Chi Man Yip
- "A Robust Mortensen and Pissarides Model" (with Ying Tung Chan)
Abstract: The present paper generalizes the Mortensen and Pissarides (1994) model by allowing incomplete information about the data generating process in the economy. An agent could use an approximating model but at the same time worry about model misspecification. Using robust control techniques as in Hansen and Sargent (2008), a robust agent optimally chooses his model, according to his degree of ambiguity aversion, to make decisions on job search, job destruction and vacancy creation. Our robust model not only provides analytically tractable framework, but also provides additional insight on how ambiguity aversion influences workers' and vacancies' decision, and preserves most of the generally intuitive comparative statics as in the original model even under workers' and vacancies' ambiguity aversion. Recent literature criticizes the Mortensen and Pissarides (1994) model in its inability to generate the volatility of market tightness in the United States. Our generalized model succeeds in replicating a similar volatility of the market tightness as in the US data, resolving the unemployment volatility puzzle. Our calibration exercise supports that incorporating ambiguity aversion is a substantial improvement over standard search and matching models.
- "On the Ambiguity of Job Search" (with Ying Tung Chan)
Abstract: Unemployed workers and unfilled vacancies confront uncertainty about the distribution of match specific productivity, and are averse to this ambiguity. However, prior search and matching models assume either complete information on the distribution or ambiguity neutrality. This paper constructs a search and matching model that features ambiguity aversion using the recursive specifications of Hansen and Sargent (2008). The model predicts that a robust agent tends to believe that higher match specific productivity is less likely to be realized, causing his/her outside option value and thus the reservation wage to fall. Hence, an unemployment rate is lower under a higher degree of ambiguity aversion of workers. We propose a procedure to compute the ''ambiguous'' unemployment rate, which is defined as the change in the unemployment rate if all agents became ambiguity neutral. Our simulation exercise uncovers quarterly ambiguous unemployment rate during 1948-2007 in US. We find that ambiguous unemployment falls with real output. When frictional unemployment were below 5%, ambiguous unemployment were close to zero. But the ambiguous unemployment could be as large as 19% of frictional unemployment in slump. In addition, our results indicate that ambiguity aversion amplifies the volatility of market tightness, potentially resolving the Shimer (2005) puzzle. Also, the impact of unemployment benefits on other labor market outcomes, such as market tightness, are shown to be larger under ambiguity aversion. This implication calls for a reexamination of the robust unemployment insurance scheme. Our result also shows that an efficient decentralized equilibrium is no longer guaranteed under the Hosios (1990) condition under ambiguity aversion. We generalize the Hosios (1990) condition in a search and matching framework under ambiguity aversion.