Contribution of Services to Output Growth and Productivity in Indian Manufacturing Pre- and Post-Reforms
As an input into the production process, services are playing an increasingly important role in manufacturing industries the world over. Yet, this has received very little attention in the empirical economic literature on producer behaviour and productivity. This paper estimates the contribution of services to output growth and productivity in Indian manufacturing using the capital-labour-energy-materials-services production function. Panel data estimations are undertaken for 148 three-digit level industries for 18 years, 1980-81 to 1997-98 and sources of growth are analysed. The results show that the contribution of services input to output and productivity growth in manufacturing (organised) has increased substantially in the 1990s. One of the major causes for this is found to be the trade reforms undertaken in the post-1990s period.
RASHMI BANGA, BISHWANATH GOLDAR
I Role of Services as Input to Manufacturing
E
Cross-country analysis reveals that the share of services in aggregate output and employment rises with the level of development [Francois and Reinert 1996; Kongsamut et al 1997]. Making a comparison across countries, a positive relationship between the level of per capita income and the intensity of use of services in manufacturing industries has been found [Francois and Reinert 1996]. A part of the growth in the use of services in manufacturing is due to splintering, i e, outsourcing of indirect production activities [Bhagwati 1984]. But, a bigger part is attributable to certain structural changes in manufacturing industries [Francois and Reinert 1996]. These are changes taking place within the manufacturing industries, raising their demand for services as intermediate input.2
Industrial firms make use of services procured from outside because of the advantages these services offer. The use of services adds value to the firms’ produce and helps in cutting down costs of operations, both contributing to the productivity of firms. Since procurement of services from an outside agency would generally be cheaper than in-house provision of the services, outsourcing saves costs. Globalisation, adoption of modern manufacturing practices such as just-in-time and build-to-order, and increasing export orientation make it necessary for industrial firms to use high quality services of various kinds, which have to be procured from specialised service providers. Thus, the extent to which a manufacturing firm will use, in the production process, services procured from outside depends on (1) the pressure on the firm to raise efficiency, reduce costs and improve competitiveness, which in turn depends on the domestic and international competition it is facing; (2) the availability of services, which depends on the level of development of the services sector in the economy; and (3) the relative cost of in-house provision of services as against their procurement from outside agencies, which depends on, among other factors, the size of the firm and the wage rate prevailing in the firm.
There is growing recognition that services procured from service provider firms are increasingly becoming an important input in manufacturing. The growing complexity of manufacturing production and distribution resulting from the application of new technologies, and increasing problems of coordination caused by these changes in manufacturing firms are raising the service content of manufactured goods [Guerrieri and Meliciani 2003]. Yet, this fact has so far received very little attention in empirical economic literature on producer behaviour and productivity. The empirical studies on the production function or productivity in industries undertaken prior to the 1980s mostly used a two-input framework in which value added was taken as the measure of output, and labour and capital as two inputs. This framework did not explicitly recognise the role of services in the production process. Many studies undertaken in the 1980s and later used the capital-labour-energy-materials (KLEM) production function framework in which the role of materials and energy found explicit recognition but that of services did not. How these studies treated services while measuring inputs was not always clearly spelt out. However, it seems that many of the studies employing the KLEM framework took services as a part of materials input
M. Thus, the price index for materials was used for deflating the value of services as well, which can obviously be questioned. Also, the lumping of services with materials involved restrictive assumptions about the substitutability between materials and services and the substitutability of these inputs with other inputs, such as labour, capital and energy.
In recent years, there has been some appreciation of the need for taking services as a separate input in the production function for carrying out productivity analysis. The Organisation for Economic Cooperation and Development (OECD) productivity manual [OECD 2001], for instance, generalises at the conceptual level, the KLEM model to a capital-labour-energymaterials-services (KLEMS) model by including services as an input. However, there are very few actual empirical studies on production functions or productivity in which services have been taken as a separate input. The present paper aims at filling this gap in the literature. To this end, a study of the contribution of services to output growth and productivity in Indian manufacturing industries is undertaken using the KLEMS production function framework.
To indicate the importance of this issue, it may be pointed out that there has been a rapid growth in the services sector in India in the last two decades, especially in the 1990s – a decade of major trade and industrial reforms3 in India [Gorden and Gupta 2003; Virmani 2004]. During 1981-90, services sector output grew at a rate of 6.6 per cent per annum. During 1991-2000, the growth rate was 7.5 per cent per annum, ahead of the growth rate of industry at 5.8 per cent per annum and that of agriculture at 3.1 per cent per annum. Business services have been the fastest growing sector in the 1990s, attaining a growth rate of about 20 per cent per annum. Some of the other services sectors, which have grown relatively fast in the 1990s are communication (14 per cent per annum) and banking (13 per cent per annum), both of which are extensively used by industries. Against this backdrop, three questions addressed in the paper are:
The rest of the paper is organised as follows. The next section briefly discusses the data sources used in the study and the measurement of output and input of manufacturing industries. Section III presents estimates of a KLEMS production function for Indian manufacturing, using which, a supply-side analysis of sources of growth in manufacturing is carried out for the period 1980-81 to 1999-2000 and separately for the 1980s and 1990s, based on the growth accounting framework. The principal aim is to assess the contribution of services to industrial growth. In Section IV, an analysis of the effect of services input on industrial productivity is undertaken. For this purpose, a multilateral total factor productivity index is constructed for 41 major industry groups (comprising the manufacturing sector) for the period 1980-81 to 1999-2000, with and without services. To assess the effect of services input on productivity, the total factor productivity index is regressed on a set of explanatory variables including technology acquisition intensity and the ratio of services input to employment. Section V presents the results of a multiple regression analysis undertaken to explain inter-industrial and inter-temporal variations in the intensity of use of services in manufacturing industries. The purpose is to ascertain if the growing use of services in Indian industries in the 1990s had something to do with the economic reforms, particularly trade reforms. This analysis is based on data for 41 major industry groups for the period 1980-81 to 1999-2000.
