Mechanism design with costly verification and limited punishment 

Economics

LI Yunan 

Published in Journal of Economic Theory, January 2020

A principal (e.g., the head of personnel for an organization) has an object (e.g., a job) to allocate among a finite number of agents (e.g., job applicants). Each agent values receiving the object and has private information about the principal's payoff from assigning the object to him. The object is allocated based on the agents' reports, and additional information about the agents arrives only after the object has been allocated for a set period of time. For example, each applicant has private information about their ability. The head of personnel may need to choose one job applicant for the job by reviewing their resumes and conducting interviews. The head of personnel can then monitor an employee's performance and fire them after the probationary period but cannot ask them to repay their salary or other employment benefits. 

How does the principal maximize her expected payoff in such an environment? New research by Dr Yunan Li at the Department of Economics and Finance shows that a simple procedure can work well.  

“When there are many agents, the object can be optimally allocated via a shortlisting procedure,” says Li. 

“In the shortlisting procedure, agents whose reported values above a prespecified threshold are shortlisted. The principal randomly chooses one agent from the shortlist to receive the object and inspects him with certainty.”   

There are several other important economic environments that roughly correspond to this model. A development aid agency may need to allocate a grant to one of several local organizations to serve a beneficiary group, and each local organization privately knows its management ability. The development aid agency can evaluate the recipient's management ability by reviewing its expenditure report and disbar it from future grant applications but cannot recover the funds already allocated to the organization. A venture capital firm may need to choose which startup to fund, and each startup has private information about its technology skill. The venture capital firm can audit the progress of a startup and cut off future funding in the event of unsatisfactory progress but often cannot ask the startup to return its initial investment.