Efficiency vs. market power in retailing: Analysis of supermarket chains

ARTICLE IN PRESS Journal of Retailing and Consumer Services 16 (2009) 61–67

Contents lists available at ScienceDirect

Journal of Retailing and Consumer Services journal homepage: www.elsevier.com/locate/jretconser

Efficiency vs. market power in retailing: Analysis of supermarket chains Ricardo Sellers-Rubio , Francisco J. Ma´s-Ruiz ´micas y Empresariales, Universidad de Alicante, Ap. Correos 99, Alicante 03080, Spain Departamento de Economı´a Financiera, Contabilidad y Marketing, Facultad de CC. Econo

a r t i c l e in fo

Keywords: Retailing Efficiency Market power Supermarket chains

abstract The objective of this article is to analyze the relationship between market structure and performance in the Spanish retail sector. As a new contribution, this paper considers a direct measurement of efficiency, which allows us to test different hypotheses explaining profitability in the generic framework of the theories of market power vs. efficiency. The results of the empirical application on a sample of 147 supermarket chains operating in Spain find that the competitive situation that best characterizes this sector is that of the modified efficient structure. Although efficiency does contribute positively to explaining differences in profitability, market share, which would capture the effect of market power, also has a positive effect. & 2008 Elsevier Ltd. All rights reserved.

1. Introduction In the field of industrial organization the analysis of the relationship between market structure and profitability has given rise to abundant literature of both theoretical and empirical natures. Generically, two alternative theories (market power and efficiency) have been put forward to explain the relationship between market structure (market concentration and market share) and performance. On the one hand, the traditional hypothesis of market power (Bain, 1951) posits that market concentration favors collusion. High market concentration and/or market share are/is associated with less favorable prices for consumers, thus generating greater profits for companies. On the other hand, the efficient structure hypothesis (Demsetz, 1973) posits that concentration of the market is the result of the greater efficiency of some firms, which consequently gain in market share and are more profitable. Traditionally, the most usual way of testing both hypotheses has been to introduce concentration and market share as explanatory variables of profitability, on the assumption that market share will reflect the effect of efficiency. However, market share may also be a manifestation of the residual influence resulting from market power (i.e. market power is not only obtained by collusion as firms may have market power when they enjoy high market shares) or other factors unrelated to efficiency (Maudos, 1998). Thus, the main implication of the analysis of the relationship between market structure and profitability is the need to individually identify each effect (market power and efficiency) through an efficiency control, as otherwise there could

 Corresponding author. Tel.: +34 965 90 39 67.

E-mail address: [email protected] (R. Sellers-Rubio). 0969-6989/$ - see front matter & 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jretconser.2008.10.001

be confusion between the respective impacts of market power and efficiency (Akhavein et al., 1997; Timme and Yang, 1991; Berger, 1995; Maudos, 1998). In this line, the present paper offers new evidences on the hypotheses that explain the relationship between market structure and profitability in the Spanish grocery retail sector. To reach this goal, we develop and test various alternative hypotheses explaining the relationship between market structure and profitability. The main contribution of this paper is the use of a direct measurement of efficiency to analyze this issue in retailing. Thus, we first apply a stochastic frontier production function to estimate efficiency and, secondly, we apply various regression analyses to examine the influence of efficiency and market structure (market share and concentration) on profitability. This analysis will allow us to evaluate the existing competitive situation of the Spanish retail sector. The empirical application is made with data from a sample of 147 supermarket chains operating in Spain in 2004. To reach the proposed objectives, this study is organized into the following sections. In the second we propose and argue various hypotheses related to the competitive situation of the retail sector in Spain. In the third we present the proposed methodology along with the sample and variables used. The fourth section presents and discusses the results obtained and the final section summarizes the main conclusions of the study.

2. The influence of market share, concentration and efficiency on profitability: hypothesis of efficiency vs. market power In the field of industrial organization two alternative theories have been put forward to explain the relationship between market structure (market concentration and market share) and performance. On the one hand, the traditional hypothesis of market

ARTICLE IN PRESS 62

´s-Ruiz / Journal of Retailing and Consumer Services 16 (2009) 61–67 R. Sellers-Rubio, F.J. Ma

