The Job You Find Is Part of the Mechanism

A reemployment-wage estimate is an endpoint, not an explanation. This chapter examines the IZA paper’s decomposition of that endpoint into worker, firm, job, and residual components, then uses its no-duration-response comparison to ask how wages can change when unemployment duration does not. The practical judgment is that a benefit extension may affect the destination firm or the wage outcome directly, so treating duration as the sole pathway to wages is not justified.

Four pieces of a reemployment wage

The IZA paper estimates a linked employer–employee wage model that separates several sources of pay. A worker fixed effect represents persistent worker characteristics that influence wages over time. A firm fixed effect captures how much a firm tends to pay compared with other firms for workers with similar observed and persistent characteristics. A job fixed effect captures the particular worker–firm match: complementarities, bargaining conditions, or other features that are specific to that pairing. A residual component collects time-varying factors that the model does not assign to the worker, firm, or match.

These components give the phrase “a better job” more structure. If a worker’s wage changes because the worker component changes, an individual-level explanation might be plausible. If it changes because the firm component changes, the worker may have moved into a firm that pays more or less to many comparable employees. If the job component changes, the outcome may reflect the fit or bargaining conditions of one particular match. A residual change is evidence that something moved, but it gives less guidance about which mechanism was responsible.

The decomposition is especially useful because the components behave differently under a job transition. A worker fixed effect is intended to capture characteristics that persist across employment, so it is not supposed to change simply because potential benefit duration changed at the separation point. A firm effect can change when the worker enters a different employer, even if the worker’s own characteristics do not. A job effect can change when the worker and firm form a different pairing. The model therefore turns a single wage comparison into a question about which part of the employment relationship moved.

That distinction does not make the components pure. Firm and job effects are estimated from linked employment histories, and the interpretation depends on mobility and model assumptions. A lower firm effect can be consistent with sorting into a lower-paying employer, but it does not reveal whether the worker chose that firm, was selected by it, or accepted it because alternatives were limited. The decomposition narrows the plausible location of the change; it does not observe the negotiation that produced it.