Has cement manufacturing across states in India become more equitable after decontrol?

Mukhopadhyaya, Jayanta Nath ; Roy, Malabika ; Raychaudhuri, Ajitava 等

INTRODUCTION

We have chosen the cement industry in India to carry out this empirical analysis due to several reasons. Firstly, the Indian cement industry is large, even by global standards. The Indian cement industry, which was the 4th largest in terms of production in 1998 (after China, Japan and USA) had climbed to the second position by 2000 and now stands only after China. Secondly, there are a large number of firms (more than 50) with the structure of the cement industry being fragmented. Although the Indian cement industry is fragmented with a large number of players, yet general reports would mention that there is dominance of the top firms and the structure of industry is oligopolistic. It can be seen that the large cement plants account for more than 90% of the total installed capacity in India. There have been few detailed studies of the cement industry in India. Moreover, the existing published studies have not, in general, focused on the post 1980s period, during which industries experienced deregulation and liberalisation.

In this paper we study how the distribution of cement industry has changed across states after liberalisation due to decontrol of the industry. This is analysed in terms of changing pattern of inequality as well as concentration for the period 1989-2006. We also provide an economic rationale behind such growth patterns and formulate some policy lessons.

The geographical distribution of industry has significance for economic efficiency, therefore, a study of the changes in inequality and concentration at the states level is important. India being a very large country, there are many states which are similar in size to a small country in Europe. As India follows a system of federalism where the State governments enjoy considerable authority in framing regulations relating to trade, commerce and industry, we feel it is important to capture the local factors by disaggregating and analysing their workings at the state level.

In the present paper, we not only examine the concentration and inequality of cement capacity across different states of India, but also identify the significant explanatory variables which can explain the growth of cement sales in the different states.

This study is relevant for several reasons. As few detailed studies have been done on the cement industry of India, the findings of this study shall have policy implications because the geographical distributions of firms play a significant role in economic efficiency. Moreover, the infrastructure and housing sector, which are the two very important sectors of the Indian industry, have very strong linkages with the cement industry.

LITERATURE REVIEW

There are not many comprehensive studies of major individual industries. So far as the Indian cement industry is concerned, there are only a few detailed studies.

In a pioneering work in India, Gokarn and Vaidya (1993) made an attempt to evaluate the performance of the cement industry after decontrol and found that the structure of the industry had undergone a qualitative change. The industry, after decontrol, was characterised by a competitive outcome in terms of price and profit performance. The nature of interfirm heterogeneities suggests that major differences arose out of technical factors. The performance of firms across strategic groups is different, with firms operating relatively new and large plants appearing to have an advantage. Gokarn and Vaidya (1994) carried out a firm level analysis in the StructureConductPerformance format of the modern theory of industrial organisation. The discussions suggested that the structural variables in the cement industry have been far from stable after 1982. The concentration ratio has been changing, the technology has undergone a sea change, and the nature of government controls faced by the industry has also been changing. Due to the instability of the structural variables, there seems little chance of drawing causal links between structure and performance.

Apart from this pioneering study, there has been a set of studies on Indian cement industry throwing light on different aspects of the industry. In another study by Jha et al. (1991) analysis was done at the aggregate level and covered the period 1960-61 to 1982-83. They found that that this industry is characterised by a large and allocative efficiency.

Pradhan (1992) had raised doubts about the success of the decontrol of the cement industry because he found that the rate of decrease in concentration fell after the decontrol.

Mukhopadhyaya et al. (2007) had found in their study, which focused on the period 1989-2003, that the capacity addition per year in the cement industry had accelerated during the period of partial decontrol and increased further during the period of total decontrol. The four firm concentration which initially had started to go down after the period of decontrol, started going up again from the year 2000. The quartile-wise analysis had revealed that the firms in the top two quartiles had gained market shares at the expense of the firms in the lower quartiles.

Mukhopadhyaya et al. (2012) had studied some aspects of the cement Industry by dividing India into four regions. The study, focusing on the period 19892006, found that the change in concentration and inequality after liberalisation in the cement industry of India at the regional level has been significant.

The literature on regional growth pattern of industries is scanty as well as scattered. Krugman (1991) had developed a model to explain the economic rationale of geographical concentration or dispersion of industries, which focused on the impact of interaction of supply and demand side factors on locational choice.

The study by Ellison & Glaeser (1999) provided a strong reaffirmation of the previous wisdom and found almost all industries to be somewhat localised. They also found that some of the most extreme cases of concentration were likely due to natural advantages. However, in many industries, they found the degree of localisation to be slight and found a great deal of heterogeneity still remained to be explored.

Fujita et al. (2001) identify some market and nonmarket forces, which they categorise as centrifugal and centripetal. The interplay of market forces drives or spreads economic activity away from the urban centres.

Wen (2004) did an interesting study on the relocation and agglomeration of Chinese manufacturing industry from 1980 to 1995. He primarily used the Gini coefficient and panel regression as the analytical econometric tools.

