Time series tests of the AK model of endogenous growth.

摘要

The AK model of endogenous growth takes output as a linear function of broad capital alone and predicts that capital accumulation is the primary determinant of the long-term growth rate of output. Permanent changes in government policies affecting investment rates should lead to permanent changes in a country's growth of GDP per capita. Jones (1995) saw no evidence of the prediction of the AK model for 15 OECD countries using data from 1950-1988. He found that rates of investment, especially for equipment, have risen persistently over time while GDP growth rates have not. In the first chapter of this dissertation, we check the empirical consistency of the AK model using Penn World Table 6.2 data for 112 countries for the period 1950-2004. We do not find evidence against the findings of Jones (1995) even with a broader selection of countries and a wider time frame. This suggests that the AK model cannot be applied to understand the growth and development experiences of most economies of the world, including India. The major exception is China. Using data (1978-2004) from Dekle and Vandenbroucke (2006) and unit root tests that permit multiple structural breaks, we show in the second chapter that the Chinese growth and investment/GDP series (at the aggregate and industrial levels) are both segmented trend stationary, have trends of the same direction and experience structural breaks in approximately the same years. We also find a stable positive correlation between China's investment rates and growth rates, and establish that the direction of causality runs strongly from high past investment shares of GDP to current growth. This supports the view that China's growth may have been predominantly driven by the accumulation of physical capital rather than by increases in capital productivity. In the third and final chapter, we test the planning exercise dubbed the 'financing gap' by aid agencies. It involves the calculation of the aggregate investment rate required to achieve an economic growth rate, assuming a Harrod-Domar model of growth. It assumes that public capital, perhaps supported by foreign borrowing, can be substituted for any shortfall between the realized and target investment rates. This requires that private and public investments are equally effective in promoting growth. We ask whether support for a linear relationship between growth and measured investment rates can be restored for India if we allow the efficiency of private and public investments to differ. We find that the dynamics of public investment and growth rates are jointly inconsistent with a role for public investment in driving growth in a Harrod-Domar framework in India. Using Vector Autoregression analysis on growth and private investment rates we conclude that shocks to private investment have only short-term effects on growth. This suggests that effects of private investment on growth work through the demand-side rather than through the supply-side long-run accumulation of capital. Some comparisons with the Chinese experience in this regard are offered.