5. Inequality of Opportunity in Education

Education stands as a cornerstone for achieving the United Nation’s Sustainable Development Goals (SDGs), particularly Goal 4, which aims to ‘ensure inclusive and equitable quality education and promote lifelong learning opportunities for all’. The significance of education transcends individual benefits, extending to societal progress and economic development. However, educational inequality poses a significant obstacle in achieving not only SDG 4 but also other interconnected SDGs such as reduced inequalities (Goal 10) and poverty alleviation (Goal 1). The educational inequalities and their resultant socio-economic consequences are more pronounced in the developing countries compared to the developed countries. Post-1980s, with the advent of globalization, education systems in developing countries faced further challenges. Structural adjustment policies imposed by the World Bank and other international financial institutions constrained national budgets, leading to cuts in education spending and weakening secondary and higher education systems.

The Global Education Monitoring Report 2023 by UNESCO highlights significant disparities in educational access across different regions. Globally, 9% of children of primary school age are out of school. This figure varies significantly across regions. In Europe and North America, only 1% of primary school-aged children are out of school, while in Sub-Saharan Africa, the rate is much higher at 20%. Other regions with high out-of-school rates include North and West Africa (9%), Oceania (7%), and South Asia (7%). Among all the regions, the situation is more challenging for older children in Sub-Saharan Africa and South Asia: 33% of lower secondary and 48% of upper secondary school-aged children are out of school in Sub-Saharan Africa, while 13% of lower secondary and 39% of upper secondary school-aged children out of school in South Asia. This shows that the proportion of children out of school is relatively low at primary level schooling in South Asia compared to other regions, whereas, the rate increases significantly for older age groups. Among South Asian countries, the rates of out-of-school children are relatively high in India (13%), Bangladesh (10%), and Bhutan (10%) for lower secondary school-aged children. However, the situation is even more critical for upper secondary school-aged children, with Afghanistan (56%), the Maldives (50%), and India (44%) exhibiting significantly high out-of-school rates. These disparities highlight the urgent need for targeted educational interventions and policies to ensure that all children in South Asia have access to quality education. This high rate of out-of-school children at both lower and upper secondary levels in India underscores the need for urgent attention.

Further, the level of education among the labour force aged 15 and older in India shows significant disparities compared to developed countries like the USA and Germany, as well as developing countries such as Brazil and South Africa (Fig. 5.1). A large portion of India's labour force has less than a basic education, meaning many people do not even finish primary school. Additionally, a significant number of workers in India only have basic education. This uneven educational distribution is further highlighted by the relatively low percentage of individuals achieving intermediate and advanced education levels, which are crucial for developing a skilled and competitive workforce. In contrast, higher rates of intermediate and advanced education are seen in developed nations like the USA and Germany, as well as in developing countries such as South Africa and Brazil. This situation reflects persistent educational inequality in India, manifesting in various forms, including unequal access based on socio-economic background, gender, and geography (Tilak, 2007; Kingdon, 2010; IER, 2024).

This educational disparity has deep-rooted causes. Economic barriers, such as poverty, prevent many families from sending their children to school or keeping them in school long enough to complete higher levels of education. Social inequalities, including caste and gender discrimination, further limit educational opportunities for many individuals. Additionally, inadequate infrastructure, such as poorly equipped schools and a shortage of qualified teachers, exacerbates the problem, particularly in rural and underserved areas. Despite significant policy efforts, such as the Right to Education Act in 2009, which aims to provide free and compulsory education to children, these initiatives have not fully addressed the underlying issues. In 2020, the Government of India also introduced a new education policy aimed at enhancing early childhood education and improving overall educational quality. Despite its promise, the policy has yet to be fully implemented across many states in the country. Studies suggest that improving educational outcomes in India necessitates targeted policies that tackle specific challenges, including the quality of education, investment in educational infrastructure, and support for disadvantaged groups. Addressing these issues is crucial for reducing educational inequality. Understanding and addressing the inequality of educational opportunities—an essential aspect of educational disparity—requires recognizing and overcoming barriers that hinder equitable access to quality education for individuals across different socio-economic background, gender, and geographical locations (Das, 2022; Kundu, 2023). By tackling these root causes, India can work towards building a more equitable and skilled workforce. In this context, this chapter examines the inequality of educational opportunities (IOp) in India and investigates the factors contributing to these disparities.

This chapter is organized into five sections. Following the introduction, the second section outlines the study's framework, detailing the theoretical and conceptual foundations underpinning the research. The third section delves into the methodology and data sources, explaining the approach and data used to analyse educational IOp. The fourth section presents the results and discussion, offering an in-depth analysis of the findings and their implications. The final section summarizes the key findings and provides concluding remarks, highlighting the main insights and their relevance to understanding and addressing educational IOp in India.

