# How The Stock Market Contributes To Economic Growth

## Introduction

The debate of whether stock market is associated with economic growth or the stock market can be served as the economic indicator to predict future. According to many economists stock market can be a reason for the future recession if there is a huge decrease in the stock price or vice versa. However, there are evidence of controversial issue about the ability of prediction from the stock market is not reliable if there is a situation like 1987 stock market crashed followed by the economic recession and 1997 financial crises. (Stock market and economic growth in Malaysia: causality test).

The aim of the study is to find the relation between the stock market performance and the real economic activity in case of four countries The UK, The USA, Malaysia and Japan. With my limited knowledge I have tried to find out the role of financial development in stimulating economic growth. A lot of economists have different view about stock market development and the economic growth.

If we focus on some related literature published on this topic one question arises:

Is economic development is affected by stock market development?

Even though there are lots of debate on some are saying that stock market can help the economy but the effect of stock market in the economy especially in the economy is very little. Ross Levine suggested in his paper published in 1998 that recent evidence suggested stock market can really give a boom to economic growth. (REFERENCE)

It is not really possible to measure the growth by simply looking at the ups and down in the stock market indicator and by looking at the rates of growth in GDP. A lot of things can cause in the growth of stock market like changes in the banking system, foreign participation in the in the financial market may participate strongly. Apparently it seems that these developments can cause development of stock market followed by the good economic growth. But to check the accuracy one required to follow an appropriate method which would meaningfully measure whether stock price is really effecting the economic growth or not?

In my work I have tried to find out the co integrating relationship between Stock price and GDP and tried to check if there is a long run and short run relationship between the stock price and GDP.

The method used for the studies is Engle Granger co integration method. To do this I have used ADF (Augmented Dickey Fuller Test) to check for the stationary behaviour of the variables and then I have performed the Engle Granger Engle Granger co integration method followed by residual based error correction model. To check for the short run relationship I have used 2nd stage Engle Granger co integration method.

To check the causal effect of the four countries stock market and economic growth I used Granger Causality Method. In this paper I have reviewed some studies of scholars which I have discussed on the literature review part. This paper contains five parts

Part two is about the literature based on the past wok of scholars. Part Three discussed about the Data. Part four is about the methodology, Results are discussed on part five and part six is all about the summary and conclusion of the whole study.

In my work I have founded there is no long run relationship between stock market and economic growth in all four countries. In addition there is no causal relation between stock index yield and the national economy growth rate. The empirical results of the thesis concludes that the possibility of seemingly abnormal relationship between the stock index and national economy of these for countries.

## Literature Review:

Stock market contributes to economic growth in different ways either directly or indirectly. The functions of stock market are savings mobilization, Liquidity creation, and Risk diversification, keep control on disintermediation, information gaining and enhanced incentive for corporate control. The relationship between stock market and economic growth has become an issue of extensive analysis. There is always a question whether the stock market directly influence economic growth. A lot of research and results shows that there is a strong relationship between stock market and economic growth. Evidence on whether financial development causes growth help to reconcile these views.

If we go back to the study of Schumpeter (1912) his studies emphasizes the positive influence on the development of a country’s financial sector on the level and the potential risk of losses caused by the adverse selection and moral hazard or transaction costs are argued by him how necessary the rate of growth argues that financial sectors provides of reallocating capital to minimize the potential losses.

Empirical evidence from king and Levine (1983) show that the level of financial intermediation is good predictor of long run rates of growth, capital accumulation and productivity. Enhanced liquidity of financial market leads to financial development and investors can easily diversify their risk by creating their portfolio in different investments with higher investment. Demiurgic and Maksimovic (1996) have found positive causal effects of financial development on economic growth in line with the ‘supply leading’ hypothesis. According to his studies countries with better financial system has a smooth functioning stock market tend to grow much faster as they have access to much needed funds for financially constrained economic enterprises by the large efficient banks.

Related research was done for the past three decades focusing on the role of financial development in stimulating economic growth they never considered about the stock market. An empirical study by Ming Men and Rui on Stock market index and economic growth in China suggest that possible reason of apparent abnormal relationship between the stock Index and national economy in china. Apparent abnormal relationship may be because of the following reason inconsistency of Chinese GDP with the structure of its stock market, role played by private sector in growth of GDP and disequilibrium of finance structure etc. The study was done using the cointegration method and Granger causality test, the overall finding of the study is Chinese finance market is not playing an important role in economic development. (Men M 2006 China paper).

