Identifying Causal Effects in Information Provision Experiments

TL;DR

This paper reveals that standard estimators in information experiments underestimate the true causal effects of beliefs on outcomes and proposes a new estimator to better capture these effects, reanalyzing existing studies.

Contribution

It introduces a local least squares (LLS) estimator that recovers unbiased average effects in belief-outcome studies, improving upon traditional methods.

Findings

Reanalysis shows increased estimates of causal effects in five studies.

In two studies, effects more than double after reanalysis.

Standard estimators understate the true causal effects.

Abstract

Standard estimators in information provision experiments place more weight on individuals who update their beliefs more in response to new information. This paper shows that, in practice, these individuals who update the most have the weakest causal effects of beliefs on outcomes. Standard estimators therefore understate these causal effects. I propose an alternative local least squares (LLS) estimator that recovers a representative unweighted average effect in a broad class of learning rate models that generalize Bayesian updating. I reanalyze six published studies. In five, estimates of the causal effects of beliefs on outcomes increase; in two, they more than double.