Professional Documents
Culture Documents
12 Chapter4
12 Chapter4
Table 4.1.1 shows us the descriptive statistic summary of the selected variables. The
Table 4.1.1 highlights that the average i.e. Mean of the Output (LR_GDP) is highest, 10.411
followed by Real Export (LR_X) reported as 7.862, Remittances inflow (LR_REM) as 6.335,
Net FDI as a share of GDP (FDI_GDP) as 0.77, Money Supply as a share of GDP (M3_GDP)
as 0.527, Gross Fixed Capital formation as a share of GDP (GFCF_GDP) as 0.258 and the
average of Total Govt. Expenditure as a share of GDP (TEXP_GDP) is lowest, 0.154 among
the variables. The table also reveals that the highest Median value among the variable is
Output (LR_GDP) with a Median value 10.422 and the lowest Median value variable is Total
Govt. Expenditure as a share of GDP (TEXP_GDP) with a value of 0.153.
Table also declares the Coefficient of Variations (C.V.) of the variables, among the
variables Net FDI as a share of GDP (FDI_GDP) has the highest C.V. (117.532%) followed
by Money Supply as a share of GDP (M3_GDP) (33.776%), Gross Fixed Capital formation
as a share of GDP (GFCF_GDP) (20.155%), Remittances inflow (LR_REM) (19.842%),
Real Export (LR_X) (14.334%),Total Govt. Expenditure as a share of GDP (TEXP_GDP)
(9.740%) and the lowest has acquired by the Output (LR_GDP) (6.416%).
47
According to Table 4.1.1 all the variables are positively skewed except Remittances
inflow (LR_REM), which is negatively skewed. Furthermore information we can gather from
the table 4.1.1 that all the variables are platykutric i.e. the value coefficient is less than 3 on
the other hand only one variable, Net FDI as a share of GDP (FDI_GDP) is leptokurtic
whose value of coefficient is more than 3.
The graphical representation of the time series data of the variables is presented in Figure
4.1.1. As can be seen from Figure 4.1.1, there is a trend in each of the series of the selected
variables and therefore, the means and variances of the time series data of the variables are
changing over time, so the series in their original form may not be stationary.
The first step in this regard is to check whether the concerned variables are stationary or
not or in other words the order of integration. The results of the Augmented Dickey Fuller
(1981) tests considering at first the assumption of only intercept and further considering the
assumption of both intercept and trend are reported in the Table 4.1.2, 4.1.3 and the results of
PP Peron (1988) test considering at first the assumption of only intercept and further
considering the assumption of both intercept and trend are revealed in Table 4.1.4, 4.1.5.The
essence of this test is to avoid spurious regression problems normally associated with time
series econometric modeling (Granger and Newbold, 1974).
48
Figure: 4.1.1Graphical representation of time series data of the variables
11.2 8 .8
.7
10.8 7
.6
10.4 6
.5
10.0 5
.4
9.6 4 .3
9.2 3 .2
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
3
.18 9
2
.16 8
1
.14 7
0
.12 -1 6
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
GFCF_GDP
.36
.32
.28
.24
.20
.16
1975 1980 1985 1990 1995 2000 2005 2010 2015
Source: Computed by the researcher
49
Table 4.1.2: Augmented Dickey Fuller (ADF) Unit Root Tests (Intercept)
Table 4.1.3: Augmented Dickey Fuller (ADF) Unit Root Tests (Intercept & Trend)
50
Table 4.1.5: PP Unit Root Tests (Intercept & Trend)
As we have found all the variables are stationary at 1st difference now as mentioned
earlier the long run relationship of Output (LR_GDP), Remittances inflow (LR_REM) and
other control variables have enquired by Johansen (1988, 1995) cointegration technique.To
determine the existence and the number of cointegrating vectors, Johansen cointegration
technique is used. There are two hypothesis tests used for cointegration testing, called the
Trace (λ trace) test and Maximum Eigen value (λ max) test. The cointegration test results at
5% level of significance are presented in Table 4.1.6 and 4.1.7.
