00Thaler_FM i-xxvi.qxd

MARKET EFFICIENCY AND BIASES 405

Table 11.1.

Excess Returns before, at, and after Analyst Buy Recommendations

Differentiated by Underwriting Relationship

Excess returns (size-adjusted mean and median buy-and-hold returns) are calculated
for periods before, at, and after the added-to-buy recommendation event date given
on First Callfor the 214 observations in our sample. Size adjustment is calculated by
subtracting the buy-and-hold return from the appropriate value-weighted CRSP
decile. We define “by underwriter” as recommendations made by equity research an-
alysts of the lead manager of the IPO and “by non-underwriter” as recommenda-
tions made by other brokerage firm analysts. “Days after IPO date” is the number of
days after the initial IPO date until the added-to-buy recommendation. T-statistics
are calculated using the cross-sectional variance in the excess returns and assume in-
dependence. The Z-statistic from the Wilcoxon rank-sum test compares the distribu-
tions of the underwriter and non-underwriter recommendations non-parametrically.

T-Statistic/
All By By Non- Z-Statistic of
Added-to-Buy Buy Recs Underwriter Underwriter the Difference
Recommendations N= 214 N= 112 N= 102 U vs. Non-U

Excess Return,
prior 30 days
Mean 1.2% −1.6% 4.1% 2.36
Median 0.7% −1.5% 3.5% 2.71

Excess Return,
3-day Event
Mean 3.5% 2.7% 4.4% 1.55
Median 2.5% 2.2% 2.8% 1.15
Days after IPO
date, Mean 83 66 102 2.60
Days after IPO
date, Median 50 47 63 3.48

Excess Return,
Event+ 3 mos.
Mean 7.8% 3.6% 12.5% 2.43
Median 6.3% 3.3% 8.0% 2.44

Excess Return,
Event+ 6 mos.
Mean 8.2% 3.2% 13.8% 1.69
Median 5.7% 3.9% 7.8% 1.58
Excess Return,
Event+12 mos.
Mean 3.5% −5.3% 13.1% 2.29
Median −5.1% −11.6% 3.5% 2.71

*Significant at 0.05 level.
Source: Michaely, Roni, Womack, 1999. Conflict of Interest and the Credibility of Under-
writer Analyst Recommendations. The Review of Financial StudiesSpecial 1999, 12 (4):
653–86.