1 Introduction

Although Kerala’s net migration rate is considerably higher than other states of India, of late the state started attracting large number of migrants from different parts of India. There is no proper database on the in-migrant workers in Kerala. Gulati Institute of Finance and Taxation (2013) estimated that there were more than 25 lakh in-migrant workers in Kerala in 2012 who were employed mostly in informal sectors. Interestingly, at the same time, there were as much as 21 lakhs out-migrants from Kerala to various parts of the world in 2014 (Rajan 2016). Kerala is reported to be facing a scarcity of manual labors, which might explain the reason for the in-migration to the state. There are several other pull factors that attract migrant workers to Kerala. They include availability of work and better wages and conditions of work and living. Lack of employment opportunities and lower wages at the place of origin have been the major push factors (Narayana et al. 2013).

The history of migration of workers to Kerala from other states can be traced back to 1980s. The migrant workers during this period were mostly from the states of Tamil Nadu, Karnataka, and Andhra Pradesh. However, this trend has changed since the last 10 years and one of the most striking features of the labour market in Kerala in the past decade was the increasing presence of migrant workers from the states like West Bengal, Uttar Pradesh, Odissa, Gujarat, Assam, Rajasthan, and from the neighboring country of Nepal. They are usually engaged in manual labor such as earth work, road works, unskilled works in construction, and in services like ironing clothes, and as cooks and waiters in small hotels and eateries, brick kilns, jewellery work, bakeries, vegetable shops, and in manufacturing industries (Kumar 2011).

The in-migrants who come to Kerala are socially, culturally and linguistically a heterogeneous group. Prasad (2016) noted that the in-migrant workers in Kerala are linguistically categorised as Hindi speaking people. But they are not a homogeneous group linguistically and culturally, and among them, there are people who speak Assamese, Bangla, Oriya, Bhojpuri, Nepali, Gujarati, and several other tribal languages. Also there are people who do not know Hindi language. Normally migrants use their local language to converse among themselves. Since Hindi is not known to all natives, it is often difficult to have interactions between the migrants and natives. Only those migrants who get a chance to mingle with the native people learn the local language of Malayalam. The image and label of in-migrants as “outsiders” who do not share the language and culture of the natives have led to their discrimination in various spheres of their life and work in Kerala. Studies have highlighted the discriminatory practices of natives and state authorities including the police against the in-migrants. For instance, Kumar and Pramod (2016) found that the native people and the police set eyes on them with suspicion. Non-proficiency of the regional language also sometimes makes it difficult for them to prove their innocence in front of police and legal system, if booked for any alleged violation of laws. The language barrier further leads to stress and strain as they are not able to communicate properly with health providers, merchants, bankers, and policemen. These make them feel alienated as well (Prasad 2016). Also, studies showed that they tend to ignore the living conditions, housing facilities, and sometimes even health to earn and save as much as money as possible during their stay in Kerala (Prasad 2016; Narayana et al. 2013). This attitude of the workers is often exploited by the contractors, employers and even house owners and it increases their precarity and informality at work and the hardship of living. There are also several factors that influence the work and wage of the in-migrant workers, namely nature of work, gender, and working hours in a day, skill level and willingness to do overtime work. Besides these, variables like age, marital status, education, caste, geography, working days in a week and bargaining power have significant influence on the wage rate and wage differences with regard to in-migrant laborers. For instance, studies have shown that the migrant workers concentrate largely in the lower-income labor market with higher risks of exposure to unsafe working condition (Chatterjee 2006).

While higher wage rate in Kerala is one of the major pull factors of migration, it is not clear whether in-migrants in Kerala receive wages at par with the natives or above the minimum wages stipulated by the state government for various sectors. Available studies on wage differentials in Kerala focused on specific sectors, especially the construction sectors (Beju and Shamna 2019, Parida et al. 2020). Given this background, this paper attempts to examine whether there are differentials in the wages of in-migrant workers as compared to the natives. If so, what could this differential be attributed to? To be more specific, whether in-migrant workers in Kerala face wage discrimination based on the markers of region and whether there are further differences in wages within in-migrants based on their skill level, language, religion, ethnicity and caste.

