Friday, August 22, 2014

The Small BIG

Steve Martin, Noah Goldstein and Robert Cialdini have a new book coming out later this month called 'The Small BIG' and their publisher was kind enough to send me an advance copy. This is the second collaboration by the three co-authors after they published 'Yes! 50 Secrets from the Science of Persuasion' in 2007. Robert Cialdini, aside from being a highly-cited social psychologist within academia, is also the author of the 1984 book 'Influence' which continues to be very popular among both lay audiences and nerds like myself. Their latest book is in a similar mould of describing how persuasion techniques, contextual factors and environment cues can be used to influence others in business and social situations.

When it comes to popular science books I generally take a chicken wing approach of gobbling up the science and skimming the popular aspects as efficiently as possible; in other words my copies of 'Predictably Irrational', 'Blink', Freakonomics' and so on are full of underlined passages and notes next to where the academic papers are referenced. When I want to look up something which I found particularly interesting in one of these books, I thumb through and quickly find the relevant study using this method.

With that said I found 'The Small BIG' refreshingly direct in its presentation. The book is made up of 52 main chapters, each around 4 pages in length and providing a "small" psychological insight that can have a potentially large effect on behaviour (hence the book title). The chapter titles follow the same structure, such as "How can a small change in venue lead to big differences in your negotiations", "What small big can be used to make defaults more effective?" and "What small big could help motivate others (and yourself) to complete a task".

All the chapters draw on the authors' obviously deep knowledge of the social psychology literature in order to relate real-life tips and applications. For example, the chapter "What small big could ensure you're dressed for success?" cites a paper which found that wearing a stethoscope led a nurse's health messages to be recalled more successfully (Castledine 1996), and an older study (Lefkowitz et al. 1955) which found that people were considerably more likely to follow a man (illegally) crossing the street on a red light when he wore a suit compared to casual clothes. Another chapter examining order effects (i.e. the order in which you present options may influence people's reaction to them) cites a paper which found that people perceived "580 [hours of tv] for $285.90" as a better deal than "$285.90 for 580 hours" (Bagchi & Davis 2012). Interestingly this effect disappeared when consumers were presented with a computationally easier choice ("600 hours for $300").The authors then suggest how these kind of findings might apply in a business context, such as framing sales orders by item-first rather than price.

To give a final example, the authors discuss the findings of a recent paper examining how people perceive opportunity cost (Frederick et al. 2009), one of the most fundamental insights of economic thought and a concept which can really change your life if you deeply internalize it. That paper conducted an experiment wherein one group were offered the chance to purchase a DVD for $15 with the options "Buy the DVD" or "Not buy the DVD". The second group were offered the same product but with the options "Buy the DVD" or "Keep the $15 for other purchases". This subtle reframing led to a reduction in purchase rates in the second group from 75 to 55 percent, a finding which sales professionals could take either way in terms of highlighting or underplaying the opportunity cost of following their communications (e.g. look at the phrase used 20 seconds into this De Beers advertisement).

With 117 references in the 52 chapters, I found this book a very effective way to dive into the persuasion/influence literature and I learned of many new studies by reading it.

Wednesday, August 20, 2014

Behavioural Science and Public Policy

Below is cross-posted from and is my attempt to summarise behavioural science and public policy for Irish policymakers (but clearly similar issues in other countries): 

I have posted here on a number of occasions about the relevance of the growing literature on behavioural economics and public policy for the Irish context. This post updates this with some new material and I hope people don’t mind if I draw on some from previous posts.

Increasingly, behavioural science is being used as a term to encapsulate the integration of psychological factors into understanding economic decision-making. This is basically an attempt to preserve the phrase “behavioural economics” to refer to explanations with explicit utility-theoretic foundations and also to avoid a lot of work from psychology simply being repackaged as “behavioural economics”. It is not a wholly satisfactory compromise as the phrase “behavioural science” means different things to different people but it is certainly helping to form a shared set of ideas and methodologies and looks likely to continue as the main way of describing this work.

There are a number of reasons for the explosion of interest in this area including the award of the Nobel prize to Daniel Kahneman in 2002 and the adoption of the book “Nudge” by the Obama and Cameron administrations. I think also the sense of purely neo-classical microeconomics being bound up with the regulatory failures surrounding the financial crisis is also fueling an appetite for more realistic accounts of decision-making. It is likely that a lot of what is now called economics will increasingly move towards a disciplinary more blurry field in particular in areas like financial regulation.

