Macroeconomic variables and stock market performance of emerging countries

This article seeks to fill the gap of severe data limitations on the link between macroeconomic variables and stock market performance. A panel data of 41 emerging countries for the period 1996 to 2011 was used to estimate the results. The model used by Sangmi and Mubasher (2013) was adopted and modified to determine the effect of macroeconomic variables on stock market capitalization. The four techniques to investigate the effects were robust ordinary least squares (OLS), FGLS, dynamic ordinary least squares (DOLS) and then Newey-West. It was discovered that depreciation in exchange rate in dollars and reduction in consumer price index affects stock market development negatively, while increase in money supply does influence stock market positively. The findings highlight the significance of macroeconomic factors such as consumer price index, exchange rate, money supply and GDP in explaining the stock market performance in emerging stock economies.


INTRODUCTION
The stock market plays a vital role in the modern economy since it acts as a mediator between lenders and borrowers.Financial markets, especially stock markets, have contributed considerably to the development of emerging economies over the last two decades.This trend is recorded at the same time that these economies are characterized with stable macroeconomic variables.The market capitalization of emerging stock markets rose from $604 billion to $3,074 billion for the period of 1990 to 1999.The trend continued in 2000 with countries like Malaysia, Jordan, Jamaica, Chile, Saudi Arabia, Thailand, and Philippines accounting for the rise in stock market capitalization.This trend is supported by Figure 1.It could be deduced that after 2000 most markets saw an increase in Stock Market Capitalization (SMC) as shown by markets sampled in this article.
Interestingly, countries cited as having high stock market capitalization over the period under study recorded low average GDP.The trend shows an inverse relationship between GDP and stock market capitalization which do not conform to literature reviewed in this article and hence raises questions worth researching.Various macroeconomic variables affect stock market behavior in line with intuitive financial theory (Maysami and Koh, 2000) for which existing literature provides number of theories illustrating the link between stock market behavior and macroeconomic variables.The effect of macroeconomic variables on the stock market characteristics is deep-rooted in literature.However, more studies are focused on the developed countries such as the US, UK and Japan (Fama, 1981;Hamao, 1988;Chen, 1991;Poon and Taylor, 1992) than we have for emerging economies.
The work of Garcia and Liu (1999) established that macroeconomic volatility does not affect stock market performance, while Maku and Atanda (2010) established that stock market performance in Nigeria is mainly affected by macro-economic factors in the long run.Ting et al. (2012) established that Kuala Lumpur Composite Index is consistently influenced by interest rate, money supply and consumer price index in the short run and long-run in Malaysia.Mehwish (2013) recognized a negative relationship between real interest rate and stock market performance in Pakistan.Consumer price index and interest rate have significant impact on the stock market performance in Bangladesh according to the findings of Jahur et al. (2014).
A regression analysis conducted by Aduda et al. (2012) reported that there is no relationship between stock market development and Macro-economic stabilityinflation and private capital flows.Mongeri (2011) established that foreign exchange rates have a negative significant impact on stock market performance.Also, Songole (2012) established that market interest rate, consumer price index and exchange rate have a negative relationship with stock return.Ochieng and Adhiambo (2012) established that 91day T-bill rate has a negative relationship with the NASI while inflation has a weak positive relationship with the NASI.Kimani and Mutuku (2013) showed that there is a negative relationship between inflation and stock market performance.
There has being no research in an attempt to explain the current performance of stock markets in emerging economies in relation to macroeconomic variables that have seen remarkable improvement for emerging economies over sampling period of this article 1996 to 2011.We argue that macroeconomic instability and ceteris paribus negatively impacts stock market development.
In contrast to this study, many researchers such as Black et al. (1997), Hamao and Campbell (1992), Chen et al. (1986), Cochran et al. (1993), Fama and French (1989), Harvey et al. (2002) and Schwert (1990) have based their analysis on business cycle variables or stock market valuation measures such as the term spread or dividend yield.These variables are usually found to be stationary which is the reason why they were not accounted for.
The main objective of this article is to examine the effect of the selected macro-economic (consumer price index, money supply, and exchange rate in dollars) and GDP on stock market performance in emerging economies.

Hypotheses
i. H 0 : There is no significant relationship between the designed macroeconomic variables and stock market performance of emerging countries.This hypothesis tests the relationship between consumer price index, money supply, and exchange rate in US dollars.H 1 : There is significant relationship between the designed macroeconomic variables and stock market performance of emerging countries.This hypothesis tests the relationship between consumer price index (-), money supply (+) and exchange rate in dollars (-).

