Dieses Kapitel untersucht das sich entwickelnde Konzept der Chancenungleichheit (IOp) und verlagert den Fokus von ergebnisbasierter Ungleichheit auf die Chancenungleichheit, die wirtschaftlichen und sozialen Ungleichheiten zugrunde liegt. Sie zeichnet die theoretische Entwicklung des IOp nach, von traditionellen Ansätzen des Wohlfahrtsstaates hin zu differenzierteren Rahmenwerken, die zwischen "fairen" und "unfairen" Ungleichheiten unterscheiden. Das Kapitel vertieft sich in die Messung des IOp und diskutiert die Prinzipien von Belohnung und Entschädigung sowie die Ex-ante- und Ex-post-Ansätze. Er hebt die Grenzen traditioneller Methoden hervor und führt maschinelle Lernalgorithmen wie konditionierte Rückschlussregressionsbäume als vielversprechende Werkzeuge für genauere und aufschlussreichere Schätzungen des IOp ein. Die Analyse konzentriert sich auf Indien und nutzt Erhebungsdaten der privaten Haushalte, um die Trends und regionalen Unterschiede bei Verbrauch und Einkommen zu untersuchen. Sie zeigt, dass der elterliche Hintergrund, insbesondere Bildung und Beruf, eine entscheidende Rolle bei der Bestimmung ungleicher Chancen spielt. Das Kapitel unterstreicht auch die Notwendigkeit gezielter politischer Maßnahmen, um diese tief verwurzelten Ungleichheiten anzugehen, und betont die Bedeutung der Verbesserung des Zugangs zu qualitativ hochwertiger Bildung und der Umsetzung progressiver sozialer Maßnahmen. Die Ergebnisse bieten wertvolle Erkenntnisse für Entwicklungsländer, die sich mit zunehmenden Einkommensunterschieden und den anhaltenden Herausforderungen der Chancenungleichheit auseinandersetzen.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
This chapter explores the growing interest among scholars and policymakers in measuring inequality of opportunity (IOp) in income. It is well documented that focusing solely on inequality based on outcomes does not fully explain the rising income disparities within and across countries. It first traces the evolution of IOp concept and later presents empirical findings for India using the data from National Sample Survey (NSSO). The study employs both traditional and machine learning techniques to estimate income IOp. The results show that about 26–27% of income inequality in India can be attributed to factors beyond an individual's control. Parental education and occupation emerge as the most significant contributors to income IOp. Among regular workers, these parental backgrounds play a key role. For self-employed, gender is the primary driver of IOp, while for casual workers, geographical has the greatest impact. These findings highlight the importance of addressing unequal circumstances to promote a fairer and more inclusive society.
Hinweise
Disclaimer: The presentation of material and details in maps used in this chapter does not imply the expression of any opinion whatsoever on the part of the Publisher or Author concerning the legal status of any country, area or territory or of its authorities, or concerning the delimitation of its borders. The depiction and use of boundaries, geographic names and related data shown on maps and included in lists, tables, documents, and databases in this chapter are not warranted to be error free nor do they necessarily imply official endorsement or acceptance by the Publisher or Author.
Earlier version of this chapter was published as a working paper by Institute for Human Development (IHD), and in the IASSI Quarterly.
2.1 Introduction
Inequality, whether economic, social, or political, has consistently been a major concern globally. Over the past three decades, the rising economic inequality across nations has drawn significant attention from policymakers and scholars alike (Morelli & Rohner, 2015). This growing interest has led to numerous studies on the measurement of inequality focusing predominantly on income or consumption expenditure. In other words, traditionally, the past studies employed a welfarist approach to measure income inequality, with inequality in outcomes as their primary focus. This classical approach, while valuable, has faced several criticisms for its insufficient consideration of the multifaceted factors that contribute to inequality (Dworkin, 1981b). This critique sparked a philosophical debate in the late twentieth century, emphasizing responsibility-sensitive egalitarian justice (Roemer, 1993, 1998). This approach differentiates between ‘fair’ and ‘unfair’ inequalities, highlighting the role of individual responsibility in distributive justice. It shifts the focus from merely examining ‘inequality of outcomes’ to addressing ‘inequality of opportunities (IOp)’. This notion of IOp provides a framework to distinguish between ‘good’ and ‘bad’ inequalities in a society, crucial for achieving both economic efficiency and social cohesion. Unequal societies may inadvertently hinder certain segments of the population while favouring others, making it essential to understand how IOp influences these dynamics.
In this background, it is important to understand to what extent the inequality is driven by IOp. However, the traditional methods and data sources for measuring IOp have several limitations. These conventional approaches often struggle to capture the complexity of IOp and its determinants accurately. To address these shortcomings, the application of Machine Learning (ML) algorithms emerges as a promising solution. ML techniques offer advanced analytical capabilities, potentially providing more nuanced and robust estimates of IOp by analysing large and complex datasets. In this light, the first objective of this chapter is to explore the theoretical evolution of the concept of IOp, while second objective is how emerging ML techniques can offer new and insightful perspectives on IOp. By leveraging ML algorithms. The aim is to overcome the constraints of traditional methods and enhancing the understanding of the contribution of IOp in overall inequality, and factors contributing to IOp.
