CAUSAL RELATIONSHIP BETWEEN STOCK MARKET AND REAL ECONOMY IN INDIA A PROJECT REPORT SUBMITTED IN PARTIAL FULFILLMENT OF THE DEGREE OF BACHELORS OF MANAGEMENT STUDIES SUBMITTED BY: SAMYAK CHAUDHARY HAHEED SUKHDEV COLLEGE OF BUSINESS STUDIES APRIL 2016
36
Embed
Causal Relationship between Stock market and Real Economy in India using Granger Causality test
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
CAUSAL RELATIONSHIP BETWEEN STOCK MARKET AND
REAL ECONOMY IN INDIA
A PROJECT REPORT SUBMITTED IN
PARTIAL FULFILLMENT OF THE
DEGREE OF BACHELORS OF MANAGEMENT STUDIES
SUBMITTED BY:
SAMYAK CHAUDHARY
HAHEED SUKHDEV COLLEGE OF BUSINESS STUDIES
APRIL 2016
2
CERTIFICATE
This is to certify that the project report entitled “Causal Relationship between Stock Market and
Real Economy of India” is the project work carried out by Samyak Chaudhary at Shaheed
Sukhdev College of Business Studies for partial fulfillment of BMS.
This report has not been submitted to any other organization for the award & any other Degree
/Diploma.
Samyak Chaudhary
3
TABLE OF CONTENTS
Chapter no. Title of the Chapters Page no.
ACKNOWLEDGEMENT 6
ABSTRACT 7
1 INTRODUCTION 8
1.1 Scope of the research 9
2 THEORETICAL PERSPECTIVE 10
3 LITERATURE REVIEW 13
4 RESEARCH METHODOLOGY 15
4.1 Data Collected for the tests 15
5 DATA ANALYSIS 19
5.1 QUARTERLY SERIES ANALYSIS 20
5.2 ANNUAL DATA ANALYSIS 23
6 FINDINGS AND CONCLUSIONS 25
7 LIMITATIONS OF THE PROJECT 27
8 FURTHER SCOPE 27
9 APPENDIX 30
10 REFERENCES 34
4
LIST OF TABLES AND FIGURES
Numbers Tables and figures Page no.
Table: 1 Second difference stationarity result GDP quarterly 20
Figure: 1 Level correlogram result GDP quarterly 20
Figure: 2 Second difference correlogram result GDP quarterly 21
Table: 2 Lag length Criteria result GDP and Sensex quarterly (lag2) 21
Table: 3 Lag length Criteria result GDP and Sensex quarterly (lag3)
21
Table: 4 Serial correlation LM test GDP and Sensex quarterly
22
Table: 5 Granger Causality result GDP and Sensex quarterly 22
Table: 6 Granger Causality result GDP and Nifty50 quarterly 22
Table: 7 Granger Causality result GDP and Sensex Yearly 23
Table: 8 Granger Causality result GDP and Nifty50 yearly
24
Table: 9 ADF test results 25
Table:10 Lag terms Results 25
Table 11: Granger Causality Test Results 25
Table 12: Market Capitalization % of GDP 28
Table 13: Market Capitalization and GDP (current billion US$) 29
5
ACKNOWLEDGEMENT
I wish to express my sincere thanks to Dr. Poonam Verma, Principal of SSCBS, for
providing me with all the necessary facilities for the research.
I place on record, my sincere thanks to Mr. Neeraj Kumar Sehrawat, Assistant professor,
SSCBS, for the continues encouragement and support as a faculty guide.
I take this opportunity to express gratitude to all of the Department faculty members for
their help and support. I also thank my parents for the unceasing encouragement, support and
attention.
I also place on record, my sense of gratitude to my colleague, who directly or indirectly, has lent
their hand in this venture.
