臺大管理論叢 NTU Management Review VOL.30 NO.3

Time Paths of Weather-Induced Mood Effects on Stock Returns 64 Table 5 Impulse Responses of the Stock Return to One Standard Deviation Shock of Individual Weather Variables Weather Variable Impulse Response (Day) 0 1 2 3 4 Air Pressure 1.24E-05 -1.48E-04 -1.23E-04 -9.01E-05 -6.38E-05 20.2795 37.4796* 34.5617 32.5428 29.9484 Cloud Cover 2.83E-04 1.76E-04 1.77E-04 4.03E-05 4.82E-05 25.9132 22.1745 26.3584 18.9866 21.2916 Ground Visibility -5.52E-05 7.63E-05 5.56E-05 3.26E-05 1.47E-05 28.2545 26.0435 23.1761 21.5046 18.2329 Rainfall -8.11E-05 4.70E-05 1.93E-05 5.13E-06 1.51E-06 31.6102 20.4855 7.1738 5.6239 3.1311 Relative Humidity -5.00E-05 2.62E-04 1.55E-04 9.32E-05 5.30E-05 24.6737 13.9074 14.1904 14.1558 13.4907 Temperature -1.10E-04 -1.51E-04 -8.99E-05 -5.88E-05 -3.86E-05 14.6581 19.0426 18.9985 18.4133 17.3674 Wind Speed -1.83E-05 -1.47E-04 -2.95E-06 -2.71E-05 7.65E-06 20.4562 50.8701*** 42.7959** 33.7250 30.0898 Note: In each row, the numbers in the upper line are the average impulse responses for years 1992 to 2017, and the numbers in the lower lines are the χ 2 26 statistics for the joint hypotheses that the impulse responses are zero in all years from 1992 to 2017. *, **, and *** = significant at the 5.7 The Roles of Investor Trading Behavior The mechanism through which weather-induced mood affects the stock return and volatility is the trading behavior of marginal investors and their roles in the price formation process. Hence, the trading behavior of marginal investors is driven by weather conditions. Previous studies suggest that weather-sensitive investors could be individual investors (Schmittmann et al., 2015) or institutional investors (Goetzmann et al., 2015). There are four major investor groups in the SET‒local individual investors, local institutional investors, local proprietary investors, and foreign investors. Their average trading shares from 1992 to 2017 were 57.66%, 7.08%, 8.38%, and 26.89%, respectively. In order to understand which investor groups contribute to the return behavior in Table 2, I estimate equation (1) with X' t = [W t v i t r t ]. v i t is the net buying volume of investor group i scaled by market capitalization. The trading volume data are retrieved for the SET database.