Anzeige
This issue is particularly important in developing countries, where income inequality has been widening over the last few decades, with some recent reversals in countries like India. As discussed in the introductory chapter, despite the impressive economic growth witnessed in India since the economic reforms of the 1990s, the benefit of growth has not led to equitable improvements in economic and social welfare of the people (Sharma, 2015). This persistent and widening inequality underscores the importance of addressing IOp to achieve meaningful reductions in overall inequality. Existing studies have extensively analysed outcome-based inequality in India, but research specifically focusing on IOp remains limited. This gap in the literature highlights the need for a comprehensive examination of IOp within the Indian context, serving as a case study for developing countries. Thus, the third objective of this chapter is to measure and identify the factors contributing to IOp in India, utilizing both conventional methods and ML algorithms. By integrating these approaches, this chapter aims to provide a clearer understanding of IOp in India and offer insights that can guide policy development for India as well as other developing countries. This can help in addressing the root causes of inequality and create equal opportunities for all individuals in society.
The remainder of this chapter is structured as follows: Sect. 2.2 outlines the evolution and theoretical underpinnings of IOp. Section 2.3 details the methodologies and approaches used in the analysis. Section 2.4 discusses about the data sources, and Sect. 2.5 presents the findings from the India-based analysis, and Sect. 2.6 concludes with summary, policy recommendations and future research directions.
2.2 Evolution of IOp Concept
The concept of IOp has evolved significantly over time, influenced by broader philosophical and economic debates on justice and fairness. To understand the development of IOp, it is essential to trace its roots and the theoretical shifts that have shaped its current interpretation as briefly reviewed below.
Origins and Early Theoretical Frameworks: The idea of equality of opportunity is deeply rooted in the principle of distributive justice. Historically, this principle has often been associated with welfarist approaches, where equity assessments are based on the distribution of individual achievements—such as welfare, utility, or satisfaction—across a population. Utilitarianism, a prominent version of this tradition, aggregates individual achievements to form a social objective function, aiming to maximize overall utility. However, utilitarianism and its welfarist foundations faced significant challenges in political philosophy and normative economics (Dworkin, 1981b).
Anzeige
John Rawls and Egalitarianism: John Rawls (1958, 1971) introduced a critical shift with his egalitarian perspective, challenging utilitarianism. Rawls proposed the concept of primary goods—basic liberties, rights, income, and wealth—as central to justice. He argued that justice requires institutions that maximize the distribution of these primary goods to the least advantaged people of the society, ensuring a fair starting point for all individuals. Rawls's theory posits that true equality of opportunity is achieved when individuals have equal access to the basic goods necessary to pursue their interests and ambitions.
Critiques and Alternatives: The notion of equality of opportunity faced critiques from scholars like Ronald Dworkin (1981b, 1981c). Dworkin questioned the feasibility of equalizing welfare, noting that individuals have diverse material needs and preferences. He argued that achieving equality of welfare might require different amounts of wealth or income for different individuals. Dworkin introduced the concept of ‘equality of resources’, which extends beyond tangible goods to include biological traits and personal talents. Arneson (1989) and Cohen (1989) further critiqued Dworkin’s approach. Arneson proposed ‘equal opportunity for welfare,’ suggesting that equality is achieved when every individual has the same set of possibilities to satisfy their preferences. Cohen's ‘equal access to advantage’ aligns closely with Arneson’s view but includes a broader range of advantages beyond mere welfare.
Amartya Sen Capability Approach: Amartya Sen (1985, 1992, 1999) contributed another significant perspective with his ‘capability approach.’ Sen argued that focusing on individuals’ capabilities—their ability to achieve desired functioning or states of being—provides a more comprehensive measure of equality. This approach integrates elements of Arneson and Cohen’s proposals, emphasizing the freedom and capability to achieve various life outcomes rather than just the distribution of resources or goods.
Roemer’s Conceptualisation: The work of John Roemer (1998, 2002) represents a critical development in the formalisation of IOp concept. Roemer conceptualised IOp as the outcome resulting from two sets of factors: those within an individual’s control (efforts) and those beyond their control (circumstances). Efforts include factors like the number of hours worked or educational choices, while circumstances encompass aspects such as family background, socio-economic status, gender, and ethnicity. Roemer’s framework provides a clear distinction between what individuals can be held accountable for and what they cannot, thus defining IOp as inequality arising from circumstances rather than efforts (Roemer & Trannoy, 2016a, 2016b).
Empirical Developments and Methodological Approaches: The theoretical underpinnings of IOp have spurred considerable empirical research. Scholars have developed various methods to estimate IOp and analyse its impact on economic outcomes. Important contributions include: Bourguignon et al. (2007): provided early empirical work on estimating IOp and its relation to economic inequality. Ferreira and Gignoux (2008, 2011, 2014): offered methodologies for measuring IOp and its effect on income inequality. Barros et al. (2009a; 2009b) explored the role of IOp in shaping consumption and income disparities. Fleurbaey and Schokkaert (2009) contributed to the understanding of how IOp affects welfare outcomes. Cecchi and Peragine (2010) expanded on methods to quantify IOp and its implications. These studies have advanced the measurement of IOp, offering valuable insights into its determinants and consequences. The next section delves deeper into the concept and measurement of IOp, reviewing the existing literature and discussing methodological advancements.
2.3 Measurement of IOp
The measurement of IOp hinges on two fundamental ethical principles: the principle of reward and the principle of compensation. These principles guide how we understand and evaluate inequality in relation to individual efforts and circumstances.
