Abstract
Notwithstanding the large decline in poverty in Bihar and UP during 2004–2005 and 2011–2012, incidence of poverty among Dalits continues to remain high in comparison to OBCs and upper castes. The article examining the nature of poverty among Dalits in Bihar and UP situates the high incidence of poverty among them in structural and institutional factors. More than 90% of Dalits living in the two states are either landless or own marginal land holdings and have an abysmal representation in the ownership of private enterprises. As a result, the majority of Dalit households eke out a livelihood through casual wage work, where many of them earn less than the stipulated minimum wage. The article, using the Oaxaca–Blinder model, re-establishes the fact that Dalits and other marginalised social groups face discrimination in the wage labour market. About 85% of Dalit workers in Bihar and 90% in UP are either working poor or ‘vulnerable to poor’, and even a small economic shock can push them below the poverty line.
Introduction
The period from 2004–2005 to 2011–2012 in India is marked by a significant decline in the incidence of poverty, from 37% to 21.9% at the all-India level. During this period, two of the poorest Indian states, Bihar and Uttar Pradesh, also registered a sharp reduction in income poverty: from 54% to 34% in Bihar and from 41% to 29% in UP (Reserve Bank of India, 2021). The number of people who came out of poverty during this period in the two states was over 26.47 million. The scheduled castes (SCs) or Dalits, who comprise around 20% of the population in each of the two states and are known to have a high incidence of chronic poverty, also witnessed a sharp fall in poverty. Poverty among them declined from 77.3% to 51% in Bihar and from 55.2% to 40.9% in UP. In fact, in UP, the rate of reduction in poverty remained the highest among SCs during 1998–1999 and 2004–2005 (Ojha, 2007), as well as during 2004–2005 and 2011–2012. But notwithstanding this decline, poverty among them remained high. Notably, at the all-India level, poverty among Dalits declined from 50.9% to 29.4% during the later period (Panagariya & More, 2014).
Given the high population of Dalits in Bihar and UP, the two states accounting for 28.7% of the Dalit population in the country, it is worthwhile to study the nature and causes of high poverty among Dalits in the two states.
In the existing literature, the high incidence of poverty in the two states is ascribed to certain structural and institutional factors that have been at work for a very long time. The land distribution is highly skewed in favour of upper castes, and Dalits have a high extent of landlessness (Pai, 2004; Sharma, 1995; Tsujita et al., 2010). Factors like low education and skill status, lack of social networks, unstable employment contracts, insecure land tenure and possession of poor-quality land also account for high incidence of poverty among Dalits in UP (Kozel & Parker, 2003). Illiteracy among women is another factor for the persistent vulnerability of Dalit households in comparison to other castes (OC) in UP (Mehrotra, 2006).
Data Source and Methodology
The article, using different data sources, estimates inequality across social groups in different indicators of well-being like monthly per capita expenditure (MPCE), stunted-underweight children, infant and child mortality rates (CMRs) and prevalence of anaemia among women. It then investigates factors driving the high incidence of poverty among Dalits by looking at the distribution of landholdings, ownership of private enterprises and wage earnings across social categories. The article also estimates wage differential across social groups using the Oaxaca–Blinder decomposition (OBD) technique. The OBD technique captures wage differences between social groups due to differences in group-level characteristics (predictors) and due to differences in coefficients. The wage difference between groups due to differences in coefficients is often attributed to discrimination. The OB method is widely used to analyse inter-group differences in the labour market (Banerjee & Knight, 1985; Gupta & Kothe, 2022; Madheswaran & Attwell, 2007; Thorat et al., 2023).