Samyak Chaudhary
6
ABSTRACT
This study tries to find out the direction of causal relationship between stock market and the real
economy of India by applying Granger Causality test on Nifty 50 and BSE Sensex data from the
period 1996-2014 and 1990-2014 respectively with real GDP values both, quarterly and
annually. By applying ADF Unit root test to check stationarity, Vector Autoregression (VAR) to
find out optimum lag structure and finally Granger causality test, the report finds out that there is
no statistically significant Granger Causal relationship between stock market and the real
economy of India for quarterly periods and a unidirectional Granger causality from real GDP to
market indices for annual figures. They may have some other causal relation which is not
explained by the Granger test assumptions or they might still be independent of each other in
India.
Being a very interesting issue to determine the cause and effect relation between stock market
and economy, this study explains the theories related to such a relationship and also emphasize
on the fact that why considering real economy for such a study is beneficial.
The results of this report show that stock market and real economy of India Granger Cause each
other but in very different scenarios and thus motivates to assess all the different scenarios. The
yearly direction for causality is from real GDP to market indices. Hence, it says that either of the
variables can be explained by their own historical values or by some exogenous variable, and
also by one another but only for yearly figures.
7
1. INTRODUCTION
The stock market and economy are very common terms that we hear every day and are one of the
strongest entities based on which the entire future of a country depends. They affect lives of
millions of people and make to the news headlines every day. Also, it is interesting for any
researcher or academician to ponder over the causal relationship between these two entities. The
prime aim of this research is to evaluate the causal relationship between stock market and real
economy of India and most importantly find out the direction of causality.
This topic may appear generic yet is widely debated as one cannot claim to know the true
relation between stock market and economy. There exist many theories that offer arguments and
counter-arguments to this issue. Also, many studies have been conducted related to this, so this
issue is very interesting to study upon. One can be curious to know whether it is a good economy
that causes a better performing stock market or is it vice-versa. Hence, this can be considered as
the motivation to conduct this study and effort is made to discuss the theoretical aspects of this
issue along with statistically studying the relationship using the popular method of Granger
causality test to know the direction of causality.
This topic may also help to re-evaluate the causal relationship between these two entities and can
motivate academicians to work upon a model in which stock market performance can be
considered an indicator of good economic conditions or vice-versa if such direction of causality
prevails.
Also, the use of Real economy is an important addition to this report. Based on the previous
researches either macroeconomic variables such as inflation, industrial production, CPI etc. or
nominal GDP was used. By using real GDP at factor cost, the report tries to capture the actual
output value of goods and services after adjusting for inflation.
Real economy may be the part of the economy which actually takes into account produced goods
and services. So it excludes the effects of inflation or deflation. So if prices are assumed to not
go up, then this gives a realistic measure of assessing the growth or the output. Real GDP is
assumed to be the variable that can most represent real economy.
8
As per Bart Hobijn and Charles Steindel (CURRENT ISSUES IN ECONOMICS AND
FINANCE , Volume 15, Number 7, 2009) GDP, especially real GDP, is considered the central
measure of overall economic activity primarily because its long and short run movements are
correlated with many factors. An important example is that the real GDP growth is closely
associated over the long run with the improvement in living standards. Subsequently, GDP
growth measured in current dollar prices and the long-run growth of the tax base are closely
correlated and thus affect tax revenues.
1.1 Scope of the research
The prime objective is to identify direction of Granger causal relationship between stock market
and real economy. This is important to note that the report only theoretically explains the
relationship between the entities based on previous literature, and only finds the presence and the
direction of the causality to know what causes what. This is done on the basis of considering that
there indeed exists a relationship between the two entities, believing the theories to be true and
only discusses them for basic understanding. So the scope is limited to the extent of determining
the direction of causality and not empirically providing a relationship for the Indian entities.
Scope of the study is limited to Indian context and data is therefore is completely based on
availability. Nifty5o was not available before 1990 and GDP values were not for period before
1988 (1960 on World Bank). So it is highly dependent on the short duration data available.
