臺大管理論叢 NTU Management Review VOL.29 NO.1

19 NTU Management Review Vol. 29 No. 1 Apr. 2019 Market constant ILLIQ SIZE B/M logSD R j,t-2-4 R j,t-5-13 R 2 Asia-Pacific Markets (16 countries) Mean 4.760 0.099 -0.298 332.171 -0.760 0.007 0.002 11.9% ( t -statistic) (4.65) (3.16) (-4.15) (1.81) (-6.38) (2.87) (1.88) Median 4.569 0.093 -0.278 278.415 -0.675 0.006 0.002 10.7% % positive 100.0% 87.5% 0.0% 93.8% 0.0% 75.0% 68.8% p -value 0.000 0.002 1.000 0.000 1.000 0.038 0.105 Table 3 presents the cross-market average values of the regression coefficients in equation (2). All firm characteristics in equation (2) (i.e., size, book-to-market, volatility, and past returns) show up as significant predictors of stock returns. Consistent with prior literature, we find that stocks of smaller firms, stocks with less volatile return, stocks of value firms and stocks of past winners earn higher future returns. The findings hold for the global sample as well as for the three sub-groups corresponding to emerging, developed, and Asia-Pacific markets. Importantly, we find that stock returns are positively related to stock illiquidity in international markets, after controlling for the effects of the above stock characteristics. For the sample of all 44 markets, the estimated coefficient of ILLIQ averages to a significant 0.081 ( t = 4.71). The median value of the regression coefficient is similar at 0.07. We also find that about 86% of the coefficients associated with ILLIQ are positive, which rejects the null hypothesis that the positive coefficients are due to chance ( p -value < 0.001). Significant evidence on a positive illiquidity premium is also present in all three sub-groups. As shown in Table 3, the average regression coefficient is positive and significant in the subgroups of emerging markets, developed markets and Asia-Pacific markets at 0.086 ( t = 2.61), 0.077 ( t = 4.17), and 0.099 ( t = 3.16) respectively. Using the coefficient b1 c of lagged illiquidity as a measure of the illiquidity premium, we replicate the estimation above to test whether the illiquidity premium thus estimated differs significantly between the Asia-Pacific markets and the rest of the world. Specifically, we regress b1 c on the two dummy variables defined above: b1 c = a0 + a1 *DUM-ASIAPAC c + a2 *DUM-EMERGE c We find the following results (in parentheses are the t -statistics): a0 = 0.069 (2.80), a1 = 0.028 (0.76), a2 = 0.003 (0.07) R 2 = 0.02.