A risk averse buyer and seller contract over the trade of an item. At the time of trading they each privately know their value s and cost r respectively, but these are not known when the contract is drawn up. The contract specifies a Bayesian revelation mechanism for implementing a trading rule and prices, as functions of their types s and r. An optimal (second-best) contract balances the goal of efficient trading and risk sharing against the need to provide the agents with incentives to reveal their type truthfully. An optimal contract is characterized. First, it is efficient to have some insurance from a third party: even though neither the buyer nor the seller will be fully insured, there is no need for the buyer to be exposed to the seller's risk or vice versa. Second, there will be less than first-best trade (underproduction). Third, once they have privately learnt their type—but before they have played the mechanism—both the buyer and the seller prefer that the final outcome will be trade rather than no trade. Fourth, although the trade prices increase with s and r, the rest of the contract need not be monotonic. This lack of monotonicity means that the standard methodology fails: it is not enough simply to appeal to local incentive compatibility, since global incentive constraints may bind. A nonstandard technique has to be used to find the nature of the second-best distortions.

If a firm does not know the individual ex post outside opportunities of its contracted workforce, then it can use hours, wages, and redundancy payments to screen them. Will a second-best contract have underemployment inefficiency, or overemployment? And, with a stochastic contract, are those workers randomly selected for layoff worse off than their retained colleagues (involuntary layoff), or vice versa (involuntary retention)? The answers depend on the nature of the workers' preferences. Two polar cases are looked at, corresponding to permanent and temporary layoff. The former is characterized by overemployment and involuntary retention; the latter by underemployment and involuntary layoff. Also examined are “simple contracts” where all retained workers are paid a common wage and all laid-off workers receive a common redundancy payment. Handling asymmetric information in stochastic contracts has led to two technical innovations. First, a number of results are proved even though all the truth-telling constraints are explicitly included. Second, a new regularity condition is found under which the local constraints are sufficient to ensure global incentive compatibility at an optimum.

The paper characterizes the choice rules that can be implemented when agents are unable to commit themselves not to renegotiate the mechanism.

The authors extend E. Maskin's results on Nash implementation. First, they establish a condition that is both necessary and sufficient for Nash implementability if there are three or more agents (the case covered by Maskin's sufficiency result). Second--and more important--they examine the two-agent case (for which there existed no general sufficiency results). The two-agent model is the leading case for applications to contracting and bargaining. For this case, too, they establish a condition that is both necessary and sufficient. The authors use their theorems to derive simpler sufficiency conditions that are applicable in a wide variety of economic environments. Copyright 1990 by The Econometric Society.

This paper examines the use of stage mechanisms in implementation problems and provides a partial characterization of the set of subgam e perfect implementable choice rules. It is shown that, in many economic environments, virtually an y choice rule can be implemented. To illustrate the power of this approach, the paper discusses a number of models in which it is possible to implement the first-best (although it wouldn't have been possible to do so without using stage mechanisms). The diversity of these models suggests that subgame perfect implementation may find wide application. Copyright 1988 by The Econometric Society.

We address the monopoly problem of designing and pricing a product line of goods distinguished by different quality and warranty levels. Consumers vary in their evaluations of these attributes, so that the problem is one of screening. It is sufficiently complex that the local approach commonly used does not work. Instead, we use new techniques for dealing with incentive constraints between nonadjacent consumer types. These techniques allow us to characterize optimal allocations that may not be monotonic. In particular, although the more eager types of buyer do pay higher prices and yield the monopoly higher profit, they may receive lower quality or lower warranty coverage. We find preference restrictions that restore monotonicity: concave risk tolerance implies that warranty coverage increases in type, and constant absolute risk aversion implies that quality increases in type.

Japan has been in a slump for the past decade. After GDP had been growing by on average 4 percent during the 1980’s, the growth rate dropped to 1 percent in the 1990’s. Asset prices also fluctuated significantly: capital gains on stocks and real estate in the 1980’s, followed by capital losses in the 1990’s, were both on the order of a few years’ worth of GDP, even after taking inflation into account. Together with production and asset prices, the fraction of nonperforming loans fluctuated substantially. These are by no means all bank loans. For the nonfinancial corporate sector in Japan, the ratio of financial assets to total assets is about 40 percent, much higher than in the United States. Such financial assets include loans to and securities of other private agents. That is, nonfinancial institutions simultaneously borrow from and lend to each other on a significant scale. Many nonperforming loans are interlocked, paralyzing the financial system. It is important to recognize that these swings have been experienced by almost all sectors of the Japanese economy. Yet in other countries, comparable movements in asset prices have had less widespread consequences. For example, the recent fluctuations in the NASDAQ index in the United States have been no smaller than those of asset prices in Japan, but the damage appears to be contained to closely related sectors. Although U.S. equity-holders, particularly pension funds, have lost value, the level of nonperforming loans is relatively limited up to now. The question is: Why does there appear to be more contagion in some countries than in others? Has contagion anything to do with the nature of financing or the extent to which there are inter-locking loans? In this theoretical paper, we examine two different mechanisms by which contagion may occur. In both cases, propagation is through balancesheet effects. First, through the indirect effects that fluctuations in asset prices have on collateral values. Second, through the direct effects that default on or postponement of debt repayments have when there are chains of credit.

Plenary talk delivered at the annual meeting of the Society for Economic Dynamics.(This abstract was borrowed from another version of this item.)