Descriptive statistics of emerging stock market sampled
To understand the economic importance of the stock market in the sample of 41 countries, the stock market capitalization ratio was examined.The choice of countries and times series data for this article rests on the availability of data.Data for this article are from Worldwide Governance indicators, World Development Indicator (WDI) and Global Finance and Development (GFD).The stock market capitalization ratio is defined as the value of domestic equities traded on the stock market relative to GDP.As can be observed from Appendix 1, stock market development indicators exhibit a considerable variability across countries, according to the stock market capitalization ratio.The top ten countries in terms of mean stock market capitalization for the period under review are South Africa, Malaysia, Jamaica, Jordan, Chile, Zimbabwe, Saudi Arabia, Thailand, Philippines and India in that order.The countries with lowest stock market capitalization are Ecuador, Slovak Republic, Bangladesh, Paraguay and least Uruguay.As can be seen in stock market development in terms of total value trade as percentage of GDP, South Africa moved from the first to third position with Saudi Arabia occupying the first position from our sample.Stock market capitalization has very little to do with the size of a country.China, which has the largest economy by far among these countries, has a smaller average market capitalization than Hong Kong over the period.South Africa and Taiwan approached China in terms of stock market capitalization despite vastly smaller population and GDP.Again even though Nigeria has a larger economy than Ghana, Ghana is ahead of Nigeria in terms of stock market capitalization as a measure of development of the capital market.
A National Bureau of Economic Research (NBER) Working Paper in April, 2013on Financial Development in 205 Economies, 1960 to 2010, has gathered substantial evidence that financial institutions (such as banks and insurance companies) and financial markets (including stock markets, bond markets, and derivative markets) exert a powerful influence on stock market development, poverty alleviation, and economic stability.Stock market development has been central to the domestic financial liberalization programs of most emerging markets.Apart from their role in domestic financial liberalization, the stock markets have also been very important in recent years as a major channel for foreign capital flows to emerging economies.Net equity flows to the emerging markets have grown over the years, providing an important source of capital for development.The share of foreign direct investment and portfolio equity in the finance mix of many developing countries has grown in recent years.Equity flows accounted for 80% of total external financing to developing nations during 1999 to 2003, compared with just 60% during 1993to 98 (Global Development Finance, 2005).Cross-border capital flows, which include lending, foreign direct investment and purchases of equity and bonds, rose to a peak of $11.8 trillion in 2007, primarily due to the acceleration in interbank lending with a smaller share being the flow of funds to real economy borrowers.According to a McKinsey Global Institute (MGI) study, as of 2012, cross-border capital flows had declined by 61% from the 2007 peak to $4.6 trillion.Most of this reduction was in intra-European flows, thus raising the share of global capital flows to emerging economies to 32% in 2012 ($1.5 trillion) from 5% in 2000.Capital flows out of developing countries rose to $1.8 trillion in 2012.
Development of stock markets in emerging market does not imply that even the most advanced emerging stock markets are mature.Trading occurs in only a few stocks which account for a considerable part of the total market capitalization.Beyond these actively traded shares, there are serious informational and disclosure deficiencies for other stocks.There are serious weaknesses in the transparency of transactions on these markets.The less developed of the stock markets suffer from a far wider range of such deficits.Compared with the highly organized and properly regulated stock market activity in the US and the UK, most emerging markets do not have such a well-functioning market.Not only are there inadequate government regulation, private information gathering and dissemination firms as found in more developed stock markets are inadequate.Moreover, young firms in emerging stock markets do not have a long enough track record to form a reputation.As a result, one expects share prices in emerging markets to be arbitrary and volatile (Tirole, 1991).Empirical evidence indicates that share prices in emerging markets are considerably more volatile than in advanced markets.
Despite this volatility, large corporations have made considerable use of the stock market.For example, the Indian stock market has more than 8,000 listed firms, one of the highest in the World.Looking at the corporate financing pattern in emerging markets it was found that contrary to expectation, emerging market corporations rely heavily on external finance and new equity issues to finance long term investment and the stock markets have been successful in providing considerable funds.
Market liquidity is one the measures of stock market development.Market Liquidity is ability for investors to buy and sell shares.Stock market performance was measured using total value traded as a share of GDP, which gives the value of stock transactions relative to the size of the economy.According to the work of Levine and Zervos (1998) this measure is used to gauge market liquidity.This is because it measures trading relative to   Of the economies sampled nine of them are from Africa   and thirty-two from other continents.Stock market capitalization which is a measure of stock market development had being relatively stable for emerging African economies sampled for this article.The proxy for this measure is stock market capitalization relative to GDP in percentage terms.South Africa and Zimbabwe are the only African economies sampled that have stock market capitalization making more than 50% of their GDP as shown in Figure 3.All the other African countries sampled were below 50% of their GDP.
The market capitalization of emerging stock markets rose from $604 billion to $3,074 billion for the period 1990 to 1999.The trend continued in the 2000 with countries like Malaysia, Jordan, Jamaica, Chile, Saudi Arabia, Thailand, and Philippines accounting for the rise in stock market capitalization as portrayed in Figure 4.In terms of stock market capitalization most of the economies sampled are making less than 50% of GDP.With the African economies sampled economies with high stock market capitalization it is only South Africa, Morocco, and Egypt.Botswana with GDP like South Africa in percentage terms is cited as having low stock market capitalization and Zimbabwe with high stock market capitalization cited with low GDP as shown in Figure 5.
In the case emerging economies outside Africa countries cited with high stock market capitalization are cited in Figure 6 with relatively low GDP.Slovenia with low stock market capitalization is cited here as the country with the highest GDP so is Czech Republic.

Macroeconomic variable and stock market development
It is often argued that stock prices are determined by some fundamental macroeconomic variables such as the interest rate, the exchange rate and the inflation.Fama (1981) highlights that there exists a significant relationship between stock returns and other macroeconomic variables namely: inflation, national, output and industrial production.Stock market-output nexus has also been extensively studied (Habibullah and Baharumshah, 1996;Habibullah et al., 1999).These results indicate that there exists a long run relationship between stock returns and output.The levels of real economic activity, money supply M2, exchange rate and interest rate will likely influence stock prices through its impact on corporate profitability in the same direction.Shiller (1989) argues that changes in stock prices reflect changes in investor"s expectations about future values of certain economic variables that affect directly the pricing of equities.
The link between Capital market development and interest rate has in recent time been an issue among researchers (Ologunde et al., 2006;Anthony and Kwame 2008).It is asserted that the financial structure of a firm, that is, the blend of debt and equity financing, changes as economies develop.It moves towards equity financing through the stock market.If the rate of interest paid by banks to depositors is increased, investors will patronize the banks the more and fewer investors will invest on the capital market.This will lead to a decrease in capital investment in the economy.Hence, stock market performance and development will be lowered because the allocation of capital resources plays a crucial role in the determination of the rate of the nation"s output.Osei (2006) investigates both the long run and the short run associations between the Ghana stock market and macroeconomic variables.The paper establishes that there is co-integration between the macroeconomic variables and Ghana stock market.The results of the short run dynamic analysis and the evidence of co-integration mean that there are both short run and long run relationships between the macroeconomic variables and the index.In terms of Efficient Market Hypothesis (EMH), the study establishes that the Ghana stock market is information ally inefficient particularly with respect to inflation, treasury bill rate and world gold price.Kuwornu and Owusu-Nantwi (2011) examined the relationship between macroeconomic variables and stock market returns in Ghana using monthly data.Macroeconomic variables used were consumer price index (as a proxy for inflation), crude oil price, exchange rate and 91-day Treasury bill rate (as a proxy for interest rate).Full information maximum likelihood estimation procedure was used in establishing the relationship between macroeconomic variables and stock market returns.The empirical findings reveal that consumer price index (inflation rate) had a positive significant effect, while exchange rate and Treasury bill rate had negative significant influence on stock market returns.On the other hand, crude oil prices do not appear to have any significant effect on stock returns.Eita (2012) investigates the macroeconomic determinants of stock market prices in Namibia.Using VECM econometric methodology revealed that Namibian stock market prices are chiefly determined by economic activity, interest rates, inflation, money supply and exchange rates.An increase in economic activity and the money supply increases stock market prices, while increases in inflation and interest rates decrease stock prices.The results suggest that equities are not a hedge against inflation in Namibia, and contractionary monetary policy generally depresses stock prices.Fama (1981) argues that expected inflation is negatively correlated with anticipated real activity, which in turn is positively related to returns on the stock market.Therefore, stock market returns should be negatively correlated with expected inflation, which is often proxied by the short-term interest rate.Kaul (1990) studied the relationship between expected inflation and the stock market, which, according to the proxy hypothesis of Fama (1981) should be negatively related since expected inflation is negatively correlated with anticipated real activity, which in turn is positively related to returns on the stock market.Spyrou (2001) also studied the relationship between inflation and stock returns but for the emerging economy of Greece.Consistent with Kaul (1990) results, Spyrou (2001) found that inflation and stock returns are negatively related, but only up to 1995 after which the relationship became insignificant.Kyereboah-Coleman and Agyire-Tettey (2008) used cointegration and the error correction model techniques to show how macroeconomic indicators affect the performance of stock markets by using the Ghana Stock Exchange as a case study.The findings of the study reveal that lending rates from deposit money banks have an adverse effect on stock market performance and particularly serve as major hindrance to business growth in Ghana.Again, while inflation rate is found to have a negative effect on stock market performance, the results indicate that it takes time for this to take effect due to the presence of a lag period; and that investor"s benefit from exchange-rate losses as a result of domestic currency depreciation.Chow et al. (1993) using monthly data for the period 1977 to 1989 found no relationship for monthly excess stock returns and real exchange rate returns.When repeating the exercise, however, with longer than six months horizons they found a positive relationship between a strong dollar and stock returns.

