“Stock market performance and GDP growth ": evidence from the Netherlands”.

(1)

University of Amsterdam

Amsterdam Business School

Master in International Finance

Master Thesis

“Stock market performance and GDP growth. Evidence from the Netherlands”

Author: Bhoj Adhikari Thesis Supervisor: Dr. R. Matta

(2)

Abstract

It is generally assumed that stock market acts as a leading indicator of the future economic activity. This paper investigates the effectsof positive or negative stock market performance on GDP growth of the Netherlands

.

By applying three popular time series econometric techniques (cointegration technique, vector error correction model and generalized impulse response function), we examine the long run relationship between stock market performance and GDP growth of the Netherlands from 2000 to 2010. The empirical results can be summarized as follows. Firstly, Johansen’s cointegration test shows that stock market performance and GDP growth are cointegrated, implying that positive stock market performance will lead to positive GDP growth and vice-versa . Secondly, the vector error correction model (VECM) results show that if there is a disturbance in the whole system, stock market performance acts as a stabilising force bringing the whole system back to equilibrium in long run despite short run divergence. Thirdly, generalized impulse response function shows that spill over effect from positive shock in stock market performance lasts longer on GDP growth, than the other way around. Therefore, results from this paper add more weight to the view that the stock markets are bellwether of the future economic activity .

(3)

1. Introduction and Purpose

It is generally assumed that stock market acts as a leading indicator of the future economic activity. Persistent stock market declines can lead to lower consumer confidence and business outlook, which, in turn, typically leads to lower consumption and investment spending, thus, increasing the likelihood of economic slowdown. On the contrary, persistent increase in stock market will lead to increased consumer confidence and business outlook leading to higher consumption and investment, thus, increasing GDP growth. Investors are also willing pay higher for the price of share if they believe that future growth will lead of higher corporate earnings increasing the shareholder return.

The connection between the stock market and the economy is of vital importance to the wellbeing of a country. Stock markets are a vital component of economic development as they provide companies with a platform to raise long-term capital and also provide investors with platform to invest their surplus funds. Stock market therefore plays vital role in encouraging investors with surplus funds to invest in financial instruments that matches their liquidity preference and risk appetite. Thus, this helps in efficient allocation of savings, which is ultimately translated into investment thus leading to increased economic activity

.

There is consensus that stock markets provide the useful information about the future development of an economy, i.e. stock market acts as the leading indicator of future economic activity. Fama [1] and Schwert [2] claim that there are three explanations for the strong link between stock prices and real economic activity: “First, information about future real activity may be reflected in stock prices well before it occurs this is essentially the notion that stock prices are a leading indicator for the well - being of the economy. Second, changes in discount rates may affect stock prices and real investment similarly, but the output from real investment does not appear for some time after it is made. Third, changes in stock prices are changes in wealth, and this can affect the demand for consumption and investment goods”. A similar research done by Barro [3], using US data between 1927 and 1988, found that stock market predicted eight of the nine periods generally designated recession.

The objective of this study is to analyze the much debated topic of whether the stock market performance can predict GDP growth and the extent of relationship between them. In order to test this relationship, we conduct a cointegration analysis of the stock market performance of the leading Dutch index and GDP growth of the Netherlands for 10 year time period. Cointegration can be viewed as the statistical expression of the nature of equilibrium relationships, with cointegrated variables sharing common stochastic trends. The results of the study will aid us to gain insight into how stock market performance impacts GDP growth in the Netherlands. Therefore, the paper aims to help policymakers as well as investors help predict the future economic activity.

From the empirical analysis we can infer that stock market performance and GDP growth have long run relationship such that positive stock market performance will lead to positive GDP

(5)

5 growth in the Netherlands and vice versa. The results show that 1% increase in stock market leads to roughly 2% increase in GDP growth. Similar research done by Arturo and Mishkin [4] for US economy found that stock prices are useful with one- to three-quarter horizons. Henry, Olekalns, and Thong [5] examined this relationship for OECD plus 5 South East Asian economies found out that stock returns contain information useful for forecasting GDP growth but the magnitude of the effect is small. Ayalward and Glen [6] who studied whether stock prices could be predictive for GDP growth for 23 countries (including US and UK) found out that 10% increase in stock prices results in 0.5%-1% rise in GDP. The magnitude of predictive power in this paper is much more pronounced than other previous papers, including those mentioned above. The discrepancy in result is due to the fact that the stock market has fallen by half during the decade of 2000, while GDP has grown at the pace of 0.9% per year. Investors piled into stock market during 1990’s as GDP growth boomed, but as GDP growth failed to keep up with the investors’ expectation during decade of 2000, the stock market fell sharply. Despite this discrepancy, the overall results point to the fact that all stakeholders of economy can look to stock market for signs of future economic activity.

This research paper is divided into roughly five sections. The first section will review the research that has been done previously by academics to uncover the link between stock market performance and GDP. The next section will outline the data and methodology that will be used to perform the empirical analysis. The third one will discuss the econometric methodology used in this paper and discuss the findings. The fourth section will outline the limitations of research and conclude the papers with conclusion. The last section will outline the references and output of the empirical tests.

2. Motivation of Research

Economic theory suggests that stock prices reflect the expectations about future corporate earnings, which are likely to be correlated with health of economy. The forward looking nature of stock market returns would imply that stock prices should be valuable as leading indicators of economic growth. Despite the significant amount of empirical literature that outlines the importance of stock market performance as important predictor of future economic activity, there are still doubts whether one can definitively look to stock market as good predictor of future economic activity. Barro [3] reports that stock market erred in predicting three recessions in US, that didn’t occur in 1963, 1967, and 1978. Binswanger [7] presents evidence that there has been a breakdown in the relationship between stock market performance and future real activity in US economy since early 1980’s. Hu [8] argues that yield spread between long term and short term government bonds is a better predictor of future economic activity than stock market performance in developed countries.

