2 Two models in which results from adjustment are tested. One, an (s,S) pricing model, assumes lump-sum adjustment and predicts firms will make relatively large, infrequent changes. The other assumes convex adjustment and predicts frequent, partial adjustments. Survey data of firms'price behavior reveal patterns consistent with the (s,S) model. However, many of the patterns are also consistent with partial-adjustment rules, although the high percentage of firms which fix prices for a quarter or more casts doubt on the plausibility of the partialadjustment hypothesis. I. INTRODUCTION In new Keynesian economics price stickiness arises not because of wage rigidity so much as because either the gains to individual firms from changing prices are negligible or the of doing so deter full and immediate changes. As a result, nominal shocks which affect demand for the firm's product, as well as real shocks, give rise to changes in output. The nature and the timing of the effects of demand changes, such as those initiated by monetary policy, thus depend on how individual firms respond to the signals they receive. Sticky-price models based on of adjustment take at least two different forms. One assumes a lump-sum to changing prices, which leads to predictions firms generally keep prices fixed and make infrequent, relatively large changes. The other assumes convex adjustment and leads to the prediction firms make relatively small and frequent partial adjustments toward a target level. Evidence drawn from a series of surveys conducted by the National Federation of Independent Business will be used to test the predictions from these two types of models. While the data do not allow a definitive test between the two models, the available evidence supports the model with lump-sum of changing prices with the important proviso there be substantial heterogeneity in the relative to firms of making changes so the frequency and size of changes vary considerably across firms. II. (s,S) RULES WITH LUMP-SUM COSTS OF PRICE CHANGES How frequently should a firm change its price? When a change takes place, how large should it be? Sheshinski and Weiss [1977] and Barro [1972] provided the first answers to these questions. In their models an individual firm incurs a lump-sum when changing its nominal price, and the firm's profits depend on its real price, where real is defined as the firm's nominal divided by an index of the prices of all other firms in the economy. Sheshinski and Weiss assume the firm tries to maximize the present value of its real profits. Their solution amounts to what is known in the optimal inventory literature as an (s,S) rule, in which S-s units are ordered when stocks run down to s. As applied to a pricing rule, when prices rise elsewhere in the economy, the firm's real falls to a lower bound s, at which point the firm raises its nominal so its real jumps to S. They prove a more rapid general inflation rate calls for a lower s and higher S. Furthermore, subject to a monotonicity condition, a higher general inflation rate calls for a shorter time between changes. After showing a number of numerical examples, Sheshinski and Weiss report that numerical experiments with a quadratic profit function ... give high intervals between changes (1-2 years) even with very low adjustment costs [1977, 300]. When the profit function is flat in the neighborhood of the profit-maximizing real price, there is little benefit from changing price. A similar argument appears in the menu cost explanation for real effects of nominal disturbances. In the absence of private incentives for individual changes, there may be real aggregate effects of nominal disturbances. For example, see Mankiw [1985], Akerlof and Yellen [1985], Kuran [1986], Blanchard and Kiyotaki [1987]. …
Read full abstract