Dynamic competition and price adjustments.

Feinberg, Robert M.

I. Introduction

The link between price rigidity and market structure has long been a topic of interest to researchers in both macroeconomics and industrial organization. Research on this subject, with its antecedents in the widely debated topic of administered inflation, has gone through several stages. In recent years the focus of research has been on explaining the speed of price adjustment [8; 7; 11; 4; 2; 3]. A parallel line of research has been on the role of entry in disciplining the price behavior of incumbent firms [15; 14; 5].

The current study extends previous work on price adjustment in several important ways. Most importantly, we introduce dynamic elements of competition as measured by several disaggregated measures of domestic and import entry. We are able to evaluate the extent to which price effects of entry into U.S. manufacturing are due to changes in desired prices or to changes in the speed of adjustment to desired prices. We depart from previous studies in that we use pooled data to estimate, by nonlinear regression methods, both the desired (target) price change and the speed of adjustment across industries.

II. Theoretical Considerations and Previous Results

While there is a growing body of evidence to confirm the phenomenon of sticky prices, there is no consensus on a theory to explain why prices in some industries are inflexible. Means [16] argued that market power enabled firms to circumvent the normal workings of the market and to control prices and as a result prices tend to be changed less often in administered markets. Since Means' study, many alternative theories have been proposed.(1) These theories include the kinked demand curve, intertemporal substitution, the ability to store inventory, differences between produce-to-stock and produce-to-order industries, differences in the transaction costs of price changes, asymmetries in information between buyers and sellers, flexibility of production [17], and contractual relationships arising from transactions cost factors [4].

Despite different theoretical underpinnings, previous empirical tests have for the most part been similar in terms of both explanatory variables and econometric modeling. The typical model used is a partial adjustment equation which has the following form:

[Mathematical Expression Omitted]

where

[P.sub.t] = price in period t

[Mathematical Expression Omitted] = target price in period t

[Lambda] = the partial adjustment coefficient.

Price is adjusted to a target price (or equilibrium) level in each period, and it is a weighted average of the target price and the previous period's price.

The target price is usually described as a function of cost, demand and the degree of competition or market structure. With the introduction of the adjustment factor to price studies, the focus of research has shifted from examinations of concentration's impact on price changes [1] to tests of its effect on the speed of adjustment, where concentration is the conventional proxy for market structure.

In the present study we adopt an alternative approach, used recently in several studies,(2) based on taking first differences in equation (1) and examining [Delta][P.sup.*] rather than [P.sup.*]:

[Mathematical Expression Omitted]

We start out with a supply and demand framework where the supply equation includes a variable for the number of firms in the industry. After taking first differences the latter variable is transformed into domestic entry and, similarly, changes in import levels represent import entry. Given that our focus is on the effects of entry, equation (2) which is in terms of changes is the appropriate model.(3) Additionally, differencing has the advantage that it results in the removal of industry specific effects.

III. Empirical Model

The primary focus of the current study is on the impact of dynamic competitive factors on price changes. We can determine the extent to which price discipline from entry is reflected in smaller desired price changes or a slower adjustment to mostly higher prices.

Following the discussion in the previous section the relevant explanatory variables for [Delta][P.sup.*] are cost and demand changes and changes in actual and potential competition.(4)

[Delta][P.sup.*] = g([Delta]cost; [Delta]demand; domestic entry; [Delta]imports) (3)

The expectation is for a positive relationship between cost and demand changes and [Delta][P.sup.*], while domestic and import entry should generally have a restraining effect on [Delta][P.sup.*].

The partial adjustment coefficient, [Lambda], is assumed to be affected by factors reflecting static and dynamic competitive conditions.(5) Specifically, we include concentration as an indicator of internal competitive conditions, while domestic entry and changes in imports are indicators of dynamic factors and external shocks.

Increased market concentration is expected to prompt a slower price response to changing market conditions. This prediction was initially proposed by Means [16]; more recently, the argument has been made that concentrated industries can afford a long run perspective and hence feel less compulsion to respond to every change in supply and demand with a price change [6, 713]. The impact of entry and import changes on the speed of adjustment is subject to conflicting hypotheses. Entry and import changes act as shocks that may serve to disrupt the transmission process and delay the attainment of the desired price change.(6) Alternatively, intensified competition resulting from entry may lead to an emphasis on short run profitability and quick price adjustments as opposed to a long run view that would suggest less frequent changes consistent with long run objectives.(7)

[Lambda] = h (concentration; domestic entry; import changes) (4)

The empirical model follows from equation (2) which contains two unobservable variables, [Lambda] and [Delta][P.sup.*]; consequently, equations (3) and (4) have to be identified and estimated indirectly.(8) Because of overidentification and nonlinear restrictions present in equation (2) we adopt nonlinear least squares (Gauss-Newton) estimation.(9) We model [Delta][P.sup.*] and [Lambda] in the following way:(10)

