A+| A| A-

Concentration-Markup Relationship in Indian Manufacturing Sector

While the positive relationship between market concentration and price-cost margin or profitability is well documented in industrial organisation literature, the present paper makes an attempt to examine the concentration-markup relationship in Indian manufacturing sector in the post-liberalisation era using dynamic panel data model. It is observed that the traditional positive concentration-markup relationship does not hold in a dynamic context, when controlled for various structural aspects of the market, firms' strategies and policies of the government. In other words, industries with greater market concentration do not necessarily enjoy higher pcm in the long run, possibly due to entry of new firms, x-inefficiency of the incumbents and deceleration in industrial production.

REVIEW OF INDUSTRY AND MANAGEMENT

Concentration-Markup Relationship in Indian Manufacturing Sector

Pulak Mishra

While the positive relationship between market concentration and price-cost margin or profitability is well documented in industrial organisation literature, the present paper makes an attempt to examine the concentration-markup relationship in Indian manufacturing sector in the post-liberalisation era using dynamic panel data model. It is observed that the traditional positive concentration-markup relationship does not hold in a dynamic context, when controlled for various structural aspects of the market, firms’ strategies and policies of the government. In other words, industries with greater market concentration do not necessarily enjoy higher PCM in the long run, possibly due to entry of new firms, X-inefficiency of the incumbents and deceleration in industrial production.

The author is grateful to Rakesh Basant for having useful discussions at the initial stage of writing this paper. The comments and suggestions received from the anonymous referee and Bhagirath Behera were of immense help in revising the paper. Thanks are also due to Subhendra Nath Saha for extending immense help in sourcing and compiling the dataset. Usual disclaimers apply.

Pulak Mishra (pmishra@hss.iitkgp.ernet.in) is at the Indian Institute of Technology, Kharagpur.

Economic & Political Weekly

EPW
september 27, 2008

T
he positive relationship between market concentration and price-cost margin (PCM) or profitability is well documented in industrial organisation literature.1 Indeed, most of the major theories of oligopoly imply a positive relationship between concentration and margin, though the strength of the relationship differs depending on the conduct by the firms.2 This positive relationship between concentration and margin is based on the basic proposition that a high degree of market concentration may cause monopolistic or oligopolistic behaviour leading to higher monopoly profit and lesser competition.3 Alternatively, it can be said that a higher PCM, other things remaining the same, signals enhanced market concentration.

This “market power explanation” of the positive concentrationmargin relationship is one sided. In fact, there might be cases where higher profits are made by large firms not due to their market power but due to their superior efficiency [McGee 1971; Demsetz 1973]. This means that market concentration is not a necessary condition for larger PCM. Not only that, as Hay and Morris (1991) pointed out, even if concentration is necessary for higher profitability, it may not be sufficient. This is specially so if there are a few or no barriers to entry. In the absence of effective entry barriers, excess profit margin may be wiped out by new entrants. Further, as the recent developments in the structure-conductperformance (SCP) framework show, margin may also be influenced by firms’ conduct like advertising, marketing, distribution and technology related strategies, integration of firms, etc, as well as by various other structural aspects of the market and policies of the government. This means that the concentration-profit margin relationship is dynamic in nature rather than a static as it is generally conceived. So, attempts are made in recent empirical research to analyse profitability in a dynamic context.4

While under static conditions, market concentration is likely to be positively associated with PCM, in a dynamic situation a set of diverse forces operate. Lagged profitability becomes a significant determinant of current profit margin. On the one hand, higher lagged profitability attracts new firms into the industry and thereby, creates a threat to current profit margin. On the other hand, this higher lagged profitability raises the willingness and ability of firms to spend on technology and selling. This not only creates barriers to entry of new firms but also enhances their strategic competencies. Besides, the size of the market, demand conditions in the market and policies of the government may also change in the long run. In other words, greater market power or operational efficiency may help firms record a higher margin in the short run but due to the consequent entry of new firms into

REVIEW OF INDUSTRY AND MANAGEMENT

the industry in the absence of effective entry barriers, strategic inefficiencies of the existing firms or changes in the policies of the government, the relationship between concentration and margin may not be significant in the long run.5

It is, therefore, necessary to examine the relationship between market concentration and PCM in a dynamic framework controlling for these diverse forces. The present paper is an attempt in this direction. The objective of the paper, therefore, is to analyse the inter-industry variations in the concentration-margin relationship in the Indian manufacturing sector controlling for a variety of structure, conduct and performance and policy variables. Such an effort has important policy implications as the economists and policymakers are against unchecked monopoly power that results in the misallocation of resources and undesirable transfer of resources from the public to the monopolist. The rest of the paper is divided into four sections. Section 1 specifies the functional model of the dynamic concentration-margin relationship. Section 2 discusses the database and estimation techniques used in the paper. Section 3 presents the empirical results and analyses the same. Section 4 concludes the paper.

1 Concentration-Margin Relationship

The following sections discuss the model specifications and the impact of explanatory variables.

