Overcoming Bias

How To Fix Wage Bias

It seems reasonably well documented that tall, slim, pretty people get higher wages.  A 2005 review:

Hamermesh and Biddle found that the "plainness penalty" is 9 percent and that the "beauty premium" is 5 percent after controlling for other variables, such as education and experience. In other words, a person with below-average looks tended to earn 9 percent less per hour, and an above-average person tended to earn 5 percent more per hour than an average-looking person. For the median male in 1996 working fulltime, the respective penalty and premium amounted to approximately $2,600 and $1,400 annually. The corresponding penalty and premium for the median female worker are $2,000 and $1,100. ... In a separate paper . ...They found evidence of a beauty premium for attorneys that increases with age, at least for the 1971-78 classes. Five years after graduating, a male lawyer from these classes with a beauty rating of one rank above average had approximately 10 percent higher earnings than his counterpart with a rating of one rank below average. Fifteen years after graduation, the beauty premium increased to 12 percent.

Social Scientists Know Lots

For me, the year's best reading is usually the many thoughtful answers to The Edge's annual question.  This year they ask: 

What have you changed your mind about?  Why?

I'd love to hang out with these folks, so maybe I should audition.  My answer:

Social scientists know lots.

As a physics student and computer science researcher, I assimilated the usual "hard science" perception that "social science" is an oxymoron -- no one knows much about it, so your opinion is as good as anyone's.  When I finally decided I needed social science credentials, to turn my institution hobby into a career, I focused on experimental economics, the only sort a hard scientist could trust, and Caltech, with impeccable hard science credentials.  But I was soon thoroughly convinced: social scientists know tons. 

Why then do so many people think otherwise?  Many say it is because social scientists are stupid, or the social world is too complex or uncontrollable.  Better answers are that social expertize conflicts with our overconfidence about familiar experience, or with our democratic ideology that everyone's political opinions should get equal weight.  But the best answer, I think is that most public talk by social experts reflects little social science.  That is, what social experts say in legal or congressional testimony, or in newspapers or magazines, mostly reflects what they and we want and expect to hear, instead of what expert evidence reveals.   

Econ of Longer Lives

I'm a little late to the party, but in December's Cato Unbound debate, Aubrey de Grey and Ronald Bailey argued that much longer healthy lifespans would be good, while Diana Schaub and Daniel Callahan had doubts.  Some samples:

Daniel Callahan: My standing complaint against de Grey and his enthusiastic colleagues is that they defend themselves by hypothesizing a variety of changes in our present way of life that would make our extended lives a kind of heaven on earth. We would be so healthy and energetic we would want to keep working indefinitely. We could start new careers, new families, new ways of life. That we might get tired of it all, or bored, is not allowed into their calculations. Nor is any imaginative effort to imagine the deleterious social effects allowed.

Ronald Bailey: So what about the social consequences of radically longer and healthier lives? In that regard, Diana Schaub in her reaction essay raises many questions for reflection about those consequences, but curiously she fails to actually reflect on them.  Schaub ... simply recapitulates the standard issue pro-mortalist rhetorical technique of asking allegedly "unnerving questions" and then allowing them to "fester in the mind." Sadly, all too many bioethicists think they've done real philosophic work by posing "hard" questions, then sitting back with steepled hands and a grave look on their countenances.

This issue has sparked many debates, conferences etc. over the last few years.  The invited participants have naturally been intellectuals who have published on the topic recently, mainly activists and bioethicists.  We economists have not published on this topic, and so have not been included.  But this is not because we have nothing to say.  Instead, no economist has anything special to say.  We can all easily see that standard economic theory seems to say longer healthy lives are a good thing.  So none of us thinks any of us should get precious academic publication credit for saying such an obvious thing.  As a result, life extension debates ignore economic theory. 

Of course appearances may be deceiving, so perhaps there are good economic theory reasons against longer healthy lives.  And perhaps economists would typically let their "judgment" overrule economic theory on this issue.  But it still seems to me a shame that observers of this debate can remain unaware of what standard economic theory seems to say on this subject. 

