tỷ lệ mục tiêu khu vực trao đổi, và tương đồng giữa nhập cư và di động vốnHầu hết các kết quả chính là nguồn gốc cho cả hai nước nhỏ và các trường hợp nền kinh tế thế giới. Bảy chương đầu tiên bao gồm các mô hình của nền kinh tế thực sự, | 311 Asset Pricing G ệ fi rm - Ejrm - ftp ậ - 1 rm - Etrm . C1 C1 Taking conditional expectations of both sides above yields Íc Ì Cl J recall footnote 32 . Using this estimate to eliminate the covariance term in eq. 54 yields the approximation we seek ÍC - l rm r G1 ÍCP Ì - BStdi rm-r Cl J where K Corrj C2 C1 rm r is the conditional correlation coefficient between consumption growth and the excess return on equity and the conditional standard deviation Stdi is the square root of the conditional variance. 37 Mankiw and Zeldes 1991 point out that taking averages over the entire Mehra-Prescott data set K the standard deviation of per capita consumption growth is per year and that of the excess equity return is per year. If we assume that 1 r P 1 then the model fits the excess average annual 1889-1978 equity return of E rm - r - only if p is roughly 26 38 As Mankiw and Zeldes 1991 also observe the equity premium puzzle is even more severe in postwar . data than in the original second moments. Ingersoll 1987 pp. 259-262 describes the class of compact distributions for which this will be the case if the trading interval is sufficiently small. 37. When variables are jointly lognormally distributed one can obtain an exact expression for the equity premium that closely parallels this one. See for example Aiyagari 1993 . 38. Another way to see the puzzle is to let mrs2 be the ex post marginal rate of intertemporal substitution realized on date 2 and rewrite eq. 51 as 1 E mrv2 l Covi znr. 2 1 rm Ei znr52 Ei l r 1 which implies that 1 Ei mrj2 Ei l rm Covi mrs2 1 rm . Observe that Stdi mrj2 Stdi l rm Cov m i2. 1 r l CotT . 1 rm where Corri Ímrỉ2 1 r the correlation coefficient is below 1 in absolute value. Thus 312 Uncertainty and International Financial Markets Mehra-Prescott sample with values of p as high as 100 needed to justify the observed premium on stocks. Most economists feel that such high levels of risk aversion are .
