1
Introduction

Why do the prices of some products fall little or even rise during business downturns while the prices of others fall dramatically and rebound during economic booms? It is not surprising that in highly competitive industries prices fluctuate with shifts in demand and supply, but what explains the dearth of price declines in markets where firms have more direct control of prices? This question is central to an understanding of business cycles. Although economists have proposed many explanations of downward price rigidity, no widely applicable theory has firm empirical support. Perhaps this is so because such support requires knowledge that is difficult to obtain, such as understanding the objectives of price setters and the constraints they face.

Hoping to find insight into price formation and rigidity, I imitated Alan Blinder and his coauthors (Blinder et al., 1998) by interviewing businesspeople who participate in price setting.1

Blinder and his associates used interviews to evaluate empirically twelve theories of price stickiness. The interviewers asked many questions, but their core approach was to describe each of the theories to respondents and ask them to assess its applicability to their company. On the basis of the answers, the authors rated on a numerical scale the importance of each theory as an explanation of price stickiness at that firm. The average of all the scores from all the interviews is a measure of the overall relevance of the theory. Blinder and his coauthors randomly chose the companies to be solicited for interviews, where the probability that a company was chosen was proportional to its value added.

Although I found the work of Blinder and his coauthors to be useful, I did not follow them by asking respondents to comment on the relevance of particular theories of price rigidity. Instead, when arranging interviews I explained that I hoped respondents would tell me what I needed to know to understand pricing in their company and industry. Before interviews, I emailed respondents a list of questions designed to make clear what topics interested me. During interviews, I used what respondents came up with as a basis for further questioning. A disadvantage of this approach was that not all respondents dealt with the same topics, so that the incidence of responses of a given type may not be a reliable gauge of its importance. Interpretation must rely on learning how respondents think, on the logic in the responses, and on circumstances described in them. An advantage of the approach is that I learned things I would not have thought to ask about. One reason I avoided the approach of Blinder and his coauthors is that economists may have overlooked correct theories of price rigidity. In fact, Blinder and his coauthors did not ask about the explanation of price rigidity that I found to be the most widely applicable.2 Another reason for not asking about theories is that in my experience such questions can cause respondents subtly to stop cooperating, if they think a theory is silly or if they feel they are being drawn into intellectual competition with a professor.

Because of my interviewing method, I could not select respondents randomly. Businesspeople are reluctant to participate in loosely structured interviews, because they worry that they might inadvertently reveal confidential information or say something that would embarrass their company. So, I had to gain trust, which I did by using what is called the snowball sampling method. I started by interviewing friends and acquaintances and, at the end of every interview, I asked for referrals to other possible respondents, while indicating what kinds of companies and people interested me. I promised full confidentiality to everyone I asked for interviews. This approach was for the most part successful, though slow. Sometimes, more than a year passed before a company's lawyers agreed to let me interview in their firm. I had disappointments. For instance, I never penetrated the Internet commerce industry and I never had an interview with a plate-glass manufacturer. In requesting referrals, I sought variety in the types of businesses but also strove adequately to cover the main industries in the American economy and the most important companies within each. I did not record the number of interview requests that were refused, but believe I talked to key people in most of the categories of businesses I studied. Interviews usually took place in the respondent's office, or in a restaurant, and lasted about ninety minutes. Most interviews were tape-recorded and later transcribed, though some respondents refused to be recorded. I checked for accuracy all the transcriptions that I did not do myself.

One might expect in a study like this to read about negative productivity shocks and Federal Reserve Bank inflation targeting or forward guidance, because these topics are much discussed in macroeconomics. Neither topic ever came up spontaneously, except when a few respondents mentioned the effects of drought on agriculture and of low water levels on river and lake shipping. I occasionally brought the topics up, but eventually desisted, because respondents implied they were irrelevant to pricing. They may have reacted this way regarding Federal Reserve Bank policy, because there was little inflation at the time.

I came up with three main findings, each of which is a phenomenon, together with explanations of why it occurs. The phenomenon in one finding is that the prices of highly differentiated products seldom decline and tend to increase only sluggishly in response to changes in demand or supply, whereas the prices of undifferentiated products respond so quickly to such shifts that many are volatile. By a highly differentiated product I mean one produced by only one firm and that is not a good substitute for any other product. A strong brand, for instance, makes a product highly differentiated.

The phenomenon in another main finding is that marginal variable costs of manufacturing firms tend to remain constant or to decline as a function of output until capacity is reached, at which point marginal variable costs rise abruptly. This assertion may seem counterintuitive, since, presumably, as output increases in the short-run more labor is used with a fixed amount of capital equipment. The proposition that marginal variable costs may be constant over a wide range of outputs is not new. For instance, Robert Hall suggests this idea (Hall, 1986), and Blinder and his coauthors present empirical evidence supporting it (Blinder et al., 1998, ch. 12).

The impact of declining or constant marginal variable costs on the ability of manufacturing firms to make money creates a link between the behavior of marginal variable costs and product differentiation. Firms with constant or declining marginal variable costs lose money if they set price equal to marginal variable cost, unless they produce at a level so close to capacity that marginal variable costs exceed average variable costs. This predicament no doubt helps explain the drive of many firms to differentiate their products, because differentiation normally enables firms to charge more than marginal variable cost and hence perhaps to cover fixed costs and earn a profit.

Constancy of marginal variable costs suggests an explanation of price stickiness that should be considered. The explanation is that if a product's price equals a constant markup over marginal variable costs, then constancy of marginal variable costs implies price rigidity. The rigidity to be explained is the lack of response of price to the business cycle. If we assume that a firm's marginal variable costs are roughly proportional to the average wages and salaries it pays, then it is reasonable to assume that nominal marginal variable costs do not fall during recessions, since economy-wide averages of nominal wages and salaries seldom fall.3 The proportionality between the price of labor and marginal variable costs is, however, questionable, since variable costs in manufacturing are often dominated by the costs of variable factors other than labor, costs that may have little to do with labor costs. The assumption that prices are a constant markup over marginal variable costs should also be questioned, because profit margins often decline during hard times. A weaker form of this assumption may be approximately valid, however, because many manufacturing firms set the price of each of their products to be roughly equal to a markup over the sum of average variable costs, and an assignment to the product of a share of the firm's average overhead and fixed costs. Costs including this assignment are known as fully absorbing costs. If price is a markup over fully absorbing costs, then the markup of price over average variable costs is positive, even if no additional margin is added for profits. If we add the assumption that average variable costs are constant, then this line of thinking leads to a weak form of downward price rigidity; namely, that in a recession prices will not fall below some level that exceeds the constant level of average variable costs. This reasoning does not imply upward price rigidity, because wages and salaries are not upwardly rigid. A weakness of the theory is that some firms reduce the markup of price over average variable costs when demand slumps, by using what is called contribution margin pricing. A contribution margin price is one between average variable costs and fully absorbing costs.

The views of businesspeople do not support the attribution of downward price rigidity to cost-based pricing and constancy of marginal variable costs. None of my respondents came up with this idea...