« Bryan Caplan: Special-Interest Secret | Main | Did Tax Cuts Fuel Economic Growth? »

Saturday, May 12, 2007

Sticky Prices: Differences in the Behavior of Aggregated and Disaggregated Prices

Tim Harford discusses sticky prices:

The Mystery of the 5-Cent Coca-Cola, Why it's so hard for companies to raise prices, by Tim Harford, Slate: The price of the first serving of Coca-Cola was 5 cents in 1886, about a dollar in today's money. Coke no longer sells for a nickel, and that is not terribly surprising. What is surprising is that it took more than 60 years for the price of Coke to change.

Economists call this "nominal price rigidity." My salary is not tweaked each month to reflect the latest inflation figures, and neither is yours. Restaurants do not reprint their menus, nor wholesale firms their catalogs, if the cost of their inputs changes by a penny.

That might be a problem. ...[P]rices need to be able to change relative to one another to reflect demand and the underlying costs of production. If prices don't adjust smoothly for any reason, then the economic consequences could be serious. If wages can't fall in a recession, then people will lose their jobs instead. If the price of a car or a restaurant meal can't fall when demand does, sales will collapse with much the same effect.

Coke was clearly an exceptional example of rigid prices. Daniel Levy and Andrew Young, the economists who analyzed the case, report that Coke's price stayed at 5 cents a serving while the price of other products bounced all over the place. ...

Part of Coke's problem was the cost of replacing vending machines that accepted only nickels—and the fact that the alternative, dimes, represented a 100 percent price hike. (The boss of Coca-Cola wrote to his friend President Eisenhower in 1953 to suggest, in all seriousness, a 7-and-a-half-cent coin.)

Most companies don't wait so long to change prices if they need to. Researchers have tended to conclude that many prices change every year or so, often sooner. Levy and some colleagues looked at supermarket pricing in the mid-1990s and found, based on detailed accounting data, that to change the price of a single type of product in a typical supermarket cost 52 cents in printing, labor, and errors. The total of all such changes was about $100,000 per store, per year—still less than 1 percent of revenue.

Technology makes it ever easier to change prices using bar codes, Web sites, and laser-printed menus. Amazon always seems to be changing book prices. Coke vending machines now take very little effort to reprogram. So, should we conclude that "menu costs" no longer matter?

That would be too optimistic. Economists have long argued that even small "menu costs" could cause large economic distortions... A prize-winning paper from Carlos Carvalho recently showed that it does not even help if many prices adjust quickly, because those that change slowly will distort the rest. ...

Let me add a bit from a paper I read not too long ago. According to this work, the degree of rigidity in prices depends upon the level of aggregation examined, the degree of monopoly power, and the type of shock hitting firms:

Sticky Prices and Monetary Policy: Evidence from Disaggregated U.S. Data, by Jean Boivin, Marc Giannoni, Ilian Mihov , NBER WP 12824, January 2007 [open linkAbstract This paper disentangles fluctuations in disaggregated prices due to macroeconomic and sectoral conditions using a factor-augmented vector autoregression estimated on a large data set. On the basis of this estimation, we establish eight facts: (1) Macroeconomic shocks explain only about 15% of sectoral inflation fluctuations; (2) The persistence of sectoral inflation is driven by macroeconomic factors; (3) While disaggregated prices respond quickly to sector-specific shocks, their responses to aggregate shocks are small on impact and larger thereafter; (4) Most prices respond with a significant delay to identified monetary policy shocks, and show little evidence of a "price puzzle," contrary to existing studies based on traditional VARs; (5) Categories in which consumer prices fall the most following a monetary policy shock tend to be those in which quantities consumed fall the least; (6) The observed dispersion in the reaction of producer prices is relatively well explained by the degree of market power; (7) Prices in sectors with volatile idiosyncratic shocks react rapidly to aggregate monetary policy shocks; (8) The sector-specific components of prices and quantities move in opposite directions.

The main contribution of the paper is to divide shocks into two types, economy-wide shocks that hit all sectors, and sector specific shocks that hit individual sectors, and to do so in an econometrically defensible manner.

In previous results, when shocks were not disentangled to separate economy-wide and sector specific shocks, empirical results tended to find a large degree of price stickiness at both the aggregate and sectoral levels lending support to sluggish-price, menu-cost* type models.

This paper follows in the path of previous work showing that individual prices, or at least highly disaggregated prices, are far more volatile, i.e. flexible, than typically assumed in sluggish price models.

Restating the results, when the authors separate shocks into aggregate and sector specific shocks, they find that:

1. Most of the price volatility at the sectoral level is due to sectoral shocks, not aggregate shocks.

2. Differences in price-setting behavior across sectors are mostly due to differences in the response to sector-specific shocks, most sectors respond similarly to aggregate shocks.

3. There is persistence (stickiness) in the response of disaggregated prices, but this is due to the response to aggregate shocks. There is little persistence in the response of disaggregated prices to sector specific shocks. In addition, the response of disaggregated prices to sector specific shocks is more immediate than the response to aggregate shocks.

4. The differences in the responses across sectors is explained by differences in market power. More monopoly power implies more stickiness.

5. There is no longer a price puzzle (the annoying tendency for prices to rise after the Fed tightens in empirical models used for policy analysis).

Overall, then, what this paper finds is that firms exhibit considerable flexibility in responding to sector specific shocks - there is not much price stickiness on display - but a much more sluggish response to aggregate shocks.

One thing that is needed is to explain these results theoretically. The paper outlines some recent attempts in this direction, but there is much more work needed on both the empirical evidence about price stickiness and on the supporting theoretical models.

*Be careful with the term menu costs - it includes all costs to the firm from changing prices, not just the literal cost of, say, printing new price lists. For example, customers may have a preference for prices that do not change often (for long-term planning, e.g.) and menu costs include any loss of sales from customers switching to firms who change prices less often, and with lots of advance warning.

    Posted by on Saturday, May 12, 2007 at 12:33 PM in Academic Papers, Economics, Macroeconomics | Permalink  TrackBack (1)  Comments (13)


    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Sticky Prices: Differences in the Behavior of Aggregated and Disaggregated Prices:

    » News of the World #34 from Economic Investigations

    Elsewhere Coercion Coercive Regulation and the Balance of Freedom, Edward Glaeser doesnt pay attention to Klein definition either and conflates coercion with use of or threat with force. Ed Glaeser on Utility, ... [Read More]

    Tracked on Sunday, May 13, 2007 at 03:22 AM


    Feed You can follow this conversation by subscribing to the comment feed for this post.