Friday, April 27, 2018

Andrew Gelman on the Economics Profession.

Professor Gelman is the Director of the Applied Statistics Center at Columbia University.
From his Statistical Modeling, Causal Inference and Social Science blog (bolded bits, our emphasis):

A quick rule of thumb is that when someone seems to be acting like a jerk, an economist will defend the behavior as being the essence of morality, but when someone seems to be doing something nice, an economist will raise the bar and argue that he’s not being nice at all.
Like Pee Wee Herman, act like a jerk
And get on the dance floor let your body work

I wanted to follow up on a remark from a few years ago about the two modes of pop-economics reasoning:
You take some fact (or stylized fact) about the world, and then you either (1) use people-are-rational-and-who-are-we-to-judge-others reasoning to explain why some weird-looking behavior is in fact rational, or (2) use technocratic reasoning to argue that some seemingly reasonable behavior is, in fact, inefficient.
The context, as reported by Felix Salmon, was a Chicago restaurant whose owner, Grant Achatz, was selling tickets “at a fixed price and are then free to be resold at an enormous markup on the secondary market.” Economists Justin Wolfers and Betsey Stevenson objected. They wanted Achatz to increase his prices. By keeping prices low, he was, apparently, violating the principles of democracy: “‘It’s democratic in theory, but not in practice,’ said Wolfers . . . Bloomberg’s Mark Whitehouse concludes that Next should ‘consider selling tickets to the highest bidder and giving the extra money to charity.'”

I summarized as follows:
In this case, Wolfers and Whitehouse are going through some contortions to argue (2). In a different mood, however, they might go for (1). I don’t fully understand the rules for when people go with argument 1 and when they go with 2, but a quick rule of thumb is that when someone seems to be acting like a jerk, an economist will defend the behavior as being the essence of morality, but when someone seems to be doing something nice, an economist will raise the bar and argue that he’s not being nice at all.
I’m guessing that if Grant Achatz were to implement the very same pricing policy but talk about how he’s doing it solely out of greed, that a bunch of economists would show up and explain how this was actually the most moral and democratic option.
In comments, Alex wrote:
(1) and (2) are typically distinguished in economics textbooks as examples of positive and normative reasoning, respectively. The former aims at describing the observed behavior in terms of a specific model (e.g. rationality), seemingly without any attempt at subjective judgement. The latter takes the former as given and applies a subjective social welfare function to the outcomes in order to judge, whether the result could be improved upon with, say, different institutional arrangement or a policy intervention.
To which I replied:
Yup, and the usual rule seems to be to use positive reasoning when someone seems to be acting like a jerk, and normative reasoning when someone seems to be doing something nice. This seems odd to me. Why assume that, just because someone is acting like a jerk, that he is acting so efficiently that his decisions can’t be improved, only understood? And why assume that, just because someone seems to be doing something nice, that “unintended consequences” etc. ensure he’s not doing a good job of it. To me, this is contrarianism run wild. I’m not saying that Wolfers is a knee-jerk contrarian; rather I’m guessing that he’s following default behaviors without thinking much about it.
This is an awkward topic to write about. I’m not saying I think economists are mean people; they just seem to have a default mode of thought which is a little perverse.

In the traditional view of Freudian psychiatrists, which no behavior can be taken at face value, and it takes a Freudian analyst to decode the true meaning. Similarly, in the world of pop economics, or neoclassical economics, any behavior that might seem good, or generous (for example, not maxing out your prices at a popular restaurant) is seen to be damaging of the public good—“unintended consequences” and all that—, while any behavior that might seem mean, or selfish, is actually for the greater good.

Let’s unpack this in five directions, from the perspective of the philosophy of science, the sociology of scientific professions, politics, the logic of rhetoric, and the logic of statistics.

From the standpoint of the philosophy of science, pop economics or neoclassical economics is, like Freudian theory, unfalsifiable. Any behavior can be explained as rational (motivating economists’ mode 1 above) or as being open to improvement (motivating economists’ mode 2 of reasoning). Economists can play two roles: (1) to reassure people that the current practices are just fine and to use economic theory to explain the hidden benefits arising from seemingly irrational or unkind decisions; or (2) to improve people’s lives through rational and cold but effective reasoning (the famous “thinking like an economist”). For flexible Freudians, just about any behavior can be explained by just about any childhood trauma; and for modern economists, just about any behavior can be interpreted as a rational adaptation—or not. In either case, specific applications of the method can be falsified—after all, Freudians and neoclassical economists alike are free to make empirically testable predictions—but the larger edifice is unfalsifiable, as any erroneous prediction can simply be explained as an inappropriate application of the theory.

From a sociological perspective, the flexibility of pop-economics reasoning, like the flexibility of Freudian theory, can be seen as a plus, in that it implies a need for trained specialists, priests who can know which childhood trauma to use as an explanation, or who can decide whether to use economics’s explanation 1 or 2. Again, recall economists’ claims that they think in a different, more piercing, way than other scholars, an attitude that is reminiscent of old-school Freudians’ claim to look squarely at the cold truths of human nature that others can’t handle....MORE
Previously from the SMCISS blog:
Taking on the Nobel Prize Winner: An Update On That "Middle Class White Guys Are Dying" Report
"Big Oregano Strikes Again"
Elon Musk Frenemy Peter Thiel is writing another book!
See also our posts:

Modelling vs. Science
A subject near and dear to our jaded hearts, some links below.
If an experiment is not reproducible it is not science.
If an hypothesis is not falsifiable it is not science.  

"Scientists Can’t Replicate AI Studies. That’s Bad News"
We're with Popper and Feynman on the overarching premise: If what you're doing isn't falsifiable, if what you're doing isn't replicable, what you're doing isn't science.
And if what you're doing was funded by the public in any way the law should consider the resulting code to be owned by the public.
There, three different concerns dispensed with in two sentences.

The Next Time Someone Tells You Economics is a Science Remind Them of Mendeleev
From the Royal Society of Chemistry:
What is a mark of a great scientist? Good scientists discover new information and make sense of it, linking it to other data. They may go further by giving an explanation of this linked data which, maybe not immediately, other scientists accept as a correct explanation. However the outstanding scientist goes further in predicting consequences of his ideas which can be tested. This boldness identifies the great scientist if the predictions are later found to be accurate. One such person was Russian chemist Dmitri Mendeleev.....
Nitrogen Upgraded, Potash Target Lowered; Ununquadium Decayed (AGU, CF; TRA; POT)
To my mind one of the goals of science or investing should be enough mastery and understanding to enable prediction. The greatest example in science was probably Dmitri Mendeleev's creation of the periodic table and his insight that he should leave spaces for elements not yet discovered.
His prediction of the properties of gallium, germanium and scandium contrasts with pseudo-science in that it is testable.

If you ever want to piss an economist off, tell them that just because they use a tool of science (mathematics), that alone doesn't make economics a science. Science is falsifiable. Mendeleev had a swing-and-a-miss on the atomic weight of tellurium, the prediction was falsified, showing that this is a true science.
Ununquadium is element 114, discovered in 1998. We're now up to 118, ununoctium, with a gap at 117. The prediction of the properties of the undiscovered element includes a half-life of 3 nanoseconds.

This ramble was triggered by a discussion last night on alternatives to gallium (in CIGS) and tellurium in CdTe thin films. It was the first time I realized an argument could be made that Mendeleev is the father of thin-film solar. Continuing our trip through the periodic table, on to the nitrogen story...
 And many more, use the "search blog" box, upper left if interested.