J Ind Compet Trade DOI 10.1007/s10842-013-0161-1

Who Benefits from Industrial Concentration? Evidence from U.S. Manufacturing Rigoberto A. Lopez & Elena Lopez & Carmen Lirón-España

Received: 21 November 2012 / Revised: 4 April 2013/Accepted: 15 April 2013 # Springer Science+Business Media New York 2013

Abstract This article estimates the impact of industrial concentration on aggregate welfare as well as consumer and producer surpluses taking into account market power and cost efficiency effects. Using a sample of 232 U.S. manufacturing industries, empirical results indicate that an across-the-board increase in concentration would enhance aggregate welfare in 69 % of the industries due to widespread efficiency gains, although these accrue mostly to producers and are not passed on to consumers. Further results indicate that greater concentration is likely to enhance aggregate welfare in industries with low or moderate initial concentration that exhibit economies of scale and have greater exposure to international trade. However, consumers benefit only from increased concentration in industries whose initial levels are low and which face more import competition, lower exports and smaller markets. Producers benefit in symmetrically opposite ways, except for the case of low initial levels of concentration. In the absence of government intervention, Pareto improvements wherein everyone benefits from greater concentration are only guaranteed in industries with low levels of initial concentration in which efficiency gains yield price reductions that benefit consumers as well as producers. Keywords Concentration . Welfare . Economies of size . Market power . Manufacturing JEL codes L11 . L60 . D43 . D61 . F12

R. A. Lopez (*) Department of Agricultural and Resource Economics, University of Connecticut, Storrs, CT 06268, USA e-mail: [email protected] E. Lopez Department of Economics, University of Alcalá, Madrid, Spain e-mail: [email protected] C. Lirón-España Massachusetts Department of Energy Resources, Market Intelligence, Boston, MA 02114, USA e-mail: [email protected]

J Ind Compet Trade

1 Introduction The conventional view that high levels of concentration yield excessive price–cost margins and, therefore, welfare losses, was challenged in the late 1960s by Williamson (1968) and has been studied more recently by Dickson and He (1997), Berger and Hannan (1998), Blair and Harrison (1999), Bian and McFetridge (2000), Feltovich (2001), Focarelli and Panetta (2003), Whitley (2003) and Brouwer (2008), as well as by other authors who have analyzed the consequences of the U.S. Department of Justice and the Federal Trade Commission (DOJ/FTC) 2010 Horizontal Merger Guidelines (Carlton 2010; Blair and Haynes 2011; Salinger 2011). By considering both the price and cost changes resulting from industrial concentration, these authors raised the possibility that concentration could give rise to welfare gains. However, their studies are constrained by specific market structure or conduct assumptions and do not establish a broader structural link between concentration and welfare consequences. European Union horizontal merger policy toughened after 1998 but became more in line with U.S. practices after 2004 in terms of the scope of the efficiency argument to counterbalance the price increasing argument (Brouwer 2008). However, the efficiency argument still carries a heavier weight in the U.S. than in the EU, where even mergers which lowered prices through efficiency, thus benefiting consumers, were considered harmful to competitors. EU Guidelines made merger regulation enforcement more closely resemble that of the U.S. in terms of the use of the Herfindahl index and the introduction of efficiency and failing firm defenses (European Council Regulation 2004). Thus, examining the efficiency and market power effects of concentration in the U.S. can provide useful insights for EU policy. In this article we estimate an oligopoly model that allows one to capture the effects of price and cost changes which result from further concentration. We then compute their impact on welfare changes to consumers and producers for 232 U.S. manufacturing industries and assess under which circumstances further concentration is more likely to be beneficial. We find that although concentration enhances aggregate welfare in 69 % of the industries due to widespread efficiency gains, consumers lose in 64 % of the cases while producers gain in 82 % of the cases, as efficiency gains accrue mostly to the industries themselves. Further findings show that increases in concentration augment social welfare in industries which are large, have economies of size as well as inelastic demand functions, are export oriented, produce consumer-oriented goods, and have low or moderate levels of initial concentration.

2 The model 2.1 Firm equilibrium The econometric model draws from the work by Azzam (1997) and Lopez et al. (2002). Consider P n firms, each producing the quantity qi (i=1,…, n) of a homogeneous product so that Q ¼ i qi is total industry output. The industry faces a market demand function given by Q = f(P, z), where P is the output price and z is a vector of demand shifters. A firm’s cost function is represented by ci(qi, w, t), in which w is a non-negative vector of exogenous input prices and t is the state of technology.

J Ind Compet Trade

Each firm is assumed to choose its output level in order to maximize profits (πi) given by: p i ¼ PðQ; zÞqi  ci ðqi ; w; tÞ:

ð1Þ

The first-order condition of (1) w.r.t. qi yields: P¼

mci ðqi ; w; t Þ  ; ÞSi 1 þ aþð1a η

ð2Þ

where α is Clarke and Davies’ (1982) collusion parameter and Si = qi/Q is the firm’s market share, η = ∂ln Q/∂ln P is the price elasticity of demand, and mci is the firm’s marginal cost (∂ci/∂qi <0). The cost function is assumed to take the restricted Generalized Leontief form: ci ðqi ; w; t Þ ¼ qi

m X m X j

m m X X  :5 g jk wj wk þ qi t g jt wj þ q2i bj wj : j

k

ð3Þ

j

where +jk, +jt and βj (j=1,…,m) are fixed parameters. Taking the derivative of [3] w.r.t. to qi and wj, one obtains the firm’s marginal cost and input demand equations. 2.2 Industry equilibrium Summing expressions [2] and [3] across firms in the industry, using the market shares as weights, one obtains the supply relation and the input demand functions for a particular industry: m P m P

