Cost economies: a driving force for consolidation and concentration?

• Author: Morrison Paul, Catherine J.

1. Introduction

   Many industries in the United States and other developed countries have experienced increased concentration and consolidation in the past few decades, stimulating concerns about market power and questions about the factors underlying such trends. Evaluation of market power often takes the form of measuring perceived output demand relationships and imputing the implied price-marginal cost deviations. However, understanding market structure patterns in more depth requires obtaining comprehensive information about the cost (supply) structure, which is precluded by the restrictive cost assumptions of most market power models. Since cost economies imply greater efficiency of larger scale or more diversified operations, this may seriously limit the insights obtainable from such models.

   Cost economies--including short-run (utilization), long-run (scale), scope (jointness), and multiplant (information- or risk-sharing) economies--may provide economic incentives for expanding throughput, size, diversification, and plant numbers. Thus, information on the existence and extent of a variety of potential cost economies is key to the relevant construction, interpretation, and use of market structure and power measures. It provides insights about why concentration might have increased, whether these trends could be welfare enhancing rather than harmful, and what trends to expect in the future that are crucial for guiding policy measures to monitor and control market power.

   The importance of the cost structure as a potential driver of market structure patterns and thus of a comprehensive empirical representation for its evaluation has been recognized at least since the advent of the New Empirical Industrial Organization (NEIO) literature. (1) The literature on regulation and natural monopoly, targeting primarily public utilities, has also focused on the extent and form of cost economies and how they affect appropriate policy formulation. (2) But for most industries, the cost structure has received little attention in models of market structure, power, and welfare, which are often based on assumptions of constant returns to scale or a simple proportional cost-output relationship. Even when the existence of cost economies is acknowledged, consideration is typically restricted to scale economies. The cost interactions that drive economic behavior and potentially motivate firms to expand and diversify are, however, much more complex than this suggests. Recognition of a broader range of cost economies is therefore required to facilitate analyses of cost and market structure patterns and interactions.

   One industry (which has generated significant concern about market power because it has been dominated by large firms, has exhibited high measured concentration levels, provides a fundamental consumption commodity, and draws its supplies from the primary agricultural sector) is the U.S. beef packing industry. Although concentration in this industry declined after reaching very high levels in the late 1800s to mid-1900s because of both technological changes and regulatory measures, (3) plant and firm size again increased dramatically after various structural changes occurred in the late 1900s.

   As documented by USDA/GIPSA (1996), (4) the four largest (steer and heifer) beef packers accounted for only 36% of the market in 1980, but this percentage had increased to 72% by 1990 and 82% by 1994. Concern about this trend is evident from the many studies of market structure for this industry in the past two decades. (5) Although a crucial factor underlying such patterns is the cost or technological structure, the cost side of the problem has usually been finessed in these studies, severely limiting the appropriate interpretation and use of resulting market structure and power measures. (6)

   In particular, the role of cost economies is often alluded to but not effectively addressed in this literature. USDA/GIPSA (1996) stresses that understanding such economies is central to assessing whether "potential efficiency gains of larger firms offset potential adverse market power effects of concentration" (p. iii) and determining "the role of Federal regulation in preventing large firms from abusing potential market power, and in monitoring the industry" (p. iii). Queries are posed about whether cost (scale and scope) economies result in enhanced efficiency of large or diversified operations and so have a role in driving observed structural changes. Questions arise about the importance of maintaining high utilization levels, given large capital stocks and rigidities. The potential that these technological cost economies occur in combination with pecuniary economies, reflecting input market power if large operations exhibit buying power in cattle markets, is underscored. But the lack of evidence about th ese patterns in the existing literature is also recognized and lamented.

