Recent times have witnessed a gradual shift towards hiring more temporary or contractual workers globally (Lass & Wooden, 2019; Kalleberg et al., 2000). We can expect these trends to accelerate in the times to come as firms increasingly hire temporary or contractual workers in lieu of directly hired permanent workers to cope with COVID-19 induced economic uncertainty. This phenomenon of the ‘fissuring of the workplace’ (Weil, 2014) has important welfare implications. This is because contractual and temporary workers typically have lower wages, reduced access to social security benefits and limited or no security of tenure when compared with regular full-time jobs (Kalleberg et al., 2000).
The proliferation of non-standard employer-employee work arrangements is particularly worrying in developing countries such as India, where a large share of the workforce is already employed in the informal sector and is engaged in low-paid and low-productivity work (Kapoor, 2020). This is best illustrated in the case of India's manufacturing sector, which has been characterised by its dualistic structure: the prevalence of a formal sector, which coexists with a large informal sector. Employment estimates from establishment surveys of the formal and informal manufacturing sector show that the share of the latter has declined over the time period 2000–01 and 2015–16.
Results from the Annual Survey of Industries (ASI), which covers formal plants registered under the Factories Act report that total employment in the formal manufacturing sector rose from 7.7 million in 2000–01 to 13.7 million in 2015–16.
The objective of this study is to examine what factors have driven contractualisation in the formal manufacturing sector. In particular, we seek to understand whether the lower wages paid to contract workers and the savings made on the expenditure of worker benefits incentivize plants to hire contract workers. Using plant-level data from the Annual Survey of Industries (ASI) for the time period from 2000–01 to 2013–14, we study the relationship between the workers’ wages, and the extent of contractualisation in a plant in the formal manufacturing sector.
Although there exists a vast literature that attributes the widespread use of contract labour to India's rigid employment protection legislations, it is noteworthy that labour regulations have not become more rigid over the time period when contract worker intensity has surged. The argument that it is inflexible labour regulations alone, which have incentivized plants to substitute directly hired workers with contract workers deserves closer scrutiny for several reasons. First, even states, which made amendments to their labour laws to make them more amenable to employers have witnessed a sharp increase in contract worker usage. Second, common wisdom suggests that rigid labour regulations impact labour-intensive industries more than capital-intensive industries. Thus, we should expect to see a greater increase in contract worker usage in the former. On the contrary, we find that it is large capital-intensive and not large labour-intensive industries, which have seen a larger increase in contract worker usage. These findings reiterate that plants are clearly induced to hire contract workers for reasons other than rigidities in labour regulations. Further, we find that the real wages of directly hired workers are on average about one and half times those of contract workers over the last decade. While this encourages plants to employ contract workers, the wage differential between contract and directly hired workers has indeed fallen over the last decade. Real wages of contract workers grew at 1.92% per annum, while those of directly hired workers remained stagnant.
The fact that the period over which the share of contract workers increased over time coincided with the years, which witnessed a faster growth rate of contract wages relative to those of directly hired workers is puzzling. A possible explanation is that the presence of contract workers in a plant's workforce enables the plant management to curb the bargaining power of the directly hired workers and depress their wages. Contract workers act as an alternative workforce, similar to
The structure of the paper is as follows. Section 2 presents a discussion of the existing literature on the contractualisation of India's workforce. Section 3 describes the data and key variables used in our analysis. Using this data, we present important stylized facts on contract worker usage in India in Section 4. In Section 5, we develop a model of firm-worker bargaining which attempts to explain why plants choose to maintain this duality in the workforce. Section 6 outlines the empirical analysis using ASI plant-level data and presents the results. The last section presents the conclusions.
The share of contract workers in total employment increased sharply from 15.5% in 2000–01 to 27.9% in 2015–16, while the share of directly hired workers fell from 61.2% to 50.4% in the same period (Figure 1). The increasing use of contract workers in India's formal manufacturing sector has been a subject of much attention (Sood et al., 2014; Das et al., 2015; Goldar & Suresh, 2017). The widespread use of contract labour has been ascribed to rigidities in India's employment protection legislation, in particular, Chapter VB of the Industrial Disputes Act (IDA). It is indeed largely because of the procedural difficulty of having to obtain prior government permission to lay off just one worker for plants covered by the IDA, that India's labour laws have been ranked stricter than those of all but two OECD countries.
