has been rather harsh on efforts to base macroeconomics on models of rational forward looking utility maximisers in Nash or Walrasian equilibrium recently.
One might almost think that I agree with my PhD supervisor Larry Summers who once said and I quote from memory (but believe me the words had enough of an impact that I remember them and where we were and who else was there).
“You know what I think ? I think that in the history of economic theory the day in which they came up with the idea of the utility function wasn’t an especially good day.”
To make things clear, I do think that major insights have been obtained by economists who introduced the concept of the utlity function not just to economics but also to macroeconomics. The problem is that I almost think that those economists are named either Milton or Friedman. After the jump, I will attempt to argue that the concept of the utility function is sometimes useful and sometimes even useful to people trying to study aggregates.
First of all, context. Larry was proposing an effort to explain the S&P 500 index using models of deterministic chaos. Existing models with chaotic dynamics had no forward looking element — they were models of weather or insect populations and no one assumed that either clouds or grasshoppers had rational expectations.
So I read up on deterministic chaos and programmed up the models he considered. They generated time series. The time series didn’t look like the S&P because they were highly predictable. I concluded that you really have to consider agents who look at the data and try to predict the future and can tell if there is a simple pattern. My version is that these agents eliminated the simple cycles and left the chaotic dynamics because they couldn’t figure them out. But see I was back to agents with an objective who were at least learning.
Of course my discovery can be found in basically every introductory micro textbook.
OK so what about Milton and Friedman. I’d say he had two major contributions which have something to do with considering economic agents’ objectives. One is the accelerationist critique of the Phillips curve (see “The role of monetary policy” a presidential address in the AER). Here he says that the old Phillips curve which related nominal wage growth to unemployment and nothing else couldn’t possibly be an economic law true in all times and all places, because incentives all concern real wages and it’s just not possible that people won’t notice a constant inflation rate of x%. Note, this is a critique of the totally stubborn and dumb expectations hypothesis. You don’t have to assume rational expectations to conclude that expectations have something to do with reality.
Good point. Definitely a true claim as the Phillips curve fit UK data and, as noted by Samuelson and Solow US data but didn’t fit German data from 1918 through 1924 so well.
Ned Phelps made the same point, so I have to admit that the concept of the objective function was put to good use by macroeconomists whose last name begins with the same sound as the phrase “false dawn”.
then there is the permanent income hypothesis. Here the glass is definitely at least half full. It is really true that economists struggled to explain strange patterns in the data, all of which seemed to be not in the slightest strange once Friedman and Franco Modigliani (OK macroeconomists one of whose names begins with the same sound as the phrase “false dawn”) began thinking about consumption savings choice as a problem in optimization.
For example, why is it that the rich save a larger fraction of their income than the poor yet as the economy grows and people on average get richer the ratio of consumption to income stays about the same (or later grew) ?
Why do African Americans have a higher savings rate than European Americans with the same income ?
Why is there a lower regression coefficient of consumption on income for US farmers than for other Americans ?
All these facts are exactly what you would expect if at least part of consumption corresponded to consumption in the PIH.
Now when one attempts to test the PIH (is the glass fully full) one discovers the same 2 things one always discovers. When on attempts to test hypotheses based on rational utility maximization one rejects them, but one always has to make auxiliary hypotheses and someone will argue that the rejection was due to the auxiliary hypothesis.
The rejection of the PIH with rational expectations is especially robust as those people will argue that rejection is due to an auxiliary hypothesis, because they can relax that assumption and get to a model in which the null is rejected at the 5% level but not at the 1% level.
But one can also reject the null that the PIH has nothing to do with consumption along with auxiliary unless and until someone finds a way to explain the stylized facts listed above, without assuming that at least part of consumption of at least some people corresponds to PIH consumption.
Milton Friedman has many disciples. Have they added a third victory based on applying micro theory to macro ? This is not a rhetorical question, I am asking for information.