The globalization has accelerated since 2000’s and economies around the world are more inter-connected than any time in past. This pace has been supported by countries around the world who have tore down the trade barriers, and companies which are relentlessly expanding abroad. Some countries are more dependent on global economy to generate economic growth

(6)

6 than others. The Netherlands falls into this category as it has very open economy and well developed stock market. It’s exports account for more than 65% of GDP making it one of the most trade dependent countries in the world. In other words, Dutch economy relies highly on other countries economic wellbeing to derive wealth and investment. Increasing (diminishing) economic activity in its export market should be reflected in Dutch economic activity. On the other hand, equity ownership in Netherlands (Table 1) is not as widespread as in US economy. Under such economic circumstances, it is easy to discount the impact of stock market on country’s economy. In this paper, we try to find the answers to these following questions. Is Dutch domestic stock market cointegrated with economy of country ? Can stock prices really predict the future economic activity of Netherlands (open trade dependent country)? How the change in global economic landscape since 2000 altered the link between stock market and economy ?

Table1: Stock market capitalization as percentage of GDP

Country Name 2005 2006 2007 2008 2009 Belgium 76% 99% 84% 33% 55% Germany 44% 56% 63% 31% 39% Netherlands 93% 115% 122% 45% 68% United Kingdom 133% 155% 137% 70% 128% United States 135% 146% 143% 83% 108%

Source: The World bank

This paper aims explore whether all the stakeholders of economy can look into the stock market as a reliable indicator that can predict economic growth(recession). Unlike previous studies which have mostly focused on US and other large developed economies, this paper will examine if the same relationship is true for small, open and prosperous economy that relies heavily on exports to generate economic growth. Thus, the study aims to bridge the gap in the literature.

3. Literature Review

Economic theory suggest that stock price reflect expectation about the future corporate earnings. Future corporate earnings depends on performance of economy. Economic growth inarguably impacts the all facets of society. Economic growth increases employment , consumption, investment and overall well-being of society whereas recession does exactly opposite by lowering living standards thereby increasing unemployment, decreasing consumption and investment. Sustainable economic growth is very important for overall well-well being of society. Therefore, economies which achieve large increases in output over extended periods of time, not only enable rapid increases in standards of living, but also have serious changes in the economic, political and social landscape of a nation. Wurgler [9]

(7)

7 contends that financial markets improve the allocation of capital, and those with developed financial sectors increase investment more in their growing industries, and decrease investment more in their declining industries, than those with undeveloped financial markets thus promoting healthy growth.

There are several empirical literatures that examine the role of financial markets in promoting economic growth. Bencivenga and Smith [10] and Levine [11] were among the first to propose growth models to identify the channels through which financial markets affect long-run economic growth. The two papers emphasized that financial markets help diversify agents’ liquidity and investment risk, attract more savings into productive investment and prevent the premature withdrawal of physical capital invested in the long-term projects. Consequently, the existence of financial markets means that more capital can be kept in productive investments, which in the end raises the rate of economic growth.

King and Levine [12] suggested another approach to identifying the channel of transmission between finance and growth, and in their model they pointed at innovation as the engine of growth. In this aspect, financial markets evaluate the potential innovative projects, finance the most promising ones and monitor the carrying out of investment. This way financial markets help in the function of efficient resource allocation. Demetriades and K. Hussein [13] state an economy with proper functioning financial markets will experience a higher growth rate of productivity. The link between macroeconomic fundamentals and the equity market is naturally appealing given the importance of macroeconomic variables in determining company cash flows. The dividend discount model and the arbitrage pricing theory, on the other hand provide important theoretical frameworks that show the channel through which the behaviour of macroeconomic variables are factored into stock prices. These models predict that any anticipated or unanticipated arrival of new information about GDP, production, inflation, interest rates, exchange rates, etc, will affect stock market performance.

3.1. Effects of Stock Market Performance on Economic activity

In theory, stock price movements may have direct effects on investment and consumption. Tobin in collaboration with Brainard [14] developed concept of q-channel and argued that the ratio of the stock price to the replacement cost of capital, which has become known as Tobin's q, should be considered as a good indicator of a company's incentive to invest, and that, moreover, this variable is the only relevant determinant for a company's investment. If Tobin's q is greater than one, then capital is more valuable if employed inside the company, and the (increase in the) company's market value is greater than it costs to produce it. As rising stock prices directly result in an increase in Tobin's q, it would be profitable for the company to expand its capital stock, leading to an increase in investment spending, aggregate demand and aggregate output.

The next channel through which stock market performance may influence GDP was suggested by Modigliani [15]. His proposition operates through the impact that the wealth variable has on consumption. A permanent increase in security prices results in an increase in the individual’s wealth holdings, and therefore in higher permanent income. Through the permanent income hypothesis, Modigliani postulated that temporally, consumers smoothen consumption in order to maximize their utility. An increase in permanent income will therefore enable consumers to re-adjust upwards their consumption levels in each period.

(8)

8 Stock price movements may influence investment decisions not only of listed companies, but of all companies holding stocks in the balance sheet. The ability of companies to borrow depends, among other factors, on the value of collateral they can provide. As stock prices increase, the value of collateral of companies holding stocks increases, enhancing their access to external funds for investment. Bernanke et al. [16] contend that in the case of declining stock prices will lead to decrease in investments because of lack of financing. This may be accompanied by general deterioration in economic conditions as profits and cash flows fall that might crimp a company's ability to finance investment spending by internal funds, starting a self-destructing loop known as financial accelerator. Many attribute Bernanke's belief in the financial accelerator may account for some of his policy decisions, such as cutting interest rates in the wake of the credit crisis of 2008-2010, which lead to equity market and confidence returning to market. In addition to the balance sheet effect affecting a company's investment decisions, changes in net wealth caused by movements in stock prices can also have an impact on both the lending behaviour of banks and the consumption of private households. Van den Heuvel [17] finds that with stock prices falling and the resulting impairments being recorded on the asset side of banks' balance sheets, banks' equity might fall to such a low level that the bank might be forced to reduce lending because of regulatory capital requirements and the increased cost of issuing new equity. Similarly Mishkin [18] , argues that declining stock prices will worsen private households’ balance sheet will result in reduced consumption, if falling stock prices reduce the value of collateral and thereby worsen households' ability to raise additional loans. This will result in households to stop purchasing durable consumers goods. Regarding the effects of stock price fluctuations on private consumption, however, most of the empirical studies have focused on the wealth level and consumption. For households holding stocks, a permanent increase in stock prices implies an increase in financial wealth. Assuming that consumers aim at smoothing their consumption over time, the increase in financial wealth results in higher current and future consumption, stimulating aggregate demand and output. Ludwig and Slok [19] argue that channels through which stock price movements directly affect the investment and consumption demand of companies and consumers having access to equity finance or holding stocks, stock prices may also affect investments indirectly via confidence effects. Stock prices are often used as leading indicators of cyclical developments and may therefore also influence consumption and investment decisions albeit not causing them. A decline in stock prices may, for example, be interpreted as a signal for increased downward risks to future economic activity and employment. This may hurt consumer confidence and current aggregate consumption of households - even if these do not own stock. Likewise, a general decline in stock prices may also lead companies (irrespective of being able to issue new equity in the stock market) to lower their profit expectations and curtail investment plans.

