Trade Liberalisation and Productivity Growth in Manufacturing: Evidence from Firm-

Level Panel Data

Author(s): Pulapre Balakrishnan, K. Pushpangadan and M. Suresh Babu
Source: Economic and Political Weekly, Vol. 35, No. 41 (Oct. 7-13, 2000), pp. 3679-3682
Published by: Economic and Political Weekly
Stable URL: http://www.jstor.org/stable/4409837
Accessed: 01-07-2017 19:24 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend access to
Economic and Political Weekly

This content downloaded from 138.246.2.11 on Sat, 01 Jul 2017 19:24:53 UTC
All use subject to http://about.jstor.org/terms
 Trade Liberalisation and Productivity
Growth in Manufacturing
Evidence from Firm-Level Panel Data
Using panel data comprising firm-level information drawn from groups within manufacturin
industry which have experienced the most significant tariff reduction, this study investigates t
trend in productivity growth since 1988-89. The sanmple of 2,300 firms and 11,009
observations, spanning the period 1988-89 to 1997-98 is very likely the largest assembled
the purpose thus far. We find no evidence of acceleration in productivity growth since the
onset of reforms in 1991-92. The result is evaluated in relation to the changes till date in the
policy regime in the Indian economy.

PULAPRE BALAKRISHNAN, K PUSHPANGADAN, M SURESH BABU

In a review of research on total factor tors where trade has been liberalised. While and productivity growth in different econo-
productivity growth (TFPG) in manu-we are aware of many dimensions to trade mies for different periods [Harrison 1994;
liberalisation, for the purpose of this study Srivastava 1996; Krishna and Mitra 1998],
facturing industry in the 1980s we had
observed that there remained a puzzlewe in define trade liberalisation as signifi- and we have proceeded accordingly.
cant reductions in the tariff rate.
that two differing methods of estimation Specify the production function for firm
- the growth accounting method and the To investigate the existence of a shift i in industry j at time t as:
econometric one - produced divergent in the growth of productivity since theYijt = AjtfitG(Ljt, Kijt, Mijt) ...(1)
results [Balakrishnan and Pushpangadan introduction of trade reforms in the Indian
where Y, K, L and M stand for output,
economy, data for a panel of 2,300 firms capital, labour and materials inputs, re-
1998]. In this paper we do not address this
issue which remains of some importance spread over five industry groups (Appen- spectively, Aj is an industry-specific
dix) at the two digit level of the NIC 1987,
to scholarship on the estimation of produc- index of Hicks-neutral technical progress
tivity growth in India. Instead, we moveyielding over 11,009 observations was
and fit is a parameter allowing for firm-
on to study the 1990s. This follows fromassembled from the data base on electronic
specific differences. Totally differentiat-
the fact that both our perception and medium
our (PROWESS) of the Centre for ing (1) and dividing through by Y, we have
own priorities have evolved since our Monitoring
last the Indian Economy (CMIE).
Onasthe basis of the record of tariff reduc-
(dY/Y)ij = (8Y/6L)(dL/Y).t
paper. Principally, it has struck us that
+ (8Y/6K)(dKY)ijt
the best received work had posited a tionre- since 1991 the industry groups chosen + (8Y/6M)(dM/Y).j
lationship between TFPG and what may were machinery, transport equipment and
+ (dA/A)jt + (df/f)it ...(2)
be referred to as liberal policy regimes, parts,
the textiles, textile products and chemi-
From the first order conditions for profit
1990s are a more appropriate period cals.
for As far as possible, industries sub-
maximisation of a firm in Cournot equi-
a test of the relationship between TFP
jected to significant tariff reductions are
librium the expression for the physical
growth and the change in the policy included.
re- The data on tariff reductions
marginal product(s) can be written as:
used
gime. For after all in the 1980s we had 'not for this study is presented in
seen nothing yet'. The changes initiated Balakrishnan, Pushpangadan and (6Y/6L)jt = (w/p)jt{ 1/[1+(sij/ej)]} =
in 1991 dwarf anything by way Suresh of Babu (2000; henceforth BPS). The (w/p)jtgij ...(3a)
liberalisation that may have taken place period 1988-89 to 1997-98 was chosen for (6Y/K)it = (r/p)jt 1/[ l+(s /ej)]} =
during the preceding decade. If this viewthe study, 1997-98 having been the last
is acceptable, and to us it appears emi- financial year for which data was available
(r/p)jtltij ...(3b)
nently so, the data of the 1990s would at the time of commencement of the study. (6Y/6M)ijt = (n/p)jt{ l/[l+(sij/ej)]} =
serve as a better test bed than any otherThe study investigated a shift in produc- (n/p)jtjlij ...(3c)
period thus far. tivity growth from the year 1991-92. where p is the product price, w, r and n
In'this paper we present results of a test are the price of labour, capital and mate-
Model
for a shift in productivity growth since rials, respectively, si is the market share
1991. We would expect any shift in pro- From Hall (1988) we have a methodo- of firm i in industry j, and t is the mark-
ductivity to occur in those sectors logy of whereby estimation of a single equa- up (price-marginal cost ratio).
manufacturing where the reforms have beention yields both an estimate of the price-Anticipating the estimation to follow,
most pronounced. It is widely held that marginal
the cost ratio and of productivity.which takes the form of estimating
This
defining character of these reforms is the methodology has been widelyproduction functions for whole industries,
greater openness of the Indian economy. applied in empirical analyses of the con-
it is assumed that the mark-up varies across
Therefore we have focused on those sec- industries alone, and not between firms.
sequence of trade reforms for competition

