11 Rule of Law
There is a wide agreement amongst empirical researchers that the Rule of Law
should be considered a key factor of economic development. For example, Curott
(2010) found that countries that put emphasis on securing the Rule of Law are
considerably more productive than countries who did not. Rule of Law and eco-
nomic development goes conjointly, but which one that causes the other is yet
to be determined in the empirical research. When deducting these, Curott mean
that Rule of Law and the security of private property are major causes of economic
growth. Econometric studies also support the economic theory implying that Rule
of Law is an important determinant of economic growth. See, for example, the
study made by Alesina and Tabellini (1989) mentioned in the previous chapter.
4.2.12 Political Stability/Absence of Violence
The political instability is regarded by economists to create a serious drawback
of economic growth. It may also lead to switching policies more frequently and
thus create volatility in Macroeconomic indicators. Aisen (2012) has analyzed
29
the impact of political instability on economic growth. They found that political
instability signicantly reduces economic growth.
4.2.13 Merger and Acquisition Transactions
The act when two organizations bring their businesses together, generally by of-
fering stockholders from one company securities in exchange for the surrender of
their stock, is dened as a merger and the act when an organization purchases a
second organization is dened as an acquisition. The acquirer has thereby acquired
ownership rights over its assets, stocks, operations, business lines and products.
Ly (2011) has looked into various deal and bidder specic characteristics and have
identied a large number of determinants that signicantly create abnormal re-
turns. Research show a positive impact on the nancial performance when consid-
ering companies in Greater China that engage in Merger and Acquisition activities
with a target company from Mainland China.
4.2.14 Internet Users
Internet users is dened by all people with Internet access in a country. When the
usage of Internet increase, Bogan (2008) concludes that the market participation
rate increases. They also nd that the stock transaction costs decreases when
using the Internet and that computer usage increases the probability of owning
stock.
4.2.15 Mobile Cellular Subscriptions
The Mobile Phone variable we are using includes post-paid and pre-paid subscrip-
tions of mobile phones and the subscriptions have access to the public telephone
network. According to Williams et al (2012), the impact of 3G penetration on
GDP growth is 0.15 percentage points when 10 percent of the 2G users switch
from 2G to 3G. They have also researched how much the impact would be if there
would be a doubling of mobile data. Since the penetration of 3G connections,
there has been a massive growth in the use of mobile data, and the impact on
GDP will be 0.5 percentage points if it doubles. In developing markets the impact
30
of mobile phones is expected to be a 4.2 percentage points increase in the long run
when there is an increase of 10 percentage points in the mobile penetration.
4.2.16 Compensation of Employees
An additional variable that we want to test is how the level of employee compen-
sation aect the stock market development. This variable show how much of total
expenses is paid out as salaries to employees, but there have not been any signif-
icant amount of research targeting on this topic. We expect that an increase in
employee compensation would indicate an improved Socioeconomic environment.
Hence, we expect this variable to have a positive correlation with the stock market.
4.2.17 Data on the Financial Variables
The data we have used consists of 249 069 data points, and have been found on the
World Bank, CIA - World Fact Book, Standard and Poors Capital IQ, Thomson
Reuters DataStream and Thomson Reuters Eikon. The base used in the research
is to cover the 98 countries, listed on World Bank, from year 1992 to 2011. The
variables used have been transformed into returns, using the following formula:
r
variable
=
x
t
x
t1
ABS(x
t1
)
(23)
This formula has been applied to be able to study the growth impact on the de-
pendent market capitalization variable. This formula is applied to all data except
Rule of Law, Political Stability and Absence of Violence and Control of Corrup-
tion. These indicators are based on 30 underlying data sources that report the
perception of a large number of surveys. They are estimated in ranges from ap-
proximately -2.5 to +2.5, where +2.5 show the best governance performance. To
apply the returns formula to these datasets we had to make some modications.
Since the range is not xed to -2.5 or +2.5 we have applied
v
new
= v
old
+ABS(x
min
) (24)
31
Where v
new
is the new value, v
old
is the old value and ABS(x
min
) is the absolute
value of the year minimum value. The nal step is the calculation of the yearly
returns.
v
final,t
=

v
new,t1
v
new,t
x
max
+ABS(x
min
)
2
if v
new,t
< v
new,t1
v
new,t
v
new,t1
x
max
+ABS(x
min
)
2
if v
new,t
> v
new,t1
(25)
Where v
final,t
is the nal value which we use in the regression, v
new,t
is the new
value from the preceding step at time t, v
new,t1
is the new value from the preceding
step at time t-1 and x
max
is the year maximum value.
