Laurence Ball, N. Gregory Mankiw, and David Romer 5 SMALL NOMINAL FRICTIONS AND LARGE NOMINAL RIGIDITIES The recent literature on nominal rigidities enters an argument that Keynesians appeared to be losing. Members of the new classical school that developed in the 1970s challenged Keynesians to explain the rigidities in Keynesian models. In response, Keynesians sometimes cited costs of adjusting prices. But as the classicals pointed out, these costs, while surely present, appear small. Indeed, the frequently men- tioned"menu costs-the costs of printing new menus and catalogs, of replacing price tags, and so on-sound trivial. Thus the impediments to nominal flexibility in actual economies appear too small to provide a foundation for Keynesian models A common but mistaken response is that there are many obvious sources of large wage and price rigidities: implicit contracts, customer markets, efficiency wages, insider-outsider relationships, and so on. The problem is that these phenomena imply rigidities in real wages and prices, while the Keynesian theory depends on rigidities in nominal wages and prices. Real rigidities are no impediment to complete flexibility of nominal prices, because full adjustment to a nominal shock does not require any change in real prices. The absence of models of nominal rigidity reflects the microeconomic proposition that agents do not ca about nominal magnitudes. The only apparent departures from this proposition in actual economies are the small costs of nominal adjust ment Thus recent work begins with the premise that it is inexpensive to reduce nominal rigidity and asks how substantial rigidity nonethele arises. The central answer of the literature is presented by mankiw, Akerlof and Yellen, Blanchard and Kiyotaki, and Ball and romer. Journal of Economic Literature, vol. 19(June 1981), pp. 493-530, and the discussion of externalities from nominal rigidity in Charles L. Schultze, Microeconomic Efficiency nd Nominal Wage Stickiness, American Economic Review, vol. 75(March 1985) 6. N. Gregory Mankiw, ""Small Menu Costs and Large Business Cycles: A Macroe of Monopoly, Quarterly Journal of Economics, vol, 100(May 1985), pp A, Akerlof and Janet L, ye ics,vol.100(1985 Supplement), pp. 823-38; Oliv erican Economic Review, vol, 77 (September 1987), pp. 647-66 David Romer, Are Prices Too Sticky? Working Paper 2171(NBER, February 1987)
Brookings Papers on E Activity, 1: 1988 Second-Order Private Costs and First-Order Business Cycles. Mankiw and Akerlof and Yellen make a simple but important point. They study imperfectly competitive economies and show that the cost of nominal dities to price setters can be much smaller than the macroeconomic ffects. An example that illustrates the cost to price setters is a firm that initially sets its price at the profit-maximizing level but does not adjust after the money supply falls. We let m () denote the firms profits as a function of its price and let P be the firms predetermined price and P* its profit-maximizing price, which it would set if it adjusted. Using a Taylor expansion, we can approximate the firms profit loss from not (1)丌(P*)-丌(P)≈丌(P)P*-P)-丌"(P*)(P*-P)2 But since P* maximizes profits, '(*)is zero. Thus the profit loss from nonadjustment is second order-that is, proportional to the square of (P*-P). As long as the predetermined price is close to the profit- maximizing price, the cost of price rigidity to the firm is small But rigidity can have first-order macroeconomic effects. An increase in nominal money with nominal prices fixed leads to a first-order increase in real aggregate demand, and hence in real output. For example, if the aggregate demand curve is simply y= M/P, rigid prices imply a change in output proportional to the change in money The effect on social welfare is also first order, as follows from the assumption of imperfect competition. Under imperfect competition, the profit-maximizing price is socially suboptimal. The price is too high and output is too low. Thus at P* the first derivative of welfare with respect to the firms price is negative: welfare would rise if the price fell below P*. Nonadjustment to a fall in money implies P greater than P*, given the negative first derivative of welfare, the welfare loss is first order. Because the cost of rigidity to a price setter is second order while the macroeconomic effects are first order, the latter can be much large This finding resolves the puzzle of why price setters refuse to incur the mall costs of reducing the business cycle through more flexible prices Despite the large macroeconomic effects, the private itives are Aggregate Demand Externalities. blanchard and Kiyotaki provide an important interpretation of the result in Mankiw and Akerlof-Yellen:
Laurence Ball, N. Gregory Mankiw, and David Romer the macroeconomic effects of nominal rigidity differ from the private costs because rigidity has an aggregate demand externality 'A fev equations make this clear. Suppose the demand for the product of firm i depends on aggregate spending and on the firms relative price For simplicity, aggregate demand is given by a quantity equation? M Combining equations 2 and 3 yield MVP Y According to equation 4, firm is demand depends on its relative price and on real money, which determines aggregate demand. Changes in real money shift the demand curve facing firm i, and the firms price determines its position on the demand curve. If M falls and firm i does not adjust, the second- order cost to firm i is that Pi/P does not adjust to the new profit-maximizing level. The externality is that rigidity in firm is price contributes to rigidity in the aggregate price level. Given the fall in nominal money, rigidity in P implies a first-order fall in real money, which reduces demand