【正文】 of action in order to in?uence the behavior of the other party. Then once that party’s behavior has been determined, the bene?t of the mitment disappears and there is now an incentive to depart from agreements have been entered into are subject to revision because both parties can typically be made better o? by “renegotiating” the original agreement. The possibility of renegotiation puts additional restrictions on the kind of contract or agreement that is feasible (we are referring here to the contract or agreement as executed, rather than the contract as originally written or conceived) and, to that extent, tends to reduce the welfare of both parties ex ante. Anything that gives the parties a greater power to mit themselves to the terms of the contract will, conversely, be welfareenhancing. Dewatripont and Maskin (1995) (included as a chapter in this section) have suggested that ?nancial markets have an advantage over ?nancial intermediaries in maintaining mitments to refuse further funding. If the ?rm obtains its funding from the bond market, then, in the event that it needs additional investment, it will have to go back to the bond market. Because the bonds are widely held, however, the ?rm will ?nd it di?cult to renegotiate with the bond holders. Apart from the transaction costs involved in negotiating with a large number of bond holders, there is a freerider problem. Each bond holder would like to maintain his original claim over the returns to the project, while allowing the others to renegotiate their claims in order to ?nance the additional investment. The freerider problem, which is often thought of as the curse of cooperative enterprises, turns out to be a virtue in disguise when it es to maintaining mitments. From a theoretical point of view, there are many ways of maintaining a mitment. Financial institutions may develop a valuable reputation for maintaining mitments. In any one case, it is worth incurring the small cost of a suboptimal action in order to maintain the value of the reputation. Inplete information about the borrower’s type may lead to a similar oute. If default causes the institution to change its beliefs about the defaulter’s type, then it may be optimal to refuse to deal with a ?rm after it has defaulted. Institutional strategies such as delegating decisions to agents who are given no discretion to renegotiate may also be an e?ective mitment device. Several authors have argued that, under certain circumstances, renegotiation is welfareimproving. In that case, the DewatripontMaskin argument is turned on its head. Intermediaries that establish longterm relationships with clients may have an advantage over ?nancial markets precisely because it is easier for them to renegotiate contracts. The crucial assumption is that contracts are inplete. Because of the high transaction costs of writing plete contracts, some potentially Paretoimproving contingencies are left out of contracts and securities. This inpleteness of contracts may make renegotiation desirable. The missing contingencies can be replaced by contract adjustments that are negotiated by the parties ex post, after they observe the realization of variables on which the contingencies would have been based. The inplete contract determines the status quo for the ex post bargaining game (., renegotiation)that determines the ?nal oute. An important question in this whole area is “How important are these relationships empirically?” Here there does not seem to be a lot of evidence. As far as the importance of renegotiation in the sense of Dewatripont and Maskin (1995), the work of Asquith, Gertner and Scharfstein (1994) suggests that little renegotiation occurs in the case of ?nancially distressed wisdom holds that banks are so well secured that they can and do “pull the plug” as soon as a borrower bees distressed, leaving the unsecured creditors and other claimants holding the bag. Petersen and Rajan (1994) suggest that ?rms that have a longer relationship with a bank do have greater access to credit, controlling for a number of features of the borrowers’ history. It is not clear from their work exactly wha