Principle of Reward and Principle of Compensation: The principle of reward is rooted in the idea that differential rewards should be preserved when they arise from individual responsibility and effort. In contrast, the principle of compensation asserts that individuals should be compensated for disadvantages arising from circumstances beyond their control (Fleurbaey, 1994). These principles shape how one should categorize and address IOp. To refine these concepts, the literature distinguishes between two key groups: types and tranches. Types refer to groups of individuals who share the same set of circumstances or opportunities. In contrast, tranches denote groups of individuals who exert the same level of effort. According to Peragine (2004a; 2004b), ‘within-type’ inequality arises from differences in individual effort, which is considered morally permissible. Conversely, ‘between-type’ inequality is driven by disparities in circumstances and is seen as inequitable, thus warranting compensation. Roemer (1998) contends that the principle of reward, or the concept of ‘within-type’ inequality, is less central to the measurement of IOp. His focus is primarily on the principle of compensation, which emphasizes the need to address inequality arising from circumstances beyond an individual's control. Consequently, while empirical evidence on ‘within-type’ inequality is limited, substantial research supports the principle of compensation in understanding and measuring IOp.
Ex-Ante and Ex-Post Approaches: The measurement of IOp is further informed by two distinct perspectives: the ex-ante and ex-post approaches (Fleurbaey & Peragine, 2013a; 2013b). The ex-ante approach examines inequality between types—groups of individuals who have identical circumstances. This approach is predicated on the idea that individuals should have equal opportunities to achieve similar outcomes, given the same circumstances. In contrast, the ex-post approach focuses on inequality between tranches—groups of individuals who have exerted the same level of effort. This perspective assesses how different efforts yield varying outcomes and the role of individual responsibility in these outcomes. The distinction between these approaches arises from differing views on the nature and measurement of effort. The ex-post approach has been less popular due to challenges in accurately measuring effort, which requires strong assumptions and is often difficult to operationalize (Fleurbaey & Peragine, 2013; Ramos & Van de Gaer, 2016). As a result, most empirical applications and studies predominantly utilize the ex-ante approach. This approach is more straightforward, as it focuses on comparing outcomes based on equal circumstances, without delving into the complexities of effort measurement.
Overall, the measurement of IOp involves navigating between the principles of reward and compensation and choosing between ex-ante and ex-post approaches. The ex-ante approach, focusing on inequalities between types, is more prevalent due to its relative simplicity and fewer methodological challenges compared to the ex-post approach. This framework guides our discussion and analysis of IOp, aligning with the predominant empirical focus in the field.
Regression Approach to Measuring IOp: Over time, various methods have been proposed to assess ex-ante IOp, with the regression approach emerging as one of the most popular and widely utilized techniques across different countries and outcomes. This has become a widely adopted method because of its flexibility and applicability across different contexts and outcomes. This method primarily involves relating an outcome variable to a matrix of circumstances, either through parametric or non-parametric regression methods. Ferreira and Gignoux (2011) applied this method using income as the dependent variable, estimating the relationship with both ordinary least squares (OLS) regression and non-parametric methods, such as averaging over types. The method can be summarized as follows: Let y represent the outcome variable of interest, such as individual earnings or income, and X denote the matrix of circumstances beyond an individual's control, such as race, gender, parental education, and occupation. The relationship between the outcome variable and the vector of circumstances can be described by the expected conditional outcome, represented by the equation:
$$y_{i} = f(X_{i} ) + \epsilon_{i}$$
(2.1)
where f(Xi) denotes the function that captures the impact of circumstances Xi on the outcome yi, and ϵi represents the error term. This relationship can be estimated using various methods depending on the research question and the nature of the dependent variable. Regardless of the estimation method used, IOp is computed using a common inequality measure applied to the outcomes derived from the regression model as expressed in Eq. 2.1. This involves comparing the inequality observed in the outcome variable to the inequality attributed to circumstances. The measure of IOp can be quantified by:
$$\theta_{r} = I(\hat{y})/I(y)$$
(2.2)
where \({\theta }_{r}\) is IOp, I(ŷ) denotes the inequality measure applied to the predicted outcomes based on circumstances, and I(y) represents the inequality measure of the actual outcomes. The choice of inequality measure—such as the dissimilarity index (Barros et al., 2009), mean logarithmic deviation (Ferreira & Gignoux, 2011), or variance (Ferreira & Gignoux, 2014)—depends on the nature of the outcome variable. For instance, binary outcomes may require different inequality metrics compared to continuous outcomes.
Decomposition of IOp Measure: To understand the extent to which various circumstances contribute to overall IOp, decomposition methods are employed. One important decomposition method is the Shapley value, derived from cooperative game theory, which offers a way to quantify the contribution of each circumstance to the total inequality observed. The Shapley value method involves two steps: the first step is estimating inequality measures by computing the inequality measure for all possible permutations of circumstance variables (Shapley, 1953). The second step is computing marginal effects by calculating the average marginal effect of each circumstance on the IOp measure. The Shapley decomposition is computationally intensive, requiring 2K permutations where K represents the number of circumstances. Its advantages include being order-independent and ensuring that the sum of individual contributions equals the total value of IOp.