Macroeconomic variables and investment
One way of linking macroeconomics variables and stock market returns is through arbitrage pricing (APT) (Ross, 1976).The q approach to the transmission mechanism increases the macroeconomic significance of stock markets which now take on an important role in managing the process of capital accumulation.APT focused on individual security returns (for selection of relevant studies see Fama, 1981Fama, , 1990;;Fama and French, 1989;Schwert, 1990;Ferson and Harvey, 1991;Black et al., 1997).It is also used in an aggregate stock market framework, where a change in a given macroeconomic variable could be seen as reflecting a change in an underlying systemic risk factor influencing future returns.Most of the empirical studies on APT theory, linking the state of the macro-economy to stock market returns, are characterized by modeling a short run relationship between macroeconomic variables and the stock price in terms of first difference, assuming trend stationarity (Andrew and Peter, 2007).
Portfolio optimization problems under partial information are becoming more and more popular, also because of their practical interest.They have been studied using both major portfolio optimization methodologies, namely Dynamic Programming (DP) and the "Martingale Method" (MM).While DP has a longer tradition in general, also MM has been applied already since some time for the cases when the drift/appreciation rate in a diffusion-type market model is supposed to be an unknown constant, a hidden finite-state Markov process, or a linear-Gaussian factor process.Along this line are the papers Lakner (1995Lakner ( , 1995)), and more recently Sass and Haussmann (2004).We consider the portfolio maximization problem under a hidden Markov setting, where the coefficients of the security prices are nonlinearly dependent on economic factors that evolve as a k-state Markov chain.
No satisfactory theory would argue that the relation between financial markets and the macroeconomics is entirely in one direction.However, stock prices are usually considered as responding to external forces.By the diversification argument that is implicit in capital market theory, only general economic state variables like inflation, money supply exchange rate and GDP will influence the pricing of large stock market aggregates.