Hypothesized No. of CE(s) Eiganvalue Trace Statistic 0.05 Critical Value Prob
51
Table 4.1.7: Johansen Cointegration Test Results
Unrestricted Cointegration Rank Test (Maximum Eiganvalue)
From the results above so far, we found that the variables are stationary at 1st difference
i.e. I(1) and are cointegrated. Therefore they have a long-term relationship. So their short-
run fluctuation can be described by their first-differences, which are stationary. The
interactions in the short-run fluctuations may therefore be described by a VAR system in first
differences. We determine the optimal lag length for the VAR system by using the Schwarz
(1978) Criterion (SC) and the Akaike (1974) Information Criterion (AIC). We used a VAR
system of “i” lags and estimate it for various lag lengths and found that the optimal lag
lengths for Output (LR_GDP) and Remittances Inflow (LR_REM) and other control
variables, to be 2.So from the foregoing discussion and the review of the literature, the effect
of Remittances inflow (LR_REM) on the Output (LR_GDP) can be represented as follows:-
ΔLR_GDPt = α + Ʃni=0 β1i ΔLR_GDPt-i + Ʃni=0 β2i ΔLR_REMt-i + Ʃni=0 β3i ΔTEXP_GDPt-i
+ Ʃni=0 β4i ΔM3_GDP t-i + Ʃni=0 β5i ΔGFCF_GDPt-i + Ʃni=0 β6i ΔFDI_GDPt-i + Ʃni=0 β7i
ΔLR_Xt-i + λ ecmt-1 + Ɛt
According to the test it reveals that the variables have a long run relationship at 1% level
of significance. The results of cointegration test are interpreted through Vector Error
Correction Model (VECM) which is a restricted version of VAR model designed for non
stationary time series. VECM captures the linear relationship among multiple time series by
52
adding error correction features. The error correction term (ECT) tells how much the error is
being corrected or adjusted from short run disequilibrium to long run equilibrium. The result
of VECM is stated in Table 4.1.8.
In the above study VECM was used to examine short runs dynamics and correct for short
run disequilibrium among the variables LR_GDP, LR_REM, TEXP_GDP, GFCF_GDP,
LR_X, M3_GDP, and FDI_GDP. The co-efficient of error correction term (ECT) is equal to -
0.2952 and is statistically significant at 5%. The sign of (ECT) conforms to the restriction of
negativity and less than one (Gujarati, 2009). It also implies that the rate at which short runs
disequilibrium is being corrected to arrive to the long run equilibrium at a rate of 29.52% per
annum.From the above results we can conclude that the Output (LR_GDP)is determined by
lag values of Remittances inflows (LR_REM), Total Govt. Expenditure (TEXP_GDP), FDI
(FDI_GDP) and Money supply (M3_GDP).We have provided empirical evidence showing
that remittances inflow has a negative impact in both the 1st lag (-0.1270 at 1% level of
significance) and 2nd lag (-0.0815 at 5% level of significance) on the Output of India.
53
On the other hand empirically it is observed that Total Govt. Expenditure (TEXP_GDP)
has negative impact at 5% level of significance which can be justified that initially Govt.
Expenditure may have a negative impact or no impact as a Govt. Expenditure requires sum
lag period to give a positive thrust to the economy. However the positive impact of FDI
(FDI_GDP) in the 1st lag and in the 2nd lag at 5% and 1% level of significance respectively,
FDI through its spillover effect have direct positive economic growth of the host countries,
which support the view of positive correlation among foreign capital inflow and economic
growth.
Whereas it is empirically also evident the theoretical opinion that the increase in money
supply can spur the economic growth through demand side. According to standard
macroeconomic theory, an increase in the supply of money should lower the interest rates in
the economy, leading to more consumption and lending/borrowing. In the short run, this
should increase the Output (GDP) of the economy. Here Money supply positively influences
Output in the 2nd lag at 5% level of significance.
The empirical result VECM also reveals that there is no autocorrelation detected in the
sample with D-W value of 2.01. The Durbin Watson (DW) statistic is a test
for autocorrelation in the residuals from a statistical regression analysis. The Durbin-Watson
statistic will always have a value between 0 and 4. If the value comes out to be 2.0 means
that there is no autocorrelation detected in the samples. Values from 0 to less than 2 indicate
positive autocorrelation and values from 2 to 4 indicate negative autocorrelation.
Now if the estimated dynamics is maintained at the near future, then how the Output
(LR_GDP)profile get constituted through these endogenous random innovations of control
variables are ascertained through a variance decomposition analysis. This analysis has a
scope for identifying the relative strengths of innovations affecting the Output (LR_GDP)
profile in the out of-sample forecast horizon. The result of Variance Decomposition is shown
in Table 4.1.9.