India's labour regulations are largely perceived as being one of the strictest in the world in terms of employment protection legislation (OECD, 2007). However, these regulations cover less than 10% of the total workforce and large masses of the workforce engaged in the informal sector are left unprotected against any contingencies and arbitrary actions of employers. Importantly, these legislations are quite poorly enforced (Nagaraj, 2004). How stringent or relaxed labour laws are in practice thus depends on how well enforced they are in a particular context. For instance, Sapkal (2016) finds the effect of strict employment protection legislations and enforcement intensity on the incidence of temporary contract workers to be positive and statistically significant across Indian states.
Although much of the discussion on the impact of labour market regulations on the contractualisation of the workforce has focused on the IDA, there is another critical legislation pertaining specifically to contract workers that needs to be discussed. This is the Contract Labour (Regulation and Abolition) Act (CLA) of 1970 and applies to establishments in which twenty or more workers are employed or were employed on any day of the preceding twelve months as contract labour.
Recently, some state governments have amended this Act making it applicable to establishments employing 50 contract workers.
Another key incentive for plants to hire contract workers is the lower wages paid to them compared to directly hired workers and the savings made on the expenditure of their worker benefits. ASI data indicates that the real wages of directly hired workers have on average been about one and a half times those of contract workers over the last decade (Figure 2). Recent studies also highlight the role of the wage differential. Saha et al. (2013) find that increased import competition has led to the informalisation of industrial labour since the lower wages of informal workers and the savings made on the expenditure of worker benefits help in reducing costs and thus improving competitiveness.
The disparity in wages of directly hired and contract workers exists despite the CLA requiring wage parity between them.
The ASI dataset provides data on the total wage bill of the plant. Further it reports the disaggregated wage bill for each category of employee i.e., directly employed workers, contract workers and supervisory and managerial staff. The total wage bill in each category is divided by the number of employees in the respective category to arrive at the average wages. Directly employed workers receive several types of non-wage compensation in addition to their normal wages and salaries.
The presence of contract workers in a plant's workforce enables the management to curb the bargaining power of directly hired workers and thereby depress the wages of directly hired workers. This argument has been discussed in Saha et al. (2013) and Goldar (2016). Braun & Scheffel (2007) find an erosion in the bargaining power of low-skilled unionized workers with increased labour outsourcing in Germany. Thus, the use of contract labour reduces labour costs directly and indirectly (Goldar, 2016). While the direct effect comes from the lower wages paid to contract workers, the indirect effect comes from the fact that the presence of contract labour reduces the bargaining strength and wages of directly hired workers. However, this indirect effect is yet to be explored empirically in the Indian context.
Much of the above-mentioned literature (barring Chaurey, 2015 and Goldar, 2016) uses ASI's aggregate state-industry data. While they discuss the role of the lower wages paid to contract workers, they do not explicitly compute wages. They use the minimum wages of the state as a proxy for contract workers’ wages and estimate the effect of the former on the share of contract workers in a given industry at the state level. The use of the aggregate state-industry ASI data does not permit the computation of wages of contract and directly hired workers. A plant/establishment level analysis using micro-data is required to take this discussion forward, and that is precisely what we attempt to do in this paper.
This study uses ASI plant-level data for the period from 2000–01 to 2013–14 to obtain an unbalanced panel of registered manufacturing plants in India. The database covers factories registered under Sections 2m(i) and 2m(ii) of the Factories Act, 1948 i.e. those factories employing 10 or more workers using power; and those employing 20 or more workers without using power. The ASI data provides information on output, value-added, fixed capital, the value of plant and machinery, materials, fuel, total persons engaged, workers and wages and salaries to all employees (directly hired workers, contract workers, supervisory and managerial staff and unpaid family workers).
Our measure of capital in this study is the net value of plant and machinery at the end of the fiscal year.
Importantly, there are three different national industrial classifications (NIC) used in the ASI dataset for the time period under study- NIC-1998 (used between 1998–99 and 2003–04); NIC-2004 (used between 2004–05 and 2007–08) and NIC-2008 (2008–09 onwards). We undertake a concordance exercise across these different classifications to make the dataset comparable as per the NIC-2004 classification.
