Economic Reform and TFP Growth in Indian Manufacturing Industries

Generally, SEC increased in the chemical, non-metal, and transport industries over the sampling years, but decreased in the other industries (food, metal, machinery, and textiles industr[r]

Economic Reform and Total Factor Productivity Growth in Indian Manufacturing Industriesrode_652152 166 Sangho Kim and Muthusamy Saravanakumar*

Abstract

A stochastic frontier production function model is applied to Indian manufacturing industries, to decompose the sources of total factor productivity growth into technical progress, technical efficiency, scale efficiency, and allocative efficiency Empirical results based on data from 2000 to 2006 suggest that increased investment needs time to deliver increased productivity and efficiency, because new technology combined with fresh investment requires higher numbers of skilled workers, better managerial practices and an advanced input mix, all of which generally take time to develop Thus, the Indian economy must boost technical efficiency by providing skilled workers and high quality managers to further economic reform

1 Introduction

The new industrial policy and the process of economic reform in the Indian economy started in the late 1970s, as the Indian government deregulated the barriers to entry and expansion However, a structural break from the previous policy regime occurred in 1991, when the government restructured the domestic capital market by liberalizing interest rates and lifting the restrictions on investment by large industrial houses and foreign controlled companies, under the MRTP (Monopoly Restrictive Trade Policy) Act This reform process of 1991 also reduced administrative controls and barriers, which had acted as obstacles to the free flow of exports and imports (Ahluwalia, 2002) With liberalization, the Indian industrial sector underwent significant changes Most of all, the reform enhanced competition among domestic firms, not only by increasing entry threats to existing firms but also by making the environment equally competitive for new firms, with improved potential resources and opportunities Thus, the reform made raising productivity a key survival criterion (Bhaumik et al., 2009)

Empirical studies, however, show mixed results regarding the impact of trade liber-alization on total factor productivity growth (TFPG) in the Indian manufacturing industry in the 1990s Some studies report that the liberalization policies improved the productivity of the manufacturing industry (Chand and Sen, 2002; Unel, 2003; Driffield and Kambhampati, 2003) However, many other studies have found a negative trend or no significant improvement in productivity growth since the onset of economic reform in 1991 (Trivedi et al., 2000; Goldar and Kumari, 2003)

The slow TFPG during the 1990s is puzzling because the period witnessed a number of changes that were expected to have favorable effects on industrial productivity One possible explanation for the declining productivity growth notes the delay or failure of domestic firms in adapting to increased competitive pressure This suggests that Indian

* Kim: Department of International Trade, Honam University, Gwangju 506-714, South Korea Tel:+

82-62-940-5394; Fax:+82-62-940-5116; E-mail: shkim@honam.ac.kr Saravanakumar: Lecturer in Economics, Sree

Saraswathi Thyagaraja College (Autonomous), Pollachi—642 107, Tamilnadu, India E-mail: econsaravanan@ yahoo.com

manufacturing failed to catch up with the additional technological pressure created by the greater inflow of foreign capital after reform (Parameswaran, 2004) Another explanation is related to gestation lags This requires some consideration of the spurt in investment activity that occurred in the 1990s, in response to economic reforms (Uchikawa, 2002) Increased investment needs time to deliver increased productivity and efficiency, because the new technology that accompanies fresh investment requires higher numbers of skilled workers, better managerial practices, and an advanced input mix, all of which generally take time to develop

In this regard, this paper investigates the relationship between reform and produc-tivity using a dataset covering the period 2000–2006 to capture the overall effect of liberalization a sufficient number of years after the reform The paper contributes to the existing literature by providing productivity estimates based on firm-level panel data obtained from the CMIE (Centre for Monitoring Indian Economy) PROWESS database Most previous studies adopted growth accounting methods to estimate pro-ductivity growth in the Indian manufacturing industry by using the Annual Survey of Industries (ASI) data, which is aggregated at industry level.Thus, our study can comple-ment the studies using the ASI data by applying a stochastic frontier approach to estimate the firm-level productivity of the Indian manufacturing industries

Additionally, this paper adopts a stochastic frontier analysis to decompose produc-tivity into four components: technical progress (TP), technical efficiency change (TEC), allocative efficiency change (AEC), and scale efficiency change (SEC) By doing so, the paper estimates SEC and AEC arising from production scale and market distortions, respectively, which may affect TFPG significantly, especially in a rapidly changing economy like India Thus, the paper will broaden the scope of the existing literature based on stochastic frontier analysis, which decomposes productivity into TEC and TP (Kalirajan and Bhide, 2004; Madheswaran et al., 2007)