OBD
A common way to study the dispersion of individual wages is to estimate a linear regression as:
where,
As we are particularly interested in comparing wage differential between two social groups (SC and other castes (OC)), the wage equation for two groups can be denoted as:
The mean wage difference
As
Based on the above equation, it can be deduced that inter-group difference in wage earnings may arise due to differences in the mean value of predictors, X, and due to differences in coefficients,
This is a ‘twofold decomposition, 1 ’ in which the outcome differences are divided into two parts. The first part on the right-hand side of Equation (6) is the explained component, that is, the component of the group differences in wage explained by the differences in predictors (the endowment effect). The second part of the right-side equation is the unexplained component (the coefficient effect), that is, the component of the group differences attributed to differences in coefficients. It would persist even if the SCs have attained the same average level of measured predictors as OC (Sen, 2014).
The decomposition shown in Equation (6) is formulated from the viewpoint of OC, considering OC’s coefficient can apply to both OC and SC in the absence of discrimination. However, this approach of estimating the effect of discrimination might involve the familiar index number problem (Oaxaca, 1973). In the problem, choice of the reference group (non-discriminatory group’s coefficient) affects the explained and unexplained component of the gap. Therefore, to avoid the index number problem, several alternative decomposition models are estimated along with the above for capturing the effect of discrimination. Those are based on assuming different forms of non-discriminatory coefficients (Cotton, 1988; Neumark, 1988; Reimers, 1983).
A general model for it is given in Jann (2008) and can be formulated as follows:
Let
The first component of the right side of Equation (7) is the endowment effect (explained part), and the second component (unexplained part) measures discrimination.
Reimers (1983) proposes using average coefficients over both groups as non-discriminatory coefficients. Cotton (1988) suggests taking both groups weighted average coefficient as a non-discriminatory coefficient. Weight is the proportion of the group’s sample in the total sample. Neumark (1988) has suggested to take a coefficient estimated from the pooled regression model. The article has used all the suggested non-discriminatory coefficients and provides comparative results.
The article shows that due to caste-based discrimination in the labour market, Dalits (SCs), on average, are paid lower wages in comparison to OC. In the absence of other income-earning assets (such as land and private enterprises), they remain working poor and consequently, poverty among them remains high and persists across generations.
The data sets used for the analysis in this article are NSSO Consumption Expenditure Survey (68th round, 2011–2012, type 2), Land and Livestock Survey (77th round, 2018–2019), Periodic Labour Force Survey (2022–2023), National Family Health Survey-5 (2021–2022) and Sixth Economic Census (2012–2013).
Poverty Across Social Groups in Bihar and UP
Figure 1 provides poverty estimates based on the Rangarajan committee’s recommendations. It shows a high incidence of poverty among Dalits vis-à-vis other social groups in Bihar and UP as well as at the all-India level. The incidence of poverty in the two states, in contrast to the all-India scenario, is higher in urban areas for all social groups.
Poverty Across Social Groups 2011–2012 (Rangarajan Committee Methodology).
The monetary poverty estimates may not always reflect the true nature of the overall poverty as an income-rich household may also perform poorly on other indicators of well-being and vice-versa. However, Dalits not only have a high incidence of monetary poverty, but even the other indicators of well-being fare poorly for them. Their average MPCE is the lowest among all social groups in both states (Figure 2). In Bihar and UP, it was only 54% and 80%, respectively, of the MPCE of ‘Others’. The MPCE of Dalits in the two states was also significantly lower than the MPCE among them at the all-India level, being 76% and 72%, respectively.
Average Monthly Per Capita Expenditure (MPCE).
The low MPCE translates into low-calorie consumption among Dalits, which in turn leads to poor health outcomes for them. The proportion of stunted children under age five is much higher among them in comparison to ‘Others’ and OBCs in the two states. These significant gaps are visible in both rural and urban areas (Table 1). Although both states perform poorly on child stunting with respect to the national level figures, the prevalence of child stunting is higher in Bihar, not only among Dalits but also among OBCs and ‘Others’. The proportion of underweight children is also higher for Dalits vis-à-vis OBCs and ‘Others’ in both the states, in both rural and urban areas. However, UP has a lower proportion of underweight children than Bihar (Table 2).
Proportion of Stunted Children Below Age Five Years, 2019–2021.
Proportion of Underweight Children Below the Age of Five in 1,000, 2019–2021.
