The Working Poor--A Statistical Artifact?

We examine the effect of measurement error on estimates of the size of the working poor population. Using a unique data set, which includes both self-reported and employer-reported earnings, we find that inaccurately reported earnings are common. Among those with very low self-reported earnings are many individuals with underreported earnings. However, this is offset by a high incidence of over reporting among those who actually have low earnings. We find that, by fortunate coincidence, these counterbalancing sources of measurement error cancel each other out exactly. Estimates of the working poor population based on self-reported earnings are extremely accurate. This result is robust to changes in the designated poverty threshold.