The Espositos hypothesize that percentage change in price is a function of share-weighted percentage changes in unit labor costs, unit material costs and the ratio of depreciable assets to output plus the percentage change in the ratio of inventories to sales (a demand indicator). They replace the change in unit labor cost by measures of wage change and productivity change to test their oligopoly model for which the price response to wage increase is expected to be larger than the response to productivity increase. No lagged variables are introduced.

For both time periods they conclude that the highly concentrated industries respond faster and more fully to changes in labor cost or material cost than the moderately concentrated industries. In the 1963-66 period the least concentrated industries were also least responsive to cost changes, and the only group to respond to the demand variable. In 1966-69 the low group showed about the same response to cost change as did the high group (both were more responsive than the moderately concentrated industries) and none of the groups was directly responsive to demand. This finding is consistent with the expectation that uncertainties of conjectural variations are most important in industries at the middle level of concentration. The Espositos reason that the middle group was less responsive to cost changes in 1966-69 than in 1963-66 because during the inflationary period of 1966-69 cost changes were larger and more frequent than they had been in the earlier period.

They also find that the introduction of separate wage and productivity variables leads to improvement in the fit of the price adjustment equation for the moderately concentrated industry group. This tends to indicate a reliance on a standard cost pricing rule. They find no significant difference in overall fit for the high and low groups in 1966-69 but both differ from the moderate group. Prices in the moderate group responded only partially to cost change and thus tended to lag behind prices in the unconcentrated and the highly concentrated industries during 1966-69.1/

1/ There is an unfortunate technical error which may affect Some or all of the conclusions of the Esposito paper. They have regressed price ratio on the other variables also in ratio form. They rely on the correct theorem that when cost variables are unweighted, regression with ratios will give coefficients identical to those obtained using percentage changes to justify using ratio in a regression with shareweighted cost variables. The weights make a difference. See Appendix C. The Espositos have, in effect, cited the equivalence of Equations (la) and (lb), but then treated Equation (3) as equivalent to (2). This is not justifiable.

Cagan's first results are those reported in Table 8. These assume the same response to weighted cost changes for industries at all levels of concentration but allow for an effect of the concentration ratio. As was noted before (p. 35 supra), in 1967-69 the CR coefficient is negative, indicating smaller price increase for the more concentrated industries; then, in the expansion of 1970-71 the coefficient becomes strongly positive.

When the industry groups are allowed to have different response rates to cost changes, the middle group is least responsive, at least for the period 1967-69.1/ This result confirms that of the Espositos, although the coefficient differences do not appear to be statistically significant. In 1969-70 and 1970-71 no clear pattern emerges as to the size of response coefficients. In general, it appears that material costs are passed on quickly by all three groups; 2/ the response to labor cost change is quite variable, and none of the coefficient estimates for labor cost is significant after 1967-69. Quantity sold is introduced in some regressions to indicate demand. The coefficients are never significant and the coefficients of other variables are not greatly affected by the presence or absence of the quantity change.

Cagan interprets his results as showing differential lags in response to cost changes as between concentrated and unconcentrated industries. The evidence is not inconsistent with that interpretation, but neither is it compelling. Even after allowing for the estimated effects of current cost changes, the concentration ratio has a significant negative coefficient in 1967-69 and significant positive coefficient in 1970-71,3/ but that in itself does

1/ Cagan, "Inflation and Market Structure, 1967-1973," Table 3, p. 18.

2/ Exception: the high concentration group in 1969-70. 3/ A technical point of possible importance: Cagan has not done independent regression for the three separate data sets, as the Espositos did. When he allows different cost response coefficients he has still forced the intercept to be the same for each group. There is no good reason for so requiring. And the results he reports in Table 5 give one reason to think quite different estimates would be obtained. The detected effect of CR may simply be to accomplish the unpermitted shift of the intercept. If it should turn out for 1970-71 that the material cost response coefficients remain approximately one for all 3 groups, and the labor cost response coefficient about zero, but the CR (Footnote continued on following page).

એ

not suffice to demonstrate that the increase of the later period is a lagged response to earlier cost increases, or that it is limited to a "catch-up." What about the fact that the response to labor cost change is so erratic? Why not introduce lagged cost changes directly and attempt to capture the hypothesized effect in a regression?

