can arise if the frictions are combined with real rigidities arising from efficiency wages, customer markets, and the like. For example, substan tial nominal rigidity can arise from a combination of real rigidity in the labor market and imperfect competition and menu costs in the good market. If firms pay efficiency wages, for instance, then real wages may be set above the market-clearing level, so that workers are off their labor supply curves. In this situation a fall in labor demand can greatly reduce employment without a large fall in the real wage even if labor supply is inelastic The importance of real rigidities for explaining nominal rigidities is not settled, because there is no consensus about the sources an magnitudes of real rigidities in actual economies. In particular, phenom- na like efficiency wages and customer markets increase nominal rigidity to the extent that they reduce desired responses of real wages and real prices to demand shifts, but economists are still unsure of the sizes of these effects. Further research on real rigidities will lead to a better understanding of nominal rigidities Staggered Price Setting. Even when real rigidities are added, the models surveyed so far cannot fully explain the size and persistence of the real effects of nominal shocks. In these models the effects of shocks are eliminated when nominal prices adjust In actual economies, reces- sions following severe demand contractions can last for several years and while individual prices are fixed for substantial periods, these period are generally shorter than several years. Thus models with sticky prices must explain why the effects of shocks persist afterall prices are changed An explanation is provided by the literature on staggered price setting which shows that if firms change prices at different times, the adjustment of the aggregate price level to shocks can take much longer than the time between adjustments of each individual price. o The price levelinertia caused by staggering implies that nominal shocks can have large and long-lasting real effects even if individual prices change frequently. 10. John B. Taylor, " Stagger Review, vol 69(May 1979, Papers and Proceedings, 1978), pp. 108-13: Taylor, " Aggregate Dynamics and Staggered Contracts, Journal of Political Economy, vol. 88(February 1980), pp. 1-23; Olivier J. Blanchard, Price Asynchronization and Price Level Inertia, in Rudiger Dornbusch and Mario Henrique Simonsen, eds, Inflation, Debt, and indexation (MIT Press, 1983),pp 3-24; Olivier J. Blanchard, The Wage Price Spiral, ' Quarterly Journal of Economics, vol. 101(August 1986), pp. 543-65
Laurence Ball, N. Gregory Mankiw, and David Romer A simple example makes clear the importance of the timing of price anges. Suppose first that every firm adjusts its price on the first of each month, so that price setting is synchronized. If the money supply falls on June 10, output is reduced from June 10 to July 1, because nominal prices are fixed during this period. But on July 1 all prices adjust in proportion to the fall in money, and the recession ends Now suppose that half of all firms set prices on the first of each month and half on the fifteenth. If the money supply falls on June 10, then on June 15 half the firms have an opportunity .o adjust their prices. But in his case they may choose to make little adjustment. Because half of all nominal prices remain fixed, adjustment of the other prices implies changes in relative prices, which firms may not want. (In contrast, if all prices change simultaneously, full nominal adjustment does not affect relative prices. If the June 15 price setters make little adjustment, then the other firms make little adjustment when their turn comes on July 1 because they do not desire relative price changes either. And so on. The price level declines slowly as the result of small decreases every first and fifteenth, and the real effects of the fall in money die out slowly. In short, price adjustment is slow because neither group of firms is willing to be the first to make large cuts. As Blanchard emphasizes, if staggering occurs among firms at differ ent points in a chain of production, its effects are strengthened. 12A I1. a natural question is why firms change prices at different times if this exacerbates aggregate fluctuations. One obvious answer is that different firms receive shocks at different times and face different costs of price adjustment. Laurence Ball and david Romer, " The Equilibrium and Optimal Timing of Price Changes, Working Paper 241 (NBER, October 1987), show that, because of externalities from staggering, idiosyncrat hocks can lead to staggering even if synchronized price setting is Pareto superior. but diosyncratic shocks cannot explain all staggering For example, some firms with two-year labor contracts set wages in even years and some set them in odd years, and this does not orrespond to deterministic two-year cycles in the arrival of shocks. Another explanatio or staggering is that it arises from firms' efforts to gain information. This source of gering is discussed in Arthur M. Okun, Prices and Quantities: A Macroeconomic Analysis(Brookings, 1981), and formalized in Laurence Ball and Stephen G, Cecchetti Imperfect Information and Staggered Price Setting, Working Paper 2201 (NBER, April 1987). For example, a firm wants to set wages in line with the wages of other firms. If all wages are set simultaneously, each firm is unsure of what wage to set because it does not know what others will do. This gives each firm an incentive to set its wage shortly after he others. The desire of each firm to"bat last, as Okun puts it, can lead in equilibrium o a uniform distribution of signing dates 12. Blanchard, "Price Asynchronization