P¼

j

k

m m  :5 P P g jk wj wk þ t g tj wj þ 2HQ bj wj j j   ; ÞH 1 þ aþð1a η

ð4Þ

m X m  :5 Xj X ¼ wk =wj þ g tj t þ HQb j ; Q j k

ð5Þ

n P where Xj is the total industry employment of the jth input, H ¼ Si2 is the Herfindahl Index, si = qi/Q α is the share-weighted conjectural variation, and η iis the price elasticity of demand. Let the share-weighted marginal cost (the numerator of [4]) be expressed by MC = B0 + m P m m m  :5 P P P g jk wj wk þ t g tj wj is the intercept and B1 ¼ b j wj is its 2QHB1, where B0 ¼ j

k

j

j

slope in quantity space. Thus, B1 equals zero, a negative, or a positive value for constant, increasing, and decreasing economies of size, respectively. Since B1 interacts with H, its sign determining the nature of efficiency impacts of concentration. Note that the industry-level Lerner index of oligopoly power can be expressed as L = −(α+(1−α)H)/η in the denominator of [4]. When α=0, then L = −H/η, which is the Cournot outcome, whereas when α=1, L=−1/η, which is the monopoly outcome. The perfectly competitive outcome is when both H, α=0, i.e., an extreme Cournot outcome with

J Ind Compet Trade

a totally unconcentrated industry, or when η approximates infinity. Let the output demand function take the logarithmic form:   X r P þ d l ln Zl ; ð6Þ ln Q ¼ d 0 þ η ln d l¼1 where d is a price deflator, Zl denotes exogenous demand shifters, and δ0, δl and η denote fixed parameters. Note that η, the price elasticity of demand, also appears in [4]. Market equilibrium in a particular industry is reached when P and Q fulfill the equations for the supply relation and demand function simultaneously. Alternatively, equilibrium occurs when the industry-level marginal revenue (MR)equals industry-level marginal cost (MC), where both MR and MC shift with concentration. Consequently, the elasticity of price with respect to the Herfindahl index (εP, H) is obtained by applying the implicit-function theorem to Eqs. (4) and (6): "P;H ¼



 MC ½1  aPH 2QHB1 þ ¼ l "L;H þ "CH ; MC þ 2HQB2 η MCη MC

ð7Þ

where εP, H is the sum of the Lerner index (market power) elasticity εL, H and the costefficiency elasticity εC, H, multiplied by the price-quantity adjustment λ. While the market power elasticity is always positive, the sign of the efficiency elasticity and the strength of the price-quantity adjustment will depend on the sign of B1. For instance, if B1 < 0, the efficiency elasticity is negative and the price-quantity adjustment is greater than 1. A rise in concentration in any industry can thus lead either to an increase or a reduction of the product’s price as well as changes in cost, yielding alternative welfare consequences for consumers and producers. The next section outlines three possible outcomes from an increase in concentration, including welfare impacts on consumers, producers and for the society as a whole.

3 Welfare impacts of changes in concentration 3.1 Measures of welfare changes Whether a rise in concentration enhances or reduces social welfare depends on the sign and value of the market power and marginal cost effects. The market power effect is represented by a downward shift of the perceived marginal revenue (MR). If the industry faces diseconomies of size, an increase in concentration generates cost efficiency losses that result in an upward shift of the industry marginal cost curve as a larger share of the output is produced by fewer firms. On the other hand, if there are economies of size, further concentration results in a downward shift of the industry marginal cost curve, as output is redistributed among larger firms that are cost efficient. The final impact of increased concentration can thus lead to any of the following three scenarios: (a) a cost efficiency loss with a price increase; (b) a cost efficiency gain with a price reduction; or (c) a cost efficiency gain with an increase in price. The change in social welfare is given by the sum of the changes in consumer and producer surplus. The change in consumer surplus (dCS) will always be negative if the increase in concentration yields a higher price and less quantity and will be

J Ind Compet Trade

positive when its impact on price is negative and positive on quantity. Analytically, dCS can be measured as: ZP1 dCS ¼ 

APη dP;

ð8Þ

P0

where P0 and P1 are the equilibrium prices before and after a change in the level of concentration.1 The term A subsumes exogenous demand shifters that remain constant. For a price increase, the associated change in consumer surplus can be divided into two parts. The first is the transfer of welfare from consumers to producers resulting from paying a higher price. The second is a deadweight loss from purchasing less quantity. As a percentage of sales, all one needs to know to assess consumer losses is the price elasticity of demand and the price elevation. The change in producer surplus due to increased concentration is given by the change in revenue minus the change in cost. The change in revenue arises from the fact that producers sell a different amount at a different price. For a price increase, the change in revenue from an increase in concentration can, thus, be denoted as the sum of the two elements: (1) a pure transfer from consumers to producers due to a higher price from a given quantity, and (2) a reduction in revenue caused by a reduction in output. Changes in cost can also be decomposed into two elements: (1) a change in cost due to a change in quantity produced; and (2) a change in the per unit cost of production due to concentration. Thus, the change in producer surplus can be represented by: " dPS ¼ ½P1 Q1  Q0 P0 

Q R0

#

ðB0 þ 2B1 QH1 ÞdQ ¼   ¼ ½P1 Q1  Q0 P0   Q2 0 B1 ðH1  H0 Þ þ B0 ðQ1  Q0 Þ þ H1 B1 Q1 2  Q0 2 0



