Just last month I was discussing Ed Banfield’s work on “present orientedness” and the work it did in getting him to make mass incarceration the platform of the Republican party. I pointed out that his use of thicker psychological descriptions of individuals in justifying longer and more punitive criminal penalties was very consistent with behavioral economics, and wondered why current behavioral economics wasn’t used for more reactionary and conservative ends.
Karl Smith is concerned after reading this paper by Bryan Caplan and Scott Beaulier – Behavioral Economics and Perverse Effects of the Welfare State. It argues the conservative critique that welfare leaves people worse off using behavioral economic rationality, since normal microeconomics argues you can’t make someone worse off by giving him or her more choices. It even quotes Ed Banfield.
None of Smith’s commenters get his question, so I’ll give it a shot. Please kick in comments, as my behavioral economics is not very good and not at all formally learned (disclosure: conceptually I think it’s a dodge from having to make a real social critique). I understand why Smith’s commenters want some formalism, as the behavioral economics is being sprayed everywhere in an aerosol can manner. Take his main example. A lot of what you think about this kind of exercise will derive from what you think about this kind of example:
A simple numerical example can illustrate the link between helping the poor and harming them. Suppose that in the absence of government assistance, the true net benefit of having a child out-of-wedlock is -$25,000, but a teenage girl with self-serving bias [unrealistically optimistic and overconfident] believes it is only -$5000. Since she still sees the net benefits as negative she chooses to wait. But suppose the government offers $10,000 in assistance to unwed mothers. Then the perceived benefits rise to $5000, the teenage girl opts to have the baby, and ex post experiences a net benefit of -$25,000 + $10,000 = -$15,000.
But shouldn’t loss aversion, one of the central pillars of behavioral economics, amplify the up-front costs of having a child? The pain and stress on the body are going to be a sure loss. And since government assistance is spread out over time, shouldn’t a hyperbolically discounting poor person care less about it? It seems a bit pick-and-choose.
As for Karl Smith’s specific worries about the welfare state: shouldn’t loss aversion, overconfidence, time-inconsistent discounting and difficulty following statistical probabilities lead a behavioral person to be made better off by universal health care and old age pensions? The probability of managing an individual investment portfolio as well as estimating the chances and subjective experience of being destitute in old age is difficult to estimate individually and the tail risk of the loss that occurs in poverty would be particularly painful. The same statement should be true for health care.
Same for unemployment insurance. One of the more curious behavioral responses is that people hate unemployment. They hate not being part of their productive community, they hate not contributing, they hate the loss of identity that one gets as someone who works. To an economist that’s b-a-n-a-n-a-s. Unemployment should be a pleasant vacation! But, last time I checked, it wasn’t (is that consistent with the latest frontiers in happiness research?).
That people are bad risks judges and hate losses should redouble our faith in social democratic insurance, not detract.
Rortybomb 
If we define “behavioral economics” as meaning something like “people don’t behave like the econ-101 textbook,” we should include Charles Karelis’ “The Persistence of Poverty” idea, which holds that some very poor people might work MORE if you just give them money. (It has something to do with the slopes of labor supply curves on account of people dealing with pain differently than they deal with pleasure.)
This is a “behavioral” argument for more social welfare policy expenditures.
I should have mentioned that. I really like that book, it reviewed it here. (No Kindle edition yet. Boo.) Highly recommended.
You may still see bingeing, etc. if you give people money in this situation, so it fits but it is actually making people very much better off than worse off.
Bryan Caplan does not actually appear to believe that welfare mothers are that responsive to the incentive to have kids.
IMO people are stupid but society is smart and norms will change over time. Never the less I though a libertarian am not against welfare for the poor, it is the quixotic attempts to subsidize the middle class and rich (SS, medicare, Gov. schooling free to all) that get me. I would like to see welfare done efficiently.
“I … wondered why current behavioral economics wasn’t used for more reactionary and conservative ends.”
Oh it will be. Look, you’re just going to have to be a little patient. It takes time to organize, and build the cadres, and so on, but we are eagerly studying and learning to apply the most modern theories and understand, and the most successful and proven techniques, in shaping and manipulating public opinion. We don’t have time to march through the institutions, but we can simply smash them when the time is right. But you can rest assured we’re not going to leave the arrow of behavioral economics in the quiver.
Mike, I think you’re saying that if poor people have preferences such as hyperbolic discounting as the behavioral economics stuff suggests, then their decisions are optimal (for themselves). We might call those decisions bad, but hey, who’s to say what’s bad.
I believe the disconnect is that Beaulier/Caplan, and most behavioral economists and their spectators, are evaluating subjects behavior from a neoclassical norm. In other words they’re saying if that poor guy didn’t have hyperbolic discounting (i.e. if he was a neoclassical agent like you and me) he would make different decisions. Those “different” decisions are then labelled “bad”. If policy is to stop people from doing “bad” things then the point of the B/C paper is to show that some policies can backfire. All of this depends on the normative assumption that neoclassical preferences produce “good” behavior.
In your example of pensions and universal health care, you slip into using the neoclassical norm, too. These non-neoclassical agents are better off by those policies if you think the neoclassical norm (e.g. smoothing consumption) controls what is best for that agent even if they don’t have neoclassical preferences.
Personally, you have to take a stand on what norm to use. If you think the neoclassical norm is controlling enough to use coercive policy to change people’s behavior, then B/C critique of some policies holds. If you think that people with weird preferences are already doing what’s best for them simply because they’re optimizing over their weird preferences, then that doesn’t leave much room for policy.