Brookings Papers on Economic Activity, 1: 1988 firms profit-maximizing price is tied to both the prices of its inputs and the prices of goods for which its product is an input(the latter influence demand for the firm s product). Thus a firm does not want to adjust its price to a shock if these other prices do not adjust at the same time. This reluctance to make asynchronized adjustments causes price level inertia. Blanchard shows that the degree of inertia increases the longer the chain of production: it takes a long time for the gradual adjustment of prices to make its way through a complicated system The literature on staggered price setting complements that on nominal rigidities arising from menu costs. The degree of rigidity in the aggregate price level depends on both the frequency and the timing of individual price changes. Menu costs cause prices to adjust infrequently. For a given frequency of individual adjustment, staggering slows the adjust- ment of the price level. Large aggregate rigidities can thus be explained by a combination of staggering and nominal frictions the former mag- nifies the rigidities arising from the latter. Asymmetric Effects of Demand shocks. We conclude this part of our discussion by mentioning a little-explored possibility for strengthening Keynesian models. The models surveyed imply symmetric responses of the economy tories and falls in nominal aggregate demand. Forexample, in menu cost models the range of shocks to which prices do not adjust is symmetric around zero, and so is the range of possible changes in output. But traditional Keynesian models often imply asymmetric effects of demand shifts In undergraduate texts for example, the aggregate supply curve is often drawn so that decreases in demand lead to large output losses while the effects of increases are mostly dissipated through higher prices. Such asymmetries are intuitively appealing, and they greatly strengthen the Keynesian view that demand stabilization is desirable stabilization raises the average levels of output and employment as well as reducing the variances. It is unclear whether plausible modifications of new Keynesian models can produce asymmetries. Asymmetric effects of shocks could arise from asymmetric price rigidity--prices that are ticky downward but not Ing no that is difficult to formalize. 3 Adjustment Costs
Laurence Ball, N. Gregory Mankiw, and David Romer THE NEW ASSUMPTIONS IN NEW KEYNESIAN MODELS Aside from the specific arguments outlined above, recent research establishes the general point that nominal rigidities can result from optimizing choices of agents in well-specified models. This contrasts with the ad hoc imposition of rigidities in many of the Keynesian models of the 1970s. Recent progress is largely a result of two innovations in modeling: the introduction of imperfect competition and greater empha sis on price rather than wage rigidities Imperfect Competition. Microeconomists have long recognized that sticky prices and perfect competition are incompatible. 4 Inacompetitive market, a firm does not set its price, but accepts the price quoted by the alrasian auctioneer. Only under imperfect competition, when firms set prices, does it make sense to ask whether a firm adjusts its price to a shock. Nonetheless, Keynesian models of the 1970s, most clearly disequilibrium models, imposed nominal rigidities on otherwise Walras ian economies. The result was embarrassments in the form of unappeal ing results or the need for additional arbitrary assumptions. Many recent models simply generalize earlier models by allowing the firms'demand curves to slope down. This single modification sweeps away many of the problems with older models. Specifically, the new models with imperfect competition offer six advantages tion, the gains from nominal adjustment are large. For example, if nominal demand rises and prices do not adjust, there is excess demand In this situation, an individual firm can raise its price significantly and still sell as much output as before, which implies a large increase in profits. In contrast, under imperfect competition a higher price always implies lower sales. Starting from the profit-maximizing price-q antity ombination, the gains from trading off price and sales after a shock are second order Anticipated Inflation, and Output, Quarterly Journal of Economics, vol. 101(May 1986 Honor of Bernard Francis Haley(Stanford University Press, 1959), pp 41-. 'Essm ,in 14. See, for example, Kenneth J, Arrow, Toward a Theory of Price Adjustment, Moses Abramovitz and others, eds, The Allocation of Economic Resources
Brookings Papers on Economic Activity, 1: 1988 -Output is demand determined. When price rigidity is imposed on a Walrasian market. so that the market does not clear. it is natural to assume that quantity equals the smaller of supply and demand, so that output falls below the Walrasian level when price is either above or below the Walrasian level. But Keynesians believe that when prices are rigid, increases in demand, which mean prices below Walrasian levels raise output, just as decreases in demand reduce output. This result is built into many Keynesian models through the unappealing assumption that output is demand determined even if demand exceeds supply. For example, in the gray- Fischer contract model, firms hire as much labor as they want, regardless of the preferences of workers. i5 In contrast, under imperfect competition, demand determination arises naturall Firms set prices and then meet demand. Crucially, if demand rises, firms are happy to sell more even if they do not adjust their prices, because under imperfect competition price initially exceeds marginal cost. Thus changes in demand always cause changes in output in the same direction - Booms raise welfare. Under perfect competition, the equilibrium level of output in the absence of shocks is efficient. Thus increases in output resulting from positive shocks, as well as decreases resulting from negative shocks, reduce welfare. In the gray- Fischer model, for example, half the welfare loss from the business cycle occurs when workers are required to work more than they want. In actual economies unusually high output and employment mean that the economy is doing well. 16 And this is the case in models of imperfect competition. Since imperfect competition pushes the no-shock level of output below the social optimum, welfare rises when output rises above this level. ployment through low aggregate de- mand. In 1970s models with sticky nominal wages, unemployment occurs hen prices fall short of the level expected when wages were set, so that real wages rise and firms move up their labor demand curves. In actual economies, however, firms often appear to reduce employment because demand for their output is low, not because real wages are high. This fact is not necessarily a problem for Keynesian theories if the goods market is imperfectly competitive. In this case, a firms labor demand 15. Fischer, ""Long-Term Contracts, and Gray, "" On Indexation. 16. Of course economists worry that low unemployment may be inflationary. But sticky-price models with perfect competition imply that low unemployment is undesirable