Yifan Wang

[Hide]

Current productivity and organization show how a firm produces today, not how it could produce after a new technology arrives. This paper asks which observations determine the gain from reorganizing. A firm divides tasks among roles, trading off the cost of tasks far from a worker’s training against handoffs between roles. Current organization and cost pin down the level and slope of the firm’s cost curve but not its curvature, so they do not determine the gain. Demand variation reveals the firm’s menu of organizations, but only its upper edge: a complete history determines the gain if and only if the technology does not lift an organization never chosen above that edge. In general equilibrium, consumption with organizations held fixed is identified and the reorganization gain is bounded. In a preregistered blind test on simulated economies, forecasts built on these results nearly halve the error in predicted reorganization.

with Chang Gao  ·  [Hide]

LGFV issuance spreads rise with local government leverage — under OLS, under a predetermined maturity-rollover instrument, and under the 2015 central-government debt swap. Housing-price growth is hump-shaped in leverage and turns negative past 18–23% of GDP. We read these patterns through a general-equilibrium model in which a leverage-dependent financial friction creates a debt threshold beyond which infrastructure investment collapses, so the bust needs no exogenous shock.

[Slides]  [Hide]

Some desirable jobs are not priced. They are rationed by quota and handed out by examination: in 2021 China’s civil service took about 20,000 people from 1.5 million applicants. This paper models entry into such a sector alongside wage-clearing sectors and an outside option. Admission rises with an applicant’s rank inside the pool of people who chose to apply, so the admission rate is an equilibrium object rather than a parameter. Neither instrument is monotone. A more generous outside option at first enlarges the queue, since it makes failing cheaper to risk; past a threshold it is simply taken instead. A larger prize at first improves selection and then dilutes it, as the marginal applicants arrive from further down the ability distribution. The two humps have different engines: one works on the size of the queue and survives with the admission schedule held fixed, the other on who is in the pool. Because both instruments move that pool, they cannot be set apart from each other.