SPECIAL ARTICLEEconomic & Political Weekly EPW april 11, 2009 vol xliv no 1539Productivity and Unit Labour Cost in Indian Manufacturing: A Comparative PerspectiveAbdul Azeez ErumbanUp to date results on prices, labour productivity and unit labour costs for Indian manufacturing in comparison with some advanced and developing countries are presented here. The results indicate that the labour productivity levels in Indian manufacturing are much lower than those of Germany, the US, South Korea, Hungary and Poland, but higher than those of Indonesia, Brazil and Mexico. The unit labour cost in Indian manufacturing is the lowest among the countries in our sample, indicating strong cost competitiveness of Indian manufacturing vis-à-vis these countries. However, a comparison with China reveals a fast erosion of Indian manufacturing competitiveness in the recent period.This is a revised version of the background paper prepared for the Organisation for Economic Cooperation and Development Economic Survey for India. The author is thankful to Sean Dougherty for initiating the project and to OECD for financial assistance. Comments and suggestions by Marcel Timmer, Bart van Ark and an anonymous referee are acknowledged. Thanks are also due to M Parameswaran and V K Anil Kumar for providing some of the Indian data and to Gerard Ypma and Edwin Stuivenwold for assistance in dealing with the German data. The usual disclaimer applies.Abdul Azeez Erumban (a.a.erumban@rug.nl) is at Groningen Growth and Development Centre, Faculty of Economics and Business, University of Groningen, The Netherlands and the Conference Board, New York.The industrial sector in India has witnessed significant policy changes since the country’s independence in 1947. The state-dominated policy within high protective barriers has disappeared partly by the initiation of a liberal policy regime in the late 1980s, and almost fully by the introduction of a new industrial policy in the 1990s. This shift towards a liberal industrial policy paradigm was expected to make the country’s manufacturing sector more efficient and competitive.1 Though previous studies have tried to understand the productivity of Indian manufacturing sector in light of the policy reforms,2 therehas not been any recent study that has looked in detail at the comparative performance of this sector from an international perspective. This is important because such an analysis will help us understand the international competitiveness of this sector. Previous studies that examined Indian manufacturing from an international perspective have shown only slight indications of a catch-up in the later part of the 1980s, after a period of relative stagnation in labour productivity (Timmer 2000; Timmer and Szirmai 1999). This paper provides new evidence on the catch-up potential and the competitive position of India’s organised manufacturing sector. We compare output, labour productivity and unit labour costs (ULC) levels of Indian manufacturing over the period 1980-2004 with a set of developed and developing countries. This is done by linking in India into a set of already existing comparisons from the International Labour Organisation’s (ILO) Key Indicators of the Labour Market (KILM) data. Using a similar methodology as in theKILM, India is linked in with other coun-tries via a bilateral comparison with Germany, as there is no unit labour costs data available for Indian manufacturing through theKILM. In order to make this bilateral comparison meaning-ful, we have expressed the output in both countries in a common currency, that is, in euros. For this, we use unit value ratios (UVRs) (instead of exchange rates or expenditure purchasing power parities (PPPs),3 derived using International Comparisons of Output and Productivity (ICOP) methodology for the bench-mark year 2002. The sectoralUVRs are derived based on relative output prices of representative baskets of goods using methods developed in the ICOP. The paper is organised in four sections. Section 1 provides a discussion on the ICOP methodology to obtainUVRs and produc-tivity measures. Section 2 discusses the basic data used in the present study and their sources. Section 3 provides the results for UVRs, labour productivity and unit labour costs and the last section concludes.
SPECIAL ARTICLEapril 11, 2009 vol xliv no 15 EPW Economic & Political Weekly401 Methodology1.1 UnitValue RatiosTheUVR-based method in the industry-of-origin approach was pioneered by Paige and Bombach (1959) for a comparison of the United Kingdom (UK) and the United States (US). The methodo-logy used in the present study follows in the footsteps of the ear-lier studies,4 but was further refined within the framework of the ICOP project at the University of Groningen (Maddison and van Ark 2002). We explain the methodology in more detail below.5In the industry-of-origin approach, the industry-specific con-version factors are derived on the basis of relative product prices. As a first step, unit values (uv) are derived by dividing ex-factory output values (o) by produced quantities (q) for each product i in each country: oiuvi = qi ...(1)The unit value can be considered as an average price, averaged throughout the year for all producers and across a group of nearly similar products. Subsequently, in a bilateral comparison, broadly defined products with similar characteristics are matched. For each matched product, the ratio of the unit values in both coun-tries is taken. ThisUVR is given by: AB uvAiUVRi = uvBi ...(2)with A andB the countries being compared, B being the base country. The product UVR indicates the relative producer price of the matched product in the two countries. WithinICOP the total manufacturing sector is divided into sev-eral homogeneous branches (equal to the two or three digit ISIC level), which are subsequently subdivided into different indus-tries (equal to four digitISIC level). Product UVRs are used to de-rive an aggregateUVR for manufacturing industries, branches and total manufacturing. This requires the choice of a particular weighting scheme. The most simple aggregation method is to weight each product UVR by its share in total manufacturing gross output (or in practice, total sales). In a comparison between two countries not all products in an industry j can be matched. This is because of lack of value or quantity data, difficulties in finding corresponding products, and because of the existence of country-unique products, etc. There-fore, matched products in an industry can be seen as a subset of all the products within an industry.The industryUVR (UVRj) is given by the weighted mean of the UVRs of the sampled products. Product UVRs are weighted by their output value as more important products should have a big-ger weight in the industryUVR: AB IjUVRj =∑ wij UVRABij i=1 ...