Macroeconomic variables and stock market development
For the purpose of this empirical study, the unit of analysis is the 41 emerging economies stock market.Here, we will draw upon theory and existing empirical work as a motivation to select a number of macroeconomic variables that we might expect to be strongly related to the real stock price.The real stock price depends upon the expected stream of dividend payments and the market discount rate.Hence, any macroeconomic variable that may be thought to influence expected future dividends and/or the discount rate could have a strong influence on aggregate stock prices.The macro-economic variables selected as explained under theoretical model of this article are; money supply (MS), consumer price index (CI) and foreign exchange rate in US dollars (EXCH).The objective here is to test the effect of economic growth measured by GDP, and macroeconomic variables (MS, CPI, and EXCH) on stock market capitalization of emerging economies.In this paper, we will draw upon theory and existing empirical work as a motivation to select a number of macroeconomic variables that we might expect to be strongly related to the real stock price.
In this study, the model used by Sangmi and Mubasher ( 2013) was adopted and modified.In this empirical chapter least squares regression is again considered due to the numerous advantages that it has over other estimation techniques.The analytical model for the macroeconomic determinants of stock market performance is depicted by the modified model of Sangmi and Mubasher (2013).GDP was interacted with all the other macroeconomic variables one at time to determine the actual effect of these variables on stock market performance.The following models were run and the significance levels were tested at α=0.05 using different Robust OLS and FGLS, respectively.
Where GMS is the interaction of GDP and MS, GCPI is the interaction of GDP and CPI, GEXCH is the interaction and EXCH.The parameters were estimated using OLS technique.The least squares method produces the best straight line.However, there may in fact be no relationship or perhaps a nonlinear relationship between GDP, CPI, MS, EXCH and stock market capitalization hence a straight line is likely to be impractical.We assess how well the linear model fits the data.A model results in predicted values close to the observed data values.The fit of a proposed regression model should therefore be better than the fit of the mean model.It is assumed that the errors or disturbances have the same variance across all observation points.When this is not the case, the errors are said to be heteroskedastic and the model is corrected by using robust standard error to determine the significance of the parameters of interest.
The test of significance (α=0.05) for this model sought to establish the determinants of stock market performance in emerging economies.The inferential statistics such as the Pearson Product Moment correlation coefficient R 2 and the coefficient of determination R of the data set, as well as p-value and F-test statistics were used.The general use of differencing has been found to reduce the possibility of spurious regression results (Philip, 1986).Studies by Adams (1992) and Anyanwu and Udegbunam (1996) conclude that first-differencing achieves stationarity of variables and thus reduces the possibility of spurious results.Based on the suggestions of the aforementioned studies, and to roughly gauge the robustness and consistency of our estimation results, the regression Equation 1 is also estimated in first difference form.Differencing Equation 1 yields the following equations, which gives models 2 to 5. The stationarity of the variables are tested at α=0.05 significance level with the following empirical model, using the following techniques; Dynamic OLS and Newey-West, respectively.
We estimate the parameters of the linear regression model by the DOLS since it correct serial correlation and endogeneity problems in models.The dependent variable is the stock market performance.This measure equals the stock market capitalization divided by GDP.
The assumption behind this measure is that overall market size is positively correlated with the ability to mobilize capital and diversify risk on an economy-wide basis.This is consistent with Kemboi et al. (2012), Yartey (2008) and Levine and Zervos (1998).
Based on theory underpinnings discussed in the literature reviewed, we hypothesize a positive relation between exchange rate and stock prices.Mukherjee and Naka (1995) and Wongbangpo and Sharma (2002) among others, indicate that both exchange rate levels and changes affect the performance of a stock market.That is currency depreciation will have a favorable impact on a domestic stock market.The opposite should hold when the currencies of the country appreciates against foreign currencies.
The effect of money supply on stock prices can be positive or negative.Since the rate of inflation is positively related to money growth rate (Fama, 1981), an increase in the money supply may lead to an increase in the discount rate and lower stock prices.However, this negative effect may be countered by the economic stimulus provided by money growth, which would likely increase cash flows and stock prices (Mukherjee and Naka, 1995).Following Geske and Roll (1983), Chen et al. (1986), Wongbangpo and Sharma (2002), we hypothesize a negative relation between stock prices and consumer price index (CPI).The levels of real economic activity (proxied by CPI) will likely influence stock prices through its impact on corporate profitability in the same direction: an increase in real economic activity (fall in the consumer price index) may increase expected future cash and hence, raise stock prices, while the opposite effect would be valid in a recession.Consumer price index is used as a proxy for inflation rate.It is chosen because of its broad base measure to calculate average change in prices of goods and services during a specific period.Inflation is ultimately translated into nominal interest rate and an increase in nominal interest rate increases discount rate which results in reduction of present value of cash flows.An increase in inflation is expected to negatively affect the equity prices.
Consumer price index is used to measure macroeconomic stability.Macroeconomic stability may be an important factor for the development of the stock market.It is expected that the higher the macroeconomic stability the more incentive firms and investors have to participate in the stock market.The stock market in countries with stable macroeconomic environment is expected to be more developed.Consistent with previous studies inflation has been used as a measure of macroeconomic stability.Although there is no agreement on the relationship between macroeconomic stability and stock market development, it is argued that higher levels of macroeconomic stability encourage investors to participate in the stock market largely because the investment environment is predictable.Furthermore, macroeconomic stability influence firms profitability, and so the prices of securities in the stock market is likely to increase.Investors whose investments are experiencing a capital gain are more likely to channel their savings to the stock market by increasing their investments, and so this will enhance stock market development.This variable is proxied with consumer price index.The selection of these variables was based upon the present value model (PVM) theory and literature discussed.This study investigates the effect of macroeconomic variables on stock market performance in emerging economies for the period 1996 to 2011.
The technique used to estimate the coefficients of the linear regression model is the least squares method.Although the ordinary least squares (OLS) estimator is consistent in the presence of a serial correlation in the error term and it is well known that the OLS estimator contains the so-called second-order bias.Focus is on the dynamic ordinary least squares (DOLS) estimator instead of fully modified OLS estimators (FMOLS).The Newey-West estimates are also used to correct for the heteroskedasticity and serial correlation in the results.

Descriptive analysis of the variables
Table 1 summarizes the basic statistical features of the data under consideration including the mean, the Purchasing power parity (PPP) is a theory which states that exchange rates between currencies are in equilibrium when their purchasing power is the same in each of the two countries.This means that the exchange rate between the two countries should equal the ratio of the two countries' price level of a fixed basket of goods and services.When a country's domestic price level is increasing (that is, a country experiences inflation), that country's exchange rate must depreciate in order to return to PPP.In this article we proxy EXCH with PPP.The average EXCH for the period under investigation is 347.56 per US dollar.The huge difference between the minimum EXCH and maximum EXCH explains the high standard deviation of 1349.48.
In general, the precise evaluation of the normal distribution is given by the values of Skewness and Kurtosis.The Skewness show the amount and direction of skew (departure from horizontal symmetry), while the Kurtosis shows how tall and sharp the central peak is, relative to a standard bell curve.
The table also shows that most of the variables skewed positively, which means that there is a lack of symmetry, in other words, there is a deviation from symmetry of the distribution of data set.That is to say the large positive change is more common than large negative change in the variables.
Regarding peakness, the table shows that the excess kurtosis is larger than 3 for stock market capitalization and exchange rate hence the observed distribution has higher peak compared to the normal distribution.These suggest that the distributions of the variables are leptokurtic, that is non-normal.The data set are not exactly normally distributed since their respective mean, mode and median are not exactly the same, but the data was sufficiently appropriate for the purpose of the study.The mode values were not shown in the table due to space.To confirm the accuracy of the normality assumption, the JB statistics and the equivalent p-values were employed.The findings indicated that all variables are rejected at 1%.
The table revealed that all the variables possess the state of normal distribution, except and which are moderately skewed to the right.and have kurtosis values of more than three, and the series are called leptokurtic.As for the remaining variables, the values of kurtosis are less than three, and the series are called platykurtic (Bulmer, 1965).
The study results revealed that the volatility of the variables measured by the standard deviation is high for GDP and consumer price index.To confirm the accuracy of the normality assumption, we employed the JB statistics and the equivalent p-values.The findings indicated that all variables are rejected at 1%, except for consumer price index and policy rate at 1%.

