Using repeated cross-sectional data to examine the role of immigrant birth-country networks on unemployment duration: an application of Guell and Hu (2006) approach

Guell and Hu (J Econom 133:307–341, 2006) propose a robust econometric estimator using repeated cross-sectional data in the context of duration/survival analysis where otherwise limited or no results are possible because reliable panel data at an individual level are missing. We present a detailed exposition of the Guell and Hu (GH) strategy and show a Monte Carlo simulation for unemployment duration where the GH method produces better estimates compared to panel data with individual matching errors. We further apply the GH model to examine the immigrant unemployment duration in the USA using the Current Population Survey data from 2001 to 2013 and focus on the role of the birth-country networks on the unemployment duration around the Great Recession. We find that birth-country networks measured at the state level significantly lower unemployment duration for all immigrants, and this effect is stronger during the pre- and post-recession periods than during the recession. We also find that networks are more effective in lowering duration for immigrants unemployed for 1–2 months than for immigrants who are unemployed for longer periods, and this effect is the strongest during the post-recession period. The findings are robust to different specifications and measures of networks including those measured at the local MSA level.