Methodological Issues in Measurement of productivity-

An alternative with Singareni Company Empirical

Illustration
*Dr. A.K. VasudevAchary
Introduction:

Productivity, in general, is defined as a ratio of what comes out of a business to what

goes into the business. The outcome of a business can be expressed in terms of value of
goods produced and the efforts can be expressed in terms of total value of inputs
required for that. The productivity would then be the value of goods divided by the value
of inputs. That is in this context, this paper raises and discusses certain issues and
problems that are often faced by empirical investigators. They pertain to measurement of
production factors and equation of productivity trends. We also examine the relevance of
popular methodology adopted in empirical studies in the developing economies. A by-
product of the discussion of the issues is the development of an alternative methodology,
which is nearer to reality. Singareni Collieries Company Limited episode is considered
for the purpose.

For orderly discussion, the paper divides into three Sections. In section I are discussed
problems in the measurement of inputs and output. It also examines the relevance of the
concepts of Gross output and Value Added for different purposes. Section II scans
difficulties inherent in the quantification and interpretation of Partial and Total factor
productivity indices. It also assesses the relevance of the popular Total factor productivity
approach, in the mixed planned economies with unorganized and non-competitive market
structure. In the final section an attempt is made to evolve an alternative new approach to
compute factor inputs in terms of surrogates by taking time series data of Singareni Coal
Industry. The paper ends with a brief summing up.
I
Quantification of Inputs and Output – Problems:

In a multi-plant multi-product industry, both the inputs and outputs are heterogeneous.
Researchers face conceptual problems in measuring them, breakup data availability apart.
Economists devote much time in underpinning numerous problems in evaluating capital.
There is some justification to single out capital whose neat measurement in dynamic
setting is almost impossible. Admittedly, capital is a crucial factor in almost all economic
analyses dealing with production and distribution. Nonetheless, both conceptually and
statistically, the problem of measurement has not been solved adequately, Moreover, as
capital is amalgam of heterogeneous means of production built up over time and capable
of serving over a stretch of period, they lose efficiency over time, However, some
*Associate Professor, Department of Economics, Osmania University, Hyderabad-7
(A.P): E-Mail: akvchary@rediffmail.com
 2

machine work with perfect efficiency until they break down and others depreciate
progressively over time.(1)

Aside, in mature industrial economies, machines are often scrapped not due to decay but
due to technical obsolescence caused by rapid innovations (2). Hence, depreciation
Accounting is not an easy job. One method to get rid of the problems of depreciation
accounting is to consider gross fixed capital and its complement of gross value added
(GVA) instead of net fixed capital and net value added. Unless depreciation is properly
estimated, valuation of capital is not possible. Economists of the two Cambridge’s across
the Atlantic debated for two decades on production function and theory of capital
(3).Joan Robinson(4), and Sraffa(5), among others, contested the Neo classical
postulation of monotonic unique relation between interest/profit rate and capital intensity.
They successfully argued and later accepted by all the economists that capital, as a value
magnitude has no precise meaning independent of interest/profit rate.

It should be admitted that like capital, labor and land too are heterogeneous and so also
outputs. However, since labor is only hired and not purchased, the problems are not
vexing. As to the measurement of output also problems are not severe. One reason for the
serious problems in measuring capital is that, while capital has no technical units of
measurement of its own, labor and land has viz., manhour/mandays, and acres/hectares
respectively. But, capital has to be measured in value terms only. Let us note here that,
value magnitudes are affected both by changes in distribution of income and techniques
of production.

An empirical researcher, taking cognizance of all these imponderables should not

cultivate nihilism. Within the constraints, one should make compromises with fine theory
of capital and, attempt to aggregate heterogeneous capital at least approximately; Kaldor
too finally holds that measurement of real stock of capital must be based on some more or
less arbitrary conventions(6). Having examined problems in the measurement of capital,
we switch over to evaluation of output.

Gross And Net Value Added Outputs

Like inputs, output too is not homogeneous. Hence, here also we face problems of
aggregation. Output may be evaluated by considering either gross output (sales value) or
value added, and both may be excluding or including depreciation. Although at macro
level, value added concept is appropriate, at micro level both gross output and value
added may be considered. If one is interested in measuring income generated within the
industry the choice falls on value added concept. Studies dealing with income distribution
and productivity evaluation often consider value added concept.