The data collected from the ASI are at current prices need to be deflated. An obvious candidate for this is the wholesale price index (WPI) series. However, we do not use the WPI directly because while ASI follows the NIC classification of industries, WPI is constructed with a view to capturing price movements based on the nature of commodities and final demand. Therefore, we construct a WPI for each of the industries in the analysis by approximating commodities based on the nature of economic activities and mapping NIC activities to WPI commodities.
Capital is deflated using the WPI created for NIC 29. We use 1993–94 as the base year to splice the data in order to develop a comprehensive and continuous series.
The raw data consist of about 746,073 observations over 14 years, with an average of about 53,290 plants surveyed each year. We only study observations corresponding to open plants and plants with positive values of output, plant and machinery and total persons engaged. We also drop the states and union territories that lack of information on employment legislation. The final sample consists of plant-year observations in 19 states.
It is important to note that the ASI data has a large number of outliers. To reduce their influence on our estimates, we winsorize the data (Dougherty et al, 2014). This procedure essentially involves top-coding and bottom-coding the 1% tails for each plant-level variable.
Next, we turn to the variables which are not obtained from the ASI database. The quantification of differences in labour market regulations (LMR) across states has been an extremely contentious subject. Much of the existing literature relies on the Besley-Burgess index (2004), which summarized state-level amendments to IDA between 1958 and 1992 and classified states’ labour regimes as having a pro-worker, pro-employer, or neutral stance. Despite the extensive use of this index in the literature, it has received much criticism. Bhattacharjea (2006, 2009a, 2009b) argues that the Besley-Burgess scoring system can erroneously classify a state as pro-employer or pro-worker with just one or two amendments to the IDA in the 50 years covered by the index. Nagaraj (2004) points out this index focuses only on IDA, when there are in fact several other labour laws, which impact industrial performance.
Given these and several other concerns, the measure of LMR used in this paper is from a study by Gupta et al. (2009). They have developed a composite measure of LMR across states by combining information from three key studies --Besley & Burgess (2004), Bhattacharjea (2009a), and OECD (2007). Using the score assigned to a state in each of these three studies (1 for flexible, 0 for neutral and −1 for inflexible), Gupta et al. (2009) create a composite classification of states’ labour market regimes by adopting a simple majority rule across the three studies. The final classification used is reported in Appendix (Table A3). As noted by Bhattacharjea (2017), the use of an index based on a mechanical reading of a single labour law may lead to flawed results. However, creating a new index is beyond the scope of this paper.
The empirical analysis conducted in this paper also requires data on the minimum wage rate and absenteeism rate of directly hired workers at the state level. We obtain this data from the reports on the working of the minimum wage law and the ASI's Volume II reports published by the Labour Bureau (Government of India) respectively. Additionally, we also use gross enrolment data for primary schooling at the state level. This is obtained from the Ministry of Human Resource Development, Government of India and the District Information System for Education.
Over the first decade of the 21st century, contractual workers have steadily substituted directly hired workers in the formal manufacturing sector (Figure A2 in the Appendix). Kapoor (2015) notes the rising trend of contractualisation and argues that a more disaggregated analysis is required as there are significant differences in trends of contractualisation across states and industries. In this paper, we take Kapoor's (2015) discussion forward to present a more comprehensive and rigorous analysis of variations in contract worker usage across different states, industries and plant sizes using factory-level data.
If it were only stringent labour regulations driving the contractualisation of labour, we would have only witnessed contractualisation of the workforce across those states which had inflexible labour regimes. However, Figure 3 indicates that this is not the case.
Figure A3 in the Appendix presents the share of contract workers across states in the organised manufacturing sector in 2000–01 and 2013–14.
There has been an increase in the usage of contract workers across industries (Table A2 in the Appendix). What stands out, however, is that the industries where contract worker intensity increased the most are in fact capital-intensive industries.
We define capital intensity as the ratio of real net value of plant and machinery to total workers. In order to classify industries as labour or capital-intensive or ambiguous, we calculate the labour intensity for all industries in the organised manufacturing sector for every year from 2000–01 to 2013–14. We classify an industry as labour-intensive if its capital intensity is below the median value for the manufacturing sector throughout the decade. NIC codes: 16, 17, 18, 19, 20, 28, 29, 33, 35 and 36 are labour-intensive industries; codes 21, 22, 23, 24, 25, 26, 27, 30, 32 and 34 are capital-intensive industries. The remaining industries are classified as ambiguous.