Economic and Political Weekly October 7, 2000 3679

This content downloaded from 138.246.2.11 on Sat, 01 Jul 2017 19:24:53 UTC
All use subject to http://about.jstor.org/terms
Now, substituting (3a)-(3c) into (2) and re- an impact of trade reforms on the scale are mutually consistent. First, we have
arranging terms, we have: parameter in some economies it would be presented here TFP estimates based on
advisable to allow for it in estimation. This econometric estimates of firm-level data
(dY/Y)ijt = g.[(wL/pY)(dL/L)
+(rK/pY)(dK/K) is done by adding an interactive slope from the CMIE. Secondly, one of us has
+(nM/pY)(dM/M)]ijt dummy to 'dk' which yields: produced estimates of TFPG for industry
+ (dA/A)it + (df/f)it ...(4) dyijt = Boj + Bjdxijt + B2j[Ddxijt] aggregates from ASI data using the growth
Denote the factor shares (wL/pY), (rK/pY) accounting method [Suresh Babu 2000].
+ B3dkij + BjD + B5j[Ddkijt]
and (nM/pY) as ac, ak and cm, respec- However, the coverage did vary purpo-
tively, (4) may be re-written as:
+ git + it...(8) sively. Both the studies use gross produc-
(dlnY)ijt = gj[oal(dlnL) + cm(dlnM) Estimation tion as the output variable precluding any
debate about the measurement of value
+ ok(dlnK)]ijt + (dA/A).t
Equation 8, was estimated by OLS with
+ (df/f)it and without group dummies and by fea-
added likely to emerge otherwise. Neither
Denoting the sum of factor shares as of these studies reveal an acceleration in
sible generalised least squares, yielding
P/p., where 3 is the returns to scale para- productivity growth since 1991. Note, then,
the OLS, 'within' and 'between' estima-
meter, we can rewrite (5) as the production that the problem of divergent trends ob-
function in intensive form: tors, respectively. The Hausman Test (Chi served in the case of the estimates for the
square: 384.5) favoured the within trans- 1980s does not exist in the case of the
dyij= j[aCdl+amdm].t + (1j-l)(dK/K).t formation applicable to a fixed effects
+ (dA/A).t + tdf/f)jt . model. The estimation of firm-level pro-
estimates provided by us for the 1990s.
where the variables y, I and m equal What is of greater interest to us though is
duction functions is particularly prone to
In(Y/K), ln(L/K) and In(M/K), respectively.' to find the original result in Balakrishnan
the classical problem of the endogeniety
In an estimate of (6), the term (dA/A)jt, and Pushpangadan (1994) holding good
of the regressors [Grilliches and Mairesse
which can be thought of as the rate of across time periods, data sources, cover-
1998]. We take the view that in the short
productivity growth for industry j, is cap- age, definitions of the output variable and
run the capital stock is not a variable for
tured by a constant term Bq. Next, (df/f)it the methodologies of estimation.
the firm while employment is a quasi-fixed
may be decomposed into a firm-specific As far as we are aware, there exist two
input [Klette 1999:462] - as opposed to other studies of TFPG in Indian manu-
effect git and a disturbance term uit. The
capital services and manhours, respectively,
resulting specification can be used to test facturing that use firm-level panel data and
which may be varied with current output are based on the econometric estimation
for changes in the extent of competition
- allowing their use as instruments for the
and in productivity growth due to trade of the production function. The near ex-
reforms. The change in the price-cost ratio
endogenous input 'materials'. Thus dxij, emplary work of Srivastava (1996), how-
was instrumented by its predicted value
