This paper uses daily data on household expectations to examine what causes households to adjust their expectations about the future of the economy. We analyze several macro variables of policy interest and find that households respond primarily to movements in the unemployment rate. Further, these responses are non-linear and asymmetric, with households displaying higher sensitivity to larger shocks and to negative information indicating a worsening of the economy. We also find heterogeneity across local labor markets: Households in areas with higher local unemployment are more sensitive to changes in national unemployment than those in areas with lower local unemployment. We further examine whether media plays a role in influencing household expectations, and find that news about unemployment rises sharply during a recession, consistent with the response of expectations.
Work in Progress
Labor Force Participation in the 21st Century
The labor force participation rate in the United States has exhibited a sharp drop after the 2008 recession. Using the Current Population Survey, this paper first disentangles the trends in the participation rate along various demographic groups using shift-share analysis. Next, using time series, cross-section and individual variation, it examines the determinants of an individual's participation decision. Finally, it concludes with a discussion of the plausibility of various theories explaining the decline in participation from a labor supply lens.
In developing countries, weak enforcement of labor and tax regulations creates an opportunity for employers to collude with employees and manipulate the unemployment insurance system. In particular, workers who are eligible for benefits have an incentive to agree with their employers to transition into informality and share the proceeds from unemployment checks. We study the importance of this mechanism for the design of optimal unemployment insurance policies in a quantitative Diamond-Mortensen-Pissarides model calibrated with Brazilian data.