Whether value added should be net or gross of depreciation, i.e., whether net value added
(NVA)or gross value added(GVA) is to be preferred? Theoretically, NVA is appropriate.
 3

But in arriving at NVA, many a turbulent problems connected with depreciation

accounting are faced. Hence, there is a tendency on the part of the empiricists to opt for
GVA. There is an additional reason for preferring GVA in third world countries. Due to
paucity of funds and supply constraints in procuring adequate capital equipments, an
existing machine is often used beyond its technical life(7).

GVA may be measured either directly or indirectly. As GVA is synonymous with

incomes generated within an enterprise, it should be equal to wages and surplus.

GVA= Wages + Surplus (Profits and Depreciation)

Alternatively, GVA may be computed directly by deducting inter-industry purchases

from gross output.

GVA = Gross Output – Inter-Industry purchases.

In empirical research, often the indirect method is employed due to difficulties in the
availability of detailed data to distinguish components of inter-industry purchases from
capital assets.
.
Reverting to GVA concept, it is pertinent to take cognizance of a limitation, when we
identify GVA with the contribution of only labor and capital employed in the industry,
we lose sight of contributions made by the outsiders i.e., economies achieved in the use
of intermediate inputs, the productivity gains in the case of an individual industry will
have some portions to explained in terms of gains realized in other related industries.
This fact is to be borne in mind while drawing inferences from micro level productivity
studies. Thus crux of the discussion carried thus far is: whether inter-industry purchases
are to be netted out (value added) or gross production (sales value) is to be considered.

Evaluation of Output at Constant Prices

In temporal studies, all the variables including output have to be evaluated in physical
terms to insulate from price effect, due to heterogeneity in the product-mix, output is
counted first in money units and later the revenue product is transformed into physical
units by the method of deflation. One issue connected with deflation is whether double or
single deflation is to be preferred. It is held that depending on the purpose, either may be
chosen, Theoretically, although double deflation method is superior, it does not always
enable to know true amount of incomes available for distribution among the factors of
production employed in the enterprise.

In the foregoing, our attention was on deflation methods, now we raise another issue
connected with the nature of price deflators relevant to convert GVA in current Rupees
into constant Rupees. Use of a particular deflator depends on the purpose in hand,
Despite the relevance of consumer goods prices and capital goods price deflators, due to
difficulties in obtaining data and computing a composite price index, researchers
sometimes lean to-wards the soft option of choosing product prices for deflation.
 4

II
Partial and Total productivities

In the foregoing, we discussed a few issues related to the choice of output concept,
deflation methods, nature of proper deflators and their relative merits, Having done the
spade work, we shift to discuss problems in measurement and interpretation of partial and
total productivity indices. We begin with partial productivity.

Partial Productivity Indices:

For long time, economists interpreted productivity to mean labor productivity. This is
fine under certain circumstances and from a broader outlook. So long as the role of
capital and technical progress embodied in machines and equipment was not significant,
growth in productivity may be viewed as growth in labor productivity.

Despite, the relevance of partial productivity analysis, it is held that since productivity is
output per unit of input, ascribing all the increments in production to labor alone is
inappropriate. Hence, economists started computing and analyzing partial productivity
indices along with total productivity indices. Admittedly, partial productivities such as
O/L serve limited purpose, this is more so when structural changes occur and K/L ratio
undergoes change. These changes compel researchers to take up total productivity
analysis.

Total Productivity Indices:

As noted above, limitations of the partial productivity analysis led to the popularity of
total factor productivity studies since the fifties. The raisen d’etre for the popularity of the
total productivity concept is obvious. Since production is the result of combined efforts of
all the factors, total productivity rather than labor or capital productivity is relevant. In
case, the output is homogeneous and is functionally dependent on a single fact like labor,
the growth in factor productivity i.e., growth in output per worker will be equal to growth
in output less growth in labor. However, the real world is far way from the aforesaid
description. Out world is with multi-outputs and multi-factors, Therefore, partial
productivity measures though use ful, fail to distinguish between output growth due to the
use of more of all inputs per laborers and output grows due to improved efficiency. There
fore, reliance is to be placed on total factor productivity.

