SPECIAL ARTICLE Trade Liberalisation and Labour Demand Elasticity in Indian Manufacturing
Bishwanath Goldar
The hypothesis that trade liberalisation raises labour demand elasticity is tested for Indian industries, and inter-temporal changes in the elasticity during 1973-74 to 2003-04 are analysed. Econometric results indicate that trade liberalisation in India had a positive effect on the labour demand elasticity. However, the estimated elasticity for the post-reform period (1991 onwards) is found to be lower than that for the pre-reform period. A closer examination reveals a marked upward trend in the labour demand elasticity after the mid-1990s, which seems attributable, among other factors, to trade liberalisation, weakening of trade union bargaining power and labour market reforms.
An earlier version of the paper was presented at the Fourth Annual Conference on Economic Growth and Development, held at the Indian Statistical Institute, Delhi, 17-18 December 2008. I have benefited from the comments received at the conference.
Bishwanath Goldar (bng@iegindia.org) is with the Institute of Economic Growth, Delhi.
Economic & Political Weekly
EPW
T
There are reasons to believe that trade liberalisation will lead to an increase in the (absolute value of) labour demand elasticity, i e, the elasticity of labour demand with respect to wage rate (Hamermesh 1993; Rodrik 1997), and the above-mentioned studies have tried to verify this hypothesis empirically. The postulated increase in labour demand elasticity arising from trade liberalisation has important implications for the labour market outcomes, especially in developing countries. As argued by Rodrik (1997), an increase in labour demand elasticity will lead to larger employment and wage shocks emanating from shocks in productivity or output demand. Also, greater volatility in employment and wages would cause an erosion of the bargaining power of l abour vis-à-vis capital in sharing of supernormal profits. Thus, the effect of trade liberalisation on labour demand elasticity is an important issue to examine.
Why should trade liberalisation lead to an increase in labour demand elasticity? This is traceable to a substitution effect and a scale effect (Hasan et al 2007). Under competitive conditions, the elasticity of demand for labour of a firm depends on: (a) the elasticity of substitution between labour and other inputs, (b) the price elasticity of demand for the products produced by the firm, and
(c) the share of labour cost in total cost of production. Trade liberalisation is expected to raise the elasticity of substitution between labour and other inputs since more and better intermediate inputs become available. Opening up the domestic markets to imports is expected to raise the price elasticity of demand for products of domestic firms since there is greater availability of substitutes for any product. Accordingly, one would expect the labour demand elasticity to increase with trade liberalisation. This, however, need not always happen. Trade liberalisation may lead to a lowering of the cost share of labour because semi-finished or unassembled products may be imported by industrial firms for their use in the production process instead of manufacturing from the raw materials stage, and this may neutralise the effects of increased elasticity of substitution among inputs and increased price elasticity of demand for the products of domestic industrial firms.
Studies undertaken for developing countries have, in general, not found empirical support for the above-mentioned hypothesis that trade liberalisation raises the labour demand elasticity in i ndustries. Empirical results obtained in the studies undertaken by Krishna et al (2001) for Turkey, Haouas and Yagoubi (2004) for Tunisia, and Fajnzylber and Maloney (2005) for Mexico, Chile and Columbia provide no support or only weak support to the hypothesis that trade liberalisation raises labour demand elasticity. By contrast, the study undertaken by Hasan et al (2007) for Indian industries has come up with empirical evidence that clearly indicates a positive impact of trade liberalisation on labour demand elasticity.
This is the second study on the same issue carried out for Indian industries. The aim is to verify the findings of Hasan et al (2007) using a somewhat different data set.1 Also, inter-temporal changes in labour demand elasticity are studied for a period e xtending to more recent years. The purpose is to judge whether any marked increase in the elasticity has taken place in the postreform period (1991 onwards) in response to the trade liberalisation (and other reforms) undertaken.
The rest of the paper is organised as follows: The first section describes the data sources, construction of variables and the models estimated. The econometric results on the relationship between trade liberalisation and labour demand elasticity in I ndian industries are presented and discussed in Section 2. Intertemporal changes in the labour demand elasticity in Indian i ndustries in the pre- and post-reform periods are analysed in S ection 3. An attempt is made to relate inter-temporal changes in labour demand elasticity to easing of labour market rigidities in Section 4. Finally, Section 5 summarises and concludes.
1 Data, Model and Variables
We now delve into the sources of the data, construction of the variables, and specification of the model.
1.1 Data Sources
The basic source of data for this study is the Annual Survey of I ndustries (ASI) published by the Central Statistical Organisation, government of India. Three datasets, formed on the basis of ASI, have been used for the analysis – one is used for the analysis in Section 2 and the other two (an industry-wise panel and a statewise panel) are used for the analyses in Sections 3 and 4.