II Data Sources and Measurement of Output and Input
The study uses two data sets. One is at a more aggregated level, i e, for 41 major industry groups (comprising the organised manufacturing sector) for the period 1980-81 to 1999-2000. It was prepared for a research project undertaken at the ICRIER on the impact of tariff policy reform on Indian industries [Virmani et al 2003, 2004]. The other data set is at a more disaggregated level, covering 148 three-digit level industries4 for the period 1980-81 to 1997-98. Some details on the data sources and measurement of variables are given below.
Data Sources
For measuring output and input, data have been drawn mainly from the Annual Survey of Industries (ASI), published by the Central Statistical Organisation (CSO), government of India. The Economic and Political Weekly has created a systematic, electronic database using ASI results for the period 1973-74 to 199798 (hereafter, EPW database). Concordance has been worked out between the industrial classifications used till 1988-89 and that used thereafter (NIC-1970 and NIC-1987), and comparable series for various three- and two-digit industries have been prepared. ASI data at three-digit industry level according to NIC-1987 could be obtained for 1998-99 and 1999-2000 from a special tabulation of ASI results, which was done by the CSO and made available to the Indian Council for Research on International Economic Relations (ICRIER) for a study on trade protection and its impact on industrial productivity [Das 2003a,b]. The time series on value of output and input at nominal prices at the level of three-digit industries have been aggregated to obtain the series for the 41 industry groups (a list of industry groups is given in Annex I). The series at nominal prices have been deflated to obtain real output and input series.
For the purpose of deflating output and input, wholesale price indices have been used, taken from the official series on Index Number of Wholesale Prices in India. Construction of materials and energy price indices requires input-use weights, for which the input-output matrix for 1993-94 prepared by the CSO has been used. A similar approach has been taken for constructing a deflator for services.
The EPW database mentioned above has been used to obtain data on output and input for 148 three-digit industries for the period 1980-81 to 1997-98. The time span for the disaggregated data set has not been extended beyond 1997-98 because ASI changed its industrial classification after 1997-98. The measurement of output and input for this data set has been done more or less in the same manner as for the data-set for 41 industry groups. This is discussed further in the next section.
The regression analysis presented in Section IV uses variables representing technology acquisition and foreign direct investment as determinants of productivity along with export intensity of industries and the ratio of services input to output. Data on export intensity of industries, foreign direct investment and technology acquisition intensity have been taken from the Prowess database of the Centre for Monitoring Indian Economy (CMIE), Mumbai. Export intensity is measured by the ratio of exports to sales. The share of foreign companies in total sales of the industries has been taken as the indicator of the level of foreign direct investment. An index of technology acquisition intensity inflow has been constructed using data on research and development (R&D) expenditure, payment of royalty and technical fees for technology imports, and capital goods imports. The construction of the index has been done in two steps. First, the relevant ratios (e g, R&D expenditure to sales) have been constructed for the 41 major industry groups from firm level data taken from the Prowess database of the CMIE. Next, applying the principal component analysis and taking the first principal component, the index has been formed. The index combines the three technology related variables using factor loadings as weights.
For the regression analysis presented in Section V, industrywise tariff rates and non-tariff barriers are used as explanatory variables. These have been taken from the data set prepared for the ICRIER research project on impact of tariff policy reforms [Virmani et al 2003, 2004] mentioned earlier. The time-series on tariff and non-tariff barriers on imports for the 41 industry groups have been compiled from several sources. The main data source on tariff rates and non-tariff barriers (percentage import coverage by quantitative restrictions) is the previously mentioned study on trade protection undertaken at the ICRIER [Das 2003a]. Since Das has not covered all three-digit industries, it has been necessary to use other sources. Tariff rates and non-tariff barriers at the level of industrial groups (66 sectors of input-output table) have been taken from Goldar and Saleem (1992), NCAER (2000) and Nouroz (2001). For some industry groups, it has been necessary to interpolate the tariff rates or import coverage ratios, as these are not available for all the years of the period under study. For some industries, the import coverage ratio is not available for years prior to 1988-89. For such industries, the figure for 1988-89 has been applied for all earlier years of the 1980s. This should not introduce any serious error in the data on non-tariff barriers because quantitative restrictions covered a very high proportion of imports of manufactures throughout the decade.5
Measurement of Output and Input
Output: For each industry group and each three-digit industry, real gross output has been obtained by deflating the nominal figures by the wholesale price index for the group or industry.6 The best deflator that could be formed from the wholesale price index series has been used. Labour: Total number of persons engaged is taken as the measure of labour input. This includes working proprietors. Capital: Net fixed capital stock at constant prices is taken as the measure of capital input. The construction of the net fixed capital series has been done by the perpetual inventory method. The method of construction of fixed capital series is explained further in Annex II. Materials input: The reported series on materials has been deflated to obtain materials input at constant prices. Following a common practice among productivity studies, a deflator for materials has been constructed with the help of an input-output (IO) table. The 1993-94 IO table prepared by the CSO has been used for this purpose. The table has 115 IO sectors of which 66 belong to manufacturing. The deflator for each industry group is formed as a weighted average of price indices for various IO sectors (for each sector including agricultural and mining products, the best price series available from the official series on wholesale price indices has been used). For each IO sector or group of sectors corresponding to an industry group considered in the study, the column(s) in the absorption matrix give(s) the purchases of materials made by the industry group from various sectors including intra-industry transactions. This information is used for constructing weights.7
The 41 industry groups were mapped into the 66 sectors of the IO table belonging to manufacturing for construction of material price indices. To get the price indices for the three-digit industries, the price indices were first constructed for the above mentioned 66 IO sectors and then the price index for each sector was applied to all the three-digit industries belonging to that sector. Energy input: Energy input at constant prices is obtained in a manner similar to that for materials. For each industry group, a price index for energy is formed considering the relative expenditures on coal, petroleum products and electricity made by the constituent IO sectors, as given in the IO table for 1993-94, and using the wholesale price indices for these three categories of energy inputs.
