Laurence Ball, N. Gregory Mankiw, and David romer depends on real aggregate demand as well as the real wage, because changes in aggregate demand shift the firms product demand (see - Real wages need not be countercyclical Imperfect competition can remedy an embarrassing empirical failure of traditional models based on sticky nominal wages--the cyclical behavior of real wages. We can tautologically write P= uW/MPL, where P is the price level, w is the wage, MPL is the marginal product of labor, and u is the markup of pric over marginal cost. If the markup is constant and marginal product of labor is diminishing, as many 1970s models assumed, then the real wage, W/P= MPL/u, must be countercyclical In actual economies, however, real wages appear acyclical or a bit procyclical. This fact can be explained if the marginal product of labor is constant, as suggested by Hall, or if the markup is countercyclical, as suggested by rotemberg and saloner and by Bils. 7 Thus there need not be a link between changes in employment and changes in real wages Nominal rigidities have aggregate demand externalities. As we explained, since real aggregate demand affects the demand curve facing individual firms, nominal rigidities have externalities. Rigidity in one firms price contributes to rigidity in the price level, which causes fluctuations in real aggregate demand and thus harms all firms. These externalities are crucial to the finding that small frictions can have large macroeconomic effects. The externalities depend on imperfect compe tition, for under perfect competition, aggregate demand is irrelevant to individual firms because they can sell all they want at the going price Product Market rigidi nd most Keyne rigidities in nominal wages. But recent work focuses largely on rigidities in product prices. The change offers two advantages -Goods are sold in spot markets. Although there is clearly much wage rigidity in actual economies-in U.S. labor contrac example, wages are set up to three years in advance-the allocative effects of this rigidity are unclear. The implicit contracts literature shows that it may be efficient for contract signers to make employment 17, Robert E, Hall, * Market Structure and Macroeconomic Fluctuations, BPEA 2: 1986, pp. 285-322; Julio J. Rotemberg and Garth Saloner, "A Supergame-Theoretic Model of Price Wars during Booms, American Economic Review, vol 76 ( June 1986) pp. 390-407: Mark Bils, *Cyclical Pricing of Durable Luxuries, Working Paper 83 ( University of Rochester Center for Economic Research, May 198
16 Brookings Papers on Economic Activity, 1: 988 independent of wages. That is, given long-term relationships with their workers, firms may choose the efficient amount of employment rather than moving along their labor demand curves when real wages change I8 In many product markets, on the other hand, buyers clearly operate on their demand curves. Forexample, the local shoe store has no agreement, explicit or implict, from its customers to buy the efficient number of shoes regardless of the prices. Instead, rigidity in the stores prices affects its sales of shoes -Real wages need not be countercyclical. As we argue above acyclical real wages are possible even if nominal rigidities occur only in wages. But it is easiest to explain acyclical or procyclical real wages if prices as well as wages are sticky. In this case, the effect of a shock real wages depends on the relative sizes of the adjustments of prices and Despite the advantages of studying rigidities in goods markets, we are ambivalent about the deemphasis of labor markets, because the apparent rigidities in nominal wages may have important allocative effects Further research on the relative importance of wage and price rigidities is needed DISCUSSION We conclude this section by discussing several issues concerning the importance of recent theories and their plausibility The importance of Nominal rigidities. Nominal rigidities are essen- tial for explaining important features of business cycles. As we have emphasized real effects of nominal disturbances, such as changes in the money stock, depend on some nominalimperfection. The only prominent alternative to nominal rigidities is imperfect information about the aggregate price level, an explanation that many economists find implau- sible. It is possible, of course, to maintain that money is neutral in the short run-that Paul Volcker, for example, had nothing to do with the 1982 recession-but this also appears unrealistic to many economists 18. Early expositions of this idea appear in Martin Neil Baily, Wages and Employment under Uncertain Demand, Review of Economic Studies, vol 41(January 1974), pp 37 50; Costas Azariadis, ""Implicit Contracts and Underemployment Equilibria, Journal of Political Economy, vol. 83(December 1975),pp. 1183-1202; and Robert E Hall, The gidity of Wages and the Persistence of Unemployment, BPEA, 2: 1975, pp. 301-35
Laurence Ball, N. Gregory Mankiw, and David Romer Thus it is difficult to explain the relation of output to nominal variables without nominal rigidities Nominal rigidities are also important for explaining the effects of real shocks to aggregate demand, resulting for example from changes in government spending or in the expectations of investors. The point is clear if we interpret M in the aggregate demand equation, y= M/P, as imply a shift term, in which case real disturbances that shift demand affect output through the same channels as changes in money Not all explanations for the output effects of real demand shocks pend on nominal imperfections. robert Barros model of government purchases, for one, does not. 9 But such explanations invoke implausibly large labor supply elasticities. Thus nominal rigidities, while not the only explanation for the effects of real demand, are perhaps the most appea In the models we have surveyed slow adjustment of prices implies that shocks cause temporary deviations of output and employment from their"natural rates. Recently, however, models of hysteresis, in which shocks have permanent effects, have become popular. For example Blanchard and Summers argue that the natural rate of unemployment in European countries changes when actual unemployment changes, so that there is no unique level to which unemployment returns. 