Endogenous assimilation and immigrant adjustment in longitudinal data

We create a longitudinal data set by matching immigrants in Israel's censuses for 1983 and 1995. These panel data reject the Immigrant Assimilation Hypothesis (IAH), which predicts that immigrants with shorter durations in 1983 should have experienced faster earnings growth between 1983 and 1995. By contrast, IAH is corroborated by the synthetic cohort methodology (SCM) over the same period. We suggest that SCM is subject to survivor bias, which increases the apparent degree of assimilation. We show that since the return to destination-specific skills increased during this period because of the very large immigration, the assimilation curve changed its shape in a way that made it difficult to estimate even using panel data.