[Delta][P.sup.*] = [[Alpha].sub.0] + [[Alpha].sub.1] COSTCHG + [[Alpha].sub.2]DEMCHG + [[alpha].sub.3]ENTRY1 + [[Alpha].sub.4]ENTRY2 + [[Alpha].sub.5]ENTRY3 + [[Alpha].sub.6]CHOECD + [[Alpha].sub.7]CHNOECD (5)

[Lambda] = [[Beta].sub.0] + [[Beta].sub.1]CONC + [[Beta].sub.2]ENTRY1 + [[Beta].sub.3]ENTRY2 + [[Beta].sub.4]ENTRY3 + [[Beta].sub.5]CHOECD + [[Beta].sub.6]CHNOECD. (6)

The sample consists of 31 4-digit industries over 2 time periods. The variable definitions and sources are the following (descriptive statistics are reported in Table I):

PPCHG = percentage change in 4-digit SIC level producer price index (1972 = 100), between 1972 and 1977, and 1977 and 1982, and for the lagged price change ([Delta][P.sub.t-1]) we also use 1967 to 1972 data [24; 23];

COSTCHG = percentage change in average variable cost, measured by payroll plus materials cost as a share of value of shipments multiplied by the appropriate GNP price deflator, calculated at the 2-digit SIC level and assigned to all included 4-digit industries [19; 21];

DEMCHG = percentage change in apparent domestic consumption (value of shipments + imports - exports) deflated by the appropriate GNP deflator, calculated at the 2-digit SIC level and assigned to all included 4-digit industries(11) [20];

CONC = the 4-firm seller concentration ratio, as adjusted by Weiss and Pascoe [25] for 1977;

ENTRY1 = number of new entrants diversifying from another industry by changing production mix from an existing plant, as a percentage of incumbent firms in the previous time period;

ENTRY2 = number of new entrants diversifying from another industry by building a new plant, as a percentage of incumbent firms in the previous time period;

ENTRY3 = number of new entrants not previously in manufacturing entering by building a new plant, as a percentage of incumbent firms in the previous time period; (ENTRY1, ENTRY2, and ENTRY3 are available for 1972-77 and 1977-82 from Dunne, Roberts, and Samuelson [10]);

CHOECD = 5-year change in OECD-imports as a percentage of U.S. apparent domestic consumption;

CHNOECD = 5-year change in non-OECD imports as a percentage of U.S. apparent domestic consumption; (CHOECD and CHNOECD are for 1972-77 and 1977-82 [22]).

Table I. Summary Statistics (N = 62)

VARIABLE MEAN STANDARD DEVIATION MINIMUM MAXIMUM

PPCHG 50.2 23.4 -9.4 111.9
COSTCHG 50.0 14.2 28.7 87.5
DEMCHG 2.7 13.0 -27.0 22.9
ENTRY1 0.1 0.09 0 0.4
ENTRY2 0.03 0.03 0 0.1
ENTRY3 0.15 0.1 0 0.4
CHOECD 0.4 1.7 -3.7 7.8
CHNOECD 1.5 3.6 -5.0 14.8
CONC 50.4 19.6 17.0 91.0

Because the focus is on the impact of entry on price changes, and given that entry data are available only for periods of five years, we have to use a corresponding period for price changes. With these rather lengthy time intervals we are, in a sense, estimating average tendencies, involving a mix of short-run and longer-run adjustments.(12) It should also be noted that for 95 percent of the observations, [Delta]P represents increases.

With respect to [Delta][P.sup.*], we expect positive coefficients for COSTCHG and DEMCHG and negative coefficients for new capacity entry (ENTRY2 and ENTRY3) and the two import change variables. The sign for ENTRY1 is less clear. The added output should have a moderating influence on [Delta][P.sup.*] but as ENTRY1 does not involve new capacity, a weak impact may be expected. In fact, a potentially offsetting influence, applicable to ENTRY2 as well, is that the removal of the entering firms as potential competitors may have a positive effect on the desired target price. With respect to the variables in the [Lambda] equation, concentration is expected to have a negative effect, whereas the impact of dynamic competitive factors, as explained before, is subject to alternative hypotheses.

IV. Empirical Results

In Table II we present results of nonlinear estimation (Gauss-Newton). In the [Delta][P.sup.*] equation, cost changes display a significantly positive effect. A one percentage point increase in costs results in a 0.74 percentage point increase in the desired price change. Demand change has the expected positive sign, however its effect is statistically insignificant.(13)

The results obtained for the dynamic competitive factors are particularly interesting. With regard to domestic entry's effect on the desired price change we see that domestic entry's impact varies by type of entry. The adding of new capacity by firms new to the manufacturing sector (ENTRY3) results in a significantly negative impact on desired price changes. At the mean rate of entry of this type, 0.15, the regression results imply an 11.0 percentage point decline in the desired price change, consistent with the procompetitive effect usually associated with entry.