Some recent very useful overviews of this area include: Shafir’sBehavioural Foundations of Public Policy is excellent; Sunstein’s lengthy “Empirically-Informed Regulation” provides a strong overview; Nudge is obviously important; a recent paper by Brigitte Madrian outlines the behavioural approach to policy; this excellent short paper by Beshears et al makes the case for the limitation of revealed preferences and the need for other mechanisms; one of the researchers in our group has put together a data-base of studies employing what can loosely be called “Nudges” in various areas of policy; Publications of the Behavioural Insights Team in the Cabinet Office are available here; I have also put together a fairly detailed reading list on behavioural economics and public policy, including legal and ethical issues; The Brookings Institute publication “Policy and Choice: Public Finance through the lense of behavioural economics” is one of the best available introductions to this area.

In terms of why Irish policy-makers should care about this area, below is not intended to be exhaustive but is an attempt to summarise the main areas.

1. The use of “nudges” to encourage saving is the most developed behavioural policy literature. This has reached national policy significance in the roll-out of pension auto-enrolment in the UK. The Irish pension framework was to see the entire private sector begin to be auto-enrolled in 2014 but subject to an economic recovery that has not yet materialised sufficiently. The psychology behind how people react to default settings in pensions is very interesting with a lot of opportunities and threats, among the latter the possibility that people will anchor too much to the default contribution and under-save as well as the possibility that naive consumers will simply be ripped off by providers who can charge higher fees with this less savvy group.

2. The role of behavioural science in financial regulation is a key question. The Financial Conduct Authority has been exploring this area actively. This excellent FCA occasional paper examines the potential implications of behavioural economics for financial regulation. In the US context, this very interesting report by Barr, Mullainathan and Shafir from 2008 outlines a new approach to consumer regulation based partly on the notion of “sticky defaults” whereby firms would be required to default people into the most desirable option based on their characteristics and only move them if they make choices following being provided with clear information. Such models are discussed in relation to two markets fraught with behavioural bias and consumer exploitation, namely credit cards and mortgages. The document also sets out proposals for changing the incentives of brokers.

As noted in another post, this literature is leading to a lot of very interesting questions for financial regulation that are hard to ask in a neo-classical setting. Below are some examples but obviously a small subset.

Should credit card variable and teaser rates be banned or at least taken out of the regular offers made to consumers?

Should mortgage providers be forced to disclose better deals available to their customers?

Should pay-day lenders be granted full access to the Irish market? If so, how do you regulate them?

Should auto-enrolment proceed in Ireland, what provisions should be put in place so that companies do not exploit naïve consumers by charging fees well in excess of regular rates?

Do behavioural biases prevent annuities markets from functioning optimally?

3. The implications of behavioural science for the design of welfare and taxation policies is another active area with applications across the Irish policy sphere in everything from structuring environmental taxes to design of incentive systems to encourage employment. Cass Sunstein, who is one of the main figures in this area, recently released a new book called “Simpler: The Future of Government“. It outlines an approach to government that emphasises making regulations, laws and taxes less confusing and more robust.

4. The search for alternative measures of welfare and social progress is a big concern of the emerging literature (see summary and readingsfrom recent conference on this). The Stiglitz-Sen commission is becoming a standard reference on this topic and it is pretty comprehensive. Understanding how we go from the empirical literature in this area into meaningful indicators is an important direction for this literature. As well as interest in measuring well-being, there is growing interest in the bidirectionality of well-being and economic activity with a lot of recent work looking at impact of mental health in particular on economic functioning. (See Layard: Mental Health: The Frontier of Labour Economics). Related to this, an increasing literature has been examining the economic importance of ensuring good child mental health. This literature is helping us to understand better the interplay between poor child mental health and later economic outcomes. A recent PNAS paperby Goodman, Joyce and Smith gives a good indication of the type of research being conducted in this area. This is an extremely important area of research at the interface of psychology and economics.

5. A lot of recent research has begun to examine more closely the mechanics of what happens during job search from a more psychological perspective. Some of this research is explained in accessible form in this Brookings Institute publication. There is no question that traditional labour supply models are not a complete guide for understanding the behaviour of people who have been laid off and the literature on job activation needs badly more cross-disciplinary work to understand what is shaping behaviour and what environmental changes people might respond to.

6. James Heckman and colleagues have been working on a large programme to integrate personality psychology and a theory of human development into economics. This is extremely important in terms of providing a theoretical and empirical basis for allocation of spending in health and education. Many of these papers are available on Heckman’s IDEAS webpage. Colleagues in Geary are involved in a collaboration looking at early childhood development. Some of these ideas are presented in accessible form on this website.