power (Bain, 1951) proposes that high market concentration and/or market share are/is associated with less favorable prices for consumers, which will in turn generate higher profits for producers. Companies operating in concentrated markets can adopt collusive behavior, charging higher prices for their products. In fact, in a concentrated market, firms have incentives to cooperate as opposed to competing because if they are able to coordinate their actions and there are no potential entrants into the market, they can behave monopolistically and maximize the joint profits of the industry. On the other hand, the efficient structure hypothesis (Demsetz, 1973; Peltzman, 1977) proposes that concentration and/or market share are/is positively correlated with company efficiency, so that the most efficient companies grow more and obtain dominant market shares. Under this hypothesis, high concentration and market share are associated with more favorable prices for consumers if some of the savings made through efficiency are passed on to consumers (possibly as part of the process of reaching dominant market shares), and the greater efficiency of companies in more concentrated markets and with higher market shares would also produce higher profits. In summary, this hypothesis holds that efficiency explains both profitability and concentration and market share, meaning that the positive association between profitability and concentration is due to the greater efficiency of large companies and not to the exercise of monopolistic power in the industry. Empirically, the most usual way of testing both hypothesis has been to introduce concentration and market share as explanatory variables of profitability, on the assumption that market share will reflect the effect of efficiency. However, Akhavein et al. (1997) indicate that there is a debate over whether market share represents the exercise of market power for differentiated products (e.g. Rhoades, 1985; Shepherd, 1986) or the efficiency of the company, which is not included in the model (e.g. Smirlock et al., 1984; Smirlock, 1985). Recent analyses (Berger, 1995; Berger and Hannan, 1997; Maudos, 1998) attempt to resolve this problem by adding direct measures of efficiency, and find that concentration and market share have a reduced impact on firm performance after controlling for efficiency. In the case of retailing, Dobson and Waterson (1996) consider that perfect competition is not the norm in most situations and that, conversely, the presence of market power (at least to a limited extent) is more common. In fact, Lamm (1981), Marion et al. (1979), Cotterill (1986, 1999) and Aalto-Seta¨la¨ (2002) find positive relationships between market structure (concentration) and prices. These findings could support the market power hypothesis, although we have found no studies analyzing the impact of market structure (market concentration and/or market share) on profitability. In the Spanish case, one of the main characteristics of the grocery retailing sector, as with the rest of Europe, is the presence of a small group of companies with high market shares, along with an atomized market structure with a large number of agents, which creates a situation in between the conditions established in the models of perfect competition and monopoly. Only two papers (Cruz-Roche et al., 2003; Yagu¨e, 1995) analyze the relationship between market structure and prices in the Spanish market: CruzRoche et al. (2003) find a positive relationship between market structure (concentration) and prices, which supports the market power hypothesis; whereas Yagu¨e (1995) detects a negative relationship between market concentration and prices, which supports the efficient structure hypothesis. Yagu¨e (1995) considers that the market structure generated by an increase in market concentration can have repercussions to both ends of the distribution channel. These repercussions can be favorable or unfavorable for consumers according to whether or not cost

reductions in the production of distribution services are passed on to consumers. In this sense, price increases derived from market power could be offset by potential price reductions coming from efficiency improvements (Dobson and Waterson, 1996). In an attempt to analyze the competitive structure that best describes the Spanish retail sector (collusion vs. efficient structure) we consider a direct measurement of productive efficiency to analyze the impact that market structure (market concentration and market share) and efficiency exert on profitability. This solution avoids the problem of having to use market share as a proxy variable of efficiency as several authors (Timme and Yang, 1991; Berger, 1995; Maudos, 1998) agree that the main implication is the need to individually identify each effect (market power and efficiency) by controlling for efficiency. Along this line, recent developments (e.g. Maudos, 1998) distinguish between these effects, using the following hypotheses, which have not been tested in grocery retailing. Firstly, the ‘‘pure efficient structure’’ hypothesis (Demsetz, 1973) assumes that more efficient companies have lower costs and, therefore, make higher profits, thus gaining market share and increasing concentration. Given that efficiency leads to higher market share and concentration, this hypothesis assumes that concentration should not show a significant relationship with company results. Consequently, we propose that: H1. The main determinant of the results of a supermarket chain is efficiency. Note that the efficient structure hypothesis requires, moreover, that efficiency leads to higher market share and concentration. In other words, a necessary additional condition of this hypothesis is that efficiency is an explanatory variable of market share and concentration, and that they are positively correlated. Secondly, the ‘‘modified efficient structure’’ hypothesis (Shepherd, 1986) assumes that variations in company results are explained, fundamentally, by differences in efficiency and, residually, by market share, a consequence of factors such as differentiation, product quality and market power (larger companies can offer a greater range of products with better quality and at more attractive prices, which leads to higher profits). In this sense, as with the pure efficiency hypothesis, the modified efficient structure hypothesis holds that concentration does not directly affect company results, as concentration is explained by greater efficiency, which in the end leads to larger market share. Put another way, concentration should not affect company results once we have controlled for the direct effect of efficiency and the residual effect of market share. In virtue of the above, we propose that: H2. The main determinants of the results of a supermarket chain are market share and efficiency. Note that this hypothesis also requires that efficiency leads to higher market concentration. This means that a necessary additional condition of this hypothesis is that efficiency is an explanatory variable of market concentration, and that they are positively correlated. Thirdly, the ‘‘pure market power’’ hypothesis (Bain, 1951) presumes that the main determinant of company results is market concentration, in such a way that higher concentration leads to better company results. Accordingly, we propose that: H3. The main determinant of the results of a supermarket chain is market concentration. The ‘‘relative market power’’ hypothesis (Maudos, 1998) assumes, unlike the pure market power hypothesis, that the

