摘要：Equity returns are typically higher correlated during market downturns than during bullish times. This paper develops a novel approach how investor expectations for such correlation asymmetries can be quantified from forward-looking data. Based on option implied volatilities, it is found that the correlation asymmetry is significant, rejecting the use of the classic mono-correlation assumption. Further, the spread between expected down and up correlations is time-varying and positively dependent on the current market mood: stock diversification is more difficult when it is needed the most. Thus, the three main advantages of the proposed model are (ⅰ) the distinction between up- and down-correlations, (ⅱ) it actually captures investor expectations as traded in current market prices and (ⅲ) the immediate response to the current market outlook. Practical relevance of this paper is highlighted by the computation of expected up-/down CAPM betas.
