volatility in inflation affects output growth negatively through its effect on returns on
investment and unexpected policy. It is important to mention that initially the estimation
includes variable of interest rate (RBR) also, but the coefficient of the variable was found
to insignificant and the Wald test of variable deletion favored dropping of the interest rate
variable and hence, dropped.
A same exercise is repeated with GDP per capita growth rate as the dependent variable for
consistency. Even then, the result is quite similar (see Table 1). Thus, it can be said that in
long run, inflation and inflation volatility is not good for output growth in India. But this
linear relationship result does not provide any clue about the level of inflation, which is
harmful to the output growth.
(Table 1 about here)
Now, following Khan and Senhadji (2001), extra inflation (INFt -Π*) is introduced in the
growth equation, so as to check for possible presence of threshold level of inflation. Result
with variable of extra inflation (presented in Table 2a and 2b) shows that there exist a
threshold level at 6 percent of inflation.2 Positive coefficient of inflation, once, extra
inflation is introduced in the growth equation indicates that below the threshold level of
inflation, inflation does not hurt output growth.
(Table 2a about here)
However, as Singh and Kalirajan (2003) points out that at some value of Π*, R-square of
the regression is maximized but the value of Π* at which the sum of the coefficients of
extra inflation and Π* significantly change sign may be less than at Π* .
Plot of sum of the coefficient of inflation and extra inflation and R-square against Π* for
all twenty estimated equation is used to see whether the estimated threshold level does
qualify the Sarel (1996) sense of threshold (see figure 1 and 2). Looking at the figure 1 and
2, where the R-square are plotted against different values of Π* , it can be said that the
threshold level of inflation is 6 percent. Now if we take a look at the Table 2 it is observed
that for GDPGR and GDPPCGR, sum of INFt and (INFt -Π*) both the coefficient is
negative and significant. Whereas, looking at the plot of figure 3 and 4 for the sum of
coefficients of INFt and (INFt -Π*), it is clear that the sum of coefficients is negative.
These findings indicate that there is no threshold level of inflation—in Sarel (1996)
sense—for India.
Both the equations (equation 1 and 2) are also estimated for quarterly data with slight
change in variable, but the results were not very convincing as the coefficient of threshold
variable was coming out to be insignificant, so the idea of further analysis using quarterly
data was dropped. Results of quarterly data can not be directly compared with annual data
2 Table 2 only presents the regression result for threshold inflation i.e. at 6 percent of inflation to conserve
space. Result is available on demand from author for entire spectrum of threshold inflation.