De-biasing and the ‘Nudge’ Unit

by Ken Gibb

The media reported today that the Government’s Behavioural Insights Team will be partially privatized (Government is to hold a competition to find a commercial partner). The doyens of libertarian paternalism, the ‘nudge’ unit employs ideas from behavioural economics and finance (as well as other sympathetic ideas) in order to influence behavioural change and de-bias behaviour in such a way as to increase social welfare and of course improve public service efficiency and generate savings. The initiative started in the backwash of the Coalition’s election. It has had quite a controversial time despite the outwardly attractive rejection of rational behavior for a recognition of a series of biases and heuristics. Despite some large claimed savings or benefits associated with policies working with a range of public sector agencies, the unit has been criticized widely in the UK by commentators and opposition politicians. Only today the Guardian ran a story criticizing the use of psychometric testing as part of job seeker requirements, apparently part of a Nudge unit exercise. The BIT website is a good place to see what they have been up to [1]. It charts the work of the unit, its links to key US academics, a number of news stories and its own reports and newsletters. I would recommend one specific paper on random controlled trials, co-authored by Ben Goldacre – not quite the same thing as behavioural nudging but a nice (if strongly pro) summary of social policy applications of RCTs.

It strikes me that aside from the whole question of the acceptability of libertarian paternalism, there are empirical questions about the size of the benefits associated with different policy changes that need to be assessed on a case by case basis. Many experimental projects demonstrate useful and important policy findings on issue like self-control, auto-enrolment and the like, but, more fundamentally, the policy objectives reflect the government’s wider agenda. So they may be deploying such ideas in areas that one does not think policy should be directed in the first place. For instance, policies on issues like extending the right to buy, or the bedroom tax can be argued to be problematic per se regardless of whether nudging is involved or not.

In my own work, Alex Marsh and I have long been pursuing the role of heuristics, loss aversion, mental accounting and other features of the behavioural economics body of work, as ways of thinking about how housing choices are made, about housing finance innovation and with regard to housing equity withdrawal. We actually submitted a paper in the late 1990s involving these ideas – though a modernised version of it only finally saw the light of day in 2011! Apart from the long road to publishing acceptability, we have always taken the view that the behavioural turn has to be augmented with a recognition of institutional features of the housing market and the reality of pervasive uncertainty about the consequences of decisions in the future and those made now for the future. Despite its attack on standard assumptions in neo-classical economics, it is far more likely that behavioural economics will be subsumed in a broader mainstream economics; rather than the other way around. This, to our way of thinking, cannot capture the reality of housing processes and choices and market activity. But that is not to say that there are not many useful ideas and insights gleaned by the work of Kahneman, Camerer, Thaler and others.

In the airport the other day I picked up a new book by Rolf Dobelli – The art of thinking clearly (Sceptre). This turns out to be a highly readable (and it is translated) pithy account of all of the heuristic and cognitive biases, plus a few other decision-making problems. Apart from being quite funny in places it does a great job of defining quite technical things in crystal clear layman’s terms (the short chapter on hyperbolic discounting is a great example). However, the interesting thing is that in each chapter he makes an attempt to suggest what one needs to do to overcome such biases and to make better decisions. Bearing in mind that the author identifies more than 95 of theses biases and problems we are actually confronted with a massive set of rules, considerations, algorithms, criteria and other ways of overcoming the biases. It is actually much more challenging to live making decisions like this (and the responses are not always consistent) than arguably would be the requirements to be a substantive rational utility maximiser. There is a real danger that decision-making becomes impossible or significantly delayed if we are to take responding to recognizable biases (having worked through how we do identify them when they arise) as seriously and do so comprehensively as the logic of this analysis implies.

Simple one-off nudges may turn out to be the informationally efficient alternative to trying to de-bias decision-making complexity consistently.