An Empirical Analysis of Stock Market Performance and Economic Growth: Evidence from India

Engle-Granger Cointegration Test Results

Table 4a: Engle-Granger Cointegration test for BSE and IIP (on monthly series)

^Regression based Engle-Granger Cointegration Method( bse t 			 0   1 iip t  u t )
Variables	Coefficients	t-test (probability values)
Constant	-2.09	-4.46 (0.00)*
IIP	2.05	22.85 (0.00)*

R-squared value 0.77
Table 4a: Engle-Granger Cointegration test for BSE and IIP (on monthly series) - continued

^Residual based Regression Method ( z t   z t  1   t ^)
Variable Residual	ADF-test (probability value) -2.38 (0.02)*

Where ‘*’ implies that the rejection of null hypothesis at 5 % or lower level of significance
Table 4b: Engle-Granger Cointegration test for NSE and IIP (on monthly series)

Regression based Engle-Granger Cointegration Method (nse t			  0  1iip t  ut )
Variables	Coefficients	t-test (probability values)
Constant	-3.31	-8.00 (0.00)*
IIP	2.05	26.00 (0.00)*

R-squared value 0.81

Residual based Regression Method ( ^^zt^^^zt^^1^^^t)
Variable Residual	ADF-test (probability value) -2.57 (0.01)*

Where ‘*’ implies that the rejection of null hypothesis at 5 % or lower level of significance
Table 5a: Engle-Granger Cointegration test for BSE and GDP (on quarterly series)

Regression based Engle-Granger Cointegration Method (bse t			  0  1 gdp t  ut )
Variables	Coefficients	t-test (probability values)
Constant	-15.35	-7.18 (0.00)*
GDP	1.82	11.21 (0.00)*

R-squared value 0.72

Residual based Regression Method ( ^^zt^^^zt^^1^^^t)
Variable Residual	ADF-test (probability value) -2.93 (0.04)*

Where ‘*’ implies that the rejection of null hypothesis at 5 % or lower level of significance
Table 5b: Engle-Granger Cointegration test for NSE and GDP (on quarterly series)

Regression based Engle-Granger Cointegration Method (nset   0  1 gdp t  ut )
Variables	Coefficients	t-test (probability values)
Constant	-16.70	-8.65 (0.00)*
GDP	1.83	12.50 (0.00)*

R-squared value 0.76

Residual based Regression Method ( ^^zt^^^zt^^1^^^t)
Variable Residual	ADF-test (probability value) -3.01 (0.04)*

Where ‘*’ implies that the rejection of null hypothesis at 5 % or lower level of significance
The table 4a and 4b presents Engle-Granger cointegration test results on monthly series. Where, study estimated regression based cointegration method by using non stationary data i.e., I (1) of BSE (or NSE) and IIP and then residual values are taken out and applied ADF test to see whether the series is stationary or not at levels. The residuals result of ADF test reject the null hypothesis of non- stationarity at 5 % level of significance and provides evidence that there is a cointegration relationship between BSE and IIP in the long run. Similarly, study also estimated regression model of NSE and IIP, and found that there is also long run relationship. The table 5a and 5b unveils quarterly data results of Engel-Granger cointegration test. The study has estimated residual based cointegration method of BSE
(or NSE) and GDP. The residual results of ADF test reject the null hypothesis of non stationarity at 5

% significance level. This suggests that there is a cointegration relationship between the considered variables in the study.

