Written January 6, 2011

The Allied Social Science Associations meetings are currently being held in Denver. These meetings are the largest set of Economics meetings on the planet. At 2:30pm today, I had 65 different sessions to choose from! And there are a similar plethora of sessions to choose from at nine different time-slots, three slots per day over the course of three days. The choice problem is considerable given that there are usually many interesting sessions at any time-slot and the attendee is forced to choose the “best of the best”. Every once in a while I get to pat myself on the back for making the right decision, and today is such a day. Of the 65 sessions available, I chose a Rational Expectations retrospect and prospect panel discussion and it was fantastic.

Rational Expectations was introduced to macroeconomic modeling in the 1970s. The idea behind Rational Expectations is that economic decision-makers making decisions about the future, who obviously face uncertainty, incorporate all relevant information into their decision-making process. They use this information to forecast what the future will look like, and when new information and/or events occur they update their information set and their forecasts. This updating process implies a corollary: that decision-makers do not systematically make errors.

The speakers included some of the great economists: Michael Lovell, Robert Lucas, Dale Mortenson, Robert Shiller, and Neil Wallace. Some of the discussion was about the historical precedents and developments surrounding Rational Expectations which I will not discuss here.

I came away with impressions that I do wish to share. I was very impressed by Robert Lucas. A parenthetical comment: the Rational Expectations research agenda and related policy prescriptions are usually in disagreement with Keynesian research and policy prescriptions. In the 1970s, those disagreements were extensive, and Mr. Lucas was in the middle of this debate as a pro-Rational Expectations researcher. Macroeconomic theory at this time still contains this disagreement although most would agree that the Rational Expectations side is dominant in academia. This is a source of angst for the Keynesians.

Despite the conflict, Professor Lucas described what was important about the Rational Expectations research program without using it as a club to beat those that he might disagree with. He resisted overly simple characterizations that were suggested by the moderators. Another parenthetical comment: the moderators did not ask unreasonable questions: they were natural ones given that Rational Expectations has powerful and controversial public policy implications. Lucas explained to the audience how certain Rational Expectations research efforts made important contributions at the time but were not so broad such that sweeping generalizations could be made. He also implied that Rational Expectations will not solve or explain every aspect of economic behavior, and that the power of Rational Expectations was that it was a more rigorous model of expectations than some of its predecessors and counterparts.

Listening to this panel provided me with a better understanding of the differences of opinion within the Economics profession about Rational Expectations. This will be a characterization that Professor Lucas might find over-simplified, but it aids in understanding a complex topic.

I start with a definition: Knightian uncertainty, named after an economist named Frank Knight, is an uncertain event that the decision-maker does not know could even happen. This is in contrast with uncertainty, which consists of a number of possible future events that are all known to be possible, where what is unknown is the probability of each possibility. Mr. Knight labeled this type of uncertainty risk.

Now suppose that there is a situation where Knightian uncertainty exists. Those economists who disagree with the Rational Expectations paradigm would say that this proves that Rational Expectations are inadequate, because the decision-maker had no way of forming a “rational” forecast.

Those economists who subscribe to the Rational Expectations modeling paradigm would say that once a previously unknown event occurs, then economic decisions from that point forward would incorporate the possibility of such an event. This would be tantamount to a change, an improvement, in the model and could result in a change in the model’s implications.

I see both sides of this argument and to me the difference of opinion here is akin to a semantical argument. They are talking about the same thing but disagreeing on how to interpret the implications. It might boil down to this: if Knightian uncertainty occurs often then the decision-making process will always be in a state of revision (an objection of those who do not like Rational Expectations) and if it occurs infrequently then the decision-making process will be easier and we can see how people could do this on average without making systematic errors.

Another disagreement is that Rational Expectations detractors argue that people are not the efficient information gathering and forecast updating machines that Rational Expectations theory argues they are. Proponents argue that all that is needed for the Rational Expectations theory to go through is that decision-makers update their information sets and forecasts often enough that this is the relevant behavior “on average”.

This is a more substantial disagreement than the one over Knightian Uncertainty and should be a key area of research. There might be empirical analysis that could shed light on this question. This could consist of devising tests of whether Rational Expectations holds across various situations where some decision-making situations were ones known to be easy to gather information and then others with progressively more difficult information gathering situations. Rational expectations should hold more easily when information gathering is easier and then perhaps it may not hold in situations where information gathering is highly difficult. A better understanding of this would be a relevant contribution to economics and could help resolve the debate.