Why do firms decide to stop their share repurchase programs?

Abstract

We explore the distinguishing characteristics of firms that completed or ended share repurchase programs. Our findings help further understanding of the economic reasons for cancelling such programs. Based on a U.S. sample of 457 completed and 79 non-completed repurchase programs, we find a significant drop in systematic risk around completed buybacks. This suggests a response to deteriorating investment opportunities. In contrast, the systematic risk of non-completers decreases prior to the announcement, followed by an increase that peaks during the event period. This suggests that firms cancel their repurchase intentions when growth options move into the money.

Appendix: Linear approximation

Because the number of calendar days between the announcement and completion or cancellation differs for every share repurchase, we apply Malmendier et al.’s (2016) linear interpolation to normalize this time interval to $T_{S} = 250$ synthetic trading days. Specifically, we calculate normalized equity betas, $\widehat{\beta }^{E}_{i,t_{s}}$, and normalized cumulative abnormal returns, $\widehat{CAR}_{i,t_{s}}$, for company i for every synthetic trading day $t_{s}$ on the basis of realized equity betas, $\beta ^{E}_{i,t}$, and realized cumulative abnormal returns, $CAR_{i,t}$. We then calculate realized betas and CARs for the empirical event window, $T_{i}$, beginning at the announcement day, $T_{A}$, and ending with the completion or cancellation day, $T_{C/W}$. For each firm, we either condense realized beta values to our synthetic time interval if $T_{i} \ge T_{S}$, or we extend them to $T_{S}$ if the empirical event window has less than 250 calendar days. As per Malmendier et al. (2016), we assign the event-specific variable $\beta ^{E}_{i,t}$ to every synthetic trading day after ${t}_{s}=t_{s}\frac{T_{i}}{T_{S}}$ days:

$$\begin{aligned} \widehat{\beta }^{E}_{i,t_{s}} = \beta _{i,{t}_{s}}. \end{aligned}$$

(2)

For example, if the cancellation announcement occurs 500 days after the initiation announcement, i.e., $T_{i}=500$, then the normalized equity beta after 50 synthetic trading days, $\widehat{\beta }^{E}_{i,50}$, will equal the realized equity beta after 100 empirical trading days, $\beta ^{E}_{i,100}$, because $\widehat{t}_{s}=50\frac{500}{250}=100$. However, the index that identifies the respective real trading day may not always be an integer. In that case, as per Malmendier et al. (2016), we use the following linear interpolation:

$$\begin{aligned} \widehat{\beta }^{E}_{i,t_{s}} =(1-\gamma _{i,t_{s}})\cdot \beta ^{E}_{i,\lfloor {t}_{s} \rfloor } + \gamma _{i,t_{s}}\cdot \beta ^{E}_{i,\lfloor {t}_{s} \rfloor +1} \end{aligned}$$

(3)

where $\gamma _{i,t_{s}}={t}_{s}-\lfloor {t}_{s} \rfloor$, and $\lfloor {t}_{s} \rfloor$ refer to the largest previous real number.

Thus, for a share repurchase completed 150 days after initiation, i.e., $T_{i}=150$, the normalized equity beta after 12 synthetic trading days will be a weighted average of realized equity betas after seven and eight empirical trading days, ${\widehat\beta }^{E}_{i,12}=0.8\cdot \beta ^{E}_{i,7}+0.2\cdot\beta ^{E}_{i,8}$, because ${t_{s}} = t_{s} \frac{T_{j}}{T_{S}}= 7.2$, and $\gamma _{i,12}= 7.2- \lfloor 7.2 \rfloor = 0.2$.