Correlation analysis
Although it is not possible to comment on causation, the results reported in Table 2 revealed information on the strength of the relationships connecting the nine macroeconomic variables.It shows strong positive relationship between stock market capitalization and money supply and a negative correlation between consumer price index, exchange rate and market capitalization on the other hand.
These results support the inclusion of these macroeconomic variables in our analysis.Levine and Zervos (1998) established that measures of stock market development are positively correlated with measures of financial intermediary development.We examine if this complementary relationship exist in emerging economies.Data permitting, we average the data over the 1996 to 2011 period so that each country has one observation per variable.We compute the correlation between stock market development (measured by market capitalization) and all the other explanatory variables for this empirical chapter as shown in Table 2.
The correlation analysis reveals that the data sets are highly correlated with each other.
is found to correlate much more with and as compared with the rest of the variables.Also notable is that is highly correlated with both , and .
is found to be highly correlated with and respectively.Our finding confirms the work of Demirguc-Kunt and Levine (1996b).
The financial intermediary development and stock market development are complements rather than substitutes.In general, the data sets are highly correlated; meaning a change of one of the variable would result to a substantial change on the other variables which is expected for such macro-economic variables.

Regression analyses and hypothesis testing
However, before the regression analysis, we sought to establish the trend of the four data sets in order to establish the trend of the involved macro-economic variables.For the heterogeneity across the countries and heterogeneous serial correlation structure of error term, we employ three different panel unit root tests.The research considers three statistical tests for testing if each series in each panel are integrated of order one, otherwise known as stationarity test.These tests are Levin et al. (2002) test, Im et al. (2003) test and Hadri (2000) test for stationarity.
The LLC test is employed to test the stationarity of the panel for it allows heterogeneity of individual deterministic effects and heterogeneous serial correlation structure of the error terms, assuming homogeneous first order autoregressive parameters (Chiawa and Asare, 2009).LLC model tests the null hypothesis of the presence of unit roots against alternative of stationarity.Im et al. (2003) broadened the LLC test by presenting a more flexible and computationally simple test structure.The IPS test made the estimation for each of the "i" sections possible.IPS tests the null hypothesis of unit root against heterogeneous alternative hypotheses which specify that some series in the panel are non-stationary.Hadri (2000) test is distinctive from other two tests mentioned for testing the absence of unit roots, that is, variance of the random walk equals to zero.He proposes a parameterization which provides an adequate representation of both stationary and non-stationary variables and permits an easy formulation for a residual based Lagrange-Multiplier (LM) test of stationarity.Here, it is assumed that the time series for each cross-sectional unit is stationary around a deterministic level or trend, against the alternative hypothesis of a unit root.
Table 3 shows the results of panel unit root tests for each variable in the panel at level and at first difference.The results show that all the panels contain unit roots at level.However, at a differenced level, the panels are said to be stationary, though there may be possibility of non-stationary series in a stationary panel as the panel unit root test will not identify the particular series that is not stationary.This is only a drawback of the panel unit root test, nevertheless stronger and higher degree of power is gained in panel setting than in the usual single cross-sectional setting.This is as a result of the combination of information from time series and cross-sectional data which leads to improvement of power of test (Im et al., 2003).The tests are conducted in two folds.First, is carried out with the inclusion of individual effects followed by the inclusion of individual effect plus deterministic time.The results show that some of the panels contain unit root only at the inclusion of time trend while others confirm the presence of unit root at both levels of testing.All the variables are tested at 5% level of significance and the p-values displayed with their corresponding t-statistic in parenthesis.The results from these three tests provide support for treating all the individual series as non-stationary in their levels but stationary in their first differences.
In order to establish whether there exists a relationship between stock markets performance of emerging economies and macroeconomic variables, a regression analysis was conducted where the stock market performance is regressed against the four predictor variables; gross domestic product (GDP), consumer price index (CPI), money supply (M2), and exchange rate in dollars (EXCH) using robust standard errors.It is established that least squares method produces the best straight line.However, there may be in fact no relationship or perhaps no linear relationship between the explanatory variables and the dependent variable.By this a straight line model is likely to be impractical.Because of this it is important that we assess how well the linear model fits the data by employing standard error of estimates, coefficient of determination and analysis of variance.Four predictors were used (GDP, money supply, consumer price index and exchange rate), while the criterion variable was stock market capitalization.It was assumed that the selected macroeconomic variables were the best predictors for stock market performance; if not, then there was need to conduct a further tests in order to eliminate any potential biases to make the OLS regression estimated best linear unbiased estimators (BLUE).According to Addelbaki (2013), in conducting a quantitative research, one of the means of testing objectively the relationship among variables is to engage in an inquiry by having assumptions clearly stated and testing for theories deductively while guarding against bias, controlling for substitute clarifications, and be skillful to generalize and replicate findings.
For Table 4, the relationship between dependent variable (SMC) and independent variables (GDP, MS, CPI and EXCH) were determined.All the variables were not significant at all three traditional significant levels.The F test was significant indicating that the model fits the data set.The relationship was then viewed with MS, CPI and EXCH each at time using models 3.In each of this case also, the interaction effect was also determined.
The relationship between money supply and stock market performance was tested with model 3 as shown in Table 5.The intercept is 3.915 which is the stock market performance when all the independent variables are zero (0).It is misleading to interpret particularly if zero (0) is outside the range of the values of the independent variables.The relationship between the variable of interest money supply and stock market performance is described by 0.076.For every 100% increase in money supply (MS), stock market performance increases by 7.6%.The sign is as expected.The value of the test statistic t is 1.66 which implies that there is not enough evidence to infer the existence of a linear relationship between the MS and stock market performance.The interaction effect of the MS and GDP is also not statistically significant but there is enough evidence to infer linear relation between GDP and stock market performance.
In model 3, the effect of consumer price index (CPI) on stock market performance of emerging markets is considered (Table 6).This relationship is expressed by 0.008 with standard error of 0.0048 which yield a t-statistic of -1.67 assuming that all other factors are zero.The sign is not as expected.By implication, there is no evidence to conclude that the coefficient of CPI is not equal to zero (0).This may mean no evidence of linear relationship or there is linear relationship but because of the problem of multi-collinearity we fail to reject the null hypothesis.
The interaction effect of CPI and GDP on SMC is significant which implies that the coefficient of CPI when all other factors are zero is misleading since the effect CPI on SMC is also influenced by GDP.To determine the actual effect of CPI on SMC, interesting values of GDP must be plugged in to obtain the partial effect.The mean value of GDP is 18.64, so at the mean GDP, the effect of CPI on SMC is -0.048.The standard error of this coefficient is 0.016 which yields a t-statistic of -2.99.With relation of GDP to SMC, there is still enough evidence to conclude that there is a linear relation between them, confirming the relationship in model 3.
The relationship between exchange rate (EXCH) and stock market performance when all other factors are zero is significant so is the interaction effect of EXCH and GDP as shown in Table 7.This implies the coefficient of -0.053 is not appropriate.The actual effect of EXCH at the mean value of GDP is 0.842 with a standard error of 0.658.This implies there is not enough evidence to infer a linear relationship between EXCH and SMC.The sign is as expected.We test for serial correlation and Heteroskedasticity in the error term in each model using DW.This assumption is formally expressed as E (e i e j ) = 0 for all i ≠ j, which means that the expected value of all pair-wise products of error terms is zero.If indeed, the error terms are uncorrelated, the positive products willcancel those that are negative leaving an expected value of 0. If this assumption is violated, although the estimated regression model can still be of some value for prediction, its usefulness is greatly compromised.The estimated regression parameters remain unbiased estimators of the corresponding true values, leaving the estimated model appropriate for establishing point estimates and the model can be used for predicting values.However, the standard errors of the estimates of the regression parameters are significantly underestimated which leads to erroneously inflated t-values.Because testing hypotheses about the slope coefficients and computing the corresponding confidence intervals rely on the calculated t-values as the test statistics, the presence of correlated error terms means that these types of inferences cannot be made reliably.
A DW test of 0.351 implies the presence of positive autocorrelation in the error term at 5% significance level.That is the error covariances are not zero (0) and this will underestimate the variance of the parameters in the model and also can cause the rejection of the null hypothesis when it is true.Breusch-Pagan is used to test the null hypothesis that the error variances are all equal versus the alternative that the error variances are a multiplicative function of one or more variables.A large chi-square of 37.83 indicates that heteroskedasticity is present.The presence of heteroskedasticity alone does not cause bias or inconsistency in the OLS point estimates.The consequence of this is that the standard errors and t-statistics for the models are invalid.Because the Durbin-Watson statistic is far from 2 (the expected value under the null hypothesis of no serial correlation) and well below the 5% lower limit and upper limits, it is concluded that the disturbances are serially correlated.To address the problem, the variables are made stationary by first difference of all the variables.
Heteroskedasticity has serious consequences for the OLS estimator.Although the OLS estimator remains unbiased, the estimated SE is wrong.Because of this, confidence intervals and hypotheses tests cannot be relied on.In addition, the OLS estimator is no longer BLUE.Put more simply, a test of homoscedasticity of error terms determines whether a regression model's ability to predict a dependent variable is consistent across all values of that dependent variable.For heteroskedasticity, the null hypothesis of constant error variance is rejected.Heteroskedasticity has serious consequences for the OLS estimator.Although the OLS estimator remains unbiased, the estimated SE is wrong.
Because of this, confidence intervals and hypotheses tests cannot be relied on.In addition, the OLS estimator is no longer BLUE.
One possible way to address this problem is just to use heteroskedasticity-robust standard errors.OLS assumes that errors are both independent and identically distributed; robust standard errors relax either or both of those assumptions.Hence, when heteroskedasticity is present, robust standard errors tend to be more trustworthy.
The VIF test was performed in order to measure the extent to which the repressors were related to other repressors and to find out how the relationship affected the stability and variance of the regression estimates.Variance inflation factor of 4.54 shows that model have relatively moderate multicollinearity problem.Severe multicollinearity is problematic because it can increase the variance of the regression coefficients, making them unstable.
The F-probability for the model provides statistical evidence that the macroeconomic variables and their interaction to GDP simultaneously and jointly affect SMC.But a firm conclusion cannot be drawn based on these results because the regression results displayed are based on level, non-stationary data series and could represent a spurious problem.The presence of serial correlation in the error terms invalidate the use of R-squared and adjusted R-squared.
Since the variables under consideration are not stationary, the first differences of the variables are used to confirm the results using DOLS and Newey-West estimation technique.It was also realized that that correlation of first difference of the data series are not significant as shown in Table 8.This reduces the possibility of multicollinearity problem.In the analysis the null hypothesis of no autocorrelation and also the assumption of homoscedasticity for all the models discussed is rejected.The method of generalized least squares (GLS) is introduced to improve upon estimation efficiency when var(SMC) is not a scalar variance-covariance matrix.This technique allows estimation in the presence of AR(1) autocorrelation within panels and cross-sectional correlation and heteroskedasticity across panels.Although these conditions have no effect on the OLS method per se, they do affect the properties of the OLS estimators and resulting test statistics.Hypothesis testing based on the standard OLS estimator of the variance covariance matrix becomes invalid.
Using GLS gives the following results as shown in Tables 9 to 12.All the explanatory variables were significant in explaining variations in SMC.That is modeling heteroskedasticity and autocorrelation in the models, the variables were significant in explaining variations in SMC.
The p-value of Wald test for all the models were significant at all the traditional significant levels.