Interior Economics

Imagine you asked what fraction of whipped cream should be air, and someone responded "it is impossible to have whipped cream without air."  Or imagine you asked how much discretion judges should have, relative to following clear legal rules, and someone responded "it is impossible to eliminate all discretion from legal decisions."  You'd think they were avoiding the question.  Similarly, I asked:

For good policy advice, what is the best weight to place on economic theory, versus (individual or cultural) intuitive judgment?

And offered a tentative answer:

My guess is over 75% weight, so I try to mostly just straightforwardly apply economic theory, adding little personal or cultural judgment.

Tyler Cowen "answers":

There is no such thing as "straightforwardly applying economic theory." ... Theories are always applied and interpreted through our personal and cultural filters and there is no other way it can be.  Robin believes in an Archimedean point for using theory, I do not. ... Robin's post is the clearest example I have seen of what I call Robin's logical atomism.

Gender Tax

Last week I noted that, in an hour of searching, I couldn't find any defenders of the economic-theory-supported tall tax, even among economists.  Today I report it was easy to find supporters (1, 2, 3, 4, 5, 6, 7, 8) of Alesina, Ichino, and Karabarbounis's proposed man tax, also well supported by economic theory:

Gender Based Taxation ... changes spouses' implicit bargaining power and induces a more balanced allocation of house work and working opportunities between males and females. Because of decreasing returns to specialization in home and market work, social welfare improves by taxing conditional on gender. When income sharing within the family is substantial, both spouses may gain from [it].

So riddle me this: if careful economic analysis had instead favored taxing men less than women, how many supporters do you think that proposal would have found, even among economists?  And what does that say about how much economic theory influences economists' policy conclusions?

P.S.  The best "man tax" is not a fixed tax amount for being a man, nor is the best "height" tax a fixed amount per inch of height.  Instead, each gender or height has a different income tax schedule.  So there is no special reason to fear some "couldn't pay the tax."

Added 20Dec:  Here's a paper on a more general tax on genetic features. 

Economist Judgment

Weak tall-tax support among economists suggests to me this question:

How much does economic theory influence economists' policy recommendations?

Or more precisely:

Considering all policy judgments of all economists (made as economists, thinking economic theory relevant), what fraction of judgement variance is explained by the fact that they know a common economic theory (including frameworks, default assumptions, and standard data sources)?

I asked (a version of) this question at lunch yesterday.  Tyler C. suggested about 15%, while Bryan C. thought the number could be much higher for some average of economist judgments.  (Alex T. and I withheld judgment.)  Low estimates suggest economic theory usually says little, or that we are usually unwilling to give it much weight relative to our considerations (legitimate or illegitimate). 

I know of two relevant studies.  In one study, economist (and physician) policy choices had no relation to their beliefs on health economics facts.  Similar results were found for labor policy.  This raises two more questions:

Which economists rely more on economic theory, relative to other considerations, for their judgments?

For example, are libertarians better able to judge economic theory implications because libertarians explicitly reject such implications as a basis for policy?  Also:

For good policy advice, what is the best weight to place on economic theory, versus (individual or cultural) intuitive judgment?

My guess is over 75% weight, so I try to mostly just straightforwardly apply economic theory, adding little personal or cultural judgment.   Which is why I endorsed the tall tax.  What weights are other economists shooting for?

Added: That lunch seems to have also inspired Alex's post today.

Tax The Tall

The New York Times Magazine "Year In Ideas" issue is great again.  One of the 70 ideas covered is Mankiw & Weinzierl's proposal, floated in April, to tax the tall: 

Should we tax tall workers at a higher rate than their shorter peers? The answer - yes - "follows inexorably" from reigning academic theories of taxation, argues Greg Mankiw. ... Using optimal-taxation formulas, Mankiw and Weinzierl crunch the numbers and come up with a "tall tax" amounting to 7 percent of a tall person's income. Short people would receive a 13 percent rebate.