In addition to these transmission channels having an impact on the macroeconomic development via domestic consumption and investment, the repercussions of stock market movements on the economies of other countries are also relevant to the economy's exposure owing to companies' international capital and trading links. Besides effects on foreign demand, the state of the foreign equity markets also plays a role for the above-mentioned transmission mechanisms. Retail and institutional investors are increasingly holding foreign equities and domestic companies may even be quoted on foreign stock exchanges. Changes in foreign stock prices are therefore associated with changes in financial wealth of domestic companies and consumers, or with changes in the market value of capital, possibly affecting domestic demand via the consumer-wealth and the balance sheet-channel or (in case of a foreign listing) the q-channel.

(9)

9

3.2. Effects of Economic Activity on Stock Market Performance

It is generally assumed that financial markets react to the changes in economic activity. Pearce and Roley [20] examined the daily response of stock prices to announcements about the money supply, inflation, real economic activity, and the discount rate and found that the expectation of positive or negative announcement affect stock prices. The expectation of good economic news results in stock market going higher whereas, news below expectation leads to decline in stock markets. Other authors including Humpe and Macmillan [21] examined whether industrial production, CPI, long run interest rate and money supply influence to the stock prices in Japan and US found evidence that stock price is negatively related to CPI and long run interest rate but positively related to industrial production.

N.F. Chen et al. [22] investigated the relationship between stock prices and macroeconomic variables assuming that macroeconomic fluctuations are influential on stock prices through their effect on future cash flows and the rate at which they are discounted. An example of such work is Van Nieuwerburgh et al. [23], in which the authors studied the relationship in Belgium and established that growth in GDP growth caused stock market development. However, daily stock market volatility cannot be solely attributed to the economic variables as investors perception of individual company, supply and demand also affect the equity markets considerably. Wasserfallen [24] found the empirical evidence that the effects of macroeconomic news are either very small or obscured by a low signal to noise ratio.

Garcia and Liu [25] observed that a reciprocal relationship between financial system development and economic growth exists whereby economic growth makes the development of financial intermediation system profitable. In other words, economic growth and stock market are complimentary to each other as establishment of an efficient financial system permits faster economic growth. The researchers stressed that the financial and real sectors interact during all stages of development and that there is; at no stage, only a one-way relationship between financial development and economic growth prevails. However, Foresti [26] indicates that stock market prices can be used in order to predict growth, but the opposite is not true. More and more authors therefore, prefer describing the relationship as a two-way causation or uni-directional relationship or the feedback effect, and in these studies, they do not always establish the direction of the causality between these two variables, and those that do seek to identify the direction of the causality often lead to ambiguous conclusions.

3.3. Other Empirical Studies

There is considerable amount of empirical literature that have sought to cement the relationship between stock market performance and GDP. An incomplete list of studies that have focused upon short run relationship between stock returns, macroeconomic and financial variables include: Jain [27], Asprem [28],Gjerde and Saettem [29]. Each has found that stock returns and various macro-economic factors are, to varying degrees, correlated, using either US or international data.

Alyward and Glen [6] conducted similar analysis using average data on 23 markets: the G-7 countries plus Australia and 15 other emerging markets over a sample period from 1951-1993. Estimation results showed that 12 out of 23 countries in the sample were found to have positive coefficients on lagged stock price variable. Mauro [30] conducted similar analysis on mix of 17

(10)

10 developed countries and 8 emerging countries Results showed that lagged stock returns 5 out of 8 emerging market countries and 10 out 17 developed countries were significantly and positively associated with output growth. Hence, given the divergent views and results, the debate in the literature on the link between stock prices and the economy remains inconclusive.

4. Data and Methodology

To formally test the hypotheses and identify the relationship between stock market and GDP growth, we run the cointegration test based on time series data.

4.1. Data

In investigating the empirical relationship between stock prices and other GDP growth for the Netherlands, one objective was to assemble time series on stock market movements over reasonable long period of time. We have extracted stock market data for the time period of 10 years from 2000-2010 to make sure that data is representative of modern economic times. The stock market index used in the analysis is the AEX index, derived from Amsterdam Exchange index, is a stock market index composed of Dutch companies that trade on Euronext Amsterdam, formerly known as the Amsterdam Stock Exchange. Started in 1983, the index is composed of a maximum of 25 of the most actively traded securities on the exchange. The quarterly index data is obtained from Bloomberg (www.bloomberg.com). The index prices were then used to calculate the quarter on quarter returns, which will serve as a time series data for stock market performance.

Gross domestic product (GDP) is the market value of all officially recognized final goods and services produced within a country in a given period of time. GDP is calculated by adding consumption, investment, government expenditure and net exports within the given time period. This means GDP represents the aggregate economy activity in a country that encompasses all the macroeconomic variables that influence the stock market. That’s why, we have used GDP in our empirical research instead of other macroeconomic variables like inflation, money supply, industrial production, etc. GDP quarterly data is obtained from the Central Bureau of Statistics (www.cbs.nl), the Netherlands .