1.1 Model Specification

The concentration-margin relationship can be best analysed with the help of the SCP paradigm – the predominant framework of analysis in industrial organisation.6 The traditional SCP paradigm was based on the early work of Edward Mason in the 1930s and was developed further by Bain (1959). The conventional SCP paradigm postulates a unidirectional relationship between market structure, conduct and performance with structure (concentration, conditions on entry, etc) influencing performance (profits, etc) via conduct (price and non-price behaviour). Successive developments in industrial organisation literature in the 1980s and 1990s make a marked deviation from this traditional unidirectional relationship. A review of this literature suggests that the variable market structure is not exogenous as it was traditionally assumed. Instead, to a large extent it is influenced by the basic conditions relating to demand and supply, such as the material inputs required, eco nomies of scale and scope, market size, price elasticity of demand and heterogeneity of consumers’ needs and preferences. Further, these basic conditions also depend on market structure and firms’ conduct.

Moreover, the causal relationship between structure, conduct and performance may not necessarily be unidirectional. Instead, dual causalities between structure and conduct, between conduct and performance and between structure and performance are very likely.7 Another important development in the modern SCP paradigm is the inclusion of public policies relating to taxes, subsidies, international trade, investment, etc, and welfare related issues [Basant and Morris 2000; Mishra 2005]. Not only may these public policies affect market

76 structure and firms’ conduct and performance, there might be a feedback effect from structure, conduct and performance to policy as well.8

Thus, market concentration is no longer an exogenous variable. Instead, an endogenous one as it is influenced by basic conditions, firms’ conduct, performance and government policies. Further, the relationship among various structural, strategic, performance related and policy-oriented variables may not be instantaneous. It is very likely that there will be a time lag in the relationship. Therefore, the concentration-markup relationship should be explored in a multidimensional and dynamic SCP-policy framework.

Given this, the ideal way of modelling the concentration and Pcm relationship is to apply a simultaneous equation approach. However, non-availability of systematic data on some important corporate strategies, such as mergers, acquisitions, alliances, etc, restricts such modelling. However, on many occasions, lags in the relationship among the variables can lead to their nonsimultaneous or sequential determination. The present paper attempts to examine the relationship between concentration and markup, taking adequate care of the potential lags to control for the problem of simultaneity in the envisaged relationships.

It should be mentioned that the use of lags in the explanatory variables is a standard treatment for the problem of endogeneity and deeper lags of these variables are likely to be better in struments. In the present paper, three-year average values of the dependent variables instead of their annual values are used to have adequate control over the problem of endogeneity. For the dependent variable also a three-year average is used instead of its annual value to account for dynamic behaviour in a greater way.9

To begin with, let us hypothesise that PCM10 of the firms in an industry depends on structure of the market (MST), firms’ conduct (FCN), pervious performance (PER) and government policy (GP), i e, PCM = f (MST, FCN, PER, GP).

Here, the lagged value of the dependent variable (PCM_1) stands for previous performance, whereas, growth rate of sales (GRS) is included to take into account the changes in demand conditions in the market. In addition, we capture the structure of markets by using four variables, viz, market concentration (CN), market size (MSZ), minimum efficient scale (MES) and capitalintensity ratio (KIR).11 For firms’ conduct, we use four variables, viz, selling strategies (SELL), in-house research and development (R&D), efforts (RDI), foreign technology purchase (FTPI) and export intensity (EXP).12

In addition, to control for technology strategies of the firms, FTPI is also expected to capture policy changes by the government with respect to foreign technology. The new industrial p olicy (NIP) of 1991 introduces liberal policy measures for foreign technology agreements and the extent differs across industries depending on their priority. Industries with automatic approval provision for foreign technology are likely to have high foreign technology purchase intensity. Similarly, export intensity (EXP) can also be seen as a variable to control for export-related policy changes made by the government. As the industry-specific p olicies or measures by government to promote exports vary,

september 27, 2008

REVIEW OF INDUSTRY AND MANAGEMENT

industries with favoured exports measures are expected to record greater export intensity. Hence, the above functional relationship can be rewritten as: PCM = f (PCM_1, GRS, CN, MES, MSZ, KIR, SELL, RDI, FTPI, EXP).

1.2 Possible Impact of Explanatory Variables

Previous Performance (PCM_1): In the present paper, lagged price-cost margin (PCM_1) is used as a measure of previous performance. On the one hand, a high level of PCM strengthens firms’ position in the industry as well as enables them to develop manufacturing, selling and technology related complementary assets. As a result, the incumbents are expected to enjoy a higher PCM in the next period. On the other hand, a high level of PCM attracts new firms into the industry. In the absence of effective entry barriers, such an entry may reduce the future PCM of firms. High current profitability can also result is X-inefficiency and hence, lower PCM in the future. The effect of lagged PCM on its current level, therefore, depends on the relative strength of these diverse forces.

Growth of Industry Sales (GRS): PCM may increase or decrease with the change in market demand. In the present paper, GRS is used as a proxy for growth of market demand. GRS is expected to influence PCM in three possible ways. First, high GRS is likely to create opportunities for existing firms to expand their business and thereby, to achieve greater efficiency through economies of large-scale operations [Kambhampati 1996]. Secondly, high GRS allows suppliers to charge higher prices and thereby, induces new players to enter the industry. This reduces the level of concentration and PCM [Ghose 1975]. Finally, high GRS may create pressure on inputs and raise input prices [Goldar and Aggarwal 2004]. The ultimate impact of GRS on PCM, therefore, depends on the relative strengths of these diverse forces.

