We study identification and estimation of causal effects in settings with panel data. Traditionally researchers follow …

With panel data important issues can be resolved that can not beaddressed with cross--sectional data. A major drawback … from the original populationas replacements for units who have dropped out of the panel, aso--called refreshment sample … different results than models that assume that themissing data process is ignorable or conventional econometric modelsfor panel …