As discussed in detail in Chapters 2 and 3 earlier, the concept of equality of opportunity stems from the idea of distributive justice. This idea pertains to the principles and rules governing the distribution of resources and evaluates outcomes against expectations (Sabbagh & Schmitt, 2016). John Rawls (1971) pioneered the theory of justice by proposing a normative framework for creating a just society through social, political, and economic institutions. These institutions play a significant role in regulating the distribution of goods and social burdens among individuals, thereby shaping their life choices (Sabbagh & Schmitt, 2016). According to Rawls (1999), the principles of justice should assign basic rights and duties and determine the proper distribution of benefits and burdens for social cooperation. He believed that the goods to be distributed in a society should be primary goods that every rational individual should possess.

Rawls’ conception of justice was challenged by Dworkin (1981a, 1981b), who argued that people in a society have different material needs and tastes. Therefore, distributing primary goods fairly would involve distributing different amounts of wealth or income to different individuals, where those with expensive tastes would need more. Arneson (1989) and Cohen (1989) disagreed with Dworkin’s view, arguing it held individuals responsible for factors beyond their control. They shifted the focus from resource distribution to the availability of opportunities. Arneson (1989) advocated for ‘equal opportunity for welfare’, suggesting that every individual should face the same set of possibilities for fulfilling their preferences. Cohen (1989) proposed the idea of ‘equality of advantage,’ extending beyond welfare to include broader aspects of advantage. Despite their different views, Rawls, Dworkin, Arneson, and Cohen all emphasized that an equitable society is one where everyone has an equal opportunity to achieve their desired outcomes (Mehta & Dhote, 2022).

Roemer (1998) further refined this idea by distinguishing between effort and circumstances. Effort refers to factors within an individual’s control, such as hours dedicated to work or study, occupational choices, and quality of work. Circumstances, on the other hand, are factors beyond an individual's control, such as family background, ethnicity, gender, and socio-economic status. This relationship can be mathematically represented for a population of individuals from 1, …, N, where each individual’s outcome (y_i)_, is determined by their circumstance (C_i) and effort (e_i):

Roemer (2002) proposed a framework for achieving equality of opportunity by categorizing individuals based on circumstances (types) and effort levels (tranches). The literature on inequality of opportunity (IOp) includes the ex-ante and ex-post approaches. The ex-ante approach focuses on inequalities between individuals with different circumstances or types, while the ex-post approach assesses inequality among people with the same effort level (Fleurbaey & Peragine, 2013; Ramos & Van de Gaer, 2016). The ex-ante approach considers the type-specific outcome distribution as the opportunity set for individuals, with the value of this set determined by the mean outcome of each type (Ferreira & Gignoux, 2011).

The counterfactual distribution (Ý_EA): Ý_EA is constructed by replacing the actual outcomes of individuals with the mean outcome of their respective types. This is done to isolate the effect of circumstances (types) on the outcomes, removing the influence of individual effort. The type-specific outcome distribution is considered the opportunity set, with its value determined by the mean outcome of each type:where

By replacing each individual's outcome y_i with the mean outcome of their type μ^k, a new distribution \({\acute{Y}}_{EA}\) is constructed. This new distribution reflects only the differences between types, ignoring within-type variation due to effort or other factors.

Calculating Ex-Ante Inequality of Opportunity (IOp_EA): The ex-ante inequality of opportunity IOp_EA is then measured by applying an inequality index I (such as the Gini coefficient or another inequality measure) to the counterfactual distribution \({\acute{Y}}_{EA}\).

The transformation from the individual outcomes to the counterfactual distribution \({\acute{Y}}_{EA}\) and the subsequent calculation of IOp_EA allows us to measure the inequality that arises solely from differences in circumstances, abstracting away the influence of individual effort. This provides a clearer picture of inequality of opportunity in the given context.

Education is a crucial factor for distributive justice because educational institutions distribute a variety of resources that play a significant role in shaping individuals’ futures. These resources include not just academic knowledge, but also attention, support, care, and respect (Resh & Sabbagh, 2016). By distributing these resources, schools create different learning opportunities and social experiences that influence individuals’ motivation, academic achievements, and ultimately their life chances (Bills & Wacker, 2003; Hurn, 1985; Oakes et al., 1992). In the context of education, the concept of Equality of Educational Opportunity (EEO) asserts that every child has the right to education, regardless of their family background, nationality, ethnicity, socio-economic status, religion, or gender (Resh & Sabbagh, 2016). The idea behind EEO is that a common schooling system can act as a great equalizer, allowing individuals effort and ability to overcome differences in their initial circumstances and educational outcomes (Resh & Sabbagh, 2016).

The interpretation of EEO has evolved over time, reflecting a shift in understanding from ‘equality of input’ to ‘equality of output’ (Resh & Sabbagh, 2016). ‘Equality of input’ focuses on distributing resources equally among all individuals. In contrast, ‘equality of output’ emphasizes addressing the different starting points of individuals. This means compensating for disadvantages faced by weaker and marginalized groups to provide them with equal opportunities to achieve similar outcomes (Coleman, 1968; Kellough, 2005). The participation in the labour market is significantly influenced by level of education, as it plays crucial role in ensuring a smooth transition from school to decent work.

This analysis indicates that education as a part of distributive justice means making sure all individuals get fair resources and support, helping them reach their full potential no matter their background. The evolving idea of Equality of Educational Opportunity (EEO) emphasizes not only distributing resources equally but also addressing the different starting points of children to achieve true educational fairness. This approach helps to ensure their smooth transition from school to work after completing the education and contributes to building a skilled labour force.