The analysis presented in Section 2 is based on data for 137 three-digit industries for the period 1980-81 to 1997-98 (drawn from an electronic database brought out by the Economic and Political Weekly Research Foundation (EPWRF) in 2002 using the results of the ASI).2 This is the same data set as used in Goldar and Aggarwal (2005). The main data source on tariff rates and non-tariff barriers (percentage import coverage by quantitative restrictions) is a research project undertaken at the Indian Council for Research on International Economic Relations (ICRIER), the results of which are reported in Das (2003). For a majority of the three-digit industries, data on import barriers could be obtained from this source. Since Das has not covered all three-digit industries of ASI, 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). In a number of cases, the estimate available for an input-output sector has been applied
52 to all three-digit industries belonging to that sector. It has also 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, as quantitative restrictions covered a very high proportion of imports of manufactures throughout the decade.3
The EPWRF has recently brought out an updated version of their electronic database on Indian industries using ASI results. This database is for the period 1973-74 to 2003-04. Data for 23 two-digit industries4 (belonging to manufacturing) for the period 1973-74 to 2003-04 have been used for the analysis of trends in labour demand elasticity in Section 3. Data on tariff and nontariff barriers could not be obtained at two-digit industry level for the entire period, and therefore these variables representing trade protection have not been incorporated in the estimation of labour demand function done at two-digit industry level in Section 3. A similar analysis based on state-level data (15 major states,5 aggregate ASI for each state) is also presented in this section. In this case again, variables representing trade protection have not been included in the estimation of labour demand function.
1.2 Model and Variables
The model used for the econometric analysis in Section 2 is similar to one of the specifications used by Hasan et al (2007). This is shown below. L = f(L-1, w, w*TR, w*QR, Y) ...(1)
In the above equation, L is labour, L-1 labour with one year lag, w real wage rate (product wage), TR tariff rate (per cent), QR quantitative restrictions (import coverage ratio, per cent), and Y output (real gross value added). A log-linear specification is used. Thus, the equation estimated may be written as: ln(L) = α+ β1 ln(L-1) + β2 ln(w) + β3 ln(w)*TR +β4 ln(w)*QR
+ β5 ln(Y) + u. ...(2) where u is the random error term.
Panel data for 137 industries for 18 years has been used for estimating the above equation. The estimation of parameters has been done by the one-step Generalised Method of Moments I nstrument Variable (GMM-IV) estimator.
Deflation of gross value added of different industries has been done with wholesale price indices (the best index that could be obtained for each industry from the official series). The number of persons engaged is taken as the measure of labour. The wage rate is computed by dividing total emoluments by the number of persons engaged. The wage rate so computed for each industry is deflated by the wholesale price index that has been used to deflate gross value added of the industry to obtain the product wage rate.
The coefficient of Y is expected to be positive and that of w negative. The coefficient of lagged labour variable, L-1, is expected to be a positive fraction. The hypothesis that trade liberalisation raises the elasticity of employment with respect to wage rate r equires β3 and β4 to be positive.
An alternative approach taken to estimate the employment function is to regress the five-year difference in ln(L) on five-year
august 22, 2009 vol xliv no 34
SPECIAL ARTICLE
Table 1: Labour Demand Function, Estimate Based on Three-digit Industry Data, GMM-IV
Explanatory Variables Regressions
(1) (2)
ln(L) 0.159 (6.9)*** 0.166 (7.2)***
ln(Y) 0.402 (33.7)*** 0.403 (33.6)***
ln(w) -0.179 (-6.1)*** -0.176 (-6.0)***
ln(w)*TR 0.0003 (3.9)***
ln(w)*QR 0.0003 (2.3)**
No of observations 2192 2192
Sargan test of over-identifying restrictions, χ2 (135) =445.0 χ2 (135) =451.7 Chi-square and probability value P=0.000 P=0.000
A-B test of autocorrelation#, z and z = 1.23 z = 1.39 probability value P= 0.22 P= 0.17
Wald test, Chi-square χ2 (4) =1275.5 χ2 (4) =1253.8
#Arellano-Bond test that average auto covariance in residuals of order 2 is 0. Values in the parentheses are z-statistics; *** indicates significance at 1% level, and ** 5% level. The equation is estimated from panel data for 137 industries for 18 years, 1980-81 to 1997-98.
differences in ln(w), ln(Y), ln(w)*TR and ln(w)*QR. The estimation of the regression equation is done by the Ordinary Least Squares (OLS) method. This follows the approach taken by Hasan et al (2007). In these estimates, the lagged labour variable is dropped.
As mentioned earlier, the analysis presented in Section 3 covers the period 1973-74 to 2003-04 using a data set at two-digit industry level and another one at state-level. In these cases, the trade protection variables could not be incorporated in the estimated employment function. The model therefore gets simplified to: ln(L) = α+ γ1 ln(L-1) + γ2 ln(w) + γ3 ln(Y) + u. ...(3)
The short-term elasticity of employment with respect to wage rate is given by γ2 and the long-run elasticity is given by γ2/(1-γ1). Since trade liberalisation and other changes in the economy are expected to influence the employment function parameters, this aspect needs to be incorporated into the analysis. One possibility is to introduce intercept and slope dummy variables for different sub-periods. This has not been done. Instead, the model shown in equation (3) above has been estimated separately for different sub-periods to examine the inter-temporal changes in labour d emand elasticity. Estimation of the model has been done by the one-step GMM-IV estimator.
2 Effect of Trade Liberalisation
Table 1 presents estimates of labour demand function for Indian manufacturing industries based on the specification in equation (2). The estimated coefficients have the expected sign and are of plausible magnitude. The coefficients are statistically significant. The interaction terms involving wage rate with tariff rate and quantitative restrictions have been used in separate equations, because there is a high correlation among the two variables representing trade restriction, and the coefficient of one of the interaction terms (ln(w)*QR) becomes statistically insignificant when both terms are included in the same equation.
