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

Time Paths of Weather-Induced Mood Effects on Stock Returns 62 Table 4 Robustness Tests Specification Impulse Response (Day) 0 1 2 3 4 Triennial Setting a -0.0156 -0.0125 -0.0013 -0.0003 0.0000 7.6766 14.5231 9.9157 8.0682 4.9086 Triennial Setting b -0.0163 -0.0174 -0.0082 -0.0062 -0.0034 7.2526 18.2542** 11.4129 7.6260 5.7863 Annual Sample with Control Variables -0.0219 -0.0165 -0.0066 -0.0041 -0.0013 43.3618** 31.2410 16.7637 13.4386 11.0613 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 m statistics for the joint hypotheses that the impulse responses are zero in all years from 1992 to 2017. The degrees of freedom m are 5.4 Nonsignificant Contributions of the Price Adjustment Process Wermers (1999) argues that the price adjustment process could cause permanent weather effects. In this study, the full sample test do not support this thesis. However, when I check the results for each year in the sample, I find the significant cause at the 95% and 90% confidence levels for 1992 and 2005, respectively. These findings suggest that the contribution of the price adjustment process to the permanent weather effects exists. However, the contribution is small. 5.5 Recursive Causal Ordering In the three-variable VAR model, I impose a recursive causal ordering from W t to σ 2 t and σ 2 t to r t . It is possible that the results for the impulse responses change with alternative causal ordering (Enders, 2015). To check for robustness, I consider the ordering from W t to r t and r t to σ 2 t . The ordering is consistent with the leverage effects (Black, 1976) under which the volatility increases (decreases) with falling (rising) prices. The results for the impulse and cumulative responses of the variance to the weather shock are exactly the same because the shock stays at the first order. The absolute correlation |ρ r,σ 2 | changes slightly in size. The qualitative conclusions for |ρ r,σ 2 | = 0 and |ρ r,σ 2 | = 1 remain unchanged. 5.6 The Effects of Individual Weather Variables There are seven weather variables: air pressure, cloud cover, ground visibility, rainfall, relative humidity, temperature, and wind speed‒used in this study. I choose the fourth principal component to summarize the common factor to simplify the analysis.