Machine Learning Algorithm—Conditional Inference Trees: Traditional regression methods face limitations, such as the challenge of selecting which variables to include in the model. Omitting relevant variables can lead to downward bias while including too many can introduce upward bias (Brunori et al., 2019a; 2019b, Hufe et al., 2017). Machine Learning (ML) algorithms address these issues by learning from data and making decisions based on patterns identified during the training phase. ML methods, including conditional inference regression trees, are gaining popularity for their ability to mitigate biases associated with model selection and to provide standardized approaches for estimating IOp. These methods offer several advantages such as reduction of bias: ML algorithms minimize arbitrary choices in model specification and address issues of upward and downward biases in IOp estimation, and improved interpretability: Conditional inference trees, a type of supervised ML method, provide clear graphical representations of opportunity structures, enhancing the comprehensibility of IOp analysis. Conditional inference trees improve upon traditional methods by using recursive binary splitting to identify relevant splits in the data based on circumstance factors. The algorithm operates through two stages: first step is initial splitting: Performs a hypothesis test (e.g. t-test) for each circumstance to determine if there is a significant split, while second step is tree growth: If the hypothesis test suggests a significant split, the algorithm uses this circumstance to partition the data and grows the tree accordingly. As if the p-value is greater than the pre-specified significance level or alpha level, no split is made. Otherwise, the selected circumstance becomes the splitting variable, and the tree is expanded accordingly, producing a visually interpretable hierarchical opportunity structure.
This approach generates opportunity trees that visually represent how different circumstances contribute to inequality. It allows for an in-depth, hierarchical analysis of opportunities, making the results accessible to a broader audience (Brunori & Neidhofer, 2020; Brunori et al., 2018a, 2018b; Lefranc & Kundu, 2020).
2.4 Data Sources and Variables
This analysis utilizes household-level survey data from several key sources: the Employment and Unemployment Surveys (EUS) for the years 2004–05 and 2011–12, and the Periodic Labour Force Survey (PLFS) for the years 2019–20 and 2022–23. These surveys are conducted by the National Statistical Office (NSO), Government of India. The cross-sectional data from these surveys are representative at both the national and state levels. The outcome variables used in this analysis include household consumption expenditure, total household income, monthly wage income (both regular and casual), regular salaried/wage income, self-employed income, and casual wage income. It is important to note that data on the earnings of self-employed individuals is not available for the years 2004–05 and 2011–12. Additionally, the weekly wages or earnings of casual and regular workers have been converted into monthly earnings for consistency in the analysis.
The circumstance variables considered in this study include: Parent’s Education: Categories include no education, primary education, secondary education, higher secondary education, and graduate or higher. Parent’s Occupation: Classified as high-skilled, medium-skilled, low-skilled, and unskilled. Social Group: Divided into scheduled castes, scheduled tribes, other backward classes, and others. Gender: Male or female. Place of Birth: Regions include north, east, central, northeast, south, and west, and Location: Rural or urban. The analysis focuses on individuals within the working-age range of 15–64 years. The final sample sizes, after excluding cases with missing parental background information,1 are as follows: 149,909 individuals in 2004–05; 112,103 individuals in 2011–12; 105,020 individuals in 2019–20; and 75,708 individuals in 2022–23. These samples represent approximately one-third of the total sample covered in each respective survey year (34% in 2022–23, 35% in 2019–20, 33% in 2011–12, and 33% in 2004–05).
2.5 Findings and Discussions
2.5.1 Characteristics of the Sample Population
This section examines the profile of the working-age population (15–64 years) based on various factors, including location, region, social group, and parental background, and their relation to outcome variables such as consumption expenditure, total income, wage income, regular salaried income, casual labour income, and self-employed income. Detailed procedures for sample and variable selection are outlined in Appendix 1.
Geographic Distribution and Demographic Characteristics: A significant portion of the sample population having individuals with parental information resides in rural areas, although this proportion has decreased over time. In 2004–05, 73.6% of the population lived in rural India, whereas this figure declined to 62.5% by 2022–23 (Table 2.1). This shift reflects broader socio-economic changes and urbanization trends in India. The distribution across regions shows that the central region has the highest proportion of the sample population (21.6% in 2022–23), followed by the north, east, south, northeast, and west regions (Table 2.1). This regional distribution highlights the demographic diversity within the country and the varying economic conditions across regions. In terms of gender, the sample is predominantly male, with males constituting 66.1% of the population in 2022–23, reflecting the tendency to record the household head, who is often male, in these surveys. The representation of women has increased over time, but males remain the majority. Regarding social groups, Other Backward Classes (OBC) represent the largest share (41.4%), followed by the general caste (24.3%), Scheduled Castes (SC) (18.5%), and Scheduled Tribes (ST) (15.9%) (Table 2.1). This distribution indicates the significant presence of historically disadvantaged groups in the sample population.
Table 2.1
Profile of sample population (in %)
2004–05
2011–12
2019–20
2022–23
Sector
Rural
73.6
70.2
68.5
62.5
Urban
26.4
29.8
31.5
37.5
Total
100.0
100.0
100.0
100.0
Region (zone)
North
12.3
13.2
14.5
17.8
East
20.3
20.7
19.5
17.3
Central
24.0
24.5
26.2
21.6
North East
3.7
3.5
3.9
14.5
South
24.2
22.8
20.8
17.2
West
15.4
15.3
15.1
11.5
Total
100.0
100.0
100.0
100.0
Gender
Male
75.8
73.9
71.7
66.1
Female
24.2
26.1
28.3
33.9
Total
100.0
100.0
100.0
100.0
Social group
ST
7.9
8.0
8.5
15.9
SC
19.6
19.1
20.5
18.5
OBC
40.8
43.4
43.5
41.4
Others (general)
31.8
29.5
27.5
24.3
Total
100.0
100.0
100.0
100.0
Source Authors calculations from EUS, 2004–05 and 2011–12, and PLFS, 2019–20 and 2022–23
Parental Education and Occupation Profile: The educational attainment of parents has seen notable changes over the years. In 2022–23, more than three-quarters of parents (75.3%) had education levels below secondary, including those who were illiterate (26.3%) (Table 2.2). However, there has been an increase in the proportion of parents with higher secondary education (from 11.7% in 2011–12 to 17.4% in 2022–23) and those with graduate-level education (from 4.5% in 2011–12 to 7.3% in 2022–23). This trend reflects improvements in educational attainment over the past two decades. The occupational profile of parents reveals a predominance of low-skilled and unskilled jobs. The share of parents in low-skilled jobs increased marginally from 67% in 2004–05 to 68.7% in 2022–23, while the share of those in unskilled manual jobs declined from 27.9% to 23.7% during the same period. Conversely, the proportion of parents in high-skilled jobs has consistently increased, rising from 2.1% in 2004–05 to 5.4% in 2022–23, whereas their share in medium-skilled jobs has declined marginally over the same period. However, the increasing share of high-skilled jobs and the declining share of unskilled manual jobs indicate a growing demand for high-skilled positions and a decreasing demand for unskilled ones in the economy. This trend also highlights a growing disparity in job quality and skill levels among parents, potentially affecting the socio-economic outcomes for their children.