This paper concludes that manufacturing industries have become more geographically concentrated following the economic reforms.

We did not find any study on state level growth pattern of industries in the Indian context. So, we feel that our study will address an important lacuna in the existing literature on Indian industry and will encourage further investigations.

METHODOLOGY

The Cement Manufacturers' Association (CMA) is the principal source of data for this paper. We have mainly used their annual publications named the "Cement Statistics" for the period 1989-2006 to collate our data. We also had discussions with their officials to get an insight into the data.

CMA was publishing data for nineteen states in 1989, the starting point of our study period. During the period under study (1989-2006), there was a change in the categorisation of states reported by CMA. Two new states were created: Chattisgarh and Jharhkhand, and CMA has been reporting their data since 2001. However, in order to maintain consistency in our analysis we continued our analysis with the earlier classification of nineteen states. Jharkhand was earlier a part of Bihar, so its data has been added back to that of Bihar. Similarly Chattisgarh was earlier a part of Madhya Pradesh, hence its data has been added back to that of Madhya Pradesh.

In this paper, we analyse the changes in inequality and concentration from the years 1989 to 2006 at the inter-state level. We study the trends of inequality and concentration across the nineteen states, based on the shares of the capacity of these states. The capacity of the respective state divided by All India capacity gives the capacity share of the state. Needless to mention, all states do not have equal capacity and their growth rates over the years are also different, which gives rise to inequality and concentration. As there are a large number of states, we break them up into quartiles and study the trend and characteristics of these quartiles separately. We study these trends, particularly focusing on the beginning year of our study--1989, the terminal year of our study--2006 and the median year of our study--1997. We study the inequality of the cumulative production versus cumulative capacity at the Inter states level as well.

To get a measure of concentration, we use the Herfindahl index (H index). It has the merit of combining information about the shares of all states, not just the largest states. If there are n states in a country and Si is the market share of state i, the H index is measured as follows:

H = [n.summation over (i=1)][S.sub.i.sup.2]

Herfindahl index going up would imply that concentration is going up at an inter-states level and vice versa. (1)

To study inequality, we use the Gini coefficient. Gini coefficient (G) is calculated as:

G = [[summation].sup.n.sub.i=1][[summation].sup.n.sub.j=1][absolute value of ([x.sub.i] - [x.sub.j])]/2[n.sup.2][mu]

Where [x.sub.i] is the capacity of the ith state, [x.sub.j] is the capacity of the jth state, n is the number of states and [mu] is the mean value of the capacity levels. Gini is the "relative mean difference" i.e., the mean of the absolute value of differences between every possible pair of individual states, divided by the mean size [mu].

We also study the inequality of the cumulative production versus cumulative capacity at the interstate level. For this purpose, we use a different concept of Gini coefficient, based on the Lorenz curve. If the Lorenz curve is represented by the function Y = L(X), where Y and X represents the cumulative share of production capacity from the lowest to the highest states, respectively. Using integration, the value of G can be found as:

G = 1 - 2[[integral].sup.1.sub.0]L(X)dX

Using the Gini coefficient, we study trend of inequality of the cumulative production versus cumulative capacity. The first measure of Gini was telling us how unequal is the distribution of capacity in the states (without any reference to production from those states). This second measure would tell us how unequal the distribution of relative production is from the available capacities.

Further, to confirm whether the change in trends of the Gini coefficient or Herfindahl index are significant or not, we regress the indices against time. To analyse the growth patterns of these indices, we use a log-linear model with log of indices as the dependent variable and time as the independent variable. The coefficients of the time variable would indicate the direction of growth. We also run the Cusum square test to see if there are any structural breaks.

In the last section, we carry out a regression analysis to identify the explanatory factors that have led to the particular pattern of growth of cement sales at the state level for the industry. A dataset is formed with the sales quantity of cement for each state being taken as the dependent variable and explanatory variables being taken as the independent variable. Therefore, we have both a cross-section and time series data leading to a panel dataset. We carry out a panel regression analysis to identify the significant factors affecting the distribution of cement industry across states.

DISCUSSIONS

1. Distribution of capacities across different states and their trends

In this section, we take the different state's share of cement capacity in India and study their trends. Table 1 presents the movement of these capacities in the beginning year 1989, the median year 1997 and the terminal year of our study 2006.

As there have been changes in the capacities of most states, we break up the states into quartiles. The state-wise break up into the quartiles is reported in table A.4 (see Appendix) and the quartile wise analysis is presented in Table 2.

If we see the overall movement between 1989-2006, we find that three quartiles viz., the top quartile, the second highest quartile and the lowest quartiles had lost shares. However, an interesting thing is that if we look at the interim period 1989-1997, three of the four quartiles including the top and the bottom quartile had actually gained shares. So, the top quartile had started losing its share only during the period 1997-2006, when the second quartile had actually made a gain.