(3)withi=1,.., Ijthe matched products in industryj; wij = oij/oj the output share of the ith commodity in industryj; and oj = ∑Iji=1 oij the total matched value of output in industryj. In bilateral com-parisons the weights of the base country (B) or the other country (A) can be used, which provide a Laspeyres and a Paasche type UVR, respectively. That is for Laspeyreswij = oBij/oBj and for Paasche wij = oAij/oAj. The geometric average of the Laspeyres and Paasche indices provide Fisher index, which is used when a single currency conversion factor is required.The next aggregation step is made by using the gross output6 of industries to obtain an industry-weighted mean of all industry UVRs in a branch: AB JkUVRk =∑ wjk UVRABjk j=1 ...(4)withj=1,..,Jkthe number of industries in branchk for which a UVR has been calculated (the sample industries);wjk = ojk/ok; and ok = ∑Jkj=1 ojk. Again gross output weights from base country B and the other country A can be used to arrive at Laspeyres and Paasche indices of the branchUVRs. The latter step is repeated for the final aggregation step from branch level to the level of total manufacturing.The representativity of the UVR for a given industry or sector can be statistically tested on the basis of the coefficient of varia-tion of the UVRs. Statistically, large variations in unit value ratios signal a greater unreliability of the measures. By adjusting the variance for a finite population correction, it is ensured that with an increasing coverage of products, the variance goes down(Timmer 2000, Chapter 3). Together with measures of the Paasche/Laspeyres spread between the unit value ratios, which indicate differences in production structure between countries, measuresof output covered by matched products, and the number of product matches, the variance ofUVRs gives a good indication ofthereliability of the unit value ratios. The FisherUVRs are used to calculate value added at the branch level into comparable prices. Thus, value added can be compared between the two countries. Using value added at com-parable prices, relative labour productivity and unit labour costs are derived by adding information on employment and wages. The definition of unit labour costs requires some further discussion.1.2 UnitLabourCostsUnit labour cost is defined as the cost of labour required to pro-duce one unit of output in a particular industry, sector or the ag-gregate economy. It is a ratio that is constructed from a numera-tor reflecting the major cost category in the production process (which is labour compensation) and a denominator reflecting the output from the production process (GDP or value added), i e, ULC=LC/Y, where LC is the nominal labour compensation and Y is the real output. ULC indices can be directly compared between countries. A comparison of relative levels of unit labour cost al-lows comparisons of cost competitiveness in absolute terms not just in relative terms. Countries with a low level of ULC relative to other countries may be regarded as competitive. The meaning of the ULC concept might be even better under-stood, when expressed in terms of the ratio of labour compensa-tion per unit of labour (e g, the wage or the total labour cost per employed person or per hour worked) and the productivity of labour (measured as output per employed person or per hour). That is, defining labour productivity asY/L and wage rate asLC/L, where L is a measure of labour such as number of employees or 
SPECIAL ARTICLEEconomic & Political Weekly EPW april 11, 2009 vol xliv no 1541total hours worked, ULC can be measured as a ratio of wage rate to labour productivity. It implies that a country can improve its competitiveness either by decreasing its labour cost per person (hour) employed or raising the productivity performance. This suggests that an economy can apply different strategies to improve competitiveness, e g, by moderating wage growth in order to cuton cost, raise productivity to create more output, or, find an appropriate mix of both strategies.A specific characteristic of unit labour cost measures is that thenumerator, which reflects the labour cost component of the equation, is typically expressed in nominal terms, whereas the denominator, which is output or productivity, is measured in real or volume terms. This implies that, when comparing unit labour cost levels across countries, the level of wages or labour compen-sation is converted at the official exchange rate: it represents the cost element of the arbitrage across countries. In contrast, output or productivity relates to a volume measure as it resembles a quantity unit of output. Hence, for level comparisons, output needs to be converted to a common currency usingPPP instead of the exchange rate, so that comparative output levels are adjusted for differences in relative prices across countries.Hence, the unit labour measure represents the current cost of labour per “quantity unit” of output produced. For an analysis in terms of comparative levels between countriesA andB this implies:ULCAB = [(LCA/ERAB)/LCB]/[(YA/PPPAB)/YB] ...(5)where ULCAB stands for unit labour cost between countryA and B, LC for total labour compensation, Y for total output (or value added), ERAB for the official nominal exchange rate between countriesA andB andPPPAB for the PPP for output in country A relative to countryB. Dividing labour compensation and output by employment or total hours worked, gives the wage rate (lc) and labour productivity (y):ULCAB = [(lcA/ERAB)/lcB]/[(yA/PPPAB)/yB] ...(6)Equation (6) can be rewritten to decompose the difference in unit labour cost between countryA and country B into three compo-nents, i e, the difference in nominal labour cost per person, the dif-ference in nominal labour productivity (that is unadjusted for dif-ferences in price levels) and the differences in relative price levels:ln(ULCAB) = ln(lcA/lcB) – ln(yA/yB) – ln(ERAB/PPPAB) ...(7)All these components contribute in their own way to differ-ences in cost competitiveness between the two countries. It may, however, be noted that the ULC index should not be interpreted as a comprehensive measure of competitiveness for several reasons. First,ULC measures deal exclusively with the cost oflabour. Even though labour costs account for the major share of inputs, the cost of capital and intermediate inputs can also be crucial factors for comparisons of cost competitiveness between countries.7 Second, the measure reflects only cost competitiveness. In the case of durable consumer and investment goods, e g, competi-tiveness is also determined by other factors than costs, notably by technological and social capabilities and by demand factors. Im-provements in product quality, customisation or improved after-sales services are not necessarily reflected in lower ULC. Third, measures of cost competitiveness may be distorted by the effects from, e g, bilateral market access agreements, direct and indirect export subsidies and tariff protection (see van Ark et al 2005 for an elaborated discussion). 