It is seen from the table that the coefficients of ln(w)*TR and ln(w)*QR are positive and statistically significant, which implies that trade liberalisation tends to increase labour demand elasticity.6 This corroborates the findings of Hasan et al (2007).
The estimates of the labour demand function based on fiveyear differences in the variables are presented in Table 2. As in the results reported in Table 1, the estimated coefficients presented
Economic & Political Weekly
EPW
in Table 2 have the expected sign, are of plausible magnitude, and are statistically significant. The coefficients of ln(w)*TR and ln(w)*QR are positive and statistically significant. This indicates that lowering of tariff rates and relaxation of quantitative restrictions on imports would raise the labour demand elasticity. This is again in conformity with the results reported in Table 1 and with the estimates of a similar equation reported in Hasan et al (2007).
Since the lagged labour variable is dropped from the estimated equation based on five-year differences, the estimated parameters may be interpreted as long-run elasticities. It would be noted that the coefficients of ln(Y) and ln(w) in the results reported in Table 2 are higher than those in Table 1.
3 Inter-Temporal Changes in Labour Demand Elasticity
Although the econometric results presented in Tables 1 and 2 indicate a positive effect of trade liberalisation on labour demand elasticity, the estimates of employment function based on the specification in equation (3) do not show any increase in labour demand elasticity in the post-reform period as compared to the pre-reform period. The estimates based on the industry-wise panel data of 23 industries (two-digit industries, 1973-74 to 2003-04) are presented in Table 3. The estimated function for the period 1973-74 to 1990-91 gives a labour demand elasticity of 0.41 (absolute value) compared to which the estimate for the period 1991-92 to 2003-04, at 0.27, is lower. The same pattern is observed when a comparison is made between the estimates of employment function for the periods 1980-81 to 1990-91 and 1991-92 to 2003-04. Given that there was a sharp reduction in the tariff rates7 and an almost complete removal of quantitative restrictions (QRs) on imports of manufactures in the post-reform
Table 2: Labour Demand Function, Estimate Based on Three-digit Industry Data, OLS on Five-Year Differences
| Explanatory Variables | Regressions | ||
|---|---|---|---|
| (1) | (2) | (3) | |
| ln(Y) | 0.506 (46.4)*** | 0.507 (46.6)*** | 0.506 (46.5)*** |
| ln(w) | -0.291 (-10.8)*** | -0.299 (-11.0)*** | -0.305 (-11.1)*** |
| ln(w)*TR | 0.0002 (2.6)*** | 0.0002 (1.9)* | |
| ln(w)*QR | 0.0004 (3.2)*** | 0.0003 (2.7)*** | |
| No of observations | 1781 | 1781 | 1781 |
| R-squared | 0.56 | 0.56 | 0.56 |
Values in the parentheses are t-statistics; *** indicates significance at 1% level, and * 10% level. Dependent variable: L (labour). The equation is estimated from panel data for 137 industries for 18 years, 1980-81 to 1997-98, with five-year differences in the variables.
Table 3: Labour Demand Function, Estimate Based on Two-digit Industry Data, GMM-IV
| Explanatory Variables | Regressions for the Period | ||
|---|---|---|---|
| 1973-74 to 1990-91 | 1980-81 to 1990-91 | 1991-92 to 2003-04 | |
| ln(L-1) | 0.485 (13.3)*** | 0.414 (8.3)*** | 0.635 (18.0)*** |
| ln(Y) | 0.261 (11.6)*** | 0.270 (10.4)*** | 0.268 (9.9)*** |
| ln(w) No of observations | -0.411 (-12.0)*** 368 | -0.394 (-9.2)*** 253 | -0.271 (-6.6)*** 299 |
| Sargan test of over-identifying | χ2 (135) =205.7 | χ2 (120) =167.7 | χ2 (298) =275.3 |
| restrictions, Chi-square and | P=0.001 | P=0.003 | P=0.823 |
| probability value | |||
| A-B test of autocorrelation#, | z = 0.26 | z = -0.63 | z = 1.03 |
| z and probability value | P= 0.80 | P= 0.53 | P= 0.31 |
Wald test, Chi-square χ2 (3) =686.1 χ2 (3) =385.9 χ2 (3) =725.9
#Arellano-Bond test that average autocovariance in residuals of order 2 is 0. Values in the parentheses are z-statistics; *** indicates significance at 1% level. The equation is estimated from panel data for 23 industries for different sub-periods during 1973-74 to 2003-04.
p eriod,8 a hike in the labour demand elasticity is expected. But, this is not borne out by the estimates of employment function for the pre- and post-reform periods.
To analyse the trends in labour demand elasticity further, the employment function (based on the specification in equation (3)) has been estimated for various eight-yearly interlocking samples within the period 1973-74 to 2003-04.9 The estimates of shortand long-run elasticity of labour demand with respect to wage rate obtained for various sub-periods are shown in Figure 1.
From the elasticity estimates presented in Figure 1, it appears that there was a downward trend in labour demand elasticity in Indian industry in the pre-reform period. This trend was arrested and reversed from the mid-1990s. The estimated short-run labour demand elasticity for the period 1996-97 to 2003-04 is markedly higher than the estimate for the period 1985-86 to 1995-96, 0.35 as against 0.21 (the long-run elasticity estimates are 0.7 and 0.5, respectively).
