Competition in Indian Manufacturing Industries A Mobility Analysis
This paper devises an improved turnover index and applies it to examine the mobility of firms in the Indian manufacturing sector during the post-reform period. The new index is used to test the stability of size ranks and analyse the changes in the degree of mobility. The paper studies the change in size distributions of industries and their inter- and intra-class mobility, and tests for the relationship between the dynamic index of competition and the direction of mobility of firms among manufacturing industries.
K PUSHPANGADAN, N SHANTA
O
In the literature the mobility of firms in an industry is identified as one important dimension – a dynamic one, of what constitutes effective competition.4 According to Singh and Whittington, the study of mobility of firms in an industry by enquiring into the amount of mixing and reordering that takes place in the size ranks of individual firms, focuses on the dynamic aspects of industrial structure.5 In this context it would be useful to point out here that public policy might be concerned more with a measure of mobility of firms or the turnover of firms in an industry, which would provide a better index of what constitutes “equality of opportunity” rather than the usual measures of concentration.6 In the discussions on what constitutes effective competition, Ijiri and Simon point out that a “frequent and sizeable change in the ranks of firms in an industry would indicate vigorous competition in the industry. On the other hand if the ranks do not change frequently, it is an indication of little competition.” On measuring the dynamics of competition, according to Baldwin,7 “much of what happens during the competitive process will be manifested by changes in relative firm position. Mobility measures provide a direct measure of the intensity of competition”.
Again, the turnover measure can throw light on important dimensions of competition, which cannot be captured by the structural indices such as the concentration ratios. For instance the n-firmconcentration ratio can remain the same, but the identity of the firms in the group can change because of competition. This can be captured only by the turnover measure.8 Besides, for identifying the market structure of an industry an understanding of the transition of firms is essential.9 In short given the multidimensional nature of competition10 and the special attributes of rank analysis, it is an important “structural indicium” for understanding competition.
This, however, does not mean that this measure is free from controversies. One major debate in the literature is the utility of the turnover index as a measure of market behaviour. At one end you find this measure being encouraged as a useful tool for measuring competition and a replacement for concentration ratios,11 while at the other end there were others12 who did not consider it a useful concept for the reason that it recorded events of little economic significance and dismissed it as a useless measure.13 Midway between the two some find it a useful supplement and additional idicia for understanding competition.14
Our critical review of the literature has led to the conclusion that if the limitations of the turnover index are taken care of, it can provide several important insights on competition which cannot be captured by other indicators and is therefore an important tool for analysing competition.
The objective of this paper is therefore to devise an improved turnover index, which overcomes the limitations of the traditional turnover index, and to apply it for measuring industrial competition. More specifically, using the new method this study examines the mobility of firms in the Indian manufacturing sector for the post-reform period.
Towards this end, the specific objectives of this paper are the following: (1) critically examine the turnover index and provide an alternative; (2) using the new index, test the stability of size ranks and analyse the changes in the degree of mobility; (3) study the change in size distributions of industries and their inter- and intra-class mobility; and (4) test for the relationship between the dynamic index of competition and the direction of mobility of firms among manufacturing industries.
The paper is organised into four sections. Section I gives the data and concepts used. Section II examines the main criticisms against the traditional turnover index and provides an alternative. Section III gives the empirical results based on the new turnover index followed by the summary and conclusions in Section IV.
I Data Sources and Measurement of Variables
The period of study for the analysis is from 1988-89 to 2000-01, one that largely coincides with the period of reforms. Data relate to financial not to calendar years as seen by the overlapping of the years throughout the paper. In order to test this model we require firm level data. The best published source available for this purpose is the data assembled by the Centre for Monitoring Indian Economy (CMIE) in electronic form commonly known as Prowess.15 The nature of this data is such that the sample size changes every year, some being dropped and some being added. In other words the database is an unbalanced panel. The reasons for this are probably non-availability of data and need not necessarily imply entry into or exit from any industry. Therefore, we cannot include entry and/or exit of firms in our analysis with this data, forcing us to consider only a balanced panel (firms that have existed throughout the period) covering the period 1988-89/2000-01. By this criterion, we have 14 industrial groups covering a wide range of products as seen from Table 1.
Measurement of Size
In the literature various measures of size such as net value added, sales, total assets and employment are used and the problems associated with them are discussed in detail16 and there is no agreement on the best measure of size. In this situation we have selected the appropriate size variable based on a correlation analysis amongst the three commonly used measures for which we have data.
Table 1: Sample Size of Balanced Panel by Industry Group,1988-89 to 2000-01
| Sl No | Industry | Sample Size |
|---|---|---|
| 1 | Beverages and tobacco | 1 4 |
| 2 | Cotton and blended textiles | 5 1 |
| 3 | Drug and pharmaceuticals | 3 0 |
| 4 | Electrical machinery | 4 7 |
| 5 | Electronics | 3 0 |
| 6 | Food products | 4 1 |
| 7 | Iron and steel | 2 5 |
| 8 | Metal products | 2 2 |
| 9 | Non-electrical machinery | 5 9 |
| 1 0 | Non-ferrous metals | 1 6 |
| 1 1 | Non-metallic minerals | 4 8 |
| 1 2 | Synthetic textiles | 2 2 |
| 1 3 | Transport equipment | 7 2 |
| 1 4 | Wood and paper | 2 0 |
| All industries | 49 7 |
Source: CMIE, Prowess (1997, 2001).