We find that the states of MP and AP retain the top two slots held in 1989, but with a reduced all India share of capacity in 2006. It is interesting to see that the capacity share of MP actually increased in the intermittent period from 1989-97 before losing significantly from 1998 till 2006. The second largest state AP initially lost considerable share before going up. The other two large states Rajasthan and Gujarat in the top quartile gained relative shares during this period. Again, the smallest state in the top quartile, Tamilnadu gained shares during this period. The combined impact is that though the top quartile gained relative capacity share during 1989-97, it lost significant shares during 1997-2006 and the overall impact for the full period is that the top quartile has lost shares. (Table A.5 shows the movement of states across quartiles.)

If we take the second largest quartile as per 1989 share, then all the five states viz., Karnataka, Maharashtra, UP, Bihar and Orissa lost share for the period 1989-2006, though Maharashtra and Bihar had gained share in the interim period 1989-97. In the third largest quartile, i.e., between 1st and 2nd quartile, two states viz., Himachal Pradesh and West Bengal added significant share during this period. This has helped this quartile to be the only quartile which has gained share not only for the overall period but also during the two interim periods.

The smallest quartile in 1989 comprises the five states of Delhi, J&K, Punjab, Assam and Meghalaya. Except Punjab, all states stagnated and had a share of less than 0.5% in 2006. However, Punjab added significant capacity during this period and its share went up from 1.2% in 1997 to 2.7% in 2006. To understand the impact of these changes in inequality and concentration, we study the trends of Gini coefficient and Herfindhal index in the subsequent sections.

2. Inter State inequality and concentration

The trend of inter-state inequality is presented in figure 1.

There is a downward trend in Gini which fell from 0.586 in 1989 to 0.559 by 2006, a fall of 4.6%. It is also interesting to see that the inequality rose in the initial years before its downward journey. It rose to 0.602 by 2000 before its secular downward trend till 2006, a drop of 7.1%.

As discussed earlier, the significant drop in capacity share of the largest state in cement capacity-MP and limestone rich deposit states like AP, Karnataka and simultaneously, the increase in capacity share of relatively much smaller states like Punjab, W. Bengal have led to the reduction in inequality.

[FIGURE 1 OMITTED]

The regression of the log of Gini coefficient of interstate capacity with time is reported in Table 3. It gives a significant relationship with a negative coefficient, clearly showing a significant downward trend with time.

The Cumulative sum of square (Cusumsq) test for the Gini of Inter-State capacity is performed to check for structural breaks. The plot of the cumulative sum of squares of recursive residuals shows breaks in 2000, 2001 and 2002 at 5% significance level, as shown in Figure 2. On a deeper study to find an economic reason for these structural breaks, we found that the apprehension regarding possible withdrawal of state level tax incentives in the late 1990s could have been a significant factor. The Government was considering Sales Tax (ST) reforms by replacing ST by a system of Value Added Tax (VAT) as early as 1995. Finally, a committee of the State Finance ministers was constituted in November 1999 by the Union government, to examine all aspects of sales tax reform, including the introduction of VAT. Since this was expected to be a major reform, having significant impact on the revenue of the states, the discussions were expected to be protracted. One of the important decisions taken in 1999 was that the sales-tax-related industrial incentive schemes would also have to be discontinued with effect from January 1,2000. (2)

[FIGURE 2 OMITTED]

In anticipation of the withdrawal of tax incentives, the manufacturers possibly went for large scale capacity expansion to avail of the incentives before they were withdrawn. This could explain the structural break in capacity. We study the change in inter-state concentration through the Herfindahl Index, presented in Figure 3.

[FIGURE 3 OMITTED]

The Herfindahl Index shows a similar downward trend. It has fallen from 0.122 in 1989 to 0.110 by 2006, a fall of 9.8%. In the initial years, it moved in a narrow band and then it moved continuously downward after reaching 0.129 in 2000, a drop of 14.7% from its peak. We find a similar trend in both the Inter State concentration and inequality. In both the cases it rises till 2000 before starting a secular downward trend till 2006. This resembles an inverted U curve as seen in figure 3. However, the drop in concentration is much more pronounced than the drop in inequality. The drop in inequality for the entire period is 4.6% whereas the drop in concentration is much higher at 9.8%. From the peak attained in 2000, the drop in inequality is 7.1%, whereas the drop in concentration is 14.7%.

To understand this better, we study the quartile wise change in shares for the two interim periods 1989-2000 and 2000-2006, as reported in Table 4. We find that from 2000, when the secular downward trend started, it is the top quartile which lost significant share and the second and third largest quartile gained shares. This leads to a drop in inequality but the drop in concentration is more as the top quartile is losing significant share. Going deeper, this could be explained by the most dominant state Madhya Pradesh loosing significant capacity share.

