Does Prospect Theory explain Trump and Brexit votes?

The quantity of votes in favor of the Brexit and in favor of Trump surprised most of us. We were surprised that Brexit and Trump voters would reject pretty safe options—remain in the EU and continue with a qualified Democrat president—in favor of obviously risky options. The uncertainty that comes with Brexit and with Trump is, and was, I think, apparent to everyone. If it is true that many people chose to vote for Trump and Brexit in spite of knowledge of the risks, then Prospect Theory can explain why.

Prospect Theory is a major theory in the field of behavioral economics. It explains why and when people are willing or reluctant to take risks. In many cases, people prefer not to take risk. For instance, most people prefer winning a sure $50 rather than taking a risky bet in which they can toss a coin and either win $100 or nothing. But in some cases, people prefer risky options. They do so when they feel that they have lost something. Imagine, for instance, that you’ve been playing roulette and that you’ve just lost an important sum of money. To recoup your losses, you might well take important risk. The choice set that you face is then something like this: either you lose $50 for sure, or you take a risky bet with a probability ½ of recovering your loss, leaving with $0, and a probability ½ of losing $100. The risks and potential gains, are exactly the same as when your options are either to earn $50 for sure or take a bet with ½ chance of either winning nothing or $100. Yet, when it concerns loss, people are much more willing to take risks. While in the winning frame, most people would go for the sure option, in the losing frame, most people would go for the bet. Prospect Theory systematically describes these changing attitudes towards risk: people are either risk seeking or risk averse depending on whether the stakes are conceived as gains or as losses. This in turn depends on whether the potential consequences of choices are states of affair that are better or worse than some reference point—what is considered ‘normal.’

How is that relevant to understanding Trump and Brexit votes? In 1988, Quattrone and Tversky, two behavioral economists, published a paper in which they show that the same attitudes towards risk—risk aversion or risk seeking—manifest themselves when deciding for whom or for what to vote. If you perceive that you or your country has been losing lately, then you’re likely to be risk seeking. If you do not have that perception, then you’re likely to be risk averse. Many of the explanations for the Brexit and Trump votes that we recently read in the news emphasized how frustrated citizens were with their economic situation, and the effect of fear mongering discourses and appeal to lost grandeur. Frustration with the economic situation is directly caused by the Great Recession and the growth of inequalities. Many perceive their current economic and social situation as one of loss domain in the sense that it is much worse than what it should normally be. The political discourses of the Trump and Brexit advocates have framed the stakes in terms of losses rather than gains. The slogans “Make America great again” and “Take back control” clearly refer to the lost grandeur of the past. This sets the reference point as a lost state that was much better than the current one. Also, fear mongering is by definition talking about the frightening (negative) state in which we find ourselves. All this motivates citizens to favor risky options: the gains, even if they are unlikely, are so strongly desired that they induce discounting the very likely losses.

Expected utility of presidents’ potential performances in a loss frame, according to Prospect Theory.

The striking consequences of risk seeking choices, is that one may prefer an option that has on average worst prospects and high uncertainty (wide range of possibilities) to an option that has better prospect on average and low uncertainty. This point can be seen with the graph, which pictures the function that ascribes utility, or desirability, to president’s performance. The shape of the function is specified by Prospect Theory. In particular, the fact that the curve of this function is convex in the loss domain expresses the risk seeking attitude. In this graph, Clinton is represented as having a better expected performance than Trump. But because there is great uncertainty with regard to Trump’s performance and because, in this situation, people are risk seeking, Trump’s presidency ends up being more desirable: the expected utility of Trump’s presidency is higher than Clinton’s in spite of the fact that its expected performance is lower. (I determined the expected utility, pictured by the arrows on the desirability axis, by assuming equiprobability of performances within each range).

With Trump and Brexit voters, we face people who, perceiving that they have lost something, are willing to take high risks. Economic frustrations cause risk seeking preferences, which contribute to motivating votes for Trump and Brexit.


Quattrone, G. A., & Tversky, A. (1988). Contrasting rational and psychological analyses of political choice. The American Political Science Review, 82 : 719-736.