The data range from the first quarter of 2000 to the last quarter of 2009 was used to carry out the empirical analysis. Figure1 shows the GDP growth and stock index returns in stacked line graph( trend graph) follow each other closely. However, further analysis will be needed to conform the view. Please note that in this paper INR refers to stock market performance and GDP denotes GDP growth.

(11)

11 Figure1 : Dutch GDP growth and AEX index returns (2000-2010)

Source: Own calculations, quarterly data

4.2. Methodology

4.2.1. Unit Root Test

An economic series that follows a random walk process is called non-stationary over time . A prerequisite before implementing a cointegration analysis is that the non-stationary variables studied are integrated of the same order. In order to test this characteristic in our time series data, we utilize popular unit root tests method, Augmented Dickey-Fuller Test (ADF) [31]. Our study will test each time series individually to ensure non-stationarity at the levels of the data, and also run the unit root tests on the first differences to ensure I(1). The equation for the ADF is given below

:

k

ΔY

t

=α+β

t

+γY

t-1

+Σρ

i

ΔY

t-1

+ε

t (1)

i=1 -40.00% -30.00% -20.00% -10.00% 0.00% 10.00% 20.00% 30.00%

Stock index vs GDP growth

(12)

12 where Yt represents the selected stock index return and GDP growth for time period , ΔYt = Yt-

Yt-1, is a constant representing a possible drift term, β measures if any time trend is present,

ΣiρiΔYt-1 are lagged values of Yt and εt is a residual that should be a white noise process. The

coefficient of particular interest is γ and the ADF test measures the t-value of γ. The test is one-sided in which the hypotheses are shown by:

H0: γ = 0 vs Ha: γ = 1

If γ = 0, it is possible to conclude that a stock market is non-stationary. The critical values that are applicable are always negative and have been identified by Dickey & Fuller through a Monte Carlo simulation. The size of the critical values depends on the lag length and whether a constant or a trend component is included in the regression. To matters easy,while testing for non-stationarity , the correct value for k (number of lags) is determined by reference to a commonly produced information criteria, Schwarz-Bayesian criteria, suggested by Campbell and Perron [32]. The aim being to maximize the amount of information. We have performed the test assuming that data is trending and stochastic, as it is the most sensible thing to do with financial/macroeconomic time series data.

4.2.2. Cointegration Analysis

Cointegration analysis can be used to examine the co-movement between two (more) non-stationary time series. Firstly, cointegration analysis establishes a long term relationship by calculating long-run equilibrium asset prices. Next, correlations within an error correction model are estimated. Therefore, stochastic trends common to the respective time series are found prior to the cointegration analysis.

If the cointegration analysis indicates that there is a cointegrating vector, we infer that the tested series will not drift apart in the long-term, and will revert to equilibrium levels following any short-term drift that may take place. In the context of this study, the presence of a cointegrating vector means that stock market and GDP co-move, one affecting the other. In contrast, if no cointegrating vector is found, we infer that stock market performance and GDP donot have impact on each other. Our cointegration analysis will include GDP growth of the Netherlands and AEX returns during the decade of 2000.

Cointegration analysis was introduced by Engle and Granger in the early 1980s, with improvements and additions made in subsequent years. Cointegration is a modelling process that incorporates non-stationarity with both long-term relationships and short-term dynamics. To examine time series in financial data using cointegration, the time series in its level form should be non-stationary and integrated of order 1, written as I(1). Integrated of order 1 means the series becomes stationary after differentiating it once. Variables are said to be cointegrated if they are I(1) and have a linear combination which is stationary without the need to differentiate the data. Cointegration is the underlying methodology we use to analyze the relationships between the stock market performance and GDP growth to determine possible predictive power of stock market.

There are two main cointegration methods that have consistently been used throughout past studies which are: 1) Engle-Grangers Two Step Estimation Method; and 2) Johansen’s Maximum Likelihood Method using either the Trace Statistic and/or the Maximum Eigenvalue Statistic. Our study uses the Johansen’s Method due to reasons mainly relating to the shortfalls of Engle-Grangers Two Step Estimation Method. The Two Step Estimation Method is very easy

(13)

13 to run, however it needs a larger sample size to avoid possible estimation errors and can only be run on a maximum of two variables (Brooks [33]). It also doesn’t allow for hypothesis testing on the cointegrating relationships themselves, unlike Johansen’s method (Brooks [33]).

4.2.2.1. Johansen’s Cointegration Test

After completion of unit root testing on our time series, assuming all our time series data are integrated of the same order, we conduct , the bivariate cointegration methodology of Johansen [34] and Johansen and Juselius [35] to examine the long run relationship between two (or more) variables.

The Johansen’s process is a maximum likelihood method that determines the number of cointegrating vectors in a non-stationary time series Vector Autoregression (VAR) with restrictions imposed, known as a vector error correction model (VEC). Johansen’s estimation model is as follows:

p

ΔX

t

= μ + ΣΓ

i

ΔX

t-i

+ αβ’ X

t=i

+ ε

t (2) i=1

Xt= (n x 1) vector of all the non-stationary indices in our study

Γi= (n x n) matrix of coefficients

α= (n x r) matrix of error correction coefficients where r is the number of cointegrating relationships in the variables, so that 0 < r < n. This measures the speed at which the variables adjust to their equilibrium. (Also known as the adjustment parameter)

β= (n x r) matrix of r cointegrating vectors, so that 0 < r < n. This is what represents the long-run cointegrating relationship between the variables.