20 If these theories are correct, then nominal rigidities cannot fully explain unem ployment, because nominal prices eventually adjust to shocks; some dditional explanation, such as the insider-outsider model in Blanchard and Summers, is needed for the persistence of unemployment. But nominal rigidities may be crucial for explaining the initial impulse unemployment. For example, after rising during the late 1970s, unem- ployment in Britain has remained high, suggesting hysteresis. But the best explanation for the original increase is arguably a conventional one low adjustment of wages and prices to shocks like tight monetary policy and increases in import prices The Importance of Externalities from Rigidity. Externalities from minai rigidity, the central element of menu cost models, are essential 19. Robert J. Barro, Output Effects of Government Purchases, " Journal of political Economy, vol. 89(December 1981), pp. 1086-1121 20. Olivier J. Blanchard and Lawrence Summers,""Hysteresis and the European Unemployment Problem, " "in Stanley Fischer, ed, NBER Macroeconomics AnnuaL, 1986 ( MIT Press,1986),pp.15-78
Brookings Papers on Economic Activity, 1: 1988 for a plausible theory of rigidities. If rigidities exist, one of the following statements must be true: rigidities do not impose large costs on the economy; rigidities have large costs to the firms and workers who create them, but these are exceeded by the costs of reducing rigidities; or rigidities have small private costs, and so small frictions are sufficient to create them, but externalities from rigidity impose large costs on the economy. The problem with the first statement is the difficulty of explaining apparently costly events, such as rises in unemployment following monetary contractions, without nominal rigidities. The second seems implausible: it would not be costly for magazine publishers to print new prices every year rather than every four years, as they typically do. 2I Thus the third statement is the best hope for explaining rigidities What Are Menu Costs? Models of nominal rigidity depend on some cost of full flexibility, albeit a small one. The term menu cost may be misleading because the physical costs of printing menus and catalogs may not be the most important barriers to flexibility. Perhaps more important is the lost convenience of fixing prices in nominal terms-the cost of learning to think in real terms and of computing the nominal price changes corresponding to desired real price changes. More generally we can view infrequent revision of nominal prices as a rule of thumb that is more convenient than continuous revision. Thus, rather than referring to menus, we can state the central argument of recent papers as follows Firms take the convenient shortcut of infrequently reviewing and chang ing prices. The resulting profit loss is small, so firms have little incentive to eliminate the shortcut but externalities make the macroeconomic At a somewhat deeper level, we can interpret the convenience of fixing nominal rather than real prices as that of using the medium of exchange, dollars, as a unit of account. 22 Alternatively, following Akerlof and Yellen, we can view simple rules of thumb as arising from"ne 21. Stephen G. Cecchetti, The Frequency of Price Adjustment: A Study of the Newsstand Prices of Magazines, ' Journal of Econometrics, vol 31(August 1986), pp 255-74. The cost of reducing nominal wage rigidity may be significant if rigidity is reduce through shorter labor contracts, which require more frequent negotiations between unions and management. But wage rigidity can also be reduced through greater indexation or by having the nominal wage change more often over the life of a contract, neither of which appears to have large costs 22. Bennett T, McCallum, On 'Real and 'Sticky-Price Theories of the busine Cycle, "Journal of Money, Credit, and Banking, vol 18(November 1986), pp, 397-414
Laurence BalL, N. Gregory Mankiw, and David Romer rationality, ' a small departure from full optimization. 23 In any case, the precise source offrictions is not important. The effects of nominal shocks are the same whether rigidity arises from printing costs, near-rationality or something else Inflation, the Frequency of Adjustment, and the Phillips Curve Recent research shows that nominal rigidity is possible in principle- that one can construct a model with firm microeconomic foundations in which rational agents choose substantial rigidity. But the validity of Keynesian theories is not thereby established. For these theories to be convincing, they must have empirical implications that contradict other macroeconomic theories, and these predictions must be confirmed by evidence. This section derives implications of recent Keynesian models and the next section tests them. As explained in the introduction, the main prediction is that the real effects of nominal shocks are smaller when average inflation is higher. Higher average inflation erodes the frictions that cause nonneutralities, for example by causing more fre- quent wage and price adjustments This section studies a specific model of the class described in the previous section. In the model, a cost of price adjustment leads firms to change prices at intervals rather than continuously. In addition te providing a basis for the empirical tests of the next major section, the model is of theoretical interest. Previous models of nominal rigidity are highly stylized; for example, most menu cost models are static. Our model is dynamic and has the appealing feature that the price level adjusts slowly over time to a nominal shock. The speed of adjustment which is treated as exogenous in older Keynesian models, is endogenous It depends on the frequency of price adjustment by individual firms hich in turn is derived from profit-maximization. 24 e first present the model and show that high average inflation educes the output effects of nominal shocks. We also show that highly variable aggregate demand reduces these effects. We then investigate 23, Akerlof and Yellen, A Near-Rational Model 24. The speed of adjustment is also endogenous in Laurence Ball, " "Externalities from Contract Length, American Economic Review, vol 77(September 1987), pp, 615-20