Correlation coefficients between various measures of size (sales, total net assets and net value added) have been computed for 14 industrial groups at the beginning of the period, 1988-89, and at the end of the period, 2000-01, and the results are reported in Table 2. It is interesting to note that all are highly correlated except for one or two industries for both the years. Since the correlation coefficient between all the three measures are quite high and significant, for almost all the industries, they can be interchangeably used. In our analysis, however, we use total net assets as the measure of size.
II The New Turnover Index
Although several limitations of the turnover index have been raised in the literature, we focus our analysis on the two major criticisms, most relevant to the measurement of competition, raised by Hymer and Pashigian (1962). The first relates to the rank ordering by size and the second to the relationship between turnover and market behaviour.
The main drawback of the rank ordering of size is that, it shows only whether the size of a firm is higher or lower than another firm but not by how much. This is a serious problem in a rankshift analysis. Hymer and Pashigian using the example of the shoe-industry from Joskow’s paper convincingly argue this point. To quote, “At the top of the industry, a difference of rank is large absolutely and relatively; some substantial force is needed to close the gap. But the great bulk (say seven-eighths) of the rank differences in Joskow’s sample are trifling absolutely and relatively, and so are changes in rank... such rank changes signify nothing”.17 In other words, while there is almost no competition, such rank shifts signify otherwise.
The second major criticism is that the changes in market shares which is the really significant phenomenon for understanding the market does not get reflected in the rank analysis. Hymer and Pashigian graphically illustrate this by examining the changes in the market shares and ranks of two leading firms, Ford and General Motors, from the US automobile industry.18 They show that even when there were large changes in market shares, there were no changes in ranks.
In short, the main argument against the use of rank correlation to understand competition is that it does not take into account size variations. That is, it is based on rank ordered data derived
Table 2: Correlation Coefficient of Alternative Measures of Size by Industry, 1988-89 to 2000-01
| 1988-89 | 2000-01 | |||||
|---|---|---|---|---|---|---|
| Industry | Sales/Net | Sales/Total | Net Value Added/ | Sales/Net | Sales/Total | Net Value Added/ |
| Value Added | Net Asset | Total Net Asset | Value Added | Net Asset | Total Net Asset | |
| Beverages and tobacco | 0.93 | 0.87 | 0.96 | 0.99 | 1.00 | 0.99 |
| Cotton and blended textiles | 0.89 | 0.12 | 0.42 | 0.52 | 0.31 | 0.92 |
| Drug and pharmaceuticals | 0.96 | 0.89 | 0.87 | 0.97 | 0.92 | 0.88 |
| Electrical machinery | 0.97 | 0.95 | 0.94 | 0.85 | 0.96 | 0.82 |
| Electronics | 0.92 | 0.88 | 0.98 | 0.83 | 0.88 | 0.78 |
| Food products | 0.86 | 0.74 | 0.81 | 0.69 | 0.73 | 0.84 |
| Iron and steel | 0.99 | 0.88 | 0.87 | 0.98 | 0.97 | 0.92 |
| Metal products | 0.96 | 0.92 | 0.91 | 0.92 | 0.88 | 0.90 |
| Non-electrical machinery | 0.99 | 0.98 | 0.99 | 0.99 | 0.99 | 0.99 |
| Non-ferrous metals | 0.88 | 0.60 | 0.58 | 0.91 | 0.95 | 0.96 |
| Non-metallic minerals | 0.93 | 0.93 | 0.77 | 0.90 | 0.92 | 0.94 |
| Synthetic textiles | 0.95 | 0.94 | 0.84 | 0.24 | 0.88 | 0.44 |
| Transport equipment | 0.96 | 0.82 | 0.83 | 0.68 | 0.90 | 0.70 |
| Wood and paper | 0.73 | 0.56 | -0.07 | 0.97 | 0.96 | 0.92 |
| Source: Same as in Table 1. | ||||||
| Economic and Political Weekly | September 30, 2006 | 4131 |
only from the order of the numbers and not the difference between them. Such ordinal measures provide information regarding greater than or less than status, but not how much greater or less. This problem is overcome by normalising the data through an order preserving transformation as detailed below.
Let there be n firms in an industry and S1, .., Srepresent their
n sizes in the ascending order of magnitude. To get the order preserving ranks incorporating the size differences, the following transformation is undertaken.
Zj = [Sj – min (S1, .., S)]/[max (S1, .., S)
nn
– min (S1, .., S)] j = 1, .., n. ...(1)
n
By this, the sizes of firms (S1,…, S) have been transformed
nto (Z1, .., Z )with the interval ranging from 0 to 1. In order to
nreplace the zero in this transformation, we have cumulated the series by adding the differences between two consecutive Zjvalues beginning from the highest value of one to obtain the
––
transformed size ranks (Z1,…, Z). The same procedure can be
nrepeated to get transformed ranks for any year and the paired values for any two years can then be used for the calculation of rank correlation coefficient. We further proceed to establish that the order preserving transformation is also related to the market shares of firms. It may be noted that Zj is related to the share of jth firm as shown below. Dividing the numerator and the denominator in equation
n
(1) by Σ Si
i=1
Sj/ΣSi – min(S)/ΣSi Zj = ——————————— max(S)/ΣSi – min(S)/ΣSi
Sj/ΣSi min(S)/ΣSi = —————————– – —————————— max(S)/ΣSi – min(S)/ΣSi max(S)/ΣSi – min(S)/ΣSi
= k1Mj-k2 ...(2)
where Mj = Sj/Σ Si j = 1, .., n
k1 and k2 are constants.