The lag hypothesis is the best we have but there is much more to be done before we understand the role of market structure or concentration in price adjustments. And even more work is needed to see how concentration may affect the adjustments of wages or other costs to demand and anticipated price change.

IV

Empirical evidence supports the view that prices in the concentrated industries behave somewhat differently than prices in the unconcentrated industries over the cycle. There is no recognizable long-run disparity in the average rates of price change between concentrated and unconcentrated industry, but there does seem to be more cyclical stability in the prices of concentrated industries. On average they appear to rise less than competitive prices in expansions but have recently fallen less or risen more in recessions. There is great disparity in industrial price movements all the time. Level of concentration in no period explains directly a large part of the variation in prices.

The price adjustment process is not well understood and has not been extensively researched. Such evidence as there is suggests that prices in neither concentrated nor unconcentrated industries respond strongly to demand changes; measurement of demand change may be a major problem in this result, however. Oligopolistic or moderately concentrated industries tend to show smaller within-period response to changes in cost than do unconcentrated industries. This is consistent with a hypothesis that concentrated industries

(Footnote continued from previous page).

coefficient becomes zero and differing intercepts are found, what should one conclude? We still have only conjectures which would explain the differences in intercept. Of course, it may also happen that the response coefficients change once the arbitrary restriction on the intercept is abandoned.

experience a lagged reaction to cost changes; this might cause prices to rise more slowly in expansion and then continue up in a subsequent recession. But no attempt has been made to introduce lags and estimate this delayed reaction directly. Nor is the evidence of smaller initial response especially clear cut or strong; unless the lag is sizable it is not likely to carry over in reversals during later periods.

Research on the dynamics of the problem must also take into account the effects of demand changes and expected price changes on costs. If concentrated industries transmit demand shifts in such a way as to speed or increase wage or other cost changes in other sectors first then it may be possible that prices in concentrated industries lag not because of delayed reaction to cost changes but through prompt reaction to relatively delayed cost increases. There is little that is well understood here.

Ackley, Gardner. "Administered Prices and the Inflation Process," American Economic Review, XLIX, 2 (May 1959), 419-430.

Backman, Jules. "Economic Concentration and Price Inflexibility," Review of Economics and Statistics, XL, 4 (November 1958), 399-404.

Blair, John M. "Administered Prices: A Phenomenon in Search of a Theory," American Economic Review, XLIX, 2 (May 1959), 431-450.

"Market Power and Inflation: A Short-Run Target Return Model," Journal of Economic Issues, VIII, 2 (June 1974), 453-478.

"Means, Thorp and Neal on Price Inflexibility," The Review of Economics and Statistics, XXXVIII, 4 (November 1956), 427-435.

ed. Roots of Inflation (New York: Burt Franklin and Co., Inc., 1975).

Bronfenbrenner, Martin, and Holzman, Franklyn D. "A Survey of Inflation Theory," American Economic Review, LIII, 4 (September 1963), 593-661. Cagan, Phillip. "Changes in the Recession Behavior of Wholesale Prices in the 1920's and Post-World War II," Explorations in Economic Research, 2, 1 (Winter 1975).

The Hydra-Headed Monster: The Problem of Inflation in the United States (Washington: American Enterprise Institute for Public Policy

Research, 1974).

Dalton, James A. "Administered Inflation and Business Pricing: Another Look. The Review of Economics and Statistics, LV, 4 (November 1973), 516-519.

DePodwin, Horace J., and Selden, Richard T.

"Business Pricing Policies and Inflation, Journal of Political Economy, LXXI, 2 (April 1963), 116-127.

Eckstein, Otto, ed. The Econometrics of Price Determination (Washington: Board of Governors of the Federal Reserve System and Social Science Research Council, 1972).

and Fromm, Gary. "The Price Equation," American Economic Review, LVIII, 5, pt. 1 (December 1968), 1159-1183.

Eichner, Alfred A. "A Theory of the Determination of the Mark-up Under Oligopoly," Economic Journal, 83 (Decmeber 1973), 1184-1200.