For the three-digit industries, a similar method has been applied. The price index for energy has been constructed for each of the 66 IO sectors engaged in manufacturing, and then the price index for each sector has been applied to the three-digit industries belonging to that sector. Services: ASI does not provide data on services used by industrial enterprises. However, data are reported on materials, energy (coal, petroleum products, wood, electricity, gas, etc) and total input. According to the definition of “total input” in the ASI, it includes besides the cost of material, power and fuel, the following cost items: (a) cost of contract and commission work done by others on materials supplied by the factory; (b) cost of materials consumed for repair and maintenance of factory’s fixed assets including cost of repair and maintenance work done by others to the factory’s fixed assets; and (c) inward freight and transport charges, postage and telephone charges, insurance charges, banking charges, etc. The difference between total input and the cost of materials and energy (hereafter referred to as “other input cost”) has been taken as a measure of services purchased by the industrial units. This is justified because a major part of “other input cost” is likely to be on account of services procured from other agencies.8 The series, so obtained, has been deflated to correct for price changes. For this purpose, a deflator for services has been constructed from the national accounts statistics (NAS). The IO table for 1993-94 provides information on the purchases of services (transport, banking and insurance, etc) made by manufacturing industries in that year. For various services sectors, the NAS reports GDP at current and constant prices, which have been used to compute implicit deflators. The IO table indicates the weights to be used for combining them; these are the flows from the services sectors to the manufacturing industries. Thus, a weighted average of the implicit deflators of different services sectors has been taken and a deflator of services purchased has been formed for each of the 41 industry groups.
To obtain a price index for services for three-digit industries, these have been computed for the 66 IO sectors engaged in manufacturing (applying the methodology described above), and then the index computed for a sector is applied to all three-digit industries belonging to that sector.
III Contribution of Services to Industrial Growth
As mentioned earlier, the analysis of supply-side sources of growth of Indian manufacturing is based on a KLEMS production function estimated from panel data. The production function may be written as:
Qit = f (Kit, Lit, Eit, Mit, Sit; Ait) …(1)
where Q denotes gross output, K capital, L labour, E energy (fuel and power), M materials, and S services. The subscripts i and t are for industry and time (year). The term Ait represents “technology”. Through this term, inter-industrial and inter-temporal variations in total factor productivity are incorporated into the production function.
For empirically applying the above equation, the functional form needs to be specified. To keep the analysis simple, a Cobb-Douglas functional form has been chosen. The efficiency term Ait has been specified as exp(ci) + exp(λt), where t denotes time. Accordingly, after logarithmic transformation, the equation to be estimated becomes:
ln(Qit) = ci+λt +αln(Lit) +βln(Kit) +γln(M) +δln(Eit)+ νln(Sit)+ εit …(2)
where ε is the random error term.
The data set on output and input for 148 three-digit-level industries for 18 years, 1980-81 to 1997-98, described in Section II, has been used for estimating the production function given in equation (2). The estimates are presented in Table 1. Estimation has been done by both the fixed-effects model and the randomeffects model.
It is seen from Table 1 that the estimates obtained by the fixedeffects and the random-effects models are quite close to each other. Going by the p-value for the Hausman test, the estimates obtained by the random-effects model should be preferred.
The estimated coefficients of all the inputs are statistically significant at the 1 per cent level. The coefficients of inputs are positive and less than one, consistent with the underlying theory of producer behaviour. The sum of the four coefficients is 0.984 in the case of the fixed-effects model and 0.992 in the case of the random-effects model. The hypothesis of constant returns to scale is not rejected by the estimates of the production function.
Analysis of sources of output growth in manufacturing is presented in Table 2. Separate analyses have been carried out for the periods 1980-81 to 1989-90 and 1989-90 to 1999-2000, and for the entire period 1980-81 to 1999-2000. The estimated coefficients of the KLEMS production function obtained by the random-effects model have been used for the decomposition of output growth (guided by p-value for the Hausman test). The estimates of parameters α, β, γ, δ and ν give the elasticities of output with respect to labour, capital, materials, energy and services respectively. The contribution of each input to output growth is computed by multiplying the trend growth rate of the input9 with the elasticity of output with respect to that input. Adding together the contributions of inputs, the growth rate of total input is obtained. The gap between the growth rate of output and the growth rate of total input gives the growth rate of total factor productivity.
The estimates of the growth rate in real gross output and the five inputs for the two sub-periods and the entire period 1980-81 to 1999-2000 are based on the output and input estimates for aggregate manufacturing for the period 1980-81 to 1999-2000 made by Goldar (2004). The measures of output and input used in Goldar (2004) are the same as that used in this study.
From the growth decomposition analysis presented in Table 2, it is seen that the trend growth rate of real gross output of aggregate manufacturing was 7.21 per cent per annum during 1980-81 to 1989-90 and 8.12 per cent per annum during 1989-90 to 1999-2000. The growth rate of total input was 6.33 per cent per annum during 1980-81 to 1989-90 and 7.86 per cent per annum during 1989-90 to 1999-2000. Thus, the growth rate in total factor productivity (TFP)10 was about 0.9 per cent per
Table 1: Estimates of KLEMS Production Function, Indian Manufacturing
Dependent Variable: ln(Q) Period: 1980-81 to 1997-98
| Explanatory | Fixed Effects | Random Effects | ||
|---|---|---|---|---|
| Variables | Coefficients | t-Statistics | Coefficients | t-Statistics |
| ln(K) | 0.048* | 5.19 | 0.051* | 5.95 |
| ln(L) | 0.038* | 2.58 | 0.049* | 4.21 |
| ln(E) | 0.098* | 9.29 | 0.095* | 10.41 |
| ln(M) | 0.672* | 61.33 | 0.666* | 67.97 |
| ln(S) | 0.128* | 16.77 | 0.131* | 17.71 |
| t (time) | 0.003* | 3.10 | 0.003* | 3.4 |
| No of observations | 2655 | 2655 | ||
| Overall R2 | 0.987 | 0.987 | ||
| Hausman statistics | 8.57 | |||
| Wald Chi2 (6) | 58978.35 |
Notation: Q = real value of gross output, K = capital input, L = labour, E = energy, M = materials, and S = services.
* statistically significant at the 1 per cent level.
Notes: (1) Due to data gaps, a few observations have been left out. (2) The null hypothesis for the Hausman test is that the coefficients in the fixedeffects and random-effects specifications are not systematically different. A rejection of the null implies that the random effects are correlated with the other regressors, and hence the estimates from the random-effects specification are biased.