can be investigated by adding an interac- ever, looks at a period prior to 1991, which,
with the prediction equation having con-
tive slope dummy to the sum of the changes due to the subsequent change in the policy
tained the current capital stock and employ-
in variable inputs in (6) and a shift in the regime, is now only of academic interest.
ment per period. The instrumental variables
level of productivity growth can be accounted On the other hand, Krishna and Mitra
estimate of(8), now without the time dummy
for by an intercept dummy. Incorporating (1998) study productivity growth after 1991
for the mark-up, maintaining a fixed ef-
these would give the estimating equation: and report statistically significant increase
fects model, is presented below.2 The esti-
in productivity growth in four industry
dYijt = Boj + Bljdxijt + B2j[Ddxijt] mated coefficient (B4) in equation (8), the
groups chosen by them. However, we
+ B3jdkijt + B4jD + git + uit ...(7) time dummy of the intercept, as reported would like to express some reservations
below indicates no improvement in pro-
where B I = p, B3 = (p-1), dx = [otdl+Oamdm], regarding their data base. We find that the
ductivity growth. Indeed it signals a sta-
dk = dK/K, and D is a dummy accounting construction of the input variables, both
tistically significant decline in the growth
for the (policy) regime in place during any capital and labour, in their work leaves
of total factor productivity after 1991-92.
particular historical phase. Given our inter- something to be desired. Here it is of
est in this study the dummy takes the value B B3 B4 B5 interest to note the comment of Grilliches
zero prior to 1991 and one from that date on. (1994) that measurement is one of the
1.07 -0.27 -0.01 0.07
Moreover, in this study we remain inter- single most important problems in the
(0.015) (0.024) (0.001) (0.025)
ested only in the behaviour of productivity estimation of productivity. He has gone on
growth. to state
(Standard errors in parentheses; all coef- eight sources of measurement error
Notice from (6) that there is a firm- ficients are statistically significant.)among which is "the improper measure-
specific factor in the form of fit to take care The above regression is based ment on the of inputs over time". In our view, a
of. We follow standard practice and allow assumption of 'poolability' of the separate
major contribution of our paper is to move
for the possibility of this standing either for industry group panels in the data set.a little
There-closer to the best practice, without
fixed or for random effects. In this study of course satisfying all of Grilliches'
fore, a test for poolability was conducted
separate estimates under each assumption and resulted in an F-statistic of 1.1 requirements.
imply- Interested readers may wish
are presented and the Hausman test used ing that the null hypothesis that the
to compare the methodology of variable
to choose among these [Hsiao 1986].. data may be pooled cannot be rejected
construction used by us, provided in the
Finally, if B3 is not significantly dif- [Baltagi 1995:49]. Appendix and that used by other authors.
ferent from zero in an estimate of (7) we That the reforms initiated in 1991 may
Conclusion
may conclude that the technology is not have led to noticeable productivity
characterised by constant returns to scale. We would like to draw attention to the gains does not particularly surprise us. In
Further, since there has been evidence of fact that we now have a set of results which our earlier work [Balakrishnan 1996;