Johansen [32] defines two different test statistics for cointegration under his method: the Trace Test and the Maximum Eigenvalue Test. The Trace test is a joint test that tests the null hypothesis of no cointegration (H0: r = 0) against the alternative hypothesis of cointegration (H1: r > 0). The Maximum Eigenvalue test conducts tests on each eigenvalue separately. It tests the null hypothesis that the number of cointegrating vectors is equal to r against the alternative of r+1 cointegrating vectors (Brooks [34]).

g

λ

trace

(r) = -T Σ ln(1-λ

i

)

(3)

i=r+1

λ

max

(r, r+1) = -T ln(1-λ

r+1

)

(4) r = number of cointegrating vectors under the null

λ = estimated ith ordered eigenvalue from the αβ’ matrices

The researchers need not to specify which variables are endogenous or exogenous, the VAR model has proved to be one of the most successful, flexible, and easy to use models for the analysis of bivariate and multivariate time series data. One problem everyone need to notice is

(14)

14 how to determine the optimal lag length of VAR model. Generally, two approaches are used more often. One way is the likelihood ratio test, and the other is the information criteria, such Akaike’s (AIC) and schwarz’s Bayesian Information Criteria (SBIC) However, the information criteria method is more powerful than LR. If AIC and SBIC provide the contradictive lag length, SBIC criterion is preferred in this paper. The reason is that SBIC will deliver the correct model with few lags, while on average AIC will choose a model with too many lag orders (Brooks [33]).

4.2.2.2. Vector Error Correction Model

15 4.2.3. Generalised Impulse Response Function

An impulse response function gives the time-path for a variable explained in a VAR model, when one of the variables in the model is shocked. We get a picture of how the variable in question responds to the shock over several periods of time. An impulse response function is essentially a type of conditional forecast.

Estimated model (VECM) was then used to calculate generalized impulse response (GIR) as proposed by Koop et al. [36]. An attractive feature of this approach is that it does not suffer from variable ordering. Generalized impulse response compares the conditional expectation of a variable in the model given an arbitrary current shock vt and history wt, to the conditional

5. Empirical Analysis and Discussion

This section details all results from the testing we conducted. All testing was conducted using EViews 6.0 statistical software. Please note that the full results of empirical tests carried out for this research are presented in appendix section.

5.1. Order of Integration

Since only integrated variables of the same order can be co-integrated, the first step is to determine the integration of relevant variables. Augmented Dickey Fuller test were calculated to test for unit roots in the variables of the system. The results of the calculation are shown in table 2; t-statistic, significance level and order of integration are displayed for both time series data, where GDP represents GDP growth and INR represents stock market performance.

According to the results of Augmented Dickey-Fuller test, summary (Table 2), we conclude that there is absence of unit root according to the P-values of all the two series as the P-values are significant. The values of computed ADF test-statistic of the two series are smaller than the critical values at 5%, and 10% levels of significance, respectively with same lag lengths (based on Schwarz Information Criterion). However in the series GDP at significance level of 1%, we cannot reject the null hypothesis because t-statistic is bigger than critical value. But we consider

(16)

16 the series to be integrated on same order because in this research because we consider 5% to be the significance level for rejection, and p-value is significant. Therefore, we reject the null hypotheses that mean all two series do not have a unit root. We conclude that two series are stationary according to the results of Augmented Dickey-Fuller (Table. 2). Finally, we get the stationarity (there is absence of unit root in ADF test) at same levels of significance with similar lag lengths.

Since the two series are stationary or they are cointegrated, which means that they are integrated of the same order and errors are stationary, in which case the model defines a long run equilibrium relationship among the cointegrated variables. So, we can now proceed with Johansen co-integration test.

Table 2: Augmented- Dickey Fuller (ADF) test

Variables Critical level t-statistic ADF-statistic Results GDP 1% Level -3.610453 -3.009129 I(1) GDP 5% Level -2.938987 -3.009129 I(1)* GDP 10% Level -2.607392 -3.009129 I(1)* INR 1% Level -3.610453 -5.124186 I(1)* INR 5% Level -2.938987 -5.124186 I(1)* INR 10% Level -2.607392 -5.124186 I(1)*

Source: own calculations

Note: I(1)* denotes the series is integrated on first order.

5.2. Johansen’s Cointegration Test

Since ADF test have confirmed that the series are integrated in same order, and there is no likelihood of spurious regression, we can now proceed with Johansen’s cointegration test. The next step is to determine the appropriate lag length. If the lag length is not optimum, the error term may not be Gaussian and the inference of the estimation may be invalid. Because the data sample is small, the author only estimates the unrestricted VAR model with all variables from the lag 0 to lag 4. After estimating unrestricted VAR with different lag length, the corresponding Akaike information criterion (AIC) and Schwarz criterion (SC) are recorded and compared. The optimal length is determined by the minimized value of both information criteria. We got contradictive results of AIC and SC, we prefer SC to select the optimum lag length. Table 3 shows that the optimal lag length for unrestricted VAR model recommended by both AIC and SC, which are contradictory and we prefer SC over AIC. Therefore, lag zero is used in the Johansen cointegration analysis and Vector Error Correction Model (VECM), which is the optimal lag length minus one.

(17)

17 Table 3: AIC and SC for Optimum Lag Length in Unrestricted VAR

Lag Length 0 1 2 3 4

Akaike AIC -7.64831 -8.348807 -8.429847 -8.502583 -8.337485 Schwarz SC -7.560337 -8.084887 -7.989981 -7.88677 -7.545725

After choosing the appropriate lag length. We proceed with the Johansen’s cointegration test. Table 4 and 5 contains the tests for number of cointegrating vector. The model is estimated under the hypothesis that the linear trend is fully contained in the intercepts of the cointegrating regressions. The results include both max-eigen value test and trace test . The first two entries in table specify the null hypothesis and the alternative whereas the last two represent the test-statistic along with critical values at the 95% level, for both trace and max-eigen value test. The critical values of the test are from MacKinnon-Haug-Michelis .

The results indicate the existence of cointegrating vector between stock index returns and GDP growth. The Trace Test in Table 4 indicates the existence of 2 cointegrating equation at the 5% significance level. This cointegrating equation means that one linear combination exists between the variables that force these variables to have a relationship over the entire 10 year time period, despite potential deviation from equilibrium levels in the short-term

.

Thus, the null hypothesis that stock index returns and GDP growth are not cointegrated (r=0 and r≤1) against the alternative of one or more cointegrating vectors (r≥1 and r≥2) is rejected, since and trace statistics exceeds critical values at 5% significance levels.