Market Concentration (CN): The Herfindahl-Hirschman Index (HHI) is used as the measure of seller’s or market concentration. This index satisfies all the desirable properties of a concentration measure as it combines both the number and size distribution of firms in the industry. Further, by squaring market shares the HHI weights more heavily the values for large firms than for small ones. Therefore, when precise data on the market shares of very small firms are unavailable, the resulting errors are not large.13 As it is generally conceived, CN is expected to have a positive relationship with PCM.

Minimum Efficient Scale (MES): Given the size of the market and its growth, MES can affect the margin in an industry in two ways. On the one hand, when the level of MES is high in an industry and many of firms operate below this level, the average cost of operation in the industry may become higher. This may reduce the profit margin of the firms in the industry. On the other hand, such operation of the existing firms below the level, MES is expected to discourage new firms from entering into the industry in the long run. The effect of MES on PCM will therefore depend on how these two opposite forces work.

Economic & Political Weekly

EPW
september 27, 2008

Market Size (MSZ): Size of the market (MSZ) affects PCM from both demand and supply sides. On the one hand, a larger market may comprise larger number of players and therefore, result in lower PCM in the industry. On the other hand, a larger market may facilitate the firms to operate at optimal scale. This lowers average costs of operation and, thereby, raises PCM. The eventual impact of MSZ, therefore, depends on the relative strength of these two opposite forces.

Capital-Intensity Ratio (KIR): KIR serves as an absolute barrier to entry. A high KIR is likely to reflect the existence of large sunk costs that create entry barriers and thereby, give rise to monopoly profit [McDonald 1999]. Besides, capital market imperfections may lead to discrimination by offering preferential lending rates to large established firms in capital-intensive industries. This higher cost of capital makes the small firms less competitive and thereby, restricts them from entering into the industry [Basant and Saha 2005]. On the other hand, high capital intensity may result in lesser flexibility in terms of adjusting to market turbulence and hence, have a negative impact on margin. Thus, the nature of the relationship between KIR and PCM is ambiguous and it depends on the relative strength of these diverse forces.

Selling Intensity (SELL): SELL is defined as the ratio of advertising, marketing and distribution related expenditure to sales. This variable is used to capture strategies towards product differentiation and building up marketing and distribution network. Image related entry barriers, product differentiation through advertising and development of marketing and distribution related complementary assets help the incumbent firms to improve their financial performance. Therefore, industries with greater selling efforts may be expected to have higher PCM and vice versa.

Research and Development Intensity (RDI): This variable is used to control for the impact of in-house R&D efforts by the firms on their financial performance. The variable is expected to have positive impact on PCM as product development through in-house R&D strengthens and/or extends market orientation while new processes reduce cost of production.14 High RDI may also act as an entry barrier yielding higher profit margin to the incumbents.

Foreign Technology Purchase Intensity (FTPI): Like RDI, FTPI is included in the model to control for the impact of firms’ efforts towards acquiring foreign technology on their financial performance. Greater access to foreign technology not only enhances competitive edge in the market place but also helps the incumbents to create strategic entry barriers. Therefore, the industries with greater efforts of the firms towards purchasing foreign technology are likely to record higher PCM.

Exports Intensity (EXP): Export markets provide the domestic firms opportunities to achieve optimal scales of operation particularly when domestic demand constraints force them to do so. This helps the firms to reduce their costs of operations and, thereby, to raise their PCM. Hence, EXP is expected to have positive association with PCM.

REVIEW OF INDUSTRY AND MANAGEMENT

2 Data and Estimation Techniques

The above functional relationship is examined with a pooled dataset for 119 product groups for the period 1992-99 collected from the Prowess database of the Centre for Monitoring Indian Economy, Mumbai to capture variations in the variables both across the product groups and over time. This relaxes the assumption made in cross section analyses that the same structureperformance relationship prevails among all industries at one point of time. Further, this pooled dataset also provides greater degrees of freedom and therefore, raises the efficiency of the estimates by increasing the sample size. Details on measurement of the variables are provided in Appendix I (p 81).

In order to explore the relationship among the variables in the long run, we estimate dynamic panel data model using Arellano-Bond regression [Arellano and Bond 1991]. The estimated model is of the form:

m yit = α + βyi,t–1 + Σγj xj, it + uit

j=1

Here, yit is the natural logarithm of PCM in industry i in period t, is the natural logarithm of predetermined and endogenous

xj, itvariable xj in industry i in period t and α, β, and γj (for j = 1, …, m) are the parameters to be estimated. Here, we transform all v ariables into their natural logarithms to

Table 1: Summary Statistics of the Variables

control for h eteroscedasticity and scale and two-step estimator, ensuring their

Variable Number of Mean Standard Minimum Maximumeffects.15 Observations Deviation Value Value goodness of fit. However, though the

The Arellano-Bond dynamic panel PCM 816 -3.44 0.68 -10.44 -1.84 model is statistically significant value of GRS 833 -1.72 0.88 -9.12 1.07

estimation technique uncovers joint Wald-Chi2 statistic is much higher in case

CN 833 -1.74 0.76 -4.09 0.00

effects of the explanatory variables on of two-step estimator.