An article by Indrani Chakraborti based on the case of India presented in a seminar in kolkata in October, 2006 provides some information about the existence of long run stable relationship between stosk market capitalization, bank credit and growth rate of real GDP. She used the concept of the granger causality after using both the Engle-Granger and Johansen technique. In her study she found GDP is co-integrated with financial depth, Volatility in the stock market and GDP growth is co integrated with all the findings the paper explain that the in an overall sense, economic growth is the reson for financial development in India.(Chakraboty Indrani).

Few writers from Malaysia found that stock market does help to predict future economy. Stock market is associated with economic growth play as a source for new private capital. Causal relationship between the stock market and economic growth which was done by using the formal test for causality by C.J. Granger and yearly Malaysia data for the period 1977-2006. The result from the study explain that future prediction is possible by stock market.

A study focused on the relationship between stock market performance and real economic activity in Turkey. The study shows existence of a long run relationship between real economic activity and stock prices…………………………………… Result from the study pointed out that economic activity increases after a shock in stock prices and then declines in Turkish market from the second quarter and a unitary (Turkish paper)

An international time series analysis from 1980-1990 by By RAGHURAM G. RAJAN AND LUIGI ZINGALES shows some evidence of the relation between stock market and economic growth. This paper describes whether economic growth is facilitated by financial development. He found that financial development has strong effect on economic growth. (Rajan and Zingales, 1998)

The study of Ross LEVINE AND SARA ZERVOS on finding out the long run relationship between stock market and bank suggest a positive effect both the variables has positive effect on economic growth. International integration and volatility is not properly effected by capital stock market. And private save saving rates are not at all affected by these financial indicators. (Levine and Zervos 1998)

Belgium Stock market study with economic development shows the positive long run relationship between both the variables. In case of Belgium the evidences are quiet strong that Economic growth is caused by the development of the stock market. It is more focused between the period 1873 and 1935, basically this period is considered as the period of rapid industrialization in Belgium. The importance of the stock market in Belgium is more pronounced after liberalization of the stock market in 1867-1873. The time varying nature of the link between stock market development and economic growth is explained by the institutional change in the stock exchange. They also tried to find out the relationship to the universal banking system. Before 1873 the economic growth was based on the banking system and after 1873 stock market took the place. (Stock Market Development and economic growth in Belgium, Stijin Van Nieuwerburg, Ludo Cuyvers, Frans Buelens July 5, 2005)

Senior economist of the World Bank’s Policy research department Ross Levine has discussed about Stock market in his paper Stock Markets: A Spur to economic growth on the impact of development. Less risky investments are possible in liquid equity market and it attracts the savers to acquire an asset, equity. As they can sell it quickly when they need access to their savings, and if they want to alter their portfolio. Though many long term investment is required for the profitable investment. But reluctance of the investors towards long term investment as they don’t have the access to their savings easily. Permanent access to capital is raised by the companies through equity issues as they are facilitating longer term, more profitable investments and prospect of long term economic growth is enhanced as liquid market improves the allocation of capital. The empirical evidence from the study strongly suggests that greater stock markets create more liquidity or at least continue economic growth. (Levine. R A spur to economic Growth)

Another paper was focused on the linkages between financial development and economic growth using TYDL model for the empirical exercises by Purna Chandra Padhan suggests that both stock price and economic activity are integrated of order one and Johansen-Juselias Coin-integration tests for this study found one co integrating vector exists. It is proved by the spurious relation rule in this study the existence of at least one direction of causality. He described that bi-directional causality between stock price and economic growth meaning that economic activity can be enhanced by well developed stock exchange and vice-versa.

## ( Title: The nexus between stock market and economic activity: an empirical analysis for India Author(s): Purna Chandra Padhan Journal: International Journal of Social Economics Year: 2007 Volume: 34 Issue: 10 Page: 741 – 753 DOI: 10.1108/03068290710816874 Publisher: Emerald Group Publishing Limited)

Chee Keong Choong (Universiti Tunku Abdul Rahman Malaysia) Zulkornain Yusop (Universiti Putra Malaysia) Siong Hook Law (Universiti Putra Malaysia) Venus Liew Khim Sen (Universiti Putra Malaysia) Date of creation: 23 Jul 2003 tried to find out the importance of the causal relationship of Financial development and economic growth. The findings of their study usin autoregressive Distributed lag (ARDL) describes about the positive long run impact on economic growth Granger causality also suggest same results.

However, another study on Iran by N. Shahnoushi, A.G Daneshvar, E Shori and M. Motalebi 2008 Financial development is not considered as an effective factor to the economic growth. The study was focused on the causal relationship between the financial development and economic growth. Time series data used for the study from the period 1961-2004. Granger causality shows there is no co integrating relationship between financial development and economic growth in Iran only the economical growth leads to financial development.

