Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Estimates of the poverty rate and of the probability of entering or exiting poverty are biased when income is observed with error. I estimate a variance components model of income which contains a white noise error term and then treat this component as an approximation of the error in observed income. By comparing poverty rates calculated with and without this estimated measurement error, I conclude that observation error causes the poverty rate to be overestimated around two percentage points on average. However, eliminating observation error substantially reduces the probability of transiting either into or out of poverty. These reductions imply that the amount of permanent poverty is underestimated when measurement error is ignored.