2 Overview of the Indian Manufacturing Sector

The Indian government initiated a significant economic reform by introducing a new industrial policy in 1991 The policy not only freed the Indian economy from unneces-sary bureaucratic control but also integrated the Indian economy with the world economy by removing restrictions on direct foreign investment and by liberating domestic entrepreneurs from the restrictions of the MRTP Act

After these measures were launched, industry and trade policy reforms were accel-erated, while public investment contracted sharply to reign in the fiscal imbalance Financing of industrial development was changed considerably by the financial sector reform that cut directed lending and abolished subsidized credit for designated sectors The economic impacts of reform can be easily observed from the transformation of the Indian economy since the 1970s There was a major shift away from agriculture towards the service sector and the industrial sector, especially after the 1991 reform During the 1990s, there was a sharp rise, by about percentage points, in the share of the service sector and an almost equivalent drop in the agricultural sector, with a continuing increasing trend for the industrial sector

The food industry accounts for about 8% of the GDP for the manufacturing sector in India The performance of the food processing industry is experiencing a 20% annual growth rate in 2010 The food processing industry represents significant value addition to agricultural production and also provides a link between the rural and urban eco-nomies India exported about US$4.5 billion worth of products in the category of food and processed food in 2001

The chemical industry is an integral component of the Indian economy, contrib-uting around 6.7% of the Indian GDP According to our sample, the industry has the largest capital intensity ratio among the seven manufacturing industries (see Table 1) This industry, including petro-chemicals and alcohol-based chemicals, has grown at a pace outperforming the overall growth of the Indian manufacturing indus-try However, the industry was burdened with heavy costs, being a raw material-intensive industry, and most of the materials used cost much more in India than elsewhere

The metal industry has the second greatest capital intensity ratio among the seven Indian manufacturing industries In 1990–91, six plants—five of which were in the public sector—had a total output of 10 million tons Another 180 small plants accounted for the rest of the country’s steel production—4.7 million tons Most of these firms were in the private sector The Indian steel industry has the world’s largest steel makers with ArcelorMittal and Tata Steel Also, India’s aluminium industry is booming, with the highest growth rates in the world, at more than 7% per year

The transport industry has grown at an annual rate of 6.43% during the period of 2007–10 The automotive industry’s contribution to the GDP in 2006–2007 was 5% In the same year, the automotive industry produced more than 11 million vehicles, regis-tering a growth of 13.56% and achieving a turnover of US$34 billion Regarding employment, the automotive industry provided direct and indirect employment to more than 13 million people in 2006–2007 However, there is also a large unorganized sector that contributes 30% to total employment, 13% to capital, and only 1.5% to total output in the automotive industry in India

The machinery industry is the fastest growing industry in India, with an average annual growth rate of 10.4% during 2007–10 The period from 1991/92 to 1995/96 witnessed a sharp 70% decline in the tariffs on capital goods After incurring a steep adjustment cost in the initial years, this industry, in fact, responded very positively, and successfully restructured, eventually experiencing very high growth rates in 1995/96 (24.8%) and 1996/97 (17.9%) The highest rates of growth of the industry coincided with the deepest cuts in import tariff, suggesting that the forces of competition and the access to foreign technology benefited the sector significantly

For the non-metal industry, India’s diverse and significant mineral resources render the nation the world’s leading producer of some minerals The share of the mineral sector in the GDP of the country is around 3.5%, while accounting for a 10% share in the index of industrial production Although 80% of the mines are in the private sector, 91% of production in terms of size comes from government-owned mining ventures Mining employs over million people

3 Decomposition and Functional Form Decomposition of Total Factor Productivity

A stochastic frontier production function is defined by

yit = (f x tit, )exp(−uit), (1) whereyitis the output of theith firm (i=1, ,N) in thetth time period (t=1, ,T),

f(·) is the production frontier,xis an input vector,tis a time trend index that serves as a proxy for technical change, andu30 is output-oriented technical inefficiency

Totally differentiating the logarithm ofy in (1) with respect to time, the change in production can be represented as

y TP jxj du dt

j

= +∑ε − ( / ), (2)

whereej=∂lnf/∂lnxjis the output elasticity of factor inputj, and a dot over a variable

indicates its rate of change Overall productivity change is affected not only by TP and changes in input use but also by the change in technical inefficiency