4.3 Summary and Conclusion
The variables used in the research have been carefully selected by running several
dierent regressions and reading up on previous research. The sixteen variables
listed above have been the most signicant variables and therefore used. Since
we have been looking for growth impact to the dependent market capitalization
variable, all of the variables have been transformed into returns. We have used
several dierent sources to nd data but the source that covers the most is the
World Bank and therefore used unless stated otherwise.
32
5 Evidence From Dynamic Panel Data Models
5.1 Introduction
The empirical testing was done using a Dynamic Panel Data analysis using a Dif-
ference GMM model. The empirical analysis has primarily been conducted using
STATA and EViews. This model does not provide us with an intercept, but pro-
vides us with the most statistically signicant and tting estimates when looking
at the autocorrelation- and Sargan-tests. Other models that were considered were
the Fixed or Random Static Panel Data Model, but as we found a very signicant
correlation between the lagged dependent variable (used as an explanatory vari-
able in the Dynamic Model) and the determinant, indicating that the dependent
(market capitalization) is depending on its own past realization, we concluded that
the Dynamic Panel Data model would provide us with the best estimates.
The variables were selected based on research done by the authors, and the
motivation behind the selection is listed in the previous chapter. The testing was
divided into Macroeconomic- and Socioeconomic variables. We chose to divide
them as they signicantly dier from one another. We did however include GDP
in both models as previous research have found a strong relationship between GDP
growth and stock market growth, and it relates both to the Macroeconomic- and
Socioeconomic variables.
5.2 Model 1
The base test was done using all selected variables in the Dynamic GMM Model.
The rst test (Table 5.1) did not perform too well in the Sargan test or the test
for second order autocorrelation, as the model proved to have second order auto-
correlation on the p < 0.10 level. Should this be the case in Model 2, the base
assumption of the GMM-test would not be met; hence the model would not be
useful without rst correcting for this in the estimation. The Sargan test was sig-
nicant on the p < 0.10 level, which indicates that there might be autocorrelation
in the model. Using robust standard errors followed by running the Arellano and
Bonds test for autocorrelation would therefore be a more suitable approach. Our
second test (Table 5.2) performed well both in the Sargan test and the test for
33
second order autocorrelation.
After the initial testing, the least signicant variables which failed to satisfy
the 90 percent condence interval was removed from the tests.
5.3 Model 2
After removing the insignicant variables, model 2 was developed. This is the
model which we will use in our ranking system. In this test, both the Macroeconomic-
and Socioeconomic variables performed well in the Sargan- and second order au-
tocorrelation tests, and the GMM assumptions are therefore true in both Model 2
tests. Seven of the signicant variables in the tests showed the expected signs and
two variables showed unexpected coecients. Gross Domestic Product (GDP),
Mergers and Acquisitions (MNA), Gross Fixed Capital Formation (GFCF), Do-
mestic Credit to Private Sector (DCP), Rule of Law (RULE), Political Stability
(POL), Mobile Users (MOB) having a positive impact on the change in Market
Capitalization and Ination (INFL) and the Ocial Exchange Rate (EXCH) show-
ing a negative correlation. The two negative variables were the ones not showing
the expected signs.
GDP was found to aect the growth in Market Capitalization by 0.55 in Table
5.1 and 0.65 in Table 5.2. As it did not show the exact same coecient, we used
an average of the two in the point system model. MNA proved to have a positive
impact on growth by 0.011 on the 95 percent signicance level, indicating that
an increased MNA activity has a very weak, but signicant, eect on the growth
in Market Capitalization. Gross Fixed Capital Formation was also found to be
signicant with a coecient of 0.34, meaning that this is the second most inuential
of the Macroeconomic variables in our test. The last signicant Macroeconomic
variable was domestic credit to the private sector, with a coecient of 0.014.
Looking at variables with a negative impact, an increase in Ination proved to
have a slight negative eect of -0.0055 on the stock market. The Exchange Rate
had a negative eect of -0.02, indicating that an increasing exchange rate has a
larger impact on the growth of the Market Capitalization than Ination, even
though the eect was quite small.