Table 2.2
Parents’ education and occupation profile (in percentage)
2004–05
2011–12
2019–20
2022–23
Education
No education
49.3
42.0
37.3
26.3
Below secondary
34.5
37.3
38.3
49.0
Secondary/higher secondary
11.7
14.7
17.2
17.4
Graduate and above
4.5
6.0
7.2
7.3
Total
100.0
100.0
100.0
100.0
Occupation
High skilled
2.1
3.8
4.6
5.4
Medium skilled
3.0
3.4
4.1
2.2
Low skilled
67.0
62.9
66.3
68.7
Unskilled
27.9
29.9
25.0
23.7
Total
100.0
100.0
100.0
100.0
Note ‘High Skilled’ includes, Professionals; ‘Medium Skilled’ includes Associate Professionals; ‘Low Skilled’ includes Clerks, Service Workers, Skilled Agricultural Workers, Craft workers, and Plant and Machinery Operators; ‘Unskilled’ includes workers in elementary occupations
Source Authors calculations from EUS, 2004–05 and 2011–12, and PLFS, 2019–20, and 2022–23
Household Monthly Consumption Expenditure (MPCE): In 2022–23, the average Monthly Consumption Expenditure (MPCE) for households was INR 12,646, with a marginal disparity between rural and urban areas. Rural households reported an average MPCE at INR 11,439, while urban households average MPCE stands at INR 15,939 on average (Table 2.3). The real value of average MPCE has fluctuated over time. It increased from INR 11,391 in 2004–05 to INR 13,173 in 2011–12, before declining to INR 10,652 in 2019–20, and increased to 12,646 in 2022–23. This decrease in average MPCE in 2019–20 may be attributed to the economic impact of the COVID-19 pandemic, which included a nationwide lockdown for several months in 2020 aimed at curbing the spread of the virus, the increase in average MPCE can be attributed to the post pandemic recovery. The lockdown likely led to reduced consumer spending due to restricted mobility and economic uncertainty. Conversely, the average monthly income of households in 2022–23 was, at INR 12,135, marginally lower the average monthly consumption expenditure.
Table 2.3
Average monthly consumer expenditure and income in real value2 (in INR) in 2004–05, 2011–12, and 2019–20
Rural
Urban
Total
Average monthly consumer expenditure
2022–23
11,493
15,839
12,646
2019–20
8839
14,587
10,652
2011–12
10,758
18,864
13,173
2004–05
9863
15,662
11,391
Average monthly income (Self-employed + Regular salaried + Casual labour)
2022–23
10,018
17,647
12,135
2019–20
13,812
24,078
17,049
Source Authors calculations from EUS, 2004–05 and 2011–12, and PLFS, 2019–20, PLFS, 2022–23
This disparity indicates that households had a lower income relative to their expenditure during 2004–05 to 2022–23. The fall in incomes is mainly attributed to a decline in income in regular and casual work (Table 2.3). Higher consumption compared to income also could be due to the various welfare schemes, such as free rations, launched by the government to absorb the economic shocks of the pandemic.
Employment Status and Trends: In 2022–23, the employment status among the sample with parental information in the working-age population reveals that approximately 51.4% were engaged in self-employment, 28.9% in regular salaried jobs, and 19.7% in casual labour (Table 2.4). Over time, there has been a significant shift in employment patterns: The proportion of individuals in regular salaried jobs has increased substantially from 13.0% in 2004–05 to 28.9% in 2022–23. Conversely, the share of those engaged in casual labour has decreased from 32.1% in 2011–12 to 19.7% in 2022–23. The percentage of individuals in self-employment has also declined, from 54.9% in 2011–12 to 51.4% in 2022–23. These shifts indicate a growing trend towards regular employment, though self-employment remains a significant component of the labour market.