3680 Economic and Political Weekly October 7, 2000

This content downloaded from 138.246.2.11 on Sat, 01 Jul 2017 19:24:53 UTC
All use subject to http://about.jstor.org/terms
Balakrishnan and Pushpangadan 1998] we output and 70 per cent of the value added.
components along with depreciation. From
have referred to the absence of conclusive We commenced with a sample of 3,000this investment can be obtained as the
arguments linking a liberal policy regime firms from industry groups with signifi-
difference between the current and lagged
and productivity growth in theory, or of cant tariff reduction. Firms for which values. This enables one to use the per-
any overwhelming evidence of the same. unacceptable values were recorded for petual inventory method (PIM) to arrive
Interestingly, this is also a view held by at an estimate of capital stock. However,
certain variables, such as negative or zero
observers who are not mainstream econo- values for fixed assets, and those for which
straightforward application of PIM is not
mists. Thus we have an observation that a continuous time series was unavailable possible as the balance sheet figures for
was made as early as on the eve of the firstwere subsequently excluded from the
capital are at historic cost, which has to
be converted into asset value at replace-
general election in the country held aftersample. As it is not mandatory for the firms
the launching of the economic reforms:to report the balance sheet and other details ment cost. To appreciate this, note that the
"One kind of reforms pertain to a fewto the data collecting agency, information perpetual inventory method implies
crucial decisions: such as the decision to
for some years was missing for certainPt+l Kt+1 = Pt+l1(-6) PtKt + Pt+l It+l
firms. This does not mean that the firms
reduce tariff barriers, to delicense indus- where K is the quantity of capital, P its
have exited from the industry. Inference
tries and dismantle the licence-permit raj. price, I is investment and d the deprecia-
Another, more subtle reform measure drawnis from the study, thus, must be lim-tion rate.
the setting up of new institutions and ited
of to the public limited companies and However, this procedure is valid only if
the adjustment of existing ones. These cannot be used to understand the entry and the base year capital stock can be written
exit behaviour of firms. The final data set
include independent regulatory structures as PoKO. But this is not the case as in the
to coagulate (sic) investment, to improvecomprised information for 2,300 firms forbase year too the firms' asset mix is valued
the efficiency of labour and capital and the
in 10-year period 1988-89 to 1997-98.at historic cost. The value of capital at
general to step up on productivity and For the precise number of firms and thereplacement cost for the base year is ar-
efficacy" [Mishra 1996]. We believe that number of observations in the separaterived at by revaluing the base year capital.
this an apposite appraisal. industry groups see BPS (2000). The method adopted for this generally
Our understanding of the relevant theory involves an element of arbitrariness and
Variable Construction
is that the route from increased competi- the most one can arrive at is an approxi-
tion and/or the liberalisation of trade to As the balance sheet data is provided mation.
in Our estimation rests on three crucial
higher productivity growth is less than nominal terms the conversion of these assumptions.
clearly defined. So much so that a reason- values into a measure of the underlying (a) We treat 1997 as the base year due
able position to adopt, it appears to us, quantities is the principal data processing
to the availability of a greater number of
would be that it is an empirical issue. Theinvolved in the estimation of productionobservations if we take a later year in the
discovery here of the absence of any signi- sample as the base. We assume that the
functions. This was done by deflating these
ficant improvement in productivity growthnominal values using appropriate price earliest vintage in the capital mix dates
since 1991-92 lends itself to two interpre- from 1977, or from the year of incorpo-
deflators. We discuss the procedure in detail.
tations. Either the period studied is too Output: CMIE provides information ration if it is after 1977. The year 1977
soon after the launching of reforms for on the value of output of firms in an itself
in- was chosen because the life of
there to have emerged the, allegedly inevita-dustry group. This was deflated by the
machinery is assumed to be 20 years as
ble, increase in the rate of productivity industry specific wholesale price index.
noted in the 'Report of the Census of
The source of the price index usedMachine
growth or the policy instruments employed is Tools 1986' of the Central
are inadequate to the task. However, we'Index Numbers of Wholesale Prices in Machine Tools Institute, Bangalore
India,
would like to propose a furtherdimension to base 1981-82=100', Ministry('National
of Accounts Statistics: Sources and
the interpretation of this result. The re-
Industry, government of India. Methods', Central Statistical Organisation,
forms have thus far remained mainly macro- Capital: A daunting task awaits the
New Delhi, 1989).
economic in nature. Productivity growth empirical researcher setting out to measure(b) The price of capital changes at a
may well have a strong microeconomic -the capital stock. While we are aware of
constant rate 7t=Pt/Pt,1 from 1977 or the
foundations which remain to be addressed. the controversies and debates with regardyear of incorporation up to 1997. The
values for n were arrived at from a series
to this issue we believe that the empirical
Appendix of price deflators constructed from CSO's
procedure applied in the present study lives