Next, we examine the intensity of contract worker usage across plants of different sizes. This is pertinent as several studies have attempted to look at the distribution of contract workers across different size bins. These studies argue that if plants were hiring contract workers to circumvent rigidities in labour regulations, we should observe the highest intensity of contract worker usage in the size bin with 50–99 workers i.e., the threshold below which Chapter VB of IDA kicks in (Industrial Disputes Act, 1947). Here too, we report the distribution of employment across different size bins but instead of constructing the size bins using the total number of workers, we do so using the total number of directly hired workers. This is because using the data for total workers would include contract workers, who are not borne on the rolls of the factory and therefore, not protected by IDA (Bhattacharjea, 2017).
We divide plants into the following three bins-- 0 to 49, 50–99, and 100 or more directly hired workers. We find that the share of contract workers in the total workforce has increased substantially across all size bins, not just the 50–99 bin (Figure 5). For instance, plants which have 100 or more directly employed workers, and are therefore above the IDA threshold, see a 10 percentage point increase in the share of contract workers. If circumventing labour regulations was the sole motive of hiring contract workers, such a significant increase in the share of contract workers would not have been witnessed in this bin
A caveat vis-à-vis construction of size bins merits clarification. The way in which employment is calculated for the purpose of determining whether or not it meets IDA thresholds does not correspond to employment levels reported in ASI (Bhattacharjea, 2017). Therefore, it is not possible to accurately divided plants into size classes that match the coverage of different chapters of IDA.
Additionally, we also examine plants in labour and capital-intensive industries in the three size bins separately (Figure A4 in the Appendix). The reliance of large plants in capital-intensive industries on contract workers is striking. This is in sharp contrast to large labour-intensive plants, where the share of contract workers is lower than all other subgroups. This is contrary to what we expect. Furthermore, we find that large capital-intensive plants expanded via contract workers in both states with flexible and inflexible labour regulations (Figure A5 in the Appendix). On the other hand, large labour-intensive plants have done so more in states with inflexible regulations as compared to states with flexible labour regulations. The share of contract workers in large labour-intensive plants in flexible states has in fact declined over the time period under study.
As discussed in Section 2, the wages of contract workers are significantly lower than those of directly hired workers, although the wage differential between the two has narrowed with wages of the former growing faster than those of the latter.
In labour surplus developing countries such as India there is no single labour market and there are forces beyond demand and supply that determine wages. Beyond the segmentations that occur due to the co-existence of the informal and formal sectors, there are complexities and segmentations owing to locational immobility of labour or capital, imperfect knowledge, skill levels beyond formal education, linguistic prowess, cost of living, reservation wages, and a host of social factors like gender, caste, kinship (Acharya, 2017). Additionally, Acharya (2017) notes that there are factors like capacity to pay. For example, in certain sectors older units pay much less compared to the newer ones since they do not make much profit and many actually sustain losses. On the other hand, in select cases employers retain workers at wages higher than those prevalent in the market, based on social contracts and personal trust (ibid). Therefore, wages could differ for the same jobs across establishments, locations, size and sectors amongst other factors. In line with the existing literature (Saha et al, 2013) in the formal manufacturing enterprises we assume that wages are determined by a bargaining process between directly fired workers and management of the firm.
Next, we examine the wage differential between contract and directly hired workers across different sized plants (Figure A6 in the Appendix). Here, the wage differential appears to have narrowed only in large plants. This decline has largely been driven by a fall in the real wages of directly hired workers in these plants. In small and medium-sized plants, on the other hand, the differential has remained almost constant over time. That large plants have witnessed substitution towards contract workers despite the fact that they are beyond the IDA threshold and that they have witnessed a decline in real wages of directly employed workers (and decline in wage differential between contract and directly employed workers) is striking and reflects the possibility of the existence of the bargaining channel mentioned in Section 2. In the following sections, we will examine this channel theoretically and empirically.
Two other noteworthy stylized facts emerge from Figure A6 in the Appendix. One, wages in medium-sized plants (both to contract and directly hired workers) are higher than wages in small-sized plants. And two, contract wages paid in medium-sized plants are roughly comparable to wages paid to directly hired workers in small plants.