2 Methods

The paper is based on a primary study conducted in Kozhikode district which is situated in the Northern part of Kerala. The selection of respondents was based on quota sampling method as there was no systematic information on in-migrant workers available. Data were collected from 300 in-migrant workers those who worked in primary, secondary, and tertiary occupational sectors using a semi-structured interview schedule from different sites in the district. Care was given to include respondents from all three occupational sectors. Primary occupational sector covered in this study includes fisheries and quarrying and allied sectors. Secondary occupational sector includes manufacturing of footwear, iron and steel, plastic, ice cubes, and gold works. Tertiary occupational sector includes mechanics, electricians and those who worked in eateries, restaurants, hotels, etc.

2.1 The Decomposition of Wage Differentials

In order to estimate the wage discrimination between in-migrant workers, Blinder (1973) and Oaxaca (1973) decomposition method is used in our study. Blinder (1973) has explained the wage discrimination among white and black in USA, the important factors were education, work experience, and family background such as parents' income and educational attainment, and the place of birth. The whites have upper hand in these factors, on the other hand, Blacks generally gain less wage by having an educated father, lose more by coming from a poor family, and gain less by being born in an urban area (Blinder 1973). This method is used widely to understand the wage differentials and its linkages to factors such as discrimination in various contexts across the world. A study conducted in Spain on gender wage discrimination using the model and found a male–female average wage differential (Gardeazabal and Ugidos 2003). The same result found in Bangladesh as well that there exists a large wage gap between male and female, and it is more in urban areas than rural areas (Ahmed and Maitra 2010). The gender-based wage differential also explained with the decomposition method, an empirical study on gender wage discrimination in Indian urban labor market with the unit level data sets of 50th, 61st, and 68th NSSO employment and unemployment survey, reveals that there is a significant average gap between male and female workers in urban labor market during the study period, however, the wage gap is declining over the period of time (Padhi et al. 2019). In addition, Agrawal (2014) and Sengupta and Das (2014) also found the wage differential between male and female in both rural and urban areas. The wage discrimination is high among rural workers in both public and private sectors as compare with the discrimination among urban workers (Das 2012). Singh and Pattanaik (2020) and Duraisamy and Duraisamy (2017) by using NSSO rounds and Periodic Labor Force Survey (PLFS), they explained the wage discrimination between ST/SC and Gen/OBC women workers and it is significantly increased last decade (1999–2000 and 2017–2018) and the less paid workers face higher discrimination than the high paid workers. Madheswaran and Attewell (2007) illustrated the discrimination against SC/ST workers is higher in private sector than in public sector in urban labor market in India. The contribution of endowment difference to raw wage gap is higher in public sector.

There are studies on wage discrimination among migrant workers from various parts of the world. A study conducted in China found that the wage gap between migrant and urban workers. The urban workers make more than migrant workers, which suggests wage discrimination (Lu and Song 2006). The wage rate and compensation rate differentials between migrant and urban workers are quite large in China. Urban workers' hourly wage is almost double than the migrants, while their hourly compensation is three times more (Lee 2012). However, very few studies that have used decomposition method in order to decompose the wage gap into endowment component and discrimination component are found in the Indian context. Khan (2016) analyzed the wage differential between migrant and non-migrant workers in India’ labor market with NSSO 55th (1999–2000) and 64th (2007–2008) rounds and found that there was wage discrimination between migrant and non-migrant workers. Further, the study found that the female, SC and ST migrant workers faced more wage disadvantages. However, Parida (2020) studied on migrant and non-migrant wage differential using the data from NSSO 55th and 64th rounds of migration specific surveys and the results are contradictory to other wage differential studies. For instance, the study found that the regular salaried and casual migrant workers earned more than non-migrants in all occupational sectors except agriculture and fishery. These earning differences were the highest among highly skilled workers and vice versa. By using Oaxaca and Ransom decomposition method, Thorat et al. (2021) analyzed the wage discrimination in public and private sector in India and found a wage gap between SC and Higher Caste (HC) and that the incidence of discrimination is more in private sector. A study conducted in Kerala found that there is wage differential existing among migrant and local workers in construction sector and illustrated that the migrant workers receive 31 percent less wages than the local workers for the same work (Beju and Shamna 2019). Another study also found that wage differentials exist between migrant and non-migrant workers in the construction sector in Kerala and the local workers earn 1.5 times more than the counterpart migrant workers (Parida et al. 2020).