Difficulties in computing Total Factor Productivity Indices:

As said earlier, total factor productivity by its very nature suggests that output is to be
related to all the inputs employed. This is conceptually sound. But the trouble is how to
overcome conceptual difficulties and how to aggregate the heterogeneous inputs. One
question as raised earlier is whether all factor inputs and current inputs are to be
considered or only labour and capital. It is worth to report here that total factor
 5

productivity and technical change are inter-changeably used in macro economic

production analysis. Also total productivity is known as output per unit of combined
inputs, Residual or measure of Our Ignorance(8). It Is Called “Residual “ because it
represents net change in output after having accounted for individual effects of all inputs.

Treatment of total factor productivity change and technical change has been
conceptualized initially, by Solow(9), through the method of shift in production function
over-time. Later, the method was popularized and refined by other learned economists,
such as Dennison(10), Hulten(11), Gollop(12). However, the post-Keynesians such as
Kaldor are dead against the neoclassical conception of exogeneous and disembodied
technical progress and method of explanation through shifts in production function.
Solow himself realized the deficiency and attempted to explain technical progress
through vintage models (13).

Furthermore, recently attention is focused attention on the problem of correcting labor

input for quality changes, economists such as Grilliches(14), Denison(15), and Kendrick
(16). Education and training have been recognized as the most important factors affecting
quality of labour input. Numerous problems are faced in incorporating education to
evaluate labour input. Estimating the contribution of the stock of human capital to growth
of output raises the well-known problems of aggregation (index numbers problem). It
also poses the problem of knowing relation of human capital with physical capital in
terms of complimentarity and substitutability.

Issue of Proper Weighting Diagram And Base Year:

Another area of major concern for researcher is the selection of proper weights and base
year. In the literature on productivity studies, we find different systems of weights. For
example, in aggregating labour and capital while Solow(17) used base year rates of
return, (wage rate and rate of Profit) as weights, Kendrick (18) preferred base year
relative shares.

As a corollary to the aforesaid, we ask another question: whether the weights assigned
should be constant or variable for the whole period (say 10 or 20 Years) Solow appears to
have given shifting rather than constant weights. However, Kendrick gave base year fixed
weights. In ‘Divisia’ index construction popularized by Solow, each time a pair of
consecutive years is considered in the continuum of the time period chosen. In other
words, in the Divisia Approach, changing weights are considered based on each pair of
years.

Another issue is that in case factor proportions do not change, is there a need to give
weights? We know that in the real world in the long run, factor proportions have a
tendency to change. If so, is it not convenient to give weights based on every 4 th or 5th
year?

When constant or shifting weights are adopted, what do they imply? Do they not warrant
or purport to reflect competitive equilibrium conditions of production? If yes, are they
 6

seen in the third world countries like India? If not, what purpose is achieved?
Conceptually, weightage system is justified and without weightage distorted picture is
likely to be painted. Therefore, unless weighting problem is imaginatively and adequately
solved, empirical analysis will not reveal true state- of affairs.

Problems in Evaluating Inputs in Flow Terms:

In productivity analysis, it is widely recognized that input should be in flow terms. Often
researchers reduce capital stock for capacity under-utilistion(19). When capital stock is
adjusted for capacity under-utilisation, logic demands that labour too is to be adjusted.
When labour is employed on permanent basis supported by trade unions, will the labour
not susceptible to under-utilisation ? Alternatively, we ask the question, is it necessary to
adjust the inputs (capital and labour) for levels of utilization? If yes, how to calculate the
levels? This itself is a major area of research and an individual researcher cannot think of
investing his limited resources on measuring reasonably capacity utilistion. Vexed with
the problem Solow in his study (1957), simply adjusted capital by (labor) un-employment
rate, assuming capital too suffers unemployment at the same rate as labour. However,
Kendrick (1960) did not adjust capital for under utilization, counting this as an aspect of
efficiency with which capital is used.

Limitations of Popular TFP Analysis:

Although there has been out-pouring of studies on total factor productivity (TFP) and
production function ever since the publication of Abromovitz’s work( 20) in 1956, all
the studies suffer from serious infirmities. Despite many improvements in methodology
and quantification made by quite a few learned analysts, the studies suffer from major
limitations, especially those pertaining to 3rd world countries. TFP studies are rooted in
the Neo classical world of competitive conditions, producer equilibrium, application of
marginal productivity theory, constant returns to scale, imagination of well-behaved
production function, neutral technical progress, technical progress viewed mainly as
disembodied unitary or constant elasticity of substitution, production with variable
proportions, etc. we are aware that some of these assumptions are relaxed in a few
studies. However, still implicitly major loaded assumptions are made in the recent
studies. A look at the structure and functioning of Indian economy and the enterprises
clearly tells how far removed these assumptions are.