MES 833 0.55 0.29 -2.67 1.41

the explained variable while controlling Further, as the Sargan test statistic is

MSZ 833 0.16 0.16 -0.66 0.61

for potential bias due to endogeneity of not statistically significant, the estimated

KIR 833 -1.59 0.53 -4.11 0.40

the explanatory variables including the model does not suffer from the problem

SELL 833 -4.37 0.66 -7.37 -2.86

lagged dependent variable.16 The lagged of over-identified restrictions. However,

RDI 786 -7.85 1.51 -14.46 -3.74

dependent variable in the model accounts the statistics for Arellano-Bond test of

FTPI 773 -6.49 1.44 -13.09 -2.91 for the dynamic effects. In other words, EXP 831 -3.78 1.12 -8.57 -1.14 both auto regressive (AR) (1) and AR (2)

the above model acknowledges the dyna-These summary statistics are based on natural logarithmic value of are not statistically significant implying

the variables.

mic relationships as well as interdependencies among the variables.17 Further, in order to test for a utocorrelation and the validity of instruments we use the A rellano-Bond test for auto-covariance and Sargan test r espectively.

In this context it should be noted that there are two versions of the Arellano-Bond estimator, viz, a one-step estimator and a twostep estimator. The asymptotic standard errors of one-step estimators are unbiased and are reliable to draw inference on the individual coefficients. However, in the one-step estimator, the Sargan test over-rejects the null of over-identifying restriction in the presence of heteroscedasticity. While the robust standard errors of the one-step estimator control for heteroscedasticity, the distribution of the Sargan test statistic is not known in this case. The two-step estimator, on the other hand, yields standard errors that are asymptotically robust to both heteroscedasticity and autocorrelation. However, as it has been shown by the Monte Carlo studies, the asymptotic standard errors of the two-step estimator can be severely downward biased in small samples. So, the present paper uses both the one-step and two-step estimator. While the statistics based on the two-step estimator are used for model specification testing of the over-identifying restrictions and to test for overall significance of the model, one-step estimates with robust standard errors are used to draw inference on the regression coefficients.

Further, the present paper uses three-year average of both the dependent and explanatory variable. Such measurement of the dependent variable is, however, unlikely to reduce power of the Sargan test as the number of cross sectional units is very large as compared to the time-series units.18 Further, since the number of cross sectional unit is quite large, the Arellano-Bond test for autocorrelation remains valid despite use of three-year average of the dependent variable.19

3 Regression Results and Interpretations

Table 1 provides the summary statistics for the variables used in the model. We estimate the above model using the Arellano-Bond generalised method of moments (GMM). In order to control for potential bias due to endogeneity of the explanatory variables, we use the two-year lag value of the lagged dependent variable and one-year lag value of other explanatory variables as instruments. The regression results including the tests for autocorrelation and the validity of the instruments (Sargan test) are summarised in Table 2 (p 79). It is observed that Wald-Chi2 is statistically significant for both one-step estimator

that the model is free from the autocorrelation problem. Hence, the results of the two-step estimator show that the estimated model is not only statistically significant but is also free from the problem of over-identified restrictions and auto-covariance.

The z-statistics of the regression coefficients for the one-step estimator are based on robust standard errors and hence, are controlled for heteroscedasticity and autocorrelation. As mentioned above, these z-statistics are used for inference of the individual coefficients. It is observed that the coefficient of lagged PCM, GRS and SELL are statistically significant at the 5 per cent level of significance, whereas, that of MSZ is statistically significant at 10 per cent level of significance. Further coefficient of lagged PCM, GRS, MSZ and EXP are positive and that of SELL is negative. Coefficient of CN, MES, KIR, RDI and FTPI are, however, not statistically significant.

3.1 Interpretation of Results

From Table 2 it is clear that PCM of an industry in any period is significantly and positively influenced by its lagged value. A high level of PCM not only strengthens firms’ position in the industry,

september 27, 2008

REVIEW OF INDUSTRY AND MANAGEMENT

it also enables them to develop manufacturing capabilities and on PCM. Similarly, technology strategies in the form of in-house selling and technology related complementary assets. Further, R&D or foreign technology purchase fail to improve PCM in the these positive effects outweigh possible negative impacts of com-industry in the long run. This may be so especially when the petitive threat in the form of entry of new firms or X-inefficiencies success rate of the in-house projects is very low or the domestic of the incumbents. As a result, firms in the industries with higher firms fail to cope up effectively with the foreign technology. PCM in the current year are likely to enjoy It is important to note that examining

Table 2: Regression Results for the Arellano-Bond Model

higher PCM in the next year. market concentration and markup rela-

Dependent Independent Variable One-Step Two-Step

Positive and statistically significant Variable Estimator Estimator tionship in a dynamic context with

PCM Intercept -0.0872 -0.1072

coefficient of GRS, MSZ and EXP imply adequate control for other structural

(-1.72)# (-3.33)*

that industries with greater size of the aspects, strategic issues and policy

PCM_1 0.6552 0.6490 domestic market, higher rate of growth (7.16)* (12.24)*changes in the present paper makes