ARTICLE IN PRESS ´s-Ruiz / Journal of Retailing and Consumer Services 16 (2009) 61–67 R. Sellers-Rubio, F.J. Ma

benefits of larger size can exist independently of market concentration, meaning that companies with higher market shares (a consequence of larger size) will have better results. In other words, H4. The main determinant of the results of a supermarket chain is market share. Finally, the hybrid hypothesis of ‘‘market power-efficiency’’ (Schmalensee, 1987) holds that efficiency determines results and that concentration favors collusion, but that the influence of market share is residual, i.e. H5. The main determinants of the results of a supermarket chain are market concentration and technical efficiency.

3. Research design In this section we present the methodology used to estimate the influence of efficiency and market structure (concentration and market share) on profitability. Additionally, we describe the sample and the variables used. 3.1. Methodology The methodology proposed to reach the objectives of this study comprises, firstly, the estimation of the efficiency of the supermarket chains in the Spanish grocery retailing sector in 2004 and, secondly, an analysis of the influence of efficiency, market share and concentration on the profitability of these supermarket chains. To estimate efficiency we use a model based on a parametric stochastic frontier, which considers the existence of certain random factors out with the control of the management units (random disturbance, measurement errors, etc.), which could affect the results obtained for each unit. This implies dividing the deviation from the efficient frontier into two components: a first component that captures the relative inefficiency of the observations with regard to the efficient frontier, and a second component that reflects the deviations from the frontier produced by external random effects out with the control of each unit and by observation measurement errors. The stochastic production frontier is estimated though the following equation (Battese and Coelli, 1995): Y i ¼ f ðX i ; bÞeðvi ui Þ

(1)

where Yi is the output of chain i, f (  ) represents production technology, Xi is the vector of inputs of chain i, b is the vector of unknown parameters to be estimated. Deviations from the production function are captured in the two error terms: (i) vi are i.i.d.N(0, sv2) and represent random errors. The v-error accounts for measurement error in outputs and the effects of misspecification in the production technology; and (ii) ui are nonnegative random variables. The u-error is associated with technical inefficiency of production. ui are assumed to be independently distributed, such that ui is the truncation (at zero) of the normal distribution with mean Zi, and variance su2. The efficiency of chain i is defined as the ratio of the output obtained to the maximum output value that chain could have obtained: EFi ¼ exp(ui), so that 0pEFip1. Efficiency is equal to 1 only if the supermarket chain is efficient, otherwise the chain analyzed is inefficient. The lower this index, the more inefficient the chain evaluated. The estimation of EFi ¼ exp(ui) entails consideration of the effects of inefficiency ui, which is unobservable. Even if the true value of the parameter vector b in the stochastic frontier model was known, only the difference

63

ei ¼ vi–ui could be observed. Therefore, the best predictor for ui is the conditional expectation of ui, given the value of vi–ui: E(exp(uit)jeit) (Battese and Coelli, 1995). 1  FðsA þ geit =sA Þ expðgeit þ s2A =2Þ (2) 1  Fðgeit =sA Þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where sA ¼ ð1  gÞs2u ; eit ¼ lnðyit Þ  X it b and F(  ) is the density function of a standard normal random variable. Secondly, in order to analyze the competitive situation of the grocery retailing sector, we analyze the influence of the efficiency estimated, of market share and of market concentration on the results (profitability) of the supermarket chains. Given that the variables market share (MS) and market concentration (CR) are endogenous in the efficient structure hypothesis, the tests of the different hypotheses are made through the estimation of a system of simultaneous equations in which MS and CR depend on efficiency (Maudos, 1998; Berger, 1995):

Eðexpðuit Þjeit Þ ¼

Resi ¼ a0 þ a1 MSi þ a2 CRi þ a3 EF i þ Aj Mi1 þ i1

(3a)

MSi ¼ b0 þ b1 EF i þ i2

(3b)

