Pettengill, Sundaram, and Mathur 111 FIGURE 1 Average Portfolio Return: Positive and Negative Market Excess Return Periods 6.00% 4.00 Positive Excess 0.00% a-Negative Excess Returns 00% 6.00% Portfolio asonality in risk-Return Relation We test for seasonality in the risk-return relation by segmenting the data by months and reestimating Equations(3)and (4). Table 3 reports the regression coefficients from Equation( 3). Examination of these results shows the rejection of the null hypothesis of no risk-return relation only for the months of January (t= 5. 13)and February(I= 2.23). This observed seasonality is consistent with Tinic and West(1984), who find a significant and positive risk-return relation only in the month of January. Further, six out of 12 months report a negative slope coefficient, implying an inconsistent risk-return relation This observed inconsistency may be primarily due to the bias from the con- ditional relation between beta and realized returns. This contention is tested by examining the regression coefficients of Equation(4) with data segmented by months. The results are presented in Table 4. When market excess returns are positive, a significant positive relationship exists between beta and return for each of the months. The null hypothesis of no risk-return relation is rejected at the 0.01 level for each of the months except June, September, and October(null rejected at the 0.05 level). When market excess returns are negative, a significant negative relation exists between beta and portfolio returns for all months except January (t=-0.92).8 These results firmly support a consistent relation between beta and returns throughout the year when the conditional relationship between beta and realized returns is considered The ensign relationship in January may be explained by the small firm effect. Small firms, nuary(see Reinganum (1983), possibly causing high beta portfolios to have relatively high returns even during periods of negative market excess returns