Following up on yesterday's post on how little time unemployed people spend searching for a job, I claim that we need to model unemployment differently. We need to realize that there are some people who activily search, and others that consider an unemployment spell to be a vacation (or a temporary layoff).
This is the premise that Fernando Alvarez and Robert Shimer have in their latest paper. They extend the Lucas-Prescott island model to factor in work, search unemployment, rest unemployement and inactivity. The idea is that you search when you look for work in a different sector, while you rest when you just wait in your sector for things to improve. What this model points out is that rest unemployment is actually efficient: why waste resources searching when sectoral conditions will improve shortly?
One consequence of this is that unemployment insurance has an important role here: it allows unemployed workers to wait for the jobs to reopen and avoids forcing them to search. A somewhat similar argument has been made earlier by Daron Acemoglu and Robert Shimer, showing that unemployment insurance allows to wait and find better matches. But in this case, everybody searches all the time.