Looking at the Socioeconomic variables, an increase in Rule of Law had the
34
largest impact on the growth of the stock market with a coecient of 1.26. Po-
litical Stability also had a signicant positive eect of 0.22. Our technological
indicator, Mobile Phone Users per 100 persons, also showed a signicant posi-
tive relationship with the stock market growth with a coecient of 0.43. Internet
Users did not show a signicant relationship with the dependent variable in model
2 and was therefore removed from this model, despite it being signicant on the
90 percent-level in model 1. Other variables not showing a signicant relationship
was Foreign Direct Investments, Exports of Goods and Services, Compensation of
Employees, Life Expectancy and Corruption.
35
36
37
6 Results From the Point-System
When developing the point system, we can see that the variables that have the
largest impact are Rule of Law followed by GDP and Mobile Subscriptions. This
means that the country with the best improvement in these variables will most
likely be located at the top of the list. In Table 6.1 we nd Mongolia, Papua New
Guinea and El Salvador in the top, all of which have a geographical location next
to or close to either a BRIC or Next-Eleven country.
In Table 6.1 below the returns from year 2010 to 2011 is listed for the top 4
countries, which the points are calculated from in Table 6.2.
The points for the top 4 countries are listed in Table 6.2 below and which the nal
scores were calculated from. Some of them have a negative impact on growth and
that is listed in the regression output in Table 5.1 and Table 5.2.
38
39
7 Conclusion and Discussion
The main goal for this thesis has been stated in the rst chapter of this thesis.
The aim was to assess which of the selected Macroeconomic and Socioeconomic
parameters that are likely to aect the stock market in the investigated country.
We also aimed at creating a ranking system (SDI) in order to rank the dierent
countries analyzed in the thesis.
When conducting the econometrical analysis, we started o with 16 variables,
did extensive testing and then used the 9 signicant variables to create our point
system. Our ndings are in line with previous research of how the variables af-
fect the stock market and the general growth in a country on all variables except
for ination and the exchange rate, which have generally shown a positive rela-
tionship with stock market returns in previous research. The variables that was
found to have the most signicant impact on the determinant variable (market
capitalization) were the change in: Gross Domestic Product, Ocial Exchange
Rate, Ination, Gross Fixed Capital Formation, Mergers and Acquisitions, Mobile
Users, Political Stability, Rule of Law and Domestic Credit to Private Sector. The
Dynamic Panel Data Model provide us with seven variables that have a positive
impact on the market capitalization and two variables that have a negative im-
pact on the market capitalization. Change in Ination and Ocial Exchange Rate
are the two variables that have a negative impact. Ination turned out to have
a negative correlation to the stock market development, which somewhat contra-
dicts the ndings by Mohanasundaram (2012). A possible explanation for this is
that we used the change in the variable instead of the actual level of ination. A
large change in the ination-rate might be a result of hyperination, which have
shown to severely reduce the economic growth for a long time. It did however
not prove to have a very strong negative impact in our tests, even though there
was a negative correlation. An increasing Exchange Rate also showed a negative
relationship, contradicting previous ndings by Kogid et al (2011). Our hypothesis
of why this might be true, is that it will be more expensive for other countries to
trade with the country, hence decreasing the exports which might in turn have a
negative impact on the stock market development. On the other hand, exports of
goods and services did not show a signicant relationship with the stock market
40
capitalization. We conclude that more research on this relationship is needed in
order to draw any general conclusions on whether the relationship is positive or
negative. We did however still include this parameter in our point system.
The major part of the variables had a positive impact, with Rule of Law as
the variable with the largest impact on the growth in market capitalization. If an
investor wants to invest in a country, we conclude that a detailed study of how laws
are set up and followed in a country should be of major interest to investors. They
must be transparent, protect fundamental rights and justice should be delivered by
competent representatives. Rule of Law is followed by the Gross Domestic Product
as the variable with the second highest impact on the market capitalization. We
therefore come to the same conclusion as previous research, increasing economic
development is closely correlated to growing stock markets. If the Mobile Users
increases in a country, both our research and previous research indicates that it
will have a positive impact on both market capitalization and the gross domestic
product. One likely explanation for this is that if people have the possibility to
talk to each other and use mobile applications, it might lead to an increase in small
businesses and a wider geographic connection to other companies and people. The
Gross Fixed Capital Formation is a component of the GDP and demonstrates
how much of the new added value that is invested. If investments are made in
xed assets, it will lead to a higher domestic market capitalization and would
be positive for the nancial development. Political Stability also had a positive
relationship with the determinant, which support the previous research by Barro
(1991). At last we have two more variables with a similar level of impact on
the determinant. These two are Domestic Credit to Private Sector and Mergers
and Acquisitions, which showed a slight but signicant impact on the dependent
variable. The general conclusion from our models is that both Macroeconomic- and
Socioeconomic variables play an important role in the uctuations in the market
capitalization. Surprisingly, the tested Socioeconomic variables seem to play a
more important role than the Macroeconomic variables in how the stock market
develops, as their elasticity was signicantly higher than the Macroeconomic ones.