Data Base up to the task of capturing this inputdatato on gross fixed capital formation
the extent of providing a reasonable esti-
published in various issues of the National
The data for the present exercise ismate of the variable. While most of the
drawn Accounts Statistics (NAS).
from the CMIE's database PROWESS. current studies use the book value of fixed (c) Similar to the price of capital we
CMIE provides information for approxi- assets deflated by an investment goods assume that investment also changes by a
mately 7,000 firms registered withdeflator-
the mostly the wholesale price index constant rate g = It/It 1. The growth of fixed
Bombay Stock Exchange, limiting itself fortomachinery and machine tools, this is capital formation at 1980-81 prices, taken
public limited companies. It shouldplain be wrong, for it makes no allowance for from various issues of NAS, is applied in
noted that public limited companies in
vintage. The methodology we have used the case of all the firms. Depending on the
India account for almost 50 per cent is based
of on that-in Srivastava (1996). It year of incorporation firms will have dif-
employment and 80 per cent of the fixed is explained below. ferent annual average growth after 1977.
CMIE provides information from bal-
capital of the private sector factories while Using the values of n and g, we arrive
contributing to around 60 per cent of thesheets on gross fixed assets and their at a revaluation factor RG to arrive at a
ance

Economic and Political Weekly October 7, 2000 3681

This content downloaded from 138.246.2.11 on Sat, 01 Jul 2017 19:24:53 UTC
All use subject to http://about.jstor.org/terms
 measure of the capital stock at replacement ratio of the scale parameter to the mark-up see
revaluation factor for the net capital stock.
Harrison (1994). For the derivation of (6)
cost for the base year. This is done as This demands the use of accounting de-
incorporating this see BPS (2000). Note that
follows. Suppose the gross fixed assets at preciation rates as well as economic de-writing (6) implicitly assumes Cobb-Douglas
historic costs can be defined as preciation rates. Economic depreciationproduction technology.
GFAth = Ptit + Pt- It- + Pt-2'It-2+ rates can be exogenously determined, 2 All results are available upon request from the
which can be rewritten as authors.
endogenously determined or arrived using

