Dr. Smith and the Asset Bubble

I sometimes hear professors (including myself) bemoan our lack of power. Then I remind them that we have the amazing power to mint A’s–a power many undergraduates would love to have. It is just that in an institution well-constructed to minimize self-interested decisions, one doesn’t get to have the power one really wants, say the power to set one’s own salary –but some other power, such as the burdensome power and duty to assign grades.

Another–but this time totally welcome–power that professors often have is to grant Ph.D.’s.  20 hours ago, after Noah’s dissertation defense, Bob Barsky (sadly for me, now of the Chicago Federal Reserve Bank), Yusufcan Masatlioglu, Uday Rajan and I drew up the documents that will grant Noah Smith his Ph.D.  

As has been the tradition in economics at least since I was in graduate school a quarter century ago, Noah’s dissertation has three chapters, each of which it is hoped will, within a few years (after the usual many rounds of revisions demanded by editors and the referees they enlist) become an article in an economics journal. Noah’s first two chapters are on asset bubbles–rapid, and somewhat mysterious, rises and then falls of the price of some asset, such as the rise and fall of internet stock prices around the turn of the millennium and the more recent rapid rise and fall of home prices. Noah describes the research in his first chapter and job market paper “Individual Trader Behavior in Experimental Asset Markets” in his post “What are asset bubbles and why do they happen?” The remarkable thing in “the literature” (the array of academic papers) on asset price bubbles that Noah adds to is that asset price bubbles happen even in the laboratory of a stock trading game with real people in which everyone is told exactly how the payouts of a “stock” are determined, say by flipping a coin to determine whether the “dividend” is 10 cents or zero each turn. So it is almost inevitable that asset bubbles will happen in the real world, which is a lot harder to figure out. To some, this may seem an obvious point, but a large share of opinions that economists give about the financial markets are based on intuition from theoretical models in which asset bubbles never, ever occur, because the “agents” in the models–who are intended as stand-ins for real people–are too smart.

Let me be clear. Studying models of financial markets in which agents are too smart for asset bubbles to ever occur is a worthy, and even at times noble, endeavor. But when it comes to real-world policy decisions, economists need to bring to bear everything we know about the world, even things that we have had a tough time analyzing in formal economic models. And we need to keep in mind especially those facts about the world that remind us of our own ignorance.  When I took a first-year macro class in graduate school from Larry Summers, Larry had a practice of presenting a model and then promptly proceeding to tear it down. I went up after class to complain that this was very discouraging. Larry Summers said something I have never forgotten: “It isn’t easy to figure out how the world works."  Larry Summer’s maxim

It isn’t easy to figure out how the world works

is an excellent mantra for an economist to repeat to himself or herself several times before giving policy advice.

The reason it is nevertheless OK to study models that leave out important facts about the world is that as economists, we need to walk before we can run, and models in which the agents are very smart are a lot easier to analyze than more realistic models in which some agents are very smart and others do strange, not-so-smart things. (I was tempted to put in the word "suboptimal” in place of “not-so smart.” “Suboptimal” is a word economists use for an action that is not the best action one could be taking from the standpoint of one’s own objectives.) Brad Delong, Andrei Shleifer, Larry Summers and Robert Waldmann published a more realistic model like this in 1991 that was a huge advance. But it is a baby model of this type compared to the complexity of the real world. 

One reason that (although often forgivable in an academic paper), assuming people are smart enough to avoid bubbles is at violence with reality is that, as Noah’s research shows, it is not enough for each trader to understand asset payouts herself or himself. The suspicion that other traders don't understand can lead to speculative trading. Also, Noah argues that people can understand what is going on one minute, but then doubt themselves the next minute when they see asset prices that suggest other people don’t agree. All the changes to experimental setups that are known to do the most to extinguish asset bubbles have one common property: they not only help each trader to act in a smarter way, they also help make each trader believe that other traders will act in a smarter way. Here are some examples:

  1. Running the asset market over and over from the beginning with the same set of traders. (See Vernon Smith, Gerry Suchanek and Arlington Williams.)
  2. Calling the “stock” a “stock in a depletable gold mine,” which helps people understand that the total amount of payouts left will go down as the game draws to a close. (This comes from an unpublished working paper “Thar she bursts–a critical investigation of bubble experiments” by Michael Kirchler, Jurgen Huber and Thomas Stockl.) 