2 DataTo construct the unit value ratios using the ICOP methodology, we require data on values and quantities of products manufactured in India and in Germany. For this purpose, we primarily make use of two data sources. They are the PRODCOM product database for Germany, available through Eurostat, and the Annual Survey of Industries (ASI) for India, published by the Central Statistical Organisation. In what follows we provide a brief discussion of these datasets.2.1 Basic Data on Values and Quantities at Product LevelPRODCOM Data for Germany: The survey onPRODucts of the EuropeanCOMmunity (PRODCOM) provides statistics on produc-tion of manufactured goods together with related external trade data for member states of the European Union. We make use of thePRODCOM data on production which provides physical quan-tity and value of output sold. This data is based on a product list containing about 4,500 products. Since each product is classified by an eight-digit code, which is in accordance with four-digit NACE code, it is easy to attribute each product to a particular manufacturing branch under NACE. We take this data for Ger-many for the year 2002. Note that we take Germany as the base country to construct the UVRs. PreviousICOP studies have used theUS as the base country. However, given the fact that the avail-ability of the US product statistics has been deteriorating over years and German statistics provide more detailed data on prod-uct quantities and values, we get more accurate UVRs when using Germany as the base country. Since the PRODCOM data refers to aggregates across firms for each product we could directly use them without much cleaning on original product data. Each product has data on value of sales in euros, and quantities pro-vided in specific units of measurements (unit, tonne, litre etc.). Bydividing the total sales value of a given product by its sold quan-tity, we obtain the basic price of the product under consideration.ASI Plant Level Data for India: ASI provides information on quantities and values of products manufactured and sold, gross output, total persons employed, total compensation and value added among others for Indian manufacturing. For this study we obtained data for the financial year 2002-03, which we compare with the calendar year 2002 for Germany. It is important to mention that theASI data covers only the organised segment of Indian manufacturing, i e, those factories which employ 10 or more workers with power and 20 or more workers without power. The data is available for almost 5,500 products classified under ASI Commodity Classification (ASICC). Unfortunately, theASICC product classification does not have any direct link with any in-ternational product classifications, making it difficult to classify these products under different industry branches.8 Therefore, it was essential to look at each and every product in detail before making a final decision on match. Moreover, since this data is plant level data, we had to clean the data before aggregating to product level.SPECIAL ARTICLE
In fact, the product data for India is directly taken from individual plant level data on Indian manufacturing.The original dataset on quantities and values provided under block j of the ASI schedule contained 95,624 observations for 29,188 firms and 4,121 products. As a first step, we filtered out all those cases, where there is no data on either quantity or sales value or both (at firm level for India). This reduced the number of observations on products to 51,329, i e, 54% of the full sample. From the filtered list we have matched each Indian product with the corresponding German product. Thus in the final sample, we have only those products for which we have a corresponding German match. Hence, the number of observations has further declined to 20,312. We have examined the unit values, calculated as total sales value minus total taxes paid divided by total sales quantity for each product at firm level, for outliers. All those firms having extremely high/low unit values for a given product were deleted. This is because the inclusion of such firms in the sample may affect the aggregate unit value ratios for the product. However, as there was no clear-cut rule on how to delete outliers and, more importantly, the number of firms in each product varied, we have applied a variety of rules. First, we identified outliers using the Hadi’s outlier index (Hadi 1992, 1994) for each product with more than five firms. Along with this, we also computed the mean and standard deviation for each product group across firms. If the firm was found to be an outlier (within a product category) both in terms of Hadi’s index and Chebyshev’s standard deviation rules, we excluded it from the sample. For those cases for which the number of firms is lower than five, we visually observed using their mean and standard deviation (there were 790 such cases). After these cleansing procedures, we were left with 19,108 cases, with 6% of outliers.
The firm level data was aggregated across firms resulting in 925 products for which we could find match with the German data. Note that there were many cases where we had to aggregate more than one Indian (German) product to get a better match with German (Indian) product. For instance, butter in Germany is matched with different types of butter produced in India, including, for instance, ghee.9 We were careful in matching likewith-like. For example, the data on car production in Germany is available for different car types, differing in engine size, while the Indian data does not make any such distinction. We assumed that the Indian production consisted mainly of small cars. A similar approach was used for a number of machinery and equipment items. However, whenever there was an accurate match, we have opted to use that. Effectively, we had 456 product groups, consisting of 1,015 German products and 868 Indian products. Still it was not possible to use all these 456 matches due to differences in the units of measurement. Some products in India are expressed in different units compared to that of Germany.10 While we were able to convert some units using appropriate conversion factors, a large number of units could not be converted to one measure.
