Work

Hysteresis through learning by doing: Estimating endogenous productivity

Year: 2026

Type: article

Abstract: Learning by doing (LBD), a mechanism whereby productivity depends on past employment, is increasingly used in macroeconomic models to capture hysteresis and endogenous productivity. A key challenge has been calibrating its structural parameters, primarily due to the limited availability of empirical estimates. We estimate a productivity persistence of 0.91 and an elasticity of productivity with respect to lagged employment ranging from 0.13 to 0.30, with seven statistically significant estimates averaging 0.22. These results provide empirically grounded parameters and demonstrate the value of LBD as a tractable approach to capturing endogenous and procyclical productivity dynamics in macroeconomic models. • Learning by doing is a mechanism whereby productivity depends on past employment. • We estimate key parameters for learning-by-doing models using macroeconomic data. • Productivity persistence is estimated at 0.91. • The elasticity of productivity with respect to lagged employment averages 0.22. • Learning by doing effectively models endogenous, procyclical productivity.

Source: Economics Letters

DOI: 10.1016/j.econlet.2026.112818

Authors: Timothy Watson, Juha Tervala

Institutions: Services Australia, Department of Social Services, University of Helsinki

Language: English

Topic: Economic Growth and Productivity

Subfield: Economics and Econometrics

Field: Economics, Econometrics and Finance

Domain: Social Sciences

SDG: Decent work and economic growth

Open Access status: hybrid

Estimated APC paid: $3,030