ð2QB1 ðH0  H1 ÞÞdQ þ

Q R1

Q0

ð9Þ

where H0 and H1 are the initial and final levels of concentration. A total welfare standard assumes that society as a whole can potentially benefit from increases in concentration as long as the sum of the changes in consumer surplus and producer surplus is positive, i.e., dSW = dCS + dPS > 0. Figure 1 illustrates the impact of an increase in H with diseconomies of size, resulting in a higher price and less quantity as MC increases and MR decreases with concentration. In this case consumer surplus decreases by areas T and G, where T represents a pure transfer from consumers to producers and G is the deadweight loss associated with the reduction in quantity. The change in producer revenue is given by T−(F1+F2+F3), where T represents the increase in revenue due to a higher price, and (F1+F2+F3) represents the revenue foregone due to a reduction in output sold. The change in cost is represented by E - F3, where E is the cost increase due to the efficiency loss at the ex-ante level of output and F3 represents the reduction in cost due to smaller production. The net change in producer surplus is thus given by T – E - F1 - F2, which can be either positive or negative. The total welfare effect is -(G+E+F1+F2), which is always negative for a price increase with diseconomies of scale. 1

A convenient way to compare consumer losses is to express them as a percentage of sales. Let K = P1/P0 denote the ratio of the final to original price and let dcs = dCS/(P0 Q0), i.e. the change in consumer surplus as 1 ðK 1þη  1Þ . Under constant returns to a percentage of observed sales. Then (4) can be expressed as dcs ¼ 1þη scale, this expression also represents the aggregate welfare losses as a percentage of sales.

J Ind Compet Trade P

P1 P0

G

T

MC1(H1) F1 MC0(H0) F2 E D MR0(H0)

F3

MR1(H1) Q1

Q0

Q

Fig. 1 Price increase with efficiency loss

Figure 2 illustrates the impact of an increase in H with cost efficiency gains that outweigh the market power effect, resulting in a price reduction. The increase in consumer surplus is given by T+G, and the change in producer revenue is represented by the sum of areas F1+ F2+F3−T. The change in production cost is represented by F3 - E and the net change in producer surplus is F1+F2+E−T, which can be either positive or negative. 2 The change in total welfare is G+E+F1+F2, which is always positive.3 3.2 Determinants of the impacts Computationally, the size and sign of welfare changes from increased concentration depend on the pricing conduct parameter, economies of size, the price elasticity of demand, and the size of the industry. The first factor underlying the impact on welfare changes is the initial level of concentration (H0). Ever since the seminal paper by Cowling and Waterson (1976) there has been a well-established negative relation between concentration and social welfare. However, due to the possibility of economies of size, the relation between H and dSW is unpredictable a priori. The 2010 merger guidelines of the U.S. Department of Justice and the Federal Trade Commission imply that the initial level of concentration has a crucial impact on the balance of cost and market power effects. Accordingly, we expect further concentration from already high initial levels to be more likely to have a negative impact on social welfare. Economies of size are expected to have a positive effect on dSW as we anticipate that concentration fosters cost efficiency in these industries. On the other hand, we expect the social welfare effect of increased concentration to be more beneficial the more inelastic the demand function, as a lower price elasticity of demand yields larger transfers of welfare from consumers to producers and smaller consumption-related deadweight losses. Following 2

However, as shown by Harris (1988), economies of size can also act as entry barriers, thus amplifying the market power effects that could reduce welfare and reduce the pass-through of any cost savings. 3 A third scenario (not depicted graphically) is one with a cost efficiency gain that is not strong enough to outweigh the market power effect, thus leading to an increase in price, which can result in either positive or negative total welfare change.

J Ind Compet Trade P

G

P0 P1

T F1

E MC0(H0)

F2

D MR0(H0) MC1(H1)

F3 MR1(H1) Q0

Q1

Q

Fig. 2 Price decrease with efficiency gains

Pagoulatos and Sorensen (1986), consumer-ready goods (as opposed to intermediate goods used in further processing) have a lower price elasticity of demand and are more likely therefore to have a positive impact on dSW. Another element involves the role of trade and market size. As Stahlhammer (1991) shows, one can expect the market power effect of concentration to be mitigated in industries facing import competition. Although exports are likely to have little impact on domestic market power (Pugel 1980; Marvel 1980), they can importantly affect the scale of operation, which could have a significant impact on efficiency (Lopez and Lopez 2003). Finally, the size of the market matters as larger industries tend to have lower markups and a higher propensity to exploit economies of scale and to export (Holmes and Stevens 2005; Melitz and Ottaviano 2008; Weder 2003). The following equation summarizes the relation between changes in social welfare from increases in concentration and the explanatory variables discussed above: dSWi ¼ 8 0 þ H0i ð 8 1 Dloi þ 8 2 Dmei þ 8 3 Dhii Þ þ 8 4 Desi þ 8 5 CONSi þ 8 6 IMPi þ 8 7 EXPi þ 8 8 VSi þ vi ;

ð10Þ

where v is a random error and φs are parameters to be estimated. The dependent variable dSWi is the social welfare change derived from a one-percent increase in concentration in industry i as a percentage of sales. The explanatory variables include three dummy variables to depict the critical levels of concentration following the 2010 DOJ/FTC guidelines: low levels (Dlo, for H<0.1), medium levels (Dme, for 0.10.18). Des is a dummy variable that denotes industries with economies of size. Stage of processing can then be measured by the percent of shipments to consumers (CONS). VS denotes the value of shipments to denote industry size and, IMP and EXP denote the values of c.i.f. imports and f.o.b. exports as a percentage of VS. To gain additional insight, two additional regressions are performed using the same explanatory variables: one for changes in consumer surplus and one for changes in producer surplus. The equations are estimated correcting for dependent-variable heteroskedasticity. The results are presented in the following section.

