This paper describes the methods the Congressional Budget Office (CBO) uses to project individual earnings in its long-term microsimulation model (known as the CBOLT model). The CBOLT model is used primarily to assess the fiscal situation of the Social Security system, as well as the Medicare and Medicaid programs. One key difference of the CBOLT model from most other models that project Social Security finances is that CBOLT models behavior at the individual level. That is, for each individual in the model, CBOLT projects levels of educational attainment, transitions in and out of marriage, labor force and unemployment transitions, immigration and emigration behavior, and Social Security benefit claiming patterns. Importantly, CBOLT models how individual earnings patterns change over each worker’s lifetime; those lifetime earnings patterns are then the key component underlying individual Social Security benefit receipt, payroll taxes paid, and thus aggregate Social Security finances.

The methodological strategy used in this paper closely follows the method first documented by Carroll (1992), but is also informed by the work of a number of different researchers. In general, a pair of earnings “shocks” is estimated that is used to perturb observed individual earnings. The first “shock” is permanent in nature and measures the long-run gap between an individual person’s earnings compared with the overall average of that person’s group (where the group may be defined by, for example, age, sex, and education). Examples of permanent shocks include a promotion or attaining higher levels of educational attainment. The second “shock” is transitory in nature and measures any additional but temporary variation in a person’s earnings. Examples of transitory shocks include an unusually large annual bonus payment, a temporary reduction in work hours, or some sort of temporary health status change (e.g., broken leg).