SPECIAL ARTICLES

Economies of Scale in Cement Industry

G S Gupta

This paper investigates whether there are economies of scale in the Indian cement industry. It,
therefore, estimates cost-output (sales) relationships, using the time series and the cross-section data, and
at the industry level and the firm level.
At the industry level, the cost-output relationships have been estimated separately for all-India.
Bihar, and Madras — the regional classification for which time-series data are available.
The relationships between cost components — material cost, labour cost, and depreciation cost —
and output, have also been determined in order to identify the sources of economies or diseconomies of
scale.
IT is found that the cement industry is Canada, and 224 kg in UK — 10 years In terms of elasticity, economies of
dominated by the L-shaped average earlier, in 1959. As the country deve- scale exist if n> 1, diseconomies of scale
cost curve and the horizontal marginal lops, demand for cement is bound to in exist if r) > 1 > and neither exist if n = 1
cost carve. In other Words, the indus- crease at an accelerated rate. It to esti- (y) = output elasticity of total cost).
try is still operating in the first half of mated that today it is increasing at
the U-shaped average cost curve, and about 10 to 12 per cent per annum. 1 . Economic theory argues that, as pro-
thus cement firms have not yet reached duction begins, there are economies of
Cement being a bulky commodity, its
their optimum size. Significant economies scale upto a certain point, and that
transport, particularly from far-off pla-
of scale exist only with respect to the thereafter diseconomies of scale set in.
ces, has to be minimised. While pro-
Behind this is the economic law of vari-
labour cost in all-India and Bihar, and duction must increase to match increas-
able returns to factors of production. As
with respect to total cost in the all- ed demand, and as production increases
more and more units of a variable means
India cement industry and particularly m total cost increases too, there is no de-
of production are applied, marginal pro-
the four cement firms studied — viz, finite relationship between cost and out-
duction first increases and then, after a
Associated Cement Companies, Jaipur put. Nevertheless, knowledge of the
point, starts to decline. This is because,
Udyog, Mysore Cement, and Sone Val- manner of this relationship is very use-
in the beginning, some fixed factors of
ley Portland Cement Company. Signifi- ful for production planning. If cost in-
production are underutilised and, as the
cant diseconomies exist only with respect creases less than proportionately with in-
scale of operation increases, a point of
to total cost and material cost in the crease in output, it is advantageous for
their optimum use is reached further ex-
Madras industry. Inter-regional compa- the supplier to expand output and vice
pansion beyond which either results in
rison has indicated that expansion of the verm. Of course, even if cost increases inefficiency or requires a further addi-
industry outside of Bihar and Madras, more rapidly than output, production tion of fixed factors of production. The
and contraction of Madras firms, would may still be increased if the demand re- reasons for the economies of scale are
be beneficial in terms of costs. Anioni? lationship is such that the manufacturing that large firms
the four cement firms considered, Asso- company still adds to its profit. Also,
ciated Cement Companies enjoys maxi- of course, even though from the cost side (a) are able to obtain raw materials,
mum economies of scale. Its sales elas- it may pay to expand output, demand etc, at lower per unit price than
ticity to total cost at sample means is conditions may be such that the expan- small firms;
0.42. sion is not profitable. (b) can raise funds at lower rates of
interest than small firms;
The cement industry in India is pret- In a multi-firm industry such as the (c) can afford to spend substantial
ty old and is now well-established. Ce- cement industry it is possible that not sums of money on scientific, tech-
ment was first manufactured in India in all firms face the same cost-output rela- nical and marketing research,
which reduce the per unit cost;
1904 in Madras. Its production in 1969 tionship. So, those firms whose cost (ci) are able to use their plant, ma-
stood at 13,620,000 tonnes and in 1971 function is such that cost increases more chinery and services of technical
at 14,928,000 tonnes. Between 1950 than proportionately with increase in staff more continuously and more
and 1969, cement production increased output may be discouraged from increas- hence efficiently than small firms;
and point
at an average rate of 8.3 per cent per ing their output. The objective of this
(e) are able to make better use of
year. In terms of cement manufacture, paper is therefore to determine the cost- their by-products.
India will soon be among the first six output relationship in the Indian cement
countries of the world. industry. These economies of large-scale pro-
duction lead to a fall in average cost
Aggregate cement consumption in as output expands. However, the opti-
SOURCES OF E C O N O M I E S A N D
India stood at 13,485,000 tonnes in mum size of the firm (i e, the minimum
1969, and it has increased at the ave- DISECONOMIES OF SCALE
average cost) is reached when these eco-
rage rate of 8.4 per cent between 1950 Economies of scale exist if total cost nomies disappear and diseconomies of
and 1969. Nevertheless, per capita ce- increases less than proportionately with large-scale production are about to set
ment consumption in India is very low increase in output; diseconomies of scale in. As the firm size becomes too large,
— particularly compared with in the exist if total cost increases more than diseconomies arise due to
advanced countries. It was 26 kg in proportionately with increase in output;
1969, whereas it was 420 kg in Ger- and neither exist if the proportionate in- (a) inefficiencies arising out of dissi-
pation and weakening of man-
many, 341 kg in the US, 313 kg in crease in cost and output is the same. agement control;
551
ECONOMIC AND POLITICAL WEEKLY March 29, 1975

