See Caballero and Engel (1991,1993) for more detailed analyses of these issues. places. Sections 6.10-6.12 draw on D. Romer (1993a).

has the effect that the aggregate price level responds fully to changes in m. In the Taylor model, in contrast, the number of price-setters changing their prices at any time is fixed; as a result, the price level does not respond fully to changes in m.

The neutrality of money in the Caplin-Spulber model is not a robust result about settings where fixed costs of changing nominal prices cause the number of price-setters changing prices at any time to be endogenous. If, for example, inflation can be negative as well as positive, or if there are idiosyncratic shocks that sometimes cause price-setters to lower their nominal prices, the resulting extensions of Ss rules generaUy cause monetary shocks to have real effects (see, for example, Caplin and Leahy, 1991, and Problem 6.12). In addition, the values of S and s may change in response to changes in aggregate demand. If, for example, high money growth today signals high money growth in the future, price-setters widen their Ss bands when there is a positive monetary shock; as a result no price-setters adjust their prices in the short run (since no price-setters are now at the new, lower trigger point s), and so the positive shock raises output (Tsiddon, 1991).°

Thus the importance of Caplin and Spulbers model is not for its specific results about the effects of aggregate demand shocks. Rather, the model is important for two reasons. First, it introduces the idea of state-dependent price changes. Second, it demonstrates another reason that the relation between microeconomic and macroeconomic rigidity is complex. The Fischer and Taylor models show that temporary fixity of some prices can have a disproportionate effect on the response of the aggregate price level to aggregate demand disturbances. The Caplin-Spulber model, in contrast, shows that the adjustment of some prices can have a disproportionate effect: a smaU fraction of price-setters making large price changes can be enough to generate neutrality in the aggregate. Thus together, the Fischer, Taylor, and Caphn-Spulber models show that any complete treatment of price rigidity requires careful attention both to the nature of price-adjustment pohcies and to how those policies interact to determine the behavior of the aggregate price level.

Part New Keynesian Economics 6.10 Overview

The Lucas, Fischer, and Taylor models are not fully satisfactory accounts of real effects of aggregate demand disturbances. The models assume the existence of imperfections that agents could overcome easily-imperfect

The seminal papers are Mankiw (1985) and Akerlof and Yellen (1985). See also Parkin (1986); Rotemberg (1982, 1987); and Blanchard and Kiyotaki (1987).

knowledge of the price level in the Lucas model, and infrequent adjustment of prices or wages in the Fischer and Taylor models. Quite accurate information about movements in the price level is easily available, and the cost of much more frequent price or wage adjustments (through indexation or other means) is small. This raises the question of why agents would permit nominal disturbances to lead to substantial fluctuations in output rather than take the small measures needed to largely eliminate the nominal imperfections.

The central idea of much recent research on the real effects of nominal shocks is that this question apphes not just to these particular models, but to all candidate sources of nominal imperfections. Individuals are mainly concerned with real prices and quantities: real wages, hours of work, real consumption levels, and the like. Nominal magnitudes matter to them only in ways that are minor and easily overcome. Prices and wages are quoted in nominal terms, but it costs little to change (or index) them. Individuals are not fully informed about the aggregate price level, but they can obtain accurate information at little cost. Debt contracts are usually specified in nominal terms, but they too could be indexed with little difficulty. And individuals hold modest amounts of currency, which is denominated in nominal terms, but they can change their holdings easily. There is no way in which nominal magnitudes are of great direct importance to individuals.

Thus, according to this newKeymsianview, if nominal imperfections are important to fluctuations in aggregate activity, it must be that nominal frictions that are small at the microeconomic level somehow have a large effect on the macroeconomy. Much of the recent research on the microeconomic foundations of nominal rigidity is devoted to addressing the question of whether this can plausibly be the case.

For concreteness, most of this section addresses this question for a specific view about the nominal imperfection. In particular, we focus on a static model where firms face a menu cost of price adjustment-a small fixed cost of changing a nominal price. (The standard example is the cost incurred by a restaurant in printing new menus-hence the name.) But, as described at the end of Section 6.12, essentially the same issues arise with other views about the barriers to nominal adjustment. In addition, the analysis focuses on the question of whether menu costs can lead to significant nominal stickiness in response to a one-time monetary shock. As a result, the analysis is more successful in achieving the first goal of analyzing microeconomic foundations of incomplete nominal adjustment (namely, characterizing the microeconomic conditions that yield sluggish adjustment) than in achieving the second goal (namely, finding the implications of those conditions for the specifics of price adjustment, and thereby providing guidance for policy). ,

6.11 Are Small Frictions Enough?

General Considerations

Consider an economy of many price-setting firms. Assume that it is initially at its flexible-price equilibrium; that is, each firms price is such that, if aggregate demand is at its expected level, marginal revenue equals marginal cost. After prices are set, aggregate demand is determined; at this point each firm can change its price by paying a menu cost. For simphcity, prices are assumed to be set afresh at the start of each period. This means that the dynamic pricing issues that are the subject of Part of this chapter are irrelevant; it also means that if a firm pays the menu cost, it sets its price to the new profit-maximizing level.

Our focus is on the question of when firms change their prices in response to a departure of aggregate demand from its expected level. For concreteness, suppose that demand is less than expected. Since the economy is large, each firm takes the actions of other firms as given. Constant nominal prices are thus an equilibrium if, when all other firms hold their prices fixed, the maximum gain to a representative firm from changing its price is less than the menu cost of price adjustinent.

We can analyze this issue using the marginal revenue-marginal cost diagram in Figure 6.3. The economy begins in equilibrium; thus the representative firm is producing at the point where marginal cost equals marginal revenue (Point A in the diagram). A fall in aggregate demand with other prices unchanged reduces aggregate output, and thus shifts inward the demand curve that the firm faces-at a given price, demand for the firms product is lower. Thus the marginal revenue curve shifts in. If the firm does not change its price, its output is determined by demand at the existing price (Point B). At this level of output, marginal revenue exceeds marginal cost, and so the firm has some incentive to lower its price and raise output."*

The condition for price adjustment by all firms to be an equilibrium is not simply the reverse of this. As a result, there can be cases when both price adjustment and unchanged prices are equilibria. See Problem 6.15.

"The fall m aggregate output is likely to reduce the prevailing wage, and therefore to shift the marginal cost curve down. For simplicity, this effect is not shown in the hgure.

Section 6.11 shows that introducing price-setting and menu costs into an economy that is otherwise Walrasian is probably not enough to generate substantial nominal rigidity. Section 6.12 is therefore devoted to the issue of what else is needed for menu costs to have important effects. Section 6.13 considers some relevant empirical work. Finally, Sections 6.14 and 6.15 discuss some extensions and limitations of the theory.