It is unfortunate that researchers after admitting the serious limitations, still some how
feel that Indian production units are nearer to the Neoclassical World and oblige
neoclassical ideal postulations. They mysteriously forget what they stated earlier and later
boldly come out with policy suggestions, as if the assumptions are quite reasonable.
Sadly the researchers seized with the Indian problems with some social purpose take
recourse to Neoclassical format (which is mainly technocratic) simply because the
Econometric/Statistical kit-box can be easily made use of? For realistic analysis, as far as
possible, we should attempt to move away from production function analysis and popular
traditional method of constructing total factor productivity indices.
 7

III
An Alternative Approach:

We now propose an alternative method to compute partial productivities. We adopted

here GVA concept but considers surrogates for the two inputs of labor and capital Viz.,
Human wage and Machine Wage. The use of these surrogates is in like with post –
Keynesian and Sraffas thinking. Machine and Human wages resemble what Sraffa says”
Fuel for the engines or the feed for the cattle”. To conserve space, we skip off details of
computation of human and machine wages. Admittedly, our alternative too suffers from
limitations. Nonetheless, the analysis is nearer to the theoretical advancements and real
state of affairs in the third world countries like India.

An Empirical Illustration:

To conserve space, we now attempt to compute only partial productivities (but not total
productivity) by using two methods viz Conventional, Surrogate partial productivities or
practical Businessman methods. Using conventional and alternative methods, trends in
partial productivities may be visualized from Tables.

Details of output per unit of labor and output per unit of capital along with their indices
as per conventional method is given in table 1. From the scrutiny of the Indices, it is clear
that trends are un-even, but growth trends in labor productivity(26.8%) were relatively
higher. Taking the 10-year period as a whole (1991-2001), we found that growth is
relatively less in capital productivity(11.4%) . Thus judged on the basis of changes in
GVA/K, the performance of Coal Industry was rather pathetic. It may be due to under
utilization of capital assets. Although, we have not estimated levels of utilization, our
interviews held with the officials gave a reasonable evidence to suspect under utilization
of capital assets. Further, less growth in capital productivity reflects the bias in technical
progress in favor of more capital-intensive techniques.

Table 1: Trends in behavior of Partial Productivities

(Conventional Method)
Year GVA Labor Capital GVA/L GVA/K Indices 1991-92
(in ‘ooo’s) (in‘ooo (in ‘ooo’s) ratio ratio of
’s)
GVA/L GVA/k
1991 2676391 116 12365010 23072.34 0.216 100.0 100.0
1992 4228928 115 14383894 36773.29 0.294 159.4 135.8
1993 6378108 115 17183065 55461.81 0.371 150.8 126.3
1994 7340905 115 24595393 63833.96 0.298 115.1 80.4
1995 6319930 114 30563667 55437.98 0.207 86.8 69.3
1996 8001625 115 33110408 69579.35 0.242 125.5 116.9
1997 12100535 112 35174776 108040.49 0.344 155.3 142.4
1998 12597854 109 36334158 115576.64 0.347 107.0 100.8
1999 15942959 108 37691244 147619.99 0.423 127.7 122.0
2000 17820347 106 38670275 168116.48 0.461 113.9 108.9
Growth 25.6 -0.9 14.2 26.8 11.4
 8

Rate
Source: Records & Accounts of Singareni Collieries company Ltd.

Table 2: Surrogates of Behavior of Partial Productivities

(Alternative method)
Year GVA Human Machine GVA/HW GVA/MW Indices 1991-92
Wage Wage of GVA/MW
GVA/HW
1991 2676391 3895089 1667877 0.687 1.605 100.0 100.0
1992 4228928 4496934 2052589 0.940 2.060 136.9 128.4
1993 6378108 5055311 2466910 1.262 2.585 134.2 125.5
1994 7340905 5673216 2646193 1.294 2.774 102.6 107.3
1995 6319930 6412676 3032241 0.986 2.084 76.2 75.1
1996 8001625 8428906 3666827 0.949 2.182 96.3 104.7
1997 12100535 8673869 3938141 1.395 3.073 147.0 140.8
1998 12597854 9023263 4079793 1.396 3.088 100.1 100.5
1999 15942959 10547636 4244155 1.512 3.756 108.3 121.7
2000 17820347 14708961 4356829 1.212 4.090 80.2 108.9
Growth 25.6 16.3 11.5 9.1 12.5
rate
Source: Records & Accounts of Singareni Collieries company Ltd.

