What the Model Adds—and What It Cannot Prove

The evidence so far identifies several linked patterns, but it does not by itself say how much each mechanism contributes to the gendered job ladder. This chapter focuses on what the Chicago Fed paper's equilibrium model adds: a way to make workers' choices, firms' job design, and bargaining interact rather than treating each statistic as an isolated cause. The useful judgment is to read the model as a disciplined explanation of possible mechanisms, not as a direct measurement of every mechanism in the data.

Why description is not enough

A descriptive result can establish that two outcomes appear together. The study observes nearly equal job-to-job transition rates, lower conditional wage growth for women, different amenity gains, and different employment flows. It also observes survey differences in search and job preferences. But these facts interact. A lower wage can change the value of accepting an offer. A higher value placed on an amenity can change the occupation a worker chooses. A separation can change the worker's alternative to a new job. A wage offer can be shaped by the surplus available in the match and by the worker's bargaining position.

If we tried to assign each outcome to one statistic, we would risk counting the same mechanism twice or ignoring feedback between choices. The model supplies a common environment in which a worker chooses search effort, receives offers, decides whether to accept, and may separate from employment, while firms choose vacancies and amenity provision. It then asks what a stable labor market would look like when men and women can differ in amenity valuation, search, separations, nonemployment value, and bargaining power.

This is what economists mean by an equilibrium model. It is not a claim that the world is literally solved by an equation every month. It is a structured set of assumptions about how workers and firms respond to one another, combined with parameter values chosen to match selected facts. Once the pieces are connected, the model can produce counterfactuals: what would happen if one feature changed while the others were held to the model's rules? That question cannot be answered by a transition table alone.

The distinction between a target and an implication is central. The paper calibrates the model to match a log wage gap of −0.205, an employment gap of −0.125, and a zero gender gap in the job-to-job transition rate. These are targeted moments: the calibration is designed to reproduce them. The model then generates other results that were not used as the same direct targets, including a positive amenity gap of 0.062, a job-value gap of −0.142, a realized productivity gap of −0.085, a conditional wage-growth gap of −0.146, and an amenity-growth gap of 0.021. The gaps are defined as women's values minus men's values.

The signs are easier to understand than the notation.