J Ind Compet Trade

4 Data and estimation An empirical model consisting of five structural equations depicting industry equilibrium is estimated. These consist of the pricing equation in (4), three input demand equations in (5) to help identify marginal cost parameters, and the output demand equation in (6). The sample consists of annual data for the period 1972–1992 for 232 U.S. manufacturing industries at the 4-digit SIC level (see definition of industries included in the Appendix). The inputs are divided into three categories: capital (K), labor (L) and materials (M). Income (Y) and time (t) are used as demand shifters. The endogenous variables are Q, P, and input quantities XK, XL, and XM. The exogenous variables consist of wK, wL, wM, y, t, d and H. Additional instruments used in the estimation included the prices of the inputs and the square values of the exogenous variables. The non-linear 3SLS estimation procedure was implemented using the SHAZAM 8.0 software to estimate the system of five equations for each industry in the sample. The equilibria P and Q before and after a 1 % increase in the Herfindahl index (at mean values) were determined simultaneously by solving Eqs. (1), (2), and (3) with the MATLAB software. The same software was used to compute the ensuing welfare changes for each of the 232 U.S. manufacturing industries in the sample.4 The main data source for prices and quantities of outputs and inputs was the online National Bureau of Economic Research database of Barstelman and Gray (1996) for U.S. manufacturing industries. Due to lack of data on the price of capital at the 4-digit SIC level, all industries are assumed to face the same rental prices but each to have a different level of capital stock. Therefore, the rental price of capital was computed by dividing the cost of capital services (provided electronically via personal communication by the Bureau of Labor Statistics) divided by total capital assets at the 2-digit SIC level. Also due to data limitations, we used maximum entropy with market shares derived from concentration ratios (Golan et al. 1996) to extrapolate Herfindahl indices to the years 1972 and 1977 and for the intercensus years.5

5 Empirical results Table 1 presents a summary of the results obtained from the estimations of market equilibrium for each industry. The average value of the Lerner Index of U.S. manufacturing industries is estimated at 0.291, with a standard deviation of 0.10, which indicates that the exertion of oligopoly power is an extended practice throughout the industries in the sample. Approximately 78 % of the industries have cost elasticities less than one, indicating a widespread prevalence of economies of size that translates into lower unit cost with higher

4 The initial sample consisted of 445 industries from Barstelman and Gray (1996). Of these, 153 had incomplete information, especially the Herfindahl index, after a change in SIC classification in 1987 or due to confidentiality reasons. In 60 cases, the model did not converge or produced pervasive results. These industries were excluded. 5 The Herfindahl index has only been published for 1982, 1987 and 1992. Given this paucity of data, the missing H’s were estimated in two steps: (1) the estimation of the 1972 and 1977 H indexes from sales concentration ratios available for those Census years, following Golan et al. (1996) (2) interpolation of the H indexes for the inter-Census years using regression on instrumental variables, following Chow and Lin (1971). These regressions yielded an average R-square of 85 %. The use of spline functions and exponential smoothing did not significantly alter the results.

J Ind Compet Trade Table 1 Summary of estimates of pricing conduct and economics of size Number of industries:

232

Lerner index Economies of size

Price elasticity of demand

Mean Value

0.291

Standard Deviation

0.10

Mean Value

0.94

Positive Negative

78 % 22 %

Mean Value

−0.64

Inelastic

81.6 %

Elastic

19.4 %

concentration. The average value of the own price elasticities is estimated at −0.64, and nearly 82 % of the industries in the sample are price-inelastic. Table 2 indicates that although a 1 % rise in concentration in all industries would increase aggregate welfare in nearly 69 % of the industries, consumers would lose in 64 % of the cases due to higher output prices. On the other hand, producer surplus would increase in 82 % of the industries, due to a combination of price increases (64 % of the cases) and cost savings (62 % of the cases) in most cases happening concurrently. These results show that although industrial concentration generally increases allocative efficiency, it tends to benefit industries more than consumers, although in some cases consumers stand to gain as well when cost efficiency lowers prices. Table 3 presents the summary of variables used to estimate Eq. (10). The overall values of H range between 0.002 for industry SIC 3273 (ready-mixed concrete) and 0.242 for industry SIC 3511 (turbines and turbine generator sets), with most of the industries in the sample Table 2 Price, cost, and welfare effects of a 1 % increase in the Herfindahl index Total

Low concentration Moderate (H<0.18) concentration (0.18≤H<0.25)

High concentration (H≥0.25)

All industries

# industries

#=177

%

#=45

%

#=10

%

#=232 %

177

Market power Marginal cost Price

100

45

100

10

100

232

Decreases 0

Increases

0

0

0

0

0

0

0

Increases

24.29

38

84.44

8

80

89

38.36

Decreases 134

75.710

7

15.56

2

20

143

61.64

Increases

62.71

30

66.67

8

80

149

64.22

37.29

15

33.33

2

20

83

35.78

43 111

Decreases 66 Quantity

37.29

15

33.33

2

20

83

35.78

Decreases 111 Consumer surplus Increases 66

62.71 35.77

30 15

66.67 33.33

8 2

80 20

149 83

64.22 34.29

Decreases 111

64.23

30

66.67

8

80

149

65.70

Increases

80.22

38

84.44

10

100

190

81.89

19.77

7

15.56

0

0

42

18.10

Producer surplus

Increases

66

100

142

Decreases 35 Aggregate effect

Increases

128

Decreases 49

72.32

30

66.67

2

20

160

68.97

27.68

15

33.33

8

80

72

31.03

J Ind Compet Trade Table 3 Summary of regression variables used in stage 2 Variables

Mean

St. dev.