DOLS estimation of a co-integrated relation
Due to fact that the variables are non-stationary, the first difference of the variables is taken to make them stationary.To account for the problem of endogeneity and serial correlation, DOLS estimator is used.The results of DOLS estimation of model 3 of the difference in variables are shown in Table 13 and its residuals are also given.Wald chi-square of p-value 0.000 implies the model fit the data set.All the explanatory variables are significant in explaining variations in ∆SMC and the signs are as expected.
Table 14 test the effect of ∆MS on stock market performance.This relationship is expressed by 0.42 assuming that ∆GDP and the interaction effect of ∆GDPand ∆MS are zero (0).From the table there is enough evidence to conclude a linear relationship between the first difference of money supply (∆MS) and first difference of stock market capitalization (∆SMC).The interaction effect is also significant which implies the relationship between ∆SM and ∆SMC when all other variables as zero (0) is not appropriate.
To determine the statistical significance of the coefficient of the partial effect of ∆MS on stock market performance there was need to rerun the regression where the interaction variable is replaced with gross domestic product less the average gross domestic product multipled by ∆MS.This gives the new coefficient on ∆MS (the coefficient of partial effect), the estimated effect at gross domestic product of 18.64, along with a standard error.Running this new regression gives the standard error of as 0.0235, which yields t = 2.66.Therefore at the average gross domestic product, it is concluded that ∆MS has statistically significance positive effect on stock market performance.An increase in money supply will increase the liquidity in the economy resulting in an increase in the purchasing power of the citizenry.This means that more money will be available not just for consumption but also for investment hence, an increase in stock market performance.Also people tend to demand more when they have more money in their hands and thereby the prices of shares may increase which leads to stock market performances rising.These results support the real activity theorists" argument that an increase in money supply increases stock prices and vice versa.
There is also enough evidence to infer a linear relationship between ∆GDP and stock market performance for all the three models.Most industries are procyclical in nature, meaning that the firms in the industry do well as the economy does well and vice versa.If ∆GDP is high, the stock prices generally tend to be high as companies are doing better than otherwise.So, ∆GDP is an important determinant of stock prices.The results are in line with the findings of Levine and Zervos (1998), Garcia and Liu (1999), Yartey (2008) and Mishal (2011).
The relationship between consumer price index and stock market performance is significant and expressed by -0.081 when all other explanatory variables in the model are held constant.As shown in Table 15, model 3, the sign of the linear relationship is as expected.That is 100% increase in consumer price index decrease the performance of stock market by 8.1%.There is also enough evidence to conclude that there is linear relationship between the interaction of ∆CPI and ∆GDP and ∆SMC.With the interaction effect being significant then the actual effect of ∆CPI at mean GDP is -0.027 with a standard error 0.0104 which yields at test of -2.64.Therefore at the average ∆GDP, it is concluded that ∆CPI has statistically significance negative effect on stock market performance.The consumer price index is used as a proxy for inflation.In times of inflation, prices are always unstable and rising.Income is therefore devoted for consumption purposes.Savings and investment will therefore be negatively affected hence affecting stock market performance of emerging economies.The argument that the stock market serves as a hedge against inflation is based on the fundamental idea of Irving (1930), and is known as the Fisher effect.The Fisher effect states that in the long run, inflation and the nominal interest rate should move one-to-one with expected inflation.This implies that higher inflation will increase the nominal stock market return, but the real stock return remains unchanged.Therefore, investors are fully compensated.
Model 3 EXCH of Table 16 test the effect of first difference of exchange rate (dollar) on first difference stock market performance for emerging markets.The relationship is described by -0.017 with standard error 0.006 when ∆GDP and the interaction of ∆GDP and ∆EXCH are zero (0).The inverse relation is as expected.Since the interaction effect is significant, the linear relationship between ∆EXCH and ∆SMC when all other explanatory variables in model is zero is not appropriate since zero is not in the range of values for exchange rate.The partial effect of ∆EXCH at the mean GDP is expressed by -0.022 with a standard error 0.008 which yields a t-statistic of -2.75.Therefore at the average gross domestic product, it is concluded that ∆EXCH has statistically significance positive effect on stock market performance.
There are different theoretical approaches to understanding the relationship between the exchange rate and stock prices.Among these approaches, the two most prominent are the goods market approaches introduced by Dornbusch and Fischer (1980) and the portfolio balance approaches discussed by Frankel (1983).The portfolio balance approach stresses the role of capital account transactions on determining the relationship between the exchange rate and stock prices.This approach postulates a positive relationship between stock prices and exchange rates, with stock prices being the root cause of the relationship.
The results of the study support the hypothesis of a negative relationship between exchange rate and stock market capitalization of emerging economies and is consistent with the findings of Soenen and Hennigar (1988), Ajayi and Mougoue (1996) who have reported a significant, negative relationship between the exchange rate and stock return.However, it contradicts the findings of Maysami and Koh (2000).They explained that a stronger domestic currency lowers the cost of imported inputs and allows local producers to be more competitive internationally.Yip (1996) also explained that a strong exchange rate limits imported inflation and hence is perceived as favourable news for stock market performance.On the other hand, some studies, such as Bartov and Bodnar (1994)   The VIF test of 2.15; 1.94 and 1.77 for models 1 to 4 of Tables 17 to 21, respectively implies that there is not enough evidence to conclude that multicollinearity is present in the models.Hence the model does not affect stability and variance of the regression estimates.In Table 17 the relationship between macroeconomic variables (∆MS, ∆CPI and ∆EXCH) and ∆SMC are established by correcting for both heteroskedasticity and serial correlation using Newey-West technique.The variables in the model are significant and the signs are as expected.The result confirms that there is enough evidence to conclude that there is a linear relationship In Table 18, the ∆GDP, ∆MS and their interaction on the effect ∆SMC in emerging markets are examined.The linear relationship between the variable of interest MS is expressed by 0.042 with Newey-West standard error of 0.013 assuming that ∆GDP and the interaction of ∆MS and ∆GDP are constant.Since the value does not fall within the range of values for ∆GDP and also the fact that the interaction effect is significant, makes the interpretation of ∆MS tricky.To resolve this problem, we determine the partial effect of ∆MS given average ∆GDP and this coefficient is described by 0.061 with Newey-West standard error of 0.017 which yields t-statistic of 3.47.That is 1% increase in ∆MS given average ∆GDP yields of 0.061% increase in ∆SMC.It is established that ∆GDP complement MS in explaining variation in ∆SMC.The R-squared of 0.366 implies that the model explains 36.6% of the variations ∆SMC.Breusch-Pagan test of a small chi-square 0.457 implies that heteroskedasticity is probably not a problem or at least that if it is a problem it is not a multiplicative function of the predicted values.DW test serial correlation of 2.019 also failed to reject the null hypothesis of no serial correlation.Wald chi-square of 82.3 confirms that the model fits the data set.
The effect of ∆CPI on ∆SMC is expressed by -0.081 with a t-statistic of 2.91.This implies there is enough evidence to conclude that there is negative linear relationship between ∆CPI and ∆SMC assuming that other variables in the model are constant.That is as ∆CPI increases by 1% ∆SMC reduces by 0.081.It is also established that interaction effect has negative effect on ∆SMC.The partial effect of ∆CPI given ∆GDP is expressed by -0.249 with Newey-West standard error of 0.080 which yields a t-statistic of 3.11.Breusch-Pagan test the null hypothesis that the error variances are all equal versus the alternative that the error variances are a multiplicative function of one or more variables.A small chi-square 0.297 implies that heteroskedasticity is probably not a problem or at least that if it is a problem it is not a multiplicative function of the predicted values.DW of 1.92 also implies the errors are not serially correlated.Wald chi-square of 77.9 supports that the model fit the data and that the model is able to explain 34.3% of the variations in ∆SMC.
Table 20 examines ∆GDP, ∆EXCH and their interaction on the effect of ∆SMC.The linear relationship between the variable of interest ∆EXCH is expressed by -0.117 with Newey-West standard error of 0.042 assuming that GDP and the interaction of ∆EXCH and ∆GDP are constant.Since the value does not fall within the range of values for GDP and also the fact that the interaction effect is significant makes the interpretation of ∆EXCH tricky.To resolve this problem, the partial effect of ∆EXCH is determined given average GDP and this coefficient is described by -0.51 with Newey-West standard error of 0.170 which yields t-statistic of 2.99.That is 1% increase in ∆EXCH given average GDP yields 0.51% decrease in ∆SMC.The negative coefficient of the interaction variable implies that ∆GDP does not complement the ∆EXCH of the effect on ∆SMC.The R-squared of 0.369 implies that the model explains 36.9% of the variations ∆SMC.Wald chi-square of 69.5 confirms that the model fit the data set.Breusch-Pagan test of a small chi-square of 0.138 implies that heteroskedasticity is probably not a problem.DW test of serial correlation of 2.14 also fail to reject the null hypothesis of no serial correlation, making the regression result efficient and consistent.