Establishing link between savings and investment is very much important and financial market provides that. Transient or lasting growth is the ultimate affect of the financial market. Economic growth can be influenced by financial market by improving the productivity of the capital, Investment to firms can be channelled and greater capital accumulation by increasing savings. To ensure the stability of the financial market potential regulation is important due to asymmetric information, especially at the time of financial liberalization.

(Economic Development and Financial Market Tosson Nabil Deabes Moderm Academy for technology aand computer sciences; MAM November 2004 Economic Development & Financial Market Working Paper No. 2 )

## Data:

The empirical analysis was carried out using the quarterly data for The UK, The USA, Japan and Malaysia. The data were collected from the DataStream for the period 1993I to 2008III. Economic growth is measured as the growth rate of gross domestic product (GDP) of each country with the help of stock prices SP. For the software processing I used Eviews 6.0 for the planned regression in order to get the results. The empirical analysis is done from the quarterly data from the stock market indices and the and the GDP between the first quarter of 1993 and the fourth quarter of 2008. All the data has been extracted from the data stream and expressed in US$. The data for Japan share price is from Tokyo Stock Exchange. Malaysia’s Share price is form Kuala Lumpur Composite Index, UK’s is from UK FT all share price index and USA share price is taken from the DOW Jones industrial share price index.

The nature of the Data is series used for the time series regression.

## List of Variables:

UGDP

UK GDP

USP

UK Share price

LUGDP

Log of UK GDP

LUSP

Log of UK Share price

USGDP

USA GDP

USSP

USA (DOW Jones) Share price

LUSGDP

Log of USA GDP

LUSSP

Log of USA Share price

MGDP

Malaysia GDP

MSP

Malaysia Share price

LMGDP

Log of Malaysia GDP

LMSP

Log of Malaysia Share price

JGDP

Japan GDP

JSP

Japan Share Price

LJGDP

Log of Japan GDP

LJSP

Log of Japan Share price

## Methodology:

Engle and Granger (1987) first established the cointegration method. It is a method of measuring long term diversification based on data. Linear combination of two non stationary series shows that they are integrated in order one I(1) that is stationary. And this is a co integrated series.

Cointegration Long term common random trend between non stationary time series. The linear combination of both the nonstationary series can be stationary if both the variables are integrated in same order. Cointegration is a very powerful approach in the long term analysis because a common stochastic trend is shared in cointegration that mean two series will not drift separately too much. They might deviate from each other but in the long run but eventually the will revert back in the long run.

If there is very low correlation between the series still the series can be co-integrated as high correlation is not implied in cointegration. The reason for choosing the method as it will allow us to check the move between the variable in the long run even there might be a divergence in the short run.

The first step in the analysis is check each index series whether the series for the presence of unit root which shows whether the series is non stationary. The method that I followed to do this is Augmented Dickey Fuller Test (ADF). I proceed the Granger cointegration technique 1987 when the stationary requirements are met.

Cointegration long term common stochastic trend between nonstationary time series. If non-stationary series x and yare both integrated of same order and there is a linear combination of them that is stationary, they are called cointegrated series. A common stochastic trend is shared in Cointegration. It follows that these two series will not drift apart too much, meaning that even they may deviate from each other in the short-term, they will revert to the long-run equilibrium. This fact makes cointegration a very powerful approach for the long-term analyses.

Meanwhile, cointegration does not imply high correlation; two series can be co integrated and yet have very low correlations. Cointegration tests allow us to determine whether financial variables of different national markets move together over the long run, while providing for the possibility of short-run divergence. The first step in the analysis is to test each index series for the presence of unit roots, which shows whether the series are nonstationary. All the series must be nonstationarity and integrated of the same order. To do this, we apply both the Augmented Dickey-Fuller (ADF) test. Once the stationarity requirements are met, we proceed Granger bivariate cointegration (1987) procedure. 30 International Research Journal of Finance and Economics - Issue 24 (2009)

## Series Stationary Test:

In this study I have used Augmented Dickey Fuller Test (ADF) to test the stationarity of variables. ADF is test for unit root where I have checked the Unit root and strong negative numbers of unit root is being rejected by the null hypothesis (level of significance). The following regression for the unit root test in Eviews:

Is the white noise error tem. Is the difference operator.

## ,

## ()

## ()

Here with the test we can find the estimates of are equal to zero or not. Y is said to be stationary if the cumulative distribution of the ADF statistics by showing that if the calculated ratio of the coefficient is less than the critical value according to Fuller (1976). If we accept the Ho the sequence is predicted to be having unit root and if Ho is rejected then we can say that the series doesn’t have unit root. This proves that the series is stationary. The co –integration test can only be performed if both the sequences are all integrated of order I (1).

## Cointegration Test:

According to Engle and Granger (1987) to check for cointegration if both the variables and are integrated with order one the proposed method for cointegration residual-based test for cointegration (Engle-Granger method).