We find that at the inter-states level, the drop in concentration is more than the drop in inequality. An explanation of the drop in concentration and inequality in capacity, over the period, could be explained by the increase in Split location plants, which, enables cement manufacturers to pursue a hybrid locational strategy. There are two locational strategies which are usually followed by the cement manufacturers. The first strategy is to locate the plant close to the mineral deposits, so as to minimise raw material assembly costs. Given that 1.4-1.5 tonnes of limestone are required per tonne of clinker, locating the plant along the limestone deposits is the logical corollary. The second strategy is to locate manufacturing facilities near the consuming centres. In this case, outward freight is minimised and marketing flexibility enhanced at the cost of higher raw material assembly costs.

The bulk of the cement manufactured is consumed near urban centres. In the manufacture of cement, for every 1 tonne of clinker, about 1.6-1.7 tonnes of limestone and coal need to be assembled. For OPC, another 50 kg of gypsum is required while grinding the clinker down. For PPC, up to another 250 kg of pozzolonic material, such as fly ash, requires to be assembled. Thus, there can be two broad locational strategies stemming from the principal objective, which is not merely to minimise unit-manufacturing cost, but to minimise unit delivered cost as well.

As long as retention prices were the norm (3), before decontrol by the Government, outward freight was of no concern to cement companies. All the cement plants thus naturally gravitated to one of the several large limestone bearing areas in the country. With the introduction of partial, and later full, decontrol, outward freight has become a critical issue in determining a company's profitability.

A hybrid strategy exists with Split location plants becoming a strategic option, with the clinker manufacture near limestone deposits and grinding and bagging facilities near the consuming centres. The advantages of this split location strategy derive from the ease of transporting clinker in open-to-sky condition (rather than bagged cement under protective cover), lower handling losses in transit and ease of storage of clinker (as opposed to cement at the market centred grinding mills). This is especially true for blended cement like PPC7PBFS, since fly ash/slag is available from the thermal power stations/steel plants, which are located in and around the country's urban centres. Fly ash disposal by power utilities has become a contentious environmental issue.

[FIGURE 4 OMITTED]

Similarly, steel producers face problems in disposing slag. Therefore, utilisation of these materials in this manner can improve the cement company's profitability while benefiting the environment. By locating such grinding units close to the markets, the distribution costs are reduced to a great extent. (4) Split location cement capacity can come up in states, particularly Punjab, W. Bengal etc. which hardly have any limestone deposits, significantly reducing the concentration and inequality across India. (5)

The regression of the log of the Herfindahl index of inter-state capacity with time gives a significant relationship at 5% level as reported in Table 3. We get a negative coefficient, clearly showing a significant downward trend with time.

The Cusumsq test for the Herfindahl index of interstate capacity is done to check for structural breaks. The plot of the cumulative sum of squares of recursive residuals shows breaks in 2000, 2001 and 2002 at 5 % significance level, as shown in Figure 4.

This is similar to the Cusum square graph of Inter State Capacity (Gini), which shows breaks from 2000 to 2002. We can see that the year 2000 to 2002 is common in both the cases. As mentioned earlier, while discussing the structural breaks in Gini of Inter State Capacity, the Government was discussing and deciding to withdraw the Sales Tax incentives in the late 1990s. In anticipation of the withdrawal of tax incentives, the manufacturers possibly went for large scale capacity expansion to avail of the incentives before they were withdrawn. This could explain the structural breaks.

3. Gini Coefficient Inter State Production versus Capacity

In the previous section, we studied the trend of inequality of capacity at the inter states level and did not consider the production parameter separately. In this section we would like to analyse how closely the changes in production are linked with changes in capacity. We are studying 19 states which have considerably different cement capacities and production. One of the important questions we would like to understand in this section is whether the states with smaller capacities are contributing relatively larger or lesser output over the time period of our study. This has economic implications as relatively higher or lower output from smaller states would impact the inequality amongst the states. For this, we use the second concept of inequality using the Lorenz curve as discussed in the methodology section 2. From this analysis, we shall be able to conclude whether the trend is towards a more equitable distribution of production from the states of India or vice versa during the time period of our study.

[FIGURE 5 OMITTED]

The Gini Coefficient of Inter State Production versus Capacity is presented graphically in figure 5.

The Gini coefficient for the cumulative production vs. cumulative capacity figures for the Inter States has been quite high, commencing at 0.643 in 1989 and falling to 0.627 by 2006, a drop of 2.5%. Like the interstate Gini it initially moved within a narrow range before showing an upward trend and reaching 0.669 by 1999 and then dropped by 6.3% by 2006. However, the Gini remains high, which indicates states with larger capacity producing relatively much more than the smaller states.

To understand the reasons for this fall, we look at the individual state's capacity vs. production figures. We define the production by capacity in a particular state as the capacity utilisation in that state. The details of the calculations are presented in Table A.5. The capacity utilisation of relatively smaller cement producing states like UP, Orissa, Bihar, W. Bengal, Kerala and J&K increased significantly over the period 1999 to 2006. Simultaneously, larger cement producing states like MP, AP Rajasthan, Gujarat Tamilnadu also increased their capacity utilisation but at a much lower rate. Consequently, the Gini coefficient for the cumulative production vs. cumulative capacity figures for the Inter States decreases.