Minimum

Maximum

dSW ($M)

102.53

218.46

−23.69

1,859.09

dPS ($M)

121.93

267.27

−158.35

2,225.82

dCS ($M)

−19.40

54.62

−419.54

138.48

H0 DHlo

0.07 0.76

0.05 0.43

0.002 0.00

0.24 1.00

DHme

0.19

0.40

0.00

1.00

DHhi

0.04

0.20

0.00

1.00

Economies of size

0.94

0.08

0.62

1.22

Consumer shipments

0.27

0.34

0.00

0.99

Import intensity

0.18

0.31

0.00

2.88

Export intensity

0.08

0.09

0.00

0.67

4,190.72

6,391.00

163.82

45,115.26

Domestic sales ($M)

(76 %) in the low concentration category (H<0.1), 19 % in the medium range (0.10.18). On average, the industries exhibited a moderate degree of economies of size (0.94), approximately 27 % of them shipped consumer-ready, finished products (rather than intermediate goods), imports amounted to 18 % of domestic production with some industries facing no import competition while others were dominated by it. Exports accounted on average for only 8 % of production, but some industries exported two-thirds of their output. On average, this table indicates that both aggregate welfare and producer surplus increase when concentration increases, but consumer surplus decreases. The regression results provide further insight into the drivers of the changes in welfare from increased concentration. Table 4 presents the parameter estimates for Eq. (10) as well as two alternative specifications of the dependent variable to disentangle specific impacts on changes in consumer and producer welfare as well. These results show that greater concentration is beneficial to both consumers and producers for industries with initially low levels of the Herfindahl index, possibly due to cost savings that offset any increased market power. For medium levels of initial concentration, further concentration harms consumers but benefits producers as producer gains more than offset consumer losses. However, when the initial level of concentration is high, consumer losses more than offset producer gains, resulting in aggregate social welfare losses. Thus, the general pattern is that concentration is always beneficial to both society and consumers at low levels of concentration. At medium levels of concentration, there is a pattern of trade-off between producers and consumers although the aggregate change in welfare tends to be positive. At high concentration levels (H>0.18), concentration cannot be defended in terms of its effects on either consumer welfare or allocative efficiency. Thus, overall, these results support the overarching 2010 U.S. DOJ/FTC merger guidelines recommendations in terms of acceptable levels of concentration. Table 4 also shows the impact of market structure variables and conduct on changes in consumer and producer surpluses. Regardless of the level of initial concentration, when industries with economies of size concentrate, there tends to be an expansion in output that causes prices to drop, benefiting consumers but hurting industries, albeit resulting in overall social welfare gains and lending support to the simulations conducted by Dickson and Yu (1989). Welfare gains tend to occur in

J Ind Compet Trade Table 4 Determinants of welfare changes from concentration Constant

Aggregate welfare

Change in: consumer surplus

Producer surplus

(Standard errors in parenthesis)

H × low concentration dummy H × medium concentration dummy H × high concentration dummy Economies of size dummy Consumer shipments Import intensity Export intensity Domestic sales

−0.17***

−0.22***

0.12***

(−0.01)

(−0.01)

(0.008)

1.06***

0.53***

0.18***

(0.16)

(0.03)

(0.03) 0.97***

0.66***

−0.44***

(0.08)

(0.03)

(0.11)

−0.15*** (−0.03)

−0.62*** (−0.12)

0.60*** (0.16)

0.20***

0.18***

−0.04***

(0.009)

(0.006)

(0.006)

0.03***

0.1***

−0.11***

(0.01)

(0.006)

(0.007)

0.005

0.02***

−0.02***

(0.01)

(0.004)

(0.002)

0.06* (0.03)

−0.09*** (0.009)

0.1*** (0.02)

0.02***

−0.03***

0.04***

(0.003)

(0.002)

(0.004)

The levels of statistical significance 10 %, 5 % and 1 % are represented by *, **, and ***, respectively. The regressions were corrected for heteroskedasticity

more consumer-oriented industries with more inelastic demand functions. Imports have a beneficial impact on consumer welfare but a detrimental one on producer welfare as industries concentrate, which roughly offset each other, yielding a quasi-neutral impact on social welfare. Export intensity has the opposite effect: it hurts consumers as industries concentrate since products are removed from domestic markets, but it is beneficial to producers since they can better exploit economies of size. The overall effect of concentration tends to be beneficial in industries with high export intensity, which lends support to the hypothesis that further concentration enhances cost-efficiency in export-oriented industries. Finally, the coefficient of the variable VS indicates that concentration is likely to have a beneficial impact in larger industries rather than smaller ones, consistent with the notion that larger industries tend to have smaller unit price–cost margins and that they are better able to exploit economies of size.

6 Conclusion Who benefits from industrial concentration? Our results, based on a full model of industry equilibrium for a sample of 232 U.S. manufacturing industries, show that an increase in the Herfindahl index would lead to welfare gains in nearly 70 % of the cases. However, the distributive results are mixed: consumers lose in 66 % of the cases while producers gain in 82 % of the cases, mainly due to higher prices and cost savings that are not fully transferred to consumers.