So from the above method we can find the equation

By regressing with

And after that and is denoted as the estimated regression coefficient vectors.

Then,

= – - is representing the estimated residual vector. If the residual is itegrated with zero that means the series for the residual is stationary, and and are then co integrated.

An in this situation (1, -) is called co-integrating vector.

Therefore is a co integrating equation, so, from it we can say that there is long run relationship between and.

## Granger causality test:

Granger causality test is applied if the relationship is lagged between the two variables to determine the direction of relation in statistical term. It gives information about the short term relationship between the variables.

In terms of conceptual definition causality is consist of different ideas, this concept produce a relation between caused and results were agreed upon. Aristo defines that there exist a link between causes and results and without causes these results are impossible. And this strong relationship.

Some economists believe that the idea of causality is the mix of both theoretical and explanation and statistical concept. The frontline operational definition of causality is given by some economist, but Granger is the one who provided the information to understand it correctly and completely.

Granger s operational causality definition depends of below hypotheses,

Next cannot be the reason of past.

1. Next cannot be reason of past. Certain causality is possible only with past causes present time or future time. Cause is always to be come true before the result. In addition, this makes time lagged between causes and results.

2. Causality can be determined only stochastic process. It is not possible to determine the causality between two deterministic processes.

After 1990s, Granger and Engle contributed to time series literature importantly. On these developments about time series analysis, some variations were done with Granger Causality test. According to this, possible long-term relationship would be tested and if 20 variables were co-integrated, long-term regression error equation s lagged value would be included in Granger Error Correction model as error correction term. Thus, Granger Causality test should be applied.

If there is no co-integration between the variables, it can be continued with Granger Causality Test without including error correction terms. If there is a co-integration between the variables, Granger Causality Test will be failed and it will be certainly necessary to be included error correction term into the models. Granger Causality Test, which depends on time series data, is made by the estimation of the equations below with Least Squares Method (LSM).

Xt = + j t j X + i t i Y + Ut

Yt = + j t j Y + j t j X + Ut

In Granger Causality test, there are three possible situations that one directional causality from x to y or y to x, opposite direction between x and y or one affect to other and independency of x and y each other. This situation changes according to chosen of null hypothesis and lagged values randomly in equations above whose parameters are whether equal to zero or not. According to researches, randomly choice makes causality incline to deviations importantly.

To understand this test clearly it can be talked about below equation;

t (LNGDP) = 0 + t inii (LNGDP)1+ t I nii (LND1)1+ Ut

To apply Granger Causality test under null hypothesis, which illustrates coefficients of financial deepening variables (LND1) are meaningful (equal to zero) and then F-statistics can

be calculated. If null hypothesis is not rejected then it is possible to say that Granger causality

test accepts that financial deepening causes economic growth. The direction can be either negative or positive (Granger and Engle, 1987). Indicators of the economic growth and the financial deepening are variables, which are used for Granger Causality test. Moreover, this test can determine the effects of one variable on the other.

## Test result for Unit Root:

Augmented Dickey Fuller Model (ADF) is used to test the stationary of each variable. Null and alternative hypothesis describes about the investigation of unit root. If the null is accepted and alternative is rejected then the variable non stationary behaviour and vice versa is stationary. Form the result of Augmented Dickey Fuller test of the four countries variables (Log GDP and Log Share price) shows that the entire variable has unit root at level which proves that the series is not stationary. However, the result from the first difference shows the significance at 1%, 5% and 10% critical value and found to be stationary behaviour. Therefore, it suggests that all the variables are integrated of order one.

Variables

level/1st

Difference

## Augmented Dickey Fuller Statistic(ADF) test Japan

Conclusion

t statistic

value

With Trend

t statistic

value

With trend and

Intercept

1%

5%

10%

1%

5%

10%

GDP

Level

-2.653258

-3.522887

-2.901779

-2.588280

-2.693600

-4.088713

-3.472558

-3.163450

1st Difference

-9.053185

-3.524233

-2.902358

-2.588587

-9.003482

-4.090602

-3.473447

-3.163967

Share Price

Level

-2.116137

-3.522887

-2.901779

-2.588280

-2.203273

-4.088713

-3.472558

-3.163450

1st Difference

-6.899295

-3.524233

-2.902358

-2.588587

-6.844396

-4.090602

-3.473447

-3.163967

## Table 01: Unit root test for stationary Japan

If we have a look on the unit root test for the variables GDP and Share price to find out the stationary behaviour the Augmented Dickey Fuller Test with intercept and with intercept and trend in level and first difference. The t statistic value with trend is -2.653258 which is higher than the critical values in 1%, 5% and 10% critical value. The same applies with intercept and trend as the t statistic value -2.693600 is higher than the critical value in all the level of critical value. So from the nature of stationary behaviour we can say in level GDP shows nonstationary behaviour. And the first difference LnGDP is integrated with order one. In case of LnSP the results with intercept and with intercept trend in level are -2.116137 and -2.203273 which is higher than the critical values shows non stationary behaviour as they are higher than the critical value. The unit root test for the variables at first difference shows stationary as the t statistic value is than the critical value in all level and they are integrated in order one.