Equation (2) clearly shows that the transformed variable of the jth firm is linearly related to its market share of assets. The relationship between the new turnover measure and the market share instability index as formulated by Hymer and Pashigian has been empirically verified for Indian industries and found to be statistically significant.19 This clearly establishes that the new measure also reflects the changes in market shares. The superiority of this measure over market shares will be established in the empirical analysis that follows. Let us examine the application of the new turnover index (NTI) to Indian industries.
III Empirical Results
Having devised a new turnover index, we have undertaken three major exercises to get insight into the firms’ dynamics and the nature of competition: (1) Using rank correlation we test for the stability of size ranks; (2) We estimate standard deviation of relative ranks using Ijiri-Simon method, for the end points to understand the change in the degree of competition; and
(3) Following Hart and Prais20 and Singh and Whittington,21 an analysis of the change in size structure and mobility patterns is attempted using transition matrices.
Stability of Size Ranks
To test for stability of size ranks, rank correlation is used. If the rank correlation between size of firms at two different points of time is not significantly different from zero then there are frequent and sizeable changes in ranks indicating competition among the firms. On the other hand, if the correlation is significant it implies lack of competition. There is absolutely no competition if the value of correlation coefficient is unity. In reality, the correlation coefficient can take any value between the two extremes 0 and 1. The empirical results for the sample of industries are given in Table 3.
All the coefficients are significant at the 5 per cent level. More importantly they are greater than or equal to 0.90 for five industries out of 14 suggesting a situation of perfect or close to perfect rigidity. It is interesting to note that perfect mobility does not prevail in any industry although they exist in varying degrees. The varying degrees of mobility can be characterised as low, medium or high mobility according to the value of the correlation coefficient. For further analysis, industries with a coefficient of less than 0.50 are treated as those with high mobility, between
0.50 and 0.75 as those with medium mobility and above 0.75 as those with low mobility. By this classification we have only one industry with high mobility (non-metallic minerals). In the medium category we have seven industries, drug and pharmaceuticals, electronics, food products, metal products, non-ferrous metals, synthetic textiles, and wood and paper. The remaining six industries fall in the category of those with low mobility. They are beverages and tobacco, cotton and blended textiles, electrical machinery, iron and steel, non-electrical machinery and transport equipment. It is worrying to note, that about 43 per cent of the industries are characterised by low mobility and barriers to mobility in varying degrees exist in all the industries considered for the analysis.
However, what is important to our analysis is that one should not only test for stability of ranks and infer from it whether size structure is mobile or rigid, but it should be supplemented with information on the change in the degree of mobility over the
Table 3: Rank Correlation of New Transformed Size between 1988-89 to 2000-01
Industry Rank Correlation of Size (1988-89/2000-01)
Beverages and tobacco 0.98 Cotton and blended textiles 0.98 Drug and pharmaceuticals 0.60 Electrical machinery 0.90 Electronics 0.51 Food products 0.73 Iron and steel 0.94 Metal products 0.55 Non-electrical machinery 0.98 Non-ferrous metals 0.61 Non-metallic minerals 0.43 Synthetic textiles 0.69 Transport equipment 0.84 Wood and paper 0.70
Note: All the coefficients are significant at the 5 per cent level. Source: Same as above.
period, that is, the average amount of shifting in the ranks during the period. This would indicate whether competition has increased/decreased in the manufacturing sector.
Ijiri and Simon22 suggest a measure for this – a measure based on the relative ranks of firms at two different points. We consider the beginning and the end of the period for such an analysis. This measure is defined as the standard deviation of qi = ri/ri*., where ri is the rank of the ith firm at the end of a period and ri* is its rank at the beginning of the period. The standard deviation of ‘q’ is an indicator of the average amount of shifting in rank during the period. We have carried out this exercise, as suggested by Ijiri-Simon, for all the 14 industries for the two consecutive years at the beginning, 1988-89 and 1989-90, as well as at the end, 1999-2000 and 2000-01, of the period of analysis. The results are given in Table 4.
From Table 4, it can be seen that in six out of 14 industries the average shifting has come down, suggesting that competition has declined. Out of the remaining eight industries, for one industry (electrical machinery), the average shifting remained the same while for the other seven industries it has increased. This suggests that only in these seven industries competition has increased. The decline in competition varies from 18 per cent to 60 per cent while the increase ranges from 13 per cent to 304 per cent. Again while average shifting in the initial period ranges from 0.002 to 0.051, in the end period it ranges from .002 to
0.042 implying a decline in the average rate of shifting (upper value). It is also to be noted that the highest average shifting in the initial year was in the metal products industry (0.051), the lowest was in cotton and blended textiles followed by nonelectrical machinery. In the end period, it was again metal products (0.042), but the lowest being beverage and tobacco industry followed by non-electrical machinery.
The following trend in average shifting is worth noting. The largest decline in the average shifting (-60.0 per cent) was in beverages and tobacco. The least decline occurred in the metal products industry (-17.6 per cent). The largest increase in average shifting occurred in the wood and paper industry while the smallest increase was observed in the food products industry. This suggests that the pace of increase/decrease in competition also varies widely among the industries.
