Wednesday, February 13, 2008

Time to ditch Calvo Pricing

Guillermo Calvo published in 1983 an influential paper that managed to take into account that not all firms adjust prices immediately and at the same time in response to shocks, in particular monetary shocks. This way of introducing price rigidities has now been used in countless models but it always struck me as incoherent and silly.

Let me be precise: Under Calvo Pricing, every firm has a probability of changing its prices. Thus, a firm that has just adapted its prices has the same probability of adapting prices again as a firm that has not changed for a long time. Furthermore, this probability is invariant in the inflation rate. There is no way in hell this makes sense. Why would firms randomize over when they change prices? Why would they stubbornly not change them when it is high time to do so? And then change them minutely in rapid succession? And why would this strategy not change of the inflation rate increases?

Mikhail Golosov and Robert E. Lucas, Jr. take this idea seriously and build a model where firms have heterogenous prices, and they can decide to adapt them conditional on some menu cost. And surprise, surprise, monetary shocks in such an economy produce no persistent effects on real or nominal agregates. In other words, if previous models got persistent effects, it was due to the Calvo Pricing assumption, which does not make sense in the first place. It could still be used if it gave a reduced form for a more complex behavior of firms. But clearly, it does a very poor job at that.

Thus, please stop using Calvo Pricing. And stop teaching it as well.

PS: For a lighter moment, here is the daily link: How the Japanese learn English


Gabriel Mihalache said...

Yes, but aren't sS pricing rules hard to aggregate?

Nancy Stokey has some stuff on Brownian economies where firms set prices when there's a concave cost to it...

In the Calvo setup, with a single output good and no heterogeneity between firms, does it matter who does what at the firm level? -- You might as well impose a AR or MA process on prices and be done with it.

Economic Logician said...

I agree, it is hard to aggregate. But why go for the convenient/easy modeling if it leads to wrong results? There are certain times where you just have to bite the bullet and go the difficult way, and this seems like such a situation.

Anonymous said...

False. Not random decisions at all, but products of a stochastic Bellman Equation!!!!

You clearly didn't read the Lucas paper.

Anonymous said...

I strongly urge you to read the following paper if the Golosov, Lucas (2007) is your major concern with Calvo pricing:

So it appears that Calvo pricing is indeed a good reduced form way to model more complex behavior of firms.