Table 2.4
Status of employment of working sample (in percentage) in 2004–05, 2011–12, and 2022–23
Status of employment
2004–05
2011–12
2019–20
2022–23
Self-employed
54.9
50.2
49.2
51.4
Regular
13.0
18.2
29.4
28.9
Casual labour
32.1
31.6
21.4
19.7
Total
100.0
100.0
100.0
100.0
Source Authors calculations from EUS, 2004–05 and 2011–12, and PLFS, 2019–20 and 2022–23
Income Analysis by Status of Employment: In 2022–23, the average monthly income varied significantly by type of employment among the sample of individuals having parental information. Regular salaried jobs offered the highest average income at INR 14,431, followed by self-employment at INR 12,876, and casual labour at INR 5020 (Table 2.5). Over the years, there have been notable changes in average monthly income: For regular salaried jobs, income increased from INR 11,407 in 2004–05 to INR 15,801 in 2011–12, but then declined slightly to INR 15,505 in 2019–20 and further to INR 14,431 in 2022–23. This decline, particularly in the urban sector and for males, may be linked to the economic disruptions caused by the COVID-19 lockdown. Casual labour income showed a substantial increase from INR 3651 in 2004–05 to INR 6106 in 2011–12, and INR 7392 in 2019–20, but again decreased to INR 5020 in 2022–23. The fluctuations in casual labour income highlight the vulnerability of this sector to economic instability and policy changes. Overall, these variations highlight the impact of economic conditions on different types of employment, with significant implications for household income stability and consumption patterns.
Table 2.5
Average monthly income in real value3 (in INR) by status of employment in 2004–05, 2011–12, and 2022–23
Location
Gender
Total
Rural
Urban
Male
Female
Average monthly self-employed (SE) income
2022–23
11,735
15,413
13,953
5799
12,876
2019–20
10,214
17,312
12,475
6055
12,133
Average monthly regular salaried (RE) income
2022–23
13,441
15,322
14,607
13,482
14,431
2019–20
12,849
17,729
15,756
13,940
15,505
2011–12
11,500
18,595
16,283
12,375
15,801
2004–05
8979
13,216
11,842
8373
11,407
Average monthly casual labour (CL) income
2022–23
4865
5613
5133
3087
5020
2019–20
7155
8549
7581
4865
7392
2011–12
5926
7062
6291
3987
6106
2004–05
3504
4586
3842
2291
3651
Source Authors calculations from EUS, 2004–05 and 2011–12, and PLFS, 2019–20, and 2022–23
2.5.2 Trends of Inequality of Opportunity in India
Overall Inequality and IOp: The trends in inequality and IOp across different outcomes such as consumption, total labour earnings (including self-employed, regular, and casual work), and wage earnings (both regular and casual work) from 2004–05 to 2022–23 are provided in Table 2.6. Inequality is measured using the Mean Log Deviation (MLD)4 and Relative IOp, with bootstrap standard errors indicating the precision of the estimates.
Table 2.6
Inequality of opportunity in consumer expenditure, income and wages
Consumption
Income (labour earnings)
Wages earnings (regular and casual work)
2004–05
2011–12
2019–20
2022–23
2019–20
2022–23
2004–05
2011–12
2019–20
2022–23
Inequality (MLD)
0.172
0.191
0.160
0.142
0.307
0.260
0.465
0.394
0.291
0.190
Relative IOp
0.236
0.235
0.222
0.262
0.241
0.278
0.184
0.272
0.265
0.282
(0.007)
(0.010)
(0.014)
(0.015)
(0.007)
(0.009)
(0.013)
(0.015)
(0.019)
(0.017)
Note Figures in parentheses are Bootstrap standard error
Source Authors calculation from EUS, 2004–05 and 2011–12, and PLFS, 2019–20 and 2022–23
Consumption Inequality and Relative IOp: The MLD shows that consumption inequality increased from 0.172 in 2004–05 to 0.191 in 2011–12, indicating a rise in inequality during this period. However, it decreased to 0.160 in 2019–20 and further to 0.142 in 2022–23, showing a reduction in consumption inequality in recent years. The Relative IOp remained relatively stable from 2004–05 to 2011–12, slightly decreased in 2019–20, but increased to 0.262 in 2022–23. This indicates that while overall consumption inequality has decreased, IOp in consumption has risen sharply in the most recent period. This suggests that even though consumption levels are becoming more equal, the opportunities leading to those levels are becoming more unequal.
Total Labor Earnings (Income) Inequality and IOp: The MLD for total labour earnings decreased from 0.307 in 2019–20 to 0.260 in 2022–23, indicating a reduction in income inequality over this period. The Relative IOp for total labour earnings increased from 0.241 in 2019–20 to 0.278 in 2022–23. This suggests a rise in IOp in labour earnings, indicating emerging disparities in the opportunities to earn income even though overall income inequality has decreased.
Wage Earnings (Regular and Casual Work) Inequality and IOp: The MLD for wage earnings shows a consistent decline in inequality from 0.465 in 2004–05 to 0.190 in 2022–23, indicating a significant reduction in wage inequality over the years. The Relative IOp for wage earnings increased from 0.184 in 2004–05 to 0.272 in 2011–12, suggesting a rise in IOp during this period. Although it slightly declined to 0.265 in 2019–20, it increased again to 0.282 in 2022–23. This indicates that while wage inequality has substantially declined, the IOp in wages has fluctuated but generally increased, suggesting that factors beyond individual control (such as family background and education) are playing a larger role in determining wage outcomes.
These results align with earlier findings, which also show that relative IOp is higher for wages and total wage earnings compared to consumption (Lefranc & Kundu, 2020; Singh, 2012a; 2012b). Over the past two decades, overall inequality has decreased. However, the persistent and increasing IOp highlights the need for policies that address the underlying factors contributing to these disparities. This includes improving access to quality education, health care, and other essential services to ensure a more level playing field for all individuals, regardless of their circumstances. These issues are explored in detail in the subsequent chapters of the book.
Overall Inequality and IOp by Status of Employment: Further, Table 2.7 presents data on Inequality and IOp in income and wages, by employment status—regular salaried jobs, self-employment, and casual labour from 2004–05 to 2022–23. The data shows that income inequality, as measured by MLD, is highest among regular salaried employees, followed by those who are self-employed, with the lowest among casual workers. This pattern is consistent across the different time periods analysed. On the other hand, the Relative IOp is higher for regular salaried jobs compared to self-employed and casual workers.