4. Error Correction Model Results

Table 6a: Error Correction Model for IIP and BSE (on monthly series)

Regression based Error Correction Model (  iip

 

 iip

 1





bse

 

3

z t

 1

  ^) t

Variables

Coefficients

t-test (probability values)

OIIPt-1

-0.54

-7.91 (0.00)*

OBSE t

0.08

1.85 (0.07)

^Zt-1

-0.04

-2.94 (0.00)*

Regression based Error Correction Model (  bse

t

 

 bse

t  1

 

2

 iip

t

 

3

z

t  1

  ^) t

OBSEt-1

0.04

0.44 (0.66)

OIIPt

0.11

0.84 (0.40)

^Zt-1

0.04

1.61 (0.12)

Where, Z_t-1 is the residual values and ‘*’ implies that the rejection of null hypothesis at 1 % level of significance
Table 6b: Error Correction Model for IIP and NSE (on monthly series)

^Regression based Error Correction Model ( iip t   1  iip t 1   2  nse t   3 z t 1   t ^)
Variables	Coefficients	t-test (probability values)
OIIPt-1	-0.54	-8.07 (0.00)*
ONSEt	0.12	2.52 (0.01)*
^Zt-1	-0.05	-3.58 (0.00)*
^Regression based Error Correction Model ( nse t   1  nse t 1   2  iip t   3 z t 1   t ^)
ONSEt-1	0.26	3.25 (0.00)*
OIIPt	0.16	1.53 (0.12)
^Zt-1	0.02	1.03 (0.30)

Where, Z_t-1 is the residual values and ‘*’ implies that the rejection of null hypothesis at 5 % or lower level of significance
Table 7a: Error Correction Model for GDP and BSE (on quarterly series)

^Regression based^Error Correction^Model ( gdp		t   1  gdp t 1   2  bse t   3 z t 1   t ^)
Variables	Coefficients		t-test (probability values)
OGDPt-1	-0.16		-1.09 (0.28)
OBSEt	0.11		0.95 (0.34)
^Zt-1	-0.11		-2.44 (0.02)*
Regression based Error Correction Model (  bse		t   1  bse t  1   2  gdp		t ^^3 ^zt  1 ^^t ^)
OBSE t-1	0.29		2.01 (0.05)
OGDPt	0.11		0.62 (0.54)
^Zt-1	0.05		0.78 (0.44)

Where, Z_t-1 is the residual values and ‘*’ implies that the rejection of null hypothesis at 5 % level of significance
Table 7b: Error Correction Model for GDP and NSE (on quarterly series)

Regression based Error Correction Model (  gdp t   1  gdp t  1   2  nse t   3 z t  1   t )
Variables	Coefficients	t-test (probability values)
OGDP t-1	-0.19	-1.31 (0.19)
ONSEt	0.13	1.19(0.24)
^Zt-1	-0.15	-2.90 (0.01)*
Regression based Error Correction Model (  nse t   1  nse t 1   2  gdp			t ^^3 ^zt  1 ^^t ^)
ONSE t-1	0.24	1.62 (0.11)
OGDPt	0.14	0.74 (0.46)
^Zt-1	0.07	0.96 (0.34)

Where, Z_t-1 is the residual values and ‘*’ implies that the rejection of null hypothesis at 5 % level of significance
The monthly series results are presented in the table 6a and 6b. It confirms the Engle-Granger cointegration test that there is a long-run relationship between the studied variables in both monthly and quarterly series. We estimated error correction models by including the error correction term in the equations to see how the disequilibrium is corrected in the short-run. The results of monthly series reveal that when the IIP data is regressed on the BSE/NSE then the error correction term (-0.04/-0.05) is statistically significant at 1 % level and when BSE/NSE is regressed on IIP then the error correction term is statistically insignificant. This implies that once the IIP and BSE (or NSE) deviates away from the long-run equilibrium, then the IIP makes all adjustment to reestablish the equilibrium by correcting disequilibrium about 4 % (or 5 %) every month. Table 7a and 7b contains quarterly data results. The coefficients of the error correction terms (-0.11 and -0.15) in the GDP equations are statistically significant at 5 % level. The error correction terms in the BSE and NSE equations are statistically insignificant. This implies that once the deviation takes from the long-run equilibrium between stock prices (BSE and NSE) and GDP, then the GDP initiates all adjustments to reestablish the equilibrium condition by correcting disequilibrium about 11 % and 15 % on each quarter in respective equations.

5.