We then calculated the Stock Development Index score for the selected coun-
tries, using the variable coecients as weights and constructed a table over all 98
countries. At the top of the list we nd Mongolia, followed by Papua New Guinea,
41
El Salvador and Kyrgyz Republic right after. These 4 countries were the countries
that had the highest development during 2011 in terms of our selected variables
that aect nancial development.
The ndings of this thesis have some valuable implications, as it show the
major importance of having a stable political environment in order to promote
stock market growth. It also promotes less developed countries as great potential
investment opportunities, even though the ndings is by no mean a safe indication
of future growth.
7.1 Suggested Further Research
This thesis used stock market capitalization as the determinant variable, which is
set up of all listed equities in a country. In order to gain a more in-depth insight
into which sectors that benet the most from the dierent variables, additional
testing using dierent sectors market capitalization as the determinant would be of
interest. Another possibility is to analyze developing markets in dierent regions
by excluding the remainder countries in the Panel Data, gaining a more detailed
insight into how the dierent regions benet from the selected variables. Including
more variables in order to explain more of the stock price movements could also
be a positive extension on the subject. In addition, many variables are newly
developed, and do not have enough data at the moment, and could therefore
not be included in our model. An extension of the analysis with those variables
included should be possible within a few years, and could help provide additional
insight into what creates stock market growth.
As mentioned in the previous section, more research on how the exchange rate
and ination aects the stock market, preferably including exports as another
explanatory variable, could be of interest. Here, a Granger Causality Test could
also be conducted in order to assess whether ination, exchange rates and exports
might Granger Cause one another.
42
8 References
Alberto Alesina, Gerald D. Cohen, and Nouriel Roubini. Macroeconomic policy
and elections in oecd democracies. Economics and Politics, 4:130, 1992.
Daron Acemoglu and Veronica Guerrieri. Capital deepening and non-balanced
economic growth. Journal of Political Economy, 116, 2008.
Raymond Atje and Boyan Jovanovic. Stock markets and development. European
Economic Review, Elsevier, 37:632640, 1993.
Irma Adelman and Cynithis Taft Morris. Society, politics and economic develop-
ment: A quantitative approach. Johns Hopkins Press (Baltimore), 1967.
F. Angeli and H. Maarse. Mergers and acquisitions in western european health
care: Exploring the role of nancial services organizations. Health Policy,
105:265272, 2012.
Seung C. Ahn and Peter Schmidt. Ecient estimation of dynamic panel data
models: Alternative assumptions and simplied estimation. Journal of Econo-
metrics, 76:309321, 1997.
Alberto Alesina and Guido Tabellini. External debt, capital ight and political
risk. Journal of International Economics, 27:199220, 1989.
Ari Aisen and Francisco Jose Veiga. How does political instability aect economic
growth? European Journal of Political Economy, 29:151167, 2013.
Badi H. Baltagi. Econometric analysis of panel data. Wiley, 2, 1995.
Badi H. Baltagi. Econometric analysis of panel data. Wiley, 3, 2005.
World Bank. Cash surplus/decit (percentage of gdp), 2013-05-27. http://
data.worldbank.org/indicator/GC.BAL.CASH.GD.ZS.
World Bank. Compensation of employees (percentage of expense), 2013-05-27.
http://data.worldbank.org/indicator/GC.XPN.COMP.ZS.
World Bank. Control of corruption, 2013-05-27. http://info.worldbank.org/
governance/wgi/mc_countries.asp.
World Bank. Domestic credit to private sector (percentage of gdp), 2013-05-27.
http://data.worldbank.org/indicator/FS.AST.PRVT.GD.ZS.