GRS 0.1299 0.1152

of market demand or larger penetration many of the findings to deviate from that

(2.94)* (3.59)*

in the international market experience of the earlier studies. For example, while

CN 0.5037 0.0058 higher PCM in the long run. Although (1.44) (-0.03)Kambhampati and Parikh (2003) find a

MES -0.1398 0.6916

market size is by and large not considered statistically significant positive relation

(-0.19) (1.11)

in modelling concentration-markup rela-ship of advertising and R&D with profit

MSZ 3.0068 1.4338 tionship, in line with many of the exist-(1.92)# (1.14)margin at firm level, the present paper

KIR 0.0067 0.1161

ing studies, the present paper finds posi-finds an inverse relationship of selling

(0.03) (0.76)

tive impact of growth of the industry [for efforts and no statistically significant

SELL -0.7404 -0.4966 example, Rao 2001] or of exports [Kamb-(-2.76)* (-3.80)*relationship of in-house R&D with

RDI 0.0187 0.0739

hampati and Parikh 2003] on profitabili-profit margins.

(0.24) (1.61)

ty.20 A larger and growing domestic mar-What is most interesting perhaps

FTPI 0.0158 0.0219 ket coupled with greater penetration in (0.45) (0.81)is that the relationship between

EXP 0.3041 0.1991

the international market facilitates the market concentration and PCM is not

(1.66)# (2.57)*

firms to operate at a higher scale and statistically significant in the long

Wald-Chi2 214.12 884.85

thereby, to reduce average costs of pro-run. This is contradictory to the positive

Sargan test for 8.60duction. This helps the firms in these over-identifying restrictions [0.38] relationship between the two as it is

Arellano-Bond test -1.06 -0.91

industries to record higher profitability found in many of the existing studies in

for AR (1) [0.29] [0.36]

in the long-run. Indian context [Kambhampati 1996; Rao

Arellano-Bond test 0.88 0.57 The statistically significant but nega-for AR (2) [0.38] [0.57] 2001; Goldar and Agarwal 2004]. There

tive coefficient of SELL suggests that even No of observations 524 524 are three possible reasons behind such a

(i) Figures in the parentheses indicate respective z-statistic.

greater selling related efforts do not nec-finding. First, with the failure of the

(ii) The z-statistic in one-step estimation is based on robust standard error to control for heteroscedasticity and autocorrelation.

essarily result in higher profit margin. strategies towards creating entry barri

(iii) Figures in the square brackets indicate level of statistical This contradicts with the general percep-significance of the respective test statistic. ers and eventual entry of new firms,

(iv) *Statistically significant at 5 per cent level of significance.

tion that greater selling intensity through (v) #Statistically significant at 10 per cent level of significance. competition in the market place

advertising helps firms to raise PCM in the long run by creating brand image, disseminating information about the products and related services and creating strong barriers for the potential entrants. Selling efforts are also likely to provide marketing and distribution related complementary assets to the firms. Therefore, the industries with greater selling intensity are generally expected to experience higher PCM. However, such expectation may be fulfilled in the short run when strategic rivalry among the firms may not be effective enough. In the long run, on the other hand, strong strategic rivalry among the firm is very likely even to maintain market position and therefore, the firms’ may not benefit from such strategies. Under these circumstances, greater selling efforts raise firms’ total cost of operation but fail to provide an individual firm any competitive advantage over the rivals in the market. As a consequence, though selling intensity in aggregate increases, the industries record lower profit margin.

MES, KIR, RDI and FTPI do not have any statistically significant relationship with PCM in a dynamic context. This means that MES or KIR fails to create effective entry barrier for new firms and barrier to the upward mobility of less favoured firms in the long run and therefore, does not have any statistically significant impact

Economic & Political Weekly

EPW
september 27, 2008

increases. This increase in market competition reduces PCM of the firms. Secondly, high initial market concentration and PCM make the incumbents complacent. This results in X-inefficiency and hence, in lower PCM of these firms. Another possible reason might be the deceleration in industrial production.21 This argument follows the general recognition that the relationship between concentration and profit margin is weaker during the periods of recession than the periods of boom [Lukas 1999].

4 Conclusions

The present paper attempts to explore inter-industry variations in the concentration-margin relationship controlling for various other aspects of market structure, firms’ business strategies, their past performance and policies of the government in a dynamic framework. It is observed that lasgged PCM has a statistically significant positive impact on its current level. In other words, industries with greater PCM in the previous year experience greater profit margin in the current year as well. In addition, rate of growth of sales, size of the market and exports intensity also have statistically significant positive impact on profitability,

REVIEW OF INDUSTRY AND MANAGEMENT

whereas, selling efforts have negative impact on the same. greater market concentration do not necessarily enjoy higher H owever, minimum efficient scale or capital intensity ratio fail to profit margin in the long run. This may be due to entry of new create any effective entry barrier and hence, to influence PCM in a firms, X-inefficiency of the incumbents or deceleration in indussignificant way in the long run. Similarly, technology strategies trial production. This raises a couple of important questions with also do not have any significant impact on inter-industry respect to market competition and competition policy: is an v ariations in profit margin. increase in market concentration necessarily harmful for the con-

Significantly, the traditional positive concentration-margin sumers in the long run as it may be in the short run? Should the relationship is not valid in the dynamic context. This means that policymakers be worried too much for increase in market conwhen controlled for various structural aspects of the market, centration? Addressing these questions properly, however, firm’ strategies and policies of the government, industries with requires further research in this direction.