The two states also fare poorly on other indicators of child health, namely infant mortality rate (IMR), CMR and under-five mortality rate (U5MR) in comparison to all-India figures (Table 3). These indicators are worse for Dalits in comparison to OBCs and ‘Others’. Out of 1,000 children, about 59 in Bihar and 64 children in UP die before attaining the age of five years. The corresponding figure in the case of Dalits is 64 in Bihar and 73 in UP.
Situation of Infant and Child Mortality Indicators Across Social Category in UP and Bihar 2019–2021.
The above discussion foregrounds that Dalits in the two states are not only monetarily poor, but they also have a poor status on other indicators of well-being in comparison to OBCs and ‘Others’ as well as in relation to Dalits at the all-India level. On some indicators, their situation is even worse than Adivasis. The subsequent section looks at some of the structural factors that account for poor performance of these indicators for Dalits.
Land Ownership Across Social Groups in Bihar and UP
A positive association is usually found between the size of land possessed and the average MPCE, particularly in rural areas (Government of India, 2007), implying that an increase in the size of landholding would help in reducing poverty. However, land ownership among Dalits in the two states is very poor. In rural areas, about 69% of Dalit households in Bihar and 49% of them in UP are landless (excluding homestead land). Another 30% and 54% of them, respectively, in the two states own marginal landholdings (Table 4). This means that 99.6% of rural Dalit households in Bihar and 96.8% of them in UP are either landless or own only marginal landholdings, indicating their negligible presence in the higher land-size categories. It is worthwhile to mention here that Arora and Singh (2015) have found, in the case of Dalits in UP, an insignificant association between the likelihood of being poor and the size of the landholding, and that even the large Dalit landholders (owning 2.01–4.00 hectares) witnessed a significant increase in the incidence of poverty during 2004–2005 and 2011–2012. Probing the reasons for this interesting phenomenon is beyond the scope of this article, but it may be safely attributed to the poor quality of lands owned by Dalits, coupled with the lack of irrigation facilities and other kinds of institutional support for them.
Distribution of Landholding Across Social Category in Rural Areas of UP and Bihar, 2018–2019.
Distribution of Landholding Across Social Category in Rural Areas of UP and Bihar, 2018–2019.
Status of Entrepreneurship Among Dalits in Bihar and UP
Very limited access to land as well as capital, which are sine-qua-non for doing well in entrepreneurship, has its bearing on the status of entrepreneurship among Dalits. The Economic Census, 2013, reveals that Dalits own 7.6% and 12.1% of private enterprises in Bihar and UP, respectively, a proportion much lower than their population share in the two states (Table 5). The Economic Census divides establishments into three types: own account establishments (OAEs), directory establishments (DEs) and non-directory establishments (NDEs). An establishment without any hired worker on a fairly regular basis and run by household labour is termed as an OAE. A DE is an establishment with hired workers employing six or more persons daily on a fairly regular basis. An NDE is an establishment with hired workers employing less than six persons daily on a fairly regular basis. Dalits account for 5.6% of NDEs in Bihar and 7.9% in UP. Their share in the ownership of DEs is even lower, only 3.9% in Bihar and 7.3% in UP. Evidently, over 90% of Dalit-owned units in Bihar and 85% in UP are own-account enterprises. Factors like lack of social capital and caste-based discrimination by upper castes also have a bearing on Dalits’ earnings and progression in entrepreneurship (Alha, 2020; Jodhka, 2010; Prakash, 2015). The high incidence of landlessness among Dalits, coupled with their poor access to credit resources (Prakash, 2015), results in their low presence in entrepreneurship and pushes many of them towards casual wage earnings in the informal sector. The low wages received by them in such employment avenues further limit their ability to accrue capital.
Representation in the Ownership of Private Enterprises Across Social Groups 2013.