By substituting (2) into TFPG,TFP = − ∑y jS xjj, where Sj is input j’s share in

pro-duction costs,TFP is rewritten as

TFP TP du dt RTS jxj S x

j

j j j

j

= − ( / ) +( −1)∑λ +∑(λ − ) (3)

where RTS(= ∑jεj) denotes the measurement of returns to scale, and

λj=f xj j/∑l lf xl=εj/∑ =l lε εj/RTS The last component in (3) measures the inefficiency in resource allocation resulting from the deviations of input prices from the value of their marginal product Thus, in equation (3), TFPG can be decomposed into TP, the TEC (–du/dt), scale components ((RTS−1)∑jλjxj), and the AEC (∑j(λj−S xj)j) This decomposition draws on Kumbhakar (2000) and Kim and Han (2001)

Functional Form

The components of productivity change can be estimated within a stochastic produc-tion frontier framework, and the time-varying producproduc-tion frontier can be specified in translog form, as

ln ln ln ln

y x t x x

t

it j jit

j

T jl lit jit

l j

TT T

= + + +

+ +

∑ ∑∑

α α α β

β β

0

2

0 5 jj jit it it

j

tlnx + −v u , j l, =L K, ,

∑ (4)

whereyitis the observed output,tis the time variable, and thexvariables are inputs

Subscriptsjandlindicate inputs (j,l=L,K) The efficiency erroruis assumed to be

independent of the statistical error v, which is assumed to be independently and

identically distributed as N(0,σv2)

Following Battese and Coelli (1992), technical inefficiency is defined by

where the distribution ofui is taken to be the non-negative truncation of the normal

distribution, N(μ σ, u2), and h is a parameter that represents the rate of change in technical inefficiency

The maximum-likelihood estimates for the parameters of the stochastic frontier model, defined by (4) and (5), can be obtained by expressing the variance parameters in terms ofγ σ σ= u2/ s2andσs2=σu2+σv2

The technical efficiency level of companyi at time t (TEit) is defined as the ratio

between its actual output and its potential output:

TEit=exp(−uit) (6)

The elasticity of output with respect to thejth factor (input) is defined by

εj j αj βjl l β β

l j

jj j Tj

f x t x x x t j l L K

= ( ) = + + + =

≠

∑

∂ln , / ln∂ ln ln , , , (7)

The elasticity of scale is defined asRTS= ∑jεj, while the rate of TP is defined by

TP f x t t T TTt Tj xj j L K

j

=∂ln ( , )/∂ =α +β +∑β ln , = , (8)

4 Empirical Results Data and Variables

The study makes use of the firm level panel data of seven industries: food and bever-ages, textiles, chemical, non-metallic, metal, machinery, and transport equipment The firm level data were obtained from the electronic database PROWESS, and the sample consists of 1630 Indian manufacturing firms between 2000 and 2006, with a total of

11,410 observations.1Our sample accounts for about 71% of employment, 67% of the

fixed capital, and 75% of the value added in the Indian manufacturing sector during the sampling years

Gross value added, calculated by deleting total purchases of intermediate inputs from gross outputs, was taken as a measure of output, and was then deflated by the wholesale prices index of the respective industries, with the base of 1993/94=100

The PROWESS database provides total fixed assets net of accumulated depreci-ation, including capital work-in-progress and revalued assets, if any The total fixed assets were deflated by the wholesale prices index of machinery and machine prod-ucts, and thus real total fixed assets were included in the function as a measure of capital

The PROWESS database does not provide employment details To estimate the number of workers engaged in an industry, the average wage rate of the industry concerned was calculated from the ASI data for all years of the study The average wage rate was estimated by dividing the total emolument of the industry by the number of workers in the industry This average wage rate, obtained from the ASI data, was then used to divide the total wages and salary of each industry, in order to estimate the

number of workers in the industry.2At the time of this study, ASI data was available

only until 2004, and the average wage rate for the year 2005/06 was extrapolated

Labor cost (CL) consists of compensation to employees, and includes salaries and

costs (C) are calculated as the total sum of these factor costs (C=CL+CK), and the

factor share in total costs (SL,SK) is calculated as each factor’s share out of the total

costs (Sj=Cj/C,j=L,K)

For individual industry estimation, this study classifies sample firms into double-digit industries, according to the National Industrial Classification (NIC) Table presents summary statistics for variables used in the stochastic frontier production functions for India’s manufacturing sector

A huge difference in employment, capital stock and output is apparent in the manu-facturing sector, in which micro-sized firms operating with family labor coexist with mega-sized international conglomerates Of the seven manufacturing industries, the food industry is the largest in terms of value added, followed by the chemical and metal industries The chemical industry has the most employees, followed by the metal and machinery industries The chemical industry also has the greatest capital intensity ratio, followed by the metal and textiles industries