This paper used the decomposition method to analyze the wage gap between the Bengali and non-Bengali migrant workers in Kerala. Perhaps, this will be the first study to attempt the discrimination between the interstate migrant workers in Kerala. This technique decomposes the gross amount of wage differential into endowment difference and discrimination coefficient in the labor market. Theory of Discrimination is originated from Becker’s (1971) seminal work. According to Becker, the gross wage differential between male and female is considered as discrimination. In our study, we decompose the gross wage differential between Bengali and non-Bengali of in-migrant workers. Hence, the gross wage (G) can be written as

$$G = \frac{{W_{{\text{B}}} - W_{{{\text{nB}}}} }}{{W_{{{\text{nB}}}} }} = \frac{{W_{{\text{B}}} }}{{W_{{{\text{nB}}}} }} - \frac{{W_{{{\text{nB}}}} }}{{W_{{{\text{nB}}}} }} = \frac{{W_{{\text{B}}} }}{{W_{{{\text{nB}}}} }} - 1$$
(1)

This (1) can be rewritten as:

$$G + 1 = \frac{{W_{{\text{B}}} }}{{W_{{{\text{nB}}}} }}$$
(2)

Taking natural log of Eq. 2 yields the results as follows:

$${\text{Ln}}\left( {G + 1} \right) = {\text{Ln}}\left( {\frac{{W_{{\text{B}}} }}{{W_{{{\text{nB}}}} }}} \right) = {\text{Ln}}W_{{\text{B}}} - {\text{Ln}}W_{{{\text{nB}}}}$$
(3)

Further, Eq. (3) is subjected into Mincer’s (1974) semi-logarithmic augmented wage function. Hence, the separated wage equations for Bengali’s and non-Bengali’s are as follows.

$$\ln \overline{W}_{{\text{B}}} = \sum \hat{\beta }_{{\text{B}}} \overline{X}_{{\text{B}}} + \varepsilon_{{\text{B}}} \quad \left( {{\text{Bengali's wage equation}}} \right)$$
(4)
$$\ln \overline{W}_{{{\text{NB}}}} = \sum \hat{\beta }_{{{\text{NB}}}} \overline{X}_{{{\text{NB}}}} + \varepsilon_{{{\text{NB}}}} \quad \left( {\text{Non-Bengali's wage equation}} \right)$$
(5)

Both Eqs. (4) and (5) are estimated via ordinary least squares (OLS). Here \(\ln \overline{W}\) denotes the geometric mean of average amount of wage, \(\overline{X}\) is the vector of mean values of the regressors, \(\hat{\beta }\) is the vector of estimated coefficients, and ɛ is the error term. Within this framework, the gross wage differential in the logarithmic term is given by

$${\text{Ln}}\left( {G + 1} \right) = {\text{Ln}}\left( {\frac{{\overline{W}_{{\text{B}}} }}{{\overline{W}_{{{\text{NB}}}} }}} \right) = {\text{Ln }}\overline{W}_{{\text{B}}} - {\text{Ln}} \overline{W}_{{{\text{NB}}}} = \sum \hat{\beta }_{{\text{B}}} \overline{X}_{{\text{B}}} - \sum \hat{\beta }_{{{\text{NB}}}} \overline{X}_{{{\text{NB}}}}$$
(6)

Assuming that the difference in the coefficients of the two average amounts of wage functions is considered as a priori evidence of discrimination, then this can be written as:

$$\Delta \hat{\beta } = \hat{\beta }_{{\text{B}}} - \hat{\beta }_{{{\text{NB}}}}$$
(7)