High Concentration of Dalits in Casual Employment
There is a significant association between the size of the agricultural labour force and the incidence of poverty in a state (Jose, 1978, 1988). In other words, it is mostly the agricultural labour households who swell the ranks of rural poor. In rural Bihar, 17% of Dalit households have casual work in agriculture as their main source of earnings (Table 6). This proportion is much lower among OBCs (5%) and ‘Others’ (1%). In rural UP, the percentage of households who report casual labour in agriculture as their main source of earning is 6% in the case of Dalits, 2% in OBCs and only 1% in ‘Others’.
Distribution of Households as Per Their Main Source of Livelihood Across Caste in UP and Bihar, 2022–2023.
Table 6 also shows that 30% of Dalit households in rural Bihar and 32% in rural UP have casual work in the non-farm sector as their main occupation. For OBCs, this proportion is 17% and 14% in the two states, respectively. In contrast, among ‘Others’, only 7% in Bihar and 5% in UP eke out their main incomes from casual work in the non-farm sector. In urban areas of Bihar, 40% of Dalit households, 11% of OBC households and only 3% of ‘Others’ households are casual labour households. The corresponding figures in urban UP are 19% for SCs, 11% for OBCs and only 3% for ‘Others’. Quite clearly, the proportion of casual wage work, which is associated with poor earnings, is much higher among Dalits vis-à-vis OBCs and ‘Others’.
About 19% and 32% of rural Dalit households in Bihar and UP, respectively, derive their major earnings from self-employment in agriculture (Table 6). Given the huge predominance of small-sized landholdings among them, the average earnings of Dalit households engaged in cultivation in the two states is closer to the lower end of the earnings spectrum. The Situational Assessment of Agriculture Household Survey (2019) report shows that only Jharkhand, Odisha and West Bengal report lower earnings of agricultural households than Bihar and UP. The same report also shows that farming households in Bihar have an average monthly income of ₹7,542. About 33% of this earning comes from wage income and another 36% from net receipt from crop production. On the other hand, in UP, the average monthly income of farming households was ₹8,061 in 2018–2019. About 36% of this income was from wages, and another 41% was net receipt from crop production. It is to be noted that net receipts from non-farm businesses account for only 6.4% and 4.8% of the earnings of agricultural households in Bihar and UP, respectively (Statement 5.1A.1, page 129). Given that these figures reflect the average earnings of all farming households across all social groups, small-marginal farmers, a category in which most Dalit rural households are concentrated, must be earning even less. On the other hand, OBC and other castes households, on account of their possession of larger-sized landholdings, must be deriving higher earnings from cultivation.
Earning Differential Across Social Groups
In the two states, casual wages are almost equal for all social groups. At best, there is only marginal differences (Table 7). Relatively lower wage earnings in casual employment leave little scope for large variability in them, and hence, wage inequality in casual wages across social groups is also small. ST casual workers are receiving the lowest wages in both states, earning 90% of the wages received by other castes’ workers in Bihar and 83% in the case of UP.
Average Daily Wage Earning (in ₹) of Workers (15–59 years) Across Social Groups, 2022–2023.
However, earnings in self-employment show marked differences across social groups. In Bihar, a Dalit self-employed, on average, earns 35% less than his counterpart from ‘Others’ while an ST earns 23% less. OBCs in self-employment earn 78% of the daily earnings of self-employed persons from ‘Others’. These differences in UP are even sharper. The relative earnings of Dalits, STs and OBCs are only 55%, 41% and 67%, respectively, of the earnings of a self-employed person from other castes.