Also, as proxy for real economy is taken as Real GDP and other macro economic factors that
affect stock prices are not considered for analysis. This is done just to simplify the study and to
show the causal link between two major variables of interest using the Granger test. It is believed
that real GDP is the most appropriate indicator of economy as sometimes GDP growth is
synonymous with economic growth. By doing this, the result will be easier to interpret and
conclusion will be straightforward.
9
2. THEORETICAL PERSPECTIVE
Financial markets are crucial for the foundation of a stable financial system of a country. Also,
financial system is an integral part of the country’s economy. Many domestic as well as
international factors directly or indirectly affect the performance of the stock market.
Volatility in stock prices raises the interest of the general public and policy-makers into stock
market environment and increase the focus on its impact on the economy. When prices rise
sharply, it raises fears of a stock market bubble or it may happen just due to inflationary
pressures. A collapse in share prices has the ability to cause immense economic damage, a
famous stance being of the stock market crash of 1929 which acted as a key factor in causing the
Great Depression in 1930s.
However, we cannot take instances from history to simply say that this always happen, as daily
movements in the stock market can also have insignificant impact on economy. A fall in share
prices does not necessarily cause an economic downturn.
The fall in share prices may affect the average consumers as the people with money invested in
stocks experience a decline in the value of their invest when stock prices plummet. So their
wealth decreases. Karl E. Case, John M. Quigley, and Robert J. Shiller (2005) it is the causal
effect of exogenous changes in wealth upon consumption behavior.
Ricardo M. Sousa (WEALTH EFFECTS ON CONSUMPTION EVIDENCE FROM THE
EURO AREA, 2009) even concludes that financial wealth effects are relatively large and
statistically significant than Housing wealth effects. But the effect should be given too much
importance as it strongly incorporates behavioral component that may vary significantly. And
investor spending pattern may be independent of share prices.
Equity markets and its news are given great amount of importance and sometimes economic
conditions are inferred from it. But to what extent theoretical models apply to practical cause and
effect, which may still remain in dark.
One consequence of 1990s liberalization has been a significant increase in the inflows of foreign
capital into India. But for countries implementing financial liberalization for adjusting the
10
economic scenario like India during 1990s, the first step is the loosening of controls on the
inflow of foreign capital, accompanied by a shift to a more liberalized exchange rate regime5.
This may make it more susceptible to bad times of foreign markets.
Nevertheless, given that the importance of the equity markets within the Indian financial system
is expected to increase significantly in the longer term, the impact of stock price movements on
the macroeconomic development may also increase.
However, all in all we can say that share prices may appear to be a part of the economic
movement, because if condition of economy can affect the prices of goods, services or other
entities, then shares to maybe get covered under the same umbrella. Or maybe stock prices under
certain conditions pull the economy to one direction.
The more important issue is to find out why it is essential to find a decent and logical relation
between stock market and economy. It is because if economy causes the stock market movement,
then macroeconomic indicators become really helpful for an investor. On the other hand, if
opposite happens, then we can consider share prices as an economic indicator and can predict an
economic slowdown or recession.
Theories for the relationship:
Stock price is the discounted present value of a firm’s payout in an economy, as suggested by
economic theory under Dividend Discount model (Gordon). As mentioned by Gevit Duca
(2007), if the firm’s payout that is profit portion which is earned due to real economic activity,
then we can say that this payout is a function of economic activity and such relationship should
prevail. Equity earnings and cash flows are naturally correlated with economic activity and the
business cycle, as per Paulo Maio and Dennis Philip (2013)1. They further suggest that equity
discount rates needed to determine equity risk premium are related to systematic common risk
factors for which macroeconomic variables are often selected. Thus, current stock prices should
be related to future economic activity through the cash-flow channel.
There are other theoretical propositions about how stock prices may have a direct affect on
economic output, further strengthening the link in the relationship between these two variables.