TABLE 1 : TOTAL COST FUNCTIONS

(b) inefficient communications and cost (TC) curve rising and convex from We have tried all the four forms for
consequent delays in operation; below (output-axis), and U-shaped ave- various cost functions estimated in this
and
rage cost (AC) and marginal cost (MC) paper. A particular form has been
(c) labour losing its motivation due
to a sense of insignificance. curves. However, lately, there is evi- chosen for each function on the basis of
dence of a linear TC curve, a horizontal both economic theory and statistical in-
The cement industry being a capital- (parallel to output-axis) MC curve, and ference. In particular, the best form for
intensive industry, the economies of scale an L-shaped AC curve. 3 There is thus no each function has been chosen on the
may be overwhelming. In this paper, we definite idea about the nature of the basis of the a priori signs of the co-
propose to determine the optimum size cost-output relationship. efficients, significance of the co-efficients
of a cement firm. In fact, there are four different hypo- as indicated by the t-test and contribu-
theses about the form of this 'relation- tion of each new independent variable to
METHODOLOGY ship: cubic, quadratic, linear and dou- the explanation of variation in the inde-
The method adopted in this paper is ble-log. It can be easily shown that for pendent variable as measured by the co-
of estimating the (total) cost function, AC and MC curves to be U-shaped, the efficient of determination (R 2 ).
both f r o m time-aeries' and cross-section TC function should be cubic and that The various cost functions estimated
data. The theory of the firm hypothe- the coefficient of the output cube must in this paper are the total cost functions
sises cost to vary positively w i t h output be positive while the co-efficient of the for the all-India, for Bihar, and for
and factor prices, and negatively w i t h output square must be negative, A qua- Madras cement industries; also for As-
technical progress. Alternatively, cost at dratic TC function will give a U- sociated Cement Companies, for Jaipur
constant prices is postulated to vary po- shaped AC curve, only if the intercept Udyog, for Mysore Cement, and for
sitively w i t h output and negatively w i t h term and the co-efficient of output square Sone Valley Portland Cement. 3
technical progress. In such formulations, are positive; and, in such a case, the While all these are estimated from
all other relevant factors are taken to MC curve w i l l be a monotonously rising time-series data, the total cost function
influence cost through output. The rele- one, A linear TC curve implies a con- for all-India has been estimated from
vant factors are: location, skill and effi- stant MC curve and an L-shaped AC cross-section data separately for 1963,
ciency of workers and management per- curve only if the intercept term is posi- 1967, 1971, and 1972. Cost functions
sonnel, unforeseen contingencies, such as tive; if the intercept term is negative, for the separate cost components —
strikes, floods, adverse climate, non- the AC curve w i l l have an inverted L- viz, material cost, labour cost, and de-
availability of raw materials, etc. shape, A double-log TC function re- preciation cost — have also been esti-
Economic theory also suggests the sults in a constant, falling, or rising, AC mated separately for all-India, Bihar,
likely shape of the cost curves, Tradi- and , M C . curves, it never yields a U- and Madras, from the time-series data.
tionally, this is supposed to be a total shaped AC or/and MC curves. The cost functions at the firm level

