When Should Governments Invest More in Nudging? Revisiting Benartzi et al. (2017)

Avishalom Tor, Jonathan Klick

Research output: Contribution to journalArticlepeer-review


Highly influential recent work by Benartzi et al. (2017) argues-using comparisons of effectiveness and costs-that behavioral interventions (or nudges) offer more cost-effective means than traditional regulatory instruments for changing individual behavior to achieve desirable policy goals. Based on this finding, these authors further conclude that governments and other organizations should increase their investments in nudging to supplement traditional interventions. Yet a closer look at Benartzi et al.'s (2017) own data and analysis reveals that they variously exclude and include key cost elements to the benefit of behavioral instruments over traditional ones and overstate the utility of cost-effectiveness analysis for policy selection. Once these methodological shortcomings are corrected, a reassessment of key policies evaluated by the authors reveals that nudges do not consistently outperform traditional interventions, neither under cost-effectiveness analysis nor under the methodologically required cost-benefit analysis. These illustrative findings demonstrate that governments concerned with social welfare cannot simply assume the superiority of behavioral instruments and should strive instead to conduct cost-benefit analyses of competing interventions, including nudges, to identify the most efficient of the available instruments.

Original languageEnglish
Pages (from-to)347-376
Number of pages30
JournalReview of Law and Economics
Issue number3
StatePublished - 1 Nov 2022
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2022 Walter de Gruyter GmbH, Berlin/Boston.


  • behavioral regulation
  • cost benefit analysis
  • cost effectiveness analysis
  • nudge

ASJC Scopus subject areas

  • Economics, Econometrics and Finance (all)
  • Law


Dive into the research topics of 'When Should Governments Invest More in Nudging? Revisiting Benartzi et al. (2017)'. Together they form a unique fingerprint.

Cite this