for all firms' goods. In other words, there is an externality because adjustment of all prices would prevent a fall in real aggregate demand, but each firm is a small part of the economy and thus ignores this macroeconomic he importance of the externality is illustrated by a firm in a recession caused by tight money. To the firm, the recession means an inward shift of its demand curve and a resulting first-order loss in profits. The firm would very much like to shift its demand curve back out, but of course it cannot do so by changing its price. Instead, price adjustment would yield only the second-order gain from optimally dividing the losses from 7. The only essential feature of equation 3 is the negative relation between Y and P. I interpret M as simply a shift term in the aggregate demand equation. Thus, as we below, the results in recent papers concern the effects shock to aggregate demand not just changes in the money stock
Brookings Papers on Economic Activity, 1: 1988 the recession between reduced sales and a lower price. The recession would end and everyone would be much better off if all firms adjusted But each firm believes that it cannot end the recession and therefore may fail to adjust even if the costs of adjustment are much smaller than the costs of the recession This argument resembles standard microeconomic analyses of exter- nalities. Consider the classic example of pollution. Pollution would be greatly reduced, and social welfare greatly improved, if each person incurred the small cost of walking to the trash can at the end of the block But each individual ignores this when he throws his wrapper on the street because he is only one of many polluters. Because of externalities economists do not find highly inefficient levels of pollution puzzling even though the costs of reducing pollution are small. For similar reasons highly inefficient nominal rigidities are not a mystery even though menu costs are sma Externalities from Fluctuations in Demand. Keynesians believe not only that shocks to nominal aggregate demand cause large fluctuations in output and welfare, but also that these fluctuations are inefficient, and thus that stabilization of demand is desirable. The models surveyed so far do not provide a foundation for this view. As explained above, nonadjustment of prices to a fall in demand leads to large reductions output and welfare. But nonadjustment to a rise in demand leads to higher output and, because output is initially too low under imperfect competition, to higher welfare. Thus the implications of fluctuations for average welfare, and hence the desirability of reducing fluctuations, are unclear. Indeed. Ball and romer show that the first-order welfare effects of fluctuations average to zero, which means that the first order-second order distinction is irrelevant to this issue. 8 Nonetheless, Ball and Romer show, by comparing the average social and private costs of nominal rigidity, that small nominal frictions are sufficient for large reductions in average welfare. The private cost is fluctuations of a firms relative price around the profit-maximizing level The social cost is the private cost plus the cost of fluctuations in real aggregate demand Greater flexibility would stabilize real demand, but each firm ignores its effect on the variance of demand, just as it ignores its effect on the level of demand after a given shock. Although both the 8. Ball and Romer, ""Are Prices Too Sticky?
Laurence ball, N. Gregory Mankin, and David Romer average social and average private costs are second order, Ball and Romer show that the former may be much larger: fluctuationsin aggregate demand can be much more costly than fluctuations in relative prices. As a result, small frictions can prevent firms from adopting greater flexibility even if business cycles are highly inefficient STILL LARGER RIGIDITIES The papers discussed so far establish that nominal rigidities can b far larger than the frictions that cause them. But as we now describe, the simple models in these papers cannot fully explain nonneutralities of the size and persistence observed in actual economies. Therefore, we turn to more complicated models that incorporate realistic phenomena that magnify nominal rigidities, These phenomena include rigidities in real wages and prices and asynchronized timing of price changes by different firms Real rigidities. As we argue above, real rigidities alone are no impediment to full nominal flexibility. But Ball and Romer show that a high degree of real rigidity, defined as small responses of real wages and real prices to changes in real demand, greatly increases the nonneutral- ities arising from small nominal frictions. g This finding is important because, although models with nominal frictions but no real rigidities can in principle produce large nominal rigidities, they do so only for implausible parameter values. Most mportant, large rigidities arise only if labor supply is highly elastic, while labor supply elasticities in actual economies appear small. The role of labor supply is illustrated by a hypothetical economy with imperfect competition and menu costs in the goods market but a Walrasian labor market. If menu costs led to nominal price then nominal shocks would cause large shifts in labor demand supply were inelastic, these shifts in labor demand would changes in the real wage and thereby create large incentives for price setters to adjust their prices. As a result, nominal rigidity would not be an equilibrium While for plausible parameter values nominal frictions alone produce little nominal rigidity, Ball and Romer show that considerable rigidity 9. Laurence Ball and David Romer, " "Real Rigidities and the Non-Neutrality of Money, " Working Paper 2476(NBER, December 1987)