«Chapter 1 Why Study Price Stickiness? Why This Way? Nothing astonishes men so much as common sense and plain dealing. —RALPH WALDO EMERSON The ...»
Why Study Price Stickiness?
Why This Way?
Nothing astonishes men so much as common sense
and plain dealing.
—RALPH WALDO EMERSON
The Importance of Price Stickiness
In recent decades, macroeconomic theorists have devoted enormous
amounts of time, thought, and energy to the search for better
microtheoretic foundations for macroeconomic behavior. Nowhere
has this search borne less fruit than in seeking answers to the following question: Why do nominal wages and prices react so slowly to business cycle developments? In short, why are wages and prices so “sticky”? The abject failure of the standard research methodology to make headway on this critical issue in the microfoundations of macroeconomics motivated the unorthodox approach of the present study.
No one should think the question unimportant. On the contrary, sticky prices are an essential element of Keynesian economics, which is sometimes called the economics of nominal rigidities. (It is called less polite things as well.) The sobriquet is an exaggeration, to be sure, but a forgivable one, for embedding the assumption of short-run price or wage rigidity into almost any macro model will make it produce characteristically Keynesian results—such as that an injection of money raises production. The simplest illustration is
the quantity theory of money with ﬁxed velocity:
In the absence of nominal rigidities, real output, y, is essentially fixed on the supply side of the economy, so that changes in money 4 Asking About Prices must pass directly and proportionately into prices. But if P, the price level, is sticky in the short run, the very same equation implies that part of any change in M must first show up in y. Conversely, if a vertical aggregate supply curve (attributable, say, to instantaneous market clearing) is appended to an otherwise “Keynesian” IS-LM model, the real effects of fiscal and monetary policy disappear.
When a scientiﬁc discipline knows next to nothing about a question of paramount importance, it is in some trouble. How did macroeconomics get into such a predicament? Two obvious explanations can be dismissed immediately. First, it is not because macroeconomists have just discovered that wage-price stickiness is a central issue; we have known this since Keynes’s General Theory, if not before. Second, the failure does not result from lack of effort.
Scores, if not hundreds, of theorists have worked on this problem, producing many interesting theoretical explanations; and new ideas keep popping up all the time.1 Progress has not been hampered by lack of imagination.
Nor, by the way, has it been hampered by lack of observation.
Although much time and energy was wasted in the 1970s and 1980s arguing over whether or not the economy should be modeled as a giant auction hall with perfectly ﬂexible prices, a small mountain of empirical evidence testiﬁes to the fact that wages and prices adjust slowly to macroeconomic events. For example, the economist Robert Gordon (1990) summarizes the evidence that aggregate price indexes move sluggishly, while Yoram Weiss (1993) surveys some papers that provide similar evidence for individual prices. The tricky questions are two: How slow is slow? And what factors account for the sluggishness? This book is devoted to the second of these two questions. But a brief word on the ﬁrst is in order, for it helps explain why conventional methods of economic inquiry—theory and econometrics—have yielded such meager results.
Why So Little Progress So Far?
When we say that wages or prices are “sticky,” we generally mean that they move more slowly than would Walrasian market-clearing prices. Two curmudgeonly questions arise, each of which inﬂuWhy Study Price Stickiness? 5 enced the design of this study: Is the statement operational? And is it something we should care about? We take these up in reverse order.
Is Wage-Price Stickiness an Important Phenomenon?
The question here is basically whether wage-price stickiness has allocative signiﬁcance. For example, if no one ever borrowed at the credit card interest rate, which remained around 19 percent for years, then the fact that this rate was extremely sticky would hardly have mattered.2 Ever since the economist Robert Barro’s (1977) ingenious paper, macroeconomists have worried that the sticky wages we see in many labor markets may in fact lack allocative signiﬁcance. Speciﬁcally, ﬁrms may not equate the real wage to the marginal product of labor and workers may not equate the real wage to the the marginal utility of leisure. Why? Because, it is argued, employees and employers implicitly enter into long-term agreements to exchange labor for money. These contracts clear the labor market in a long-run sense, thereby tying long-run average labor supply closely to long-run average wage payments. But if workers dislike wage variability, the contract may pay steady wages month after month even if both the marginal product of labor and actual hours of work vary considerably over time.
This argument has persuaded many economists that it is hazardous to conclude from the observed stickiness of wages that there is pervasive disequilibrium in labor markets. It also calls into question the allocative signiﬁcance of the limited variability of real wages over the business cycle. But is the argument empirically important? Do ﬁrms and workers actually enter into such agreements? No one really knows. When Alan Blinder and his student Don Choi (1990) asked a small sample of personnel managers what they thought of the implicit contract theory, the results were a bottle half full and half empty.3 About half thought it ”plausible or relevant,” the rest did not.4 Irrespective of its empirical relevance, the theory has profoundly inﬂuenced the thinking of academic economists and has shifted the focus of research from wage stickiness to price stickiness. When Blinder initiated this research, he knew he would be preaching to the unconverted and was anxious to have at least some economists pay at least some attention to the unorthodox research ﬁndings. So 6 Asking About Prices he decided to study only price stickiness. Of course, the implicit contracts argument can be applied to prices, too. Sticky prices may just be installment payments on long-term agreements, rather than symptoms of non-clearing product markets. But everyone seems to agree that, however important or unimportant the implicit contract theory is in labor markets, it must surely be less important in product markets, where arms’-length, spot transactions are much more common.
Is Wage-Price Stickiness an Operational Concept?
The next question is whether and how we can breathe empirical life into the theoretical notion of price stickiness. In brief, how slow is slow? To state the issue perhaps a bit too boldly, a theory that predicts that prices adjust more slowly than market-clearing prices—and nothing else—is basically untestable, and therefore an empty theory.