   In this study, I explore these technological and market characteristics and interrelationships. I use monthly cost and revenue data from the USDA/GIPSA survey of the 43 largest U.S. beef packing plants in 1992-1993 and a cost function-based model to represent the cost structure of these plants. I incorporate profit maximization over cattle purchases and fabricated, slaughter, hide, and by-product production. Also, I take regional, firm, category (plant type), and monthly differences into account as fixed effects. The resulting estimates and constructed measures facilitate the empirical identification and evaluation of cost economies associated with utilization, scale, scope, and multiple plants.

   My results suggest that substantive utilization economies exist, causing large plants in particular to value keeping throughput (and thus cattle input) at high levels. Significant scale and scope economies also prevail, (7) providing driving forces toward large and diversified plants. Multiplant economies are evident, although they are less substantive than those embodied in individual plants. These technological economies are slightly counteracted by pecuniary diseconomies deriving from the cattle input market, although the latter appear insignificant and neutralized by utilization economies. Overall, these cost economies cause marginal costs to fall significantly short of average costs. Thus, measures of market power based on simple cost structure (and thus marginal cost) assumptions are at best limited in their interpretability and at worst may be erroneous.

2. The Cost Structure Model and Measures

   The Model

   A cost function model provides a natural foundation for a detailed structural characterization of costs for U.S. beef packing plants. Such a model may be represented in general form as TC(Y, p, r, T) = VC(Y, p, r, T) + [summation over (k)][r.sub.k][p.sub.k], where VC is variable costs; Y is a vector of M outputs, [Y.sub.m]; p is a vector of J variable input prices, [p.sub.j]; r is a vector of K restricted or control variables, [r.sub.k], with market prices [p.sub.k]; and T is a vector of external shift factors.

   This framework may be used to represent many aspects of the cost structure. (8) In particular, cost effects may arise from input substitutability, utilization changes (due to input fixities), scale economies (and biases), scope economies (output jointness), and other interdependencies such as endogenous input prices (buying power) or multiplant economies. (9) These cost structure characteristics can be represented via first and second derivatives and elasticities of the cost function if a flexible functional form including a full range of cross effects is estimated, price endogeneity for netputs potentially subject to market power is recognized, and fixed effects representing firm affiliation are accommodated.

   Modeling and measuring this range of output and input relationships also requires a detailed data set. For the present purposes, I use weekly data from a USDA/GIPSA survey of the 43 largest U.S. beef packing plants that I aggregated to the monthly level because many values had been interpolated from monthly data, and input-output relationships may not be well represented on a weekly basis. The data include information on the prices and quantities of both slaughter output and a number of fabricated (more processed) subcategories that I divided into by-product, hides, and "other" fabricated outputs (essentially boxed beef). My materials inputs include livestock (cattle), measured in terms of carcass weight and delivered price, as well as purchased or transferred beef products and values of "other" purchased materials inputs. Hours worked and the associated wages/hour comprise the labor quantity and price measures. Data on fuel and electric expenditures and quantities were used to construct my energy price and q uantity measures. Summary statistics for and some elaboration of these data are presented in Appendix A; further details may be obtained from USDA/GIPSA (1996). (10)

   For the empirical cost specification, I thus have four products in the Y vector: slaughter and fabricated meat products ([Y.sub.S] and [Y.sub.F]), by-products ([Y.sub.B]), and hides ([Y.sub.H]). My five variable inputs are labor (L), energy (E), purchased beef ([M.sub.B], where M denotes "materials" and B "beef"), "other" materials ([M.sub.O]), and the primary cattle input (C). I also include one fixed input, capital (K), represented by the reported replacement cost of the plant (as discussed further in Appendix A). The p vector includes the prices of inputs that may be assumed to be variable, to have appropriately measured prices, and to possess competitive markets. (11) Demands for these inputs are thus represented by Shephard's lemma, [v.sub.j] = [partial]VC/[partial][P.sub.j], for j = L,E,[M.sub.B]. The remaining three inputs are treated as r vector components for various reasons.

   First, [M.sub.O] is reported in (nominal) dollar values rather than (real) quantities...