|
Article Excerpt
I. INTRODUCTION
Fashions that take hold of the field of macroeconomics share important characteristics with the economy its practitioners seek to understand: approaches that are in vogue exhibit cycles with varying frequencies and amplitudes. During the 1960s, theories based on assumptions of widespread wage and price stickiness dominated the landscape. The new classical macroeconomics that emerged in the 1970s challenged the preeminence of these Keynesian theories, and by the mid-1980s, real-business cycle theories assuming widespread price flexibility received a considerable portion of macroeconomists' attention. Beginning in the early 1990s, as reviewed by Gordon (1990), Mankiw (1990), and Mankiw and Romer (1991a, 1991b), a Keynesian counterrevolution occurred, and by the conclusion of the 1990s, sticky-price theories had regained much of their luster. Today, as discussed by Woodford (2003), price stickiness is a fundamental building block of numerous theories aimed at explaining both movements in domestic variables, such as employment and real output, and changes in international variables, such as the trade balance and the current account.
There has, as we discuss below, been empirical work seeking to determine whether real-world data might be consistent with the implications of macroeconomic theories with inflexible product prices. There has been very little work, however, to determine whether the most influential approaches to testing the reasonableness of micro-based models of price stickiness support the underlying assumption of widespread price rigidities at the micro level. Instead, leading studies in the literature examine only aggregate data. We find this quite surprising given that so many modern macroeconomic models consider economies populated with representative producers that establish an array of product prices that are fixed in the short run, which in most models is commonly taken to be the current period of analysis.
Are most product prices really sticky? Bils and Klenow (2004) have addressed this question by studying the frequency of price adjustments for 123 categories of consumer goods and services. Their answer is that most prices in fact change relatively frequently, with about half of prices changing roughly every 4 or 5 too. Although they found that the prices of certain services and what they call "processed" physical goods adjust more slowly, their overall results offer relatively little support for sticky-price models.
In this paper, we also examine microeconomic data for the United States. In contrast to Bils and Klenow, however, we examine industry-level data on firm prices and costs. We employ the empirical methodology proposed by the highly influential study by Gali and Gertler (1999), which in turn is based on the micro-based analysis of imperfect competition and price stickiness proposed by Calvo (1983). Nevertheless, our conclusions are broadly similar to those of Bils and Klenow. Overall, we find that across the industries to which we are able to apply the Gali-Gertler-style approach, no more than 43% of total sales are generated by industries with estimated periods of price adjustment of 6 quarters or more. This percentage drops to below 23% when the discount factor is restricted to reasonable values, and under these restrictions, industries accounting for at least 36% of all sales in the industries we analyze have estimated price adjustment intervals of less than 2 quarters.
In all empirical approaches we consider, unrestricted analyses producing estimates of industry-specific discount factors, at least 44% of total sales of the industries we consider are generated by industries with estimated price adjustment speeds of less than 1 yr. Furthermore, in most specifications, the unweighted average price adjustment interval for all industries is below 1 yr, and the weighted average--using industry shares of total sales as weights--price adjustment interval for all specifications ranges from no higher than 3.5 quarters to as low as 2.1 quarters.
Thus, while our analysis offers some degree of support for models that seek to explain macroeconomic phenomena using models hinging on price inflexibilities, it casts considerable doubt on the relevancy of theories that presume that all prices are sticky. The industry-level data we examine suggest that the majority of U.S. product prices adjust within 6 mo to 1 yr. Our estimates indicate that firms take well over a year to adjust some product prices, but firms take considerably less than a year to change prices of at least as many products. Applying the Gali and Gertler's (1999) approach to U.S. industry-level data, therefore, provides at least as much support for the assumption of price flexibility as for the assumption of price stickiness.
In the next section, we review the extent to which modern macroeconomics has hitched its wagon to the price stickiness assumption, even though econometric evidence about the predictions of sticky-price models has been mixed and empirical support for this key assumption has been lacking. In Section III, we explain how we apply the methodology of Gali and Gertler (1999) for measuring the degree of price stickiness to U.S. industry-level data, in which we are able to utilize actual industry cost data rather than the labor income share proxy they employ in their analysis of aggregate U.S. data. Section IV summarizes our empirical results. In Section V, we discuss the implications of our analysis for modern macroeconomic theory.
II. THE STICKY-PRICE ASSUMPTION: THEORY AND EVIDENCE
The origins of today's sticky-price macroeconomic theories are Taylor's (1980), Rotemberg's (1982), and Calvo's (1983) initial work incorporating staggered adjustments into linear models that appealed to quadratic loss functions as approximations to "consumer surplus" measures, as in Aizenman and Frenkel (1985) and Horowitz (1987). Based on the roadmap first provided by Blanchard and Kiyotaki (1987), the preponderance of modern sticky-price macro models begins with microfoundations based on decision making by optimizing agents, a mode of analysis that in a dynamic, general-equilibrium context rules out equilibria in which agents make decisions at any point that turn out to be inconsistent with long-term or lifetime budget constraints. In order to generate short-run non-neutralities consistent with real-world data, the literature has revealed that there is a large range of factors from which a macroeconomic theorist might choose. These include nominal wage stickiness, asset portfolio rigidities and liquidity constraints, real wage rigidities, and product price inflexibility. The latter source of non-neutralities has received most attention in the currently predominant New Keynesian literature, which emphasizes models in which monopolistically competitive producers (which often double as consumers) set the prices of imperfectly substitutable goods.
Thus, in most New Keynesian sticky-price models, price contracting or menu costs are presumed to induce producers to leave prices unchanged over short-run intervals. Agents that function as producer-consumers possess identical intertemporal utility functions, in which the present value of an individual's utility depends positively on a constant elasticity of substitution index of the differentiated goods produced by all individuals in the economy, positively on real money balances, and negatively on work effort the individual devotes to productive activities, which in turn depends on the demand for the individual's product. Each model, of course has its own special features; among many recent examples are Chari, Kehoe, and McGratten (2000), Erceg, Henderson, and Levin (2000), and Dotsey and King (2001). In an open economy setting, the Obstfeld and Rogoff (1995a, 1995b) "redux" model, which Canzoneri, Cumby, and Diba (2002) have referred to as the new "workhorse model" of open economy macroeconomics, builds on this sticky-price, dynamic optimizing approach. Obstfeld and Rogoff assumed that the law of one price holds for consumer price indexes, but failure of product prices to adjust fully to exchange-rate variations or other external or internal shocks results in terms-of-trade variations and incomplete pass-through effects that induce violations of the law of one price for national output deflators as in Sarno (2001), Lane (2001), and VanHoose (2004). Short-run product price inflexibilities faced by representative agents ultimately cause actual output of an open economy to differ from the efficient level, so that policies inducing changes in aggregate demand at a national level have the potential to improve both domestic and global welfare in a world with interdependent economies.
Early work exploring whether there is, indeed, widespread price stickiness in real-world economies focused on prices of magazines sold at newsstands, of specific transactions for certain products, or of items listed in catalogs, as analyzed by Cecchetti (1986), Carlton (1986), and Kashyap (1995), or considered survey evidence on the frequency of price adjustment, as in Blinder (1991). As pointed out by Caplin and Spulber (1987), however, it may be problematic to infer widespread evidence of price stickiness from data limited to only a few specific industries. This perhaps helps to explain why much of the subsequent literature has focused on conducting empirical tests of price stickiness via analysis of macroeconomic data. Kandil (1994), for instance, sought to directly test for price...
|
