This post comes to us from Yuhai Xuan
at Harvard Business School.
In my paper Empire-Building or Bridge-Building? Evidence from New CEOs’ Internal Capital Allocation Decisions, which was recently accepted for publication in the Review of Financial Studies, I examine CEOs’ decision-making processes for capital allocation in the context of power and relationships within corporations by investigating whether the job histories of CEOs influence their capital allocation decisions when they preside over multi-divisional firms. I investigate the capital allocation decisions made by 265 new CEOs at 230 diversified firms after turnovers between 1993 and 2002. CEO turnovers provide a good opportunity for this study because CEOs are likely to be most vulnerable to political complications at work when they are new to the post. In particular, I focus on the 98 new CEOs in my sample who advanced through the ranks from certain, but not all, divisions in their firms. I call these CEOs specialists and separate the segments in their firms into two groups based on their affiliation with the CEOs: divisions that the CEOs advanced through the ranks from (labeled the in-group), and the rest of the divisions (labeled the out-group). The empirical analysis in the paper focuses on changes in segment capital expenditures around CEO turnovers to determine whether specialist CEOs treat the in-group and the out-group segments differently when allocating capital after succession, and if so, whether they favor the in-group (“empire-building”) or the out-group (“bridge-building”) in their allocation decisions.
My results are broadly consistent with the bridge-building hypothesis. I find that, on average, the out-group segments experience a significant increase in capital expenditures after CEO turnover relative to the in-group segments. The average change in segment investment ratio (capital expenditures over assets) after a specialist CEO takes office is 0.013 higher for the out-group than the in-group, statistically significant at the 5% level or better. This difference of 0.013 is economically meaningful as it represents more than 20% of the average pre-turnover investment ratio of 0.06. Moreover, these findings also hold for specialist CEOs hired from outside the firm and are robust to the inclusion of segment-level, firm-level, and turnover-related controls as well as changes in the test specifications including the definition of specialists, the measure for capital expenditures, the time frame around turnover, and the sample period. I further test for the bridge-building hypothesis by examining whether the in-group and out-group difference in capital allocation change around turnover is related to the specialist CEO’s relative bargaining power within the firm. I find that the difference is more pronounced if the specialist CEO does not hold a corporate-level executive title such as chief operating officer or president before succession or if the in-group segments and the out-group segments are not in related industries. The results from the finer tests are consistent with the prediction of the bridge-building hypothesis that a specialist CEO with less power should engage in more bridge-building efforts, which imply a more pronounced pattern of reverse-favoritism in capital allocation.
While my results are consistent with the bridge-building hypothesis, a key concern is the issue of endogeneity. CEOs are chosen by the board of directors, and the job histories of CEOs are observable by the board and may be an important selection criterion in the board’s choice for nomination. Even though the most obvious and natural endogeneity story is one that would lead to a bias that works in precisely the opposite direction to the empirical findings in this paper, I consider alternative versions of the endogeneity story in which the CEO might be chosen to grow the segments in the out-group or to reduce investments in the in-group, leading to the relative increase in the capital expenditures of the out-group segments observed in the data. I use two approaches to address this concern. First, I try to discriminate against this type of endogeneity story by identifying weak divisions in the firm based on segment cash flow and segment Q. I find that the in-group and the out-group segments experience differential capital allocation change regardless of segment operating performance and segment investment opportunity. The difference in capital expenditure change is significant and of the same magnitude even when one compares the strong segments in the in-group with the weak segments in the out-group, inconsistent with what the endogeneity story might suggest. Second, I estimate a segment’s propensity to be a member of the out-group based on pre-turnover segment characteristics, and use the propensity scores as a summary measure to match the out-group segments and the in-group segments. Again, I find a relative increase in the average change in capital expenditures for the out-group compared with those of the in-group after a specialist CEO takes office. The magnitude and significance level of the estimate are similar to those of the main results, further alleviating the concern that endogeneity might account for the findings.
Finally, I investigate whether having a specialist CEO affects segment investment efficiency by studying the changes in the sensitivity of segment investment to Q before and after the CEO turnover. My results show that the sensitivity of segment investment to Q increases significantly after CEO turnover in a generalist’s firm, indicating an improvement in investment efficiency. Segments under a specialist CEO, however, do not experience such improvements: the investment sensitivity to Q for these segments is virtually unchanged after the turnover. In addition, I examine the market’s reaction to the announcement of the appointment of specialist versus generalist CEOs and find that the cumulative abnormal returns around announcements are significantly higher for incoming CEOs who are generalists. The market’s response corroborates the finding that generalist CEOs are associated with improved segment investment efficiency after turnover and suggests that appointments of generalist CEOs are perceived by the market as positive news for the conglomerates.
Overall, my results suggest that the job histories of CEOs are an important determinant of their capital allocation decisions and that new specialist CEOs are affected by political concerns in the capital allocation process. New specialist CEOs use the capital budget as a bridge-building tool to elicit cooperation from powerful divisional managers in previously unaffiliated divisions.
The full paper is available for download here.