Empirical Results

Hypothesis testing The maximum-likelihood estimates of the parameters in the

trans-log stochastic frontier production function, defined by equations (4) and (5), are

obtained for each of the seven industries.3 All the estimates of g are statistically

significant at the 1% significance level

All the estimates ofhare negative, and all are statistically significant, except for the

non-metal, machinery, and transport industries A significantg, along with a negative

and significanth, implies the existence of technical inefficiency that increases over the years.4

The alternative hypotheses are tested using likelihood ratio tests.5The first

altern-ative hypothesis, that there are no technical inefficiency effects (H1:g =m=h=0) in

food, textiles, and chemical industries, is rejected at the 1% significance level If the alternative hypothesis is true, there are no frontier parameters in the regression equa-tion, and the estimation becomes an ordinary least squares estimation

The second alternative hypothesis, that technical inefficiency is time-invariant (H1:

h=0), is also rejected at the 1% significance level for the food, textiles, chemical, and metal industries, implying that, in these industries, technical inefficiency is not time-invariant, given the time-varying specification of the stochastic frontier defined by equation (4)

The third alternative hypothesis, that there is no technical change (H1:

aT=bTT=bTL=bTK=0), is rejected, except for non-metal, and the fourth alternative

hypothesis, that TP is neutral (H1:bTL=bTK=0), is rejected, except for the non-metal,

metal, and transport industries

The last alternative hypothesis, that the technology in Indian manufacturing is a Cobb–Douglas production function (H1:bLL=bKK=bLK=bTT=0), is rejected for all

industries Thus, the Cobb–Douglas production function does not provide an adequate specification of the Indian manufacturing sectors

Change in Technical Efficiency

Table represents the average annual rates of the technical efficiency (TE) and the TEC of Indian manufacturing industries from 2000 to 2006 Figures presented are weighted averages in which each firm’s share in total value added is used as a weight Average technical efficiency is highest in the chemical industry (0.468), followed by the

Table Derived Output Elastcities and Returns of Scale for Indian Manufacturing Industries, 2000–2006

Food Textiles Chemical Non-metal Metal Machinery Transport

el 0.421 0.482 0.593 0.507 0.687 0.611 0.388

(0.292) (0.076) (0.051) (0.358) (0.053) (0.065) (0.146)

ek 0.531 0.349 0.325 0.528 0.127 0.262 0.526

(0.164) (0.128) (0.092) (0.275) (0.058) (0.025) (0.145)

RTS 0.953 0.831 0.918 1.036 0.814 0.874 0.915

(0.219) (0.076) (0.103) (0.209) (0.026) (0.040) (0.074)

Notes: Standard deviations are in parenthesis.ejis the output elastivcity of factor j (ej=∂lnf/∂lnxj) and

RTS= ∑jεj

Table Average Annual Rates of Technical Efficiency (TE), Technical Efficiency Change (TEC) and Technical Progress (TP) of Indian Manufacturing Industries

Food Textiles Chemical Non-metal Metal Machinery Transport TE 2000 0.320 0.282 0.490 0.149 0.208 0.281 0.445

2001 0.324 0.271 0.488 0.199 0.197 0.289 0.458 2002 0.322 0.259 0.470 0.118 0.192 0.296 0.469 2003 0.325 0.252 0.469 0.116 0.183 0.292 0.464 2004 0.314 0.239 0.457 0.171 0.181 0.293 0.462 2005 0.296 0.237 0.456 0.165 0.177 0.297 0.460 2006 0.309 0.226 0.452 0.161 0.168 0.300 0.457 Mean 00–06 0.315 0.252 0.468 0.154 0.186 0.292 0.459 TEC 2001 0.004 -0.011 -0.002 0.050 -0.011 0.008 0.013

2002 -0.002 -0.012 -0.018 -0.081 -0.005 0.007 0.011

2003 0.003 -0.007 -0.001 -0.002 -0.009 -0.004 -0.005

2004 -0.011 -0.013 -0.012 0.055 -0.002 0.001 -0.002

2005 -0.018 -0.002 -0.001 -0.006 -0.004 0.004 -0.002

2006 0.013 -0.011 -0.004 -0.004 -0.009 0.003 -0.003 Mean 01–06 -0.001 -0.009 -0.006 0.002 -0.006 0.003 0.002 TP 2000 0.012 0.024 0.034 0.103 0.130 -0.028 0.036 2001 0.047 0.049 0.048 0.088 0.148 0.013 0.051 2002 0.079 0.055 0.083 0.068 0.155 0.046 0.075 2003 0.112 0.070 0.104 0.046 0.166 0.083 0.094 2004 0.145 0.086 0.126 0.028 0.178 0.120 0.111 2005 0.179 0.100 0.147 0.009 0.191 0.157 0.130 2006 0.214 0.118 0.168 -0.008 0.205 0.195 0.148 Mean 00–06 0.113 0.072 0.101 0.048 0.168 0.084 0.092

