Right in the Short Run

David Laibson mentioned how if you give people randomly assigned health care, and then give them the options to change two weeks later, they tend not to.  Sendhil Mullainathan mentioned how health care providers systematically recommend the same plan no matter what medical history the caller (a senior citizen) had and no matter what cocktail of medicines (s)he was on.  Thoughts?

Over a year ago, over at Marginal Revolution, Tyler Cowen asked the question, “How quickly should I go through my stock of Battlestar Galactica?”

He argues:

The Hotelling rule tells us to consume a stock so the shadow value rises at a rate commensurable with the rate of interest…or something like that. C’mon, let ’s get real. Here are a few options:

1. Set aside one day for a BSG fest. I would lose the pleasures of anticipation, so no way. (Would you want all non-currents-events-specific MR posts available all at once?) The pleasures of memory would be weaker as well.

2. Have a strict rule, such as one a day.

3. Have a stranger impose a rationing pattern. Sometimes we call this stranger the Science Fiction Channel. But what about the accumulated stock of programs on DVD?

4. Watch it when your wife lets you (not an issue for those that have married well).

5. Refuse to watch the last episode, in an attempt to deny your mortality.

6. Watch them at an increasing rate.

#3 is appealing, but so far I am opting for #6. Comments are open, if you wish to rationalize what you already have done.

Addendum: This question will become more important.

Typically, when I read a good book or watch a good film, I tend to go through it all at once. For one thing, I don’t really enjoy the “anticipation”. I just want to read the damn thing. Moreover, I’m perfectly happy to reuse these goods - I re-read and re-watch things all the time. The benefit of watching/reading things straight through is that it allows me to immerse myself in a fantasy world continuously for a whole day… so I vote #1 …

Lots of value there …

I never cease to be amazed at the power of economics in providing a language to discuss social phenomena. Take for instance, Flavio Cunha and James Heckman’s “The Technology of Skill Formation”. Abstract below:

This paper uses a simple economic model of skill formation to organize this and other evidence summarized below and the findings of related literatures in psychology, education and neuroscience. The existing theoretical literature on child development in economics treats childhood as a single period (see, e.g., Becker and Nigel Tomes, 1986; S. Rao Aiyagari et al., 2002; Roland Benabou, 2002). The implicit assumption in this approach is that inputs into the production of skills at different stages of childhood are perfect substitutes. We argue that to account for a large body of evidence, it is important to build a model of skill formation with multiple stages of childhood, where inputs at different stages are complements and where there is self-productivity of investment. In addition, in order to rationalize the evidence, it is important to recognize three distinct credit constraints operating on the family and its children. (i) The inability of a child to choose its parents. This is the fundamental constraint imposed by the accident of birth. (ii) The inability of parents to borrow against their children’s future income to finance investments in them. (iii) The inability of parents to borrow against their own income to finance investments in their children. This paper summarizes findings from the recent literature on child development and presents a model that explains them.

Thoughts?

I was in the supermarket late last night, and I got to thinking.

There once was a world filled with egg-eating people. These people were hard workers; left with little time of their own, they were forced to shop for food during their lunch breaks. It turns out that in this strange world, there were two types of jobs, and each job employed a different type of person. Let’s call these two groups the R’s and the P’s.

Although his job was not simple, every R was lucky enough to get a long lunch break, during which time he went to the market to buy eggs for his family. When an R picked out a carton of eggs, he would inspect it for damage and look for another carton if he saw that an egg or two was broken. Once he found a good carton, he would buy it and return to work. Since the R’s were so diligent about inspecting their eggs, they almost never brought home a cracked one.

The P’s worked hard, but unlike the R’s, their jobs were much more restrictive. Although P’s got a lunch break, it was a lot shorter than that of the R’s. Since P’s needed eggs to feed their family just as much as R’s did, they would use this time to run to the market, too. The P’s found that if they grabbed an egg carton and paid for it, they could get back to work in time, but when they stopped to check for cracked eggs, they would get back to work late. Sometimes P’s who returned late were penalized with docked pay, and the next day these P’s could not even afford to bring home any eggs.

It turned out that quite a few of the cartons contained cracked eggs. Since the P’s were too rushed to open the cartons before they purchased them, P’s frequently purchased these damaged cartons. The eggs were the family’s only food, and when a few eggs were cracked, everyone ate a little bit less, including the children.

After a while, some of the R’s became curious about the lifestyles of the P’s. Eventually, an R went into a P’s home and found some egg cartons, each of which had a cracked egg or two. This R went back home and told his family about what he had seen. The family agreed that the P’s were making poor decisions at the market: cracked eggs were worthless, but the P never bothered to check for them. Furthermore, this R had heard the P complaining about not having enough eggs to feed his children. How silly, thought the R, for this P needs only to check his egg cartons for cracks to be able to provide plenty of eggs for his family.

The R family told their friends, who were also R’s, all about the P household that they had seen. Pretty soon, all the R’s had heard about the way the P’s lived, and none of them could understand it. They are too stupid to purchase eggs for their family! said one. The source of their troubles is their inability to focus on egg checking! claimed another.