Notes

1 See, for example, Weiss (1974), Lindsey (1976), Raven scraft (1983), Salinger (1990), Kambhampati (1996), Rao (2001) and Goldar and Aggarwal (2004).

2 In this connection, it should be mentioned that throughout the late 1960s, there was a consensus that concentration of firms in the marketplace increased their profitability and facilitated collusion, though this positive correlation disappeared in the 1970s.

3 See, for example, Bain (1951), Chamberlin (1933) and Stigler (1964) for explanations for the proposition that concentrated industries facilitate collusion leading to supernormal profit.

4 For example, Kambhampati and Parikh (2005) attempted to examine the determinants of firmlevel profit margins in Indian manufacturing s ector using a dynamic market model.

5 Operational efficiency assures short-term success, whereas strategies matter to succeed in the long run.

6 There are some alternative approaches to the industrial organisation literature, such as, the Marshallian School, Austrian School, and ‘Workable Competition’ School. See Reid (1987) for a discussion in this regard.

7 For example, high profitability from efficiency enhancement can raise market share and hence market concentration. Similarly, high profitability can lead to greater research and development (R&D) activities, advertising, and predation whereas low profitability can result in collusion. As regards conduct-structure relationship, strategies like R&D mergers, acquisitions, advertising, predation, alliances can affect market concentration significantly.

8 For example, while policy changes of the 1990s had changed the basic environment and functioning of the Indian corporate sector in a considerable way, the wave of mergers, acquisitions and other collusive strategies of the 1990s and emerging market conditions forced to implement the Competition Bill, 2002. Similarly, while a reduction in taxes and/or increase in subsidies

can raise profitability of the firms, unsatisfactory performance of an industry may compel the government to correct policies relating to public investment, regulation and controls, taxes and subsidies, etc.

9 Using such an average measure of the dependent variable is very important in a multidimensional framework, as in such a framework the adjustment process is likely to be slow and a single lag dependent variable based on annual values as an explanatory variable may not be enough to capture the entire dynamics of the model.

10 In industrial organisation literature, financial performance of business operations is generally examined in terms of profitability. There are two types of profitability ratios – (i) profit margin ratios, and (ii) rate of return ratios. While the profit margin ratios show the relationship between profit and output or sales, the rate of return ratios reflect the relationship between profit and investment. As the objective of the present paper is to examine concentrationmarkup relationship, we have used PCM as the measure of profit margin.

11 Import is not included as a structural variable in the present model as it is very difficult to isolate the effect of import of intermediate products from that of final ones. However, such exclusion is unlikely to affect the estimated relationship in a significant way as import is more strongly related to market concentration rather than PCM.

12 The basic objective estimating this generalised model is to find out the relationship between PCM and CN with GRS, MSZ, MES, KI, SELL, RDI, FTPI and EXP as the controlling variables. This is so because, apart from market concentration, performance of an industry depends on other structural characteristics of the market, firms’ conduct and policies of the government as well.

13 It should, however, be mentioned that the choice among different measures does not matter much as the literature shows that various concentration measures are highly correlated and provide similar findings [Scherer and Ross 1990].

14 However, a negative coefficient of RDI is not surprising as the existing accounting practices allow firms to express R&D expenses entirely in the year incurred instead of amortising it to recognise its future benefits.

15 It should be mentioned that since the variables are transformed into their natural logarithms the negative and zero values of the variables like PCM, RDI, FTPI and EXP are treated as the m issing observations. This makes the panel an unbalanced one. However, considering that there are large numbers of observations in the dataset, impact of these missing values on the estimates is quite negligible.

16 Since the industry is the unit of observation in the present context, endogeneity of the explanatory variables is unlikely to be acute as it normally is when the firm or the line of business is the unit of observation [Salinger 1990].

17 The use of such dynamic models is favoured, especially, for panels that have a large number of cross sectional units with a small number of time periods, as we have in the present case. This is so because their estimation methods do not require larger time periods to obtain consistent parameter estimates.

18 Power of the Sargan Test may decline when the number of cross sectional units is small compared to the time-series units. But, this is not so in the present case.

19 The Arellano-Bond test for autocorrelation may become unreliable when the number of cross sectional units is small.

20 However, there are also studies [for example, Ghose 1975; Goldar and Agarwal 2004] that document an inverse relationship between growth of industry and profitability.

21 This argument seems to be very important in the present context, as the Indian industry sector experienced a deceleration in production during 1996-2000.

References

Arellano, M and S Bond (1991): ‘Some Tests of S pecification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations’, The Review of Economic Studies, Vol 58, pp 277-97.

Bain, J S (1951): ‘Relation of Profit Rate to Industry Concentration: American Manufacturing, 19361940’, Quarterly Journal of Economics, Vol 65, pp 293-324.

– (1959): Industrial Organisation, Wiley, New York.