Note: Figures presented are weighted averages in which each firm’s share in total value added is used as a

transport industry (0.459) This implies that the output of the chemical industry can be increased by 53.2% without increasing input, just by enhancing TE The non-metal and metal industries have the lowest and second lowest estimates, 0.154 and 0.186, respect-ively The food, textile, and machinery industries have mean technical efficiency of 0.315, 0.252, and 0.292, respectively The average TE for all industries did not change much throughout the sample period

According to the TE estimates, Indian manufacturing industries can be grouped into three types: the high TE group comprises the chemical and transport industries with 0.468–0.459, the middle TE group comprises the food, textile, and machinery industries with 0.252–0.315, and the low TE group comprises the non-metal and metal industries with 0.154–0.186 There is a large gap in TE between the high efficiency industry group and the low efficiency industry group, suggesting unbalanced growth between industries

The Indian manufacturing industries consist of a small number of mega-sized firms equipped with advanced technologies and a large number of small and medium enterprises (SMEs) Even though the largest firms are utilizing frontier production technology, the number of those firms is very small compared with the number of small firms operating far below frontier production Overall, low TE reflects not only the composition of the Indian manufacturing sector but also a wide technology gap between frontier large firms and lagging small firms

Regarding the estimates of TEC, it should be noted that this is positive in the machinery (0.003), non-metal (0.002), and transport industries (0.002), and is negative in the food (-0.001), textile (-0.009), chemical-0.006), and metal (-0.006) industries

The annual TEC presents a different picture from the mean TEC in the period 2001

to 2006 In the food industry, the mean TEC was estimated at -0.001, but yearly

estimates show a significant improvement from 0.004 in 2001 to 0.013 in 2006 In the textile industry, there was negative TEC in all years, and there was no change in the TEC of this industry between the first and last year, at-0.011 In the chemical industry

the TEC deteriorated during the sampling period, from-0.002 to-0.004 In the

non-metal industry, the mean TEC was estimated at 0.002, but TEC declined drastically

from 0.050 to-0.004 during the period 2001–2006 A similar trend was observed in the

transport industry, as TEC declined from 0.013 to-0.003 In the metal industry, mean

TEC was-0.006 from 2001 to 2006, but yearly TEC fluctuated within a small range

throughout the period Similarly, a slight fluctuation in TEC was also observed in the machinery industry, which had positive TEC throughout the sampling years, except for 2003

Negative TEC is observed in every industry and year, except for several years in the food industry and most years in the machinery industry This might reflect the fast growth of the Indian manufacturing sector As the production frontier of the economy was continuously shifted upward by large firms, the gap between the frontier and actual production enjoyed by SMEs widened, thereby causing a deterioration of the average TE

Meanwhile, positive TECs in the food and machinery industries suggest that SMEs were catching up fast to frontier firms, narrowing the gap between SMEs and the frontier firms In these industries, technology transfers between the firms were more pronounced than in the other industries that experienced a widening TE gap

According to the previous literature, TE is related to factors such as the skill of workers, managerial expertise, and input mix, among others (Kim and Shafii, 2009).This means that India’s manufacturing firms must enhance on-the-job training to lift the skill level of their workers, and hire high-quality mangers to put existing frontier production processes into practice Additionally, the Indian government should con-tinue to promote the free market by deepening its deregulation process

Opening up the domestic market to foreign investment will enhance TE by bringing in advanced managerial techniques and tight monitoring by foreign ownership Increased imports will also provide firms with foreign parts and components to improve their input mix Thus, the government should promote free trade to facilitate these changes

Technical progress Table presents the weighted averages of the rates of TP of

Indian manufacturing industries by year The average rate of TP increased continu-ously for all industries except the non-metal industry throughout the sampling period The average annual growth rates of TP were highest in the metal industry, with 0.168, followed by the food and chemical industries with 0.113 and 0.101, respect-ively The average annual growth rates of TP were slowest in the non-metal industry, with 0.048, followed by the textiles and machinery industries, with 0.072 and 0.084, respectively