J Ind Compet Trade

The key findings of the empirical analysis are that further concentration in industries is beneficial to both consumers and producers only when the initial level of concentration is low, where everyone gains with concentration, resulting in a Pareto improvement. In contrast, increased concentration in industries with high initial levels of concentration (H>0.18) tends to be detrimental to society as a whole and particularly to consumers, but beneficial to producers. Regardless of its initial level, further concentration tends to enhance aggregate welfare in industries which are large, have economies of size and high export intensity, and produce consumer-oriented goods. On the other hand, in industries facing more import competition, further concentration has a positive impact on consumers, a detrimental impact on producers, and is nearly neutral for society as a whole. These results justify the U.S. Department of Justice and the Federal Trade Commission (2010) and European Union’s attention to mergers in already highly-concentrated industries and stress the need to scrutinize not just their net social welfare impact but also—and above all– the consequences for consumers. For as this study shows, while concentration generally increases aggregate welfare, it commonly leads to higher prices, as cost savings are usually not passed on to consumers. Import competition tends to mitigate this negative impact on consumers, but exports are a double edged sword for consumers: while they allow opportunities to increase economies of size and enhance efficiency gains, they also force domestic consumers to compete with consumers abroad, and this trade off does not appear to favor domestic consumers. Although many countries, including those in the EU, increasingly resort to the efficiency standard as an argument to diminish possible price enhancement from concentration, the lack of transmission of efficiency gains, up the chain towards the consumer, raises questions about its universal applicability. Obviously, competition policy, including trade policy, could affect the nature of the distributional effects of concentration. Ideally, industrial concentration would induce efficiency gains without price increases from increased market power. In the absence of government intervention, this outcome is only guaranteed in industries with low levels of initial concentration in the sample of industries analyzed. Acknowledgments The authors wish to thank two anonymous referees who provided helpful comments that significantly improved this article. We gratefully acknowledge financial support from USDA-NIFA-201034178-20766 grant, from the University of Alcalá, and from the Zwick Center for Food and Resource Policy (formerly the Food Marketing Policy Center) at the University of Connecticut.

Appendix. Industries included in the sample Food & kindred products (N=32) 2011 Meat Packing Plants; 2013 Sausages/Prep. Meats; 2021 Creamery Butter; 2022 Cheese, Nat. & Proc.; 2023 Cond. & Evap. Milk; 2024 Ice Cream & Frozen Desserts; 2026 Fluid Milk; 2032 Canned Specialties; 2033 Canned. Fruits & Veg.; 2034 Deh. Fruits/Veg/Soups; 2035 Pickles/Sauce/Salad Dr.; 2037 Frozen Fruits & Veg.; 2041 Flour/Other Grain Mill; 2045 Blended & Prep. Flour; 2046 Wet Corn Milling; 2047 Dog & Cat Food; 2048 Prepared Feeds, N.E.C.; 2061 Raw Cane Sugar; 2062 Cane Sugar Refining; 2063 Beet Sugar; 2066 Chocolate/Cocoa Prod.; 2077 An. & Mar. Fats & Oil; 2079 Short. & Cooking Oils; 2082 Malt Beverages; 2084 Wines/Brandy/Spirits; 2085 Distilled & Blended Liquors; 2086 Bottled and Canned soft drinks; 2087 Flavoring Extracts and Syrups; 2091 Canned/Cured Seafood; 2095 Roasted Coffee; 2097 Manufactured Ice; 2099 Other Food Products