The analysis of state-level ASI data for the industrial sector of different states shows by and large the same inter-temporal pattern of changes in labour demand elasticity as revealed by the analysis of industry-level data. As mentioned earlier, data for 15 major states are considered for the state-level analysis, and the period covered is 1973-74 to 2003-04. The construction of variables, the specification of the labour demand function and the estimation method are the same as in the estimates based on industry level data.10
Estimates of the labour demand function, based on state-level data, for the periods 1973-74 to 1984-85, 1985-86 to 1995-96 and 1996-97 to 2003-04 are reported in Table 4. The estimated shortand long-run labour demand elasticity for interlocking eight yearly samples are shown in Figure 2. There is clear indication from the table and the graph that the labour demand elasticity had declined in the period 1985-95 compared to the period 1973-84 and after the mid-1990s there was a significant increase in labour demand elasticity. This is consistent with the results reported above.
The reasons for the observed downward trend in labour demand elasticity in Indian manufacturing in the pre-reform p eriod (Figures 1 and 2) are not clear. It seems one contributing factor was the fall in the share of labour cost in total production cost.11 The ratio of emoluments to total value of output almost steadily declined in the period 1973-74 to 1990-91, and the same
Table 4: Labour Demand Function, Estimate based on State-level Data, GMM-IV
| Explanatory Variables | Regressions for the Period | ||
|---|---|---|---|
| 1973-74 to 1984-85 | 1985-86 to 1995-96 | 1996-97 to 2003-04 | |
| ln(L-1) | 0.734 (12.6)*** | 0.542 (7.9)*** | 0.261 (4.2)*** |
| ln(Y) | 0.195 (5.9)*** | 0.109 (3.3)*** | 0.137 (3.8)*** |
| ln(w) | -0.309 (-6.3)*** | -0.069 (-1.0) | -0.355 (-4.1)*** |
| Dummy for 1998-2003 No of observations | 180 | 165 | -0.148 (-7.0)*** 120 |
| Sargan test of over-identifying | χ2 (89) =104.0 | χ2 (208) =158.2 | χ2 (227) =117.2 |
| restrictions, Chi-square and | P=0.13 | P=0.996 | P=1.000 |
| probability value | |||
| A-B test of autocorrelation#, | z = 2.15 | z = -0.08 | z = 0.63 |
| z and probability value | P= 0.03 | P= 0.94 | P= 0.53 |
Wald test, Chi-square χ2 (3) =365.1 χ2 (3) =116.2 χ2 (4) =128.2
#Arellano-Bond test that average autocovariance in residuals of order 2 is 0. Values in the parentheses are z-statistics; *** indicates significance at 1% level. The equation is estimated from panel data for the manufacturing sector of 15 states for different sub-periods during 1973-74 to 2003-04.
trend continued in subsequent years. This may be seen from Figure 3. It is remarkable that even though the downward trend in the share of labour cost continued beyond the mid-1990s, there was a marked increase in the elasticity of demand for labour.
Figure 1: Short- and Long-run Labour Demand Elasticity (for eight years ending each time point)
0.8
0.2 0.3 0.4 0.5 0.6 0.7 Long run elasticity Short run elasticity 0.1 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Source: Author’s calculation and ASI data.
Figure 2: Short- and Long-run Labour Demand Elasticity (estimates based on data for states, for eight years ending each time point)
0 0.2 0.4 0.6 0.8 1 Short run elasticity Long run elasticity 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Source: Author’s calculation and ASI data. Figure 3: Ratio of Emoluments to Value of Output (1973-74 to 2003-04, in %)
14
12
10 8 6
4
1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 Source: Author’s calculation and ASI data.
In view of the results reported in the previous section, the hike in labour demand elasticity from the mid-1990s seems to be a ttributable in a significant measure to trade liberalisation. It may be mentioned in this context that while substantial cuts in industrial tariff rates were made in the first half of the 1990s, the rates were quite high at the beginning of the 1990s and there was considerable water in tariff12 with the consequence that the tariff cuts in the initial years of economic reforms might not have exposed Indian manufacturing firms to any significantly enhanced import competition. The reduced level of tariff reached in the second half of the 1990s (and further cuts later) probably enabled imported manufactured goods to offer sufficient competition to the local manufactures so as to raise the price elasticity of
august 22, 2009 vol xliv no 34
SPECIAL ARTICLE
demand for locally produced manufactured products. The r emoval of QRs on imports of a large number of manufactured consumer goods in 2000 and 2001 (about 1,400 items; Goldar 2005) and other measures taken in the second half of the 1990s to ease imports of consumer goods may have substantially strengthened this force. In a sense, therefore, the trade liberalisation demonstrated its real impact in India only from the mid1990s and this probably explains why a rise in labour demand elasticity is found from the mid-1990s, and not before.
At the same time, it needs to be recognised that there were possibly other factors at work. A comprehensive study of the factors that led to an increase in labour demand elasticity from the mid-1990s is beyond the scope of this paper. Nonetheless, it would be useful to consider the impact of labour regulation on employment and how increased labour market flexibility may have contributed to higher labour demand elasticity.