11
The first theory is explained by Tobin’s Q which was suggested by Tobin (1969). Tobin’s Q is
the ratio of share prices and the replacement cost of capital. When share prices are high, the
value of the firm relative to the replacement cost of its stock of capital (Tobin’s Q) is also high.
So relatively, the cost of capital appears to be low as the numerator increases. As cost of making
investment appears to be low, investment expenditure increases and thus technically aggregate
economic output should increase. This occurs because investment would be easier through debt
(which is cheaper) and no new shares are offered in a situation of a high share price.
The second of the relationship was suggested by Modigliani (1971) from the consumer’s point of
view. The basic premise of his theory is the impact of wealth on consumption. A permanent
increase in security prices results in an increase in the individual’s wealth (wealth effect).
Through the permanent income increase hypothesis, Modigliani postulated that consumers
smooth consumption in order to maximize their utility. This increase in permanent income will
allow them to consume more and to re-adjust their consumption levels upwards.
The third possibility is when stock prices impact output which is referred to as the financial
accelerator (Bernanke and Gertler, 1989; Kiyotaki and Moore, 1997). It focuses on the impact
that stock prices have on firms’ balance sheets. The ability of firms to borrow depends
substantially on the collateral they can pledge. The collateral value firms can offer increases in
scenarios where their stock price value increases as firms can pledge their stocks itself. Thus,
higher credit can be raised which can be used for investment purposes and subsequently triggers
a growth in economic activity.
So the basic premise of the first and the third theories is that if by any means cost of credit
decreases, investment increases and further assumes that a good investment will definitely
increase in aggregate output for the whole economy.
12
3. LITERATURE REVIEW
Gevit Duca (2007) conducted a Granger causal test of stock market prices and GDP in
developed market economies which revealed that these two tend to move together over time and
have statistically significant causal relationship in some countries. The question is also raised as
to what is the reason for such a relationship. It explains many popular theories about this
relationship which my report also incorporates and uses as a base. This paper employs the
Granger causality test in order to examine causality direction. The focus of the paper is on long-
term trends with evidence presented from top five stock markets in the world in terms of market
capitalization. He found that a unidirectional causality prevails from stock index to GDP in US,
UK, Japan and France, but no causality was found in Germany.
Samveg Patel (2012) in his study tries to investigate the effect of macroeconomic determinants
on the performance of the Indian Stock Market using monthly data over the period January 1991
to December 2011. The study is done for eight macroeconomic variables, namely, Interest Rate,
Inflation, Exchange Rate, Index of Industrial Production, Money Supply, Gold Price, Silver Price
& Oil Price, and two stock market indices namely Sensex and S&P CNX Nifty. It applies
Augmented Dickey Fuller Unit root test, Johansen Cointegration test, Granger Causality test and
Vector Error Correction Model (VECM) and found that the study found that Interest Rate is I
(0); Sensex, Nifty, Exchange Rate, Index of Industrial Production, Gold Price, Silver Price and
Oil Price are I (1); and Inflation and Money Supply is I (2). It also found the long run
relationship between macroeconomic variables and stock market indices. The study also revealed
the causality from exchange rate to stock market indices to IIP and Oil Price.
A study done by ABOUDOU Maman Tachiwou (2009) incorporates a different approach from
the above papers. This examines the causal relationship between stock market development and
economic growth. The geographical area selected is the West African Monetary Union economy
over the last decade. The techniques applied are unit–root tests and the long–run Granger
noncausality test proposed by Toda and Yamamoto (1995). The results say that there is strong
causal flow from the stock market development to economic growth. A unidirectional causal
relationship is also observed between real market capitalization ratio and economic growth.