Notably, for Dalits, earnings in casual employment are higher than that in self-employment at the all-India level as well as in Bihar and UP. At the all-India level, a Dalit’s daily earning in self-employment, on average, is equal to 93% of the wage earnings of a Dalit casual worker. The corresponding figures in Bihar and UP are 86% and 76%, respectively. The lower earnings in self-employment in comparison to casual employment for Dalits and Adivasis in the two states indicate inadequate opportunities in wage employment, which results in many of the members of the two marginalised social groups making a distressed shift towards low-end, low-productivity and often caste-based occupations in self-employment as a last resort. These factors compromise the earning potential of Dalits and Adivasis in self-employment and lead to swelling of BPL households among them. Notably, a high incidence of poverty among self-employed persons in Bihar and UP is not a recent phenomenon. The share of self-employed persons living below the poverty line (PL) was as high as 42% in Bihar and 53% in UP in 2005. These proportions have come down to around 31% in Bihar and 27% in UP in 2022. It is only in the states of Jharkhand, Maharashtra and Chhattisgarh that poverty among self-employed persons is higher than that in Bihar (ILO-IHD, 2024).
On the other hand, for ‘Others’ social group, earnings in self-employment are notably higher than that in casual employment at the all-India level as well as in Bihar and UP. Among them, daily earnings in self-employment are 134% of the daily wage income of a casual worker in Bihar, 124% in UP and 144% at the all-India level. The higher earnings in self-employment for them can be attributed to ownership of larger-sized lands and capital resources, higher educational status, better access to credit and resourceful social capital.
The inter-group disparities in earnings are also visible in regular salaried employment in which Dalits, on average, earn only 66% of what is earned by ‘Others’ at the all-India level as well as in UP and 68% in Bihar.
To examine the factors causing the difference in wages between social groups, we employ the standard OBD. The decomposition results are shown in Table 8. Panels C–E of the table report alternative variants of OBD results. These variants assume different non-discriminatory coefficients. The dependent variable is the log of daily regular wage, while independent variables include age, age-square, level of education, technical education, vocational education, gender, sector and state dummies. Among these variables, differences in age, education level, technical education and gender contribute significantly in explaining the gap in wages across social groups. The results of these individuals’ coefficients are not reported here due to the paucity of space. The results reported in Table 9 indicate that the wage gap between SC, ST and OBC vis-à-vis ‘Others’ is statistically significant. Decomposing this wage gap between explained and unexplained component, the results show that both components significantly contribute to the wage gap.
Decomposition Results of Daily Wage Gap of Regular Workers Between SC–Others, ST–Others and OBC–Others in UP, Bihar and All India, 2022–2023.
Percentage Distribution of Workers Earning Less than Minimum Wage Across Social Groups, 2022–2023.
At the all-India level, 60%–70% of the wage gap between SCs and ‘Others’ is explained by differences in characteristics between the two groups, while 30%–40% of this gap is unexplained, which can be attributed to caste discrimination in the labour market. In UP, 60%–75% of the wage gaps between SCs and ‘Others’ is explained by differences in characteristics, while 25%–40% is unexplained. In Bihar, 20%–40% of the wage gap remains unexplained. The unexplained component in both states can be attributed to caste discrimination. A similar scenario is visible in the wage gap between STs and ‘Others’ in UP as well as at the all-India level. For Bihar, the explained portion has a negative though significant value, which is possible because of a low number of observations. However, in considering the non-discriminatory coefficient from the pooled model as suggested by Neumark (1988), around half of the variation in the wage gap between the groups is explained by differences in group-level characteristics, while the remaining half is unexplained. The components are statistically significant at a 1% level of significance.
Between OBCs and ‘Others’, a higher proportion of wage gap between OBCs and ‘Others’ is explained by differences in group-level characteristics in Bihar as well as at the all-India level. In UP, differences in group-level characteristics account for roughly half of the wage gap, while the remaining gap in wages remains unexplained.
Working Poor in Bihar and UP
In labour markets characterised by poor quality of jobs, a large proportion of workers, in spite of remaining fully employed, earn less than even the stipulated minimum wages. At the all-India level, such workers constitute over 40% of the total workers (Table 9). The share of Dalits, Adivasis, OBCs and ‘Others’ in workers earning less than the minimum wages is 45%, 56%, 42% and 29%, respectively. About 58% of workers in UP and only 32% of workers in Bihar receive lower than the minimum wages. The low figures in the latter are perhaps due to lower minimum wages in comparison to UP and all-India.
