Combining a Survey Experiment with Lifecycle Models to Evaluate Pronatalist Policies

We develop a three-step procedure for ex-ante policy evaluation. It combines (i) a vignette survey that randomizes out-of-sample policies (child allowances and external childcare subsidies) to infer fertility responses; (ii) two applied-theory lifecycle models estimated on external microdata; and (iii) the inserting of the randomized policies into the models’ estimated fertility decision rules to obtain model-based fertility predictions for the vignette respondents, thus cross-validating these predictions with their fertility responses. These steps, applied to the United States and Germany, enable comparable out-of-sample policy evaluation across surveys and models, suggesting how to use vignettes as a compass for model specification.