54
Table 4.1.9: Variance Decomposition of LR_GDP
According to Table 4.1.9 it is found that the Remittances inflow innovation would
account 20.98% variance of Output at the 2 forecast period horizon that is the highest
innovation among 10 forecast horizon and 10.23% of such variance at the 10 forecast period
horizon. The variance decomposition analysis, therefore, confirms that in the near future the
Remittances inflow may be an important cause for negative Output growth in India.
55
4.2. Empirical Results of Remittances and Inflation
According to our 2nd objective, to assess the possibility of the long run relationship among
the selected macro economic variables, Inflation (L_CPI), Interest rate (L_I), Real Exchange
Rate (LR_EX), Money Growth (DLR_M3), Output (LR_GDP), Remittances inflow
(LR_REM) time series data from the period 1975 to 2016 of those variables are being used as
discussed previously.
Table 4.2.1 shows us the descriptive statistic summary of the selected variables. The
Table 4.2.1 highlights that the average i.e. Mean of the Output (LR_GDP) is highest, 10.411
followed by Remittances inflow (LR_REM) reported as 6.335, Real Exchange Rate
(LR_EX) as 4.276, Inflation (L_CPI) as 3.556, Interest rate (L_I)as 2.121, and the average of
Money Growth (DLR_M3) is lowest, 0.083 among the variables. The table also reveals that
the highest Median value among the variable is Output (LR_GDP) with a Median value
10.422 and the lowest Median value variable is Money Growth (DLR_M3) with a value of
0.078.
Table also declares the Coefficient of Variations (C.V.) of the variables, among the
variables Money Growth (DLR_M3) has got the highest C.V. (51.807%) followed by
Inflation (L_CPI) (25.899%), Remittances inflow (LR_REM) (19.842%), Interest Rate (L_I)
(16.030%), Output (LR_GDP) (6.416) and the lowest has acquired by the Real Exchange
Rate (LR_EX) (6.407%)
According to Table 4.2.1 among the variables three are viz. Interest rate (L_I), Output
(LR_GDP) and Money Growth (DLR_M3) positively skewed on the other hand Remittances
inflow (LR_REM), Real Exchange Rate (LR_EX) and Inflation (L_CPI) which are
negatively skewed. Furthermore information we can gather from the table 4.2.1 that all the
variables are platykutric i.e. the value coefficient is less than 3 on the other hand only two
variables, Interest rate (L_I) and Money Growth (DLR_M3) are leptokurtic whose value of
coefficient is more than 3.
The graphical representation of the time series data of the variables is presented in Figure
4.2.1. As can be seen from Figure 4.2.1, there is a trend in each of the series except Interest
56
rate (L_I) and Money Growth (DLR_M3) of the selected variables and therefore, the means
and variances of the time series data of the variables are changing over time, so the series in
their original form may not be stationary.
The first step in this regard is to check whether the concerned variables are stationary or
not or in other words the order of integration. The results of the Augmented Dickey Fuller
(1981) tests considering at first the assumption of only intercept and further considering the
assumption of both intercept and trend are reported in the Table 4.2.2, 4.2.3 and the results of
PP Peron (1988) test considering at first the assumption of only intercept and further
considering the assumption of both intercept and trend are revealed in Table 4.2.4, 4.2.5.The
essence of this test is to avoid spurious regression problems normally associated with time
series econometric modeling (Granger and Newbold, 1974).