Table 2: Source of Growth in Indian Manufacturing,1980-81 to 1999-2000
(Per cent per annum)
1980-81 to 1980-81 to 1989-90 to 1999-2000 1989-1990 1999-2000
Growth rate of output 7.63 7.21 8.12
Growth rates of input Capital 8.76 7.41 10.34 Labour 1.46 0.62 2.44 Energy 7.37 9.27 5.16 Materials 7.26 7.49 6.99 Services 7.52 0.42 15.78
Contribution of input to output growth Capital 0.45 0.38 0.53 Labour 0.07 0.03 0.12 Energy 0.70 0.88 0.49 Materials 4.84 4.99 4.66 Services 0.99 0.06 2.07
Growth rate of total input 7.04 6.33 7.86 Total factor productivity growth rate 0.59 0.88 0.26 Relative contribution of services to
output growth (per cent) 13.0 0.8 25.5 Relative contribution of TFP growth to output growth (per cent) 7.7 12.2 3.2
Source:
annum during 1980-81 to 1989-90 and lower at about 0.3 per cent per annum during 1989-90 to 1999-2000. For the entire period, 1980-81 to 1999-2000, the growth rate of output was
7.63 per cent per annum and the growth rate of total input, 7.04 per cent per annum. The growth rate of TFP in this period was about 0.6 per cent per annum.
Real value of services used in manufacturing grew at the rate of about 0.4 per cent per annum in the 1980s and the growth rate increased sharply to about 16 per cent per annum in the 1990s. The contribution of services to output growth was a meager 0.06 per cent per annum during the 1980s, which increased substantially to 2.07 per cent per annum during the 1990s. The relative contribution of services to output growth was about one per cent in the 1980s and it increased to about 25 per cent in the 1990s.11 Thus, the results indicate that the increased use of services in industrial firms made a significant contribution to industrial growth in India in the 1990s.
IV Impact of Services on Productivity in Manufacturing
To study the effect of the increased use of services on manufacturing productivity, a multilateral TFP index has been constructed. The index may be written as:
(S S)/2 (S S)/2
+ +
bi zi ci zi
⎛⎜⎜
⎞⎟⎟⎛⎜⎜⎞⎟⎟
to 1989-90 and lower at 0.5 per cent per annum for the period 1990-91 to 1999-2000. For the entire period 1980-81 to 1999-2000, the trend growth rate in the index is 0.8 per cent per annum. By comparison, the trend growth rate in MTFPI* (index without services) is 0.7 per cent per annum for the period 1980-81 to 1999-2000. For the two sub-periods, the trend growth rates are 0.5 and 1.1 per cent per annum respectively. It is interesting to observe that while the trend growth in MTFPI* is lower than that in MTFPI for the 1980s, the opposite is true for the 1990s. This may be explained by the fact that the growth in services input in manufacturing was slower than that in other inputs in the 1980s, whereas in the 1990s, services input grew faster than other inputs (Table 2). Evidently, the estimate of productivity growth for the post-reforms period is overstated when use of services in manufacturing is not taken into account.
To assess the impact of increased use of services on total factor productivity in manufacturing, the following equation has been estimated:
ln(MTFPI)it = φ0 + φ1 ln(S/Q)it + φ2 ln(TECH)i,t–1 + φ3 ln(FS)i,t–1
+ φ4 ln(XI)i,t–1 + ξit …(4)
In this equation, MTFPI denotes the multilateral TFP index (five-input based; varying across industries and over time; textiles in 1980-81 being taken as 1.0) and ξ is the random error term. S denotes the real value of services used. For the purpose of normalisation, it has been divided by output. If the use of services
⎛⎜⎜⎞⎟⎟
Q Q X X
X X
b c
zi bi
⎠
ci zi
⎠
…(3)
TFP
bc
=
(i K,L,E,M,S)
∏
∏=
has a favourable effect on productivity, this should show up in
i
i
⎝⎠⎝
⎝
a positive coefficient of S/Q. Three other factors, which are
The index varies across industries and over time. It expresses the productivity level in industry-year b as a ratio to the productivity level in industry-year c. Q denotes real value of gross output. Xbi is the i’th input (K capital input, L labour, E energy input, M materials, and S services) for industry-year b, and Xci is that for industry-year c. Xzi is the geometric average of i’th input across all observations. Sbi and Sci are the income shares of i’th input for industry-year b and c respectively. Szi is the arithmetic average of income share of i’th input across all observations.
The multilateral total factor productivity index (MTFPI) has been computed for 41 industry groups for 20 years, 1980-81 to 1999-2000, from the aggregate level panel data set mentioned earlier.12 Based on the MTFPI, Table 3 presents the trend growth rate in TFP in Indian manufacturing for the periods 1980-81 to 1989-90 and 1990-91 to 1999-2000, and for the entire period 1980-81 to 1999-2000. To arrive at a summary estimate of TFP for the manufacturing sector, a weighted average of MTFPI across industries has been taken for each year, using value-added weights. The trends growth rates shown in the table are for the average index so computed.
An interesting question to be examined in this context is whether the inclusion of services as an input makes a significant difference to the estimates of growth in TFP. Therefore, an alternate multilateral total factor productivity index has been formed by considering only four inputs, namely labour, capital, materials and energy, and leaving out services.13 This index is denoted by MTFPI*. Table 3 presents, for different periods, a comparison of trend growth rates in TFP in manufacturing based on MTFPI and MTFPI*.
The trend growth rate in MTFPI (average across industries) is found to be 1.3 per cent per annum for the period 1980-81 expected to influence industrial productivity, have been included in the regression equation. TECH represents technology acquisition intensity. It is based on R&D intensity, technology import intensity and capital goods import intensity. As mentioned earlier, these three ratios have been computed separately for the 41 industries for different years, and then an index has been formed by applying the principal component analysis. FS denotes foreign share. It is measured by the ratio of the sales of foreign firms to total industry sales. XI denotes export intensity. It is measured by the value of exports a ratio to sales. For TECH, FS and XI, we hypothesise a positive relationship with productivity. These variables have been included in the equation with a one-year lag to take care of any problem of simultaneity that might arise between industrial productivity and these explanatory variables.