In the entire Indian manufacturing sector, TP increased remarkably quickly, by about 0.104 from 2000 to 2006 Over the sampling years, TP was increasing in all industries except the non-metal industry, in which TP moved in the opposite direction, declining continuously over the years The metal industry led steadily in TP throughout these years, but the large initial gap in TP between industries narrowed rapidly over the years The gain in TP was most impressive in the machinery industry, which jumped from the lowest, and negative, TP in 2000 to the second highest level in 2006

Every industry experienced increasing TP, particularly in the later part of the sample period, except the non-metal industry This widespread increase in TP was made possible by increased investment resulting from the series of economic reforms imple-mented by the Indian government since 1985

The government abolished most industrial licenses, and lifted restrictions on invest-ment by removing the asset limits of the MRTP in 1985 The governinvest-ment also increased the limit on foreign equity participation from 40% to 51% in 1998 Following this reform, capital investment in the Indian manufacturing sector, by both domestic firms and foreign multinationals, increased rapidly across industry Increased capital invest-ment in the manufacturing sector brought into the country the new technologies embodied in new capital, thus raising TP in an unprecedented fashion However, declining TP in the non-metal industry suggests that some industries were excluded

from this development.6

Change in scale efficiency Table presents the weighted average rates of SEC, which

measures the effects of input changes on productivity growth, and is zero if RTS is constant, and greater (or less) than zero if RTS is increasing (or decreasing), given

positive input growth The average SEC was slowest in the food industry with-0.029,

The average SEC was fastest in the chemical industry with 0.006, followed by the non-metal and transport industries with 0.003 and 0.002, respectively

Generally, SEC increased in the chemical, non-metal, and transport industries over the sampling years, but decreased in the other industries (food, metal, machinery, and textiles industries) There was a conspicuous decrease in SEC in the food and metal industries in the later years Otherwise, the SEC has been confined within the narrow range-0.02 to 0.02 since 2003, after wide fluctuations before that year

The Indian government implemented the reform process in 1991 to make the manufacturing sector more competitive The reforms expanded the market across borders, to enable the manufacturing sector to realize scale economies However, the estimated SEC was meager and negative in most industries and most years This may have resulted in more firms with an insufficient production scale entering the market after government deregulation, thus reducing the overall SEC, even though some firms did move closer to, or even realized, economies of scale with increased investment

Notably, however, the positive SECs in the chemical and transport industries since 2004 suggest a possible improvement in scale efficiency as firms increased their invest-ments, taking them closer to the optimum production scale In these industries, more firms got closer to the optimum, thus improving the overall SEC of these industries These industries are rapidly increasing in the Indian manufacturing sector, and their rapid investment, which is required to compete in this market, deters small firms from entering this market These factors may improve the overall SEC of the industries

Table Scale Efficiency Change (SEC), Allocative Efficiency Change (AEC) and Total Factor Productivity Growth (TFPG) of Indian Manufacturing Industries

Food Textiles Chemical Non-metal Metal Machinery Transport

SEC 2001 -0.030 -0.028 0.078 0.026 -0.086 -0.068 0.015

2002 -0.037 0.017 -0.054 -0.007 0.050 0.044 -0.009

2003 -0.011 -0.007 -0.007 -0.013 0.002 -0.0009 -0.0001

2004 -0.016 -0.002 0.004 0.011 -0.019 -0.009 0.002

2005 -0.026 -0.009 0.006 -0.002 -0.048 -0.009 0.001

2006 -0.054 -0.017 0.009 0.00002 -0.040 -0.017 0.003

Mean 01–06 -0.029 -0.008 0.006 0.003 -0.024 -0.010 0.002

AEC 2001 -0.039 0.070 -0.004 -0.071 0.268 -0.004 0.030

2002 0.021 -0.064 -0.120 0.033 -0.169 -0.011 0.026

2003 -0.007 -0.003 -0.034 0.045 -0.016 -0.008 0.001

2004 -0.044 0.004 0.003 -0.042 0.050 0.004 -0.012

2005 -0.001 0.002 0.005 0.024 0.088 -0.004 0.012

2006 -0.014 0.020 0.002 -0.093 0.025 -0.002 0.004

Mean 01–06 -0.014 0.005 -0.025 -0.017 0.041 -0.004 0.010

TFPG 2001 -0.082 0.030 0.101 0.006 0.260 -0.065 0.090

2002 0.002 -0.057 -0.128 0.050 -0.039 0.073 0.087 2003 0.032 -0.005 0.027 0.034 0.075 0.068 0.090 2004 0.021 0.020 0.095 -0.045 0.135 0.110 0.097 2005 0.084 0.023 0.120 -0.011 0.153 0.138 0.139 2006 0.078 0.048 0.139 -0.145 0.109 0.171 0.150 Mean 01–06 0.023 0.010 0.059 -0.019 0.116 0.083 0.109

