Economists Love The Spread

The spread is a debating tactic where you present a set of supporting arguments so wide and particular your opponents are unable to rebut them all because 1) they have day jobs and 2) they have limited space or time to address them in any particular forum.  A champion spreader is Noam Chomsky, who selectively recites facts of world history from Indonesia, Russia, to El Salvadore and 200 places in between, which no one but a professional in Comparative Economic Systems would be familiar with (indeed, comparative economic systems was a flourishing economic subdiscipline precisely because it couldn’t be easily refuted because those communist countries didn’t have a free press and made up their production data, but after 1989 it was obvious this field was simple wishful thinking dominated by deluded Marxists).

So, a paper by Michael Woodford calling for NGDP targeting is a case study of a good type of argument for this kind of thing. As it is doubtful that less than 1% of his readership understands all the strengths and limitations of the model he presents, it makes for a great spread argument by Paul Krugman, because he can be sure that most readers won’t have the time, inclination, or ability to assess the credibility of this paper. As with most spread arguments, they can be used by those with the opposite conclusion (eg, John Cochrane here), because a small tweak makes it support an opposite view, and who’s to say what is minor when references are really recondite.

The model used by Woodford is from an older paper by Eggertsson and Woodford (2003). Kids Prefer Cheese notes this following set of quotes from that paper:

For simplicity we shall assume complete financial markets and no limits on borrowing against future income.

Our model abstracts from endogenous variations in the capital stock, and assumes perfectly flexible wages (or some other mechanism for efficient labor contracting), but assumes monopolistic competition in goods markets, and sticky prices that are adjusted at random intervals in the way assumed by Calvo (1983), so that deflation has real effects.

We assume a model in which the representative household seeks to maximize a utility function Real balances are included in the utility function, following Sidrauski (1967) and Brock (1974, 1975), as a proxy for the services that money balances provide in facilitating transactions.

In other words, the basic workhorse of this model was created decades ago and has been available to thousand smart economists trying to forecast the macroeconomy. If some metric of monopolistic competition, sticky prices, inflation, and real balances, predicted future investment and consumption, we would know it because there’s a model that targets these variables directly, and whether a parameter is 7 or -0.3 is simple enough to change after the fact: if some parameterization worked, the model could rationalize it. There isn’t such a model, as evidenced by the fact that all recessions have surprised macro economists in real time.

Now, these are smart people, so they are very good at explaining the past, but that’s really unimpressive, and reflects the degrees of freedom available relative to the target variables of quarterly GDP aggregates. It’s like finding a trading rule for last year’s daily S&P moves.

Don’t be fooled by The Spread in economics.  All macro models are no more precise than the infamous Laffer curve, which while true at extremes and useful for some intuition, says very little about your average policy change and intermediate forecasts. Outsiders can’t critique such models directly, but they can say, “what did this model say in 2007?  In early 2009?”  That is, not what does the model now say about 2008 given 2007 data, but what did it actually say in 2007?