13
Another paper by Adwin Surja Atmadja (2005) seeks to examine the Granger-causality among
stock prices indices and macroeconomic variables in five ASEAN countries, Indonesia;
Malaysia; the Philippines; Singapore; and Thailand with particular attention to the 1997 Asian
financial crisis and period onwards. Monthly data is used to find that there were few Granger
causalities found between the country’s stock price index and macroeconomic variables and
concludes that the stock markets do not seem to have played a significant role in most of these
countries’ economies. Atmadja also says that macroeconomic variables are unlikely to be
appropriate indicators to predict the future behavior of other macroeconomic variables.
The paper by Pramod Kumar Naik, Puja Padhi investigates the same objectives for BSE
Sensex and five macroeconomic variables. Johansen’s co-integration and vector error correction
model are used to explore the long-run equilibrium. The analysis reveals that macroeconomic
variables and the stock market index are co-integrated and, hence, a long-run equilibrium
relationship exists between them. They say that the stock prices positively relate to the money
supply and industrial production but negatively to inflation.
14
4. RESEARCH METHODOLOGY
The objective of this dissertation is to find out whether real Indian economy and the Indian Stock
Market indices namely, Nifty50 and BSE Sensex Granger cause each other or not.
The two variables considered in this study are real GDP and stock market indices that can be
considered as the dummies, with the relationship between them being tested by the Granger
Causality test.
Real GDP is the GDP value at constant prices, taking prices of base years as reference. This can
act as a proxy for real economy.
Calculation of Real GDP is done on the basis of this formula
Real GDP at FC= Nominal GDP at FC*Real GDP index (base=2010 prices)/100
Data Collected for the tests
This study is mainly conducted on Indian data using the Indian GDP values and Indian stock
market indices. This is to apply this research in the Indian market scenario.
It uses both quarterly and annual data for real GDP and Market indices.
The quarterly period is from year 1996 Q4 to 2015 Q3.
The annual period is from the year 1990 to 2014.
The CNX BSE Sensex data has been obtained from BSE official website, whereas, Nifty50 data
is taken from NSE Website (from1990 to 1995) and Investing.com (from 1996 to 2015)
Nominal GDP at Factor Cost in Indian Rupees and the Real GDP index with base as 2010 prices
taken as 100 are the necessary data for real GDP collected from IMF database (International
Financial Statistics)
All GDP values are in Billion Rs.
15
Process
The data used is tested for stationarity. For this, augmented Dickey Fuller Test is applied.
Non-stationary series is converted to stationary at required differences.
Also, serial autocorrelation is also check alongside this and to remove it if it is there,
intercept and trend is ignored for the series and another subsequent difference is taken.
After that, the number of lagged terms included in all Granger tests conducted is going to
be determined on the basis of the VAR lag length criteria. Schwarz Information Criterion
is given preference.
Autocorrelation LM tests are applied to further select the best lag that removes serial
correlation also.
Furthermore, Granger Tests are applied first to see whether real GDP Granger causes
each of the Stock Market Indices and vice-versa is also checked.
Also, GDP is tested with each Index, one at a time. There are two indices taken for this research
(BSE Sensex and Nifty50)
1. Null Hypothesis is Stock index do not Granger cause GDP
2. Alternate Hypothesis is Stock index Granger cause GDP
Nifty50 and BSE sensex are the two most important and popular stock market indices of India,
which I believe, truly represent the Stock market of India.
16
Tests Used
The three step process to conduct this research is as follows:
ADF Unit root test to test stationarity: A Unit root test is conducted to test the stationarity of a
time series using an autoregressive model. Augmented Dickey-Fuller (ADF) has a null
hypothesis that rho = 1 or the series has a unit root.
They are first observed at level, and then p-value is checked. If p-value is significant, then we
select the series for further analysis otherwise, we check stationarity at first difference and so on.
We also begin with a simple assumption that the series might have both constant and trend.
However, p-value remains the criteria for judging irrespective of other values.