To understand whether competitive industries remained competitive or became non-competitive and vice versa, we have calculated rank correlation coefficients between the average shifting at the two end points in Table 4. We get a rank correlation of 0.70 suggesting that the position of industries in relation to competition or non-competition does not seems to have changed very much. Since average shifting for all industries has also remained almost the same (0. 0180 and 0.0182), it shows that no substantial change in the structure of competition has occurred during the period.
The rank correlation analysis based on the new turnover index and the average shifting measured by Ijiri-Simon index have provided some indication of the presence of competition in the industries. But this does not provide any indication of the impact of mobility on size structure. In order to get insights into this, we estimate standard measures based on the moments of the frequency distributions of industries such as the coefficient of variation, skewness and kurtosis for the end points for all the industries as reported in Table 5.
The coefficient of variation of size has increased in six industries, substantially in the industries of beverages and tobacco and drug and pharmaceuticals, and decreased in the remaining eight industries notably in non-ferrous metals and wood and paper. Next we examine how strong have been these trends in the concentration as indicated by the skewness of the distribution. The interesting feature of the distributions is that all of them are skewed to the right both in the initial and terminal year as indicated by the positive value of the third moment. In other words in all the industries, there is a concentration of firms in the smaller size-classes for both periods. The asymmetry has increased only in five industries; beverages and tobacco, drug and pharmaceuticals, metal products, non-metallic minerals and synthetic textiles. In the case of peakedness, as measured by Kurtosis, it has increased in four industries and decreased in the remaining 10. In other words, the majority of the cases, the peakedness has decreased.
The analysis of the moments of the distributions clearly shows that mobility analysis is incomplete unless we examine the nature of size structure in general and inter- and intra-class mobility in particular. This is accomplished by undertaking analysis of transition matrices of all the industries, which alone can capture the nuances of firm mobility. This is illustrated with the transition
Table 4: Ijiri-Simon Index of Competition, 1988-89 to 1989-90and 1999-2000 to 2000-01
Standard Deviation of Relative Ranks
Industry 1989-90/ 2000-01/ Per Cent 1988-89 1999-2000 Change 1 234
Beverages and tobacco 0.005 (4.5) 0.002 (1) -60.0 Cotton and blended textiles 0.002 (2) 0.003 (3.5) 50.0 Drug and pharmaceuticals 0.036 (12) 0.021 (10) -41.7 Electrical machinery 0.040 (13) 0.040 (13) 0.0 Electronics 0.011 (7) 0.037 (12) 236.4 Food products 0.015 (8) 0.017 (7) 13.3 Iron and steel 0.005 (4.5) 0.003 (3.5) -40.0 Metal products 0.051 (14) 0.042 (14) -17.6 Non-electrical machinery 0.0018 (1) 0.0022 (2) 22.2 Non-ferrous metals 0.0102 (6) 0.0234 (11) 129.4 Non-metallic minerals 0.0167(9) 0.0202 (9) 21.0 Synthetic textiles 0.0311 (11) 0.0167 (6) -46.3 Transport equipment 0.022 (10) 0.009 (5) -59.1 Wood and paper 0.0046 (3) 0.0186 (8) 304.3 Average for all industries 0.0180 0.0182
Correlation between col 2 and col 3 = 0.699
Note: Figures in parenthesis are their ranks respectively. Source: Same as above.
Table 5: Changes in the Coefficient of Variation, Skewness andKurtosis of Size by Industries, (1988-89 to 2000-01)
Industry CV Skewness Kurtosis 1988-2000-1988-2000-1988-200089018901 8901
Beverages and tobacco 1.77 2.87 3.43 3.71 12.34 13.81 Cotton and blended textiles 3.62 3.19 6.93 6.71 48.90 46.73 Drug and pharmaceuticals 1.02 1.55 2.47 3.50 5.91 13.96 Electrical machinery 1.26 1.34 3.07 2.24 11.85 5.09 Electronics 2.41 1.65 3.57 2.07 13.24 3.94 Food products 1.21 1.18 2.60 2.10 7.55 5.20 Iron and steel 3.60 2.48 4.73 3.60 22.94 13.94 Metal products 1.11 1.33 1.27 2.18 0.19 5.60 Non-electrical machinery 3.62 3.06 6.83 6.63 49.35 47.43 Non-ferrous metals 2.16 1.32 3.39 1.25 12.38 0.20 Non-metallic minerals 1.73 1.85 3.22 3.27 10.45 10.16 Synthetic textiles 1.15 1.14 1.82 1.95 3.03 3.61 Transport equipment 2.05 2.17 3.52 3.44 13.44 13.27 Wood and paper 2.21 1.13 3.98 1.48 16.70 1.04
Source: Same as above; CV: Coefficient of variation.
Economic and Political Weekly September 30, 2006 pattern of firms in one of the industries, the synthetic textiles.23
When we proceed to draw the transition matrix of firms, frequency distribution of firms by decilewise size-class is estimated for the end points, i e, 1988-89 and 2000-01. Each size-class is taken as double the previous size-class in order to have geometric intervals for analytical reasons. This helps to capture not only substantial mobility but also enable sharper and clearer interpretation of size and growth. It may be noted that we have taken 5 per cent as a separate size class (though strictly it does not fall within the decile definition) considering the size distribution of firms in our sample. While this analysis basically follows Hart and Prais,24 Singh and Whittington,25 etc, we have improved upon their methodology by taking deciles instead of size-class per se in order to ensure comparability across time.
