Evidence on the fundamental determinants of investors' expectations of risk
Previous research shows that a number of firm characteristics explain the crosssection of common stock returns. These characteristics either (i) are functions of stock prices, or (ii) are not functions of stock prices and hence depend only on accounting disclosures. Characteristics in the first class reflect and summarize investors’ risk opinions while characteristics in the second class contribute to the determination of investors’ risk opinions. This study draws a distinction between these two classes in order to parsimoniously characterize the accounting disclosures that determine investors’ opinions of risk and to evaluate the importance of accounting disclosures for determining investors’ opinions relative to non-accounting information. The results show that: (1) Investors’ opinions about systematic risk are determined by profitability, firm size and the growth of firm size. (2) There are strong seasonal patterns in the expected return premia of the accounting determinants of opinions. Specifically, between January and September, the premia for profitability, size and size growth are negative. Between October and December, the premia for profitability and size are positive and the premium for growth continues to be negative. (3) Between January and September, accounting determinants explain 81% to 91% of the variation in the cross-section of average returns. Between October and December, accounting determinants explain 77% to 88% of the variation in returns. This suggests that, although investors’ opinions depend on nonaccounting information, investors’ opinions are determined primarily by accounting disclosures. (4) The cross-sectional variation that accounting determinants do not explain has implications for risk measurement; its magnitude indicates that accounting disclosures do not contain sufficient information about investors’ opinions to effectively measure the risk of equities with extreme exposure to non-accounting determinants.