Another problem was due to outliers in aggregate product level unit value ratios for a given industry. All those products having extremely high/low unit value ratios were excluded; we had 115 such products, which constituted almost 6% of total output produced by the firms in the product data sample. Similarly, there
april 11, 2009 vol xliv no 15 EPW Economic & Political Weekly
SPECIAL ARTICLEEconomic & Political Weekly EPW april 11, 2009 vol xliv no 1543were 83 products for which we could not find a unit conversion factor, which, therefore, had to be excluded. Finally, we were left with 258 products for which a useful match could be made be-tween the two countries. These were classified under 43 three-digit industries and 19 two-digit branches.2.2 Aggregate DataAs mentioned in the introduction, we provide the estimates of labour productivity and unit labour cost for 2002 and also over the period 1980-2004. For this we required data on output, value added, employment, hours worked and employee’s compensation at aggregate level. Moreover, in order to derive UVRs for aggre-gate manufacturing, we require branch and industry level output data (see Equations 3 and 4). For India, this data is taken from Annual Survey of Industries, compiled and published by Economic and Political Weekly Research Foundation (EPWRF). This data is available since 1973-74 till 1997-98, under the National Industry Classification (NIC), 1987. The data for years after 1997-98 has been taken from the ASI web site (http://mospi.nic.in/mospi_asi.htm). However,ASI changed its industrial classification from NIC 1987 toNIC 1998 since 1998-99, so that we had to reclassify the data prior to 1998 using the concordance table provided by the CSO.11 Finally, we have a series, all inNIC 1998, which is largely comparable withISIC. The variables which we used are gross out-put, gross value added, total persons engaged, total emoluments and total hours worked. ASI does not provide data on man hours worked. Nevertheless, it provides data on man-days worked. We have converted man-days data to man hours by assuming that each man-day worked is assumed to be equivalent to eight hours. Also the price deflators to deflate output and value added are taken from various publications on wholesale price indices, com-piled and published byEPWRF till 1993-94. After 1993-94, we have taken the data from the Ministry of Commerce and Industry web site (http://eaindustry.nic.in). For some industries, there was no direct price deflator available. In such cases, we used the weights given in each base year for the different component of that particular industry to derive a weighted price deflator or opted to use the nearest industry deflators. Output and employment data for Germany is taken from EU KLEMS database (see Timmer et al 2008). We obtained gross value added, total persons engaged, total employees compensation and total hours worked from the EU KLEMS database.12 For gross output, we obtained the data from OECDSTAN database. 3 Results3.1 Unit Value RatiosThe main results for the UVRs and the comparative price level for 2002 are given in Table 1. Column 3 in the table shows the UVRs weighted at German quantities (Laspeyres), column 4 shows the UVRs at Indian quantity weights (Paasche) and column 5 shows the geometric average of these two (Fisher). The estimated UVR for total manufacturing is 25.2 INR to the euro, which is much lower than 48.1 INR, which is the official exchange rate against the euro in 2002. The UVRs vary significantly across industries, with the leather and footwear industries being the lowest andpetroleum, coke and nuclear fuel industries being the highest. It should be noted that in Table 1, we also have an additional row below to-tal manufacturing, i e, total excluding coke, petroleum and nu-clear fuel. This is because the latter industryisobservedtoinflu-ence the entire manufacturing sector’sresult significantlyasit has very high unit value ratios. Therefore, we opted to compare the results including and excluding this industry. Table 1: Unit Value Ratios and Relative Price Levels, India(2002)Industry IndustryCodesLaspeyresPaascheFisherComparative PriceLevel Rs/EuroRs/EuroRs/EuroIndia/ GermanyFood, beverages and tobacco 15+16 30.8 22.0 26.0 0.54Textiles 1726.514.219.40.40Clothing 1816.822.819.60.41Leather and footwear 19 11.2 10.5 10.8 0.22Wood, products of wood and cork 20 62.8 29.7 43.2 0.90Pulp, paper and paper products 21 31.2 34.5 32.8 0.68Coke, petroleum and nuclear fuel 23 127.1 99.8 112.7 2.34Chemicals 24 26.8 22.2 24.4 0.51Rubber and plastics 25 33.9 26.9 30.2 0.63Non-metallic mineral products 26 21.9 22.7 22.3 0.46Basic metals 27 24.5 19.8 22.0 0.46Fabricated metal products 28 15.4 12.5 13.9 0.29Machinery and equipment 29 35.3 13.6 21.9 0.45Office machinery 30 51.5 42.7 46.9 0.97Other elect machinery 31 15.7 22.9 18.9 0.39Radio, TV and communication eqpt 32 11.7 11.1 11.4 0.24Scientific and other instruments 33 25.4 30.9 28.0 0.58Motor vehicles 34 20.6 22.7 21.7 0.45Furniture and other mafg 36 17.7 14.7 16.1 0.33Total 29.221.725.20.52Total - Coke, petro and nuc fuel 26.0 19.1 22.3 0.46Exchange rate 48.1 Comparative price levels are calculated as Fischer UVR/exchange rate.Source: Own calculation using data from ASI, EUKLEMS and Prodcom (see text).Table 2: Reliability Indicators of Indian UVR(2002)Industry No of UVRs Coverage Ratio (%) Coefficient of Variation Germany India Laspeyres PaascheFood, beverages and tobacco 63 41.9 43.7 0.03 0.05Textiles 1921.937.50.140.12Clothing 136.544.80.210.16Leather and footwear 3 19.5 22.6 0.01 0.01Wood, products of wood and cork 5 22.1 9.3 0.07 0.12Pulp, paper and paper products 10 24.7 39.6 0.07 0.08Coke, petroleum and nuclear fuel 4 11.2 9.7 0.09 0.17Chemicals 19 11.8 12.4 0.11 0.09Rubber and plastics 11 17.1 17.7 0.16 0.16Non-metallic mineral products 10 12.0 65.6 0.03 0.03Basic metals 5 10.0 14.0 0.10 0.11Fabricated metal products 15 11.1 7.7 0.09 0.07Machinery and equipment 43 10.4 28.7 0.12 0.09Office machinery 4 27.6 29.6 0.10 0.14Other elect machinery 6 5.0 1.2 0.05 0.09Radio, TV and communication equipment 7 6.4 26.7 0.07 0.05Scientific and other instruments 9 2.8 5.7 0.15 0.09Motor vehicles 8 48.8 5.8 0.08 0.05Furniture and other mafg 4 1.0 4.0 0.33 0.07Total 258 22.123.00.03 0.04Coverage ratio indicates the share of output covered by the included products in total output of the industry. Source: As in Table 1.
