the unemployment period. If the person was not unemployed but experienced the same or longer period of work compared with the period of benefit receipt, then months of receipt are allocated to fall within the work period. If the person both worked and was unemployed and the period of benefit receipt exceeded that of unemployment, then months of receipt are assigned first to months of unemployment and then to months of work directly preceding and/or directly following the spell of unemployment.
ALLOY next allocates asset income, which is summed and divided evenly over all months of the year (net rental income is constrained to be greater than or equal to zero). The sum of all other sources of unearned income is allocated by first assigning a randomly determined duration. There are three distributions by month of cumulative duration probabilities. One is for people age 60 and older, for whom the distribution is a relatively smooth convex curve. Another is for people under age 60 who report receiving income from a “regular” unearned income source (social security/railroad retirement, private and government pensions, veterans’ benefits, and workers’ compensation). This distribution has values only for months 10 through 12. The third distribution is for people under age 60 who have zero income from regular unearned sources and positive income from all other unearned income. This distribution is a relatively smooth concave curve. A random draw between zero and 1 is compared with the appropriate distribution, and the duration in months is chosen accordingly. Then a starting month of receipt for unearned income is chosen and the months of receipt are allocated, wrapping around the year as necessary. The average monthly amount (the sum of regular and irregular unearned income sources divided by the simulated months of receipt) is then computed and assigned to each month of receipt.
HITSM
This model distributes unemployment compensation evenly over the weeks of simulated unemployment. The number of weeks of receipt is restricted so that it does not exceed the limit for the state in which the individual lives. The model distributes workers’ compensation evenly over each week in which the individual is simulated as out of the labor force, and it distributes income from all other sources evenly over each of the 52 weeks of the year. Weekly incomes are then summed to form monthly figures, where each month is assumed to include 4.33 weeks.
Comment
MATH has by far the richer procedures for assigning unearned income by months compared with the other two models.
