Wednesday, March 12, 2008

Prices are not rigid

Central banks, and in particular the US Federal Reserve, are pumping liquidity in the market like there is no tomorrow. These money injections are only going to have a real impact if prices are rigid, that is, the additional money is not going to be inflated away immediately. The trouble is, prices are not rigid.

The best study tackling price rigidity in the data is the one by Bils and Klenow, 4th most cited recent article in Economics. They use the monthly data the BLS compiles for the consumer price index. They show that on average, prices last only 4.3 months. If you take away promotions, they last 5.5 months. The article actually gives the results for 350 different types of goods, with the most volatile prices being gasoline, tomatoes, airfare, natural gas, lettuce, eggs, car rentals, girls' dresses, oranges and chuck roast, the last changing on average every 1.3 months. The most rigid? Coin-operated laundry, vehicle inspection, driver's license, mass transit, car registrations, legal fees, vehicle tolls, safe rentals and newspapers, which change prices every 30 months or so.

This study is based on data from 1995-97, when inflation was in the 2-3% range. Since then, price flexibility can only have increased, as scanners are now commonplace in retail, price changes are distributed electronically, and inflation expectations are higher. What do I take from this? If there is a real impact of monetary injections, it will be limited to a couple on months. Then inflation hits. Is it worth the effort? I say no.

PS: Now relax with the latest from Improveverywhere.


Gabriel M said...
This comment has been removed by the author.
Gabriel M said...

Seriously now, what about this:

The point there is that prices are inflexible in a way, a way that's different from what's traditionally assumed. So maybe that's a "third way".

Economic Logician said...

Gabriel: The Eichenbaum , Jaimovich and Rebelo paper is the last in a series of papers that try to deal with sales and promotions in scanner data. As the others, it highlights that prices change a lot. It also shows that for some goods, they bounce between a few reference prices.

The real question is whether the frequency where they reach the various references prices changes, or if they just follow regular cycles. This is much harder to figure out. Many goods have a substantial fraction of their turnover during sale events, so sales are still needed. But one can imagine that if prices need to be higher, such sales are less frequent and more difficult to take advantage of.

They argue that the fact that reference prices do not change a lot is evidence of price rigidity. I am not so sure about it.

Also, this data does not take into account coupons and how shoppers actually shop. The collected prices are for a particular good in a particular shop. People buy wherever they like to. The prices people actually face is even more flexible than that.

Anonymous said...

Even if you claim that, in the data, prices have a duration of a year, which is really stretching it, no sensible model can give any significant impulse from monetary policy anyway. Looking for significant price rigidity is a lost cause.