SPECIAL ARTICLEapril 11, 2009 vol xliv no 15 EPW Economic & Political Weekly44In the last column of the table we present the relative price level – i e, the ratio of Fisher UVR to the prevailing nominal exchange rate in 2002. This ratio is of great importance, as it indi-cates whether the Indian products are relatively cheaper (below 1) or dearer (above 1) than those produced in Germany. It is clear that the Indian products are cheaper than the German products. On average the Indian products are priced at only 46% of the German price level excluding petroleum and 52%, when oil is included. The price advantage of Indian manufacturing varies significantly across industries. The largest advantage is in leather and leather products, followed by radio, television and communication equip-ment, fabricated metals, other manufacturing, electrical machinery and apparatus, textiles and clothing. While the only industry with a price disadvantage is coke and petroleum, industry office and accounting machinery show an almost near price level. Also wood and wood products, paper and paper products and rubber and plastics are relatively highly priced industries.The observed differences in the compar-ative price levels between India and Germany may appear to be surprising as thelaw of one price in international trade theory sug-gests that the price of an internationally traded commodity would be the same every-where in the world, if the price is expressed in a common currency. The rationale is that people could ship goods from low-priced countries to high-priced countries in order to make a riskless profit by arbitraging, if the prices differ across countries. The law of one price is the basis of the PPP hypothesis that nominal exchange rate between two currencies should be equal to the ratio of aggregate price levels between the two countries. Nevertheless, there is hardly any empirical evidence in the literature to sup-port thePPP hypothesis in the short run. There could be a number of reasons why PPP would not hold in the short run, which include many barriers that hinder arbitrage possibilities such as tariff and non-tariff barriers, transport costs, product differentiation, sticky nature of prices, price discrimina-tion and so on (see Anderson and Wincoop 2004; Taylor and Taylor 2004; Engel and Rogers 2001). The observed differences in comparative price levels, thus, suggest that PPP did not hold true for manufacturing products in 2002.13 In Table 2 (p 43), we provide some reliability statistics for the UVR measures. As mentioned before, there were 258 UVR matches, covering 23% of Indian output and 22% of German output. The largest numbers of matches were found in food, beverages and tobacco, followed by machinery and equipment. The last two col-umns in Table 2 indicate the coefficient of variation (CV) of UVRs within industries. It shows that the UVRs for total manufacturing and most two-digit branches are quite reliable, as the coefficient is less than 0.1. A few branches, however, show a highCV. For instance, furniture and other manufacturing and clothing have shown a CV greater than 0.2 in LaspeyresUVR. The results for these branches should, therefore, be interpreted with caution. 3.2 Output, Labour Productivity and ULC: Comparing India and GermanyBasic results on India-German comparison for the benchmark year 2002 are provided in Table 3. The estimated unit value ratios pre-sented in Table 1 are used to convert output in INR into euros, so that the Indian output level can be compared with the German out-put levels. The UVR-converted output in combination with labour input, employees, hours worked and the exchange rate-converted labour compensation, is then used to derive relative output, labour productivity and ULC comparisons between India and Germany. The first two columns in Table 3 present value added, employ-ment and total hours worked by manufacturing branch, expressed as a percentage of the same measures for the corresponding branch in Germany. The table suggests that the number of per-sons engaged in the organised segment of manufacturing in In-dia is about the same as in Germany (it is about seven million workers). But the Indian employment number is significantly higher than in Germany in food, beverages and tobacco, textiles, clothing, leather and footwear, coke, petroleum and nuclear fuel, non-metallic minerals and basic metals. The industries which show a relatively lower level of employment in India are medical precision and optical instruments and machinery and equipment. This may be due to the fact these are industries for which India does not have much dominance. Also high income elasticity and lower per capita income in India may be considered as important explanations for this phenomenon. The results are quite different for value added though. Overall, the Indian-organised manufacturing constitutes only 20% of the Table 3: Relative Levels of Output, Employment, Labour Productivity, Unit Labour Cost and Wage Rate in Indian Manufacturing(Germany=100, 2002)Industry Value Total Total Value Value Unit Compen- Compen- Added/PersonsHoursAdded/Added/Laboursation/sation/ PersonHourCostPersonHourFood, beverages and tobacco 25.2 185.5 264.6 13.6 9.5 21.0 2.9 2.0Textiles 163.8 847.9 1,500.7 19.310.9 19.7 3.8 2.2Clothing 71.5430.2 791.1 16.69.0 19.9 3.3 1.8Leather and footwear 117.9 472.2 737.7 25.0 16.0 13.5 3.4 2.2Wood, products of wood and cork 1.6 26.2 36.2 6.1 4.4 35.8 2.2 1.6Pulp, paper and paper products 12.7 107.4 142.9 11.8 8.9 29.8 3.5 2.7Coke, petroleum and nuclear fuel 39.8 291.4 506.4 13.7 7.9 34.8 4.8 2.7Chemicals 37.6 152.5 260.6 24.7 14.4 15.1 3.7 2.2Rubber and plastics 11.6 63.4 104.9 18.3 11.1 21.1 3.9 2.3Non-metallic mineral products 30.9 209.7 314.6 14.8 9.8 16.8 2.5 1.6Basic metals 65.8 193.1 305.9 34.1 21.5 16.6 5.7 3.6Fabricated metal products 9.5 33.3 57.6 28.5 16.5 13.4 3.8 2.2Machinery and equipment 7.5 36.5 61.7 20.6 12.2 22.2 4.6 2.7Office machinery 8.8 39.6 75.6 22.2 11.6 24.8 5.5 2.9Other elect machinery 12.6 42.1 72.4 30.0 17.4 13.8 4.1 2.4Radio, TV and communication eqpt 46.2 64.2 86.8 71.9 53.2 7.0 5.0 3.7Scientific and other instruments 4.0 18.6 33.2 21.3 11.9 26.9 5.7 3.2Motor vehicles 7.8 30.3 57.4 25.7 13.6 17.1 4.4 2.3Furniture and other mafg 13.6 43.0 72.0 31.6 18.9 13.6 4.3 2.6Total 19.8101.3168.019.611.817.33.42.0Total - Coke petro. and nuc. fuel 20.2 100.6 166.8 20.1 12.1 16.6 3.3 2.0Total is the sum of industries listed in the table. Source: Own calculation using UVRs from Table 1 and value added, total hours and total employment data from ASI and EU KLEMS.SPECIAL ARTICLE
German value added. The Indian value added is higher than that in Germany in textiles and leather, while in clothing and basic metals it is more than 50% of the German level.