Since data on TECH, FS and XI could be obtained only for the 1990s, the regression equation has been estimated using data for the period 1990-91 to 1999-2000. As noted earlier, the use of services in manufacturing grew rapidly in the 1990s. Thus, for assessing the impact of services on industrial productivity, an analysis based on the data for the 1990s is more appropriate. The regression results are presented in Table 4. Since panel data
Table 3: Trend Growth Rate in TFP in Manufacturing Based onMTFPI
(Per cent per annum)
Period Trend Growth Rate Trend Growth Rate Based on MTFPI Based on MTFPI* (Excluding Services)
1980-81 to 1989-90 1.3 0.5 1990-91 to 1999-2000 0.5 1.1 1980-81 to 1999-2000 0.8 0.7
are used, fixed and random effects model have been estimated. Based on the p-value of Hausman statistics, the random effects model is preferred.
From the regression results presented in Table 4, it is seen that the coefficients of technology acquisition (TECH), foreign direct investment (FS) and export intensity (XI) variables are positive as expected. The coefficient of TECH is statistically significant indicating that technological improvements lead to higher productivity. The coefficient of XI is not statistically significant but the t-ratio of the coefficient is more than one. This is suggestive of a favourable effect of increased export intensity on industrial productivity.
The coefficient of services variable is positive and statistically significant at 1 per cent level. The equation was re-estimated after dropping some of the other explanatory variables, and in all cases, the coefficient of the services variable was found to be statistically significant at 1 per cent level. It seems reasonable therefore to interpret the regression results as suggestive of a positive relationship between services input and industrial productivity. Accordingly, it seems that the growing use of services in manufacturing in the post-reforms period might have contributed to better productivity performance.
V Have Trade Reforms Caused Increased Use of Services in Manufacturing?
To examine the reasons for higher use of services in the manufacturing sector in the post reforms period, we estimate the impact of the factors discussed earlier in Section I, i e, higher competition, higher relative cost of using in-house services and development of the services sector in the post reforms period. For the analysis, we use an aggregated data set for 41 industry groups for the years 1980-81 to 1999-2000.
It has been noted above that in the post-reform period, there was a marked acceleration in the growth rate of services used in Indian manufacturing. It should be pointed out in this connection that this acceleration in the growth of services used was almost across-the-board. Since competition can drive industrial firms to increase the use of services procured from outside with a view to gaining competitiveness, the observed rapid increase in the use of services in manufacturing in the 1990s may be connected with trade reforms. To examine the impact of trade reforms on the use of services in the manufacturing sector, the following equation has been estimated:
ln(S/Q)it = θ0 + θ1ln(W/P)it + θ2TRFit + θ3NTBit + θ4DUM + ζit …(5)
Here, S/Q, i e, service use intensity (value of services used as a ratio to the value of output, both at constant prices) is taken as the dependent variable. ζ is the random error term. W/P is the ratio of the nominal wage rate to the price index of services. A fast increase in wage rate in relation to the increase in price index of services would create a situation conducive to splintering because the industrial firms will find procurement of services from outside more economic than in-house provisions. Accordingly, the coefficient of this variable is expected to be positive.
Tariff rate adjusted for changes in real effective exchange rate14 (denoted by TRF) and percentage of imports covered by non-tariff barriers (denoted by NTB) are included among the explanatory variables.15 Inasmuch as lowering of tariff and non-tariff barriers on manufactured imports intensifies competition and hence induces the domestic industrial firms to improve their competitiveness by increasing use of services procured from outside, a negative coefficient of these variables is expected.
Apart from these, there are other variables that would influence the intensity of use of services in the manufacturing sector. Some important examples are the growth of the services sector and changes in the market structure due to industrial policy reforms. While these factors are important, due to of lack of data it is difficult to include them as specific explanatory variables in the regression equation. The best that could be done was to introduce a dummy variable for the 1990s (denoted by DUM). The dummy variable is expected to capture the effect of a number of variables including the growth of the services sector in the 1990s and various economic policy changes made in this decade, especially those that helped in the development of the services sector. It seems reasonable to expect a positive coefficient for this variable.
The results of the multiple regression analysis are reported in Table 5. These are based on panel data for 41 industry groups
Table 4: Effect of Services on Productivity in IndianManufacturing, 1990-91 to 1999-2000, Regression Results
Dependent Variable: ln(MTFPI)
| Explanatory | Fixed Effects | Random Effects | ||
|---|---|---|---|---|
| Variables | Coefficients | t-Statistics | Coefficients | t-Statistics |
| ln(services/output) | 0.0214** | 2.48 | 0.0256** | 2.95 |
| ln(TECH) | 0.0075# | 1.85 | 0.0080* | 1.99 |
| ln (FS) | 0.0003 | 0.24 | 0.0007 | 0.56 |
| ln (XI) | 0.0068 | 1.34 | 0.0069 | 1.39 |
| Constant | 0.1121** | 3.06 | ||
| No of observations | 410 | 410 | ||
| Overall R2 | 0.15 | 0.17 | ||
| Hausman statistics | 5.54 | |||
| Wald Chi 2 (4) | 16.73 |
** statistically significant at 1 per cent level. * statistically significant at 5 per cent level. # statistically significant at 10 per cent level Notation: TECH= index of technology acquisition intensity (based on R&D, technology imports and capital goods imports); FS= foreign share (indicator of foreign direct investment); XI = export intensity.
Table 5: Factors Determining Use of Services in Indian Manufacturing, 1980-81 to 1999-2000, Regression Results
Dependent Variable: Log (Services/manufacturing output)
Explanatory Fixed Effects Random Effects Variables Coefficients t-Statistics Coefficients t-Statistics
Tariff adjusted for
changes in real
effective exchange
rate -0.004** -10.07 -0.004** -10.07 Dummy for non-tariff
barriers (import
coverage ratio)@ -0.126** -4.47 -0.128** -4.47 Log (wage rate/price
index of services) 0.134* 1.96 0.180** 2.97 Dummy for the 1990s 0.109** 3.73 0.101** 3.47 Constant -2.34** -17.26 No of observations 820 820 Overall R2 0.07 0.08 Hausman statistics 50.78** Wald Chi 2 (4) 213.4
** statistically significant at 1 per cent level. * statistically significant at 5 per cent level. @ takes value one if NTB is more than 50 per cent, zero otherwise.
for 20 years, 1980-81 to 1999-2000. Since panel data are used for the analysis, panel data estimation techniques have been applied. Estimates of both the fixed effects and random effects model are presented in the table. The p-value of the Hausman statistics indicates that the fixed effects model should be preferred.