Two aspects of industrial labour market deserve attention here. The first is about the bargaining power of trade unions. There is a view that the trade unions became weaker in terms of their bargaining power in the post-reform period (Sharma 2006: 2083). It may be pointed out in this context that the number of disputes per 10,000 workers fell from 1.1 in 1989 to 0.5 in 2002. Also, there has been a downward trend in the number of mandays lost due to strikes and an upward trend in that lost due to lockouts.13 This hints at a possible change in the power balance in industrial firms against the workers and in favour of the management. It would not be unreasonable therefore to argue that the observed hike in the labour demand elasticity in Indian industry in the period after the mid-1990s may be connected with weakening of the bargaining power of trade unions vis-à-vis the management.
The second aspect deserving attention is the labour market rigidities arising from the enforcement of labour laws. There are indications that the level of rigidity has declined over time. Sharma (2006) notes that several Indian states have relaxed the provision of enforcement of labour laws leading to flexible practices at the ground level. There is other literature that supports this view14 (Papola 2008). It appears therefore that the increase in labour demand elasticity in the period since the mid-1990s may partly be a reflection of the growing flexibility in the industrial labour market in India.
4 Impact of Growing Flexibility in Industrial Labour Market
Data required for a thorough analysis of the impact of growing flexibility in the industrial labour market in India on labour d emand elasticity are not readily available. A rough and ready assessment is, however, possible with the help of the databases used in the previous section along with additional information on the labour market rigidities in various industries and states. With this aim, the industries and states covered in the study have been divided into groups according to certain characteristics of the labour market and then the estimated employment function for the different groups have been compared to assess which groups had a relatively greater increase in the elasticity after the mid-1990s. From the results obtained, it seems reasonable to
Economic & Political Weekly
EPW
conclude that growing flexibility in labour market has contributed to a hike in labour demand elasticity in Indian industries.
For the analysis done with the help of the industry-wise panel data, trade union membership (i e, the degree of unionisation) has been considered as a key feature reflecting labour market rigidity. A comparative analysis of labour demand elasticity for two groups of industries categorised on the basis of the degree of unionisation (relatively low unionisation and relatively high unionisation) has therefore been done.15 The results are reported in Table 5.
Table 5: Labour Demand Function, Estimate Based on Two-digit Industry Data, GMM-IV, Comparison among Industries according to the Level of Unionisation
Explanatory variables Regressions for the Industries Characterised by
Relatively Low Unionisation (8 industries)
Relatively High Unionisation (8 industries) 1985-86 to 1995-96 1996-97 to 2003-04 1985-86 to 1995-96 1996-97 to 2003-04
ln(L) 0.644*** (8.6) 0.261*** (4.1) 0.500*** (7.5) 0.551*** (6.9)
-1
ln(Y) 0.254*** (5.4) 0.322*** (7.2) 0.320***(6.5) 0.419***(4.7)
ln(w) -0.160(-1.5) -0.767***(-10.5) -0.206***(-3.3) -0.580*** (-4.7)
No of observations 88 64 88 64
Sargan test of over- χ2 (175) =81.9 χ2 (203) =89.7 χ2 (175) =87.7 χ2 (203) =61.2 identifying restrictions, P=1.0 P=1.0 P=1.0 P=1.0 Chi-square and probability value
A-B test of autocorrelation#, z = 1.15 z = -0.84 z = 1.21 z = 0.20 z and probability value P= 0.25 P= 0.40 P= 0.22 P= 0.84
Wald test, Chi-square χ2 (3) =200.3 χ2 (3) =188.3 χ2 (3) =183.6 χ2 (3) =97.3 #Arellano-Bond test that average autocovariance in residuals of order 2 is 0. Values in the parentheses are z-statistics; *** indicates significance at 1% level. The equation is estimated from panel data for 16 industries for two sub-periods during 1985-86 to 2003-04. Eight industries are chosen that rank relatively low in unionisation (in 1998 and 1999), while another eight rank relatively high in unionisation. The 23 industries covered in the study have been ranked according to unionisation and the top eight and bottom eight have been taken.
It is seen from Table 5 that in both groups of industries, marked by low unionisation and high unionisation, there has been an increase in labour demand elasticity after the mid-1990s. However, the increase has been relatively higher in industries characterised by low unionisation than that in industries characterised by relatively higher unionisation. This result is expected since higher unionisation is likely to constrain the ability of the firms to adjust employment.
Similar to the cross-industry analysis above, a comparison of labour demand elasticity across states has been made, after grouping states according to the degree of labour reforms undertaken. For this purpose, an index of labour reforms in different Indian states reported in a recent study by Dougherty (2008)16 has been used. The 15 states considered for the study have been divided into two groups. One group includes nine states in which the degree of labour reforms undertaken is above the national average. The other group includes six states in which the degree of labour reforms undertaken is below the national average.17 The estimated labour demand function for the two groups of states, states undertaking into relatively high and relatively low degree of labour reforms, are presented in Table 6 (p 56). It is seen from the table that in both groups of states there was an increase in l abour demand elasticity in the post-1995 period. However, the increase in the elasticity was relatively bigger among the states that had undertaken greater labour reforms.