Suppose for a given endowment, if “NB” have an opportunity to get wage like “B,” then the wage outcome would be as follows:

$${\text{Ln}}\overline{W}_{{{\text{NB}}}} = \sum \hat{\beta }_{{\text{B}}} \overline{X}_{{{\text{NB}}}}$$
(8)

Now, the a priori evidence in absence of discrimination is as follows:

$$\hat{\beta }_{{{\text{NB}}}} = \hat{\beta }_{{\text{B}}} - \Delta \hat{\beta }$$
(9)

Substituting Eq. 9 in Eq. 6

$${\text{Ln}}\left( {G + 1} \right) = Ln \overline{W}_{{\text{B}}} - {\text{Ln}} \overline{W}_{{{\text{NB}}}} = \left[ {\sum \hat{\beta }_{{\text{B}}} \left( {\overline{X}_{{\text{B}}} - \overline{X}_{{{\text{NB}}}} } \right)} \right] + \left[ {\sum \overline{X}_{{{\text{NB}}}} \left( {\hat{\beta }_{{\text{B}}} - \hat{\beta }_{{{\text{NB}}}} } \right)} \right]$$
(10)

In Eq. (10), the first term on the right hand side can be interpreted as endowment differences and the second term has been regarded as the discrimination coefficients.

2.2 Limitations

The study could not present a disaggregated picture of wage differentials across individual state of origin and social groups, particularly for scheduled castes and scheduled tribes since the sample size turned out to be very thin when disaggregated. Categories of SC and ST are clubbed due to this limitation. In some sectors like plastic manufacturing, there were only in-migrant workers. Hence, comparison of wages of native and in-migrant workers was not possible. Further, the paper could not verify minimum wages across all occupational sectors and hence, comparison of workers’ wages with minimum wages was limited to skill levels.

3 Results

The sample contained in-migrant workers from the different states of India. According to District Labor Statistics, nearly 50 percent of total migrant workers in Kozhikode are from West Bengal. Out of the 300 workers interviewed, most of the workers were from West Bengal, followed by Bihar, Assam, Jharkhand, UP, Odisha, and Gujarat. This is in tune with the general trend in the district. They worked in different occupational sectors in the district, but the migrants from Uttar Pradesh were found only in manufacturing sector. They are mostly from rural areas and a very few hail from the urban areas (see Table 1). More number of the respondents were Hindus. There was an equal share of Hindu and Muslim in-migrants in the sample from the states of WB and Bihar. All the workers from Odisha and Gujarat were Hindus and a large number of migrants from Jharkhand and Assam were also Hindus. Among social groups other backward class (OBC) had the highest share, primarily due to the higher representation of Muslims in the sample, which is followed by schedule caste (SC) and schedule tribe (ST). While a large number of OBC workers were from Bihar, ST migrants were from Jharkhand. All the respondents in the sample from Gujarat belonged to general category (read forward castes). The highest share of SC workers was found among migrants from Uttar Pradesh. Majority of the in-migrant workers were in the age-group of 21 to 30 years, and a few were in the age-group of 41 and above. Most of the respondents migrated without family members.

Table 1 Distribution of respondents by their demographic and socioeconomic characteristics (N = 300)
Full size table

As it is true for any migration there were several push and pull factors that determined the migration to Kerala. Push factors were unemployment, less availability of work, irregular work, poverty and family pressure and lower wages at the place of origin (see Fig. 1). Pull factors included better wages, availability of work, better conditions of work, and favorable climatic conditions in Kerala.

Fig. 1
figure 1

Source: Primary Data, 2018

Push and Pull factors of migration.