FAt t t (l+g)(l+7l)-l
the one-hoss-shay model. The first one
GFAh = PI ( l +g)( l+:) implies borrowing a set of estimates for
References
some other economy, the second one makes
Balakrishnan, P (1996): 'Economic Reforms,
and similarly gross fixed assets at replace- use of the assumption of straight lineCompetition and Productivity Growth in India:
ment costs can be written as depreciation and the third assumes that theA Panel Study of the Manufacturing Sector',
GFAr = PIt + P I PI(t2) +..... depreciation during the life of a machineBackground paper forThe World Bank's 'India:
which can be rewritten as is zero and is 100 per cent at the end ofCountry Economic Memorandum', mimeo,
Indian Statistical Institute, Delhi Centre, New
the life of the machine. Limitations regard-Delhi.
GFA r = PtIt (I+g) ing the three measures are well recognised
Balakrishnan, P and K Pushpangadan (1994):
in the literature. Ambiguity exists on the'Total Factor Productivity Growth in
treatment of depreciation due to obsoles-Manufacturing Industry: A Fresh Look',
then the revaluation factor RG , defined as
the ratio of the value of the asset at cence and depreciation due to physicalEconomic and Political Weekly, 29, 2028-35.
- (1998): 'What Do We Know about Productivity
deterioration. This poses further problems,
replacement cost to the value of the asset Growth in Indian Industry?', Economic and
as one has to deal with the concepts of
at historic cost will be, if the earliest vintage Political Weekly, 33, 2241-46.
of capital dates back infinitely, obsolescence and aging, retirement and Balakrishnan, P, K Pushpangadan and M Suresh
discarding (mortality) and the service lifeBabu (2000): 'Liberalisation, Competition and
(l+g)(l+7r)-l
RG = of the capital stock. Data required toProductivity Growth in Indian Manufacturing',
g(l+7l) mimeo, Indian Institute of Management,
untangle these issues are more than what
Kozhikode and Centre for Development
In this study we assumed that the capital is available at present. Thus we preferStudies, Thiruvananthapuram.
stock has finite economic life. Now the gross values to net values. Baltagi, B H (1995): Econometric Analysis of
revaluation factor becomes Labour: The expenditure on wages andPanel Data, John Wiley, Chichester, UK.
salaries was converted into a measure of Dennison, E F (1967): Why Growth Rates Differ:
G [(l+g)t+-1] (1+t)'t [(l+g)(l+rr)-l] Post-War Experience in Nine Western
labour input of firms by administering an
Countries, Brookings Institution, Washington, DC.
g([(l+g)(l+r)]T+1 - 1) estimated average total compensation to
Goldar, B N (1986): Productivity Growth in Indian
where t is the life of the machine. labour in the firm's industry for that year. Industry, Allied Publishers, New Delhi.
Using the revaluation factor thus ob-The resulting measure may be seen as Grilliches, Z (1994): 'Productivity, Research and
labour expressed in 'efficiency units'. The Development, and the Data Constraint',
tained we convert the capital in the base
American Economic Review, 84, 1-23.
average compensation by industry was
year into capital at replacement costs at Grilliches, Z and J Mairesse (1998): 'Production
current prices. We then deflate this valuecomputed by dividing each industry's total Functions: The Search for Identification' in
emoluments by total labour hours from the Steinar Storm (ed), Econometrics and
to arrive at a measure of the capital stock
in real terms for the base year. The priceAnnual Survey of Industries (ASI). As, at Economic Theory in the Twentieth Century:
the time of our investigation, ASI data was The Ragnar Frisch Centenary Symposium,
deflator used is the price index for machin-
Cambridge University Press, Cambridge.
ery and machine tools as plant and machin-available only up to 1995-96 we have used
Hall, R (1988): 'The Relation between Price and
ery account for 71.5 per cent of GFA (RBI extrapolated values for subsequent years. Marginal Cost in US Industry', Journal of
Bulletin, 1990, 44, 3). Subsequent years' Materials: The value of the materials bill Political Economy, 96, 921-47.
investment, GFAt - GFAtl, is added to thewas deflated by a materials input-price Harrison, A (1994): 'Productivity, Imperfect
index. The input-output coefficients for Competition and Trade Reform: Theory and
capital stock existing at every time period
Evidence', Journal of International
using the perpetual inventory method. 1989-90 have been used as weights to Economics, 36, 53-73.
It should be noted that we have used combine the wholesale prices of the rel- Hsiao, C (1986): The Analysis of Panel Data,
gross values of capital in our estimates.evant materials. The source of the weights Cambridge University Press, Cambridge.
Dennison (1967) argues that a correct is CSO's input-output table for 1989-90 Klette, T J (1999): 'Market Power, Scale Economies
and the relevant price indices were taken and Productivity: Estimates from a Panel of
measure falls somewhere between the gross
Establishment Data', The Journal of Industrial
stock, and the net stock, advocating the from
use 'Index Numbers of Wholesale Prices Economics, 67, 451-76.
of a weighted average of the two with in India, base 1981-82=100', ministry ofKrishna, P and D Mitra (1998): 'Trade
industry, government of India. {13
higher weight for the gross as the true value Liberalisation, Market Discipline and
is expected to be closer to it. Empirically Productivity Growth: New Evidence from
implementing this runs into a problem in Notes
India', Journal of Development Economics,
56, 447-62.
the Indian context as the figure for capital
[Foradvice and/ordiscussion we thank Sanghamitra Mishra, J (1996): 'Towards a Consensus on Power',
consumption is difficult to arrive at.
Das, K L Krishna, Arunava Sen and B P Vani. The Economic Times, New Delhi, April 22.
Moreover, one often encounters the ques-
Responsibility is ours. For support and financial Srivastava, V (1996): Liberalisation, Productivity
tion of the reliability of the depreciation
assistance we thank the Centre for Development and Competition: A Panel Study of Indian
values reported by the firms as mostStudies,
of Thiruvananthapuram and the Indian Manufacturing, OUP, Delhi.
Institute of Management, Kozhikode, respectively.] Suresh Babu, M (2000): 'Total Factor Productivity
these are calculated as per the allowances
in Manufacturing: Some Further Results',
by the income tax authorities. Another
1 There are two steps in this derivation. For the mimeo, Centre for Development Studies,
related problem is the computation of the
summation of factor shares being equal to the Thiruvananthapuram.

3682 Economic and Political Weekly October 7, 2000

This content downloaded from 138.246.2.11 on Sat, 01 Jul 2017 19:24:53 UTC
All use subject to http://about.jstor.org/terms