The heterogeneity in contract workforce may result in varying outcomes from negotiations between firms and directly hired workers
For the theoretical framework, we use firms instead of plant, which is consistent with previous literature. A firm can be a multi-unit enterprise owning several plants.
We consider a representative firm which uses two types of inputs: labour,
Output elasticity of directly hired workers, contract workers and other inputs are given by
Further, we assume the objective of directly hired workers is to maximize their welfare by increasing their wages. Wages of directly hired workers are determined through negotiations between the firm and these workers. Wages paid to contract workers are bounded by minimum wages on the lower end (Basu et al., 2021). The resulting objective function,
Wages of directly hired workers are fixed through bargaining between them and the firm, where the ex-ante relative bargaining power of the firm is given by
Differentiating with respect to
Using these first order conditions, we obtain the following expression.
The sufficient condition for the ratio of contract workers to directly hired workers,
The main objective of this study is to understand what induces plants in the organised sector to hire contract workers. While we exploit the state-level variation in labour regulation, our analysis seeks to examine how the wage differential between contract and directly hired workers impacts the composition of the workforce, in particular the share of contract workers. Our basic premise is that besides labour market rigidities, two other factors drive the contractualisation of the workforce. First, contract workers receive lower wages helping plants reduce their wage bill and second, they help the plant's management diminish the bargaining power of directly hired workers. Using insights from the theoretical framework, the basic specification proposed to evaluate this is as follows:
It is calculated as a ratio of expenditure on energy inputs, storage and transportation to current value of gross output.
As discussed in the previous section, the wage differential between contract and directly hired workers, and the share of contract workers are determined jointly through an equilibrium mechanism and there exists an endogeneity problem
Intuitively, when contract workers become more expensive relative directly hired workers increase (i.e., the ratio of wages of contract to directly hired workers increases); we expect to see a fall in the share of contract workers. However, this decline in the share of contract workers in the plant's workforce results in an increase in the bargaining power of directly hired workers, resulting in an increase in their wages and consequently a decline in the ratio of the wages of contract to directly hired workers. These wages are determined by respective state governments and vary across states and over time. The man-days scheduled to work are arrived at by adding the man-days actually worked and the man-days lost on account of absence of the workers due to some reason or the other. Annual Survey of Industries 2011–12, Volume II, Labour Bureau, Ministry of Labour & Employment, Government of India. Page 10. Available Online
The estimates for the OLS regression with industry fixed effects associated with equation 7 are given in Table 1. As expected, the coefficient on the wage ratio is negative and significant for all samples (Columns 1–10). However, this specification does not consider the simultaneity between the wage ratio and the ratio of contract workers to total workers. The coefficient on LMR for the overall sample (Column 1) is negative and statistically significant suggesting that plants in states with more flexible labour regulations have lower shares of contract workers. The coefficient on the time trend is also positive indicating a positive trend in the hiring of contract workers overtime as suggested by the descriptive statistics. Column 2 and Column 3 report the results of the regression for capital and labour-intensive industries respectively. The coefficient on LMR is negative and statistically significant in both cases. Next, we break down plants by size i.e. small (those having less than 20 directly hired workers), medium (those having 20–99 directly employed workers) and large (those having greater than 100 directly employed workers). In the case of small plants (Column 4) and medium-sized plants (Column 5), we find LMR to be negative and statistically significant. In the case of large plants (Column 6), the sign of LMR remains the same but is statistically insignificant. We also disaggregate the sample of large plants into large labour-intensive and large capital-intensive industries. For both large capital-intensive plants and large labour-intensive plants, the coefficient on LMR is negative but insignificant. This is not surprising since owing to their size, the labour regulations lose their bite for large plants. We further classify the plants in the sample into different categories depending on the intensity of contract worker use. Specifically, first, we compute the 75th percentile for the share of contract workers in the total number of workers across plants for each year under study. We then classify these plants into the respective categories depending on which quartile the share of contract workers in the plants lies. The coefficient on LMR remains negative and significant for plants with a low share of contract workers (Column 9), while it is statistically insignificant for plants with large shares of contract workers (Column 10). Table A4 in the Appendix presents the two-way fixed effects (at plant-year-levels) results.
