This site is undergoing review. Some annotations were human-generated, some AI-generated — all are being verified.

Inefficient Automation

Beraja, Zorzi

2022NBER Working Paper Series24 citations

Theoretical / conceptualTheoretical model

Automation / RobotsGeneral automationOccupational mobilityAugmentation vs. substitution

Summary

Beraja and Zorzi develop a dynamic heterogeneous agent model with endogenous automation, occupational reallocation frictions, and borrowing constraints to study optimal policy responses to automation, showing that the equilibrium is inefficient when firms fail to internalize displaced workers' limited ability to smooth consumption.

Main Finding

The equilibrium is inefficient when reallocation and borrowing frictions are sufficiently severe: firms fail to internalize that displaced workers have limited ability to smooth consumption while reallocating. The optimal policy slows down automation substantially (doubling the half-life from 20 to 47 years), achieving welfare gains of approximately 4% under efficiency weights and 6% under utilitarian weights, by flattening consumption profiles and raising consumption early when displaced workers value it more.

Primary Datasets

PSID

Labor outcomes

Calibration targets from multiple sources: BLS labor share data; unemployment duration from Alvarez and Shimer (2011); productivity loss from Kambourov and Manovskii (2009); income process from Floden and Lindé (2001) using PSID; occupational mobility from Kambourov and Manovskii (2008); tax progressivity from Heathcote et al. (2017)

Secondary Datasets

McKinsey Global AI

AI-focused

Graetz and Michaels (2018) for automation depreciation; McKinsey (2017) for fraction of automated activities; Bharadwaj and Dvorkin (2019) for routine occupation share; Ganong et al. (2020) for unemployment replacement rate

Key Methods: Dynamic heterogeneous agent model with overlapping generations, borrowing constraints, and reallocation frictions; constrained Ramsey problem; quantitative calibration with discrete choice occupational mobility and idiosyncratic income risk
Sample Period: 1970-2020 (quantitative model calibration period)
Geographic Coverage: United States (quantitative calibration)
Sample Size: Theoretical model; quantitative calibration uses aggregate and distributional moments from US data
Level of Analysis: Individual, Occupation, Firm
Occupation Classification: None
Industry Classification: None

Notes

The Review of Economic Studies, vol. 92, no. 1, pp. 69-96 [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run. [Claude classification]: Published in Review of Economic Studies 2025, vol. 92, no. 1, pp. 69-96. This is a theoretical paper with quantitative calibration. The model features endogenous automation choice by firms and reallocation decisions by workers facing borrowing constraints. The key mechanism is that displaced workers become borrowing-constrained during slow reallocation, creating a wedge between private and social returns to automation. The optimal policy taxes automation temporarily during the transition but not in the long run.