653
March 29, 1975 ECONOMIC AND POLITICAL WEEKLY

TABLE 2 : COMPONENT C O S T FUNCTIONS

and at the country level, using crass- cost to labour cost are used as proxy vari- The Bombay Exchange Directory pro-
section data, are estimated using cost ables for measuring the technologies of vides firm-wise data for 17 firms manu-
and sales data rather than cost and out- different firms. facturing cement in India, They are
put data; for, their balance-sheet cost Ashoka Cement ( A C L ) , Associated Ce-
THE DATA
data, w h i c h alone are published, include ment Companies ( A C Q , Chettinad Ce-
both production and distribution costs. The time-series data for all-India, ment Corporation (CCC), I n d i a Cement
As changes in inventories render sales Bihar, and Madras are obtained f r o m ( I C L ) , Jaipur U d y o g (JUL), Saurashtra
different f r o m production, the cost-sales the Census of M a n u f a c t u r i n g Industries Cement and Chemical Industries (SCC),
relationships cannot be used to infer eco- and the A n n u a l Survey of Industries. Travancore Cement ( T C L ) , A n d h r a Ce-
nomies or diseconomies of scale. The time-series data for individual firms ment (ANC), Bagalkot Udyog (BUL),
In time-series regressions, time (trend and the cross-section data for all cement D a l m i a Cement I n d i a . ( D C L ) , Madras
variable) is used as a proxy variable for units in I n d i a are obtained f r o m the Cement ( M A C ) , Orissa Cement (OCL),
technical progress; on the other hand, Bombay Stock Exchange Directory Panyam Cements and M i n e r a l Industries
in cross-section data, the ratios of mate- (1973). As the data are voluminous, (PCM), Shri Digvijay Cement (DCL),
rial cost to labour cost and of operating they are not reproduced here. and Sone Valley Portland Cement (SVP).