An alternate index of labour regulation that may be used for the analysis is the one prepared by Besley and Burgess (2004). They have coded the amendments made by various states to
Table 6: Labour Demand Function, Estimate Based on State Level Data, GMM-IV, Comparison among States according to the Extent of Labour Reforms Undertaken
Explanatory Variables Regressions for the States Marked by
Relatively High Degree of Labour Reforms Relatively Low Degree of Labour Reforms (9 states) (6 states)
1980-81 to 1995-96 1996-97 to 2003-04 1980-81 to 1995-96 1996-97 to 2003-04
ln(L) 0.642*** (10.4) 0.201*** (2.8) 0.710*** (9.2) 0.287** (2.4)
ln(Y) 0.111*** (3.0) 0.285*** (4.9) 0.104**(2.4) 0.060(1.3)
ln(w) -0.157***(-2.8) -0.431***(-3.9) -0.222***(-3.1) -0.314** (-2.3)
Dummy for 1998-2003 -0.131***(-5.1) -0.145***(-4.2)
No of observations 144 72 96 48
Sargan test of over- χ2 (263) =125.3 χ2 (227) =69.2 χ2 (263) =86.3 χ2 (227) =42.5 identifying restrictions, P=1.0 P=1.0 P=1.0 P=1.0 Chi-square and probability value
A-B test of autocorrelation#, z = 0.70 z = 0.88 z = 0.10 z = 0.23 z and probability value P= 0.48 P= 0.38 P= 0.92 P= 0.82
Wald test, Chi-square χ2 (3) =145.2 χ2 (4) =114.1 χ2 (3) =187.8 χ2 (4) =33.5
findings of Hasan et al (2007) on the effects of labour market rigidities on labour demand elasticity.
The finding of a positive association between labour market flexibility and increase in labour demand elasticity in this section could be interpreted in two ways. This may mean that the growing labour market flexibility has, by itself, made the labour demand function more elastic. It may alternatively mean that the labour market flexibility has permitted trade liberalisation to have the expected positive effect on labour demand elasticity. In this case, trade liberalisation is the prime force leading to increased labour demand elasticity, labour market flexibility permits or obstructs the effect to occur. Perhaps, the observed increase in elasticity after the mid-1990s is attributable to
#Arellano-Bond test that average autocovariance in residuals of order 2 is 0. Values in the parentheses are z-statistics; *** indicates significance at 1% level, and ** 5% level. The equation is estimated from panel data for 15 states for two sub-periods during 1980-81 to 2003-04. The 15 states are divided into two groups. The first group includes nine states that rank relatively high (above national average) in terms of labour reforms undertaken. The second group includes the other six states that rank relatively low.
Table 7: Labour Demand Function, Estimate Based on State Level Data, GMM-IV, Comparison between Pro-employer and Pro-worker States
Explanatory Variables Regressions for the State Category
Pro-employer States (5 states) Pro-worker States (5 states)
1980-81 to 1995-96 1996-97 to 2003-04 1980-81 to 1995-96 1996-97 to 2003-04
ln(L) 0.272*** (2.8) 0.100 (0.9) 0.663*** (8.4) 0.298*** (3.6)
-1
ln(Y) 0.269*** (4.8) 0.199** (2.2) 0.148***(3.8) 0.270***(4.5)
ln(w) -0.409***(-5.8) -0.606***(-3.8) -0.185***(-2.9) -0.151 (-1.3)
Dummy for 1998-2003 -0.174***(-5.4) -0.080***(-2.7)
No of observations 80 40 80 40
Sargan test of over-identifying χ2 (263) =67.8 χ2 (227) =45.2 χ2 (263) =74.4 χ2 (227) =34.0 restrictions, Chi-square and P=1.0 P=1.0 P=1.0 P=1.0 probability value
A-B test of autocorrelation#, z = 0.75 z = -0.01 z = 1.03 z = 0.13 z and probability value P= 0.45 P= 0.99 P= 0.30 P= 0.89
a combination of the two effects.
5 Conclusions
Econometric results obtained in this study corroborate the findings of Hasan et al (2007) and provide support to the hypothesis that trade liberalisation raises labour demand elasticity. Although the econometric results reported in the paper along with the findings of Hasan et al (2007) clearly indicate that trade liberalisation had a positive effect on the labour demand elasticity in Indian industries, the estimated elasticity for the post-reform period is found to be lower than that for the pre-reform period. A closer examination of the data reveals that there was a downward trend in the labour demand elasticity in Indian industries in the pre-reform period, which continued for some years even after the initiation of reforms. It appears that the downward trend in the labour demand
Wald test, Chi-square χ2 (3) =103.1 χ2 (4) =70.9 χ2 (3) =242.0 χ2 (4) =105.0 elasticity was arrested and reversed since the mid-1990s.
#Arellano-Bond test that average autocovariance in residuals of order 2 is 0. Probably the effect of trade reforms occurred with a lag Values in the parentheses are z-statistics; *** indicates significance at 1% level, and ** 5% level. The equation is estimated from panel data for 10 states for two sub-periods during 1980-81 to 2003-04. Two or the effect became stronger from the mid-1990s. It groups of states are considered: pro-employer and pro-worker. The states included in the two groups are
listed in the text.
the Industrial Disputes Act between 1958 and 1992 as pro-worker, anti-worker (pro-employer) and neutral, and on that basis an i ndex of labour rigidity has been formed for different states. The index has recently been updated by Purfield (2006).18 Based on the updated index constructed, Purfield classifies Andhra Pradesh, Karnataka, Kerala, Madhya Pradesh, Rajasthan and Tamil Nadu as pro-employer states, and Gujarat, Maharashtra, Orissa, and West Bengal are pro-worker states. Taking this classification with one change made (Kerala taken as pro-employee rather than pro-employer; see Hasan et al 2007), employment functions have been estimated separately for pro-employer and pro-employee states. The comparison is presented in Table 7.
