A Consistent Variance Estimator for 2SLS When Instruments Identify Different LATEs

TL;DR

This paper develops a consistent variance estimator for 2SLS in models with multiple instruments identifying different LATEs, correcting standard errors when traditional estimators are inconsistent due to heterogeneity.

Contribution

It introduces a new asymptotic variance estimator for 2SLS that remains valid under treatment effect heterogeneity and multiple LATEs, addressing a key gap in existing methods.

Findings

Proves the conventional variance estimator is inconsistent with multiple LATEs.

Derives the correct asymptotic distribution for 2SLS under heterogeneity.

Provides a practical variance estimator based on Hall and Inoue's method.

Abstract

Under treatment effect heterogeneity, an instrument identifies the instrument-specific local average treatment effect (LATE). With multiple instruments, two-stage least squares (2SLS) estimand is a weighted average of different LATEs. What is often overlooked in the literature is that the postulated moment condition evaluated at the 2SLS estimand does not hold unless those LATEs are the same. If so, the conventional heteroskedasticity-robust variance estimator would be inconsistent, and 2SLS standard errors based on such estimators would be incorrect. I derive the correct asymptotic distribution, and propose a consistent asymptotic variance estimator by using the result of Hall and Inoue (2003, Journal of Econometrics) on misspecified moment condition models. This can be used to correctly calculate the standard errors regardless of whether there is more than one LATE or not.