Surprisingly, entry of established firms via change in production mix (ENTRY1) seems to lead to a larger desired price change. Given that this is not a particularly robust result it is risky to read too much into it; however, as noted above, a possible explanatory factor could be the fact that potential competition is removed. The use of an alternative estimation technique, nonlinear three stage least squares, where ENTRY1 and the import change variables are treated as endogenous variables, also points to a positive coefficient for ENTRY1. Diversification through the construction of new plants appears to have little impact on the desired price change. Turning to the effects of import changes we find that both OECD and non-OECD import changes have a strong procompetitive effect.

With respect to [Lambda],(14) concentration is found to have the hypothesized negative effect on the speed of adjustment, with the coefficient significant at the 10 percent level. However, domestic entry, regardless of whether it merely involves a change in production mix, the construction of new capacity by a diversifying firm or by a brand-new firm, does not seem to have any discernible impact on [Lambda]. Changes in imports, though, from both OECD and non-OECD countries do appear to have a disruptive influence on the adjustment process and delay the move to the target price. Apparently, industries subject to large changes in import-entry cannot change prices as fast as industries that are not subject to such entry. Changes in imports from OECD nations display an effect that is larger than the effect of import changes from non-OECD sources. A one percentage point increase in the OECD import-share results in a 14 percent decline in [Lambda].

Given the 5-year periods it is not surprising that, on average, we find actual prices converging to desired prices. However, we do note that there are some large differences among individual industries in the speed of adjustment.

Table II. Price Changes and Speed of Adjustment (N = 62)

[Delta]P(*)

CONSTANT 16.64
 (12.48)

COSTCHG 0.74
 (0.24)(**)

DEMCHG 0.29
 (0.27)

ENTRY1 66.26
 (38.79)(*)

ENTRY2 40.65
 (86.48)

ENTRY3 -73.26
 (31.82)(**)

CHOECD -3.62
 (1.72)(**)

CHNOECD -1.82
 (0.89)(**)

[Lambda]

CONSTANT 1.41
 (0.32)(**)

CONC -0.0088
 (0.0049)(*)

ENTRY1 0.01
 (1.00)

ENTRY2 3.50
 (2.74)

ENTRY3 0.83
 (0.97)

CHOECD -0.14
 (0.04)(**)

CHNOECD -0.07
 (0.02)(**)

The numbers in parentheses are asymptotic standard errors.

* = significant at the 10 percent level for a 2 tail test.

** = significant at the 5 percent level for a 2 tail test.

V. Summary

The current study examines the role of dynamic competitive factors and thus provides a new perspective on price rigidity. The results indicate that new competition from imports tends to delay the adjustment process and not unexpectedly has a procompetitive effect on the magnitude of the desired price changes.

Perhaps most interestingly the effects of disaggregated domestic entry vary considerably by type of entry. New firm entry consistently reduces the desired price change, whereas capacity entry of existing firms from other industries has no discernible effect. Entry involving a change in production mix but no new capacity has a more ambiguous impact, suggesting a weak positive effect on desired prices. With regard to [Lambda], none of the three measures of disaggregated domestic entry appear to have any impact. It is clear that import-entry has a more clearly pronounced negative effect on the speed of adjustment than do measures of domestic entry.

Concentration has the expected negative effect on the speed of price adjustment, but its coefficient is just barely significant. This result, combined with the results for the entry and import competition variables, suggests that in analyzing price flexibility, the inherently static notion of concentration does not capture all dimensions of competition. Dynamic competitive factors are important determinants of the price adjustment mechanism.

We are grateful to an anonymous referee for helpful comments.

1. For a survey see Carlton and Perloff [6].

2. Encaoua and Geroski [11] and Domowitz, Hubbard and Petersen [9] use this approach. Encaoua and Geroski pay particular attention to static competitive factors, such as market concentration, import ratios and the degree of foreign ownership, as explanatory factors for [Lambda].

3. After differencing the supply and demand equations and following some manipulation we obtain an equation with [Delta]P on the left hand side. See also Feinberg and Shaanan [12].

4. For example Stiglitz [18] notes that in recessions the threat of entry is diminished and this permits firms to charge higher prices.

5. For additional factors that may affect the speed of adjustment, see Carlton and Perloff [6].

6. One suspects that a certain amount of asymmetry may exist with respect to entry's effect on [Lambda] for positive and negative [Delta][P.sup.*]. However, given that 95 percent of our data consists of increases in [Delta]P it becomes a somewhat moot point.

7. For a discussion on price responsiveness and time horizon, in the context of a price smoothing strategy, see Encaoua and Geroski [11].