CRi ¼ c0 þ c1 EF i þ i3

(3c)

where Resi is the indicator of the results (profitability) of supermarket chain i, MSi is the market share of chain i, CRi is the market concentration of chain i, and EFi the efficiency (previously estimated) of chain i. Mij (j ¼ 1,2) are specific control variables of each company and of the markets in which they operate and which could affect profitability. a0, a1, a2, a3 and Aj (j ¼ 1,2) are parameters to be estimated. The estimation of this system of simultaneous equations is made for maximum likelihood. The following estimated values for parameters a1, a2 and a3 allow us to accept/reject the five proposed hypotheses:

 a1 ¼ 0, a2 ¼ 0 and a340 ) Hypothesis of ‘‘pure efficiency’’,

   

thus accepting Hypothesis 1 and rejecting Hypotheses 2–5. Note that accepting this hypothesis also requires that b140 and c140. a140, a2 ¼ 0 and a340 ) Hypothesis of ‘‘modified efficient structure’’, thus accepting Hypothesis 2 and rejecting Hypotheses 1, 3, 4 and 5. a1 ¼ 0, a240 and a3 ¼ 0 ) Hypothesis of ‘‘pure market power’’, thus accepting Hypothesis 3 and rejecting Hypotheses 1, 2, 4 and 5. a140, a2 ¼ 0 and a3 ¼ 0 ) Hypothesis of ‘‘relative market power’’, thus accepting Hypothesis 4 and rejecting Hypotheses 1, 2, 3 and 5. a1 ¼ 0, a240 and a340 ) Hybrid hypothesis of market powerefficiency, thus accepting Hypothesis 5 and rejecting Hypotheses 1–4.

3.2. Sample, data and variables The application of the methodology proposed in the previous section is made on data from a sample of 147 supermarket chains operating in Spain in 2004, an interesting example for the objectives of this research as this commercial format has become one of the key players in the distribution of mass consumption products in almost all Spanish cities and in recent years has gained sizeable market share from traditional distribution and hypermarkets. Moreover, this subsector is characterized by an increase in the size of the companies, which has led to higher concentration. The selection of the sample used begins with the supermarket chains found in Spain in the ALIMARKET database in 2004. To

ARTICLE IN PRESS 64

´s-Ruiz / Journal of Retailing and Consumer Services 16 (2009) 61–67 R. Sellers-Rubio, F.J. Ma

ensure the homogeneity of the companies analyzed (an inherent aspect of the concept of efficiency) we exclude discount stores, as both their assortment and services offered differ from those offered in a supermarket. In this way there is an initial sample of 196 supermarket chains, of which we discard 49 companies with no available information on relevant variables. This leads to a final sample of 147 supermarket chains operating in Spain in 2004, which represents almost 70% of total supermarket sales and around 65% of the number of outlets in 2004. With regard to the selection of variables, the definition of inputs and outputs is the main problem faced by any analysis of efficiency in distribution (Alderson, 1948; Beckman and Buzzell, 1958). This has led authors such as Barros and Alves (2003) to propose that in the selection of these variables we should consider the literature review, the availability of information and the opinion of managers. Following this proposal, after the review of the literature and the available information, to estimate efficiency we consider one output and two inputs, information obtained from the Distribution Annual Report (2004) and the SABI database. The latter contains accounting information on Spanish companies with turnovers of more than 6 million Euros or with over 20 employees. Specifically, in the case of output, we have selected the monetary output ‘‘sales income’’, which considers the whole product range offered by a supermarket chain. It should be remembered that although the true value of output in distribution consists of a series of services (Betancourt and Gautschi, 1988), the commercial distribution sector is characterized by the breadth and depth of its product assortment, which hinders the collection of information on services supplied. Additionally, the scope of this study, which considers different supermarket chains, rules out the collection of this type of information. With regard to inputs, we use two controllable productive factors: (i) The number of employees (Li) of the supermarket chain, which is the representative input of the labor factor (Bucklin, 1978; Ingene, 1982; Pilling et al., 1995; Yoo et al., 1997; Thomas et al., 1998). (ii) The number of outlets (Ki) of the supermarket chain. This input represents the capital factor and acts as a proxy of the coverage that a chain has in the market. To test the proposed hypotheses through the system of simultaneous equations, we use the following variables. As a dependent variable, and to ensure robust results, we use two different measurements of company profitability: returns over investment (ROI) and returns over assets (ROA). ROI is calculated as the quotient between the net profits (after tax) of the supermarket chain and its investments (measured by fixed assets), while ROA is calculated as the quotient between the net profits of the supermarket chain and its assets (capital plus reserves). As independent variables we use the following: (i) The market share of a company (MSi), obtained as the quotient between the sales income of the supermarket chain and total sales income for the supermarket and hypermarket sector. We consider the total sales income of the hypermarket format in the calculation of the market share of a supermarket chain because one of the characteristics of the Spanish market is the intensification of intertype competition between the different commercial formats (Gime´nez et al., 2002). (ii) Market concentration (CRi), measured through the Herfindahl index (Aalto-Seta¨la¨, 2002), and acting as a proxy of the market power of the company (Berger et al., 1997). In line with Maudos (1998), its calculation entails estimation, for each province, of a concentration index such as the sum of the squares of the market shares of each supermarket chain operating in the province. Next, each supermarket chain is assigned a concentration index equivalent to the weighted average of these indices according to the geographical distribution of its sales

Table 1 Descriptive statistics of the variables used.

