Government Expenditure in India: Composition and Multipliers

Abstract

We estimate fiscal multipliers for total, capital (capex), and revenue (revex) Indian government expenditure using a two variable Structural Vector Auto-Regression (SVAR). Our quarterly data allows us to estimate both short- and long-run multipliers. We then extend and re-estimate the model including supply shocks and the monetary policy response sequentially and together and re-estimate the multipliers. The long-run capex multiplier remains much larger than the corresponding revex multiplier in all the estimations. The short run impact multiplier is the highest for revex, but does not rise after the first quarter. The capex peak multiplier in the 2nd quarter is 1.6–1.9 times larger. The cumulative multiplier is also the highest for capex, 2.4–6.5 times the size of the revex multiplier. Capex also reduces inflation more over the long-term. Despite this, capex shows greater volatility since it is more vulnerable to discretionary cuts. Monetary accommodation of capex and revex is allowed to differ. It varies in the absence/presence of supply shocks. The combination of a direct cut in capex and monetary tightening in response to a supply shock reduces the capex multiplier. The results are consistent with an elastic long-run aggregate supply. Disaggregated evaluation of spending policy, therefore, gives useful insights.

Appendices

Appendices

A1. Stationarity of variables used in estimation

Variable	t-statistic	p-value
Capex growth	− 8.38117$^{***}$	0.00
Real interest rate	− 3.239939$^{**}$	0.0224
GDP growth	− 2.70917$^{*}$	0.0783
Revex growth	− 3.53444$^{**}$	0.0103
Totex growth	− 6.63957$^{***}$	0.00
Inflation	− 4.90293$^{***}$	0.0001

1. ***Significant at 1% level of significance, **Significant at 5% level of significance, *Significant at 10% level of significance

Capex growth—Growth Rate of Central Government Capital Expenditure.

Revex growth—Growth Rate of Central Government Revenue Expenditure.

Totex growth—Growth Rate of Central Government Total Expenditure.

Inflation—WPI Inflation Rate

Real interest rate—Short-term real interest rate

The CMMR variable is stationary at 5% level of significance using the DF-GLS test.

Null hypothesis: CMMR has a unit root

Exogenous: Constant

Lag length: 0 (Automatic-based on SIC, maxlag=10)

	t-Statistic
Elliott-Rothenberg-Stock DF-GLS test statistic	− 2.231382
Test critical values: 1% level	− 2.602794
5% level	− 1.946161
10% level	− 1.613398

A2. Structural VAR Analysis

See Figs. 4, 5, 6 and Tables 5, 6, 7, 8, 9, 10

Fig. 4

IRF Capex a Including supply shocks. b IRFs Revex: including supply shocks

Fig. 5

a IRFs revex: including monetary shocks. b IRFs Capex: including monetary shocks

Fig. 6

a IRFs Revex: including monetary and supply shocks. b IRFs Capex: including monetary and supply shocks

Table 5 VDA Capex: Including supply shocks

Table 6 VDA Revex: including supply shocks

Table 7 VDA Revex: including monetary shocks

Table 8 VDA Capex: including monetary shocks

Table 9 VDA Revex: Including monetary and supply shocks

Table 10 VDA Capex: including monetary and supply shocks

A3. Robustness Checks:

See Figs. 7, 8 and Tables 11, 12

1. Introduction of additive dummy variable for the crisis period between 2008Q1 to 2009Q4

The GFC of 2008, triggered by the sub-prime mortgage crisis in the U.S. and the collapse of Lehman Brothers, saw an increase in fiscal expenditure by the Government in order to insulate the economy from the headwinds of downturn. It is thus necessary to check if this fiscal expansion and its exit, however slow and partial, starting 2010, influence the multiplier analysis.

A dummy variable ‘Crisis_Dum’ is introduced in the basic two-variable SVARs to see if the values and the behaviour of the multipliers and the IRFs change drastically once the crisis-led fiscal expansion is accounted for as an exogenous event in the model.

Fig. 7

IRFs a Total Central Expenditure Growth and GDP Growth. b Capex Growth and GDP Growth. c Revenue Expenditure Growth and GDP Growth

Fig. 8

IRFs a Total Central Expenditure Growth and GDP Growth. b Capex Growth and GDP Growth. c Revenue Expenditure Growth and GDP Growth

Accounting for this variable marginally increases the impact and peak multipliers for revex and leads to a significant increase in the cumulative multiplier. This increase can be attributed to the reduction in the fluctuations in impulse responses of revex to shocks to itself, since the crisis dummy accounts for a significant amount of movement in the revex growth (which is not the case for capex growth).

It is evident that despite accounting for the crisis dummy variable, the peak and cumulative multipliers for capex are larger than revex, implying that the long-run cumulative impact of capex is stronger even in the face of adverse exogenous shocks to the economy. The inclusion of this variable in the model does not alter the nature and magnitude of the impulse response of GDP growth to shocks in expenditure variables either.

2. Analysis with expenditure series deflated by GDP deflator series

The implicit GDP deflator series for the period 1998Q1 – 2014Q2 has been retrieved from the St. Louis Federal Reserve Bank’s Economic Research website (https://fred.stlouisfed.org/series/INDGDPDEFQISMEI). The base year is 2010-11 and the series has been seasonally adjusted.

The growth variables are derived from the expenditure variables deflated by the above series. The average GDP to real total expenditure ratio in this scenario for 1999 Q1 to 2014 Q2 was 3.39, for real revex it was 3.97, for real capex the average ratio it was 25.93. The ratios are lower than that for WPI deflated series due to the differences in the values of WPI series (base year = 1993–1994) and the implicit GDP deflator series (base year = 2010–2011).

The Impulse Response Functions (IRFs) to structural one standard deviation innovations and two standard deviation error bands are given in Fig. 8a–c.

Table 11 Multipliers with short-run restrictions

Table 12 Multipliers with short-run restrictions

The IRFs do not show any marked difference in direction and magnitude from the original real expenditure variables (deflated by WPI series). The multiplier values also show the same behaviour, with the ratio of capex Multipliers to their revex counterparts being in the same range as observed in Table 1. The values are lower on account of lower magnitude of average output to spending ratio. Nevertheless, this analysis also confirms our major finding that the capex multipliers are much larger in the long-run as compared to revex multipliers.