A similar point can be made in relation to Noah’s second paper on asset bubbles in the lab: “Private Information and Overconfidence in Experimental Asset Markets.” This paper documents that investors not only trade on their own inside information, but also make trades with other investors who have insideinformation. (Noah’s research also shows that the amount of such trading is not closely related to existing measures of overconfidence. It isn’t easy to figure out how the world works.) In theoretical models, the super-intelligent agents realize that anyone who wants to trade with you should be looked at with suspicion: “What does he or she know that I don’t?” And in fact, the insight that you should wonder what the other guy knows that you don’t is important. Let me quote Noah’s comments on the intriguing paper “Are investors really willing to disagree: an experimental investigation of how disagreement and attention to disagreement affect trading behavior” by Jeffrey Hales:

Finally, a very interesting experiment by Hales (2009) investigates the closely related question of whether traders “agree to disagree” about the value of an asset. In his experiment, subjects trade in pairs and receive private signals about asset value. Hales finds that whether traders are prompted to consider the adverse selection problem has a strong effect on whether trade occurs. When traders are asked to guess the difference between their own signal and the signal of the other trader, trade tends not to occur; however, without such prompting, trade does tend to occur. This result suggests that over-reliance on private information is not due to traders “agreeing to disagree,” but simply to their failure to consider the fact that others have information that may disagree with their own. [Emphasis added.]

This intervention (asking people to focus on the difference between their signal and the signal of the other trader), like interventions 1 (repeating the game) and 2 (calling the stock a stock in a depletable gold mine) above, gains a lot of its power from changing what each trader visualizes is going on in the heads of the other traders.

Noah’s third chapter is joint work with Bob Barsky and me: “Affect and Expectations.” (It is common for dissertations in economics to include one out of three chapters that is joint work with advisors.) This paper shows that it is misnomer to call consumer confidence indexes measures of “consumer sentiment,” if by “sentiment” one means something emotional. (To back up the idea that people really do call it “consumer sentiment” see for example the title of the wikipedia article on the University of Michigan Consumer Sentiment Index.) For a variety of purposes, I had arranged to have the University of Michigan Surveys of Consumers collect data on happiness from the people surveyed. Noah and Bob and I analyzed how movements in happiness compared to movements in expectations about the economy. We could measure a relationship, but it was very small. So movements in consumer confidence are not primarily about fluctuations in happiness. And it is hard to think of many emotions that don’t show up as having some effect on happiness, so it would not be easy to defend the idea that some other emotion is the primary mover of consumer confidence. People’s confidence in the economy does go up and down in a big way. But something that doesn’t show up in happiness movements–and so something that is probably not emotional in the usual sense–is the primary cause of fluctuations in consumer confidence.

It isn’t easy to figure out how the world works.  (Larry Summers, 1984.)

Update: Ever since I wrote this post I have been wracking my brain trying to remember whether Larry Summers told me “It isn’t easy to understand how the world works” as in an earlier version of this post, or “It isn’t easy to figure out how the world works.” At first, I dismissed “figure out” because that is a favorite phrase of mine and I feared I had substituted that into my memory. But I keep coming back to “figure out.” And it occurred to me that maybe one of the reasons I like “figure out” may be precisely Larry Summers using it in this context. In any case I have decided to edit the repeated quotation to the “figure out” form. If anyone knows from Larry’s speech patterns which he would have been more likely to say, let me know.  

By the way, this is a good place to let readers know that I have enough of a random, patchwork, perfectionistic impulse that I routinely allow myself silent edits in my blog posts without an update notification. I am not running for office, I am not on trial, and I already have tenure, so I don’t have to play the game of “gotcha.” In my case, it is my words that matter, not me, and the words that matter are the ones I am willing to stand behind in the end.

But in this case, the “It isn’t easy to understand how the world works” version of the quotation has already gone out into the world and even become the title of a blog post by Kevin Grier–“Angus” at Kids Prefer Cheese, so I feel I need to clearly signal this particular update.  On other occasions, clearly signaling an update helps to drive home the overall point of a post. And if I ever write on some other website, I will follow the norms there.