We study the trading dynamics in an asset market where the quality of assets is private information of the owner and finding a counterparty takes time. When trading of a financial asset ceases in equilibrium as a response to an adverse shock to asset quality, a large player can resurrect the market by buying up lemons which involves assuming financial losses. The equilibrium response to such a policy is intricate as it creates an announcement effect: a mere announcement of intervening at a later point in time can cause markets to function again. This effect leads to a gradual recovery in trading volume, where asset prices first decline when there is trade before the intervention, but then recover to their normal values. The design of the optimal policy is stark. When markets are deemed important and losses are small, it is optimal to intervene immediately as delaying involves fixed costs. As losses increase and the importance of the market declines, the intervention is optimally delayed and it can be optimal to rely more on the announcement effect by increasing the size of the intervention. Here it is never optimal to increase the purchase of lemons, but it can be optimal to increase the price at which lemons are bought. Search frictions are important for these results. They  compound adverse selection, making a market more fragile with respect to a classic lemons problem. They dampen the announcement effect, but make an intervention per se more powerful. They cause the optimal policy to be more aggressive, leading to an earlier intervention at a larger scale. Joint with Jonathan Chin.

Keywords: Adverse Selection, Search, Trading Dynamics, Intervention in Asset Markets, Announcement Effect

JEL Classification: G1, E6