Efficient markets models assert that the price of each asset is equal to the optimal forecast of its ex-post or fundamental value. These models do not imply, however, that the covariance between two asset prices is given by the covariance between the ex-post values they respectively forecast: these two covariances can even have opposite signs. However, it is possible to place bounds on the covariance between asset prices given the covariance matrix of ex-post values. We present such bounds for both covariances and correlations and show how such bounds can be tightened using information beyond the covariance matrix of ex-post values. The methods are used to examine whether the historical correlation between the U.S. and U.K. stock markets 1919-1989 is warranted. The bounds on the warranted covariance are very wide and include the actual correlation.
Keywords: Volatility, stock market, asset pricing, efficient markets, information