Change in allocative efficiency Table presents the weighted average rates of AEC, which refers to the distortion in the factor market that causes firms to deviate their factor use, as indicated by a factor’s relative marginal productivity The average AEC

was slowest in the chemical industry, with-0.025, followed by the non-metal industry,

with-0.017 The average SEC was fastest in the metal industry, with 0.041, followed by the transport industry, with 0.010 In the other industries the average SEC ranged from

-0.014 to 0.005

AEC increased in the textiles, chemical, metal, and transport industries, all of which changed sign from negative to positive in the later years, and it decreased in the other industries (food, metal, machinery, and textiles) AEC also improved in the machinery industry over the years, as its negative estimates decreased AEC remained significantly negative in the food and non-metal industries, even though there was some fluctuation

The Indian government reduced administrative controls and barriers in its deregu-lation process, and removed obstacles to the free flow of exports and imports Obviously, these reform efforts made the business environments of India more market-oriented, thus reducing allocative inefficiency substantially Even though the positive impacts of the reform on AEC can be observed in most industries, estimation revealed the existence of large allocative inefficiency in the food and non-metal industries Thus, the Indian government should continue working to reduce any remaining market distortions, in order to enhance efficient allocation in the manu-facturing sector

Total factor productivity growth Table presents the weighted average rate of TFPG,

which is calculated as the sum of TEC, TP, SEC, and AEC Average TFPG was fastest in the metal industry with 0.116, followed by the transport industry with 0.109 Average

TFPG was slowest in the non-metal industry with -0.019, followed by the textiles

industry with 0.010 In the other industries average annual TFPG ranged from 0.023 to 0.083

Average TFPG increased in every industry except the non-metal industry The increase was led by the machinery industry, followed by the metal, transport, and chemical industries These industries comprise the top-tier industry group with a fast growing TFPG, while the second-tier industry consists of the food and textiles indus-tries The non-metal industry is the only industry with deteriorating TFPG

Decomposition shows that TP was the key component that determined TFPG, out-weighing all other components, and setting the tone of Indian manufacturing productivity growth However, all of the other factors contributed significantly to TFPG, both individually and additively During the sampling period, TEC and SEC deteriorated in four industries, slowing down overall productivity performance, whereas AEC improved in most industries

The decomposition suggests the urgent need to improve TE in Indian manufacturing industries To so, the quality of workers in these industries should be improved by enhancing the education system and through on-the-job training The slow SEC, in turn, suggests that Indian manufacturing firms must expand more, to take advantage of economies of scale, even though their size has increased greatly in recent years AEC also indicates that the Indian government could boost the manufacturing sector by furthering its efforts to introduce market-friendly measures

should be emphasized, to boost TFPG in the food, metal, and machinery industries, whereas the lagging TEC in the textiles and transport industries should be addressed The slow TP and AEC in the non-metal industry and chemical industries, respectively, should be increased

The TFPG in this study is greater than in most previous studies that have estimated

TFPG.7 In many previous studies, lower TFP growth during the 1990s is noticeable,

despite a number of changes that were expected to have favorable effects on industrial productivity as a result of the government reform Our much greater TFPG estimates in the early 2000s suggest that increased investment needs time to deliver increased productivity and efficiency, because the new technology that arrives with fresh invest-ments requires more skilled workers, better managerial practice, and an advanced input mix, all of which generally take time to develop

5 Conclusion

This study adopts a stochastic frontier analysis to decompose productivity into four components: technical progress (TP), technical efficiency change (TEC), allocative efficiency change (AEC), and scale efficiency change (SEC)

The empirical results of this study show that Indian manufacturing suffers from both a low TE level and a low growth rate of this factor On TP, the results show that every industry, except the non-metal industry, experienced increasing technical progress, particularly in the later part of the sample period The results show that SEC increased in three industries over the sampling years and decreased in four industries, whereas AEC increased in four industries and decreased in three industries Average TFPG increased in every industry except the non-metal industry

The results show that TP was the key component in determining TFPG, outweighing all other components, and setting the tone of Indian manufacturing productivity growth However, all other factors contributed significantly to TFPG, both individually and additively