World Bank. Exports of goods and services (percentage of gdp), 2013-05-27.
http://data.worldbank.org/indicator/NE.EXP.GNFS.ZS.
World Bank. Foreign direct investment, net inows (bop, current usd), 2013-05-
27. http://data.worldbank.org/indicator/BX.KLT.DINV.CD.WD.
World Bank. Gdp (current usd), 2013-05-27. http://data.worldbank.org/
indicator/NY.GDP.MKTP.CD.
World Bank. Gross xed capital formation (current usd), 2013-05-27. http:
//data.worldbank.org/indicator/NE.GDI.FTOT.CD.
World Bank. Ination, consumer prices (annual percentage), 2013-05-27. http:
//data.worldbank.org/indicator/FP.CPI.TOTL.ZG.
World Bank. Internet users (per 100 people), 2013-05-27. http://data.
worldbank.org/indicator/IT.NET.USER.P2.
World Bank. Market capitalization of listed companies (current usd), 2013-05-27.
http://data.worldbank.org/indicator/CM.MKT.LCAP.CD.
World Bank. Mobile cellular subscriptions (per 100 people), 2013-05-27. http:
//data.worldbank.org/indicator/IT.CEL.SETS.P2.
World Bank. Ocial exchange rate (lcu per usd, period average), 2013-05-27.
http://data.worldbank.org/indicator/PA.NUS.FCRF.
World Bank. Political stability/absence of violence, 2013-05-27. http://info.
worldbank.org/governance/wgi/mc_countries.asp.
World Bank. Population ages 15-64 (percentage of total), 2013-05-27. http:
//data.worldbank.org/indicator/SP.POP.1564.TO.ZS.
World Bank. Public spending on education, total (percentage of gdp), 2013-05-27.
http://data.worldbank.org/indicator/SE.XPD.TOTL.GD.ZS.
World Bank. Rule of law, 2013-05-27. http://info.worldbank.org/
governance/wgi/mc_countries.asp.
Robert J. Barro. The ricardian approach to budget decits. The Journal of Eco-
nomic Perspectives, 3:3754, 1989.
Robert J. Barro. Government spending in a simple model of endogenous growth.
The Journal of Political Economy, 98:103125, 1990.
Robert J. Barro. Economic growth in a cross section of countries. The Quarterly
Journal of Economics, 106:407443, 1991.
S. Basu. Investment performance of common stocks in relation to their price-
earnings ratios: A test of the ecient markets hypothesis. Journal of Finance,
32:663682, 1977.
O. Bandiera, G. Caprio, P. Honohan, and F. Schiantarelli. Does nancial reform
raise or reduce saving? Review of Economics and Statistics, 82:239, 2000.
Geert Bekaert, Harvey R. Campbell, and Christian Lundblad. Does nancial
liberalization spur growth? AFA Atlanta Meetings.
Meredith Beechey, David Gruen, and James Vickery. The ecient market hy-
pothesis: A survey. Economic Research Department Reserve Bank of Australia,
2000.
George Bittlingmayer. Output, stock volatility, and political uncertainty in a
natural experiment: Germany, 1880-1940. Journal of Finance, 53:22432257,
1998.
Manoel Bittencourt. Financial development and economic growth in latin amer-
ica: Is schumpeter right? Journal of Policy Modeling, 34:341355, 2012.
Vicki Bogan. Stock market participation and the internet. Journal of Financial
and Quantitative Analysis, 43:191212, 2008.
Meysam Bolgorian. Corruption and stock market development: A quantitative
approach. Physica A: Statistical Mechanics and its Applications, 390:45144521,
2011.
J. Benhabib and M. M. Spiegel. The role of nancial development in growth and
investment. Journal of Economic Growth, 5:341360, 2000.
David Cass. Optimum growth in an aggregative model of capital accumulation.
The Review of Economic Studies, 32:233240, 1965.
N. A. Curott. Foreign aid, the rule of law, and economic development in africa.
University of Botswana Law Journal.
Thomas E. Copeland and J. Fred Weston. Financial theory and corporate policy.
Addison-Wesley, 3, 1988.
John Earwell. International nancial liberalization: The impact on world devel-
opment. Discussion Paper, 12, 1996.
Mehmet Eryigit. The long run relationship between foreign direct investments,
exports, and gross domestic product: Panel data implications. Theoretical and
Applied Economics, pages 7184, 2012.