Variables

level/1st

Difference

## Augmented Dickey Fuller Statistic(ADF) test Malaysia

Conclusion

t statistic

value

With Trend

t statistic

value

With trend and

Intercept

1%

5%

10%

1%

5%

10%

GDP

Level

-1.195020

-3.522887

-2.901779

-2.588280

-1.933335

-4.088713

-3.472558

-3.163450

1st Difference

-5.951843

-3.524233

-2.902358

-2.588587

-5.923595

-4.090602

-3.473447

-3.163967

Share Price

Level

-1.900406

-3.522887

-2.901779

-2.588280

-1.891183

-4.088713

-3.472558

-3.163450

1st Difference

-7.842122

-3.524233

-2.902358

-2.588587

-7.779757

-4.090602

-3.473447

-3.163967

The unit root test result for LMGDP and LMSP values presented in natural logarithm. And the level values with intercept and with intercept and trend for LMGDP is -1.195020 and -1.93335 respectively. The values are higher than the critical value means the series has non stationary behaviour. On the other hand the 1st difference values with intercept and with intercept and trend are -5.951843 and -5.923595 respectively. The 1st difference values are integrated with order one. And in the same way I did the ADF test to check for Stationary behaviour of LMSP in level and first difference with intercept and trend. The values in level are -1.900406 and -1.891183 with intercept and trend us higher than the critical value and the series is not integrated with order one. The first difference t statistic values are -7.842122 and -7.779757 with intercept and with intercept and trend respectively are less than the critical value in both the case implies that the series is integrated with order one.

Variables

level/1st

Difference

## Augmented Dickey Fuller Statistic(ADF) test UK

Conclusion

t statistic

value

With Trend

t statistic

value

With trend and

Intercept

1%

5%

10%

1%

5%

10%

GDP

Level

-0.690866

-3.522887

-2.901779

-2.588280

-2.377333

-4.088713

-3.472558

-3.163450

1st Difference

-7.474388

-3.524233

-2.902358

-2.588587

-7.439027

-4.090602

-3.473447

-3.163967

Share Price

Level

-1.711599

-3.522887

-2.901779

-2.588280

-1.261546

-4.088713

-3.472558

-3.163450

1st Difference

-7.254574

-3.524233

-2.902358

-2.588587

-7.391821

-4.090602

-3.473447

-3.163967

The results from Augmented Dickey Fuller test (ADF) for UK GDP in level with intercept and with intercept and trend is –0.690866 and -2.377333 respectively. Both the values in level are higher than the critical value and are integrated in order 0 shows non stationary behaviour. The t statistic values in 1st difference with intercept and with intercept and trend are -7.474388 and -7.439207 respectively. Which suggest that the critical values are less than the critical values in 1%, 5% and 10% level. So from the above hypothesis it can be said that it series is integrated with order one. When I performed the unit root test using the same method the series in level with intercept and with intercept and trend the values in are -1.711599 and -1.261546 respectively. The values are higher than the critical values implies that they are not integrated in order one shows non stationary behaviour. However, the 1st difference value of log natural share price is -7.254573 and -7.391821 with intercept and with intercept and trend respectively. So from the result we can say that the series is integrated in order one in both the cases with intercept and with intercept and trend. So the series in first difference is stationary.