Basant, R and S Morris (2000): Competition Policy in India: Issues for a Globalising Economy, project report submitted to the Ministry of Industry,

Government of India, New Delhi. Basant, R and S N Saha (2005): ‘Determinants of Entry in the Indian Manufacturing Sector’, Working Paper No 2005-01-01, Indian Institute of Management, Ahmedabad. Chamberlin, E H (1933): The Theory of Monopolistic Competition, Harvard University Press, C ambridge, Mass.

Unbound Back Volumes of Economic and Political Weekly from 1976 to 2007 are available.

Write to: Circulation Department,

Economic and Political Weekly

320-321, A to Z Industrial Estate, Ganpatrao Kadam Marg, Lower Parel, Mumbai 400 013. Circulation@epw.in

september 27, 2008

REVIEW OF INDUSTRY AND MANAGEMENT

Demsetz, H (1973): ‘Industry Structure, Market Rivalry and Public Policy’, Journal of Law and Economics, Vol 16, pp 1-9.

Ghose, A (1975): ‘Concentration and Controls in the Indian Industry’, Journal of Industrial Economics, Vol 23, No 3, pp 203-22.

Goldar, B and S C Aggarwal (2004): ‘Trade Liberalisation and Price-Cost Margin in Indian Industries’, Working Paper No 130, Indian Council For Research On International Economic Relations, New Delhi.

Hay, D A and D J Morris (1991): Industrial Economics and Organisation: Theory and Evidence, Oxford University Press, Oxford.

Kambhampati, U S (1996): Industrial Concentration and Performance; A Study of the Structure, Conduct and Performance of Indian Industry, Oxford University Press, Delhi.

Kambhampati, U S and A Parikh (2003): ‘Disciplining Firms: The Impact of Trade Reforms on Profit Margins in Indian Industry’, Applied Economics, Vol 35, No 4, pp 461-70.

– (2005): ‘Has Liberalisation Affected Profit Margins in Indian Industry?’, Bulletin of Economic Research, Vol 57, No 3, pp 273-304.

Lindsey, C W (1976): ‘Concentration in Philippine Manufacturing, 1960-70’, unpublished doctoral dissertation, University of Texas at Austin.

Lukas, P (1999): ‘Vertical Integration, Input Contracts and the Margin-Concentration Relationships’, Applied Economics, Vol 31, pp 631-40.

McDonald, J T (1999): ‘The Determinants of Firm Profitability in Australian Manufacturing’, Economic Record, Vol 75, No 229, pp 115-26.

McGee, J S (1971): In Defense of Industrial Concentration, Praeger, New York.

Mishra, P (2005): ‘Mergers and Acquisitions in the Indian Corporate Sector in the Post-Liberalisation Era: An Analysis’, unpublished doctoral thesis, Vidyasagar University, West Bengal.

Rao, P A (2001): A Study of the Determinants of Firm Profitability in Selected Industries in Post-Reform India, unpublished MPhil dissertation, Delhi School of Economics, University of Delhi.

Ravenscraft, D J (1983): ‘Structure-Profit Relationships at the Line of Business and Industry Level’, Review of Economics and Statistics, Vol 65, pp 22-31.

Reid, G (1987): Theories of Industrial Organisation, Basil Backwell, Oxford.

Salinger, M (1990): ‘The Concentration-Margins Relationship Reconsidered’, Brookings Papers on Economic Activity: Microeconomics, pp 287-336.

Scherer, F M and D Ross (1990): Industrial Market Structure and Economic Performance, Rand Mcnally, Chicago.

Stigler, George J (1964): ‘A Theory of Oligopoly’, J ournal of Political Economy, Vol 72, pp 44-61.

Weiss, L W (1974): ‘The Concentration-Profits Relationship and Antitrust’ in H J Goldschmid, H Michael Mann and J Fred Weston (eds), Industrial Concentration: The New Learning; Little Brown, Boston, pp 184-233.

Appendix I: Measurement of Variables

The modified SCP paradigm suggests that in many situations various structure, conduct and performance characteristics of an industry are simultaneously determined. In many other situations, lags in the relationship among the characteristics can lead to non-simultaneous or sequential determination of these variables. In the present paper, we attempt to examine the relationship between concentration and mark-up taking adequate care of the potential lags in the envisaged relationships. Futher, f ollowing Basant and Saha (2005), the

Economic & Political Weekly september 27, 2008

EPW

v ariables are measured with lags of three years to avoid the potential simultaneity bias among them. Such measures are also expected to make the dataset more consistent over the period of time.

PCM: The variable PCMjt is measured as equal to n n n

⎛ ⎞

⎜Σ(VAit–WSit )⎟

) Σ(VAi,t–1–WSi, t–1) Σ(VAi,t–2–WSi, t–2

i=1 i=1 i=1

+ +⎟

⎜ n n n ⎟ ⎜Σ VOit Σ VOi,t–1 Σ VOi,t–2 ⎟

⎝ i=1 i=1 i=1 ⎠

3

where PCMjt = price-cost margin of industry j in period t, VA = value added by firm i in period t, WS = wages and salaries paid by firm i in period t, and VO = value of output of firm i in period t.

GRS: In the present paper, GRS is measured as

= Sjo (1+gj)t

Sjt

Here g stands for the rate of growth of sales (S) of industry j. This function is regressed over a period of five years with a one-year lag in the starting year. Here, g is estimated by applying regression method.