Sales income (  103h) Employees Outlets ROI (%) ROA (%) Market share (MS) Market concentration (CR) Assets (  103h) Market growth rate (%)

Mean

SD

Max

Min

2593.8 2419.1 99.9 10.297 6.452 0.012 0.002 158,564.9 3.206

6110.1 4701.7 139.5 3.460 2.782 0.023 0.010 398,665.3 1.021

46,204.8 31,694 681 15.371 11.904 0.141 0.071 1,775,997.1 7.731

57.2 46 6 5.349 5.219 0.000 0.000 433.73 0.019

income. As indicated above, given the strong intertype competition between the supermarket and hypermarket formats, as a relevant market for the calculation of provincial market share we use the sales income of both company types. The calculation is made on the 500 largest distribution companies and assumes a provincial sales income distribution proportional to the number of outlets that each company has in each province (Maudos, 1998). (iii) Efficiency (EFi) estimated with the model of Battese and Coelli (1995). (iv) The assets of each supermarket chain (Assi). This control variable is used as a proxy of the size of the chain and collects the influence of factors related to the scale of production (economies of scale, for example). (v) The growth rate of the market (GMi), calculated as the percentage growth of the reference market of each supermarket chain with regard to the previous year. This control variable is constructed, as with market share and the Herfindhal index, by considering the total sales income of the supermarket and hypermarket sector and using the provincial market as a reference market. Finally, each supermarket chain is assigned a market growth index equivalent to the weighted average of these provincial indices according to the geographical distribution of its sales income. The descriptive statistics of the variables used are presented in Table 1.

4. Results and managerial implications 4.1. Results First, we estimate efficiency using panel data from a sample of 147 supermarket chains operating in Spain in 2004. We consider a flexible translog production function: ln Y i ¼ b0 þ b1 ln T i þ b2 ln Li þ b3 ðln K i Þ2 þ b4 ðln K i Þ2 þ b5 ðln K i Þðln Li Þ þ vi  ui

(4)

where Yi represents the output of chain I; Ki and Lt represent the inputs of chain I; vi are i.i.d.N(0,sv2), representing random errors; and ui are non-negative random variables i.i.d.N(Zi,su2). The results of the estimation of the proposed model, for maximum likelihood (ML), are presented in Table 2. We test various null hypotheses relative to the production function parameters through the test: l ¼ 2[L(H0)–L(H1)], where L(H0) is the value of the likelihood function in the restricted model according to null hypothesis H0; L(H1) is the value of the likelihood function in the model estimated under the alternative hypothesis H1. This statistic follows a ji-squared distribution with degrees of freedom equal to the difference between the parameters estimated under the null and the alternative hypotheses. The first test considers the global significance of the model. The value of the test of the likelihood ratio (l ¼ 144.216; critical value ¼ 11.071) rejects the null hypothesis that all the parameters (except the constant) are zero (H0: b1 ¼ b2 ¼ b3 ¼ b4 ¼ b5 ¼ 0).

ARTICLE IN PRESS ´s-Ruiz / Journal of Retailing and Consumer Services 16 (2009) 61–67 R. Sellers-Rubio, F.J. Ma

Additionally, we reject the null hypothesis that the production function that best characterizes the available technology is a Cobb–Douglas function (H0 0:b3 ¼ b4 ¼ b5 ¼ 0) (l ¼ 29.391; critical value ¼ 7.815). Finally, we reject the null hypothesis that g is zero (H00 0: g ¼ 0) (l ¼ 126.409; critical value ¼ 8.76), which shows a high relative importance for the effect of inefficiency on the specification error. Finally, and with regard to the estimation of efficiency, we find that the average efficiency for the sample analyzed was 0.758 (SD ¼ 0.231) in 2004, which indicates that the companies could have reached the same output level with 24.2% less resources. Secondly, and to test the competitive situation of the Spanish retail commercial distribution sector, we estimate the proposed system of simultaneous equations (3a)–(3c) by the maximum likelihood method (see Table 3). The estimation of this system of equations is made using two alternative measures of firm profitability: returns over investment, ROI (Eqs. (1)–(3) of Table 3) and returns over assets, ROA (Eqs. (4)–(6) of Table 3). As can be seen, the results are robust, independently of the returns indicator used (returns over assets, ROA, or returns over investment, ROI). The joint significance tests of the parameters estimated show that significant information is obtained at a level below 1% in all the equations. The results obtained show that market share and efficiency have a positive and statistically significant influence, independently of the profitability measure used. Additionally, market concentration is not significant in the explication of profitability,

Table 2 Estimation of efficiency. Variable

Parameter

Coefficient

SD

Constant Outlets (K) Employees (L) Outlets2 Employees2 Outlets  Employees

b0 b1 b2 b3 b4 b5

4.358 1.271 0.281 0.042 0.019 0.081 0.792 0.912 173.629

2.169 0.214 0.029 0.051 0.009 0.007 0.218 0.026

g s2 log L 2 u/(

g¼s

2 u+

2 v ),

s s

2

2 u+

2 v ).

s ¼ (s s

 Probo0.01.  Probo0.05.

