second subperiod, for two out of the three trading rules, the abnormal returns are always positive. The trading rule was repeated for cases where the market price was at lea five percent below the tender offer price. The results(available from the authors show that, while the number of observations falls by approximately 30 percent the average and median abnormal returns from the trading strategy increase by about two percent relative to the results reported in Table II To test whether the results depend on the market price relative to the tender price, the following regression was run XRET:=a+ B(Pri-. i)/PB.i + eir where XRETi is the abnormal return from the trading strategy for case i. All 19 cases in which the pre-expiration price was below the tender offer price are included in the sample Table IlI presents the regression coefficients and R2 for the various trading rules. The degree of underpricing (relative to the tender offer price) explains more than 75 percent of the variability of the returns generated by the trading strategy, and the slope coefficients are highly significant (t> 23). The slope coefficient of 0.85 implies that buying shares before the expiration date whenever he market price is x percent below the tender price results in abnormal returns of. 85x percent. Thus, the performance of the trading rule can be enhanced by focusing on cases where the pre-expiration price is substantially below the tender price. Note that this result is anomalous: in an efficient market, a low Pb relative to Pr would simply imply that the market expects a low Pe and/or a large A closer examination of the regression results reveals four outlier observa tions. Hence, a nonparametric Spearman rank correlation test was performed. The rank correlation coefficients for the three trading rules are 0.687, 0.667, and 0.685, respectively, and are statistically significant In Table Iv results are presented for the trading strategy which assumes that stocks are purchased six trading days prior to the expiration of the offer if the market price is at least three percent below the tender offer price. The nonre purchased shares are sold twelve trading days after the expiration of the offer The average and median abnormal returns are somewhat smaller than their corresponding values in Table II but are still substantial. On average, abnormal returns exceed 5.5 percent in all periods and are positive in about 90 percent of the cases. The trading strategy was repeated for cases where the market price was at least five percent below the tender offer price. This strategy increased the abnormal returns by about two percent relative to the results in Table Iv In order to test the relationship between the degree of underpricing and trading abnormal returns, we run the regression in equation(2). The results are consistent with the findings in Table III: a=-002(t= 3.62), 8=0.86(t= 18.85), and R 1 Regression results for the 109 cases where the market price was at least three percent below the tender price are similar to those reported in Table IlL. 12 We reran the regressions without the four outliers. The R for the three regressions in Table IV re equal to 0.58. 0.51, and 0.47, respectively. In each case, the independent variable was highly gnificant (t> 12)and the regression coefficient was larger than. 54