Efficient targeting in childhood interventions

Abstract

Using Danish administrative data, we address the issue of efficient targeting in childhood interventions. We define children to be in need of an intervention if they experience one or more socially undesirable outcomes in adulthood. Because interventions are very effective early in life, we then test if and to what extent indicators available at birth can predict these outcomes. We find fair to good levels of prediction accuracy for many outcomes, driven by a parsimonious set of predictors. We show that optimal weights for the construction of risk scores deviate from the weights typically used in targeted interventions.