Eugene F. Fama. The behavior of stock market prices. Journal of Business,
38:34105, 1965.
Eugene F. Fama. Ecient capital markets: A review of theory and empirical
work. The Journal of Finance, 25:383417, 1970.
Eugene F. Fama. Random walks in stock market prices. Financial Analysts
Journal, 51:7580, 1995.
Philip Hans Franses and Heleen Mees. Does news on real chinese gdp growth
impact stock markets? Applied Financial Economics, 21:6166, 2011.
Valeriano F. Garcia. Macroeconomic determinants of stock market development.
Journal of Applied Economics, 2:29, 1999.
Gene M. Grossman and Elhanan Helpman. Endogenous innovation in the theory
of growth. The Journal of Economic Perspectives, 8:2344, 1994.
Herschell I. Grossman. A general equilibrium model of insurrections. The Amer-
ican Economic Review, 81:912921, 1991.
Sanford J. Grossman and Joseph E. Stiglitz. On the impossibility of information-
ally ecient markets. The American Economic Review, 70:393408, 1980.
Samuel P. Huntington and Jorge I. Dominguez. Political development, handbook
of political science. Macropolitical Theory, 3:1114, 1975.
Peter Blair Henry. Stock market liberalization, economic reform, and emerging
market equity prices. The Journal of Finance, 55:529564, 2000.
Cheng Hsiao. Analysis of panel data. University of Southern California, 2, 2003.
Mori Kogid, Rozilee Asid, Jaratin Lily, Dullah Mulok, and Nanthakumar Lo-
ganathan. The eect of exchange rates on economic growth: Empirical testing
on nominal versus real. The IUP Journal of Financial Economics.
Noureddine Khadraoui. Capital movements and economic growth uctuations:
The threshold eect of nancial development. International Journal of Economics
and Finance, 4:193205, 2012.
Anders Klevmarken. Modelling labor supply in a dynamic economy. Working
Paper.
Tjalling C. Koopmans. On the concept of optimal economic growth, in the econo-
metric approach to development planning. North-Holland Publ, 1965.
Noureddine Khadraoui and Mounir Smida. Financial development and economic
growth: Static and dynamic panel data analysis. International Journal of Eco-
nomics and Finance, 4:94104, 2012.
Chi-Vi Ly. The eect of m and a announcements on the stock price behavior in
the greater china area. University of Zurich.
Ross Levine and Sara Zervos. Stock markets, banks, and economic growth. Policy
Research Working Paper Series, 1690, 1996.
Angus Maddison. Dynamic forces in capitalist development: A long-run compar-
ative view. Oxford University Press, 1991.
Burton G. Malkiel. A random walk down wall street. W. W. Norton and Com-
pany, Inc, 1973.
Harry M. Markowitz. Portfolio selection: Ecient diversication of investments.
Wiley, 1959.
R. M. Marsh. Does democracy hinder economic development in the latecomer
developing nations? Comparative Social Research, 2:215248, 1979.
W. J. McGuire. Attitudes and attitude change. Handbook of Social Psychology,
3:233346, 1985.
Mohsen Mehrara, Amin Haghnejad, Jalal Dehnavi, and Fereshteh Jandaghi Mey-
bodi. Foreign direct investment, exports, and economic growth in the developing
countries: A panel data approach. Journal of Academic Research in Economics,
2:259280, 2010.
T. Mohanasundaram and P. Karthikeyan. Impact of institutional investments and
macro-economic variables in the indian equity market. International Research
Journal of Finance and Economics, 94:69, 2012.
John F. Muth. Rational expectations and the theory of price movements. Econo-
metrica, 29:315335, 1961.
F. Nicholson. Price-earnings ratios in relation to investment results. Financial
Analysts Journal, pages 105109, 1968.
Abu Nurudeen and Abdullahi Usman. Government expenditure and economic
growth in nigeria, 1970-2008: A disaggregated analysis. Business and Economics
Journal, 4, 2010.
M. O. Odedokun. Alternative econometric approaches for analyzing the role of
the nancial sector in economic growth: Time-series evidence from ldcs. Journal
of Development Economics, 50:119146, 1996.
Adegbemi Babatunde Onakoya and Russel Olukayode Christopher Somoye. The
impact of public capital expenditure and economic growth in nigeria. Global
Journal of Economics and Finance, 2, 2013.
Jim ONeill, Dominic Wilson, Roopa Purushothaman, and Anna Stupnytska.
