The “nudge” agenda – certainly oversold and arguably underperforming

Richard Thaler and Cass Sunstein published Nudge: Improving Decisions about Health, Wealth, and Happiness in 2008. The authors laid out the prospect of changing economic, social, and political behaviour in ways that neither challenged individual freedoms nor imposed collective costs. Within a couple of years, the possibility of nudging citizen behaviour at minimal cost to the government in directions thought to be of society-wide benefit was embraced enthusiastically around the world.

The first and best-known Behavioural Insights Team (BIT), or Nudge Unit, was founded by the U.K government under David Cameron in 2010 and is now a wholly-owned unit of the innovation charity Nesta. The BIT website now urges prospective clients to “discover how giving your clients a gentle nudge using research and data focused on human behaviour can help your business to scale within your industry”.

BIT claims to have run “over 780 projects to date, including 500 randomised controlled trials in dozens of countries over the past nine years, more than the rest of the UK government combined in its history … (and as a consequence) … have a wealth of insights and results to build on in continuing to shape policy and practice”. Identified successes in the UK include:

1 – Increasing tax payments to bring forward £200 million extra revenue in 12 months;

2 – Reducing days on benefits by between 5 and 10 million each year after improving online systems for jobseekers;

3 – Reducing antibiotic prescription by three to four per cent amongst the highest prescribing G.P.s, resulting in over 70,000 fewer prescriptions over six months;

4 – Using text messaging to reduce by 150,000 the number of repossession interventions by bailiffs and saving £30 million;

5 – Adding 100,000 people to the organ donation register;

6 – Persuading 20 per cent more people to consider switching energy providers; and

7 – Doubling the number of applicants to the British Army.

However, doubts have emerged over the past five years regarding research validity and operational outcomes. Research led by Queen Mary University has shown: “that despite the widespread use of behavioural interventions across society, failed interventions are surprisingly common” (Osman et al, 2020).

Research in 2020 looked at the results of 126 randomised controlled trials run by two nudge units in the United States (The Behavioural Insights Team North America and the Office of Evaluation Sciences). The research revealed that the nudge trials had, on average, only 1.4 per cent of the expected impact (Della Vigna and Linos, 2020). This is much lower than the 8.7 per cent impact predicted in behavioural economics literature.

Then, in 2021, “the first comprehensive analysis of past research on techniques aimed at changing citizen behaviour”  covered 212 published articles involving more than two million participants (Mertens et al, 2021). The authors’ work revealed that there was only very moderate significance in the difference between nudged and not-nudged groups.

It can thereby be argued that nudging has three critical weaknesses:

1 – Impact – nudges have not proven consistently effective at steering individual choice in the right direction (behaviour is more challenging to change than expected).

2 – Implementation – nudges do not necessarily lend themselves to easy implementation in public policy; and

3 – Imposition – nudges can too easily slip into coercion or manipulation (mainly if the nudger is parsimonious with its provision of information to the nudgee).

It is in the last of these three weaknesses that the dark side of nudging is evident. In that nudging may not be about helping people make better choices; but may actually about getting people to make the choices that policymakers want them to make. While, at the same time, making it nigh on impossible to deal with deliberate manipulation by the nudging entity. This is clearly evident in two notorious government initiatives that have resulted in all three of these negative outcomes. Both programs, Australia’s Robodebt Scheme and the UK’s Post Office Scandal, demonstrate the ethical risks and damage involved in programs driven by the pursuit of savings and blind faith in behavioural insights or alogracy (government by algorithm often leading to increased inequality due to opacity, scale, and widespread damage to individuals).

Both programs were ultimately brought down by the: “optimism bias of the behavioural tsars that … led them to place too much stock in their judgement in a world of limited evidence” (Sodha, 2020) and by the nudging of staff and suppliers to perform to corporate targets, rather than to customer service benchmarks.

In the case of the UK’s Post Office Scandal, 2500 branch owner-operators were accused of crimes they did not commit, 900 were wrongly prosecuted in court, several served time in jail, and at least four committed suicide.

It is concluded that nudging is no silver bullet, but neither has it totally failed to deliver on its promise. It is just that its promise has been overblown and, in certain cases, used to benefit the nudger rather than the nudgee. It is also clear that the actual impact of any stand-alone nudging program is often both restricted in scope and limited in half-life. Thus, nudging can be expected to deliver only as one contributory element in a coordinated series of inter-institutional interventions that include compulsion, facilitation, information, and persuasion – the Nudge itself.

This not a stand-alone panacea to complex policy facilitation. Nor, when applied in an ethics-free zone, is it free of seriously harmful side-effects.

This blogpost is an edited version of the article by Fergus Neilson published in the Journal of Behavioural Economics and Social Systems (BESS) edition 6.1 in 2024. To access the full article, please go to The Nudge agenda – possibly oversold and arguably underperforming.