It may be seen from Table 7 that the estimated labour demand elasticity is relatively higher for pro-employer state, and there has been an increase in the labour demand elasticity in the post-1995 period among these states. By contrast, the labour demand elasticity is relatively lower in pro-employee states, and there has been no increase in the labour demand elasticity after the mid-1990s. This is broadly consistent with the empirical
seems reasonable to conclude that the observed increase
in the labour demand elasticity in the period after the mid-1990s is attributable in a significant measure to trade liberalisation. Also, other factors such as weakening of the bargaining power of trade unions and easing of labour regulations may have contributed to the hike in labour demand elasticity after the mid-1990s.
For the Attention of Subscribers and Subscription Agencies Outside India
It has come to our notice that a large number of subscriptions to the EPW from outside the country together with the subscription payments sent to supposed subscription agents in India have not been forwarded to us.
We wish to point out to subscribers and subscription agencies outside India that all foreign subscriptions, together with the appropriate remittances, must be forwarded to us and not to unauthorised third parties in India.
We take no responsibility whatsoever in respect of subscriptions not registered with us.
MANAGER
august 22, 2009 vol xliv no 34
SPECIAL ARTICLE
Notes
1 Hasan et al (2007) have used state-level two-digit industry data. This study makes use of three-digit industry data at the all-India level. The source of data on tariff rates also differs. However, there are some similarities. Both studies make use of Annual Survey of Industries data and thus are confined to the organised sector of Indian manufacturing (factories employing 10 or more workers with power or 20 or more workers without power). The period covered in the study of Hasan and associates, 1980-81 to 1997-98, matches with the period covered in the analysis presented in Section 2 of the paper.
2 The Economic & Political Weekly had brought out in 2002 a systematic, electronic database using ASI results for the period 1973-74 to 1997-98. For this database, concordance was worked out b etween 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 were prepared. This database has been used for the study. Though data are available for 152 three-digit industrial groups, the study covers only 137 groups. The remaining groups which have been excluded have the feature that the value of products is reported to be zero or very low in comparison with value added. Data on tariff rates and QRs on imports for three-digit industries are not available for years prior to 1980-81. Therefore, the analysis had to be confined to the period 1980-81 to 1997-98.
3 For aggregate manufacturing, the proportion of imports covered by QRs was about 90% in 1988-89.
4 The NIC (1998) codes of industries included in the study are 01 and 15 to 36. Industry coded 01 has been included because it covers cotton ginning.
5 The states are: Andhra Pradesh, Assam, Bihar, Gujarat, Haryana, Karanataka, Kerala, Madhya Pradesh, Maharashtra, Orissa, Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh and West Bengal.
6 It should be noted, however, that the numerical magnitude of the coefficients is low. A decrease in the tariff rate from 100% to 5% raises the labour demand elasticity by only about 0.03. It is argued later in the paper that the impact of trade liberalisation on labour demand elasticity probably b ecame stronger after the mid-1990s. It seems therefore that if the data set used for the estimates presented in Table 1 were extended to 2003-04, the coefficients would have been higher. It is u nfortunately not possible to extend the data set beyond 1997-98 because of the change in the i ndustrial classification from 1998-99.
7 The unweighted average rate of tariff (excluding countervailing duty and specific exemptions) on imports of manufactured products was 122% in 1986 and 129% in 1991, which declined to 40% in 1996 and 35% in 1998. There was further reduction in tariff rates from 2004 onwards, and the unweighted average tariff rate on manufactured imports came down to 12% by 2007 (Pursell et al 2007).
8 By the end of the 1980s, the manufactured products subject to QRs on imports accounted for about 90% of manufacturing value added, and this proportion declined to about 46% in May 1992 and further to about 36% by May 1995 (ibid). After 1995 and the completion of the Uruguay round, India’s remaining industrial QRs were contested at the World Trade Organisation (WTO) by other WTO members. India therefore had to phase out the remaining QRs which were not compatible with WTO rules; this process started in 1998 and finished in April 2001. By April 2007, the conventional QR coverage of manufacturing in the aggregate declined to only about half of 1% of manufacturing GDP (although two large industries namely sugar and urea remained protected) (ibid).
Economic & Political Weekly august 22, 2009
EPW9 The first sample is for the period 1973-74 to 1980-81, and the last one is for the period 1996-97 to 2003-04.
10 There has been a change in ASI coverage from 1998-99. The electricity generation and distribution industry, for instance, has been excluded from ASI from 1998-99, but is included in the data for previous years. The industry-wise panel is not affected (or not much affected) by this change in coverage. But, this introduces data incomparability in the state-wise panel. The aggregate ASI results for the states for the years 1998-99 onwards are not comparable with the results for earlier years. To take into account this data incomparability, a dummy variable for the period 1998-99 to 2003-04 has been introduced in the model.
11 There was a fall also in the share of labour income in value added. See Goldar and Aggarwal (2005).
12 The average rate of tariff on industrial goods in India was about 130% in 1991. The average implicit tariff on manufactured products (based on price comparison between domestically produced goods and imported goods) was about 30% to 40% (Pursell et al 2007). This signifies a good deal of tariff redundancy or “water in tariff”.
