Testing International Asset Pricing: Some Pessimistic Views

Abstract

IN THE LATE SIXTIES modern capital market theory started to be applied to the area of international investment (Grubel, (1968), Levy and Sarnat, (1970)). The original idea was to apply the Markowitz-Sharpe risk-return analysis to expost data on stock prices and exchange rates, in order to determine the optimal internationally diversified portfolios that investors of each country should hold. It has been argued that optimal portfolios should be different for investors of different countries because of exchange risk. This argument, supported by some empirical evidence, has been used to justify capital flows between countries (see, for example, Levy and Sarnat, (1975), Lee, (1969), etc.). This paper will argue that such a conclusion is inconsistent with the theoretical framework used. Moreover, the big issue is the economic definition of exchange risk. More recently equilibrium models of the international capital market have been proposed (Solnik, (1974), Grauer, Litzenberger and Stehle, (1976)). Their underlying definitions of exchange risk are different and lead to somewhat different pricing relations and optimal portfolio selections. Before engaging in any tests of integrated versus segmented capital markets and international pricing, it seems to be a useful time to confront the various models, especially as far as exchange risk is concerned. Empiricists consistently resort to mean-variance analysis of expost-data to test their various models and often confuse the economic results with some obvious technical implications of the mean-variance framework. This paper will attempt to give a fair representation of the various international asset pricing models, stressing their real economic conclusions. It will then claim that is is very unlikely that an empirical mean-variance analysis will ever be able to discriminate between the various views of the world.1