Previous work by Dumas and Solnik (1993) has shown that a CAPM which incorporates foreign-exchange risk premia (a so-called 'international CAPM') is better capable empirically of explaining the structure of worldwide rates of return than does the classic CAPM. In the specification of that test, moments of rates of return were allowed to vary over time in relation to a number of lagged 'instrumental variables'. Dumas and Solnik used instrumental variables which were endogenous or 'internal' to the financial market (lagged world market portfolio rate of return, dividend yield, bond yield, short-term rate of interest). In the present paper, I use as instruments economic variables which are 'external' to the financial market, such as leading indicators of the business cycles. This is an attempt to explain the behavior of the international stock market on the basis of economically meaningful variables which capture 'the state of the economy'. I find that the leading indicators put together by Stock and Watson (NBER working paper no. 4014, 1992) as predictors of the U.S. business cycle also predict stock returns in the U.S., Germany, Japan and the United Kingdom. These instruments lead again to a rejection of the classic CAPM and no rejection of the international CAPM.