  • comment-avatar
    Bill Benzon 15 November 2016 (12:12)

    One of the most interesting things I’ve read about the Trump victory. In effect, he performed the old gestalt switch: “You see an old crone? Here, look at her my way. See, she’s a beautiful young woman. Vote for me! You see a rabbit? No, it’s a duck. I’m going to cook you the most wonderful Duck à l’Orange. It’ll be fantastic! Vote for me!

    • comment-avatar
      Christophe Heintz 15 November 2016 (17:31)

      —Framing: the duck-rabbit analogy and the moral issue—

      Thank you for this comment: The analogy with the duck-rabbit illusion is helpful. The duck-rabbit illusion is such that the same picture can be perceived as a picture of a rabbit or as a picture of a duck. It’s the same picture, and yet it can be perceived differently. In decision making, the very same stakes can be presented and perceived differently, and this influences what choice is made. Behavioral economists have talked about ‘framing effects’ for naming the fact that the way the information is presented, rather than just the stakes, influence decisions. Prospect Theory accounts for one type of framing effect: positive versus negative framing will influence attitudes towards risk.

      Because Framing Effects contradict the standard theory of rational decision making, one could argue that relying on these effects in order to influence votes is manipulative: it does not consist in providing the best information for making the best decision, but it takes advantage of specific psychological mechanisms. Yet, some behavioral economists such as Thaler and Sunstein have pointed out that it is not possible to provide an information without any framing. If information is to be conveyed, this information will be presented in a specific way. This is unavoidable.
      So the question in political and moral philosophy is: When making a political discourse, is it bad to frame stakes in terms of losses rather than in terms of gains? Relatedly: should we encourage voters to be risk averse rather than risk seeking?

      Myself, I would not say yes to the above question. There is, however, one consequence of negative framing that bothers me: risk seeking voters will tend to prefer candidates that do not specify their programs. Indeed, if the program is unspecified or not really discussed, then the voter can envisage very bad as well as very good future performances. In a positive framing, this is a disadvantage, but in a negative framing, this is advantageous. (It comes with a recommendation for campaigners: the more a program is specified, the more positive should the framing of the stakes be, and reciprocally. I would hypothesize that it is what campaigners have already done, and that we have a correlation between frames and program specificity in past campaigns). This effect bothers me because for a democracy to function well, there should no advantage coming from being unspecific.

    • comment-avatar
      Christophe Heintz 15 November 2016 (17:38)

      —Framing at will ?—

      Bill’s funny remarks also suggest that framing is some kind of magic trick. This points to an empirical question: can stakes be framed at will in a positive or negative way? I would tend to think that there is always some wiggle room, but there are also constraints. A duck rabbit-picture cannot be perceived as the Eiffel tower. I guess a negative framing will be more or less successful depending on whether it resonates with personal experiences. There are quite a few discussions on the web about whether Trump voters were mainly people who where right to feel frustrated because of their economic conditions or not. Maybe Trump voters felt frustrated because of things other than their economic status, or maybe they were not feeling frustrated, which would mean the loss frame did not play a role in motivating a vote for Trump. These are empirical questions, but I do not have good analyses to point at yet.

  • comment-avatar
    Samuel Veissière 15 November 2016 (19:56)

    Thanks for this brilliant piece. My take on this is that there is more than “avoiding loss by gambling” in this story. What I am calling The Trump Effect is more of an invariant recipe for relevance and epistemic power in general: ingroup (+ prestige + frequency) biais + fundamental attribution error / person bias + moral dumbfoudedness. I am pretty sure this works as a contentless recipe for political decision-making.

    • comment-avatar
      Christophe Heintz 18 November 2016 (17:40)

      Thank you Samuel. I agree with you that risk preferences can only be one element motivating votes. As you point out and develop in your blog post, there must be quite a few psychological factors.

  • comment-avatar
    Bill Benzon 15 November 2016 (21:01)

    First, the election was quite close in the popular vote, with Clinton slightly ahead (at this point in the count I think it’s about 1%). One can imagine that various voters had various motivations. But Trump presented the same image to everyone. Back in July George Lakoff published a longish essay in which he presents Trump as strong father against Clinton an nurturing mother. So, whatever their particular motivations, everyone sees Big Daddy promising a better tomorrow. That provided just enough prospective energy to unite them in voting for Trump.

