The Common Problem of Bad Controls in Tests of the Linguistic Savings Hypothesis

A Comment on Ayres et al (PNAS, 2023) and related literature

Abstract

Languages encode the future differently, such that languages can be sorted into different types. Strong Future Time Reference (FTR) languages have an obligatory distinction between present and future, whereas weak FTR languages do not. The Linguistic Savings Hypothesis predicts this difference will affect people's patience, perhaps because weak FTR languages make the future feel closer and more salient. Three recent articles have reported significant effects on individual patience as well as firm-level income smoothing and investment efficiency. However each paper includes controls which increase bias, rather than reduce it. Due to data sharing limitations, I can only publish a reanalysis in one case; fixing the problem leaves smaller and non-significant effects. This problem does not affect all research on the Linguistic Savings Hypothesis, but the effects are not as large or widespread as previously reported.