Variables

level/1st

Difference

## Augmented Dickey Fuller Statistic(ADF) test USA

Conclusion

t statistic

value

With Trend

t statistic

value

With trend and

Intercept

1%

5%

10%

1%

5%

10%

GDP

Level

-3.244801

-3.522887

-2.901779

-2.588280

2.866507

-4.088713

-3.472558

-3.163450

1st Difference

-5.010864

-3.524233

-2.902358

-2.588587

-5.010864

-4.090602

-3.473447

-3.163967

Share Price

Level

-2.074732

-3.522887

-2.901779

-2.588280

-0.359637

-4.088713

-3.472558

-3.163450

1st Difference

-8.181234

-3.524233

-2.902358

-2.588587

-8.735399

-4.090602

-3.473447

-3.163967

Augmented Dickey Fuller Statistic in case of the variable of USA LUSSP and LUGDP I have used the same method using intercept and intercept and trend in level and first difference. The level t statistic value for LUSGDP is -3.244801 and -2.866507 respectively with intercept and with intercept and trend. The result for USA is same as the other country which is higher than the critical values. Proves that the series is not integrated with order one and is nonstationary. Whereas the first difference t statistic value for LUSGDP is less than the critical value. The t statistic value LUSGDP with intercept is -5.010864 and -5.010864 with intercept and trend. In this case both the values are lesser than the critical value implies that the series is integrated with order one in first difference. While taking the values in level and 1st difference in case of LUSSP the tstatistic value in level are -2.074732 and -0.359637 in level respectively with intercept and wit intercept and trend. Still the series is showing the same nature in level as they are higher than the critical values and the series is not integrated in order 0. The first difference value for LUSSP series with trend and with trend and intercept is -8.181234 and -8.735399 respectively which is less than the critical value implies the series is integrated with order one.

## Co integration Test:

Two step procedure of Engle-Granger cointegration is to check for the long run relationship between the variables. The first stage was run by using traditional OLS method. To do this we need to check whether the series is stationary or not. Which we have checked before by doing ADF test on each series. where the result shows that the series is integrated with order (1).

Engle-Granger representation theorem that might have an error correction mechanism is the series is integrated.

In this case the long run OLS model is as follows in case of Japan:

LJGDP = 7.97824432568 + 0.163668097988*LJSP

Dependent Variable: LJGDP

Method: Least Squares

Date: 12/17/09 Time: 20:30

Sample: 1991Q1 2009Q2

Included observations: 74

Coefficient

Std. Error

t-Statistic

Prob.

C

7.978244

0.120791

66.04995

0

LJSP

0.163668

0.048847

3.350602

0.0013

R-squared

0.134891

Mean dependent var

8.381114

Adjusted R-squared

0.122876

S.D. dependent var

0.10605

S.E. of regression

0.099321

Akaike info criterion

-1.75426

Sum squared resid

0.710261

Schwarz criterion

-1.69199

Log likelihood

66.90753

Hannan-Quinn criter.

-1.72942

F-statistic

11.22653

Durbin-Watson stat

0.310636

Prob(F-statistic)

0.001287

From the above model I have saved the residual series and performed ADF test with trend and without trend and the values are as follows in the table:

Unit Root test for residual

Series saved residual RJP

T statistic

Test critical values:

1% level

5% level

10% level

With intercept

-2.831807

-3.522887

-2.901779

-2.588280

With intercept and trend

From the above table we can see that the result is significant only in 10% level. Which suggest that there might be a long run relationship between the variables. But there is no long run relationship at 1% and 5% significant level as both the values are higher than the critical value.

2nd stage regression result:

LJGDP = 7.96681067902 + 0.170453164194*LJSP + 0.819211725701*RJP(-1)

Dependent Variable: LJGDP

Method: Least Squares

Date: 12/31/09 Time: 18:51

Sample (adjusted): 1991Q2 2009Q2

Included observations: 73 after adjustments

Coefficient

Std. Error

t-Statistic

Prob.

C

7.966811

0.064529

123.4601

0

LJSP

0.170453

0.026119

6.525992

0

RJP(-1)

0.819212

0.064206

12.75915

0

R-squared

0.747462

Mean dependent var

8.384005

Adjusted R-squared

0.740246

S.D. dependent var

0.103806

S.E. of regression

0.052906

Akaike info criterion

-3.00038

Sum squared resid

0.195932

Schwarz criterion

-2.90625

Log likelihood

112.5137

Hannan-Quinn criter.

-2.96286

F-statistic

103.5928

Durbin-Watson stat

1.958683

Prob(F-statistic)

0

2nd stage regression suggest that there is short run relationship between stock market and economic growth. As from the table values after running the regression with the help of one intercept and lagged value of the residual save from the first stage regression. Here we can see that the all the coefficient has positive values and r-sruared (0.747462) is less than the Durbin-Watson value(1.958683). so form the results we can see that there exists a short run relationship between stock market and economic growth.

Malaysia

Following the same stages on Malaysia, by running the regression on OLS to check the long run relationship between stock market and economic growth in Malasia. The equation to check the first stage regression is:

LMGDP = 8.2331829641 + 0.340689829517*LMSP

The result from the above regression are described in the following table:

Dependent Variable: LMGDP

Method: Least Squares

Date: 12/17/09 Time: 21:00

Sample: 1991Q1 2009Q2

Included observations: 74

Coefficient

Std. Error

t-Statistic

Prob.