Examining the results of the regression analysis presented in Table 5, it is seen that the coefficients have the expected sign and are statistically significant at the 1 per cent level in almost all cases. The results clearly indicate that the lowering of tariff and non-tariff barriers had a favourable effect on the use of services in Indian manufacturing. In other words, the observed acceleration in the use of services in manufacturing in the 1990s is attributable in a significant measure to the trade reforms. The coefficients of the wage rate variable are positive as expected. The coefficient of the dummy variable for the 1990s is found to be positive, as expected. Accordingly, it may be inferred that the economic policy changes made in the 1990s and other developments in this decade created a condition favourable for the increased use of services in manufacturing.
VI Conclusion
This paper examines the contribution of services to output growth and productivity in Indian manufacturing in the pre- and post-reform period. For this purpose, a KLEMS production function was estimated, explicitly recognising services as an input to production. Panel data for 148 industrial groups for the period 1980-81 to 1997-98 were used to estimate the production function.
The results brought out that the importance of services as an input to production in the manufacturing sector increased considerably in the 1990s as compared to the 1980s. Use of services in manufacturing grew at an accelerated pace in the 1990s. The growth rate was about 16 per cent per annum. The contribution of services to the growth of manufacturing output went up considerably, from about 1 per cent in the 1980s to about 25 per cent in the 1990s. The trade liberalisation undertaken in the 1990s, which increased competition in the domestic market was found to be responsible, to a certain extent, for the increase in the intensity of use of services in the manufacturing sector. It appears from the empirical results that the increasing use of services in manufacturing in the post-reforms period had a favourable effect on industrial productivity.
The acceleration in the growth of the services sector in the Indian economy in the 1990s, ahead of industry and agriculture has raised the question of sustainability of India’s overall growth rate. There is a view that due to the slow growth rate of industry, the services sector might not be able to sustain its pace of growth as it will come to face constraints emerging from slow growth in domestic demand. However, the findings of this paper suggest that the use of services is growing rapidly in the industrial sector and the increased use of services is contributing to both output and productivity growth in the industrial sector. This points to the possibility that the Indian services sector might not only succeed in sustaining its own growth but might also help in improving the growth rate of industrial sector in the near future.
EPWEmail: b_goldar77@yahoo.com
(Per cent)
| Industry | Description | Share in Value Added |
|---|---|---|
| Group | in Triennium Ending | |
| 1999-2000 | ||
| 1 | Food products | 9.13 |
| 2 | Beverages | 1.46 |
| 3 | Tobacco products | 1.83 |
| 4 | Textiles (except readymade garments | |
| and carpets) | 8.60 | |
| 5 | Carpet weaving | 0.77 |
| 6 | Readymade garments | 1.54 |
| 7 | Furniture and fixtures wooden | 0.03 |
| 8 | Wood and wood products | 0.47 |
| 9 | Paper and paper products | 1.52 |
| 10 | Printing and publishing | 1.37 |
| 11 | Leather footwear | 0.41 |
| 12 | Leather and leather products | 0.43 |
| 13 | Rubber products | 2.09 |
| 14 | Plastic products | 1.70 |
| 15 | Petroleum and coal tar products | 3.42 |
| 16 | Inorganic and organic heavy chemicals | 2.91 |
| 17 | Fertilisers and pesticides | 4.55 |
| 18 | Paints, varnishes and lacquers | 1.19 |
| 19 | Drugs and medicines | 4.71 |
| 20 | Soaps, cosmetics, glycerine | 1.49 |
| 21 | Synthetic fibres, resin | 5.13 |
| 22 | Other chemicals | 1.36 |
| 23 | Structural clay products | 0.62 |
| 24 | Cement | 2.51 |
| 25 | Other non-metallic mineral products | 3.12 |
| 26 | Iron and steel basic metals | 10.29 |
| 27 | Non-ferrous basic metals | 3.10 |
| 28 | Hand tools, hardware | 0.50 |
| 29 | Miscellaneous metal products | 1.95 |
| 30 | Tractors, agricultural implements | 0.96 |
| 31 | Non-electrical machinery except agricultural | |
| and office machinery | 4.63 | |
| 32 | Office, computing machinery | 0.13 |
| 33 | Electrical industrial machinery | 2.40 |
| 34 | Other electrical machinery | 4.72 |
| 35 | Ships and boats | 0.21 |
| 36 | Rail equipment | 0.35 |
| 37 | Motor vehicles | 4.76 |
| 38 | Manufacturing of motor cycles, scooters, bicycles | 1.35 |
| 39 | Other transport equipment | 0.38 |
| 40 | Watches and clocks | 0.19 |
| 41 | Miscellaneous manufacturing industries | 1.71 |
Annex II: Construction of Fixed Capital Stock Series
For each of the 41 industry groups, fixed capital stock series at 1981-82 prices have been constructed for the period 1980-81 to 1999-2000. The steps in the construction of the fixed capital series are as follows: (1) Implicit deflator for gross fixed capital formation for registered manufacturing is derived from the data on gross fixed capital formation in registered manufacturing at current and constant prices published in the NAS. The deflator series is constructed for the period 1971-72 to 1999-2000. The base is shifted to 1981-82 so as to be consistent with the price series used for other inputs and output. (2) From ASI, the book value of fixed capital stock (at historical prices, net of depreciation) in 1980-81 is taken for each industry group. This is adjusted for price changes by using the average value of the deflator for the previous 10 years (1971-72 to 1980-81). This provides the benchmark capital stock.16 (3) Gross investment in fixed capital is computed for each year by subtracting the book value of fixed assets in the previous year from that in the current year and adding to that figure the reported depreciation in fixed assets in the current year.17 To obtain real gross investment, the gross fixed investment series at current prices is deflated by the price series mentioned above. (4) Real net investment in fixed assets is derived by subtracting depreciation of fixed capital from real gross investment in fixed assets. The rate of depreciation is taken as 5 per cent, which the same as assumed in Unel (2003). (5) Starting from the benchmark fixed capital stock and adding real net fixed investment for successive years, the net fixed capital stock series is constructed. The capital stock estimates made for the 41 industry groups have been used to form such estimates for the three-digit industries. For each group, we compute for each year the ratio between estimated value of new capital stock at 1981-82 prices and the reported fixed capital (book value) in ASI. This ratio is then applied to all three-digit industries that belong to that group.