How solid are the brics? Global Economics Paper, 134, 2005.
John Price and Edward Kelly. Warren buett: Investment genius or statistical
anomaly? 2004.
P. Quartey and M. Gaddah. Long run determinants of stock market development
in ghana. Journal of African Business, 8:105125, 2007.
Nawaratne Gedara Shyama Chandani Ratnasiri. Economic growth, inequality
and technology adoption in transitional economies. PhD thesis, Queensland Uni-
versity of Technology, 2009.
Sergio Rebelo. Long-run policy analysis and long-run growth. The Journal of
Political Economy, 99:500521, 1991.
Ali Raza, Nasir Iqbal, Zeshan Ahmed, Mohammad Ahmed, and Tanvir Ahmed.
The role of fdi on stock market development: The case of pakistan. Journal of
Economics and Behavioral Studies, 4:2633, 2012.
Nouriel Roubini and Xavier Sala i Martin. A growth model of ination, tax
evasion, and nancial repression. NBER Working Paper, 4062, 1992.
Paul M. Romer. Increasing returns and long-run growth. The Journal of Political
Economy, 94:10021037, 1986.
Stephen A. Ross. The arbitrage theory of capital asset pricing. Journal of Eco-
nomic Theory, 13:341360, 1976.
Barr Rosenberg, Kenneth Reid, and Ronald Lanstein. Persuasive evidence of
market ineciency. Journal of Portfolio Management, 13:917, 1985.
V. Vance Roley and Lawrence D. Schall. Federal decits and the stock market.
Economic Review.
Sanjay Sehgal, Wasim Ahmad, and Florent Deisting. A re-assessment of the role
of the nancial sector in driving economic growth: Recent evidence from cross
country data. International Journal of Economics and Finance, 5:133, 2013.
John D. Sargan. The estimation of economic relationships using instrumental
variables. Econometrica, 26:393415, 1958.
Joseph A. Schumpeter. The theory of economic development: An inquiry into
prots, capital, credit, interest, and the business cycle. Harvard University Press.
G. William Schwert. Why does stock market volatility change over time? The
Journal of Finance, 44:11151153, 1989.
Standard and Poors Capital IQ. Merger and acquisition transactions (closed),
2013-05-27. https://www.capitaliq.com/home.aspx.
Mehdi Safdari, Masoud Abouie Mehrizi, and Marzie Elahi. The eect of popu-
lation age structure on economic growth in iran. International Research Journal
of Finance and Economics, 72, 2011.
D. Seck and Y. H. El Nil. Financial liberalisation in africa. World Development,
21:18671881, 1993.
Robert M. Solow. A contribution to the theory of economic growth. The Quarterly
Journal of Economics, 70:6594, 1956.
Shallu Sehgal and Suparn Sharma. Total factor productivity of manufacturing
sector in india: A regional analysis for the state of haryana. International Journal
of Management and Business Research, 4:241256, 2012.
Suparn Sharma and Shallu Sehgal. Total factor productivity of manufacturing
sector in india: A regional analysis for the state of haryana. International Journal
of Management and Business Research, 4:241256, 2012.
Antti Talvitie and Binyam Reja. Observed dierences in corruption between asia
and africa: The industrial organization of corruption and its cure. International
Public Management Review, 13, 2012.
Muhammad Usman, Muzaar Ali, Haz Waqas Kamran, and Hassan Khalid.
Impact of exports on economic growth: A case of luxemburg. Information Man-
agement and Business Review, 4:17, 2012.
P. van der Eng. Capital formation and capital stock in indonesia, 1950-2008.
Bulletin of Indonesian Economic Studies, 45:345371, 2009.
Henry C. Wallich. The spotty economy. Financial Analysts Journal, 25:2124,
1969.
Chris Williams, Gabriel Solomon, and Robert Pepper. What is the impact of
mobile telephony on economic growth? A Report for the GSM Association.
Charles Amo Yartey. The determinants of stock market development in emerging
economies: Is south africa dierent? IMF working paper, 2008.
Mehwish Zafar. Determinants of stock market performance in pakistan. Inter-
disciplinary Journal of Contemporary Research in Business, 4:10171026, 2013.
Shiyong Zhao. Privatization, fdi inow and economic growth: Evidence from
chinas provinces, 19782008. Applied Economics, 45:21272139, 2013.