Treatment Effects In Settings Research Articles

Long‐Run Effects of Dynamically Assigned Treatments: A New Methodology and an Evaluation of Training Effects on Earnings

We propose and implement a new method to estimate treatment effects in settings where individuals need to be in a certain state (e.g., unemployment) to be eligible for a treatment, treatments may commence at different points in time, and the outcome of interest is realized after the individual left the initial state. An example concerns the effect of training on earnings in subsequent employment. Any evaluation needs to take into account that some of those who are not trained at a certain time in unemployment will leave unemployment before training while others will be trained later. We are interested in effects of the treatment at a certain elapsed duration compared to “no treatment at any subsequent duration.” We prove identification under unconfoundedness and propose inverse probability weighting estimators. A key feature is that weights given to outcome observations of nontreated depend on the remaining time in the initial state. We study effects of a training program for unemployed workers in Sweden. Estimates are positive and sizeable, exceeding those obtained with common static methods. This calls for a reappraisal of training as a tool to bring unemployed back to work.

Estimating population average treatment effects from experiments with noncompliance

Abstract Randomized control trials (RCTs) are the gold standard for estimating causal effects, but often use samples that are non-representative of the actual population of interest. We propose a reweighting method for estimating population average treatment effects in settings with noncompliance. Simulations show the proposed compliance-adjusted population estimator outperforms its unadjusted counterpart when compliance is relatively low and can be predicted by observed covariates. We apply the method to evaluate the effect of Medicaid coverage on health care use for a target population of adults who may benefit from expansions to the Medicaid program. We draw RCT data from the Oregon Health Insurance Experiment, where less than one-third of those randomly selected to receive Medicaid benefits actually enrolled.

L2-Boosting for Economic Applications

We present the L2Boosting algorithm and two variants, namely post-Boosting and orthogonal Boosting. Building on results in Ye and Spindler (2016), we demonstrate how boosting can be used for estimation and inference of low-dimensional treatment effects. In particular, we consider estimation of a treatment effect in a setting with very many controls and in a setting with very many instruments. We provide simulations and analyze two real applications. We compare the results with Lasso and find that boosting performs quite well. This encourages further use of boosting for estimation of treatment effects in high-dimensional settings.

Variable selection for propensity score models when estimating treatment effects on multiple outcomes: a simulation study

It is often preferable to simplify the estimation of treatment effects on multiple outcomes by using a single propensity score (PS) model. Variable selection in PS models impacts the efficiency and validity of treatment effects. However, the impact of different variable selection strategies on the estimated treatment effects in settings involving multiple outcomes is not well understood. The authors use simulations to evaluate the impact of different variable selection strategies on the bias and precision of effect estimates to provide insight into the performance of various PS models in settings with multiple outcomes. Simulated studies consisted of dichotomous treatment, two Poisson outcomes, and eight standard-normal covariates. Covariates were selected for the PS models based on their effects on treatment, a specific outcome, or both outcomes. The PSs were implemented using stratification, matching, and weighting (inverse probability treatment weighting). PS models including only covariates affecting a specific outcome (outcome-specific models) resulted in the most efficient effect estimates. The PS model that only included covariates affecting either outcome (generic-outcome model) performed best among the models that simultaneously controlled measured confounding for both outcomes. Similar patterns were observed over the range of parameter values assessed and all PS implementation methods. A single, generic-outcome model performed well compared with separate outcome-specific models in most scenarios considered. The results emphasize the benefit of using prior knowledge to identify covariates that affect the outcome when constructing PS models and support the potential to use a single, generic-outcome PS model when multiple outcomes are being examined.