Works in Progress
Works in Progress
For a dynamic labour market, where a firm and worker arriving in each period may be an efficient match or suboptimal match, we characterise the optimal policy that maximises expected social surplus. This policy trades off the benefit of waiting for a better match in the future with the cost of being unemployed. Our model shows that strictly positive expected unemployment is efficient for markets with intermediate discount factors and that there is no unemployment when the discount factor is low and unemployment converges to zero as the discount factor approaches one. Accounting for private information, we find conditions under which the dynamically efficient allocation can be implemented via a direct mechanism.
The adoption of artificial intelligence (AI) in the workplace has coincided with declining junior hiring. We propose an explanation based on the interaction between AI, training, and human-capital accumulation. We combine an overlapping-generations framework with a principal-agent problem in which senior workers choose how much effort to devote to training juniors, who subsequently become seniors themselves. This creates a trade-off between current production and future human-capital accumulation. AI makes observed outcomes a noisier signal of underlying human capital, with this distortion outweighing its productivity gains. This weakens training incentives and hence reduces the value of hiring juniors. Our mechanism suggests that AI may not only substitute for juniors’ tasks but also weaken firms' incentives to develop human-capital, thereby providing an explanation for the decline in junior hiring.
We consider a kidney exchange market of two-way exchanges that allow incompatible and compatible pairs to participate. Patients in incompatible pairs have dichotomous preferences, while patients in compatible pairs have trichotomous preferences. We study the model in which compatible pairs participating in the kidney exchange market are only matched with a donor with a strictly better quality than their own donor. The planner's objective is to maximise the overall number of matches. We use a graph theoretic approach to identify which compatible pairs are never required to be included in the exchange market and which compatible pairs can be included to help maximise the planner's objective. We characterise the minimum number of compatible pairs required to join the kidney exchange market to maximise the number of transplants. We aim to study which compatible pairs are essential in the kidney exchange market and which have substitute compatible pairs to maximise the number of transplants.
This paper analyses and compares the efficiency and equity of different assignment mechanisms in a theoretical set up, modelling dynamic assignment of objects to queuing agents. At each time period, a new object is offered according to the assignment mechanism and agents choose whether to accept or reject the object. Applications of this model include waiting lists for elective surgeries at hospitals, social housing and schools. Under the queued to the back mechanism that I propose, the punishment upon rejecting the object, induces less selective behaviour by agents in the queue. Generally, the queued to the back mechanism achieves the desired balance between the first come first served and lottery mechanisms. The benefit of the queued to the back mechanism is that compared to mechanisms in the existing literature, the queued to the back mechanism can be applied to scenarios in which there is equal weight of importance on the characteristics: utility, fairness, misallocation and waste.