Partial Productivities based on alternative method in table 2, depicts that the labor
productivity (GVA/HW) recorded lower growth(9.1%) rate compare with the capital
productivity (GVA/MW) 12.5%. One general conclusion is that when inputs are
measured in flow terms using proxies, performance of Singareni does not look as bad as
when they are measured conventionally(except labour productivity). We concede that our
surrogates need refinements. Notwithstanding the deficiencies, we hold that labor and
capital productivities and capital intensities are to be expressed in flow terms for
meaningful assessment of the performance.

A Summing Up:

We now wind up the discussion by noting end points, although everyone aims at higher
productivity, there is ambiguity as to what constitutes productivity, how to measure it,
and how to detect empirical investigation and choose concepts which are nearer to
theoretical advances and real third world economies. Empirical investigation of
productivity trends both at micro and macro levels poses serious conceptual and practical
problems. Researchers has to make compromises in quantifying the variables and should
attempt to move nearer to the real world of state affairs. Keeping this in view, we scanned
methodologies for quantifying inputs and outputs, and computation of partial and total
productivity indices. Conventional methodology molded in neoclassical format is not
relevant to evaluate productivity in third world economies due to unrealistic assumptions.
We suggested one alternative, which is premised on the Post-Keynesian strand of
thinking. For empirical illustration, the Singareni company episode was considered. For
comparative purpose, conventional as well as alternative measures were computed and
examined. Though our alternative is not free from limitations, but quite relevant to
practical businessmen.
 9

References:

1. Lutz FA & Hague DC: Theory of Capital, Macmillan, London,1961

2. Barna T: “On Measuring Capital “ in FA Lutz (eds)Theory of Capital. Opcit p.95
3. Harcourt GC: “Capital and Growth, Penguin, Hermondsworth, (Eds) 1971 pp9-41
4. Joan Robinson: Accumulation of Capital, Macmillan, London, 1976. P 32
5. Sraffa P: Production of Commodities By means of Commodities, CUP, London,
1960 part III.
6. kaldor N: A model of Economic Growth in Essays on Economic Growth and
Stability, Gerald Duckworth, London, 1960, p 269
7. George Rosen: Industrial Change in India, Asia Publishing House, Bombay, 1959.
p.42
8. Abromovitz M: Resources and output Trends in the US Since 1870, American
Economic Review, May 1956
9. Solow RM: Technical Change and Aggregate Production Function; Review of
Economics and Statistics, 39(3), 1957
10. Denison EF: the Sources of Economic Growth in the US and the Alternatives
before US, Committee for Economic Development, Washington DC, 1962
11. Hulten CR: On the Importance of productivity change’ American Economic
Review, 69(march) 1979
12. Gallop FM: Growth accounting in an open Economy in A Dogramci (Ed),
Developments in Econometric Analysis of productivity: Measurement and
Modeling Issues. KN Publishing, Boston, 1983
13. Solow RM: Investment and Technical progress” in k Arrow. S karton and P
Sraffas (Eds) Mathematical methods in the Social Sciences, Stanford, 1960
14. Grilliches Z: the Sources of Measured productivity Growth: US Agriculture
‘Journal of Political Economy, 71, 1963
15. Denison EF: Measurement of Labour input: some questions of definition and
Adequacy of Data, Output Input and Productivity Measurement. Studies in
income and Wealth, NBER, Princeton, 1961
16. Kendrick JW: productivity Trends in the US NBER, Princeton, 1961
17. Solow RM: Technical change and Aggregate production” opcit
18. Kendrick JW: opcit
19. Goldar BN: Productivity Growth in Indian Industry, Allied, Delhi 1985
20. Abromovitz M: Resources and Output Trends in the US since 1870” American
Economic Review, 1956.