Table 2.7
Inequality of opportunity in income/wages by employment status
Regular salaried
Self- employed)
Casual worker
2004–05
2011–12
2019–20
2022–23
2019–20
2022–23
2004–05
2011–12
2019–20
2022–23
Inequality (MLD)
0.500
0.457
0.306
0.208
0.269
0.239
0.186
0.155
0.117
0.113
Relative IOp
0.163
0.238
0.242
0.229
0.179
0.194
0.153
0.173
0.162
0.208
(0.012)
(0.018)
(0.024)
(0.019)
(0.017)
(0.011)
(0.009)
(0.013)
(0.014)
(0.0100)
Note Figures in parentheses are Bootstrap standard error
Source Authors calculations from EUS, 2004–05 and 2011–12, and PLFS, 2019–20, and 2022–23
Regular Salaried Jobs: MLD for regular salaried jobs indicate inequality decreased from 0.500 in 2004–05 to 0.208 in 2022–23. This shows a significant reduction in income inequality for regular salaried jobs over this period. The Relative IOp increased from 0.163 in 2004–05 to 0.242 in 2019–20, and slightly decreased to 0.229 in 2022–23. This indicates that the IOp has increased for regular salaried jobs over the years, suggesting that income inequality for regular workers is influenced relatively more by differences in opportunities rather than just income levels.
Self-Employed: MLD for self-employed individuals decreased from 0.269 in 2019–20 to 0.239 in 2022–23. This indicates a modest decline in income inequality among the self-employed. The Relative IOp for self-employed individuals was relatively stable, ranging from 0.179 in 2019–20 to 0.194 in 2022–23. This suggests a moderate level of IOp among self-employed that did not change drastically over time.
Casual Labour: MLD for casual workers remained relatively stable, decreasing slightly from 0.186 in 2004–05 to 0.113 in 2022–23. Casual workers have consistently experienced lower levels of income inequality compared to regular salaried and self-employed individuals. The Relative IOp for casual workers increased slightly from 0.153 in 2004–05 to 0.208 in 2022–23. Despite the small increase, casual workers generally experienced lower levels of IOp compared to those in regular salaried and self-employed roles.
The analysis reveal that increase in Relative IOp among regular salaried jobs suggests that most of the rise in income inequality can be attributed to growing disparities in opportunities within this employment category. This indicates that addressing IOp in regular salaried jobs could be crucial for reducing overall income inequality. In contrast, self-employment and casual workers has had more stable opportunities, but the overall lower inequality in these categories suggests fewer barriers to equitable access in this type of employment.
2.5.3 Contribution of the Factors
To relative contribution of various circumstances to IOp has been calculated by using Shapley decomposition (Fig. 2.1). The results indicate that parental background is the most significant factor contributing to IOp in consumption (MPCE), accounting for 47.9%. This includes parental education (31.4%) and occupation (16.5%). The region contributes 23.1% to consumption IOp. For total labour earnings (MPCI) IOp, parental background remains the most significant factor, contributing 58.1%, followed by region at 19.9%. These findings highlight the substantial impact of familial and regional circumstances on economic outcomes, emphasizing the need for policies targeting these fundamental sources of inequality.
Fig. 2.1
Shapley decomposition of IOp in 2022–23.
Source Authors calculations from EUS, 2004–05 and 2011–12, and PLFS, 2019–20, and 2022–23
Further, the contributing factors by status of employment reveal that parental background also significantly influences IOp in various forms of employment. For regular salaried workers, parental education and occupation combined contribute approximately 61.5% to their income IOp. This substantial impact arises because the likelihood of securing a regular salaried position is heavily influenced by an individual’s educational attainment, which in turn is shaped by their parents’ educational levels and occupational status (Das & Biswas, 2022). In essence, parents’ educational and occupational backgrounds play a crucial role in determining access to better-paying, stable employment, highlighting a key area where inequality of opportunity is perpetuated across generations. Conversely, regional differences emerge as the primary factor contributing to income IOp in casual wage employment, accounting for 68.7%.
This suggests that geographic location and regional economic conditions are pivotal in determining the opportunities available to casual workers. Additionally, gender disparities play a significant role in self-employment, where gender accounts for 37.5% of income IOp, and regional factors contribute 29.6%. Social identity, including caste or social group, contributes to income IOp as well, though its impact is more pronounced in regular salaried positions (10.4%) compared to casual (7.4%) and self-employment (5.4%) (Das & Biswas, 2022). These findings underscore that while parental background is a major factor in regular employment, gender and regional factors are more influential in self-employment and casual wage work, respectively.
2.5.4 Conditional Inference Regression Tree
Conditional inference regression trees provide a nuanced understanding of how various circumstances affect IOp in consumption expenditure and income. The analysis based on MLD reveals that tree-based methods yield estimates for consumption IOp at 0.240 and income IOp at 0.282 that are slightly higher but closely aligned with parametric estimates. The conditional inference regression tree visually represents the types and conditions that most significantly impact IOp.