This chapter uses machine learning techniques to calculate ex-ante educational Inequality of Opportunity (IOp) in India. As discussed in the earlier chapter, ML algorithms are preferred over conventional methods, such as Ordinary Least Squares (OLS), because conventional methods often suffer from researchers’ discretion in selecting circumstance variables. Arbitrary selection can lead to the exclusion of important variables or the inclusion of too many variables. Excluding important variables reduces the model's explanatory power, leading to downward biases, while including excessive variables results in upward biased estimates (Brunori et al., 2019; Hufe et al., 2017; Ferreira & Gignoux, 2011). ML algorithms help overcome these limitations by minimizing the risks of ad-hoc and arbitrary data selection, balancing upward and downward biases (Hothorn et al., 2006; Brunori et al., 2019; Brunori & Neidhofer, 2021; Hothorn & Zeileis, 2021; Salas-Rojo & Rodriguez, 2022). The ML algorithms used in this study to measure ex-ante educational IOp are called conditional inference trees and conditional inference forests. These algorithms serve a dual purpose: they create circumstance-based types and predict the outcome variable. They have been used in several studies predicting income-based IOp (Brunori et al., 2018, 2019; Brunori & Neidhofer, 2021; Lefranc & Kundu, 2020; Mehta & Dhote, 2022; Mehta et al., 2023).

Conditional inference trees provide a visual representation of the opportunity structure for a particular outcome variable by recursively splitting the range of circumstances, enabling the identification of sub-groups with similar circumstances. Conditional inference forests, a variant of conditional inference trees, generate multiple trees and combine their results by averaging. The repetitive extraction of subsamples by the algorithm ensures the independence of each tree, enhancing the reliability of IOp estimates. An additional feature of conditional forests is their ability to provide the relative importance of factors used in creating the trees. Furthermore, a significant advantage of machine learning methods is that when a large set of circumstances is present, only those circumstances that have a statistically meaningful relationship with the outcome variable are considered.

This study uses data from the Periodic Labour Force Survey (PLFS) of 2022–23 conducted by the National Statistics Office (NSO), which is representative at the national and state levels. The outcome variable used to assess educational IOp is years of education, given in number of years. The circumstance variables used to measure educational IOp (Table 5.1) are as follows:

The sample used for measuring educational IOp includes 30,087 individuals. This sample consists of individuals who have completed their education or dropped out, are above the age of 15, and have information on their parents’ education and occupation. The choice of machine learning methods over conventional approaches in this chapter aims to provide more accurate and unbiased estimates of educational IOp by effectively handling the selection of circumstance variables and ensuring robust results.

There are significant differences in educational access between rural and urban areas, with rural areas generally having lower average years of schooling compared to urban areas. Gender disparities, while present, are less severe, but there is still variation in educational attainment among women. Social group disparities are noticeable, with Scheduled Tribes (ST) and Scheduled Castes (SC) having lower average education levels compared to Other Backward Classes (OBC) and the General category. Regional differences also play a role, with states like Central India falling behind more developed regions like the South.

The Gini coefficient is 0.22, which indicate significant disparities in education. However, the factors beyond individual control or circumstance or IOp account for about one-third (35%) of the total education inequality. Machine learning methods like conditional inference trees and forests also confirm the same, and highlight that parental education is the most important factor affecting educational inequality, with regional disparities also playing a significant role. The most influential factors contributing to educational inequality are: parental education (62.7%), region (20.6%), occupation (7.2%), sector (5.8%), social group/caste (3.0%), and gender (0.7%). Among states, educational inequality is highest in Bihar and Rajasthan, moderate in Uttar Pradesh and West Bengal, and lowest in Kerala.

The conditional tree classification analysis reveals that the most disadvantaged group in terms of educational opportunities includes individuals with uneducated parents and those whose parents are in low-skilled or unskilled jobs, especially in the northern, eastern, central, northeastern, and western regions of India. These individuals have the lowest average schooling of 8.3 years. In contrast, the most advantaged group includes those living in urban areas with parents who have skilled jobs and graduate level or higher qualifications, achieving the highest average schooling of 14.7 years.

The regional analysis indicates that less developed states like Uttar Pradesh and Rajasthan show high educational inequality, where factors such as parental background, social groups, and gender contribute significantly to this inequality. On the other hand, states like Himachal Pradesh, Maharashtra, Kerala, Punjab, Haryana, Uttarakhand, and Delhi have lower levels of educational inequality, but circumstances still play a significant role in contributing to the inequality in these states.

To address these challenges, targeted policies should focus on enhancing educational infrastructure in underserved areas and implementing programs specifically designed to support disadvantaged groups. Investing in quality education, particularly in rural and less developed regions, is crucial. Policies should also aim to improve access to education for marginalized social groups and ensure gender equity in educational opportunities. States like Kerala, which exhibit low educational inequality due to effective policies and better infrastructure, can serve as models for other regions. By adopting and adapting successful strategies from these states, India can work towards reducing educational disparities nationwide.