Firstly, the transition matrix (TM) shows whether the sizedistribution of firms has changed or not during the period as indicated by the frequency of the size class given in the last row and in the last column. For example, only in the case of the first size-class, the frequency remains the same. Secondly, TM provides the number of firms moved upward/downward and/or remained in the same size-class. The diagonal element shows the number of firms without inter-class mobility during the period. The left of the diagonal in any row indicates downward interclass mobility and the right of it upward inter-class mobility. It may be noted that in the first row only upward mobility is possible since it begins with the diagonal element. Similarly, in the case of the last but one row only downward mobility is possible since it end with the diagonal element. As a result, both upward and downward mobility can be captured only in the rows between the first and the last. For example, all the firms in the size-class (first decile-class, second row) in 1988-89 have moved out either to the higher or lower size-class by 2000-01. More precisely, three moved to second decile-class, two to fourth decile-class and the remaining two moved down to =.5 class by the end of the period.
Another noteworthy feature of the mobility analysis, is that it clearly shows that the identity of firms in a size-class can change even when frequency remains the same. This is best brought out from the first size-class itself. The number of firms is six in the first size-class at the beginning as well as at the terminal year. But the transition matrix shows that two out of the six firms in the year 1988-89 moved out, one to the second decile and another to the fourth decile. Another two firms from the second decilemoved into the first size-class leaving the size-class frequency constant. Thus, the six firms in 2000-01 are not the same six in 1988-89. This changing identity can happen in any size-class even when there is no change in the frequency within the size-class. The nature of this problem for the remaining industries will be examined later. In addition to the identity of firms in the sizeclass, the mobility analysis has implication for the market structure. It depends on the nature and direction of mobility. If mobility is only in one direction, say upward, then industry is getting more and more concentrated. If it is on either direction, upward and downward, the firms move towards an efficient scale according to traditional theory of firm under profit maximisation. If the mobility is intra-class but not inter-class and restricted to smaller classes, then it indicates the features of dominant-firm-with-fringe competition (DFFC). The DFFC model can be illustrated using the transition matrix of food products industry given in Table 7.
The mobility of firms is concentrated in the left hand top corner of the matrix and within the first two deciles (10 and 20) of the size distribution. Size-rank test of the industry shows that competition among firms in the smaller classes is high but low in the upper deciles. This is also statistically valid.26 This is an
Table 6: Transition Matrix of Firms in Synthetic Textilesby Size-class, 1988-89 to 2000-01
Size-Class <5 >5-<10 >10-<20 >20-<40 >40-<80 >80 No of (Decile-Firms
wise in Size-class 1988-89 (Decilewise)
< 5 41 1--6 >5< -<10 2-3 2 --7 >10 -<20 -2 ----2 >20 -<40 ---1 1 13 >40 -<80 --1 1 1 -3 >80 -----11 No of firms
in 2000-016 3 4 5 2 2 22
Source: Same as above.
Table 7: Transition Matrix of Firms in Food Products by Size-class, 1988-89 to 2000-01
Size-Class <5 >5-<10 >10-<20 >20-<40 >40-<80 >80 No of (Decile-Firms
wise in Size-class 1988-89 (Decilewise)
< 5 832 -1-14 >5< -<10 33 -1 2 -9 >10 -<20 32 4 ---9 >20 -<40 -1 1 1 1 -4 >40 -<80 --1 1 2 -4 >80 -----11 No of firms
in 2000-0114 9 8 3 6 1 41
Source: Same as above.
Table 8: Frequency Distributions by Size-class and by Industry,1988-89 to 2000-01
Industry Year <2nd Decile >2nd Decile <4th Decile and >4th Decile
Beverages and 1988-89 92.9(13) -7.1(1) tobacco 2000-01 92.9(13) -7.1(1) Cotton and blended 1988-89 98(50) -2(1)
textiles 2000-01 98.1(50) -1.9(1) Drug and 1988-89 56.7(17) 33.3(10) 10(3) pharmaceuticals 2000-01 86.7(26) 3.3 (1) 10(3)
Electrical machinery 1988-89 80.8(38) 12.8(6) 6.4(3) 2000-01 74.5(35) 10.6(5) 14.9(7) Electronics 1988-89 90(27) 3.3(1) 6.7(2) 2000-01 80(24) 3.3(1) 16.7(5) Food products 1988-89 78(32) 9.8(4) 12.2(5) 2000-01 75.7(31) 7.3(3) 17(7) Iron and steel 1988-89 96(24) -4(1) 2000-01 88(22) 4(1) 8(2) Metal products 1988-89 63.6(14) 13.6(3) 22.7(5) 2000-01 68.2(15) 13.6(3) 18.1(4)
Non-electric machinery 1988-89 96.6(57) 1.7(1) 1.7(1) 2000-01 96.6(57) 1.7(1) 1.7(1) Non-ferrous metal 1988-89 81.3(13) 12.5(2) 6.2(1) 2000-01 62.5(10) 12.5(2) 25(4) Non-metallic minerals 1988-89 91.6(44) 2.1(1) 6.3(3) 2000-01 89.5(43) 4.2(2) 6.3(3)
Synthetic textiles 1988-89 68.3 (15) 13.6 (3) 18.1(4) 2000-01 59.1(13) 22.7(5) 18.2(4) Transport equipment 1988-89 88.8(64) 5.6(4) 5.6(4) 2000-01 89(64) 4.2(3) 7(5) Wood and paper 1988-89 90(18) 5(1) 5(1) 2000-01 55(11) 25(5) 20(4)
Note: Figures in the parenthesis are the number of firms in each size-class by industry.