To estimate the size of working poor, we have used the method proposed by Kannan (2014). Kannan on the basis of a household’s monthly consumption expenditure has classified households into six categories of ‘extremely poor’, ‘poor’, ‘marginal’, ‘vulnerable’, ‘middle income’ and ‘high income’ group. Using the same methodology, we have estimated the proportion of ‘poor’, ‘marginal’ and ‘vulnerable’ households. Workers who, at the household level, earn less than the poverty line (PL) of income (defined by Rangarajan Committee) are termed as working poor. Workers who earn above the PL but less than 1.25 times the PL, and those who earn between 1.25 times the PL but less than twice the PL, are termed as ‘marginal poor’ and ‘vulnerable’, respectively. For estimating the share of working poor, the state-wise minimum wage fixed for unskilled workers in 2018–2019, the latest year for which this data are available, is used. The consumer price index (General Index for base year 2011–2012) is used here to update the minimum wage from 2018–2019 to 2022–2023.
At the all-India level, around 36% of workers earn wages which are unable to keep them above the PL (Table 10). Another 9% of workers are marginal poor, earning between 1 and 1.25 times the PL. Over 27% of workers earn wages between 1.25 times the PL but less than twice the PL. They are classified as vulnerable workers. While over 72% of the workers in the country fall into either of these three categories, this proportion among Dalits and Adivasis, at 80% and 84%, respectively, is noticeably higher. The corresponding figures for ‘Others’ and OBCs are 62% and 73%, respectively.
Distribution of Poor, Marginal Poor and ‘Vulnerable’ Workers (PS+SS) in India, 2022–2023.
In Bihar, nearly 79% of workers are either working poor, marginal poor or vulnerable. These figures across social categories are 85% for Dalits, 80% for STs, 79% for OBCs and 66% for ‘Others’. In UP, around 88% of workers in total fall into these categories. About 55% of them are working poor, 10% are marginal poor, and another 23% are considered vulnerable. The proportion of working poor among STs, SCs, OBCs and ‘Others’ in the state is 79%, 55%, 58% and 44%, respectively (Table 11).
Distribution of Poor, Marginal Poor and ‘Vulnerable to Poor’ Workers (PS+SS) in Bihar and UP, 2022–2023.
The low proportion of working poor in Bihar in comparison to UP and all-India is possibly due to lower stipulated minimum wages in the state. A high proportion of marginal poor and vulnerable workers among Dalits and OBCs in the two states and STs in UP is a cause of concern as even a small exogenous shock can push such workers below the PL.
In concluding this article, some broad suggestions are offered here which could help in eliminating poverty in the two states. A reduction in poverty among Dalits and Adivasis in the two states is essential to reduce the level of overall poverty. The growth elasticity of poverty is a decreasing function of the degree of inequality of the income distribution (Bourguignon, 2002). Thus, states with a more equal distribution of income experience a greater reduction in the poverty rate for a given increase in per capita income (Shah, 2016). Hence, the focus must be on reducing high levels of inequality to eliminating poverty in Bihar and UP. The caste-based inequality in land distribution, as well as a high level of landlessness among Dalits, is a huge hindrance in eliminating poverty. Given the difficulties in initiating a fresh round of land reforms to provide land to landless, pulling out a large number of landless out of agriculture towards the secondary and tertiary sectors will certainly be one way forward. This is going to be a humongous task given the huge importance of agriculture in the two states. While governments in the two states have initiated special provisions for mitigating inequalities among the marginalised castes, there is a need to boost up such efforts. Efforts should also focus on building alternative livelihood options for the landless poor, such as the development of the livestock sector and the sincere implementation of MGNREGA.
Given the high proportion of working poor among Dalits, there is a need for fixing and strictly implementing a wage floor. In order to reduce out-of-pocket expenditure on health and education, which impacts poor households more and pushes them below the PL, more resources need to be pumped in for strengthening the overall health and educational infrastructure. In fact, the state governments should aim to effectuate universalisation of health and education.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.