Do Mankiw and Weinzierl actually endorse such a system? Far from it. Rather, they argue, the proposed tax clarifies our thinking about taxation in general. They say that height is a "justly acquired endowment" ... By the same logic, they imply ... the government has no right to force someone with the "justly acquired endowment" of entrepreneurial genius to pay a higher tax rate.  Peter Diamond, an economist at M.I.T., says the paper's basic mistake is the notion "that if you can draw a silly inference from an approach, then that discredits a model."

To which Mankiw replies:

If economic theorists are allowed to embrace inferences from a model that they like and cavalierly reject those that they consider "silly," what is the point of theory? That discretion gives the theorist the freedom to always confirm his priors.

Evaporative Cooling of Group Beliefs

Followup to:  Uncritical Supercriticality

Early studiers of cults were surprised to discover than when cults receive a major shock - a prophecy fails to come true, a moral flaw of the founder is revealed - they often come back stronger than before, with increased belief and fanaticism.  The Jehovah's Witnesses placed Armageddon in 1975, based on Biblical calculations; 1975 has come and passed.  The Unarian cult, still going strong today, survived the nonappearance of an intergalactic spacefleet on September 27, 1975.  (The Wikipedia article on Unarianism mentions a failed prophecy in 2001, but makes no mention of the earlier failure in 1975, interestingly enough.)

Why would a group belief become stronger after encountering crushing counterevidence?

Heroic Job Bias

Friday Eliezer described a Superhero bias:

The police officer who puts their life on the line with no superpowers, no X-Ray vision, no super-strength, no ability to fly, and above all no invulnerability to bullets, reveals far greater virtue than Superman.

Superman risks less, yet we celebrate him more.  Eliezer followed with:

John Perry was a New York City police officer ... the fact that an atheist and a transhumanist can still be a police officer, can still run into the lobby of a burning building, says more about the human spirit than all the martyrs who ever hoped of heaven.

Actually the hero bias is far worse than Eliezer imagines, because in fact police and firefighters do not risk their lives more for you!  Yes, these workers do die more often on the job, but they die much less often because of the job.  You see, these are relatively high status jobs, and having low status is quite deadly for humans.  So in fact people risk their life more by becoming waiters, bus drivers, truck drivers, plumbers, painters and lots of other jobs!  A 1999 Demography article shows how male mortality risk varies with job category:

Publication Bias and the Death Penalty

The front page of Sunday's New York Times contained an interesting article reviewing research linking  the death penalty to homicide trends.  Adam Liptak attempts to provide a balanced account of the debate, noting first one set of findings:

According to roughly a dozen recent studies, executions save lives. For each inmate put to death, the studies say, 3 to 18 murders are prevented.

And then my own research:

The death penalty “is applied so rarely that the number of homicides it can plausibly have caused or deterred cannot reliably be disentangled from the large year-to-year changes in the homicide rate caused by other factors,” John J. Donohue III, a law professor at Yale with a doctorate in economics, and Justin Wolfers, an economist at the University of Pennsylvania, wrote in the Stanford Law Review in 2005. “The existing evidence for deterrence,” they concluded, “is surprisingly fragile.”

Surely a dozen studies is itself evidence of robustness.  Why then is then is it that we find these results are fragile?  Two words: Publication bias (also known as the file drawer problem).  Our research revealed that alternative approaches to testing the execution-homicide link can yield a huge array of possible results (positive and negative).  But if only strong pro-deterrent results are reported (and the others remain in the file drawer), this could look misleadingly like there is a pro-deterrent consensus.

It turns out that there are some rather simple tests for publication bias.  Our friends in medicine provide a useful intuition.  Imagine that there are many separate drug trials being considered - some with large samples, some with small samples.  If all results are being reported, then smaller samples should, on average, yield similar estimates to larger samples, albeit with a bit more noise (in both directions).  So the standard error of an estimate should be uncorrelated with the coefficient.  But if researchers only report statistically significant estimates, then they will only report results with t-statistics>2, yielding a strong correlation between standard errors and coefficient estimates.

You can probably guess what we find.