Peers, & predictability

Peer effects matter for consumer spending. A new paper finds that "the consumption decisions of U.S. households are to a large extent driven by upwardly directed interpersonal comparisons." That is, we spend more if our neighbours are living it large.

This is not a new finding. It corroborates Robert Frank's theory of "expenditure cascades", evidence from the effects of the Dutch postcode lottery, and the large literature on the importance of peer effects.

This means Boris Johnson had a point when he said (pdf) that keeping up with the Joneses is a "valuable spur to economic activity." But it also supports those who blame the crisis (partly) upon inequality: people's efforts to emulate the spending of the rich led to them becoming over-borrowed.

But there's an angle here that is under-appreciated. Peer effects help explain why economic forecasting is a mug's game.

The argument for forecasting is that, across millions of people, individuals' whims cancel out. If this is so, the law of large numbers means we can ignore the crooked timber of humanity and think of a representative consumer responding predictably to changes in wealth, interest rates and such like.

But if peer effects are important, the law of large numbers no longer holds. If I buy a new car (I just have), my neighbours could so so as well. If so, my spendthrift urge ceases to be idiosyncratic. Instead, it's possible that whims can "go viral."

Now, you might object that, in practice, macroeconomic forecast errors are rarely due mainly to mispredictions of consumer spending relative to incomes. However, peer effects might also apply to corporate spending decisions. One reason for this is that, because bosses talk to each other at conferences, clubs and suchlike, firms' animal spirits are as likely to be at least as infectious as households'. Also, there's a business case for peer effects to hold; if a rival invests in cost-cutting equipment, you have to do so or else be competed out of business.

My point here is simple, and unoriginal - though not recognized by the swinish multitude which thinks economists should be in the forecasting game. It's that the economy cannot be treated as if it comprised merely a "representative" consumer or firm, but rather that interactions between agents matter and these mean that the economy is a complex emergent process which renders forecast errors all but inevitable. For this reason, we should not take seriously whatever forecasts the OBR makes on Thursday.