C

8.233183

0.644484

12.77484

0

LMSP

0.34069

0.116332

2.928597

0.0046

R-squared

0.106441

Mean dependent var

10.11598

Adjusted R-squared

0.094031

S.D. dependent var

0.407894

S.E. of regression

0.388243

Akaike info criterion

0.972285

Sum squared resid

10.85275

Schwarz criterion

1.034557

Log likelihood

-33.97453

Hannan-Quinn criter.

0.997126

F-statistic

8.576678

Durbin-Watson stat

0.054361

Prob(F-statistic)

0.004557

Unit Root test for residual

Series

T statistic

Test critical values:

1% level

5% level

10% level

With intercept

-1.301997

-3.522887

-2.901779

-2.588280

With intercept and trend

From the above regression and after saving the residual I performed and ADF test with trend and without trend on the residual series. Here the result suggests that the t statistic value is higher than the critical values of 1%, 5% and 10% level. Which suggest that residual series is non stationary and there is no relationship between the variables in long run.

The estimated equation in error correction model is as follows:

LMGDP = 8.13761928798 + 0.360964712114*LMSP + 0.965225800038*R(-1)

Dependent Variable: LMGDP

Method: Least Squares

Date: 01/01/10 Time: 23:15

Sample (adjusted): 1991Q2 2009Q2

Included observations: 73 after adjustments

Coefficient

Std. Error

t-Statistic

Prob.

C

8.137619

0.147701

55.09505

0

LMSP

0.360965

0.02665

13.54478

0

R(-1)

0.965226

0.027335

35.31042

0

R-squared

0.952382

Mean dependent var

10.12619

Adjusted R-squared

0.951022

S.D. dependent var

0.401091

S.E. of regression

0.088766

Akaike info criterion

-1.96541

Sum squared resid

0.551553

Schwarz criterion

-1.87128

Log likelihood

74.7374

Hannan-Quinn criter.

-1.9279

F-statistic

700.0218

Durbin-Watson stat

2.075716

Prob(F-statistic)

0

2nd stage results are suggesting about the short run relationship between the variables. As we can see from the is less than the Durbin-Watson Statistic. So from the result we can say that there exist a co-integrating relationship between stock market and economic growth in short run.

## UK

Considering the case of UK to find out the relationship both in long and short run I used the same procedure to find out the relationship. As all the variables are integrated with order one which suggests the variables are stationary. Now by applying the Engle Granger cointegration method to estimate the co integrating vector in OLS and then examining the residual series. Cointegration for the long run depends on the residual series. Here I defined the residual series a RUK for the variables LUGDP (log of UK GDP) and LUSP(log of UK share price). If we look at the table of the unit root test for the residual series of the Co-integrating regression of LUGDP and LUSP the residual series RUK is -1.355485 with intercept and -2.426938 with intercept and trend. Where both the result for unit root test by applying Augmented Dickey Fuller test suggests that the residual series has a nonstationary behaviour in both the case with intercept and with intercept and trend. As the critical value for at 1%, 5% and 10% is -3.522887, -3.522887 and -2.588280 respectively with intercept and -4.088713, -3.472558 and -3.163450with intercept and trend. As the t statistic value is higher than the critical values in both the case, so from the result we can say that the residual series in non stationary and there is no long run relationship between the variable.

Dependent Variable: LUGDP

Method: Least Squares

Date: 12/17/09 Time: 21:10

Sample: 1991Q1 2009Q2

Included observations: 74

Coefficient

Std. Error

t-Statistic

Prob.

C

6.41427

0.52629

12.18771

0

LUSP

0.790239

0.064275

12.29475

0

R-squared

0.677363

Mean dependent var

12.87916

Adjusted R-squared

0.672882

S.D. dependent var

0.332711

S.E. of regression

0.190291

Akaike info criterion

-0.45386

Sum squared resid

2.607181

Schwarz criterion

-0.39159

Log likelihood

18.79298

Hannan-Quinn criter.

-0.42902

F-statistic

151.1608

Durbin-Watson stat

0.149084

Prob(F-statistic)

0

## Unit Root test for residual

Series residual saved

T statistic

Test critical values:

RUK

1% level

5% level

10% level

With Intercept

-1.355485

-3.522887

-2.901779

-2.588280

With intercept and trend

-2.426938

-4.088713

-3.472558

-3.16345

## 2nd stage

Dependent Variable: LUGDP

Method: Least Squares

Date: 01/04/10 Time: 17:57

Sample (adjusted): 1991Q2 2009Q2

Included observations: 73 after adjustments

Coefficient

Std. Error

t-Statistic

Prob.

C

6.375942

0.207063

30.79235

0

LUSP

0.795176

0.025265

31.47291

0

RUK(-1)

0.937553

0.046342

20.23103

0

R-squared

0.952647

Mean dependent var

12.88329

Adjusted R-squared

0.951294

S.D. dependent var

0.333094

S.E. of regression

0.073512

Akaike info criterion

-2.3425

Sum squared resid

0.378285

Schwarz criterion

-2.24837

Log likelihood

88.50121

Hannan-Quinn criter.