The regression of the log of the Gini coefficient of Inter-state production vs. capacity with time results shown in Table 3 indicates it has no significant relationship with time.

4. Identification of explanatory variables of Sales at the State level

In this section, we will identify the key explanatory variables for cement sales in the states. To this end, we carry out a panel regression to identify the factors that have led to a particular pattern of growth and distribution of cement sales at the state level.

In any industry, one of the main factors that crucially drive the growth of sales is the effective supply of inputs. Coal and limestone are the two major inputs for cement production. A specialty of cement is that it is a low value, high volume commodity. Consequently, transportation is another important supply side factor, which will decide on the movement and distribution of sales and may play an important role in deciding the growth of the cement industry. On the demand side, infrastructure and housing construction are the big drivers.

Our study period is from 1989 to 2006. However, we found that the SDP data was not available for a single base year, for this entire period. The National Accounts Statistics generally revises the base year for the data decennially. The series for the base year 1980-81 covers the period from 1980-81 to 1995-96. The new series with 1993-94 as the base year (released by CSO in 1999) covers the SDP data from 1993-94 to 2005, and in many states up to 2006. The practice has been not to make changes every now and then, but necessary major changes are kept for implementation at the time of base year revision exercise. In the past this has been revised decennially to capture the change in the working force estimates from the population census. (6)

The new series has been introduced by CSO after a comprehensive review of both the databases and the methodology employed in the estimation of various macro-economic aggregates. As there have been some major changes in the methodology by CSO while computing the 1993-94 series vis-a-vis the older series, it is observed that in some of the aggregates there are significant differences between the new 1993-94 data series and the estimates based on 1980-81 series. In view of the major changes in the two series, we have carried out our analysis, solely based on the new series i.e., the 1993-94 series. Here the data is available from 1993-94 to 2004-06 and in some cases up to 2004-05. We have extrapolated the data for one year where the SDP data is not there for 2005-06. We have data for nineteen states for thirteen years (1993-94 to 2005-06) i.e. 247 data points for our panel regression and other analyses.

Individual state wise data for coal and lime stone for the entire eighteen year period from 1989-2006 is not available. Similarly, state wise data for transportation and construction is also not available for the period. We, therefore, decided to take proxies for these factors. At the state level, disaggregated data related to State Domestic Product is regularly published by the government and is available for a long time period.

SDP mining (Sdpm) serves as a proxy for coal and limestone factor and is a supply side variable and SDP construction (Sdpc) represents the demand side variable and has been taken as a proxy for the construction. The SDP of transportation, storage and communication are clubbed together in the published data. Since transportation plays an important role in the cement industry and both railways and road transportation are important, we decided to disaggregate the data and consider the SDP-Railways and SDP-Other means of transportation separately as variables. Hence, we take SDP railways (Sdprail) and SDP other means of transportation (Sdpotr) as the other two explanatory variables.

A dataset is formed with the Sales quantity of cement for each state being taken as the dependent variable and Sdpm, Sdpc, Sdprail and Sdpotr for each state being taken as the independent variable. This is done for the period 1994-2006. Therefore, we have both a cross-section and time series data leading to a panel dataset. As we are considering data of 19 states for 13 years, we have a dataset of 247 state-years.

The regression equation that we estimate is:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

where, i represents the ith state, t represents the time period ranging from 1993-94 to 2005-06 and Sales: sales quantity of cement for each state: sdpm: SDP mining of the respective state; sdpc: SDP construction of the respective state; sdpm: SDP railways of the respective state

At the outset, we also ran a multicollinearity test of all the dependent variables. The result reported in A.6 indicates that there is no multicollinearity, as the Variance Inflation Factor (VIF) for all the variables are low. From the results reported in Table 5, based on the high value of Lagrange multiplier test, we go for the FEM/REM over pooled regression model. Also, as we have a low value of the Hausman statistic, we choose the REM over the fixed effect model.

We see that SDP mining and SDP construction are significant with positive coefficient. Positive coefficient implies that increased activity in mining and construction in states, results in higher sales of cement in those states. The surprise element is that SDP Railways factor and SDP (Other Transportation) are not significant. It appears that supply of cement is being ensured in states which have high construction, irrespective of an extensive network of road and rail network in that state. Split technology plants enabled the firms to transport clinker to the demand centres, thereby reducing pressure on transportation. Increased usage of blended cement in India, which is up to 61 % in 2006, (from 27% in 1992-93) also eases the pressure on the transportation dependence for distribution of cement.

CONCLUSIONS

As considerable variation in the capacity of cement exists in different states of India, we have conducted a study at the states level. We have found that the redistribution in capacity has led to a drop in concentration and inequality as seen from the trend of Gini and Herfindahl index at the inter-states level. Moreover, the negative regression coefficient of time confirms the downward secular trend indicating that the distribution of cement capacity is becoming more equitable at the States level in India.

