A Nonlinear Econometric Analysis of Capital Flight

Abstract

This paper develops a nonlinear econometric model of the determinants of capital flight for eighteen developing countries over the period 1978 to 1988. During most of the 1980s, capital flight proved difficult to reverse. Besides an uncertain domestic policy environment, costs associated with such an investment decision might have delayed flight reversal. Costs that characterize flight reversal could generate potentially significant barriers to continual adjustment of an investment position. This in turn would generate a nonlinear relationship between flight and its determinants, which reflect the degree of domestic macroeconomic imbalance. To detect the existence, and importance, of cost-driven thresholds we model flight within the context of a friction, or two-threshold Tobit, model. Given the panel data set, we consider a country-specific error component to account for the possibility of unobserved country-specific heterogeneity. To correct for endogeneity among the regressors in the random-effects nonlinear estimation, we implement Newey (1987). While the data seemingly do not support the existence of cost-driven thresholds for flight, the central government surplus, premium for foreign exchange in the black market and the presence of an IMF adjustment program are all significantly related to flight. The negative relationship between fiscal balances and flight highlights the desire by investors to escape future taxation directly, and indirectly via monetization of fiscal deficits, by undertaking capital flight. In devising a framework for reform, that typically entails fiscal consolidation, IMF supervision, endorsement and surveillance lend credibility to reform and reassure investors.