65

thus rejecting the pure market power hypothesis. As shown, the fact that the coefficient that accompanies the variable market share is statistically significant when we introduce the effect of efficiency suggests, in line with Berger (1995) and Maudos (1998), that in the regressions with no consideration of a direct measure of efficiency, market share cannot be considered an adequate proxy of efficiency. In other words, given that the effect of efficiency is controlled for in the regressions, the positive sign of the market share variable indicates the presence of market power, at least to a limited extent. This result upholds the ‘‘modified efficient structure’’ hypothesis of Shepherd (1986), which assumes that variations in the results of companies are explained by differences in efficiency and, residually by market share; as a consequence of factors such as differentiation, product quality and market power (large companies can offer a wider range of better quality products at more attractive prices, which leads to higher profits). Moreover, the acceptance of this hypothesis also requires that efficiency is positively correlated with concentration. As can be seen (see Eqs. (2) and (5) of Table 3), the influence of efficiency is positive and significant at 95%. Additionally, the fact that the correlation between efficiency and market share is positive, although not significant, highlights the importance of considering these two effects separately, given that market share could be capturing other aspects (differentiation, etc.) with influence over profitability. With regard to the control variables, we can see that the market growth rate is not significant in any case. The lack of significance of this variable is explained because the period of analysis (1 year) prevents us from analyzing the effects that the economic cycle could have on the margins and profitability of the retail companies. In the case of assets, we observe that this variable is positive and significant, independently of the profitability indicator used, which supports the size increases experienced by the Spanish retailers in recent years and described by Rebollo (1999) and Cruz-Roche et al. (2003). In conclusion, the results obtained (we accept that a140, a2 ¼ 0 and a340) do not accept Hypotheses 1 (pure efficiency) and 3 (pure market power); but neither do they accept Hypothesis 4 of relative market power, nor do they accept Hypothesis 5 (hybrid market power-efficiency). However, the results obtained support Hypothesis 2, meaning that they show that the situation that best reflects the competitive structure of the supermarket format in 2004 is that of the modified efficient structure, as the main determinants of company results are market share and efficiency.

Table 3 Consequent factors of technical efficiency: hypothesis of efficiency vs. market power (standard errors in brackets). Eq.

Dep. var.

Constant

Market share (MS)

1

ROI

5.83 (2.575)

2

CR

3

MS

2.747 (1.160) 1.452 (0.232) 1.553 (0.185)

4

ROA

6.216 (1.291)

5

CR

6

MS

2.211 (2.229) 2.529 (0.931) 2.936 (0.993)

 Probo0.01.  Probo0.05.

Market concentration (CR) 3.521 (6.397)

4.094 (3.001)

Efficiency (EF)

Assets (Ass)

Market growth (GM)

R2 adjust.

6.343 (1.334) 3.654 (1.83) 1.589 (1.477)

1.251 (0.502)

0.621 (1.006)

0.435

9.549 (3.221) 2.439 (1.077) 1.631 (1.477)

1.094 (0.371)

0.231 0.039

0.447 (0.991)

0.516 0.291 0.027

ARTICLE IN PRESS 66

´s-Ruiz / Journal of Retailing and Consumer Services 16 (2009) 61–67 R. Sellers-Rubio, F.J. Ma