Why? Because economists have no agreed-upon metric to use in assessing the observed speed of adjustment of any particular price, let alone the aggregate price level. So if we ﬁnd, say, that the price of candy bars changes every six months, how do we know whether this is slower or faster than the Walrasian norm?5 If a theory makes no prediction other than that prices move less rapidly than Walrasian prices, econometric testing is almost (but not quite) out of the question. To conduct a test, a complete model of supply, demand, and price adjustment must be speciﬁed, estimated, and used to derive a quantitative measure of the speed at which the market-clearing price moves. Then actual price movements can be compared to this norm. This research strategy is not a counsel of perfection; it can be implemented. Indeed, it is one of the ways that econometricians have demonstrated that prices and wages are sticky. But, of course, any such demonstration is conditional on the validity of the many maintained hypotheses used as the framework for estimation. So any such ﬁnding is open to dispute.
But the problem goes deeper than this. If we have a wide variety of models, each of which predicts that prices are sticky and nothing else, conventional econometrics will have a hard (if not impossible) time distinguishing among them, for there is no way to test one theory against another. This problem, we think, is the main reason why formal econometrics has made so little progress in weeding out invalid theories of wage-price stickiness. It is one of Why Study Price Stickiness? 7 the two problems that originally drove Blinder—in desperation!— to the interview method.
The other problem is that several of the most prominent theories of price stickiness rely on variables that are either unobservable in principle or unobserved in practice.
One example is a theory that achieved wide popularity in the 1980s, which holds that ﬁrms hesitate to cut prices in slumps out of fear that customers will misinterpret any price cut as a reduction in quality—when in fact there has been no such quality reduction.6 Notice that unobservability is crucial to the argument; if quality were easily observed, there would be no possibility of misinterpretation because everyone would recognize when quality had changed.
Another example is the “menu cost” theory, which says that ﬁrms change prices infrequently because they incur a ﬁxed cost each time they do so.7 In principle, such costs can be measured. In practice, however, we have few such measurements and are unlikely to get many.8 It might seem that this particular theory does at least carry a clear collateral implication—namely, that ﬁxed costs preclude small price changes, where the precise meaning of “small” is deﬁned by the size of the menu costs. In principle, that is correct.
But our inability to measure menu costs directly robs this implication of operational signiﬁcance. Will the ﬁrm avoid 1 percent, 5 percent, or 10 percent price changes?
When theories rely on unobservables in essential ways, econometric testing is difﬁcult, to say the least. Thus it is no accident that new theories of price stickiness have continued to proliferate faster than applied econometrics has been able to discard old ones. It was that unsatisfactory state of affairs that ﬁrst set Blinder thinking about an alternative approach.
Time for a New Approach?
In pondering this dilemma, a curious “empirical regularity” emerged. Virtually every theory of price stickiness outlines a thought process that allegedly leads decisionmakers (generally modeled as proﬁt maximizers) to conclude that it is against their best interest to change the price. But if people actually think the way one of these theories says, then they should be aware that they do—or so it seemed. Hence an idea: Why not ask them?
8 Asking About Prices This naive idea must be approached with caution. If you confront a price-setter with an open-ended question like, “Why don’t you cut your prices more (or more often) when sales sag?” you may get shrugs, blank stares, or incoherent answers. What you hear is unlikely to fit neatly into economists’ theoretical boxes.
But suppose you ask more pointed questions. Suppose you describe in plain English the chain of reasoning that, according to Theory X, goes through the minds of price-setters. If Theory X really describes their behavior, the decisionmakers ought to recognize and resonate to it. If they do not, then they are probably not behaving as the theory says. At least that was our methodological precept. If the true reasons for price stickiness are buried deep in the subconsciousnesses of decisionmakers, then interviews are unlikely to uncover them.
Here are two examples of the rather pointed way in which we posed questions about the theories.
One very old theory of why prices may be rigid over the business cycle, which enjoyed a strong revival in the 1980s, starts with the premise that proﬁt-maximizing ﬁrms with market power set price (P) as a markup over marginal cost (MC), which markup 1).9 Thus:
depends on the elasticity of demand (
P MC[ /( 1)],
where is deﬁned to be a positive number. The theory then asserts that demand curves become less elastic as they shift in so that, even though MC falls as output contracts, the markup rises to compensate. The result may be approximate constancy of the proﬁt-maximizing price over the business cycle.
In principle, this theory can be tested by conventional econometric means; all we need do is measure how the elasticity of demand varies over the business cycle in a variety of industries. In practice, however, any applied econometrician will recognize that as a tall order, unlikely to be ﬁlled with the nonexperimental data at our disposal. But now think about using the interview method as an alternative to time-series econometrics. If this theory is the real (or one real) reason for price rigidity, ﬁrms must both believe that their demand elasticities are procyclical and act on that belief.
In that case, if you ask them about it—eschewing jargon like “elasWhy Study Price Stickiness? 9 ticity,” of course—they ought to recognize the idea and feel comfortable with it.
To test this theory, our questionnaire posed the following plainEnglish question:10 B5(a). It has been suggested that, when business turns down, a company loses its least loyal customers ﬁrst and retains its most loyal ones. Since the remaining customers are not very sensitive to price, reducing markups will not stimulate sales very much. Is this idea true in your company?
If the respondent answered yes, we then asked:
B5. How important is it in explaining the speed of price adjustment in your company?
To preview some results that will be examined in more detail later, almost 60 percent of respondents accepted the premise that the elasticity of demand varies procyclically. But only about half of those (hence, about 30 percent of all ﬁrms) rated it a “moderately important” or “very important” source of price stickiness.
Our second example, the Okun (1981) “invisible handshake” theory, is an even more extreme example in that direct econometric testing seems out of the question even on conceptual grounds.