So time went on, with the R’s bringing home pristine cartons and feeding their children well, and the P’s bringing home damaged cartons and leaving their children hungry. The R’s knew about the hungry P children, and most of them didn’t like it. But in this world, the R’s felt strongly that all the P’s needed to do was concentrate a little more while shopping at the market. It never occurred to them that the P’s knew just as much as the R’s about how to check their eggs, and that many P’s would have gladly looked inside the cartons if they had the time.

Every once in a while, a P would propose that something be done about the cracked eggs.  One P said that once in a while, some of the R’s should have to give a few eggs to the P’s. Another P thought that P jobs should have a bit longer of a lunch break. But the R’s laughed at these suggestions.

In this world, little was ever done to help the P’s, and in fact the situation continues to this day.

The end.

There is some pretty interesting research being done in Economics & Psychology over at NYU. Economic theorist Andrew Caplin and Mark Dean are working on an axiomatic approach to neuroeconomics, together with the help of Paul Glimcher.

Reinforcement learning theory has produced important insights into economic behavior. Intriguingly, neuroscientists recently discovered a plausible mechanism through which reinforcement may be encoded in the brain. Yet their resulting “dopaminergic reward prediction error” hypothesis has not yet been incorporated into economics. We develop an axiomatic model that characterizes the empirical implications of this theory for an idealized data set comprising both neuroscientific measurements and choices. Our axiomatization removes the language barrier between economics and neuroscience. This will allow “neuroeconomic” experimental protocols to be developed appropriate to the questions motivating economic, as opposed to purely neuroscientific, interest in learning.

In the words of a good friend, “it’s really strange to see ‘dopamine’ and ‘metric space’ in the same sentence”.

Whether integrating experimental psychology and brain imaging results into economic theory will have practical value remains to be seen. However, in my limited comprehension of the field, it seems ripe with potential. For instance, neuroscience results on inter-temporal discounting and status-quo bias in decision-making could be translated into the language of decision theory. By axiomatizing some of the experimental results into the economic theory, models may better articulate and predict consumer behavior.

A little bit of background on myself: I love computers, and I love economics. Unfortunately, the two disciplines mix pretty infrequently. That’s why I let out a little yelp when I found this paper, called “Fact-Free Learning” by Enriqueta Aragones, Itzhak Gilboa, Andrew Postlewaite, and David Schmeidler. The paper is trying to present a model of how we understand information that is already available to us. Sometimes, we figure out something new by looking at the data we have in a new light — fact-free learning!

But fact-free learning is tough. The real breakthroughs are often unexpected, and they may happen slowly. The authors argue that some familiar tools of economics and computer science, when brought together, can explain this phenomenon. Before I get into their argument, I need to explain something called complexity, because the central argument of the paper relies on it.

Computational complexity is the study of how computer algorithms scale as the size of the problem they are trying to solve increases. Usually we are looking for some sort of bound on the amount of time it will take the algorithm to finish, given a problem of size n.

Some algorithms may not scale so well. Seriously, they might not scale so well. We’re not entirely sure! (Why? Wikipedia explains.) This class of algorithms is called NP, and we are pretty sure that any implementation of an NP-hard problem will end up taking a few lifespans-of-the-universes to complete for even a small size input.

Now let’s get back to the paper. Remember how I said that it brought together tools of both computer science and economics? Well, the economic tool these fellows bring to the table is the regression. Say you have information on lots of variables, and you want to see which variables explain some phenomenon. If you pick out a few of these variables, you can regress the measure of the phenomenon on those variables to determine which variables are relevant.

But what if you can only use a few variables — say k — in the regression at once? Then you might want to run the regression with every possible combination of k variables, looking for the one that does the best job explaining the phenomenon. The authors argue that this process — finding the set of k variables that does the best job explaining a phenomenon in a regression — is a lot like fact-free learning. But there’s a catch:

Linear regression is a structured and relatively well-understood problem, and one may hope that, using clever algorithms that employ statistical analysis, the best set of k regressors can be found without actually testing all (mCk) subsets. Our main result is that this is not the case. Formally, we prove that finding whether k regressors can obtain a prespecified value of R2, r, is, in the language of computer science, NP-Complete. Moreover, we show that this problem is hard (NP-Complete) for every positive value of r.

The implication is that fact-free learning is really difficult for computers. And if it’s difficult for computers, it’s probably really difficult for people too!

We present a model of coarse thinking, in which individuals group situations into categories, and transfer the informational content of a given message from situations in a category where it is useful to those where it is not. The model explains how uninformative messages can be persuasive, particularly in low involvement situations, and how objectively informative messages can be dropped by the persuader without the audience assuming the worst. The model sheds light on product branding, the structure of product attributes, and several puzzling aspects of mutual fund advertising.

The abstract of Sendhil Mullainathan, Joshua Schwartzstein, and Andrei Schleifer’s recent paper entitled Coarse Thinking and Persuasion.

The paper produces some comforting results:

The model sheds light on a number of phenomena. Most importantly, it explains how uninformative persuasion can be effective, especially in low involvement situations, such as evaluating cheap goods or political candidates. The model also helps understand the pervasive phenomenon of persuaders’ omitting bad payoff-relevant news from their messages. Both uninformative persuasion and omission of data are possible in our model even if the audience takes into account the strategy of the persuader.

Hopefully heuristics and persuasion literature will become more popular in coming years.