In the fourth and fifth columns of the table, we provide relative labour productivity in India compared to Germany. The relative labour productivity levels, both in terms of hours worked and number of employees, are much smaller in India than in Germany, i e, 20% of the German productivity level in terms of value added per employee, and only 12% in terms of value added per hour. This holds true for most industries except radio, television and communication equipment, whereas labour productivity in terms of employment is quite closer to Germany. Thus our results indicate that, while the Indian manufacturing employs almost the same number of employees as the German manufacturing sector do, it produces only one-fourth of the German output level, thus resulting in low labour productivity.
The last three columns of Table 3 provide India’s cost competitiveness, expressed in terms of relative unit labour cost and relative wage rates. Comparisons of unit labour costs are often used in evaluating the competitive position of countries. Following the ICOP methodology, we have calculated this as a ratio of compensation to employees in Indian manufacturing, converted to euros using the exchange rate, to value added in Indian manufacturing, converted to euros using unit value ratios. This ratio is expressed as per cent of the corresponding ratio calculated for Germany (see Equation 5).
The table shows that the ULC in Indian organised manufacturing was only 17% of Germany. This varies from 7% in radio, television and communication equipment industry to 36% in wood and wood products industry. In general, the results for relative unit labour costs for manufacturing branches shows that almost all branches have cost advantage over Germany in terms of unit labour costs. Also, as evident from the last two columns, India pays only 2% of German hourly compensation and about 3.5% of German compensation per employee. This suggests that the lower labour productivity in Indian manu facturing is matched with the very low labour compensation rates. The relative labour compensation is much lower than the relative labour productivity level, as is visible from the ULC, suggesting that the sector enjoys cost advantage over the German manufacturing.
Relative Labour Productivity Levels and ULC, 1980-2004: We also derived a series of relative labour productivity and ULC over the period 1980-2004, using the value added, employment, hours worked and labour compensation data for Germany and India for aggregate manufacturing.14 Labour productivity was extrapolated for the whole period using the 2002 benchmark labour productivity estimate.15 The results are depicted in Figure 1.
Figure 1 shows that the Indian manufacturing has improved its productivity over years, from 8% in 1980 to 22% in 2004, when value added per employee is considered. Nevertheless, the rate of catch-up is much lower, when hours worked is considered for the productivity concept. It has improved from only 6% in 1980 to 12% in 2003. This difference is largely due to the decline in the number of hours in Germany and continuous increase in hours
Economic & Political Weekly EPW april 11, 2009 vol xliv no 15
Figure 1: Relative Labour Productivity, Labour Compensation and Unit Labour Cost in Indian Manufacturing (1980-2004, Germany=100) 80
70 Unit labour cost 60 Exchange rate 50 Compensation/person 40
30
Value added/person Compensation/hour
Value added/hour
20 10 0
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004
Source: Exchange rates from Reserve Bank of India and others, own calculation.
Figure 2: Relative Labour Productivity (Value Added/Person) in Manufacturing (US=100) 60
50
40
| 30 | Brazil | Korea | China Hungary | Poland | |
|---|---|---|---|---|---|
| Mexico | Indonesia | ||||
| 20 | India |
10
0
1980 1982 1986 1988 1992 1998 2004
1984 1990 1994 1996 2000 2002 2005
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004
Source: For China, van Ark et al (2007), for India, own calculations and for other countries, ILO Key Indicators of the Labour Market.
Figure 3: Relative Labour Productivity (Value Added/Hour) in Manufacturing (US=100) 45
40
35
30 Poland
Mexico Hungary 25
Korea
20
15
India 10
5
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2005
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004
Source: As in Figure 2.
worked in India.16 More importantly, the rate of catch-up appears to be lower during the 1990s compared to that of 1980s. This is true in most industries as well. However, there is a sign of slight improvement in productivity catch-up in the 2000s. Although the relative labour productivity level has shown a deceleration in 2000 and in 2001, it has started to improve again during the last three years. Overall, the productivity catch-up in the Indian manufacturing has not been very impressive. Even after quarter of a century, and despite the significant economic reforms, the Indian productivity still hovers around 20% of the German productivity level.