Historically, cement units tended to come up near limestone deposits as it is a major cost component of cement production. Moreover, due to its high weight it cannot be transported economically to long distances. A significant change seems to be coming up, with split location plants becoming more popular, which involves the clinker manufacture near limestone deposits but grinding and cement manufacture near the consuming centres. This was a cause of the redistribution of capacity away from the traditional limestone rich states and consequently leading to a drop in concentration and inequality. Cement and limestone are high volume, low value commodity and the transportation cost is very significant. Consequently, the drop in concentration and inequality among cement producing states implies efficiency gain for this sector. This study, thus, opens up avenues for further research in the area of efficiency improvements in cement and allied industries, which would be of paramount importance in industrial economics.

A quartile-wise analysis of the States reveals that the top two quartiles have lost shares. The lowest quartile remained at the fringes and lost further capacity share and had a capacity of less than 1% of India. It was the states between the 1st and 2nd quartile which gained share.

The Gini Coefficient of Inter State Production versus Capacity also shows a downward trend. The capacity utilisation of relatively smaller cement producing states increased rapidly over the period, compared to larger cement producing states which also increased their capacity utilisation, but at a lower rate.

SDP mining and SDP construction are significant explanatory variables for cement sales in the states. The panel regression results assign a positive coefficient to both the variables, implying that increased activity in mining and construction in states results in higher sales of cement in those states. The SDP Railways factor and SDP (Other Transportation) factor is not significant at the states level.

Another interesting result is that the Hausman test prefers REM model over FEM. This indicates that the factors at the state level are less important. It appears that the factors at the macro level, which cover all states, may be more important. The policy on decontrol, de-licensing, fiscal and monetary policies of the Union Government appears to be more critical.

The present study is primarily exploratory rather than normative. However, some policy implications follow: New technology Split location plants which involve the clinker manufacture near limestone deposits but grinding and bagging facilities near the consuming centres need to be encouraged through tax breaks. Blending of fly ash or slag with Ordinary Portland Cement (OPC) to produce blended cement has the twin advantage of reducing pollution and reducing cost. With a view to promoting the use of blended cement, tax could be significantly reduced for blended cement compared to OPC. Moreover, construction codes of the Government agencies may be modified to encourage their use, where the performance requirement of constructions could be met by the use of blended cement. More Split location plants and blended cement could help in achieving the objective of growth with equity.

Caption: Figure 1: Trend of Inter State inequality

Caption: Figure 2: Cusum square Gini of Inter State Capacity

Caption: Figure 3: Trend of Inter State concentration

Caption: Figure 4: Cusumsq Herfindahl index of Inter State Capacity

Caption: Figure 5: Trend of Inter State inequality (Production vs. Capacity)

APPENDIX

Table A.1: Gini of
Inter States Capacity

YEAR GINI

1989 0.586
1990 0.581
1991 0.577
1992 0.588
1993 0.595
1994 0.592
1995 0.581
1996 0.583
1997 0.585
1998 0.593
1999 0.602
2000 0.602
2001 0.598
2002 0.581
2003 0.568
2004 0.561
2005 0.561
2006 0.559

Source: Constructed using
data from CMA publication
"Cement Statistics"

Table A.2: Herfindahl Index
of Inter States Capacity

Year Herfindahl Index

1989 0.122
1990 0.121
1991 0.119
1992 0.125
1993 0.128
1994 0.126
1995 0.125
1996 0.127
1997 0.124
1998 0.126
1999 0.129
2000 0.129
2001 0.126
2002 0.118
2003 0.113
2004 0.110
2005 0.110
2006 0.110

Source: Constructed using
data from CMA publication
"Cement Statistics"

Table A.3: Gini of Inter
States Production vs.
Capacity

YEAR GINI

1989 0.643
1990 0.636
1991 0.641
1992 0.644
1993 0.646
1994 0.647
1995 0.648
1996 0.637
1997 0.643
1998 0.659
1999 0.669
2000 0.663
2001 0.657
2002 0.647
2003 0.634
2004 0.630
2005 0.627
2006 0.627

Source: Constructed using
data from CMA publication
"Cement Statistics"

Table A.4: Quartile wise distribution of States (Capacity wise)

Quartiles 1989 1997 2006

>3rd Madhya Pradesh Madhya Pradesh Madhya Pradesh
quartile Andhra Pradesh Andhra Pradesh Andhra Pradesh
 Rajasthan Rajasthan Rajasthan
 Gujarat Maharashtra Gujarat
 Tamil Nadu Gujarat Tamil Nadu

Between Karnataka Karnataka Maharashtra
2nd and 3rd Maharashtra Tamil Nadu Karnataka
quartile Uttar Pradesh Bihar Uttar Pradesh
 Bihar Himachal Pradesh Bihar
 Orissa Uttar Pradesh Himachal Pradesh

Between Himachal Pradesh Orissa Punjab
1st and 2nd West Bengal Punjab West Bengal
quartile Haryana West Bengal Orissa
 Kerala Delhi Kerala