4.2. Managerial implications

5. Conclusions

From the above results the hypothesis of collusion in Spanish grocery retailing is clearly rejected. The evidence obtained is favorable to the modified efficient structure hypothesis, since although efficiency does contribute positively to explaining differences in profitability, market share also has a positive effect. This implies that the potential benefits associated with company size exist even when the market is not concentrated. This result should be taken into account by policy makers and managers. One of the main concerns for policy makers when analyzing mergers and/or acquisitions in retailing is their effect on consumers, especially if collusion applies. However, policy makers should bear in mind that the hypothesis of collusion is rejected by the data. Thus, large size (the consequence of mergers) allows a firm to increase its market share, which will positively affect its profitability. This large size is not necessary unfavorable for consumers if cost reductions in the production of distribution services are passed on to them. In this sense, potential price reductions can be obtained from efficiency improvements (Dobson and Waterson, 1996) derived from a large size. In the specific case of grocery retailing Cruz-Roche (1999) shows that a size increase allows for increased efficiency in distribution as it reduces costs. Moreover, returns gained from large size are used for various activities, chief among which are the following: (i) more favorable purchases and the establishment of more advantageous contracts with suppliers; (ii) realization of promotional activities; (iii) increased financial power to make essential investments, especially in technology; (iv) brand image, assurance of quality to customers; and (v) improvement and modernization of economic and administrative management. Consequently, large size allows a company to improve its profitability and increase its market share, which will ultimately positively affect its results. In fact, Cruz-Roche et al. (2003) show that the effect of scale economies, measured by the dimension of the largest companies, indicates that size can lift entry barriers that protect existing companies from potential competition. Moreover, the size of a distributor is one of the main factors generating dependence in the supplier and, therefore, power for the distributor (Aalto-Seta¨la¨, 2002). In other words, large distributors account for an important part of the sales volumes of manufacturers, who find that they cannot do without this relationship. This is especially important with regard to intermediaries with large market share. In fact, retailers find the best opportunities to increase profits in securing more advantageous purchasing prices than their competitors, but also through other discounts and services. Therefore, it is essential that retailers attain high market power, i.e. a strong participation in the retail markets in which they operate (Cruz-Roche, 1999). Further, the results obtained should be taken into account by retail firms. Retail managers should be aware of the importance that efficiency has for their own firms. The importance of efficiency in explaining the differences of profitability among firms of the sector is vital in order to guarantee their existence in the market. In recent years, growing competitiveness among retailing companies and the globalization of markets have given rise to an economic environment in which it is becoming increasingly difficult for companies to survive. Thus, efficiency is an important issue as a manager’s control and expansion strategies begin with an assessment of the relative efficiency level. Strategically, retail efficiency can be used to differentiate firms and provides the foundation to develop strategies for growth and diversification.

The objective of this study is to analyze the relationship between market structure and profitability in the Spanish grocery retail sector with data on a sample of supermarket chains operating in Spain in 2004. The main contribution of the paper relies on the use of a direct measure of efficiency to analyze the competitive situation in the Spanish supermarket sector. The results obtained indicate that the main determinants of profitability for the sample analyzed are market share and efficiency, which upholds the proposal that the competitive situation characterizing the supermarket subsector is that of a modified efficient structure. This result is valuable when putting strategies into practice. In particular, it highlights the importance of efficiency and market share as key factors to improving profitability, but it also means that the potential benefits associated with company size exist even in nonconcentrated markets. In this sense, the size of these companies appears to provide a useful reference structure in the analysis of competition in order to detect possible competitive advantages. Further, retail managers should make efforts to analyze their competitive environment from the point of view of each competitor, analyzing their own efficiency levels. Efficiency is a crucial factor to explain profitability. With regard to the limitations of the study, which call for prudent reading of the results obtained, we can highlight the following: Firstly, the analysis period is of only 1 year and is based on a single retail format, which means that the conclusions of the study should only be considered in terms of the sample used. Secondly, the impossibility of including other factors of the competitive structure, both vertical and horizontal, which could affect profitability in retail companies, gives us a model with a limited specification. Although the objective of the study is based on an analysis of the determinants of profitability in terms of market structure and efficiency, there could be problems derived from the omission of other relevant variables, which could also affect profitability. Finally, given the important implications that size increases and market concentration have on the sector, for future lines of research we suggest the distinction between the ways of realizing the concentration process, internal growth (opening new outlets) and external growth (takeovers or mergers with other companies), insofar as the greater use of mergers and takeovers supposes a greater tendency towards concentration than towards internal growth. This means that the external growth of companies could cause an acceleration of the concentration process and its possible consequences on the degree and type of competition, at the same time as allowing improvements in the efficiency of the distribution system. Conversely, if internal growth is used to create entry barriers, through the anticipation of future demand of the market, it will produce an oversizing of retail supply and, consequently, an increase to the total cost of the activity, with negative effects on the efficiency of the distribution system. Therefore, the analysis of the execution method of this process allows examination of its repercussions on the degree of efficiency of the commercial distribution system. References Aalto-Seta¨la¨, V., 2002. The effect of concentration and market power on food prices: evidence from Finland. Journal of Retailing 78 (3), 207–216. Akhavein, J.D., Berger, A.N., Humphrey, D.B., 1997. The effects of banks megamergers on efficiency and prices: evidence from the profit function. Review of Industrial Organization 12, 95–139. Alderson, W., 1948. A formula for measuring productivity in distribution. Journal of Marketing 12 (4), 442–448. Alimarket, 2004. Distribution Yearbook. Alimarket, Madrid, Spain.