OLS regression results
ln(WC/WD) | −0.067*** (0.008) | −0.097*** (0.011) | −0.054*** (0.016) | −0.092*** (0.006) | −0.162*** (0.010) | −0.218*** (0.015) | −0.224*** (0.021) | −0.196*** (0.032) | −0.073*** (0.008) | 0.002** (0.001) |
time trend | 0.013*** (0.001) | 0.017*** (0.001) | 0.008*** (0.002) | 0.003*** (0.001) | 0.017*** (0.001) | 0.037*** (0.002) | 0.037*** (0.002) | 0.037*** (0.004) | 0.019*** (0.001) | 0.008*** (0.000) |
LMR-GHK | −0.024*** (0.006) | −0.047*** (0.008) | −0.024** (0.011) | −0.022*** (0.005) | −0.013** (0.006) | −0.018 (0.012) | −0.020 (0.016) | −0.029 (0.024) | −0.024*** (0.006) | 0.000 (0.001) |
Industry fixed effects | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
110,238 | 46,325 | 34,985 | 36,670 | 40,760 | 32,808 | 14,545 | 10,762 | 107,781 | 2,457 | |
0.051 | 0.047 | 0.073 | 0.032 | 0.057 | 0.130 | 0.084 | 0.120 | 0.058 | 0.324 |
The control variables include industry fixed-effects, plant-specific characteristics such as ln(age of plant in years), ln(plant's fuel intensity) and year-wise ln(state level gross enrolment ratio). For details of capital-and labour-intensive industries, see footnote 11. The dependent variable is ratio of contract workers to total workers.
Next, to circumvent the endogeneity issue, we estimate a regression equation where we use our absenteeism and minimum wage as proxies for the bargaining channel (Table 2). As before, the coefficient on LMR across all sub-samples and the overall sample (Columns 1–10) is negative and statistically significant suggesting that plants in states with more flexible labour regulations have lower shares of contract workers. Importantly, for the overall sample (Column 1), we find the coefficient on the minimum wage to be positive and statistically significant. Typically, one would expect the sign on this coefficient to be negative because the share of contract workers in the plant's workforce declines as contract workers become relatively more expensive as a result of the increase in the minimum wages, which is the lower bound for contract worker wages. However, following the literature (Saha et al., 2013), we consider minimum wages to be a proxy for directly hired workers’ bargaining power in the literature. This is because the states mandate that contract workers be paid at least the minimum wage. Consequently, as the minimum wages increase, the contract workers become more likely to be substituted by directly hired workers (ibid). Therefore, in this scenario, the positive sign on the minimum wage suggests that the bargaining effect overwhelms the price effect. That is, despite the rising relative wages of contract workers, plants continue to hire them as they help the management suppress the bargaining power of directly hired workers. The coefficient of absenteeism is statistically insignificant.
Proxy variable regression results
time trend | 0.009*** (0.001) | 0.012*** (0.001) | 0.007*** (0.002) | 0.003*** (0.001) | 0.015*** (0.001) | 0.016*** (0.001) | 0.015*** (0.002) | 0.015*** (0.003) | 0.028*** (0.001) | 0.003*** (0.000) |
LMR-GHK | −0.047*** (0.005) | −0.059*** (0.007) | −0.060*** (0.010) | −0.013*** (0.005) | −0.012* (0.007) | −0.082*** (0.010) | −0.061*** (0.012) | −0.164*** (0.020) | −0.043*** (0.006) | −0.001*** (0.000) |
ln(absenteeism) | −0.005 (0.010) | −0.003 (0.012) | 0.112*** (0.023) | 0.074*** (0.012) | 0.068*** (0.015) | −0.041*** (0.016) | −0.020 (0.017) | 0.050 (0.039) | −0.013 (0.010) | 0.001* (0.001) |
ln(min wage) | 0.109*** (0.011) | 0.112*** (0.014) | 0.013 (0.024) | 0.029*** (0.011) | 0.057*** (0.015) | 0.102*** (0.020) | 0.135*** (0.022) | −0.002 (0.045) | 0.125*** (0.011) | −0.003*** (0.001) |
Industry fixed effects | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
N | 134,436 | 60,577 | 41,279 | 36,670 | 40,760 | 57,006 | 28,797 | 17,056 | 120,321 | 14,115 |
0.077 | 0.102 | 0.072 | 0.023 | 0.039 | 0.170 | 0.207 | 0.133 | 0.086 | 0.153 |
The control variables include industry fixed-effects, plant-specific characteristics such as ln(age of plant in years), ln(plant's fuel intensity) and year-wise ln(state level gross enrolment ratio). For details of capital-and labour-intensive industries, see footnote 11. The dependent variable is ratio of contract workers to total workers.
