FACULTÉ & RECHERCHE

How do firms allocate capital internally across units? Do more powerful and better connected managers inside a conglomerate get larger capital allocations? To answer these questions, we put together a unique five-year business-unit panel data set on planned and actual capital allocations inside a world-wide conglomerate with 5 divisions and 22 business units, and construct measures of managerial power and connections from profile data and a questionnaire carried out for unit CEOs. Our conglomerate seems to be exposed to the same kind of frictions in capital allocation and investment as documented in previous studies for other conglomerates. In the search for a potential channel behind such frictions, we test if cash windfalls at the headquarters are distributed inside the conglomerate according to managerial power or connections. In contrast to planned capital allocations, which are not affected by connections, we find that units with more powerful and better connected managers get substantially larger parts of the windfalls. Our estimates show powerful managers increasing their actual vs. planned investment between 10 and 32% more than their less powerful peers. These results are not explained by managers’ abilities or an endogenous allocation of managers across units. In support of bargaining power theories, our results provide direct empirical evidence of an important channel behind frictions in capital allocation inside firms