Losing sleep at the market: an empirical note on the daylight saving anomaly in Australia

Economic Papers (Economic Society of Australia), Dec, 2003 by Andrew C. Worthington

Once again, the signs on the estimated coefficients appear to offer prima facie support for the posited daylight saving effect. The signs on WKD and STT are always negative, with the magnitude of STT being larger, suggesting the presence simultaneously of both the weekend and daylight saving effect market anomalies. That is, the weekend effect is associated with a lower mean return than others days and the start of DST is linked with a larger negative effect as compared to other weekends. The sign on END is likewise consistence with the hypothesis that mean returns following the end of DST are positive, or at least higher, than the mean returns of DST starting and non-starting weekends. However, only in the case of DST starting weekends are the least squares estimates of market returns significant.

Breusch-Godfrey Lagrange multiplier and White's heteroskedasticity tests are used to test for higher order serial correlation and heteroskedasticity in the least squares residuals, respectively. To start with, for both the price (test statistic = 25.9878, p-value = 0.0000) and accumulation (test statistic = 26.0726, p-value = 0.0000) returns least squares regressions, the null hypothesis of no serial correlation is rejected and we may conclude the presence of higher order serial correlation in the residuals. Then the null hypothesis of no heteroskedasticity in the least squares residuals fails to be rejected where price (test statistic = 0.1309, p-value = 0.9417) and accumulation (test statistic = 0.2282, p-value = 0.8768) returns are specified as the dependent variable and we conclude the presence of heteroskedasticity in the least squares residuals.

After corrections are made for heteroskedasticity (White) and heteroskedasticity and autocorrelation (Newey-West) most of the parameters in Table 1 are significant at any conventional level, irrespective or whether market returns are specified in price or accumulation terms, or whether the transition weekends are in all Australian states and territories or only NSW and VIC. On average, weekend returns that follow the start or end of daylight saving time are not abnormally high or low when compared to other weekends or other days of the week. Furthermore, there is no significant difference between weekend returns as defined (Friday-Monday) and daily returns Monday-Tuesday, Tuesday-Wednesday, Wednesday-Thursday and Thursday-Friday.

5 Concluding Remarks and Policy Recommendations

The present study employs parametric analysis to test for the 'daylight saving effect' market anomaly in the Australian stock market. At first there would appear to be some empirical evidence to support the conjecture that the transition to (and from) daylight saving, as variously defined, is associated with a lower (higher) mean market return than either other weekends or other days. However, after adjustments are made for heteroskedasticity and/or autocorrelation, neither the transition to daylight saving nor the movement from daylight saving is associated with returns that are statistically significantly different from other days, let alone other weekends. These results lie counter to the US, UK and Canadian findings of Kamstra et al. (2000, 2002) but are similar to the results of Pinegar (2002, p. 1255) who also found that the hypothesis that "... mean weekend returns are significantly lower following changes in daylight-saving vis-a-vis other weekends is not robust". Indeed, there would also appear to be no evidence in this study to support even the well-investigated hypotheses underlying the weekend effect market anomaly. Daylight saving in Australia may be opposed on a number of policy grounds, but it would appear that adverse affects on capital markets should not be one of them.