Clarification: in my second paragraph (slaps forhead: blog comments shouldn’t be so long to have paragraphs!) the phrase “different decisons” switches reference between two things. Sorry. The paragraph should read:
I believe the disconnect is that Beaulier/Caplan, and most behavioral economists and their spectators, are evaluating subjects behavior from a neoclassical norm. In other words they’re saying if that poor guy didn’t have hyperbolic discounting (i.e. if he was a neoclassical agent like you and me) he would make different decisions. The decisions they make because of their weird preferences are then labelled “bad”. If policy is to stop people from doing “bad” things then the point of the B/C paper is to show that some policies can backfire. All of this depends on the normative assumption that neoclassical preferences produce “good” behavior.
Regarding unemployment (and ignoring your cheap shot on economists… boo… hiss…): if there’s a preference to be in the employed state — to be a part of the productive community — then this would be easily remedied by getting a job at Mickey D’s or Starbucks or starting a lemonade stand. I think preferences must be much more complicated than just a simple preference to be employed.
Regarding unemployment (and ignoring your cheap shot on economists… boo… hiss…): if there’s a preference to be in the employed state — to be a part of the productive community — then this would be easily remedied by getting a job at Mickey D’s or Starbucks or starting a lemonade stand.
No it wouldn’t.
Undeniably rock-solid argument you got there. I mean really, the 4’33” of arguments.
I can’t claim much expertise in this area but I agree with you that loss aversion could be expected to amplify the up front costs of having a child out-of-wedlock. I would have thought a behavioral economics analysis would start with that proposition.
The authors seem to be implying that after the birth of the child the teenage girl is likely to regret the choices that she made. I would have thought that she would be more likely to rationalize the choices that she made – i.e. her preferences are likely to change so that they are more favourable to the child. She is unlikely to choose to have the child adopted out even if caring for it is more difficult than she imagined.
The example seems to rest on alleged negative externalities associated with out-of-wedlock children – which seems to me to be conventional neoclassical economics.
The problem with BC’s analysis is they don’t address the “how much” question. They lay out the *possibility* that the welfare state can hurt the poor via behavioral economic mechanisms, but they don’t tell us how important that channel is. I suspect Karelis’ analysis suffers from the same problem.
The next step in a policy analysis has to be addressing this question.
Note that in this case, behavioral economics is just another way for conservatives to say “black people are stupid and lazy” without having to cop to the obvious racism.
Caplan and Beaulier are talking about means-tested welfare programs. The mechanism they describe—means-testing combined with overoptimism creating the illusion of an incentive to make oneself worse off in order to pass the means test—would not apply to welfare programs which are not means-tested.
very thought provoking. I was just picking some books from http://www.farnamstreetblog.com/2010/04/behavioral-economics-reading-list.html and this will help.
Is there any actual human being who thinks about having a child out of wedlock as an exercise in profit maximization, as opposed to having sex then, for cultural or other reasons, choosing to have the child? Shouldn’t models for economic behavior be considered only to the extent that they can be shown to reflect actual, real-world, behavior, as opposed to the mere fact that the model can be constructed?
Wait, Tom- you mean you don’t have a child out of wedlock as an exercise in profit maximization? You’re missing out man-it’s the best.
Furthermore, some people having sex expect that it will, or at least might, lead to marriage, and only later discover that it won’t; so they’re not really choosing the “out-of-wedlock” part *at all*, let alone based on a pseudo-calculation of this sort. And that part is the most salient to conservatives — they don’t want to discourage all childbearing, after all. And I assume most of them have pretty much accepted that they’re not going to roll back the clock on premarital/nonmarital sex in general.
Also, since marriages don’t necessarily last as long as they used to, ISTM that marital status at the time of birth isn’t as significant as it used to be. It’s not a terribly reliable predictor of marital status at the child’s fifth or tenth birthday (in either direction), either of which may be more relevant to the child’s actual life outcomes.
P.S. Converting the nonfinancial gains/losses to dollar terms, averaging, and then adding the government support is a flawed methodology even under assumptions that financial and nonfinancial benefits/harms are comparable; you should convert the financial outcome of each scenario with government support to its utility equivalent and then average *utility* across possible outcomes. This is more work and requires actually setting forth a money-utility conversion function, which explains why few people bother to do it, but without it the model isn’t even mathematically rigorous, let alone representative of real-world decisionmaking.
Pingback: Does Behavioral Economics Undermine the Welfare State? « The Baseline Scenario
Pingback: Is Happiness Conservative? « The Baseline Scenario
Maybe I am a bit obsessive with details, but having read the Beaulier and Caplan paper entirely and the comments it generated, one thing strikes me as missing:
Has anyone realized Beaulier and Caplan are not psychologists, and yet they not only act as if they had solved the problem of what is intelligence (for them, it’s high IQ), but also disentangled its causes and effects and proposed policy guidelines.
And the intellectual foundation they used to achieve that remarkable feat is 4 papers in psychology and one pop psychology book, highly controversial, The Bell Curve, by Herrnstein and Murray (which underpins their whole case on IQ!!!!).
And, believe it or not, that paper was actually published in a journal.
Not fair to lump all behavioural economists in with Caplan et al. Their work does not label decisions bad or good. They were/are just looking for better models of actual human behaviour. The Caplan et al is political science and pure politics as they move on to suggest policy responses to alleged bad decisions. This is clearly different from the nobel-prize work done by others!