57
Figure 4.2.1: Graphical representation of time series data of the variables
DLR_M3 L_CPI
.30 5.2
4.8
.25
4.4
.20
4.0
.15 3.6
3.2
.10
2.8
.05
2.4
.00 2.0
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
L_I LR_EX
3.0 4.8
4.6
2.5
4.4
2.0 4.2
4.0
1.5
3.8
1.0 3.6
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
LR_GDP LR_REM
11.6 9
11.2 8
10.8 7
10.4 6
10.0 5
9.6 4
9.2 3
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
Source: Computed by the researcher
Table 4.2.2: Augmented Dickey Fuller (ADF) Unit Root Tests (Intercept)
58
Table 4.2.3: Augmented Dickey Fuller (ADF) Unit Root Tests (Intercept& Trend)
Analyzing the result of Augmented Dickey Fuller (1981; AD, henceforth) and the
Phillips and Perron (1988; PP, henceforth) test of stationarity, the test reveals that the
selected variables are stationary at different order of integration i.e. I (0) and I (1) as Money
growth and Interest rate are stationary at level whereas the other variables viz. Output,
Remittances inflow, Inflation and Real Exchange rate are stationary at 1st difference. In this
situation to assess the presence of long-run relation among the selected variables and their
59
dynamics methodology suggests to apply the ARDL bound test approach of cointegration
followed by the ECM-ARDL model, which is introduced by Pesaran, Shine, and Smith
(2001).The estimable form of ECM-ARDL model is stated below:
Δ CPIt = α + Ʃni=0 βi ΔIt-i + Ʃni=0 γi ΔEXt-i + Ʃni=0 θi ΔMt-i + Ʃni=0 πi ΔYt-i + Ʃni=0 δi ΔREMt-i
+ μ1 It + μ2 EXt + μ3 Mt+ μ4 Yt + μ5 REMt + λ ecmt-1 + Ɛt
Where the parameter λ indicates error correction term or speed of adjustment to restore
equilibrium, and ‘n’ is the optimum lag-length(s) chosen for the estimation. The parameters
βi, γi, θi, πi and δi indicate short-run multiplier, while parameters μ1, μ2, μ3, μ4 andμ5 stand for
long-run multiplier. In this model the null hypothesis of no cointegration implies μ1 = μ2 =
μ3 = μ4 = μ5 = 0 and alternative hypothesis of cointegrating relation implies μ1≠ μ2 ≠ μ3 ≠
μ4 ≠ μ5 ≠0.
Having confirmed that the order of integration of the selected variables is different but
none of the series has an order of integration 2 or higher; ARDL bound testing approach to
cointegration is used to study the long-run relation among these variables. As per the
objective of our study, the result of ARDL Bounds Test is reported in Table 4.2.6.
The value of F statistic is 4.494 i.e. greater than the value of 1% level of significance i.e.
4.1.5 in case of 1 bound. So ARDL Bound Test is providing strong evidence that the model
has a long run cointegration among the variables.
Having confirmed the cointegrating relations the paper seeks to estimate the dynamics
that inflation and remittances inflow maintain with the selected policy variables. In this
regards ECM-ARDL models are estimated and the results are reported in Table 4.2.7. The
60
error correction terms obtains from these estimations are statistically significant, which re-
confirm the cointegrating relationship. The ECM derives from the cointegrating model of
inflation reveals that inflation is affected by interest rate, exchange rate, and output and
remittances inflow in short run but there is no evidence of long term relationship between
inflation and remittances inflow.
The study examines the long run and short run relationship among inflation rate,
remittances inflow, exchange rate, GDP, growth of money supply and interest rate. The
empirical findings from the error-correction model show that the inflation adjusts to its
61
equilibrium rapidly at a rate of 59%. In short run, inflation is determined by its past values,
and lag values of Money supply growth (DLR_M3), Exchange rate (LR_EX), Interest rate
(L_I), Output (LR_GDP) and Remittances inflow (LR_REM). According to the result
remittances inflow can instantaneously create deflationary pressure in the economy with 1%
level of significance with a coefficient value -0.129. Empirical findings also provide us
evidences that inflow of remittances in India has a positive significant impact to fuel up
inflation in order of 1st and 2nd lag with 5% level of significance with coefficient value 0.688
and 0.595respectively and 3rd lag with 10% level of significance. This result is supported by
the theoretical evidences discussed in section 3.
As we know output is anti inflationary the empirical result also reflects that. All the lag
values of output are anti inflationary with 5% level of significance. On the other hand
theoretically justified empirical result is observed under the ECM-ARDL model that Money
supply growth has positive impact on inflation i.e. under the result Money supply growth
instantaneously fuel up inflation at 1% level of significance and accordingly 1st lag value at
1% , 2nd lag value at 10% and 3rd lag value at 5% level of significance
Furthermore, the stability of the ARDL cointegrating model has been examined by
CUSUM& CUSUMQ tests. The time plots of the cumulative sum of recursive residuals and
also the cumulative sum of squares of recursive residuals of the model, is presented in
Figures 4.2.2 & 4.2.3, lie within the 95 percent acceptance regions. These figures indicate
that the ARDL cointegrating models are stable.
