A disequilibrium model for international capital flows

This study analyses international capital movements for sixteen industrial countries. the basic aim is to check the feasibility of an econometric approach to a subject with lots of institutional detail and rapid structural change. therefore, special attention is given to the statistical parameter estimation problem. instead of ad hoc specifications which might pretend a better fit in the sample period but do not necessarily improve the forecasting properties, all estimated equations strictly follow the theoretical model derived from the theory of international capital movements. starting with the two major developments presented recently in the theory of international capital flows, namely the portfolio model and the monetary approach to the balance of payments, we analyze the 'reduced form synthesis' of both concepts. as a complement we propose a disequilibrium model which emphasizes a structural specification and thus overcomes some problems inherent in the reduced form model. in addition, the disequilibrium model enables empirical tests of the relative importance of monetary and portfolio allocation motives as causes for international capital flows. in all sixteen countries considered the estimation results stress the importance of both monetary and asset allocation aspects. for each country significant influences of current account balance, asset behavior of monetary authorities, national income, interest rates, and exchange rates on capital flows could be traced. in addition, for most countries wealth effects and explicit speculative factors enter the model. the complete system is highly interdependent via interest rates, exchange rates, and wealth effects. despite the fact of large disturbances by singular speculative or institutional events the empirical analysis reveals strong evidence of systematic influences as causes for international capital flows. this result was obtained by using a robust estimation technique that fades out singular disturbances in the sample period .;