THREE ESSAYSIN THEORETICAL CORPORATE FINANCE

The thesis is articulated around three theoretical corporate finance research articles. All articles model intertwined and important corporate decisions and have in common the modeling of cash policy: Each article presents a new cash model. In the first article, I examine managers’ decision to pay dividends with a dynamic cash model. The model departs from the Modigliani and Miller world according to two dimensions: Agency costs and financing frictions. In this paper, I show that managers with more valuable outside options pay more dividends, but the outside option makes extremely costly for shareholders to optimally compensate managers. The model is solved by computing the opti- mal firms’ ownership which should be granted to managers and by computing the optimal payout and liquidation policy decided by managers. In the second article, I examine the impact of managerial optimism on corpo- rate policies and agency costs within a model of dynamic corporate investment for a financially constrained firm. An optimistic manager is defined as a man- ager who over-evaluates the profitability of the firm assets. The main feature of the model is to allow joint investment and saving policies and the model is solved numerically. In addition to the common sense that an optimistic manager should build an empire by investing more in average and keeping more cash under control, I find the following primary results: when the firm is highly financially constrained, an optimistic manager under-invests and can oversell the physical assets. The investment sensitivity to cash is lower. The agency costs sensitivity to cash is higher and this reinforces the need to monitor the manager. In the third article, I examine the effects of credit rationing on corporate cash holdings by modeling the precautionary demand for cash. In the model firms can pledge part of the future cash-flows to creditors when current cash- flows are insufficient to finance investment. The discrete time model with three periods is solved with closed form solutions. I show that the cash-flow sensitivity of cash and the investment to cash sensi- tivity are inappropriate indicators of financing constraint. By contrast, I show that the variation of cash holdings is monotonically decreasing with the degree of the financing constraint. An empirical study with a large sample of manu- facturing firms over the 1971 to 2011 period confirms this result. The challenge in theoretical corporate finance is to set up models that per- mit to highlight unexplored problems and trade-offs or to solve well known puz- zles. Each model proposed in this thesis has for objective to solve a puzzle or to highlight new important trade-offs. In the first paper I highlight a new trade-off between managers outside option and payout policy. In the second paper I highlight that optimistic managers can under invest and in the third I show that a financially constrained firms can optimally reduce liquidities even with future profitable investments. In general it is challenging to propose models rich enough to get non trivial, but simple enough solutions. A good model prediction should be testable empir- ically and easy to communicate. I hope that the directions of research proposed and the results found in this thesis could be useful for empirical researchers. It was an important objective. The context of this research is rather simple. Over the past two decades, dy- namic corporate finance models have taken an important part of the literature in financial economics. In particular cash models have been extensively used to provide insights and quantitative guidance for investment, financing or risk management decisions under uncertainty. Cash models are particularly useful to model intertwined decisions, because the cash is an asset used as a buffer for a financially constrained firm. This permits to analyze both cash-inflows and outflows in a natural way. The models presented in this thesis belong to this agenda. The next step and important perspective would be to use the trade-offs iden- tified to set up structural estimations. Recursive models have been extensively used to perform structural estimations so far and could be maybe adapted from the theoretical models presented in this thesis. However it is not clear whether fix points techniques could be used to solve them. The obvious improvement could be due to significantly more powerfully com- puters. It would permit the calibration of recursive models with more state variables