62
Figure 4.2.2: CUSUM Test
20
10
-10
-20
1978 1988 1998 2008 2016
The straight lines represent critical bounds at 5% significance level
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
1978 1988 1998 2008 2016
The straight lines represent critical bounds at 5% significance level
63
According to the ECM-ARDL result exchange rate depreciation leads to inflationary
pressure in the economy. The depreciation in exchange rate means the appreciation of
domestic currency which means increase in purchasing power of the domestic consumer,
which supposed to have a positive influence on inflation through creating additional demand
in the domestic market (Vargas-Silva 2009); therefore, appreciation of the value of domestic
currency may have an inflationary impact.
64
4.3. Empirical results of Remittances and Real Effective
Exchange Rate
For the last objective, to assess the possibility of the long run relationship among the selected
macroeconomic variables: Real Effective Exchange Rate (L_REER), Output (LR_GDP),
Money Supply as a share of GDP (M3_GDP), Import as a share of GDP (M_GDP), Export as
a share of GDP (X_GDP), Remittances inflow (LR_REM) time series data from the period
1975 to 2016 of those variables are being used as discussed previously.
Table 4.3.1 shows us the descriptive statistic summary of the selected variables. The
Table 4.3.1 highlights that the average i.e. Mean of the Output (LR_GDP) is highest, 10.411
followed by Remittances inflow (LR_REM) as 6.335, Real Effective Exchange Rate
(L_REER)as 4.648, Money Supply as a share of GDP (M3_GDP) as 0.527, Import as a share
of GDP (M_GDP)as 0.12 and the average of Export as a share of GDP (X_GDP)is lowest,
0.087 among the variables. It is also observed that the highest Median value is acquired by
the Output (LR_GDP) as 10.422 among the variables on the other hand Lowest Median value
variable is Export as a share of GDP (X_GDP) as 0.078.
The table also declares the Coefficient of Variations (C.V.) of the variables, among the
variables Import as a share of GDP (M_GDP)has the highest C.V. (54.166%) followed by
Export as a share of GDP (X_GDP) (45.597%), Money Supply as a share of GDP (M3_GDP)
(33.776%), Remittances inflow (LR_REM) (19.842%), Output (LR_GDP) (6.416%) and the
lowest has acquired by the Real Effective Exchange Rate (L_REER) (6.131%).
According to that Table all the variables are positively skewed except the Remittances
inflow (LR_REM) is negatively skewed also the table declares that all the variables are
platykurtic as the value of the coefficient are less than 3.No variables are claimed to be
leptokurtic as because no variable have the coefficient value more than 3.
The graphical representation of the time series data of the variables is presented in Figure
4.3.1. As can be seen from Figure 4.3.1, there is a trend in each of the series of the selected
variables and therefore, it can be said that the means and variances of the time series data of
the variables are changing over time, so the series in their original form may not be stationary
65
The first step in this regard is to check whether the concerned variables are stationary or
not. The results of the augmented Dickey-Fuller (1981) tests considering at first the
assumption of only intercept and further considering the assumption of both intercept and
trend and results of the order of integration of the variables are reported in the Table 4.3.2,
4.3.3 and the results of PP Peron (1988) test considering at first the assumption of only
intercept and further considering the assumption of both intercept and trend are revealed in
Table 4.3.4 and 4.3.5.The essence of this test is to avoid spurious regression problems
normally associated with time series econometric modeling (Granger and Newbold, 1974).
66
Figure 4.3.1: Graphical representation of time series data of the variables
L_REER LR_GDP
5.4 11.6
5.2 11.2
5.0 10.8
4.8 10.4
4.6 10.0
4.4 9.6
4.2 9.2
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
LR_REM M3_GDP
9 .9
8 .8
.7
7
.6
6
.5
5
.4
4 .3
3 .2
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
M_GDP X_GDP
.30 .20
.25 .16
.20 .12
.15 .08
.10 .04
.05 .00
1975 1980 1985 1990 1995 2000 2005 2010 2015 1975 1980 1985 1990 1995 2000 2005 2010 2015
Source: Computed by the researcher
67
Table 4.3.2: Augmented Dickey Fuller (ADF) Unit Root Tests (Intercept)
Table 4.3.3: Augmented Dickey Fuller (ADF) Unit Root Tests (Intercept & Trend)
68
Table 4.3.5: PP Paron Unit Root Tests (Intercept & Trend)
From the results of the above tables of stationarity we have found that all the variables
are stationary at 1st difference now as mentioned earlier the long run relationship of GDP,
Remittances and other control variables have enquired by Johansen (1988, 1995)
cointegration technique.To determine the existence and the number of cointegrating vectors,
Johansen cointegration technique is used. There are two hypothesis tests used for
cointegration testing, called the Trace (λ trace) test and Maximum Eigen value (λ max) test.