VAR to select optimum lag structure: Vector autoregressive (VAR) models are widely used in
forecasting. It computes various criteria to select the lag order of an unrestricted VAR. It
prompts to specify the maximum lag to “test” for and we compare the modified LR statistics to
the 5% critical values starting from the maximum lag, and decreasing the lag one at a time until
we first get a rejection. The alternative lag order from the first rejected test is marked with an
asterisk (if no test rejects, the minimum lag will be marked with an asterisk).The determination
of the lag length can be used in Granger Causality test.
Autocorrelation LM test: Test for Autocorrelation in the residuals in the VAR window. It reports
the multivariate LM test statistics for residual serial correlation up to the specified order. This is
done because no serial correlation is a basic assumption of Granger and this helps if the lag
selection criterion is showing a tie for 2 or more lag values at the time of VAR Lag structure
selection.
H0: no serial correlation is present
H1: serial correlation is present
Granger Causality Test: As regression analysis deals with the dependence of one variable on the
other and correlation tells the degree of such relationship, but they do not explain cause and
effect. The Granger (1969) approach to the question of whether x causes y is to see how much of
17
the current y values can be explained by past values of y and then to see whether adding lagged
values of x can improve the explanation.
Y is said to be Granger-caused by x if previous lagged values of x explain the current values of y
significantly better than Y’s own previous lagged values.
EViews runs bivariate regressions of the form:
The null hypothesis is that x does not Granger-cause y in the first regression and
that y does not Granger-cause x in the second regression.
It is important to note that the statement “x Granger causes y” does not imply that y is the effect
or the result of x. Granger causality measures precedence and information held but does not by
itself indicate causality in the more common use of the term.
18
5. DATA ANALYSIS
First, we collect the data in MS Excel. From here, we export to Eviews for analysis. Level of
significance for all the tests is a standard 5% and P-value selection criterion,
If p-value < α, then enough proof to reject H0, accept Alternate hypothesis
If p-value > α, then not enough roof to reject H0, accept Alternate hypothesis
0
10,000
20,000
30,000
40,000
50,000
1996 1998 2000 2002 2004 2006 2008 2010 2012 2014
GDPquarter
Nifty50
BSEsensex
0
4,000
8,000
12,000
16,000
20,000
24,000
28,000
90 92 94 96 98 00 02 04 06 08 10 12 14
GDPannual
Nifty50annual
BSEsensex
19
5.1 QUARTERLY SERIES ANALYSIS
1. ADF Unit root Test:
For GDP quarterly: Stationarity at second difference, with no intercept and trend
Null Hypothesis: D(GDPQUARTER,2) has a unit root
Exogenous: None
Lag Length: 4 (Automatic - based on SIC, mailbag=11)
t-Statistic Prob.*
Augmented Dickey-Fuller test statistic -8.173229 0.0000
Test critical values: 1% level -2.598907
5% level -1.945596
10% level -1.613719
Table: 1 second difference stationarity result GDP quarterly
There was a presence of partial correlation, as shown by Correlogram. So, GDP quarterly series
was taken at second difference without any intercept and trend. Even, though GDP at first
difference with intercept and trend was stationary, still to remove serial correlation, another
difference had to be taken and intercept and trend had to be excluded.
Figure: 1 level correlogram result GDP quarterly
New series generated as d2gdpquarter=d(gdpquarter,2)
For new series, correlogram is below which shows autocorrelation to be removed.
20
Figure: 2 second difference correlogram result GDP quarterly
For Sensex quarterly:
Series is stationary at first difference, with intercept and trend. But series at first difference is
having serial correlation. So another difference is taken at no intercept and trend. So finally
below result is for second difference.
At second difference, serial correlation is removed results of which are shown in the next figure.
For Nifty50 quarterly:
Series made stationary at second difference, with no intercept and trend to reduce serial
correlation.
For GDP and Sensex: lag length criteria optimum at both 2nd
and 3rd
lag
Lag LogL LR FPE AIC SC HQ
0 -1275.643 NA 8.88e+12 35.49008 35.55332 35.51526