Does public capital crowd-out or crowd-in private capital in India?

Abstract

This study examines the long-run effects of public capital on private capital and tests the null of Granger non-causality between public and private capital in India. Both single-equation and system estimates of the model consistently suggest the long-run crowding-in effects of public capital. The error-correction as well as over-parameterized level-VAR models consistently suggest uni-directional Granger-causality from public to private capital. The support for the significant crowding-in effects of public capital has important implications for the formulation of long-term growth and development strategies. It underlines the need to accelerate public infrastructure to induce the distortions-free and market-driven increases in private capital and to attract the inflow of foreign direct investment. The inflows of foreign capital have witnessed perceptible increases since the beginning of the 1990s. The geographical and sectoral distributions of FDI inflows remain asymmetric and skewed in favor of the select regions and the services industries. The services-sector-led growth needs to be accompanied by the commensurate performance of the goods-producing, agricultural and industrial, sectors so as to sustain the escalated trajectory of economic growth.

Keywords:

• public capital
• private capital
• crowding-out
• cointegration
• error-correction model

JEL Classification:

• C32
• C50
• E62
• O10
• O40

Acknowledgement

I am grateful to the editor and an anonymous referee of the journal for very useful comments and suggestions on the paper. I am, however, solely responsible for any errors and omissions that may remain.

Notes

1. The ADF test on ϵ(t) is performed using an AR lag of k=1. The PP and KPSS tests are each performed using the lag windows (lw) of lw=1 and lw=4. The results obtained from both the lag windows provided similar evidence for the nulls in PP and KPSS tests and are, therefore, reported only for one of the lag windows (lw=4) to conserve space.

2. The OLS estimators of a regression are ‘super-consistent’ when the model series are cointegrated. Instead of approaching their true values at a rate proportional to n ^–1/2, the OLS estimates will approach them at a rate proportional to n ^–1 (Davidson and Mackinnon 1993).

3. The SIC suggested a dynamic structure of k{–3, 0, +3}, while the AIC suggested k{–4, 0, +4} for the DOLS estimator. Both SIC and AIC suggested a dynamic structure of k{–4, 0, +4} for the NLLS estimator. The use of a larger model structure may impinge upon the efficiency of the estimates and, therefore, the model is estimated using k{–1, 0, +1} and k{–2, 0, +2}. The results remain robust to the use of higher model structures. The standard errors of the parameters from DOLS and NLLS estimators are adjusted using the HAC estimator.

4. The capital–output ratio (COR) is defined as: ; where (K/Y) is the capital–output ratio, (K/L) the capital–labor ratio and (Y/L) the output–labor ratio. The low (high) output–labor (Y/L) ratio and implied low (high) productivity of labor leads to high (low) capital–output (K/Y) ratio and implied low (high) productivity of capital. By taking first-difference transformation of K/Y and then dividing through by Y, the COR can be represented in terms of the incremental capital-output ratio (ICOR) so that ; where denotes the growth of output and indicates the investment. The ICOR = rearranged as suggests that the acceleration of growth of output requires the acceleration of investment and/or the reduction in ICOR and implied improvement in productivity.