Conclusion
Using a sample of 41 emerging stock economies over a period 1996 to 2011, it was discovered that gross domestic product, money supply, exchange rate in dollars and consumer price index are the important determinants of stock market development.Several policy implications can be drawn from this study.The government, in formulating monetary policy, must be aware of the fact that the stock market responds more favorably to an increase in the money supply.Leaders in emerging economies must also be conscious of the fact that stock prices tend to increase when the leaders implements expansionary policy to increase GDP and also depreciate exchange rates.
From the study, it can be observed that there exists a significant relationship between macroeconomic variables and the stock market performance.This relationship can either be positive or negative depending on which variable is being put under consideration.The study therefore recommends that the macroeconomic environment is very important and should be closely monitored to ensure stability.Emerging economies with stable macroeconomic environment enjoy increased activity at the stock market and hence an increased performance.Stock market performance is an indicator to the foreign investors on the stability of the stock market.It is therefore recommended that good measures should be put in place to promote the stock market activities which in turn increases the stock market performance.
It was established that financial intermediary (policy rate), stock market liquidity, exchange rate in dollars and the stabilization variable (consumer price change) are the important determinants of stock market development, while money supply does not prove to be significant.In addition, it was found that financial intermediaries and stock markets are complements rather than substitutes in development process.In order to promote stock market development in emerging economies, it is important to improve stock market liquidity, efficiently control exchange rate, develop financial intermediaries and then control inflation.
The salient conclusions drawn from this study suggest that strong macroeconomic variables are important for the stock market development in emerging country"s markets.To reverse the persistent anemic stock market performance trend in emerging economies, both domestic and external policy makers may have to place significant emphases on the maintenance of the voice and accountability, political stability, government effectiveness, rule of law, and control of corruption.The need to stabilize the macroeconomic indicators as well as improving upon the knowledge base of the citizenry is equally important for performance of stock markets in emerging economies.Although the empirical results are intriguing, they warrant further analysis.Much work remains to be done to better understand stock market development.
These findings also have important policy implications for emerging economies in relation to macroeconomic variables.Prudent management of macroeconomic variables can facilitate stock market development.
Rational management of macroeconomic variables ensures greater confidence in the stability of the economy as macroeconomic volatility magnifies the asymmetric information problem.First, macroeconomic variables such as consumer price index, exchange rate in dollars, money supply and GDP all play important role in determining the market performance.Therefore, policy makers have to maintain reasonable fiscal and monetary discipline in order to increase the demand for credit to the private sector, and subsequently influence the stock market development.