    Second, and FWIW, I grew up in western Pennsylvania in coal and steel country. Though I haven’t been back there in years, I’m in touch with some of my old high school acquaintances on Facebook and they went for Trump. They’re clearly looking for a MUCH BRIGHTER TOMORROW. “Make America Great Again” clearly resonated with them.

    • comment-avatar
      Christophe Heintz 18 November 2016 (17:35)

      Thank you Bill for pointing to George Lakoff’s essay. Since Lakoff is an influential figure in cognition and politics, this essay is clearly worth mentioning.
      As argued by Lakoff, the Strict Father Model might have played a role in motivating votes for Trump. So we have three *compatible* potential psychological factors for explaining Trump votes: risk-seeking preferences (my post), group-biases and other biases (Samuel Vessiere) and cultural models in politics (Lakoff). How would you test for their respective effect on the motivation to vote for Trump?
      Evidence for the effect of risk seeking preferences should include:

      1. Evidence that people did indeed think of the situation in terms of loss rather than gain. The framing of the debate and the demographics of Trump voters (from regions with economic recession) seem to go in that direction.
      The CNN exit pool also provides some relevant numbers:
      – Among the 31% of people thinking that life for the next generation of Americans will be worse than today, 63% voted for Trump.
      – Among the 27% who thought their financial situation was worse today than four years ago, 78% voted for Trump.

      2. evidence that many Trump voters thought of him as a risky candidate.
      For instance, exit pools show that 60% of the voters thought of Trump as not qualified to serve as president, yet 18% of those still voted for him. (Only 5% of those who thought Clinton is not qualified voted for her). Of those who thought that neither of Trump and Clinton are qualified for the presidency, 69% voted for Trump.

      I’ve used CNN exit pool:
      Admittedly, I have looked for confirmatory evidence. People are welcome to look for possible falsifications.

  • comment-avatar
    Brain molecule Marketing 18 November 2016 (15:50)

    My reading is that Beh Econ has no biological or evidence basis. However, it has been a masterful marketing job and career booster.

    Instead, I would look to the biological primacy of ethnic hatred and the fear response as the cause of the behavior. The “Pathogen Theory” is a decent correlational model of why fear or the outsider is such a universal and powerful driver of hostile and reactive behavior – fight, flight or freeze.

    Anecdotally, we can see that fear mongering targeting outsiders is the dominant media and political strategy. Biology always “trumps” ideology.

  • comment-avatar
    Christophe Heintz 18 November 2016 (18:41)

    Prospect Theory is based on a very large set of behavioral data. The paper that I cite includes the result of several experiments. You might want to question the value and significance of these experiments, but saying that there is no evidence is plainly false.

    About the biological basis of behavioral economics, I can point, for instance, to this paper about the neural substrate of risk attitudes:
    Tom, S. M., Fox, C. R., Trepel, C., & Poldrack, R. A. (2007). The neural basis of loss aversion in decision-making under risk. Science, 315(5811), 515-518.

    I share your interest in the study of adaptive cognition. About the adaptive value of the risk attitudes described by Prospect Theory, you can have a look at this paper:
    McDermott, R., Fowler, J. H., & Smirnov, O. (2008). On the evolutionary origin of prospect theory preferences. The Journal of Politics, 70(02), 335-350.

    I hope you are right when you say that doing behavioral economics is a career booster. That would mean that economists have recognized that their field need to stand on better psychological assumptions. With regard to voting behavior, the standard economic theory is limited to specifying what political preferences voters have. Behavioral economics investigates further the psychological factors, including the role of framing on attitudes towards risk.

    Regarding “Pathogen Theory”: If I understand you well, you point to yet another psychological factor that might have motivated votes for Trump. Psychological mechanisms whose evolved functions is to avoid pathogens might include intergroup prejudices; such prejudices have motivated votes for Trump. Thanks for pointing out this hypothesis. Note however that it is misleading to say that “biology trumps ideology.” A more precise characterization of the role of biology on ideologies is that the success of ideologies depends on the specifics of evolved human cognition.