For almost 30 years now, most empirical work in labor economics has relied on human-capital theory to drive statistical specifications. A parsimonious structure, the theory has proved itself capable of generating rich, testable implications. Humancapital theory is primarily a supply-side approach that focuses on the characteristics and skills of the individual workers. It pays far less attention to the environments in which workers work. As such, the humancapital framework has led researchers to focus on one class of questions, but to ignore others. Specifically, little attention has been paid to the jobs in which workers are employed. The entire notion of a job, which seems central to the thinking of businesspersons and administrators, is virtually absent from most labor-market analyses. Standard production theory treats labor as a homogeneous commodity and ignores the slots or positions in which that labor is used. The institutional literature, and especially the work on internal labor markets, asks questions that differ from those asked by human-capital theory. The research agenda discussed below returns to the questions posed by the old internal-labor-markets literature (see especially Melvin Reder [1955] and Peter Doeringer and Michael Piore [19711). Directing attention to jobs also changes the kinds of analytic techniques appropriate for empirical analysis. Human-capital theory has created a large quantitative literature that is based on a few basic empirical methods. A jobs-based theoretical structure must also guide empirical methods, but the suitable methods often differ from those best for answering human-capital-oriented questions.

There are two obvious possibilities that can account for the rise in productivity during recent recessions. The first is that the decline in the workforce was not random, and that the average worker was of higher quality during the recession than in the preceding period. The second is that each worker produced more while holding worker quality constant. We call the second effect, "making do with less," that is, getting more effort from fewer workers. Using data spanning June 2006 to May 2010 on individual worker productivity from a large firm, it is possible to measure the increase in productivity due to effort and sorting. For this firm, the second effect-that workers' effort increases-dominates the first effect-that the composition of the workforce differs over the business cycle.