CN: The variable CN is measured by using the formula

nn n( Σ S2it+ Σ S2i,t–1+Σ S2i,t–2 )

i=1 i=1 i=1 CNjt =

3

Here CNjt is the extent of market concentration in industry j in period t and si stands for m arket share of the ith firm in the industry in period t. The market share of a firm is defined as the ratio of the firm’s sales to total industry sales.

MES: This variable MES is measured as,

nn n([Σ i=1 Sit ]+log [i=1 Σ Si,t–1 ][Σ Si,t–2 ])

log +log i=1 nn n MESjt =

3

where MESjt stands for minimum efficient scale of industry j in period t, Si for sales of the ith firm in period t and n for number of firms in the industry.

KIRL: In the present study, KIR is measured by using the following formula:

nn n

Σ CEit Σ CEi,t–1 Σ CEi,t–2

i=1 i=1 i=1

+ +

n n n

(
3)

Σ Sit Σ Si,t–1 Σ Si,t–2

i=1 i=1 i=1 KIRjt =

where KIRjt stands for capital intensity ratio in industry j in period t and CE for capital em ployed by the ith firm in the industry in period t.

MSZ: In the present study, MSZ is measured as:

n n n

log ( )

Σ Sit+ Σ Si,t–1+ Σ Si,t–2

i=1 i=1 i=1 MSZjt =

3

where MSZjtis the size of the market of industry j in period t.

SELL: In the present paper SELL is measured as the sum of the ratio of distribution and marketing expenses (DM) to sales and the ratio of advertising expenditure (A) to sales, i e, the sum of distribution and marketing intensity (DMI) and advertising intensity (AI). SELLjt = DMIjt + AIjt

⎡⎛ n n n ⎞ Σ DMit Σ DMi,t–1 Σ DMi,t–2

⎢⎜⎜i=1 i=1 i=1⎟⎟

+ +

⎢⎜n n n ⎟ Σ Sit Σ Si,t–1 Σ Si,t–2

⎢⎜⎝i=1 i=1 i=1 ⎟⎠ SELLjt= ⎢ 3+⎣

⎛n n n ⎞

⎜ ⎟

Σ Ait Σ Ai,t–1 Σ Ai,t–2

⎜i=1 i=1 i=1⎟

⎜ + + ⎟

n n n

⎜ ⎟

Σ Sit Σ Si,t–1 Σ Si,t–2

⎝ i=1 i=1 i=1 ⎠ 3 ⎦

where SELLjt is the selling intensity in industry j in period t.

RDI: The ratio of expenditure on in-house R&D (RD) to sales is used as a measure of RDI, i e,

nn n

Σ RDit Σ RDi,t–1 Σ RDi,t–2

i=1 i=1 i=1 + +

n n n

(
3)

Σ Sit Σ Si,t–1 Σ Si,t–2

i=1 i=1 i=1 RDIjt =

where RDIjt stands for in-house R&D intensity in industry j in period t.

FTPI: Foreign technology purchase intensity (FTPI) is measured as the ratio of expenditure on foreign technology purchase (FTP) to sales, i e,

nn n Σ FTPIit

Σ FTPIi,t–1 Σ FTPIi,t–2

i=1 i=1 i=1 ++

nn n

Σ Sit Σ Si,t–1 Σ Si,t–2

i=1 i=1 i=1 FTPIjt=

(
3)

where FTPIjt stands for foreign technology p urchase intensity in industry j in period t.

EXP: The variable EXP is defined in the following way:

n n n

Σ Eit Σ Ei,t–1 Σ Ei,t–2

i=1 i=1 i=1 ++

nn n

(3 )

Σ Sit Σ Si,t–1 Σ Si,t–2

i=1 i=1 i=1 EXPjt =

where EXPjt is the export intensity of industry j in period t and E stands for exports by firm i in the industry in period t.

To read the full text Login

Get instant access

New 3 Month Subscription
to Digital Archives at

₹826for India

$50for overseas users

Comments

(-) Hide

EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

This article surveys recent development in Indian employment relations, broadly defined, from the perspective of a visiting British academic. The...

Mergers and acquisitions are being increasingly used the world over as a strategy for achieving larger size, faster growth in market share and...

This paper presents a new framework for the memoranda of understanding that central public sector enterprises sign with the government for...

In the mainstream international management literature the issue of the extent to which multinational corporations achieve the outcomes desired by...

Though India has a good number of Schumpeterian entrepreneurs who have made the country a frontrunner in information technology, the country also...

The corporate sector in India has witnessed a substantial growth of mergers and acquisitions since the 1990s. Through the first wave of M&As (...

In today's globalised economy, mergers and acquisitions are being increasingly used the world over as a strategy for achieving a larger size and...

Uttar Pradesh, a leading producer of sugar cane and sugar before independence, lost its pre-eminent position to Maharashtra in the 1960s, when the...

Almost all earlier studies comparing pharmaceutical prices in Pakistan and India have attributed higher prices in Pakistan mainly to the...

This paper analyses Maharashtra's data on demand, supply and load shedding. The trends in the hourly load variation and seasonal variation of...

Back to Top