In capital-intensive industries (Column 2), we find the coefficient on the minimum wages to be positive and statistically significant, suggesting that the bargaining effect overwhelms the price effect. For labour-intensive industries reported in Column 3, we find the coefficient on absenteeism to be positive and statistically significant, while the coefficient on the minimum wages is positive and statistically insignificant. This suggests that for labour-intensive industries, instead of increases in minimum wages, the inherent bargaining power of directly hired workers is driving the substitution of the directly hired worker by contract workers.
For small and medium-sized plants (Columns 4 and 5), we find the coefficients on absenteeism and minimum wage to be positive and statistically significant. In the case of large plants (Column 6), the sign and significance of minimum wage remain the same. The coefficient on absenteeism is negative and significant. However, the magnitude of the coefficient on absenteeism is smaller compared to that on the minimum wage. This suggests that for large plants, the benefit of hiring contract workers due to the effect they have on suppressing directly hired workers outweighs the costs arising from the relative increase in their wages over time. In addition, for large capital-intensive plants, the coefficient on minimum wage was positive and significant, but for large labour-intensive plants, it was insignificant. That the effect of the bargaining channel playing out in large capital and not labour-intensive plants may be a consequence of the fact that the former has a greater incentive to cut costs as compared to the latter.
Common wisdom suggests that in plants with a high share of contract workers, the bargaining channel should not be present. Since such plants already have a substantially large share of contract workers that help suppress the bargaining power of directly hired workers, they have little or no incentive to hire more contract workers simply for this particular purpose. They would hire more contract workers only if they are relatively cheaper. On the other hand, plants, which have a smaller share of contract workers, would benefit from hiring contract workers even if they become relatively more expensive as this would help suppress the bargaining power of directly hired workers. This hypothesis is confirmed as we find that for plants with a low share of contract workers — below the 75th percentile (Column 9), the coefficient on the minimum wage is positive and significant, while in the case of plants with a high share of contract workers — above the 75th percentile (Column 10), the coefficient is negative and insignificant. The coefficient on absenteeism is statistically insignificant for plants with a low share of contract workers, while it is positive and significant for the plants with a high share of contract workers. However, the magnitude of the coefficient on absenteeism is smaller than that of the minimum wage for the high-share contract worker plants (Column 10). This indicates that the bargaining channels outweigh the price effects in the plants with a low share of contract workers, while it is the reverse for the plants with a high share of contract workers. Table A5 in the Appendix presents the two-way (plant and year) fixed effects results.
As a supplementary analysis, we also estimate the instrument variable estimates using absenteeism and minimum wage as instruments for the wage ratio between contract and directly hired workers. Table A6 in the Appendix reports the IV estimates.
Much of the existing literature attributes the increasing use of contract workers across the organised manufacturing sector to rigid labour regulations. Our analysis in this paper suggests that this is not the sole factor responsible for these trends. Not only are contract workers cheaper than directly hired workers, but their presence in the plant's workforce helps diminish the bargaining power of the latter. The non-negligible growth in wages of directly employed workers and the slow growth in wages of contract workers is reflective of this.
The decline in the bargaining power of workers is not unique to the Indian manufacturing industry. Empirical literature from other countries corroborates this global trend as well (Dekker & Koster, 2018). For instance, evidence from the United States also shows that the falling labour share of income, rising income inequality and slow wage growth are not only attributable to the globalisation, technological changes, and rising monopoly power but more compellingly to the broad-based decline in worker power (Summers & Stansbury, 2020). The extent to which firms’ managements resort to practices such as the hiring of contract workers to suppress the bargaining power of workers, as discussed in this paper, is noteworthy. The implications of this phenomenon not just for income inequality but also security of tenure is deeply worrying in a post COVID-19 world which has witnessed widening disparities between workers in regular stable employment arrangements and those in precarious work arrangements.