Housing is in a relatively deep recession, yet the decline in residential construction employment (not including specialty contractors) has been relatively mild (see Chart 1).

Chart 1

One hypothesis for the relatively mild contraction in residential construction employment is that a lot of these jobs have been filled by undocumented workers in this cycle. If undocumented, these workers would not be counted when they were hired or when they were fired.

Is there any corroborating evidence that undocumented workers increasingly have been employed in the residential construction industry? Yes, the behavior of currency. Chart 2 shows a recent sharp decline in currency held by the public as a percent of the total M2 money supply. The decline in M2's relative currency component began shortly after the peak in this cycle's housing starts. Not only would undocumented workers likely not be counted in the establishment payroll statistics but they also would likely be paid in currency, not direct deposits to bank accounts. The absolute and relative slowing in U.S. currency growth in recent years is consistent with the hypothesis that there have been large layoffs of undocumented workers. In turn, this would help explain the relatively mild contraction in documented residential construction employment.

Chart 2