Source: Same as above.
Mobility Upward
Figure: Mobility and Ijiri-Simon Index of Competition, All Industries (2000-01 to 1999-2000)
Figure 1(a)
Correlation (0.30) (not significant)
0.8
0.6
0.4
0.2 0
0.000 0.010 0.020 0.030 0.040 0.050 Average Shifting (Ijiri-Simon Index)
Figure 1(b)
Correlation (0.35) (not significant)
Figure 1(c)
Correlation (0.55) (significant at 5 per cent level - 2 tailed)
drug and pharmaceuticals and non-metallic minerals) there is change in the identity of firms. In these industries, the conventional concentration ratios (N firm market shares) could remain the same but mobility has occurred. This clearly establishes the superiority of the new turnover index over the market share or N firm concentration ratio. It provides important insights on interand intra-class mobility which alone can give an accurate picture of the nature of competition in an industry.28
The second issue, the most important from the industrial organisation theory, is the insight given by mobility analysis on the market structure and size-distribution. Although there is no integration of the two in the traditional literature, which is still an unresolved issue as far as industrial economists are concerned, one model that integrates the two is the dominant-firm-withfringe competition (DFFC). In such models, the size-structure will be skewed to the right and with high intra-mobility within the smaller size-classes but very little inter-mobility between the small and the large-classes. One industry that has such market structure is the food products as demonstrated in an earlier paper.29 It may be noted that under u-shaped average cost curves, competitive market structure should have both upward and downward mobility. Since all such interesting patterns of mobility cannot be put down for each industry, we provide a summary of the mobility patterns of all the industries during the period in Table 9.
Table 9 provides estimates on the percentage of firms moving in the upward direction, downward direction and remaining in the same size-class by industries during the period. It may be noted that on an average 54 per cent of the firms remain in the same class, 24 per cent move upward and 21 per cent downward in the manufacturing sector. Further, the highest percentage of firms remaining in the same class is in the non-electrical machinery (86 per cent) with cotton and blended textiles and trans
1.00
Mobility Total
0.80
0.40
0.20
and paper whereas in the case of downward mobility it is in drug
0.00
and pharmaceuticals. Beverages and tobacco does not show any
0.000 0.010 0.020 0.030 0.040 0.050
upward mobility and non-electrical machinery indicates very
Average Shifting (Ijiri-Simon Index)
little downward mobility. so that main features of dynamic competition can be captured. If the competitive pressure were high in an industry then it
The first feature is the changing nature of the size-structure would have higher mobility in either direction under U-shaped of industries. It may be noted that the class-interval has been cost-curves in the long run. This implies that the measures of reduced to three instead of the six in the earlier transition matrices competition and mobility should be positively related among the due to inadequate number of firms within each size-class. However,
Table 9: Mobility of Firms by Industries, 1988-89/2000-01
here again each decileclass is taken as double the previous
Mobility across size-classes
decileclass in order to have geometric intervals, as done in the
Industry No Mobility Upward Downward
transition matrices. As stated earlier this helps to capture not only
1 234
substantial mobility but also enables sharper and clearer inter-
Beverages and tobacco 0.64 0 0.36
pretation of size and growth. The change in the size distribution
Cotton and blended textiles 0.82 0.13 0.04
of firms arising out of the transition process27 for the end points Drug and pharmaceuticals 0.13 0.07 0.80
is shown in Table 8. Electric machinery 0.45 0.32 0.23 The frequencies of all size-classes have changed in four in-Electronics 0.70 0.23 0.07
Food products 0.46 0.24 0.29
dustries (electrical machinery, food products, iron and steel, and
Iron and steel 0.72 0.24 0.04wood and paper). It is to be observed that in six industries out Metal products 0.36 0.23 0.41
of 14, that is for about 43 per cent, there is no change in the Non-electrical machinery 0.86 0.12 0.02 Non-ferrous metal 0.50 0.44 0.06
number of firms in the large size-class. This does not however
Non-metallic minerals 0.46 0.33 0.21
rule out the mobility of firms from the largest decile to other
Synthetic textiles 0.32 0.41 0.27 deciles and vice versa. That is there can be the identity changes Transport equipment 0.82 0.05 0.13
as observed earlier in the case of synthetic textiles. Of the six Wood and paper 0.35 0.60 0.05 Average for all industries 0.54 0.24 0.21
industries with the same number of firms in the largest size-class (> 80) during the period, in three industries (synthetic textiles, Source: Same as above.
0.60
port equipment showing the same high levels of immobility (82
Economic and Political Weekly September 30, 2006 industries. Three versions of the same hypothesis can be formulated if the average cost curves is not of the conventional type. They are: (1) upward mobility alone is positively related to competition; (2) downward mobility is alone positively related to competition; and (3) both upward and downward mobility are positively related to competition.
Empirical evidence is reported in Figure 1(a), 1(b) and 1(c) with correlation coefficients. Note that the competition is measured by Ijiri-Simon index for the endpoint (1999-2001) and the mobility of firms for the entire period. The results show that only the correlation coefficient between total mobility index (upward+downward) given in Table 9 and competitive index is statistically significant. In other words, more competition prevails in industries with mobility in either direction; otherwise it is not. This proves that as mentioned in the literature, it is the mixing and reordering of firms that constitutes vigorous/effective competition.