The cointegration test results at5% level of significance are presented in Table 4.3.6 and
4.3.7.
Hypothesized No. of CE(s) Eiganvalue Trace Statistic 0.05 Critical Value Prob
69
Table 4.3.7: Johansen Cointegration Test Results
From the results above so far, we found that the variables are stationary at 1st difference
i.e. I(1) and are cointegrated. Therefore they have a long-term relationship. So their short-
run fluctuation can be described by their first-differences, which are stationary. The
interactions in the short-run fluctuations may therefore be described by a VAR system in first
differences. We determine the optimal lag length for the VAR system by using the Schwarz
(1978) Criterion (SC) and the Akaike (1974) Information Criterion (AIC). We used a VAR
system of “i” lags and estimate it for various lag lengths and found that the optimal lag
lengths for Real Effective Exchange Rate (L_REER) and Remittances Inflow (LR_REM)
and other control variables, to be 2.So on the basis of foregoing discussion and the review of
the literature, the effect of Remittances inflow on the GDP can be represented as follows:-
ΔL_REERt = α + Ʃni=0 β1i ΔL_REERt-i + Ʃni=0 β2i ΔLR_GDPt-i + Ʃni=0 β3i ΔLR_REMt-i +
Ʃni=0 β4i ΔM3_GDP t-i + Ʃni=0 β5i ΔM_GDPt-i + Ʃni=0 β6i ΔX_GDPt-I + λ ecmt-1 + Ɛt
According to the test it reveals that the variables have a long run relationship. The results
of cointegration test are interpreted through Vector Error Correction Model (VECM) which
is a restricted version of VAR model designed for non stationary time series. VECM captures
the linear relationship among multiple time series by adding error correction features. The
error correction term (ECT) tells how much the error is being corrected or adjusted from
70
short run disequilibrium to long run equilibrium. The result of VECM is stated in Table
4.3.8.
The study analyses the long run and relationship among REER, remittances inflow,
Import, Export, GDP and Money supply. The empirical findings from the VECM show that
the REER adjusts to its equilibrium by the rate of 23.98% approx 24% per annum. In short
run, REER is determined by its past values, and lag values of Money supply (M3_GDP),
Import (M_GDP), Output (LR_GDP) and Remittances inflow (LR_REM). We empirically
analyzed that remittances inflow has a negative impact both in 1st lag (-0.1558 at 1% level of
significance) and also in 2nd lag (-0.1040 at 1% level of significance) on the REER of India in
long run that will increase the international competitiveness.
However it is observed from the above VECM it reveals that Output has a positive
impact on REER in 1st lag with 1% level of significance i.e. with the increase in Output it is
evident that REER will appreciate. The result also declares that increase in import also pulls
down to REER to depreciate in 1st lag at 1% level of significance. It supports the
71
macroeconomic view that more import will lead to increase in value of foreign currency in
terms of domestic currency that will lead to REER to depreciate.
The empirical result VECM also reveals that there is no autocorrelation detected in the
sample with D-W value of 2.38 i.e. close to 2. The Durbin Watson (DW) statistic is a test
for autocorrelation in the residuals from a statistical regression analysis. The Durbin-Watson
statistic will always have a value between 0 and 4. If the value comes out to be 2.0 means
that there is no autocorrelation detected in the samples. Values from 0 to less than 2 indicate
positive autocorrelation and values from 2 to 4 indicate negative autocorrelation.
As we have already discussed thatif the estimated dynamics is maintained at the near
future, then how the REER profile get constituted through these endogenous random
innovations of control variables are ascertained through a variance decomposition analysis.
This analysis highlights the relative strengths of innovations affecting the REER profile in
the out of 10 sample forecast horizon. The result of Variance Decomposition is shown in
Table 4.3.9.
According to the table 4.3.9 it is found that the Remittances inflow innovation would
account 57.60% variance of Real Effective Exchange Rate at the 4 forecast period horizon
and 33.39% of such variance at the10 forecast period horizon. The variance decomposition
72
analysis, therefore, confirms that in the near future the Remittances inflow may be an
important cause for depreciation of Real Effective Rate in India which in turn can increase
the international competitiveness of the country in future.
73