< 1st Delhi Haryana Delhi
quartile Jammu & Kashmir Kerala Jammu & Kashmir
 Assam Meghalaya Assam
 Meghalaya Jammu & Kashmir Meghalaya
 Punjab Assam Haryana

Source: Constructed using data from CMA publication "Cement
Statistics"

Table A.5: Trend of Inter State Inequality

State Capacity Utilization Capacity Utilization
 difference in 2006 difference in 2006
 & 1989 & 1999

Uttar Pradesh 23.7 47.0
Orissa 15.7 35.5
Bihar 37.0 27.9
West Bengal 27.8 24.9
Kerala 46.8 24.4
Jammu & Kashmir -8.0 23.5
Rajasthan 22.8 20.0
Madhya Pradesh -1.3 18.1
Karnataka 10.9 13.0
Himachal Pradesh -20.0 10.0
Maharashtra 20.6 9.6
Gujarat 11.3 4.4
Tamil Nadu 15.5 2.1
Andhra Pradesh 14.8 1.8
Assam -20.0 0.5
Haryana -100.2 0.0
Delhi -2.9 -2.2
Meghalaya 9.8 -3.5
Punjab 106.1 -5.8

Source: Constructed using data from CMA publication
"Cement Statistics"

Table A.6: Collinearity diagnostics of
the variables at the State level

Model Collinearity
 Statistics

 Tolerance VIF

1 (Constant)
 sdpm .480 2.082
 sdpc .258 3.881
 sdprail .291 3.441
 sdpotr .354 2.821

Dependent variable: Sales

REFERENCES

[1.] Cement Manufacturers' Association. Cement Statistics. Various Volumes. Delhi. India.

[2.] Credit Rating Information Services of India Ltd (CRISIL). 2005. Cris Infac Cement Annual Review. 2005. Mumbai. India.

[3.] Credit Rating Information Services of India. (CRISIL). Crisil Research Cement Annual Review. 2009. Mumbai. India.

[4.] Ellison, G. and E.L. Glaeser. 1999. The Geographic Concentration of Industry: Does Natural Advantage Explain Agglomeration? American Economic Review. Vol.89, No.2:311-316.

[5.] Fujita, M., P. Krugman, AJ. Venables. 2001. The Spatial Economy: Cities, Regions and International Trade. MIT Press, USA.

[6.] Gokarn S and R.Vaidya. 1994. The Cement Industry in India Competitive Structure, Profitability and Prospects. Investment Information and Credit rating Agency of India limited (ICRA)--ICRA sector focus series #3. New Delhi. India

[7.] Gokarn, S. and R.Vaidya.1993. Deregulation and Industrial Performance The Indian Cement Industry. Economic and Political Weekly. Vol. 28, No. 8.: 33-41

[8.] ICRA Ltd. 2006. ICRA sector analysis Industry comment. The Indian Cement Industry. New Delhi. India.

[9.] Jha, R, M.N. Murty, S. Paul and B.S. Sahni. 1991. Cost Structure of the Indian Cement Industry, Journal of Economic studies. Vol.18, No.4: 59-67

[10.] Krugman, P.1991. Increasing Returns and Economic Geography. Journal of Political Economy. Vol. 99, No.3:483-499.

[11.] Mukhopadhyaya, J.N, Malabika Roy and Ajitava. Raychaudhuri. 2007. Post-deregulation performance of the cement industry. The Icfaian journal of Management Research. Vol.6, No.7: 7-24.

[12.] Mukhopadhyaya, J.N, Malabika Roy and Ajitava. Raychaudhuri. 2012. Change in concentration and inequality after liberalization in the cement industry of India. Vilakshan. Vol IX, No.2

[13.] Nath, P and P.R.Bose. 2002. Leveraging Liberalization--The case of Indian Cement Industry. Economic and Political Weekly. Vol. 37, No 30:3199-3203

[14.] Planning Commission, Government of India. Eleventh Five Year Plan-2001-12. Oxford University Press. New Delhi. India.

[15.] Planning Commission, Government of India. Tenth Five Year Plan. 2002-2007. New Delhi. India.

[16.] Pradhan, G. 1992. Concentration in Cement Industry under New Policy Regime. The Economic and Political weekly. Vol. 27, No 9: 31-38

[17.] Wen, M.2004. Relocation and Agglomeration of Chinese Industry. Journal of Development Economics. Vol. 73, No. 1: 329-347

(1) To our knowledge Herfindahl index has primarily been used to study industrial concentration. We have not come across any studies that use this index to study concentration at the states level.

(2) Source: White Paper on State-Level VAT. by the Committee of State Finance Ministers

(3) During the control regime, when prices were fixed, government would give a fixed retention price to the cement manufacturers and reimburse the freight.

(4) According to Crisil report the production of blended cement has increased to 61% in 2006 up from 27% in 1992-93.

(5) We could not do any direct field study on this. But, we have confirmed this from research done by Crisil, ICRA, CMA etc

(6) However, as the female working force population estimates was captured better through the National Sample Survey Organization survey the new series was started from 1993-94 instead of 1990-91.