Annex III: Measurement of Services Input in Manufacturing
The ASI does not provide data on services purchased by factories. To measure this input, therefore, the difference between total inputs and the sum of materials and energy inputs has been taken. This gap, hereafter referred to as “other input cost” is accounted for by several inputs. However, services purchased are expected to form a major part. According to information obtained from the CSO (private communication from Director, Industrial Statistics Wing, CSO), the cost of work done by other units on materials supplied by the factory constituted about 12 per cent of “other input cost” in 2001-02. Some important cost items under “other input cost” are operating costs, non-operating costs, insurance charges, and cost of repair and maintenance of building, plant and machinery, and other fixed assets. Considering the definition of “total input” in ASI and available information on important cost items included in total input other than materials and energy, it appears that services form a major part of “other input cost”. To cross-check the validity of this assertion, the proportion of services to total output for the year 1998-99 was estimated for various two-digit manufacturing industries using IO table for that year, and the cost of services to total output was estimated using ASI database, employing the method described above. The results show that the proportion of services bought to total manufacturing output was around 15 per cent based on IO table, and the ratio of “other input cost” (taken as a rough measure of the cost of services) to output was around 14.4 per cent based on ASI data. Evidently the two figures are quite close. The correlation between the two series was found to be 0.20. Thus, one feels justified in using “other input cost” as a measure of services input.
Notes
[This is a revised version of a paper written earlier as a part of the working paper series of the Indian Council for Research on International Economic Relations, New Delhi (#139, July 2004).]
1 There is a great deal of overlap between the output of service industries and final product services, but the former includes purely intermediate activities as well as activities that are considered intermediate when performed for a business purchaser and as final when performed for a household (e g, repair and maintenance).
2 Two different approaches have been taken to explain the inter-relationship between the services sector and the manufacturing sector. One set of studies argue that the demand for producer services grows with development and this expansion is linked to growth in round-about production and the associated conversion of local markets into national markets [e g, Greenfield 1966; Katouzian 1970; Francois 1990]. As against that, Bhagwati (1984), in his seminal work, emphasised “splintering” or outsourcing of indirect production activities as a possible source for the apparent growth in producer services. He put forward different ways in which technical and structural changes define a continuous process during which services splinter-off goods and goods splinter-off services. He argued that services that splinter-off from goods are technically progressive and possibly capital-intensive since these services arise due to specialisation, which reflects economies of scale. Being a part of the dynamic process of change in the economic system they are not technically stagnant. But the services that are left after the goods – from services splintering process are mostly technically unprogressive and labour intensive.
3 Since 1991, India has undertaken a major economic reforms programme. Under the programme, significant and far-reaching changes have been made in industrial and trade policy. Import liberalisation has been a principal component of the economic reforms undertaken. Tariff rates have been brought down considerably and quantitative restrictions on imports have been by and large removed. For a discussion on India’s economic reforms since 1991, see Joshi and Little (1996), among others.
4 These industries together constitute almost the entire organised manufacturing sector.
5 For aggregate manufacturing, the proportion of imports covered by quantitative restrictions was about 90 per cent in 1988-89. Further note
SAMEEKSHA TRUST BOOKS
Inclusive Growth K N Raj on Economic Development
Essays from The Economic Weekly and Economic and Political Weekly
Edited by ASHOKA MODY
The essays in the book reflect Professor K N Raj’s abiding interest in economic growth as a fundamental mechanism for lifting the poor and disadvantaged out of poverty. He has also been concerned that the political bargaining process may end up undermining growth and not provide support to those who were excluded from access to economic opportunities. These essays, many of them classics and all published in Economic Weekly and Economic and Political Weekly, are drawn together in this volume both for their commentary on the last half century of economic development and for their contemporary relevance for understanding the political economy of development in India and elsewhere.
Pp viii + 338 ISBN 81-250-3045-X 2006 Rs 350
Available from
Orient Longman Ltd
Mumbai Chennai New Delhi Kolkata Bangalore Bhubaneshwar Ernakulam Guwahati Jaipur Lucknow Patna Chandigarh Hyderabad
Contact: info@orientlongman.com
that in actual application in regression analysis, a dummy variable has been created to represent non-tariff barrier, taking 50 per cent as the cutoff level. Needless to say that the use of a dummy variable in place of actual values makes the regression results less sensitive to errors in data on non-tariff barriers.
6 Note here that for deflating the output of 41 industry groups, the wholesale price index has been adjusted for changes in excise duty rate over time. Adjusting the wholesale price index for inter-temporal changes in excise duty rate applicable to the relevant product category yields in effect an index of producers’ price, which is obviously more relevant for deflation of output and value added.
7 A number of studies on productivity trends in Indian industries have constructed a deflator for materials used in manufacturing in this manner. See, Rao (1996) and Goldar and Kumari (2003), among others.
8 This is discussed further in Annex III.
9 Exponential trend equations are fitted to the time-series on output and inputs to obtain the trend rates of growth for the period 1980-81 to 1999-2000. To get the trend growth rates for sub-periods, the kinked exponential model is used. See Boyce (1986) for a discussion on this method of estimating sub-period growth rates.
10 Note here that these estimates of TFP are based on elasticities derived from the estimated production function, and hence not strictly comparable with the TFP estimates that use income shares of factors as weights.
11 This finding remains unchanged even if average income shares of inputs are used for growth accounting instead of the elasticities obtained from the estimated production function as done in Table 2.
12 We have chosen textiles (excluding carpet making and readymade garments) in 1980-81 as the base for computing the MTFPI. Our outputinput data starts from 1980-81, and in that year textiles topped the list in terms of value added. It accounted for about 19 per cent of gross value added in manufacturing (registered). This is the rationale for choosing textiles in 1980-81 as the base for making productivity comparisons.
