Sticky Prices
Matthew Martin
9/04/2012 04:43:00 PM
Tweetable
To better explain how sticky prices work, consider a scenario in which consumers--for whatever reason--decide they don't want to spend as much (that is, they experience a negative preference shock). Nothing has happened to the technological capacity of the economy, so all that "should" happen is that prices fall enough to induce households to spend as much as they did before the preference shock. However, prices are sticky, so when this negative preference shock occurs, it takes a period of time--usually about six months to a year--before they can lower prices. In the mean time, households buy fewer goods and services, meaning that GDP falls below potential output until prices can adjust (or the preferences go back to normal).
But the reality is that sticky prices are really nothing more than a mathematica trick, not a realistic description of the economy. The best example of where they fail to explain reality is to look at the "New Keynesian Phillips Curve" that results from the stick price model. In the real world, we observe a positive relationship between inflation and unemployment, which Samuelson named the "Phillips Curve" after it was first documented by A.W. Phillips in the 1950s. The sticky price model does reproduce this feature. Sort of. Actually, sticky prices alone doesn't get you there.
To see the problem with sticky prices, consider a scenario in which the economy (currently at an efficient equilibrium) experiences a permanent increase in money supply. In the long run, this will result in higher prices, but it will take time before firms are able to raise their prices, as a result of sticky prices. Hence, in the short run household have more money to spend, but prices remain constant, leading households to increase the quantity of goods and services they want to buy. Now suppose that you are a firm producing those goods and services--prices are going up, but you are unable to raise your price along with them, in the short run. The optimal response, in that case, is to reduce your output to decrease marginal costs. Less output means less employment, so what we actually see in sticky price models in which firms behave in a profit-maximizing way is a negative short-run relationship between employment and inflation--the opposite of the positive relationship in the data.
To fix this problem, New Keynesian models place a constraint on firms' behavior: firms have to sell to anyone who wants to buy their product at the advertised price. It is hard to "microfound" this assumption. In the real world, firms sell out all the time, leaving people unable to buy at the advertised price. That is, the assumption is explicitly incorrect, even though it improves the ability of the model to explain the data.
I suspect that because of issues like this, sticky price models will eventually die out. Here's the essence of the problem: if firms really do have rational expectations, then the sticky price model requires that we assume firms are not perfectly profit-maximizing. I think a more realistic model would look a lot more like Friedman's version of the phillips curve--keep profit maximization, but dispense with rational expectations. That is, simply assume that firms increase output because, although they observe the increase in demand for their particular product, they do not perfectly observe the increase in inflation. In contrast to sticky prices which require vague, unfounded institutional constraints on both the firm's supply decision and their price decision, the Friedmanite phillips curve allows firms to behave in a profit maximizing way (at least in terms of their expected profits) with respect to both the price and output decisions, while still delivering the positive observed relationship between employment and inflation.