Prof. Jayanta Nath Mukhopadhyaya, Department of Finance, Globsyn Business School; Kolkata-91; e-mail: jnmukhopadhyay@gmail.com

Malabika Roy, Department of Economics, Jadavpur University, Kolkata-32; e-mail: malabikar@gmail.com

Ajitava Raychaudhuri, Department of Economics, Jadavpur University, Kolkata-32; e-mail: ajitaval@gmail.com

Table 1: States share of All India capacity

States 1989 1997 2006

Madhya Pradesh 20.9% 24.3% 18.1%
Andhra Pradesh 18.5% 13.7% 15.8%
Rajasthan 10.9% 12.1% 12.4%
Gujarat 9.1% 9.4% 10.9%
Tamil Nadu 7.9% 5.9% 9.2%
Karnataka 7.8% 6.9% 6.9%
Maharashtra 7.8% 9.6% 7.5%
Uttar Pradesh 5.0% 3.5% 4.6%
Bihar 4.3% 4.9% 3.6%
Orissa 2.0% 1.9% 1.9%
Himachal Pradesh 1.4% 3.5% 3.0%
West Bengal 1.1% 1.0% 2.2%
Haryana 1.0% 0.5% 0.1%
Kerala 0.8% 0.5% 0.4%
Delhi 0.4% 0.6% 0.3%
Jammu & Kashmir 0.4% 0.2% 0.1%
Assam 0.4% 0.2% 0.1%
Meghalaya 0.4% 0.2% 0.1%
Punjab 0.0% 1.2% 2.7%

Note: MP includes Chattisgarh and Bihar
includes Jharkhand.

Source: Constructed using data from CMA
publication "Cement Statistics"

Table 2: Quartile wise analysis of the movement of shares of
capacity

Capacity share (1989-2006) Share '89 Share '06 Gain/loss

Top (> 3rd quartile) 67.4% 66.4% -1.0%
(Between 2nd and 3rd quartile) 26.9% 25.7% -1.2%
(Between 1st and 2nd quartile) 4.3% 7.2% 2.9%
(< 1st quartile) 1.5% 0.8% -0.6%

Capacity share (1989-1997) Share '89 Share '97 Gain/loss

Top (> 3rd quartile) 67.4% 69.1% 1.7%
(Between 2nd and 3rd quartile) 26.9% 24.7% -2.2%
(Between 1st and 2nd quartile) 4.3% 4.6% 0.3%
(< 1st quartile) 1.5% 1.6% 0.2%

Capacity share (1997-2006) Share '97 Share '06 Gain/loss

Top (> 3rd quartile) 69.1% 66.4% -2.7%
(Between 2nd and 3rd quartile) 24.7% 25.7% 1.0%
(Between 1st and 2nd quartile) 4.6% 7.2% 2.6%
(< 1st quartile) 1.6% 0.8% -0.8%

Source: Constructed using data from CMA publication "Cement
Statistics"

Table 3: Regression results of interstate capacity Gini
& H index with time

Dependent Variable Independent Coefficient &
 Variable t-ratio in
 parentheses

Log of Gini coefficient Time -0.002
(Inter-state capacity) -(2.044) *

Log of Herfindahl index Time R2 = 0.21
(Inter-state capacity) -0.006
 -(2.832) **

Log of Gini coefficient Time R2 = 0.33
(Inter-state capacity -0.00
vs. production) -(0.788)
 R2 = 0.037

** shows significance at 5% level

* shows significance at 10% level

Table 4: Quartile wise analysis of the movement of share of capacity

Capacity share (1989-2006) Share '89 Share '2000 Gain/
 loss

Top (> 3rd quartile) 67.4% 70.5% -3.1%
(Between 2nd and 3rd quartile) 26.9% 23.4% -3.5%
(Between 1st and 2nd quartile) 4.3% 5.0% 0.7%
(< 1st quartile) 1.5% 1.1% -0.4%

Capacity share (2000-2006) Share '2000 Share '06 Gain/
 loss

Top (> 3rd quartile) 70.5% 66.4% -4.2%
(Between 2nd and 3rd quartile) 23.4% 25.7% 2.2%
(Between 1st and 2nd quartile) 5.0% 7.2% 2.2%
(< 1st quartile) 1.1% 0.8% -0.3%

Source: Constructed using data from CMA publication "Cement
Statistics"

Table 5: Random Effect results of the panel
regression at the States level

Variables Coefficients
 (t-ratios)

Sdpm 1.30E-04
 (3.586) ***

Sdpc 1.31E-04
 (8.936) ***

Sdprail 1.69E-05
 -0.172

Sdpotr -1.23E-05
 (-0.564)

[R.sup.2] = 0.54; N = 247
Lagrange Multiplier: 967.8
(1 df p value: 0.000);
Hausman: 7.49
(4 df p value: 0.112)

*** significant at 1% level