Full size image

3.1 Occupation and Nature of Employment

The respondents mostly were casual workers who worked in the lower strata of primary, secondary and tertiary occupational sectors. Majority of them were full time migrants who worked throughout the year (Table 2). Their duration of stay ranged from less than a year to more than 10 years. Only a small percentage of workers in the sample worked in Kerala for less than a year. Nearly 22% of them had worked for more than ten years. The sample had nearly 11% of seasonal migrants, and they all were from the state of West Bengal. While migrants who had specific skill sets tended to find the same kind of work, what they had been doing at the place of origin unskilled and semi-skilled workers took up whatever works that were available for them in any sector at the place of destination. Nearly one-third of the respondents worked in manufacturing sector who were mostly from the states of Uttar Pradesh, Bihar, Jharkhand, and Assam. Other major sectors where respondents in the sample worked were fisheries, construction and quarrying and allied sectors. A large number of migrants from WB were in construction sector followed by fishing and allied sector. Migrants from Odisha in the sample worked only in Fisheries and quarrying sectors.

Table 2 Nature of migration and work across the State of origin
Full size table

3.2 Average Wage Differential of Migrant Workers

We have attempted to understand the wage differences of migrants within the group of migrants and as compared to the natives. It should be mentioned that we have not controlled the years of experience, which will definitely have a minor impact on the wages. What we present here is the wage differential for same category of occupation within and across sectors. Wage differentials were estimated for skilled categories of work across education. Table 3 shows the average wage differential by education and skill level. Skill level was a major factor that determined the wages of workers. For instance, skilled migrant workers who were goldsmiths, operator driver of earth movers, masons, footwear designers, mechanics, cooks, and electricians received wages which were much above the stipulated minimum wages fixed by the state for the skilled categories of work.Footnote 1 Among semi-skilled workers, fishermen received the highest wages [which was also higher than the skilled categories], mainly because of the scarcity of workers in the sector due to the dangerous conditions of work. Their wages were notably more than the minimum wages stipulated for semi-skilled occupation in Kerala, which is Rs. 425 as of 2017. However, workers in the semi-skilled categories such as steel polishing and aluminum moulding received wages less than that of the stipulated minimum wages. Interestingly, all unskilled workers in the sample received wages more than the stipulated minimum wages, which was Rs. 360/-. Also, their mean wages were more than that of some of the semi-skilled workers. However, it should be noted that their work was highly precarious and irregular in nature as compared to the semi-skilled and unskilled workers. Interestingly, level of education was not a major factor in determining wages in all these sectors. This is primarily due to the fact that these workers are in the lower rung of the job hierarchy in the sector. For instance, we have not found any migrant worker in the category of supervisor or above where educational entitlements may be translated in to remunerative wages.

Table 3 Average daily Wage Differential for in-migrant workers by Education and Skill Level (in Rs.)
Full size table

Further, we have examined the wage differentials between migrants and natives (Table 4). It is clear from the table that both the migrants and the natives receive notably higher wages than the minimum wages stipulated for respective skill category. However, we have found significant wage differentials between the wages of the migrants and the natives. Our survey showed that while the native workers got an average wage of Rs. 1000 for skilled jobs it was around Rs. 700 for migrant workers. The average daily wages of native mechanics and electricians was Rs. 900, whereas it was Rs. 583 for the migrant mechanics and Rs. 500 for migrant electrician. Similarly, the average daily wages of native driver, goldsmith, mason, and footwear designer was Rs. 1000 and a restaurant cook was Rs.1200, whereas it was Rs. 560 for a migrant cook, which is less than half of what the native workers earn although it was more than the minimum wage in Kerala. This trend was found true for semi-skilled and unskilled categories of work as well. For instance, native fishermen received notably higher wages than their migrant counterparts. Similarly, all migrant semi-skilled footwear makers received wages more than the minimum wages but less than the wages of the natives. While the native workers in the unskilled category construction work (helpers) received Rs. 750 per day, their migrant counterpart received only Rs.530. Such differences in wages between natives and migrants were found in all sub-categories of unskilled work.