The analysis implies that public policy for increasing competition should address ways and means of removing mobility barriers in the manufacturing sector.
IV Summary and Conclusions
One important expected outcome of reforms is to increase competition and efficiency in the economy. While several studies have dealt with the efficiency aspect, there are not many studies on competition. This study tries to fill the gap. Competition being multidimensional in nature needs to be looked at from different angles. In the literature the mobility of firms is identified as one important dimension – a dynamic one, of what constitutes effective competition. At the same time it is also not without criticism. The main criticism against the turnover index (the rank ordering of size) is that it only shows whether the size of a firm is higher or lower than another firm but not by how much. Therefore, the stability of the ranks may not capture the extent of competition. The second criticism is that the changes in market shares which is the really significant phenomenon for understanding the market does not get reflected in the rank analysis.
The contention of this paper is that if the limitations of the turnover index are taken care of, it can provide several important insights on competition, which cannot be captured by other indicators, like the concentration ratio and is therefore an important tool for public policy.
In this paper, we provide an alternate turnover index, which overcomes the limitations of the traditional index through an order preserving transformation of the data provided by CMIE (Prowess) on 14 major industries for the period, 1998-89/200001.The new index is tested for rank shifts of 14 industries. The stability between the ranks of the industries for the end points indicates that 43 per cent of the industries have very low mobility. The change in competition examined through Ijiri-Simon index indicating average shifting among firms belonging to 14 industrial groups, shows that 50 per cent of the industries have shown an increasing trend. But the rank correlation of change in competition or average shifting among the industries over the period does not show any shift in their relative positions during the period. This implies that some rigidity exists in the expansion of competitive forces in the manufacturing sector.
The changing nature of the size structure is examined by the skewness and kurtosis of the distributions. The frequencies of all size-classes have changed in four industries (food products, iron and steel, electrical machinery and wood and paper). It is observed that in six industries out of 14, about 50 per cent, there is no change in the number of firms in the large size-classes. This does not however rule out the mobility of firms from the largest to other smaller classes and vice versa. In other words, the identity of firms in a class can change even when the frequency
SPECIAL ISSUE
INTEGRATED CHILD DEVELOPMENT SERVICES August 26, 2006
Infant Survival: A Political Challenge — Shantha Sinha Decentralised Childcare Services: The SEWA Experience — Mirai Chatterjee Food Dole or Health, Nutrition and Development Programme? — Shanti Ghosh Infant and Young Child Feeding: An ‘Optimal’ Approach — Arun Gupta Hidden Hunger: The Problem and Possible Interventions — Tara Gopaldas Universalisation of ICDS and Community Health
Worker Programmes: Lessons from Chhattisgarh — T Sundararaman Implementation of ICDS in Bihar and Jharkhand — Nandini Nayak, Naresh C Saxena Tamil Nadu: ICDS with a Difference — Anuradha Khati Rajivan Rethinking ICDS: A Rights Based Perspective —Dipa Sinha Chhattisgarh: Grassroot Mobilisation for Children’s Nutrition Rights —Samir Garg
For copies write to Circulation Manager
Economic and Political Weekly
Hitkari House, 284, Shahid Bhagatsingh Road, Mumbai 400 001 email: circulation@epw.org.in
does not change. Of the six industries with same number of firms in the largest size-class (> 80) during the period, three of them (drug and pharmaceuticals, non-metallic minerals and synthetic textiles) exhibit the change in the identities. In these industries, it may be noted, the concentration ratios could remain the same and yet mobility could occur indicating competition.
The inter- and intra-class mobility of firms are examined for its implication for market structure using transition matrices. The analysis gives the following conclusions. On an average, 54 per cent of the firms remain in the same class, 24 per cent move upward and 21 per cent downward in the manufacturing sector. Further, the highest percentage of firms remaining in the same class is in the non-electrical machinery (86 per cent) with cotton and blended textiles and transport equipment showing the same high levels of immobility (82 per cent). The highest upward mobility has occurred in wood and paper whereas in the case of downward mobility it is in drug and pharmaceuticals. Beverages and tobacco does not show any upward mobility and non-electrical machinery indicates very little downward mobility. The study also links up patterns of mobility with market structure. It illustrates the existence of varied types of competition in terms of inter- and intra-class mobility, which cannot be captured by the traditional concentration ratios. It shows that competition is not associated with either upward mobility or downward mobility but only with upward and downward mobility.
This study clearly establishes that the mixing and reordering, upward and downward mobility, of firms in an industry, is an important dimension of what constitutes effective competition
– a dynamic dimension. Any public policy on increasing competition in the manufacturing should give top priority in understanding mobility barriers and suggest ways of eliminating such barriers.
EPWEmail: pushpangadan@cds.ac.in
Notes
[The authors are grateful to K K Subrahmanian, P Mohanan Pillai and U S Mishra and to the referee for their valuable comments. We also thank V Jayachandran for his excellent research assistance. Of course, the usual caveat applies.]
1 See Goldar (2004), Krishna (2004), Balakrishnan et al (2002), Suresh
Babu (2002), and Balakrishnan and Pushpangadan (1998), among others.
2 Joskow (1960), pp 113-116.
3 See Shepherd (1975), Scherer (1979).
4 Baldwin (1998), Ijiri and Simon (1977), Boyle and Sorenson (1971), Singh
and Whittington (1968), Gort (1963) Joskow (1960), Simon and Bonini
(1958).
