Mon, Jan 23, 2017 at noon:
H. Luke Shaefer
Although the availability of panel data has augmented our knowledge of poverty dynamics, the challenges posed by the presence of left-censored data remain. This analysis, using 1970 to 1985 data from the PSID, demonstrates that omitting left-censored observations, which is still the norm, could lead to serious selection bias in an analysis by systematically discarding individuals in the midst of long-term poverty spells and from certain demographic groups. Although there is still no single accepted way of dealing with left-censoring, several methods that may mitigate the problem are discussed.