Table 4 Average daily wage differential of Native workers and in-migrant workers by same category of occupation and skill level (In Rs.)
Full size table

3.3 Wage Differentials among In-Migrant Workers

We found from the primary survey that migrants from West Bengal work in majority of the occupational sectors, such as fisheries, quarry and allied, construction, manufacturing and service sectors. However, the other state migrants (non-Bengali workers) are concentrated only in specific occupations with limited population. Moreover, migrant workers from West Bengal enjoy a social, economic and cultural upper hand because of their long stay in Kerala, knowledge of local language and the social capital, and network they have with natives, employers and contractors as compared to other migrants from states. Hence, it is important to know whether non-Bengali migrants are facing discrimination as compared to the majority group (Bengali), since they have an upper hand in the occupational sectors. Therefore, we attempted to examine the wage gap between migrants from West Bengal and other states. The econometric analysis confirms that wage differentials exist between in-migrant workers of same skill levels especially the major difference is found between Bengali and non-Bengali workers (Table 5).

Table 5 Determinants of wage among migrant workers in Kerala
Full size table

The results show a direct correlation between age of the respondents and their wages. For instance, if the age of the migrant worker increases by 1 percent his/her wage increases by 0.29 percent in pooled migrant workers, 0.1 percent in Bengali workers and 0.16 percent in non-Bengali workers. It could be attributed to the experience that a worker gains with age. This was true for all workers. As it is already discussed, decomposition analysis also showed that educational qualification was not a major determinant of wages for migrant workers since the respondents in Kerala were in lower rung of the occupational hierarchy. Interestingly, knowledge of local language was an important determinant of wages. The pooled sample result showed the migrant workers who knew the local language earned 9.2 percent more wages than those who did not speak the local language. Among migrants higher wages attributable to knowledge of local language was more for migrants from WB than other states. Knowledge of local language might have helped them to negotiate for better wages. Muslim migrants earned wages less than that of the Hindu migrants in the pooled sample. However, among migrants from WB, Muslims wages more than their Hindu counterparts. Across social groups forward caste migrants received higher wages in the pooled sample followed by OBCs and SCs and STs. However, among migrants from WB, those who belonged to SC/STs earned more than OBCs and forward caste groups. It is because of the majority of Bengali migrants from these social groups worked in fisheries sector which was highly paid job. In precise, discrimination based on markers of religious and caste prejudices was not experienced by migrant workers from WB in Kerala. This, however, was not validated true in the case of migrants from other states.

The majority of the respondents were unskilled workers. In pooled sample, the semi-skilled workers earned 22 percent more than unskilled and 13 percent more than skilled workers for 1 percent increase in these variables. This is due to the prevalence of higher wages in the fishing sector, which is listed in the primary sector of occupation. It is important to note that migrants from WB in the skilled jobs received less wages as compared to their counterparts from other states. However, migrants from WB in semi-skilled sectors received higher wages than those who migrated from other states.

Table 6 clearly shows that discrimination exists in wages between the workers from WB and other states. Discrimination is high against non-Bengali migrants (59.1%). In short, although all migrants face wage discrimination, non-Bengali migrants faces the most within them. Table 6 also shows that the Bengali migrants had an advantage as their endowments difference is 40.1 percent over non-Bengalis.

Table 6 Oaxaca—Blinder decomposition results (as %)
Full size table

4 Conclusion

The study illustrated the wage differentials existing at two levels. First is the wage difference between natives and in-migrants irrespective of their skill levels. Second is the wage difference between migrant workers from West Bengal and other states that was established by using the decomposition method. The decomposition model used in the study also suggests that the wage differentials among migrants workers can be attributed to discrimination that face by being not sufficiently integrated to the Kerala society. The empirical results shed light on the need for state interventions to level the wages of migrants and natives on the one hand and formulating programs for integration of guest workers (as the government of Kerala refer to migrant workers) who still live on the margins of Kerala society on the other. The most important step among others is to increase their capacity for wage negotiation by educating them about the rights of workers and the minimum wages stipulated for each sector. Our results showed that migrants from the state of WB who integrated relatively well with the society could negotiate better for wages. The state government introduced educational development program to in-migrant workers to educate Malayalam in some districts or places. If the government is extending the program across Kerala, more number of migrant workers will get benefit and they will get more wage negotiation power in local language. Control of labor contractors who take a share of workers’ wages is another issue that needs attention.