J Ind Compet Trade

Textile mill and apparel products (N=25) 2211 Weaving Mills, Cotton; 2221 Weaving Mills/Synth.; 2241 Narrow Fabric Mills; 2251 Wom.s Hosiery, Ex.Socks; 2252 Hosiery, N.E.C.; 2261 Finishing Plants, Cotton; 2262 Finishing Plants, Synth.; 2284 Thread Mills; 2295 Coated Fab., Not Rubb; 2311 Men’s/Boys’ Suits/Coats; 2323 Men’s/Boys’ Neckwear; 2329 Men’s/Boys Cloth., N.E.C.; 2335 Women’s Dresses; 2339 Women’s O.wear, N.E.C.; 2341 Women’s & Children’s Underwear; 2361 Girls’ & Children’s Dresses & Blouses; 2371 Fur Goods; 2381 Fabric Gloves/Mittens; 2385 Wat. Pro. O. Garments; 2386 Leather/Sheep Lin.Cloth.; 2387 Apparel Belts; 2389 Apparel/Acc., N.E.C.; 2392 House Furn., N.E.C.; 2394 Canvas/Related Prod.; 2397 Schiffli Machine Embroideries Lumber, wood prods and furniture (N=21) 2411 Logging; 2426 Hwood Dim./Floor.Mills; 2431 Millwork; 2434 Wood Kitchen Cabinets; 2435 Hwood Veneer/Plywood; 2439 Struc.Wood Mem. N.E.C.; 2441 Nailed Wood Box/Shook; 2449 Wood Containers, N.E.C.; 2451 Mobile Homes; 2452 Prefab.Wood Buildings; 2491 Wood Preserving; 2511 Wood Household Furn.; 2512 Uphol. Household Furn.; 2514 Metal Household Furn.; 2519 Household Furn., N.E.C.; 2521 Wood Office Furniture; 2522 Office Furn., Exc. Wood; 2531 Public Building & Related Furniture; 2541 Wood Partitions & Fixtures; 2591 Dra.hdware/Blinds/Shade; 2599 Furniture/Fixtures, N.E.C. Paper printing and allied products (N=6) 2621 Paper Mills; 2631 Paperboard Mills; 2652 Setup Paperboard Boxes; 2711 Newspapers; 2741 Misc.Publishing; 2761 Manifold Business Forms Chemicals, petroleum and rubber industries (N=23) 2812 Alkalies & Chlorine; 2813 Industrial Gases; 2816 Inorganic Pigments; 2819 Industrial Inorganic Chemicals, N.E.C.; 2822 Synthetic Rubber; 2833 Medicinals & Botanicals; 2841 Soap/Other Detergents; 2843 Surface Active Agents; 2844 Toilet Preparations; 2851 Paints & Allied Products; 2861 Gum & Wood Chemicals; 2865 Cyc.Crudes/Intermediates; 2869 Ind. Org. Chem., N.E.C.; 2873 Nitrogenous Fertilizers; 2874 Phosphatic Fertilizers; 2875 Fertilizers, Mixing Only; 2879 Agri. Chemicals, N.E.C.; 2891 Adhesives & Sealants; 2892 Explosives; 2893 Printing Ink; 2992 Lubricating Oils & Greases; 3011 Tires & Inner Tubes; 3021 Rubber/Plastic Footwear Leather products (N=9) 3111 Leather Tann./Finishing; 3131 Ftwear Cut Stock/Find.; 3142 House Slippers; 3143 Men’s Ftwear, Ex. Athl.; 3144 Wom. Ftwear, Ex.Athl.; 3149 Ftwear, Ex.Rubber, N.E.C.; 3151 Leather Gloves/Mittens; 3161 Luggage; 3199 Leather Goods, N.E.C. Stone clay glass & concrete products (N=18) 3221 Glass Containers; 3229 Press./Blown Glass, N.E.C.; 3231 Prod. of Purchased Glass; 3241 Cement, Hydraulic; 3251 Brick/Structural Clay Tile; 3253 Ceramic Wall/Floor Tile; 3255 Clay Refractories; 3261 Vitr. Plumbing Fixtures; 3262 Vitr. China Food Utensils; 3264 Porcelain Elect. Supplies; 3271 Concrete Brick & Block; 3272 Concrete Products, N.E.C.; 3273 Ready-Mixed Concrete; 3274 Lime; 3275 Gypsum Products; 3281 Cut Stone & Stone Products; 3291 Abrasive Products; 3295 Min., Ground or Treated Primary metal industries (N=15) 3312 Blast Furnaces/Steel Mills; 3313 Electrometallurgical Prod.; 3315 Steel Wire/Related Prod.; 3316 Cold Finishing of Steel Shapes; 3317 Steel Pipe & Tubes; 3321 Gray Iron Foundries; 3324 Steel Inves. Foundries; 3325 Steel Foundries, N.E.C.; 3341 Secondary Nonferrous Metals; 3351 Copper Rolling/Drawing; 3353 Alum.

J Ind Compet Trade

Sheet, Plate/Foil; 3354 Aluminum Extruded Prod.; 3355 Alum. Roll./Draw., N.E.C.; 3356 Nonfer.Roll./Draw., N.E.C.; 3399 Primary Met. Prod.,N.E.C. Fabricated metal products (N=22) 3411 Metal Cans; 3412 Met. Barrels, Drums/Pails; 3421 Cutlery; 3425 Hand Saws/Saw Blades; 3429 Hardware, N.E.C.; 3431 Metal Sanitary Ware; 3432 Plum. Fit./Brass Goods; 3442 Metal Doors, Sash/Trim; 3443 Fab.Plate Work/Boi.Shops; 3444 Sheet Metal Work; 3446 Architectural Metal Work; 3448 Prefab. Metal Buildings; 3452 Bolt/Nuts/Rivets/Washers; 3462 Iron & Steel Forgings; 3465 Automotive Stampings; 3466 Crowns & Closures; 3471 Plating & Polishing; 3479 Metal Coating & Allied Services; 3489 Ordnance & Accessories, N.E.C.; 3495 Wire Springs; 3496 Misc. Fab. Wire Prod.; 3497 Metal Foil & Leaf Industrial & commercial machinery (N=19) 3511 Turb./Turbine Gen. Sets; 3519 Inter. Comb. Eng., N.E.C.; 3524 Lawn/Garden Equipment; 3532 Mining Machinery; 3533 Oil Field Machinery; 3534 Elev./Moving Stairways; 3535 Conv./Conveying Equip.; 3552 Textile Machinery; 3554 Paper Ind. Machinery; 3562 Ball & Roller Bearings; 3564 Blowers & Fans; 3566 Sp.Chang./Driver/Gears; 3567 Ind. Furnaces & Ovens; 3568 Pow. Trans. Equip., N.E.C.; 3581 Auto. Merch. Machines; 3585 Refrigeration & Heating Equipment; 3586 Measuring & Dispensing Pumps; 3589 Service Ind. Mach., N.E.C.; 3592 Carb./Pist./Rings/Valves Electronic & other electrical equipment (N=14) 3621 Motors & Generators; 3629 Elect. Ind. App., N.E.C.; 3632 Household Ref./Freezers; 3634 Elec. Housewares/Fans; 3639 Household Appl., N.E.C.; 3643 Current-Carrying Wir.Dev.; 3645 Res. Lighting Fixtures; 3646 Com.-Lighting Fixtures; 3648 Lighting Equip., N.E.C.; 3651 Hhold Aud./Vid. Equip.; 3674 Semicond./Related Dev.; 3675 Electronic Capacitors; 3677 Elec. Coils/Transformers; 3694 Engine Electrical Equip. Transportation equipment (N=10) 3713 Truck & Bus Bodies; 3714 Mot. Veh. Parts/Access.; 3721 Aircraft; 3724 Aircraft Eng./Eng. Parts; 3728 Aircraft Par./Equip., N.E.C.; 3732 Boat Building/Repairing; 3743 Railroad Equipment; 3751 Motcycles/Bicycles/Par.; 3761 Guided Missiles & Space Vehicles; 3792 Travel Trailer & Campers Measuring & analyzing instruments (N=8) 3822 Environmental Controls; 3823 Proc. Control Inst.; 3824 Fluid Met./Counting Dev.; 3825 Inst. to Measure Elec.; 3841 Surgical & Medical Instruments; 3842 Surgical Appl./Supplies; 3851 Opthalmic Goods; 3873 Watches/Clocks Miscellaneous manufacturing (N=10) 3911 Jewelry, Precious Metal; 3914 Silverware/Plated Ware; 3915 Jewelers’ Mat./Lap.Work; 3931 Musical Instruments; 3944 Games/Toys Child.s Veh.; 3951 Pens/Mechanical Pencils; 3952 Lead Pencils/Art Goods; 3955 Carb. Pap./Inked Ribbons; 3991 Brooms & Brushes; 3993 Signs/Ad. Displays Note: Four-digit 1987 SIC codes Indicated.