The Maximum Eigenvalue Test in table 5 also shows the presence of cointegrating equations at the 5% level confirming the results from Trace Test. Thus, the null hypothesis that stock index returns and GDP growth are not cointegrated (r=0 and r=1) against the alternative of one or more cointegrating vectors (r≤1 and r=2) is rejected, since and max-eigen statistics exceeds critical values at 5% significance level.

Therefore, these two tests confirm a cointegrating relationship over the 10 year sample period. The presence of conintegrating vector indicates that the stock return and GDP growth share the long-run equilibrium. This implies that stock market performance and GDP growth are correlated.

Table 4: Results from Johansen’s Cointegration Trace test Null

Hypothesis

Alternative

Hypothesis Eigenvalue Trace Statistic

Critical value (5%)

r = 0 r = 1 0.550544 40.4723 15.49471

r ≤ 1 r = 2 0.211824 9.283333 3.841466

(18)

18 Table 5: Results from Johansen’s Cointegration Max-Eigen value test

Null Hypothesis Alternative Hypothesis Eigenvalue Max-Eigen Statistic Critical value (5%) r = 0 r ≥1 0.550544 31.18897 14.2646 r ≤ 1 r ≥ 2 0.211824 9.283333 3.841466

Analyzing the normalized cointegrating coefficient in the cointegration test allows us to understand how these variables adjust in the 10 year time period

.

The results are displayed in table 6

.

Since we have identified the existence of one cointegrating equation, we can say that a stable equilibrium relationship is present. The results are normalized on the INR. Due to the normalization process, the signs are reversed to enable proper interpretation. The coefficient reveals that 1% increase stock market performance will result in 2% increase in GDP.

Table 6: 10-Year Normalized Cointegrating Coefficients Log likelihood 167.47 INR GDP 1.000000 -3.51871 (-2.01508)

Note: The standard errors are in parenthesis ().

5.3. Error Correction Mechanism

Given the finding that stock market performance and GDP growth are cointegrated in the long run, we utilize the cointegration vector to construct the error correction model (ECM). It should be noted that the cointegrating vector is obtained from the Johansen Maximum-likelihood Estimates (Normalized). Table 7 shows the Vector Error Correction Estimates.

From table 7 below, the speed of adjustment back to equilibrium is represented by the error correction term, also known as the adjustment factor. The coefficients of the error correction term of GDP growth is insignificant, but the coefficient of INR is significant. These results show that if there is a disturbance occurred in the whole system, stock market performance will have significant force tending to bring the model back into equilibrium whenever it moves too far. Put another way, the burden of adjustment back to equilibrium in the system rests on the stock

(19)

19 market performance. It is of note that the GDP growth has a slower speed of adjustment than stock market performance. This indicates that the stock market performance is the first one to display any effects from a shock, and will then disseminate this shock to GDP growth.

Table 7: Vector Error Correction Estimates Cointegrating Eq. CointEq1

INR(-1) 1

GDP(-1) -3.518711 (-2.01508) [-1.74619]

C 0.044085

Error Correction D(INR) D(GDP) CointEq1 -0.727836 0.037724 (-0.16252) (-0.00778) [-4.47832] [ 4.84942] C 0.002333 -0.000513 -0.02125 -0.00102 [ 0.10980] [-0.50418]

Note: The standard errors are in parenthesis () & Standard errors in bracket [ ].

5.4. Generalised Impulse Response

This subsection presents the results for generalised impulse response function. Figure 2 and 3 plots the response of GDP growth and stock market performance to the shock of one positive standard deviation based on Cholseky dof adjusted innovation. The results are displayed for 8 quarters.

Figure 2 illustrates that when one positive standard deviation of shock is given to GDP growth, then stock market performance reacts to it by going up initially till 6 months and then it stabilizes for remaining period. This means that positive shock in GDP impacts stock market performance positively. However, when one positive standard deviation of shock is given to stock market performance (Figure 3), the GDP growth rises sharply for next 6 months before slowing the ascend and stabilizing at 10 months The results reveal that spill over effect from shock on stock market performance last longer on GDP growth than the other way around.

(20)

20 Figure 2: Response of stock market performance following a shock in GDP

Figure 3: Response of GDP following a shock in stock market performance

(21)

21

5.5. Discussion of Findings

This subsection will detail the findings of our empirical analysis. The major take away from this empirical analysis for all stakeholders of Netherland’s economy is as follows: positive stock market performance will always lead to positive GDP growth and vice versa. The results in this research confirms the view from earlier empirical research performed by Boubakari and Jin [37], who found positive links between stock market and GDP for which stock market is liquid and highly active for 5 Euronext countries including the Netherlands. In another similar research, for Euro area, Andersson and D’Agostino [38] found out that introduction of the euro in 1999 seems to have resulted in a significant improvement in predictive power of future economic growth across most asset classes including stock prices in particular.

The results from Johansen cointegration tests suggest that there is a strong cointegrating relationship between stock market performance and GDP growth. The fact that stock market performance and GDP growth share long run relationship may be due to Equity Valuation Models, stock price valuation depends on expected future dividends. Hence, such relationship may possibly result from the fact that expected future dividends are a good proxy of future economic activity as measured by GDP. The normalised cointegrating coefficients show that 1% increase in stock market index would result in roughly 2% increase in GDP growth. The magnitude of result is much larger and sounds counter intuitive, and much less likely to happen in real world under normal economic circumstances. The divergence in result is due to the fact that despite the positive GDP growth ( although slower), the stock index has almost fallen by half in last decade. The AEX index boomed during 1990’s anticipating boom in growth as in 1990’s, but as GDP growth failed to match the expectation, the index fell much more than GDP during recessions. Error correction mechanism also reveals another interesting facet of this relationship by revealing that stock market performance reacts quickly to shocks and brings the system back to equilibrium in long run despite short run divergence. This process has been highlighted by Jian-Chiu Han [39], in which he concludes that equity valuation cannot outgrow the economy on long run.

