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

Time Paths of Weather-Induced Mood Effects on Stock Returns 52 report that analytical standard errors are at least as good as those obtained from Monte Carlo simulations and bootstrapping methods. In this study, the sample period covers 26 years. For a full sample test, the Wald statistics are the sum of statistics for all 26 years. Hence, these statistics are chi-square variables with 26 degrees of freedom (Doyle and Chen, 2009). 2.6.2 Direct and Indirect Causes of Permanent Weather Effects Significant permanent effects can be caused directly by the price adjustment process (Wermers, 1999) or indirectly by the Friedman and create-space effects (Lee et al., 2002). The resulting effects from the two causes are bundled together. To understand the contribution of each cause to the effects, I re-estimate equation (1) with X' t = [W t σ 2 t r t ]. σ 2 t is the variance of r t . The recursive causal ordering runs from W t to σ 2 t and σ 2 t to r t . This causal structure is consistent with the model in which returns are explained by their volatility (Lee et al., 2002). In the Wold representation, the permanent weather effects on the variance σ 2 t and stock return r t are lim J↑∞ Σ J j=0 Θ j 2,1 and lim J↑∞ Σ J j=0 Θ j 3,1 , respectively. I calculate the correlation ρ r,σ 2 of r t with σ 2 t due to the weather shock η W t by ρ r,σ 2 = . If the permanent effects are exhaustively explained by the direct (indirect) cause, the null hypothesis must be |ρ r,σ 2 | = 0 (|ρ r,σ 2 | = 1), and the Wald statistic is a chi-square variable with 1 degree of freedom. If the direct and indirect causes explain the permanent effects, the hypotheses |ρ r,σ 2 | = 0 and |ρ r,σ 2 | = 1 must be rejected. I do not test the hypotheses |ρ r,σ 2 | = 0 and |ρ r,σ 2 | = 1 jointly but consider each hypothesis separately. The significance result from a joint test does not necessarily imply the significance results for both hypotheses. 3. The Data The stock returns are the daily logged differences in the closing indexes of the SET index portfolio. The realized daily variances are computed by Rogers and Satchell’s (1991) adjusted extreme-value estimator. This estimator is efficient, simple, and general. The computation requires data on opening, closing, maximum, and minimum indexes readily observed during the day. The weather variables are Bangkok’s seven weather variables—air pressure (hectopascal), cloud cover (decile), ground visibility (meters), rainfall (millimeters), relative humidity (%), temperature (°C), and wind speed (knots per hour). These variables measured at Don Muang Airport by the Thai Meteorological