5 Singh and Whittington (1968), p 94.
6 Ijiri and Simon (1977), p 14.
7 Baldwin (1998). He however adds that there can be instances when there
is intense competition between firms with no change in relative position.
He however treats such instances as exceptions, and adds that it is unlikely
that there are a large number of intense struggles occurring in which no
winner emerges (p 4).
8 For empirical evidence, see, Pushpangadan and Shanta (2005).
9 For an illustration see Pushpangadan and Shanta (2004). 10 Scherer (1979). 11 Joskow (1960), Simon and Bonini (1958), Hart and Prais (1956). 12 Hymer and Pashigian (1962). 13 Ibid. 14 Joskow (1960), Simon and Bonini (1958). 15 CMIE, Prowess (1997, 2001). 16 Sutton (1997), Curry and George (1983), Shalit and Sankar (1977). 17 Hymer and Pashigian (1962), p 83.
18 Ibid.
19 For details, see Hymer and Pashigian (1962), p 85; for Indian evidence, see, Pushpangadan and Shanta (2005), mimeo.
20 Hart and Prais (1956).
21 Singh and Whittington (1968).
22 Ijiri and Simon (1977).
23 Similar transition matrices have been calculated for all the 13 industries, the results are not reported due to space limitation.
24 Hart and Prais (1956).
25 Singh and Whittington (1968).
26 See Pushpangadan and Shanta (2004).
27 These transition matrices are available with authors upon request.
28 For a detailed discussion of the superiority of mobility analysis over market share analysis (concentration ratio), see Baldwin (1998), pp 174-175.
29 See Pushpangadan and Shanta (2004) for an empirical test.
References
Balakrishnan, P, K Pushpangadan (1998): ‘What Do We Know about the Productivity Growth in Indian Industry?’ Economic and Political Weekly, August, pp 2241-46.
Balakrishnan, P, K Pushpangadan and M Suresh Babu (2002): ‘Trade Liberalisation, Market Power and Scale Efficiency in Indian Industry’, Centre for Development Studies, Working Paper Series, No 336, Thiruvananthapuram.
Baldwin, R John (1998): The Dynamics of Industrial Competition, a North American Perspective, Cambridge University Press, UK.
Basu, K (1993): Lectures in Industrial Organisation Theory, Blakwell Publishers, Oxford, UK.
Boyle, S E and R L Sorenson (1971): ‘Concentration and Mobility: Alternative Measures of Industrial Structure’, Journal of Industrial Economics, XIX No 2, April, pp118-32.
CMIE (1997): Prowess Users’ Manual, Vol 1, Mumbai.
– (2001), Prowess (Electronic Data), Mumbai. Collins, N R and L E Preston (1961): ‘The Size Structure of the Largest Industrial Firms 1909-1958’, American Economic Review 51, December,pp 986-1011. Curry, B and K D George (1983): ‘Industrial Concentration: A Survey’, Journal of Industrial Economics, Vol XXI, No 3, pp 203-55.
Goldar, Bishwanath (2004): ‘Indian Manufacturing: Productivity Trends in Pre- and Post-Reform Periods’, Economic and Political Weekly,November, pp 5033-43.
Gort, M (1963): ‘Analysis of Stability and Change in Market Share’, Journal of Political Economy, 71, February, pp 51-63.
Hart, P E, and S J Prais (1956): ‘The Analysis of Business Concentration’, Journal of the Royal Statistical Society, CXIX, Pt 2, A119, pp 150-81.
Hymer, S and P Pashigian (1962): ‘Turnover of Firms as a Measure of Market Behaviour’, Review of Economics and Statistics, Vol XLIV, pp 82-87.
Ijiri, Y and H A Simon (1977): Skewed Distributions and the Size of Business Firms, North Holland Publishing Co, Amsterdam.
Joskow, J (1960): ‘Structural Indicia: Rank-Shift Analysis As a Supplement to Concentration Ratios’, Review of Economics and Statistics, Vol XLII, pp113-16.
Krishna, K L (2004): ‘What Do We know about Sources of Economic Growth in India?’ National seminar on Planning and Development: Institutions and Markets, Dept of Economics, St Thomas College, Thrissur, Kerala.
Pushpangadan, K and N Shanta (2005): ‘An Order Preserving Turnover Index and Market Behaviour’, Centre for Developemnt Studies, Thiruvananthapuram, mimeo.
Scherer, F M (1979): Industrial Market Structure and Economic Performance, Rand McNally and Co, Chicago.
Shalit, S and U Sankar (1977): ‘The Measurement of Firm Size’, Review of Economics and Statistics, Vol LIX, No 3, pp 290-98.
Shepherd, W G (1975): The Treatment of Market Power: Antitrust, Regulation, Public Enterprise, Columbia University Press, New York.
Simon, H A and C P Bonini (1958): ‘The Size Distribution of Business Firms’, American Economic Review, XLVIII, pp 607-17.
Singh, A and G Whittington (1968): Growth, Profitability and Valuation, Cambridge University Press, London.
Suresh M Babu (2002): ‘Economic Reforms and Entry Barriers in Indian Manufacturing’, Centre for Development Studies, Working Paper Series, No 331, Thiruvananthapuram.
Sutton, J (1997): ‘Gibrat’s Legacy’, Journal of Economic Literature, Vol XXXV, pp 40-59.
Economic and Political Weekly September 30, 2006