The impulse response function shows that the spill over effect from the shock on stock market performance lasts much longer in GDP growth than the other way around. The one standard deviation of shock on stock market performance leads to sharp spike in GDP growth. On the other hand, one standard deviation of positive shock on GDP growth leads to only small spike in stock market performance. This discrepancy which has been discussed above and is due the result of sharp fall in stock index despite slower GDP growth. However, the take away from generalised impulse response test is that positive stock market performance will lead to positive GDP growth and effect will last for longer period, than the other way around.

Above results from empirical analysis unequivocally point to fact that stock market performance and GDP growth share long run positive relationship. Thus, the investors and policymakers alike can look to stock market performance as the signal of future economic activity while making important economic decisions.

(22)

22

5.6. Limitations

Following the study, a number of factors emerge that must be considered when making conclusions from the results that were reached. Subsequent research undertaken in the area of cointegration should take into account the limitations found in this study. As always, there are limitations in terms of the reliability and validity of the results. We can separate these limitations into two distinct groups: data limitations and model limitations.

Data limitations of the study come in the form of frequency, sample period and stock index (doesn’t cover the whole economy). To allow a more complete picture of the cointegration patterns that exist within stock market performance and GDP growth, it could be advantageous to include higher frequency data – perhaps monthly data, rather than the quarterly data used in this study. The 10 year sample period we selected covered approximately one business cycle, however if long term cointegration patterns were to be analyzed, a longer time period would be needed to ensure validity of the results.

Limitations of the model are mainly related to the use of the Johansen procedure of cointegration analysis. The Johansen method has a high probability of generating outliers, as well as high variance. It is also very sensitive to the lag length selected for the Vector Error Correction Model (Brooks [34]). Another criticism of the Johansen procedure is the difficulty associated with interpreting results. That is, other types of cointegration tests allow for greater clarity when comparing results across studies. In addition, the impulse response function results are affected if important variables are omitted from the system, their effects go to the residuals and hence may lead to major distortions in the impulse responses and the structural interpretations of the results.

6. Conclusion

In this study, we explored the relations between stock market performance and GDP during the decade of 2000, which has been marked by lots of volatility in both GDP growth and stock market performance. We have studied whether the stock market performance will lead to GDP growth in the Netherlands. We tested time series data (stock market performance and GDP) for the presence of unit root or for stationarity. The Augmented Dickey- Fuller test indicated that series are non-stationary and integrated of the same order. Then, we proceeded with the cointegration test and employed highly popular Johansen’s cointegration test. Along with cointegration, we performed Vector Error Correction, which measures the speed of adjustment of variables concerned. After testing for error correction mechanism, we performed the impulse response function to see how stock market performance and GDP growth would react to shock on each other.

The empirical results from the Johansen cointegration test did find evidence of the cointegration between stock market performance and GDP .The result was straight-forward as trace test and max-eigen value test indicated the presence of cointegrating vector. Overall, the findings of the Johansen test showed that stock market performance does impact the GDP growth. The normalised cointegrating coefficient showed that 1% increase in GDP growth leads to roughly 2% increase in GDP growth. However, it is advised not to overemphasise this finding because

(23)

23 data sample period of 2000-2009, included recession in early 2000’s and global economic crisis in late 2000’s. During these period stock market became much more responsive to monetary and economic stimulus than economic growth. VECM results pointed to fact that when there is disturbance in system, stock market performance acts as fulcrum bringing the system back to equilibrium in long run despite short run divergence. Last but not the least, the impulse response function showed that positive stock market performance will lead to more positive GDP growth and spill over effect lasts longer in case of GDP growth when stock market performance is shocked than the other around.

The findings in this empirical analysis could be improved further by analysing data from longer time period, which covers at least 2 business cycles. The empirical analysis could be expanded further by adding macro variables like inflation, industrial production, investment, etc in the research along with GDP. These steps would sharpen the results of findings and correct some anomalies and discrepancies we have encountered during this research.

Our results nonetheless advocate the positive link between stock market performance and GDP growth. These findings have important implications for all stakeholders of economy who are looking for any signs that could tell them how the future economic period could be predicted. This means that stock market returns are indeed a good predictor of economic growth for the Netherlands. Although the stock market performance is also impacted by many unobserved factors that move the market, we advocate that the results from this paper should be viewed the most significant signal on how economy will perform in the future. Therefore, positive stock market performance should be viewed as good sign for economic growth in coming period and vice versa.

(24)

24

7. References

[1] Fama E. (1990). Stock returns, expected returns, and real activity . Journal of Finance, 45: 1089-1108

[2] Schwert W. (1990). Stock returns and real activity: a century of evidence. Journal of Finance, 45: 1237-1257

[3]Barro R.J. (1990). The stock market and investment. The review of financial studies, 3(1): pp.115-131

[4] Estrella, Arturo, and Frederic S. Mishkin, (1998). Predicting US recessions: financial variables as leading indicators. Review of Economics and Statistics, 80.1 : 45-61.

[5] Henry, Olan T., Nilss Olekalns, and Jonathan Thong (2004). Do stock market returns predict changes to output? Evidence from a nonlinear panel data model. Empirical Economics, 29.3: 527-540

[6] Aylward A, Glen J (2000). Some international evidence on stock price as leading indicators of economic activity. Applied Financial Econometrics, 10: pp.1-14

[7]Binswanger M, (2000). Stock returns and real activity: Is there still a connection? Applied

Financial Econometrics,10, pp.379-387

[8] Hu Z. (1993) . The yield curve and real activity. Staff papers (International Monetary Fund:

Washington), 40-4, pp. 781-806

[9]Wurgler J. (2000). Financial markets and the allocation of capital. Journal of Financial

Economics, 58, 1-2, pp. 187-214

[10] V.R Bencivenga and B.D. Smith (1991). Financial Intermediation and Endogenous Growth,

Review of Economic studies, 52(2), 195-209

[11] R. Levine (1997). Financial Development and Economic Growth: Views and Agenda,

Journal of Economic Literature, 31(5)

[12] Robert King and Ross Levine (1993). Finance and Growth: Schumpeter might be right, The

Quarterly Journal of Economics, 108(3), 717-737

[13] P. Demetriades and K. Hussein (1997). Does Financial Development Cause Economic Growth: Time-Series Evidence from 16 Countries, Journal of Development Economics, 51 [14] James Tobin (1961). Money, Capital, and other Stores of value. American Economic

Review, 51, pp.26-37

[15] Modigliani, F.(1971).Consumer Spending and Monetary Policy: the Linkages, Federal

Reserve Bank of Boston Conference Series: Paper No. 5

[16] Ben S. Bernanke, Mark Getler, and Simon Gilchrist,(1996). The Financial accelerator and flight to the quality, Review of Economic and Statistics, 78, pp.1-15

(25)

25 [17] Van den Heuvel, Skander J (2000).Does Bank Capital matter for monetary Transmission ?