TABLE 3 : O U T P U T (SALES) ELASTICITIES OF T O T A L COSTS

554
ECONOMIC AND POLITICAL WEEKLY March 555, 1975

TABLE 4 : OUTPUT ELASTICITIES OF COMPONENT CotSIS

when the data from all these 17 firms these Tables all the pertinent statistics, their corresponding optimum sizes.
were not available, those relating to 16 i e, t-values of the co-efficients, co-effi-
or 14 firms are used in the cross-sec- cient of determination (R 2 ), and D u r b i n - E C O N O M I E S OF S C A L E
tion analysis. Watson statistic ( D W ) , have been In order to find out the existence or
A l l rupee variables are measured in recorded for each equation. The t- otherwise of economies of scale, we
crores of rupees, and output variables values of those co-efficients which are need to compute output elasticity of
are measured in crores of tonnes. Prices significant by the one-tail t-test, at the total cost from the estimated cost-out-
and wages variable are denominated in 5 per cent and 1 per cent levels are put relationships. In its double-log
index numbers, w i t h base 1961 = 1.00. marked w i t h * and **, respectively. form, this elasticity is constant and is
Ratios are measured in absolute num- It w i l l be seen from the results that given by the co-efficient-of-output vari-
bers. The time variable is normalised all the co-efficients have a priori cor- able; in all other forms, the elasticity
to take a value of zero in period one; rect signs. Statistically, most equations is variable and needs computation. The
one in the period two; and so on. appear to be equally acceptable. Of variable elasticities have been computed
The f o l l o w i n g notations have been the 20 equations in Tables 1 and 2, at each observation, and at the sample
14 equations explain more than 92 per means. The variable elasticities at
cent, one about 88 per cent one about sample means and their extreme values,
81 per cent, three about 81-82 per cent, along w i t h constant elasticities, are pre-
and one nearly 64 per cent of the total sented in Tables 3 and 4. Out of the
variation in the corresponding dependent 20 elasticities, six are greater than one
variable. Most co-efficients are signi- and the remaining are less than one.
ficantly different from zero, and no Thus, of the 20 cost-output (sales)
equation is subject to serially correlated relationships estimated, diseconomies of
disturbance term. scale are found in 6 and economies of
scale in 14 cases. .Furthermore, nine
It w i l l be seen from Tables 1 and 2
elasticities fall between the range of
that, out of the total of 20 cost equa-
0.89 and 1.08; they thus indicate near
tions, one is cubic, two are quadratic,
index absence of economies or diseconomies
twelve are linear, five are of double-
p K = wholesale price index of scale On the lower side, there are
log forms. F r o m the signs of the co-
and transport equipments index. efficients in all these equations, it w i l l three elasticities: output elasticities of
labour cost in all-India (0.39) and in
be clear that the cubic equation w i l l
RESULTS AND INTERPRETATION Bihar (0.28) and sales elasticity of total
result into U-shaped AC and MC
cast in A C C (.42). On the higher side
The different forms of the various curves, linear equations into L-shaped
there is one elasticity — the output
cost functions were estimated by the AC and constant MC curves, and
elasticity of total cost in Madras (1.19).
ordinary least squares method. One double-log into rising or falling AC
A l l this implies that there are large eco-
form for each equation was then select- and MC curves — depending upon
nomies of scale in labour costs in the
ed on the basis of both economic theory whether the elasticity co-efficient is
all-India and Bihar cement units and
and statistical inference. Furthermore greater than or less than unity. Thus,
in total cost in A C C , while there are
other things remaining the same, we in the I n d i a n cement industry, there is
significant diseconomies of scale in the
have preferred cubic f o r m over all a predominance of the linear TC curve,
Madias cement industry. Significant
other forms, quadratic form over linear, constant or horizontal MC curve, and
economies of scale are also found in
and double-log forms, and linear f o r m L-shaped AC curve. This does not
total costs in all-India (n = 0.56),
over double-log form. The results of mean that the Indian cement industry
J U L (N , 0 . 5 0 ) , and M C L (n- 0,61).
the best-fitting total-cost equations axe is not facing the U-shaped AC curve,
jprovided in Table X, and those of com- but it does mean that cement firms in These elasticities can be used to make
ponent cost equations in T a b l e 2. In India, in general, have not yet reached inter-regional, inter-temporal, and inter-

653
 ECONOMIC AND POLITICAL WEEKLY
March 29, 1975

Bihar, and w i t h respect to total cost in 3 It must be remembered t h a t there