As far as the ULC is concerned, it shows a decline during the 1980s from 66% in 1980 to 22% in 1991, and remained stagnant
45
Korea Hungary Poland India China 
SPECIAL ARTICLEEconomic & Political Weekly EPW april 11, 2009 vol xliv no 1547productivity has been noted by Lee et al (2006).19 It is often arguedthat China could increase its productivity by resource re-location from low productive sectors to high productive sectors (The Conference Board 2008). Though the Chinese unit labour cost is still slightly higher than that of India’s, the decline in Chinese ULC has been very signifi-cant over 1995-2002, compared to that of India. More importantly, the decline in ChineseULC is not accompanied by a decline in its average labour cost; rather it is accompanied by an increase in labour productivity. Its average labour cost (as a per cent of US) hasincreased from 2.1 in 1995 to 2.9 in 2002, while the labour productivity has registered an even higher increase, say from 5.9 to 13.7. At the same time, the labour productivity in India has declined from 12.4% in 1995 to 12.2% in 2002, along with a decline in average labour cost from 2.9% to 2.4%. In Table 4, we compare the labour productivity in India and China relative to the US for 17 two-digit industries. In almost 10 in-dustries, India’s productivity is lower than that of China. In particu-lar, the industries like food, beverages and tobacco and wood and wood products have registered very low productivity levels. How-ever, some industries such as chemicals, which have a larger share in the Indian manufacturing production basket, have shown im-pressive labour productivity levels compared to China. Other in-dustries that have shown an advantage over China include rubber and plastics, basic metals, fabricated metals and office, accounting, other electrical machinery and radio-television equipment. While only the basic metals in India has shown a labour productivity of more than 25% of the US, China has shown such a trend in more industries, including machinery and equipment and motor vehi-cles, where its productivity is as high as 40% of the US level. 4 ConcludingRemarksThis paper has presented new and up to date results on unit value ratios, labour productivity and unit labour costs for Indian manu-facturing in comparison with some developing and developed countries. These figures help one understand the competitive position of Indian manufacturing from an international perspective. Using two extensive datasets on quantities and values of manufactured products in India and Germany, we have derived the relative prices between these two countries which are subsequently used to express the output values in a common currency. We have observed that though the labour productivity in Indian manufacturing has improved over the past quarter of a century, it is still much lower than that of the advanced countries and most developing countries. More importantly, the rate of productivity catch-up is quite slow. India’s unit labour cost is the lowest among the countries we have considered for comparison. Moreover, over yearstheULC has shown a declining tendency, indicating that the country maintains a competitive advantage interms of unit labour costs. However, its relative performance compared to that of China has not been very impressive. The competitive position of Indian manufacturing seems to be quickly eroding both in terms of productivity and ULC com-pared to that of China. Thus the intrinsic features of Indian manufacturing seem to be its low and slow-growing labour productivity, low and declining unit labour cost and the erosion of cost competitiveness compared to China. While China could attain lower ULC by improving its productivity, India’s lower ULC has been largely due to lower wage rates, which may not be sustainable in the long run. It is of high importance for Indian manufacturing to gain betterproductivityinorder tocatchup with the frontier countries. An important limitation of this study is that it is limited only to the organised manufacturing sector in India. A major chunk of Indian manufacturing production takes place in the unorgan-ised segment, which is not considered for our analysis due to lack of appropriate data. Hence, future analysis should include unorganised segment of Indian manufacturing in order to get a more realistic picture. Notes 1 Recent evidence shows that the competitive land-scape is uneven across sectors and states in India (Dougherty et al 2008). 2 For instance, previous studies have shown that the total factor productivity growth in Indian manufac-turing has not been very impressive, particularly not in the later phase of the reforms (see for e g, Herd and Dougherty 2007; Goldar and Kumari 2003; Balakrishnan et al 2000). Goldar and Kumari (2003) have argued that this is because of the influ-ence of certain adverse factors rather than the eco-nomic reforms, such as the decline in agricultural growth and deterioration in industrial capacity uti-lisation. Erumban (2005) provides evidence that capacity utilisation in the manufacturing sector has been fluctuating in the post-reform period. 3 While the exchange rate is deficient in that it does not account for differences in purchasing power of different currencies, the expenditure PPPs, de-rived using final expenditure prices include the price of goods imported by a country, but pro-duced elsewhere, and exclude the price of goods and services exported from a country. They do not reflect relative producer prices. 4 See Kravis (1976) for a summary of earlier works. 5 See Timmer et al (2001) for an elaborated discus-sion on this methodology. 6 Gross output weights are more appropriate than value added weights, as the product UVRs refer to gross output level. 7 One might argue that with greater international tradability of capital and intermediate inputs, labour input is the key determinant of cost com-petitiveness as it is much less mobile across countries. 8 More importantly, it does not comply with Indian classifications itself (NIC 87 and NIC 98). For in-stance, the product code for beef in ASICC is 11202, which belongs to the industry group beef slaughtering and preparation. This code has no correspondence with the corresponding industry codes; this product belongs to industries 15112 (5-digit), 1511 (4-digit) and 151 (3-digit) under NIC 98. It is also strange that this 4-digit code 1511 which represents the meat industry in NIC 98 cor-responds with alcohol in ASICC; the ASICC for al-cohol absolute and edible is 15111. 9 Note that, however, such aggregation was possi-ble only if all products were expressed in same unit of measurement. If they were expressed in different units, we opted to exclude them from the sample. However, such cases were quite mar-ginal across the components of a given product group within India or Germany, though the prob-lem was very large, while comparing German units with Indian units.10 For instance, optical fibre cable in India is ex-pressed in kilometres, while they are expressed in kilograms in Germany. 11 It may be noted that EPWRF has recently brought out a revised ASI database that covers the period from 1973-74 to 2003-04. This data was not avail-able at the time of the analysis, and is therefore, not used in our analysis.12 We use the March 2008 release of the EU KLEMS data, which is available at http://www.euklems.net.13 As mentioned before, we have tried to make a like-with-like comparison whenever possible. However, part of the divergence could still be at-tributed to quality differences. More importantly, output values are often calculated for product groups rather than specific products causing com-parability problems at a disaggregated level. 14 We have also done this exercise for two-digit in-dustries. However, the results for two-digit manu-facturing branches are not reported here. They are available from author upon request. Also see Erumban (2007).15 Output growth has been calculated using con-stant price value added in national currencies.16 Total hours worked in India increased from 13,471 million (106% of Germany) in 1980 to 17,764 mil-lion (170% of Germany) in 2003, while the German hours declined from 12,743 million to 10,432 million.