13 The productivity is estimated using the KLEM production function.
14 Tariff rate has been adjusted for changes in the real effective exchange rate because the effect of lowering of tariff may be partly offset by depreciation in the real effective exchange rate.
15 Since both tariff rates and non-tariff barriers were lowered in the 1990s, these two variables are correlated. Therefore, the regression results are affected when both variables are included in the regression. To tackle this problem, a dummy variable for non-tariff barriers has been used. It takes value one if the level of barrier is more than 50 per cent, zero otherwise.
16 The bulk of the assets existing in the benchmark year, 1980-81 (in terms of net book value) would have been bought in the previous 10 years. This is the rationale for using the average value of the deflator for the previous 10 years for making price adjustments. It may be pointed out in this connection that assets acquired in the previous five years would constitute a much bigger part of the net book value of assets than the assets acquired between the fifth and tenth year in the past. Thus, the use of a simple average of the fixed assets price series for the previous 10 years for making price corrections introduces an upward bias in the benchmark estimate of the capital stock. It should be noted, however, that the depreciation rate used by firms for accounting purposes (allowed by income tax authorities) is much higher than the true depreciation (taken here as 5 per cent). The implication is that the reported capital stock in ASI understates the true value of net fixed assets at historical prices. These two biases tend to cancel out each other to some extent.
17 Let Bt denote the book value of fixed assets in year t and Dt the reported depreciation in that year. Then, the gross investment in year t, denoted by It, may be obtained as It = Bt - Bt-1 + Dt. It should be noted here that the ASI reports the book value of fixed assets net of cumulative depreciation.
References
Bhagwati, J N (1984): ‘Splintering and Disembodiment of Services and Developing Nations’, World Economy, 7(2): 133-43.
Boyce, J K (1986): ‘Kinked Exponential Models for Growth Rate Estimation’, Oxford Bulletin of Economics and Statistics, 48(4): 385-91.
Chenery, H B (1960): ‘Patterns of Industrial Growth’, American Economic Review, Vol 57: 415-26.
Clark, C (1940): The Conditions of Economic Progress, Macmillan, London.
Das, D K (2003a): ‘Quantifying Trade Barriers: Has Protection Declined Substantially in Indian Manufacturing?’ Working Paper No 105, Indian Council for Research on International Economic Relations, New Delhi.
– (2003b): ‘Manufacturing Productivity under Varying Trade Regimes: India in the 1980s and 1990s’, Working Paper No 107, Indian Council for Research on International Economic Relations, New Delhi.
Fisher, A G B (1935): The Clash of Progress and Security, Augustus M Kelly, New York.
Francois, J F (1990): ‘Producer Services, Scale and the Division of Labour’, Oxford Economic Papers, 42 (4): 715-29.
Francois, Joseph F and Kenneth A Reinert (1996): ‘The Role of Services in the Structure of Production and Trade: Stylised Facts from a Cross-Country Analysis’, Asia-Pacific Economic Review, 2(1): 35-43.
Fuch, V (1968): The Service Economy, National Bureau of Economic Research, New York.
Goldar, B N and H N Saleem (1992): ‘India’s Tariff Structure: Effective Rates of Protection of Indian Industries’, Studies in Industrial Development, Paper No 5, Ministry of Industry, Government of India, October.
Goldar, Bishwanath (2004): ‘Productivity Trends in Indian Manufacturing in the Pre- and Post-Reform Periods’, Working Paper No 137, Indian Council for Research on International Economic Relations, New Delhi.
Goldar, Bishwanath and Anita Kumari (2003): ‘Import Liberalisation and Productivity Growth in Indian Manufacturing Industries in the 1990s’, Developing Economies, 41(4): 436-60.
Gorden, Jim and Poonam Gupta (2003): ‘Understanding India’s Services Revolution’, paper prepared for IMF-NCAER conference on, ‘A Tale of Two Giants: India’s and China’s Experience with Reform’, November 14-16, New Delhi.
Greenfield, H I (1966): Manpower and the Growth of Producer Services, Columbia University Press, London.
Guerrieri, Paolo and Valentina Meliciani (2003): ‘International Competitiveness in Producer Services’, paper presented at workshop on empirical studies of innovation in Europe, Urbino, December 1-2.
Joshi, Vijay and I M D Little (1996): Indian Economic Reforms 1991-2001, Clarendon Press, Oxford.
Katouzian, M A (1970): ‘The Development of the Service Sector: A New Approach’, Oxford Economic Papers, 22 (3).
Kongsamut, P, S Rebelo and D Xie (1997): ‘Beyond balanced Growth’, NBER Working Paper No 6159, National Bureau of Economic Research, Cambridge MA.
Kuznets, S (1957): ‘Quantitative Aspects of the Economic Growth of Nations’, Economic Development and Cultural Change, supplement to Vote V, No 4.
NCAER (2000): Protection in Indian Industry, National Council of Applied Economic Research, New Delhi.
Nouroz, H (2001): Protection in Indian Manufacturing: An Empirical Study, MacMillan India, New Delhi.
OECD (2001): OECD Productivity Manual: A Guide to the Measurement of Industry-level and Aggregate Productivity Growth, Statistics Directorate, Directorate for Science, Technology and Industry, OECD, Paris.
Rao, J M (1996): ‘Manufacturing Productivity Growth: Method and Measurement’, Economic and Political Weekly, November 2, 31:2927-36.
Unel, Bulent (2003): ‘Productivity Trends in India’s Manufacturing Sectors in the Last Two Decades’, IMF Working Paper No WP/03/22, International Monetary Fund, Washington DC.
Virmani, A (2004): ‘India’s Economic Growth: From Socialist Rate of Growth to Bharatiya Rate of Growth’, Working Paper No 122, Indian Council for Research on International Economic Relations, New Delhi.
Virmani, A, B Goldar and C Veeramani (2003): Quantitative Assessment of Tariff Policy Changes on the Indian Industry, Report prepared for the Ministry of Industry, Government of India.
Virmani, A, B Goldar, C Veeramani and V Bhatt (2004): ‘Impact of Tariff Reforms on Indian industry: Assessment based on a Multi-sector Econometric Model’, Working Paper No 135, Indian Council for Research on International Economic Relations, New Delhi.