firm comparisons of cement factories in arc cement units in states other than
India. Output elasticity of total cost the all-India cement industry and in Bihar and Madras and hence the
is the lowest in all-India (0.56), and the the four cement firms studied. Signifi- units in these two regions do not
highest in Madras (1.19); it is 0.89 in cant diseconomies exist only with tell the story of all units in all-
Bihar. This means that cement facto- respect to the total cost and material India. '
cost in the Madras factories. A C C is the largest cement manu-
ries in places other than Bihar and facturing firm SVP is the oldest,
Madras enjoy greater economies of Inter-regional comparison has indicat- M C I . is the smallest and J U L is a
scale than do those in Bihar. This sug- ed that expansion of the industry in medium-Sized firm.
gests that, looking from the cost side places other than Bihar and Madras
alone, cement firms in places other than References
and contraction of Madras firms will be [1] Gupta, G S (1974): 'Demand for
Bihar and Madras must be encouraged beneficial from the side of costs. Among Cement in India', submitted for
more to expand than those in Bihar, the four cement firms considered, the publication in Indian Economic
and that the Madras firms should con- Associated Cement Companies with Journal.
tract. sales elasticity of total cost at 0.42 [2] Gupta, V K, (1968): 'Cost Func-
tions, Concentration and Barriers
The cost-sales relationships, from enjoys the maximum economies of to Entry in Twenty-Nine Manufac-
cross-section data of 1963, 1967, 1971 scale. turing Industries of India', Journal
and 1972, indicate no significant change of Industrial Economics, Volume
in sales elasticity of total cost over the X V I I , Number 4 (November), pp
Notes 57-72.
period 1963 through to 1972. It has
[The author is grateful to P N Misra [3] Johnston, J, (1960): "Statistical
fluctuated between 0.98 and 1.08. Cost Analysis", M c G r a w - H i l l , New
for useful discussion and to T K Patel
Coming to inter-firm comparisons, for her help in data collection and York.
the maximum economy is enjoyed by computer runs.] [4] Somani, G D, (1969): 'Cement in
the Associated Cement Companies the Fourth Plan, Commerce, (Sup-
1 Somani (1969), Gupta (1974). plement) 119:3046 (September 20),
( n= 0.42), the largest cement firm
2 Johnston (1960), Gupta (1968). pp 6-9.
in India; and the minimum economy is
available to Sone Valley Portland
Cement (n = 0.81). A l l the in-
dividual firms under study are reaping Commodity Price Stabilisation
the fruits of economies of scale. T H E 87-member Committee on Com- tive sources of finance for .stocking
Output elasticities of component costs modities set up by the United Nations programmes; and multilateral commit-
indicate, however, that economies of Conference on Trade and Development ments in trade. Attention w o u l d also
scale arise mainly from labour cost; ( U N C T A D ) agreed in Geneva recently be given to the effects of the pro-
and that material and depreciation costs that it should hold two further meetings gramme on commodity imports essential
are subject to nearly constant returns to before the end of this year aimed at to developing countries.
scale. reaching conclusions on an integrated In a unanimous resolution on assis-
programme and to formulate suggestions tance to developing countries in the
In this paper, we have not worked "for possible negotiations between multilateral trade negotiations, the
out the optimum size of the cement producers and consumers for arrange- Committee requested the UNCTAD
industry or firm. This can be obtained ments on internationally traded com- Secretary-General to continue to assist
only from the total cost-output (not modities". developing countries in multilateral
total cost-sales) relationship, and only
The decision stems from the trade negotiations w h i c h have now
when this relationship is either quadra-
U N C T A D Secretary-Generars proposals entered an active phase w i t h i n the
tic or cubic. However, from our
for an integrated programme fox com- General Agreement on Tariffs and
finding of the predominance of the
modities. They call for a comprehen- Trade. The Trade and Development
L-shaped average cost curve, it can
sive attack on the commodity problem Board should be kept informed of
be concluded that neither the industry
through simultaneous action in five developments in this regard. Further,
nor any firm has yet reached its
fields: establishment of international the Director-General of the Food and
optimum size. The industry is still
stocks for a wide range of primary Agriculture Organisation was invited to
operating on the left part of the U-
commodities; establishment of a common make available to all developing coun-
shaped average cost curve. This means
fund for financing these stocks; m u l t i - tries concerned the FAQ's expertise in
that, at least from the cost side alone,
lateral trade commitments; improved the field of commodities and commodity
there is good scope for the cement
facilities to compensate developing problems.
industry's further expansion.
countries for shortfalls in their com- There was general agreement in the
modity export earnings; and Expansion Committee that the U N C T A D secre-
CONCLUSION of processing in developing countries tariat should continue its work on
of their raw materials. marketing and distribution systems for
We have estimated the cost-output
relationship in the Indian cement The Committee recommended that primary commodities. The secretariat
industry both from time-series and the U N C T A D Secretary-General, Ga- should seek the active co-operation of
cross-section data. It is found that, in rmani Corea, elaborate his proposed governments of both producing and
general, the industry is still operating programme. In its further w o r k , the consuming countries, trade and industry
on the first half of the U-shaped aver- Committee decided that particular at- interests and intergovernmental organi-
age cost curve; thus cement firms have tention should be given to the follow- sations. The commodity coverage should
not yet reached their optimum size. ing questions: suitability of individual be extended and further work to sup-
Investigation into the economies or dis- commodities for stocking; stock manage- plement the studies already prepared
economies of scale has suggested that ment measures including principles and on bananas, cocoa, and hides and skins
significant economies exist only w i t h objectives of market intervention; should be considered in the light of
respect to labour costs in all-India and amounts, terms, techniques and prospec- comments made by governments.

556