Table 1 .
Descriptive analysis of the variables., and the Jarque-Bera test for the data in their levels.The study revealed that gross domestic product (billions of dollars) varied mostly followed by consumer price index, money supply (millions of dollars).Money and quasi money comprise the sum of currency outside banks, demand deposits other than those of the central government, and the time, savings, and foreign currency deposits of resident sectors other than the central government.The mean value of MS for the emerging markets sampled for this article is 1.41E+08 million dollars with a standard deviation of 1.21E+09 million of dollars.This implies the changes in MS in emerging markets are very volatile with a minimum growth of 2761.33 to a maximum of 1.08E+10 million dollars over the period under investigation.

Table 2 .
Correlation coefficient of macroeconomic variables and SMC (levels).

Table 3 .
Results of the panel unit root test (C).
p-values and brackets is the t-values, the significance level is α =0.05.

Table 4 .
Relationship between macroeconomic variables and Stock Market Performance.

Table 5 .
Relationship between Stock Market Performance and Money Supply.

Table 6 .
Relationship between Stock Market Performance and Consumer Price Index.

Table 7 .
Relationship between Stock Market Performance and Exchange Rate.

Table 8 .
Correlation coefficient of macroeconomic variables and SMC (differences) The dependent variable is SMC; ** and * denote statistical significance at the 0.01 a .

Table 9 .
Relationship between Stock Market Performance and Macroeconomic Variables.

Table 11 .
Relationship between Stock Market Performance and Consumer Price Index.

Table 12 .
Relationship between Stock Market Performance and Exchange Rate.

Table 13 .
Relationship between stock market performance and macroeconomic variables.

Table 14 .
Relationship between stock market performance and money supply.

Table 15 .
Relationship between stock market performance and consumer price index.

Table 16 .
Relationship between Stock Market Performance and Exchange Rate.

Table 17 .
Relationship between stock market performance and macroeconomic variables.

Table 18 .
Relationship between stock market performance and money supply.

Table 19 .
Relationship between stock market performance and consumer price index.

Table 20 .
Relationship between stock market performance and exchange rate.

Table 21 .
Unit root test of residuals DOLS.model are constant in the case of each.There is also enough evidence to conclude that these variables are significant with the right signs at 5% significant level.The DW test of 1.97 implies that we fail to reject the null hypothesis that errors are serially correlated at 5% significance level.Breusch-Pagan test of chi-square of 0.438 fail to reject the null hypothesis.The results from the unit root tests of LLC, IPS and Hadri conclude that residuals from Newey-West regression are stationary as shown in the table.This implies that the Newey-West regression is not a spurious regression.