-2.30499

F-statistic

704.1223

Durbin-Watson stat

2.248029

Prob(F-statistic)

0

## USA

In case of USA to find out the relationship between stock market and economic growth using Engle Granger cointegration method we find the following results.

LUSGDP = 6.422388123 + 0.32041281224*LUSSP

Dependent Variable: LUSGDP

Method: Least Squares

Date: 12/31/09 Time: 02:02

Sample: 1991Q1 2009Q2

Included observations: 74

Coefficient

Std. Error

t-Statistic

Prob.

C

6.422388

0.140166

45.82

0

LUSSP

0.320413

0.015722

20.38041

0

R-squared

0.852266

Mean dependent var

9.274948

Adjusted R-squared

0.850214

S.D. dependent var

0.166293

S.E. of regression

0.064359

Akaike info criterion

-2.62203

Sum squared resid

0.29823

Schwarz criterion

-2.55975

Log likelihood

99.01496

Hannan-Quinn criter.

-2.59719

F-statistic

415.3609

Durbin-Watson stat

0.124101

Prob(F-statistic)

0

## Unit Root test for residual

Series

Residual saved RUS

T statistic

Test critical values:

1% level

5% level

10% level

With intercept

-0.638033

-3.522887

-2.901779

-2.588280

With intercept and trend

-1.430799

-4.088713

-3.472558

-3.163450

After saving the residuals from the 1st stage regression RUS I did the ADF test on it where we can see the t statistic value is literally higher than the 1%, 5% and 10% critical value in both the cases with intercept and with intercept and trend. As we can see the critical values are -3.552287, -2.901779 and -2.588280 with intercept, -1.430799, -3.472558 and -3.163450 in 1%, 5% and 10% level respectively. So the possibility for having long run relationship between GDP and stock price doesn’t exist in case of USA.

## 2nd stage regression:

Dependent Variable: LUSGDP

Method: Least Squares

Date: 01/05/10 Time: 21:36

Sample (adjusted): 1991Q2 2009Q2

Included observations: 73 after adjustments

Coefficient

Std. Error

t-Statistic

Prob.

C

6.400276

0.051084

125.29

0

LUSSP

0.323107

0.005722

56.46591

0

RUS(-1)

0.972896

0.043361

22.43708

0

R-squared

0.981148

Mean dependent var

9.278975

Adjusted R-squared

0.980609

S.D. dependent var

0.163769

S.E. of regression

0.022805

Akaike info criterion

-4.683433

Sum squared resid

0.036405

Schwarz criterion

-4.589305

Log likelihood

173.9453

Hannan-Quinn criter.

-4.645922

F-statistic

1821.53

Durbin-Watson stat

2.153933

Prob(F-statistic)

0

## Granger Causality test:

Pair wise Granger Causality Tests

Date: 01/05/10 Time: 22:03

Sample: 1991Q1 2009Q2

Lags: 3

Null Hypothesis:

Observation

F-Statistic

Prob.

LJSP does not Granger Cause LJGDP

71

1.46842

0.2315

LJGDP does not Granger Cause LJSP

0.7659

0.5173

After performing the causality tests on the series DLJGDP and DLJSP with lag 3 according to the causality table to reject the null hypothesis that GDP does not granger cause LJSP. No causal relationship exists between share price and GDP in Japan.

Pair wise Granger Causality Tests

Date: 01/05/10 Time: 22:18

Sample: 1991Q1 2009Q2

Lags: 3

Null Hypothesis:

Observations

F-Statistic

Prob.

LMSP does not Granger Cause LMGDP

71

14.8418

0.0000002

LMGDP does not Granger Cause LMSP

0.65292

0.584

We can see the same result when we performed the causality test LMGDP and LMSP. Here we cannot reject the null which shows that there is no causal relationship between stock price and GDP.

Pair wise Granger Causality Tests

Date: 01/05/10 Time: 22:22

Sample: 1991Q1 2009Q2

Lags: 3

Null Hypothesis:

Observations

F-Statistic

Prob.

LUSP does not Granger Cause LUGDP

71

4.17743

0.0092

LUGDP does not Granger Cause LUSP

0.58556

0.6267

Pair wise Granger Causality Tests

Date: 01/05/10 Time: 22:24

Sample: 1991Q1 2009Q2

Lags: 3

Null Hypothesis:

Observations

F-Statistic

Prob.

LUSSP does not Granger Cause LUSGDP

71

2.50276

0.0671

LUSGDP does not Granger Cause LUSSP

0.51256

0.6751