The results suggest that, to promote productivity, specific guidelines are required in each industry Industries with relatively slow TP (non-metal) should emphasize research and development investment to speed up their innovation processes In the food, chemical, non-metal, and machinery industries, where there is considerable allocative inefficiency, a policy to enhance TFP by improving resource allocation should be pursued This can be done by promoting free markets and reducing government intervention Meanwhile, in industries where TEC is slow (food, textiles, chemicals, and metal), a policy to enhance the efficient use of existing technology is recommended, to catch up with frontier technology

The much greater TFPG in this study during the early 2000s suggests that increased investment needs time to deliver increased productivity and efficiency, because new technology together with fresh investment requires higher numbers of skilled workers, better managerial practices, and an advanced input mix, all of which generally take time to develop Thus, Indian manufacturing industries must deal with the urgent need to catch up, and then keep up, with the fast upward shift in produc-tion frontiers resulting from the spurt in new investments, implemented after economic reform For this to happen, the government must enhance its education system to provide the skilled workers and quality managers necessary to further eco-nomic reform

effi-ciency lower, even though some firms moved closer to, or realized, economies of scale with increased investments However, the study also shows the positive impact on the allocative efficiency of Indian economic reform, which reduced administrative controls and barriers in its deregulation process, and removed obstacles to the free flow of exports and imports

References

Ahluwalia, M S., “Economic Reforms in India since 1991: Has Gradualism Worked?”Journal of Economic Perspective16 (2002):67–88

Battese, G E and T J Coelli, “Frontier Production Function, Technical Efficiency and Panel Data: With Application of Paddy Farmers in India,” Journal of Productivity Analysis (1992):153–69

Bhaumik, S K., S Gangopadhyay and S Krishnan, “Reforms and Entry: Some Evidence from the Indian Manufacturing Sector,”Review of Development Economics13 (2009):658–72 Chand, S and K Sen, “Trade Liberalisation and Productivity Growth: Evidence from Indian

Manufacturing,”Review of Development Economics6 (2002):120–32

Driffield, N L and U S Kambhampati, “Trade Liberalization and the Efficiency of Firms in Indian Manufacturing,”Review of Development Economics7 (2003):419–30

Goldar, B and A Kumari, “Import Liberalization and Productivity Growth in Indian Manufac-turing Industries in the 1990s,”Developing Economies41 (2003):436–60

Kalirajan, K P and S Bhide, “The Post-reform Performance of the Manufacturing Sector in India,”Asian Economic Papers3 (2004):126–57

Kathuria, V., “Productivity Spillovers from Technology Transfer to Indian Manufacturing Firms,” Journal of International Development12 (2000): 343–69

Kim, S and G Han, “A Decomposition of Total Factor Productivity Growth in Korean Manu-facturing Industries: A Stochastic Frontier Approach,”Journal of Productivity Analysis 16 (2001):269–81

Kim, S and M Shafii, “Factor Determinants of Total Factor Productivity Growth in Malaysian Manufacturing Industries: A Decomposition Analysis,”Asian-Pacific Economic Literature23 (2009):48–65

Kumbhakar, S C., “Estimation and Decomposition of Productivity Change When Production is not Efficient: A Panel Data Approach,”Econometric Reviews19 (2000):425–60

Madheswaran, S., H Liao and B.N Rath, “Productivity Growth of Indian Manufacturing Sector: Panel Estimation of Stochastic Production Frontier and Technical Inefficiency,”Journal of Developing Area40 (2007):35–50

Parameswaran, M., “Economic Reforms, Technical Change and Efficiency Change: Firm level Evidence from Capital Goods Industries in India,” Indian Economic Review 39 (2004):239–60

Trivedi, P., A Prakash; and D Sinate, “Productivity in Major Manufacturing Industries in India: 1973–74 to 1997–98,” Development Research Group study 20, Mumbai: Department of Eco-nomic Analysis and Policy, Reserve Bank of India (2000)

Uchikawa, S., “Investment Boom and the Capital Goods Industry,” in S Uchikawa (ed.), Eco-nomic Reforms and Industrial Structure in India, New Delhi: Manohar Publishers (2002). Unel, Bulent, “Productivity Trends in India’s Manufacturing Sectors in the Last Two Decades,”

IMF working paper WP/03/02 (2003)

Notes

3 The parameter estimates are available from the authors upon request

4 Trends in TE can be different from those implied by the estimated sign ofh, because TE is a weighted average in which firm size is used as weight Thus, average TE can increase or decrease as firm size changes over the sampling period