SPECIAL ARTICLEapril 11, 2009 vol xliv no 15 EPW Economic & Political Weekly48 17 Note that for the years before euro has been intro-duced we have calculated the euro-INR exchange rate using the conversion factor between euro and German mark and the exchange rate between German mark and INR. 18 This high and increasing ULC along with high labour productivity in South Korea has been pre-viously highlighted by van Ark et al (2005). 19 Van Ark et al (2006) also provide a crude estimate for the productivity of all manufacturing in India (including unregistered manufacturing) and China (including firms below township level). This shows a much wider gap between India and China at 2.4% and 4.6% of the US level, respec-tively. Lee et al (2007) shows even larger gap be-tween India and China in manufacturing labour productivity: in 2002 the Indian productivity was less than one-third of Chinese.ReferencesAnderson, J E and E Wincoop (2004): “Trade Costs”, Journal of Economic Literature,42, 691-751.Balakrishnan, P, K Pushpangadan and M Suresh Babu (2000): “Trade Liberalisation and Productivity Growth in Manufacturing: Evidence from Firm Level Panel Data”,Economic & Political Weekly, 35(41): 3679-82.Dougherty, S M, R Herd, T Chalaux and A A Erumban (2008): “India’s Growth Pattern and Obstacles to Higher Growth”, OECD Economics Department Working Papers No 623.Engel, C and J H Rogers (2001): “Deviations from Pur-chasing Power Parity: Causes and Welfare Costs”, Journal of International Economics, 55, 29-57.Erumban, A A (2005): “Economic Reforms and Indus-trial Performance: An Analysis of Economic Capa-city Utilisation in Indian Manufacturing”,Indian Journal of Economics and Business, 4(2), 305-23. – (2007): “Productivity and Unit Labour Cost in In-dian Manufacturing: A Comparative Perspective”, GGDC Research Memorandum GD-96, University of Groningen.Goldar, B and A Kumari (2003): “Import Liberalisa-tion and Productivity Growth in Indian Manufac-turing Industries in the 1990s”,Developing Econo-mies, December, 41(4): 436-60.Hadi, A S (1992): “Identifying Multiple Outliers in Multivariate Data”,Journal of the Royal Statistical Society, Series (B), 5s4, 761-71.– (1994): “A Modification of a Method for the Detection of Outliers in Multivariate Samples”, Journal of the Royal Statistical Society, Series (B), 56, 393-96.Herd, R and S Dougherty (2007): “Growth Prospects in China and India Compared”,The European Journal of Comparative Economics, 4 (1), 65-89.Kravis, I B (1976): “A Survey of International Compari-sons of Productivity”,The Economic Journal, Vol 86, March, 1-44.Lee, B L, D S P Rao and W Shepherd (2007): “Compari-sons of Real Output and Productivity of Chinese and Indian Manufacturing, 1980-2002”,Journal of Development Economics, 84 (1), 378-416.Maddison, A and B van Ark (2002): “The International Comparison of Real Product and Productivity” in A Maddison, D S Prasada Rao and W F Shepherd (ed.),The Asian Economies in the Twentieth Century (Cheltenham: Edwar Elgar), 5-26.OECD (2007): OECD Economic Surveys, India 2007 (Paris: OECD). Paige, D and G Bombach (1959): A Comparison of Na-tional Output and Productivity (Paris: OEEC).Szirmai, A, R Ren and M Bai (2005): “Chinese Manu-facturing Performance in Comparative Perspec-tive, 1980-2002”, Economic Growth Center Work-ing Paper 920, Yale University.Taylor, A M and M P Taylor (2004): “The Purchasing Power Parity Debate”,Journal of Economic Perspectives, 18(4), 135-58.The Conference Board (2008):Performance 2008: Productivity, Employment, and Growth in the World’s Economies,Research Report, R-1421-08-RR (New York: The Conference Board). Timmer, M P and A Szirmai (1999): “International Comparison of Productivity Performance in Man-ufacturing in South-and East Asia, 1963-1993”, Oxford Development Studies, 27(1), 57-79.Timmer, M P (2000):The Dynamics of Asian Manu-facturing: A Comparative Perspective in the Late Twentieth Century (Cheltenham: Edward Elgar).Timmer, M P, B van Ark, N Mulder, L Nayman and D Ünal-Kesenci (2001): Formalisation of the ICOP Methodology for Binary Manufacturing Comparisons (Groningen and Paris: University of Groningen and CEPII) (mimeo).Timmer, M P, M O’Mahony and B van Ark (2008): The EU KLEMS Growth and Productivity Accounts: An Overview (University of Groningen and University of Birmingham).van Ark, B, A A Erumban, V Chen and U Kumar (forthcoming): “The Cost Competitiveness of the Manufacturing Sector in China and India: An Industry and Regional Perspective” in B Eichen-green, P Gupta and R Kumar (ed.),Emerging Giants: China and India in the World Economy (UK: Oxford).van Ark, B, E Stuivenwold and G Ypma (2005): “Unit Labour Costs, Productivity and International Competitiveness”, Groningen Growth and Devel-opment Centre Working Paper, GD-80.van Ark, B, J Banister and C Guillemineau (2006): “Competitive Advantage of “Low-Wage” Coun-tries Often Exaggerated”,Executive Action Series, No 212, China Centre for Economics and Business, The Conference Board. Sages