ARTICLE IN PRESS ´s-Ruiz / Journal of Retailing and Consumer Services 16 (2009) 61–67 R. Sellers-Rubio, F.J. Ma

Bain, J.S., 1951. Relation of profit rate to industry concentration. Quarterly Journal of Economics 65, 293–324. Barros, C.P., Alves, C., 2003. Hypermarket retail store efficiency in Portugal. International Journal of Retail & Distribution Management 31 (11), 549–560. Battese, G.E., Coelli, T.J., 1995. A model for technical inefficiency effects in a stochastic frontier production function for panel data. Empirical Economics 20, 325–332. Beckman, T.N., Buzzell, R.D., 1958. Productivity: facts and fiction. Business Horizon 1, 24–38. Berger, A.N., 1995. The profit-relationship in banking—tests of market power and efficient structure hypotheses. Journal of Money Credit and Banking 27 (2), 405–431. Berger, A.N., Hannan, T.H., 1997. Using measures of firm efficiency to distinguish among alternative explanations of the structure–performance relationship. Managerial Finance 23, 6–31. Berger, A.N., Leusner, J.H., Mingo, J.J., 1997. The efficiency of bank branches. Journal of Monetary Economics 40, 141–162. Betancourt, R.R., Gautschi, D.A., 1988. The economics of retail firms. Managerial and Decision Economics 9, 133–144. Bucklin, L.P., 1978. Productivity in Marketing. A.M.A., Chicago. Cotterill, R., 1986. Market power in the retail food industry. Review of Economics and Statistics 68 (August), 379–386. Cotterill, R., 1999. Market power and the Demsetz quality critique: an evaluation for food retailing. Agribusiness 15 (1), 101–118. Cruz-Roche, I., 1999. Los canales de distribucio´n de productos de gran consumo. Concentracio´n y competencia. Pira´mide, Madrid, Spain. Cruz-Roche, I., Rebollo, A., Yagu¨e, M.J., 2003. Concentracio´n y competencia en los ˜ ola 96, canales de distribucio´n alimenticios. Papeles de Economı´a Espan 112–133. Demsetz, H., 1973. Industry structure, market rivalry and public policy. Journal of Law and Economics 16, 1–9. Dobson, P.W., Waterson, M., 1996. Vertical restraints and competition policy. Research Paper 12, Office of Fair Trading, London. Gime´nez, M.L., Pe´rez, J.A., Sa´nchez, M., 2002. Intensificacio´n de la competencia intertipo entre supermercados e hipermercados. Distribucio´n y Consumo 64, 5–17.

67

Ingene, C.A., 1982. Labor productivity in retailing. Journal of Marketing 46 (4), 75–90. Lamm, M.R., 1981. Prices and concentration in the food retailing industry. Journal of Industrial Economics 30 (September), 67–78. Marion, B., Mueller, W., Cotterill, R., Geithman, F., Schmelzer, J., 1979. The food retailing industry. Praeger, New York. Maudos, J., 1998. Market structure and performance in Spanish banking using a direct measure of efficiency. Applied Financial Economics 8, 191–200. Peltzman, S., 1977. The gains and losses from industrial concentration. Journal of Law and Economics 20, 229–263. Pilling, B.K., Henson, S.W., Yoo, B., 1995. Competition among franchises, companyowned units and independent operations: a population ecology application. Journal of Marketing Channels 4 (1), 177–195. ˜ a. Rebollo, A., 1999. Concentracio´n en el sector de distribucio´n comercial en Espan Distribucio´n y Consumo 47, 29–43. Rhoades, S.A., 1985. Market share as a source of market power: implications and some evidences. Journal of Economics and Business 37, 343–363. Schmalensee, R., 1987. Collusion versus differential efficiency: testing alternative hypothesis. The Journal of Industrial Economics 35, 399–425. Shepherd, W.G., 1986. Tobin’s q and the structure–performance relationship: comments. American Economic Review 76, 1205–1210. Smirlock, M., 1985. Evidence of the (non)relationship between concentration and profitability in banking. Journal of Money, Credit and Banking 17, 69–83. Smirlock, M., Gilligan, T., Marshall, W., 1984. Tobin’s q and the structure–performance relationship. American Economic Review 74, 1050–1060. Thomas, R.R., Barr, R.S., Cron, W.L., Slocum Jr., J.W., 1998. A process for evaluating retail store efficiency: a restricted DEA approach. International Journal of Research in Marketing 15 (5), 487–503. Timme, S.G., Yang, W.K., 1991. On the use of a direct measure of efficiency in testing structure–performance relationships. Working Paper, Georgia State University. Yagu¨e, M.J., 1995. Relacio´n entre concentracio´n y precios en el comercio minorista. ˜ ola 739, 59–70. Informacio´n Comercial Espan Yoo, B., Donthu, N., Pilling, B.K., 1997. Channel efficiency: franchise versus nonfranchise systems. Journal of Marketing Channels 6 (3/4), 1–15.