8. For a similar formulation, applied to a study on the dynamics of concentration, see Geroski, Masson and Shaanan [13].

9. We also examine specifications where domestic entry and import entry are endogenous.

10. Both linear and nonlinear versions of [Lambda] are tested.

11. Both cost and demand are measured at the 2-digit level to avoid any simultaneity bias which might result from measurement at the 4-digit level. Both variables involve a price term in their calculation, demand in deflating nominal consumption, and cost in converting the ratio of cost to revenue to an average cost index.

12. For example, the price change from 1977 to 1982 is in part a short-run response to the cost and demand changes occurring between 1980 and 1982, and in part a longer-response to changes experienced in the 1977 to 1979 period.

13. Tests with changes in excess capacity as an additional variable to account for cyclical factors in [Delta][P.sup.*] did not improve the explanatory power of the equation.

14. A nonlinear specification for [Lambda] was also tested but the results are largely unaffected by this choice of specification.

References

1. Aaronovitch, S. and Malcom C. Sawyer, "Price Change and Oligopoly." Journal of Industrial Economics, December 1981, 137-47.

2. Bedrossian, Arakel and Demetrios Moschos, "Industrial Structure, Concentration and the Speed of Price Adjustment." Journal of Industrial Economics, June 1988, 459-75.

3. van Bergeijk, Peter A. G., Robert C. G. Haffner and Pieter M. Waasdorp, "Measuring the Speed of the Invisible Hand: the Macroeconomic Cost of Price Rigidity," Kyklos, Vol. 46, No. 4, 1993, 529-44.

4. Bradburd, Ralph M. and Richard E. Caves, "Transaction-Cost Influences on the Adjustment of Industries Prices and Outputs," Review of Economics and Statistics, November 1987, 575-83.

5. Bresnahan, Timothy F. and Peter C. Reiss, "Entry and Competition in Concentrated Markets." Journal of Political Economy, October 1991, 977-1009.

6. Carlton, Dennis W. and Jeffrey M. Perloff. Modern Industrial Organization. New York: Harper Collins, 1990.

7. Dixon, R., "Industry Structure and the Speed of Price Adjustment." Journal of Industrial Economics, September 1983, 25-37.

8. Domberger, Simon, "Price Adjustment and Market Structure." Economic Journal, March 1979, 96-108.

9. Domowitz, Ian, R., Glenn Hubbard, and Bruce Petersen, "Oligopoly Supergames: Some Empirical Evidence on Prices and Margins." Journal of Industrial Economics, June 1987, 379-98.

10. Dunne, Timothy, Mark J. Roberts, and Larry Samuelson, "Patterns of Firm Entry and Exit in U.S. Manufacturing Industries." Rand Journal of Economics, Winter 1988, 495-515.

11. Encaoua, David and Paul Geroski, "Price Dynamics and Competition in Five OECD Countries." OECD Economic Studies, 1986, 47-74.

12. Feinberg, Robert M. and Joseph Shaanan, "The Relative Price Discipline of Domestic Versus Foreign Entry." Review of Industrial Organization, April 1994, 211-20.

13. Geroski, Paul A., Robert T. Masson, and Joseph Shaanan, "The Dynamics of Market Structure." International Journal of Industrial Organization, April 1987, 93-100.

14. Harris, Maury N., "Entry and Long-Term Trends In Industry Performance." The Antitrust Bulletin, 1976, 295-314.

15. Levin, Stanford Lee. "The Impact of Entry on Large United States Industrial Corporations." Doctoral Dissertation, University of Michigan, 1974.

16. Means, Gardiner C. Industrial Prices and Their Relative Inflexibility. U.S. Senate Document 13, 74th Congress, 1st Session, Washington, 1935.

17. Mills, David E. and Laurence Schuman, "Industry Structure with Fluctuating Demand." American Economic Review, September 1985, 758-67.

18. Stiglitz, Joseph E., "Price Rigidities and Market Structure." American Economic Review, May 1984, 350-55.

19. U.S. Department of Commerce, Bureau of the Census, U.S. Census of Manufactures, various issues, Washington, D.C.: U.S. Government Printing Office.

20. -----, Bureau of Economic Analysis, (unpublished data).

21. -----, Bureau of Economic Analysis. Survey of Current Business, various issues. Washington, D.C.: U.S. Government Printing Office.

22. -----. Trade Statistics, (unpublished data).

23. U.S. Department of Labor. Handbook of Labor Statistics various issues. Washington, D.C.: U.S. Government Printing Office.

24. -----, Bureau of Labor Statistics. Labstat Database (unpublished data).

25. Weiss, Leonard W. and George A. Pascoe, Jr. Adjusted Concentration Ratios in Manufacturing, 1972 and 1977. Washington, D.C.: Federal Trade Commission Statistical Report, June 1986.