References Azzam AM (1997) Measuring market power and cost-efficiency effects of industrial concentration. J Ind Econ 45:377–386

J Ind Compet Trade Barstelman EJ, Gray WB (1996) NBER-CES manufacturing productivity database. National Bureau of Economic Research, Technical Working Paper 205 Berger AN, Hannan TH (1998) The efficiency cost of market power in the banking industry: a test of the quiet life and related hypotheses. Rev Econ Stat 80:454–465 Bian L, McFetridge DG (2000) The efficiencies defense in merger cases: implications of alternative standards. Can J Econ 33:297–318 Blair RD, Harrison JL (1999) Antitrust: on the cutting edge. Antitrust Bull 44:1–3 Blair RD, Haynes JS (2011) The efficiencies defense in the 2010 horizontal merger guidelines. Rev Ind Organ 39:57–68 Brouwer MT (2008) Horizontal mergers and efficiencies: theory and antitrust practice. Eur Rev Law Econ 26:11–26 Carlton DW (2010) Revising the horizontal merger guidelines. J Compet Law Econ 6:619–652 Chow GC, Lin AI (1971) Best linear unbiased interpolation, distribution, and extrapolation of time series related series. Rev Econ Stat 53:372–375 Clarke R, Davies SW (1982) Market structure and price–cost margins. Economica 49:277–282 Cowling K, Waterson M (1976) Price–cost margins and market structure. Economica 43:267–274 Dickson V, He J (1997) Optimal concentration and deadweight losses in Canadian manufacturing. Rev Ind Organ 12:719–732 Dickson V, Yu W (1989) Welfare losses in Canadian manufacturing under alternative oligopoly regimes. Int J Ind Organ 7:257–467 European Council Regulation. Guidelines on the Assessment of Horizontal Merers under the Council Regulation on the Control of Concentration between Undertakings. Official Journal of the European Union 2004/C 31/03 Feltovich N (2001) Mergers, welfare, and concentration: results from a model of stackelberg-cournot oligopoly. Atl Econ J 29:378–392 Focarelli D, Panetta F (2003) Are mergers beneficial to consumers? evidence from the market for bank deposits. Am Econ Rev 93:1152–1172 Golan A, Judge G, Perloff JM (1996) Estimating the size distribution of firms using government summary statistics. J Ind Econ 64:69–80 Harris F (1988) Testable competing hypotheses from structure-performance theory: efficient structure versus market power. J Ind Econ 34:267–279 Holmes TJ, Stevens JJ (2005) Does market size matter for the pattern of trade? J Int Econ 65:489–505 Lopez RA, Lopez E (2003) The impact of imports on price–cost margins: an empirical illustration. Empir Econ 28:403–416 Lopez RA, Azzam AM, Lirón-España C (2002) Market power and/or efficiency: a structural approach. Rev Ind Organ 20:115–126 Marvel HP (1980) Foreign trade and domestic competition. Econ Inq 18:103–122 Melitz MJ, Ottaviano GIP (2008) Market size, trade and productivity. Rev Econ Stud 75:295–316 Pagoulatos E, Sorensen R (1986) What determines the price elasticity of industry demand? Int J Ind Organ 4:237–250 Pugel TA (1980) Foreign trade and U.S. market performance. J Ind Econ 29:119–129 Salinger MA (2011) The 2010 revised merger guidelines and modern industrial organization. Rev Ind Organ 39:159–168 Stahlhammer NO (1991) Domestic market power and foreign trade. Int J Ind Organ 9:407–424 U.S. Department of Justice and the Federal Trade Commission (2010) Horizontal merger guidelines Weder R (2003) Comparative home-market advantage: an empirical analysis of British and American exports. Rev World Econ/Weltwirtschaftliches Arch 139:220–247 Whitley J (2003) The gains and losses from agricultural concentration: a critical survey of the literature. J Agric Food Ind Organ 1(article 6):18 Williamson OE (1968) Economies as an antitrust defense: the welfare tradeoffs. Am Econ Rev 58:18–36