FRBNY Economic policy Review

[18] Frederic S Mishkin (1978). The Household Balance Sheet and the Great Depression,

Journal of Economic History, 38, 4:918-937

[19] Alexander Ludwig and Torsten Slok (2002). The Impact of changes in Stock prices and House prices on Consumption in OECD Countries, IMF Working Papers

[20] Pearce, Douglas K. and V. Vance Roley (1985). Stock Prices an Economic News. Journal

of Business, Vol. 58, No. 1,pp. 49-67

[21] Humpe , Andreas, & Macmillan, Peter (2009). Can Macroeconomic variables explain long-term stock market movements? A comparison of the USA and Japan. Applied Financial

Economics, 19, 111–119

[22] N.F. Chen, R. Richard and S.A. Ross (1986). Economic forces and the stock market.

Journal of Business, 59, 383-403

[23] S. Van Nieuwerburgh, F. Buelens, and L. Cuyvers (2006). Stock market development and economic growth in Belgium, Explorations in Economic History, 43, 13-38

[24] Walter Wasserfallen (1989). Macroeconomics news and the stock market :Evidence from Europe, Journal of Banking & Finance, 13, pp.613-626

[25] Valerino F. Garcia and L. Liu (1999). Macroeconomic Determinants of Stock Market,

Journal of Applied Economics, 2, 29-59

[26] Foresti, P. (2006). Testing for Granger causality between stock prices and economic growth‟, MPRA Paper No. 2962

[27] Jain, P.C.(1988). Response of hourly stock prices and trading volume to economic news.

Journal of Business, 61, 219-31

[28] Asprem, M (1989). Stock prices, asset portfolios and macroeconomic variables in ten European countries, Journal of Banking and Finance, 13, 589-612

[29] Gjerde, R and Saettem, F (1999). Casual relations among stock returns and macroeconomic variables in a small open economy, Journal of International Financial Markets, Institutions and Money, 9, 61-74

[30] Mauro P (2000). Stock returns and output growth in emerging and advanced economies.

IMF Working Paper WP/00/89

[31] Dickey, D.A and Fuller, W.A (1979). Distributions of the Estimators for Autoregressive Time Series with a Unit Root. Journal of the American Statistical Association, 74, 427 – 431

(26)

26 [32] Campbell, J. and Perron, P. (1991). Pitfalls and opportunities: what macro-economists should know about unit roots,NBER Macroeconomic Annual 1991(Eds) O. Blanchard and S Fisher, MIT press, Cambridge MA, pp. 141-201

[33] Brooks, C. (2008). Introductory Econometrics for Finance: 2nd Edition, Cambridge

University Press, Cambridge

[34] Johansen, S (1988). Statistical analysis of cointegration vectors, Journal of Economic

Dynamics and Control, 12, 231 – 254

[35] Johansen, S., Juselius, K. (1990). Maximum likelihood estimation and inference on cointegration – with applications to the demand for money. Oxford Bulletin of Economics and

Statistics 52, 169-210

[36] Koop, G., Pesaran, M. H. and Potter, S. M. (1996). Impulse response analysis in nonlinear multivariate models, Journal of Econometrics, 74, 119–47.

[37] Ake Boubakari and Dehuan Jin (2010). The Role of Stock Market Development in Economic Growth: Evidence from Some Euronext Countries, International Journal of Financial

Research, Vol. 1, No. 1;pp.1-7

[38] Andersson, Magnus, and Antonello D'Agostino (2008). Are sectoral stock prices useful for predicting euro area GDP?. ECB Working Paper, 876

[39] Han, Jian-Chiu (2000). Equity Valuation Cannot Outgrow the Economy Over the Long Run,

(27)

27

8. Annex

(28)

(29)

29

8.2. Variable Lag Order Test

(30)

30

8.3. Johansen’s Cointegration Test

(31)

(32)

32

“Stock market performance and GDP growth ": evidence from the Netherlands”.

University of Amsterdam

Amsterdam Business School

Master in International Finance

Master Thesis

“Stock market performance and GDP growth. Evidence from the Netherlands”

Abstract

.

Contents

1. Introduction and Purpose

.

2. Motivation of Research

3. Literature Review

3.1. Effects of Stock Market Performance on Economic activity

3.2. Effects of Economic Activity on Stock Market Performance

3.3. Other Empirical Studies

4. Data and Methodology

4.1. Data

4.2. Methodology

:

ΔY

=α+β

+γY

+Σρ

ΔY

+ε

Stock index vs GDP growth

ΔX

= μ + ΣΓ

ΔX

+ αβ’ X

+ ε

λ

(r) = -T Σ ln(1-λ

)

λ

(r, r+1) = -T ln(1-λ

)

ε

= y

– βx

Δy

= αε

+ γΔx

+ u

ΔGDP

= α

+α

ΔGDP

+ α

ΔINR

+ α

ECT

+u

ΔINR

= β

+β

ΔINR

+ β

ΔGDP

+ β

ECT

+u

GIR (x

, v

, w

) = E [X

|v

, w

] – E [X

|w

]

5. Empirical Analysis and Discussion

5.1. Order of Integration

5.2. Johansen’s Cointegration Test

.

.

.

5.3. Error Correction Mechanism

5.4. Generalised Impulse Response