13 Between 1994 and 2003, mandays lost due to strikes fell from 6.7 to 3.2 million, whereas mandays lost due to lockouts increased from 14.3 to
27.1 million (source: data on strikes and lockouts provided in Economic Survey (government of I ndia) of different years).
14 Ahsan and Pagés (2007: 6) point out that while there have not been important changes in labour laws or in union formal presence or power, certain other factors may have increased flexibility in the Indian labour market. These factors, according to them, include weakening law enforcement, increasing recourse to temporary workers, increasing use of casual and contract labour and a shifting stand of the judiciary.
15 This is based on membership reported by unions submitting return (data taken from www.indiastat.com). Since only a small proportion of unions submit return (about 10-15%), the available data may not accurately portray the degree of unionisation in different industries.
16 This is based on a state-level survey recently undertaken. It covered eight major areas of labour law, identifying 50 specific subjects of possible reform many of which could be implemented by administrative procedure rather than through formal amendments to the laws. The eight areas covered in the index are the Industrial Disputes Act (IDA), Factories Act, State Shops and Commercial Establishments Acts, Contract Labour Act, the role of inspectors, the maintenance of registers, the filing of returns and union representation. Each state is given a score reflecting the extent of reforms undertaken; the maximum possible score is 50 and the average score across states is 21.
17 The states included in the first group are Uttar Pradesh, Andhra Pradesh, Gujarat, Rajasthan, Haryana, Punjab, Madhya Pradesh, Karanataka, and Orissa (in order of scores given). The second group includes Tamil Nadu, Maharashtra, Assam, Kerala, Bihar and West Bengal (in order of scores given).
18 The Besley-Burgess index has also been updated by Ahsan and Pagés (2007).
References
Ahsan, Ahmad and Carmen Pagés (2007): “Are All Labour Regulations Equal?, Assessing the Effects of Job Security, Labour Dispute and Contract Labour Laws in India”, World Bank Policy Research Working Paper No 4259, World Bank, June.
vol xliv no 34
Besley, Timothy and Robin Burgess (2004): “Can Labour Regulation Hinder Economic Performance? Evidence from India”, The Quarterly Journal of Economics, 119 (1): 91-134.
Bruno, Giovanni S F, Anna M Falzoni and Rodolfo Helg (2003): “Measuring the Effect of Globalisation on Labour Demand Elasticity: An Empirical Application to OECD Countries”, Flowenla Discussion Paper No 2, Hamburg Institute of International Economics, Hamburg.
Das, Deb K (2003): “Quantifying Trade Barriers: Has Protection Declined Substantially in Indian Manufacturing?”, Working Paper No 105 (New Delhi: Indian Council for Research on International E conomic Relations).
Dougherty, Sean (2008): “Labour Regulation and Employment Dynamics at the State Level in India”, OECD Economics Department Working Papers No 624, OECD publishing.
Fajnzylber, Pablo and William F Maloney (2005): “Labour Demand and Trade Reform in Latin America”, Journal of International Economics, 66(2): 423-46.
Goldar, Bishwanath (2005): “Impact on India of Tariff and Quantitative Restrictions under WTO”, Working Paper No 172 (New Delhi: Indian Council for Research on International Economic Relations).
Goldar, Bishwanath and Hasheem 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.
Goldar, Bishwanath and Suresh C Aggarwal (2005): “Trade Liberalisation and Price-cost Margin in Indian Industries”, Developing Economies, 43(3): 346-73.
Hamermesh, Daniel S (1993): Labour demand (Princeton, New Jersey: Princeton University Press).
Haouas, Ilham and Mahmoud Yagoubi (2004): “Trade Liberalisation and Labour-Demand Elasticities: Empirical Evidence from Tunisia”, IZA Discussion Paper No 1084 (Bonn: Institute for the Study of Labour).
Hasan, Rana, Devashish Mitra and K Ramaswamy (2007): “Trade Reforms, Labour Regulations and Labour-Demand Elasticities – Empirical Evidence from India”, Review of Economics and Statistics, 89(3): 466-81.
Krishna, Pravin, Devashish Mitra and Sajjid Chinoy (2001): “Trade Liberalisation and Labour-Demand Elasticities – Evidence from Turkey”, Journal of International Economics, 55(2): 391-409.
NCAER (2000): Protection in Indian Industry (New Delhi: National Council of Applied Economic Research).
Nouroz, Hasheem (2001): Protection in Indian Manufacturing: An Empirical Study (New Delhi: Mac-Millan India).
Papola, T S (general editor) (2008): Labour Regulation in Indian Industry, Vols 1-10 (New Delhi: Bookwell Publishers).
Purfield, Catriona (2006): “Mind the Gap – Is Economic Growth in India Leaving Some States Behind?”, IMF Working Paper No 06/103, International Monetary Fund, April.
Pursell, Garry, Nalin Kishor and Kanupriya Gupta (2007): “Manufacturing Protection in India since Independence”, ASARC Working Paper 2007/07, Australia South Asia Research Centre, Australian National University.
Rodrik, Dany (1997): “Has Globalisation Gone Too Far?” (Washington DC: Institute for International Economics).
Sharma, Alakh N (2006): “Flexibility, Employment and Labour Market Reforms in India”, Economic & Political Weekly, 27 May, pp 2078-85.
Slaughter, Mathew J (2001): “International Trade and Labour-Demand Elasticities”, Journal of International Economics, 54: 27-56.