There are clearly several empirical, ethical and legal issues with the development of this agenda across all of these areas. The enthusiasm for randomised controlled trials in this area clearly has to be tempered with an awareness of their limitations (e.g. here). Furthermore, the extent to which interest-groups constrain the types of policies that emerge will be interesting to observe.

Along with colleagues, I have organised an annual workshop on economics/psychology in Ireland and it will take place again on October 31st in the Geary Institute (sign-up page here). Anyone interested in this area is welcome to attend.

Randomised Controlled Trials

The use of randomised controlled trials in policy is a key aspect of the emerging literature in behavioural science and policy. Key researchers in development economics such as Esther Duflo (papers here) successfully advocated their use in evaluating development projects. Economics has a tradition of using RCTs including famous experiments on negative income taxation and the legendary RAND Health Insurance experiment. However, clearly for the first time they are becoming widespread across many fields. Below will not be new to anyone working in this area and is intended to get students thinking about RCTs more broadly.

It is often argued that RCTs avoid the problems of methods such as linear econometric modelling and instrumental variable analysis in that the mechanism assigning participants to treatment is completely observed by the researchers and thus it is possible to make causal inference but students should be careful in evaluating claims about RCTs. It is simply not true that, as a rule, they involve fewer assumptions than techniques such as fixed effects panel models or instrumental variable modelling. The assumptions required depend upon the type of problem being addressed and the feasibility of different designs. Many RCTs involve experimenting with a small geographical subset of the population and there are various stages leading from targeting that subset to them actually participating and many other issues such as differential attrition and so on. Going from the results of such experiments back up to the population parameter of interest requires many assumptions just as interpreting a well-constructed IV parameter does. (See Angus Deaton for an infinitely more eloquent elucidation of these issues eg here).

The phrase Local Treatment Effect is now commonly used in Economics to get across the idea that most experiments, whether randomised or natural, give you an estimate of the effect of the variable of interest that is, to a degree, specific to the type of treatment and to the group that receive it (see the very good Mostly Harmless Econometrics for a gentle, albeit deceptively so, introduction to this). Estimating local treatment effects is a more modest and achievable goal than estimating fixed average population effects. Many RCT designs estimate local treatment effects well but are often presented as giving a causal effect that will apply to different groups in different situations. The usual debates about ecological validity, replicability, representativeness etc., all apply to RCTs in policy and are interesting to think about in the local treatment framework.

Understanding the conditions under which a randomised controlled trial reveals something useful about policy is an important goal of behavioural science. For example, many RCTs carried out on large online groups may meet exactly the conditions to be informative as they are being conducted on very large samples of people who themselves will be the target for future changes. Similarly, policy experiments in small regions on self-selected individuals may be informative if those regions and individuals are precisely the group being targeted by the policies e.g. a job-centre RCT may tell us nothing about the effect on people who don't go to the job centre but this is not such a problem if they are not of interest to the policy question. Arguably, many of the trials being conducted by the Behavioural Insights team would score highly on these criteria as they are often trialling directly with programme participants with a view to changing those specific programmes. But in general, there are a lot of questions to address before going from RCT results back up to the population of interest.

This debate between Deaton and Banerjee is very informative regarding the issues at hand. This is obviously still an ongoing and at times vociferous debate. What is emerging is a much more nuanced language for describing the results of experiments and their relation to policy.

Tuesday, August 19, 2014

Vacation Framing: Experiment Results

Here are the results of the one-day, one-question experiment I started last night. The hypothesis was that participants would value 2 weeks of vacation time more in a hypothetical job offer if it was framed as a loss compared to a gain - a pretty classic endowment effect experiment. I'll put the main limitation up front - the sample is almost laughably biased towards a relatively savvy group of behavioural science nerds because I advertised it here and on twitter. Thanks to the 113 people who participated in this experiment. It was motivated by p3, paragraph 13 of this Sunstein & Thaler paper.

Participants were randomized into either a "loss frame" or "gain frame " question. They were:

In both conditions the participants are facing the same task - you have 2 weeks vacation for sure, now put a value on the 2 negotiable weeks. The only difference is whether it was framed a loss (you have 4, how much do you want for the 2) or a gain (you have 2, how much will you pay for 2 more).

Here are the $ distributions of the two groups. You can see a considerable amount of clustering at $5,000 (the maximum value possible) in the loss frame, indicating those people valued the 2 negotiable weeks very highly.

Lastly, here are the average $ values. Those in the loss frame were willing to pay 1.88 times more - neatly in line with the 2:1 ratio usually observed in loss:gain situations.

Monday, August 18, 2014

1-Question experiment

Blog readers,
If you could spare a minute, please consider answering this 1-question experiment. It's just for a bit of fun and I'll post the results in a day or two.