Conditional Inference Regression Tree for Consumption Expenditure: The conditional inference regression tree for consumption expenditure (MPCE), as shown in Fig. 2.2, identifies parental education as the most critical factor or circumstances influencing consumption IOp. For individuals whose parents have attained secondary or higher levels of education, the sector (rural or urban) emerges as the second most significant factor, followed by parental occupation for urban residents and geographical region for rural residents. Specifically, individuals with parents in high or medium-skilled occupations in urban areas, particularly in the central and northeastern regions, exhibit the highest outcome value (MPCE), which is followed by those belonging to the northern, southern, and western regions of India. This pattern indicates that individuals with parents with secondary or higher education, residing in urban areas, and engaged in skilled jobs, experience the highest MPCE values and the lowest levels of consumption IOp.
Fig. 2.2
Monthly consumption expenditure tree. Note R: Rural; U: Urban; N: North; NE: North East; S: South; W: West E: East; C: Central; Sec/HS: Secondary/Higher Secondary; GradAbv: Graduate and Above; NoEdu: Illiterate or Nor Formal Schooling; BS: Below Secondary; US: Unskilled; Low: Low Skilled; Med: Medium Skill; High: High Skilled; M: Male; F: Female.
Source Authors calculations from PLFS, 2022–23
Conversely, for individuals whose parents have less than secondary education or no formal schooling, the region becomes the second most important factor influencing consumption IOp. In these cases, the third most significant factor is sector (rural or urban) for individuals from NNESW (northern, northeastern, southern, and western) regions, and social group (caste) for those from the eastern and central regions. Further, the analysis highlights that individuals residing in rural parts of NNESW regions have the lowest MPCE values, indicating the highest levels of consumption IOp.
Conditional Inference Regression Tree for Income (Total Labour Earnings): The conditional inference regression tree for labour income or earnings, presented in Fig. 2.3, reveals that parental education is the most significant factor determining income IOp. For individuals with parents who have a graduate-level education or higher, parental occupation becomes the second most important factor. In contrast, for those with less educated parents, the region becomes the second significant factor influencing income IOp. For individuals whose parents have graduate or higher education and are engaged in high or medium-skilled jobs, geographical region is the next crucial factor, followed by the sector (rural or urban). Further, individuals from the eastern, central, and northeastern regions who reside in urban areas exhibit the highest labour income values or the lowest income IOp, followed by those from northern, southern, and western regions.
Fig. 2.3
Conditional inference regression tree for HH total labour income. Note R: Rural; U: Urban; N: North; NE: North East; S: South; W: West: E: East; C: Central; Sec/HS: Secondary/Higher Secondary; GradAbv: Graduate and Above; NoEdu: Illiterate or Nor Formal Schooling; BS: Below Secondary; US: Unskilled; Low: Low Skilled; Med: Medium Skill; High: High Skilled; M: Male; F: Female.
Source Authors calculations from PLFS, 2022–23
Conversely, for individuals with parents who have less than a graduate-level education, the sector is the third most important determinant of income IOp across all regions. In rural areas, social group (caste) is the next critical factor, particularly in the eastern and central regions, where those belonging to Scheduled Castes (SC) and Scheduled Tribes (ST) have the lowest total labour income values or highest income IOp. In urban areas, parental education remains a significant factor. In the NNESW (north, northeast, south, and west) regions, geographical region for individuals residing in rural areas, and parental occupation for those residing in urban areas emerge as key factors influencing labour income IOp. Among these groups, individuals with parents who have below secondary education or no formal schooling, for those residing in rural parts of the NNESW regions, experience the lowest labour income values or the highest income IOp.
2.5.5 Regional Analysis
Consumption IOp: A regional analysis of consumption IOp reveals significant differences across Indian states. Chhattisgarh has the highest consumption IOp (42%), while Bihar has the lowest (14%) (Map 2.1 and Table 2.8). In the Central region, Chhattisgarh and Madhya Pradesh (31%) have high IOp, while Uttar Pradesh (18%) and Uttarakhand (28%) have lower levels. In the Northern region, Delhi (33%) and Punjab (32%) have higher IOp compared to Rajasthan, Haryana, and Himachal Pradesh (19–25%), with Jammu and Kashmir having the lowest (19%). The high IOp in Delhi and Punjab is due to diverse economies and significant urban populations, which may lead to greater disparities in consumption opportunities (Bourguignon et al., 2007). In the Western region, Gujarat (29%) and Maharashtra (27%) show relatively higher IOp. Among the Southern states, Telangana (30%) has the highest IOp, while Kerala has the lowest (16%), with Andhra Pradesh, Tamil Nadu (22.8%), and Karnataka ranging between 23 and 26%. The progressive social policies in Kerala may contribute to lower IOp by promoting a more equal distribution of resources and opportunities (World Bank, 2021).
Map 2.1
Regional consumption IOp (MLD).
Source Authors calculations from PLFS, 2022–23
However, it is surprising that the relatively less developed states in the Eastern region, Odisha and Jharkhand indicate high IOp, while Bihar and also Uttar Pradesh in the Central region show low IOp. The reasons could be that states like Odisha, Jharkhand, and Chhattisgarh are rich in natural resources, which can lead to unequal wealth distribution and high IOp. The mining industry often benefits a small elite, leaving the majority with limited economic opportunities (Singh, 2012). These states have underdeveloped infrastructure and limited access to education and health care, exacerbating inequality. The lack of social services and infrastructure perpetuates disparities in consumption opportunities (World Bank, 2016). On the other hand, despite overall poverty, the more uniform socio-economic conditions in Uttar Pradesh and Bihar may result in lower consumption IOp. The widespread poverty reduces the gaps in consumption opportunities (Singh & Mohapatra, 2019).
Income IOp: A regional analysis of income IOp shows significant variations across Indian states.5 Jharkhand has the highest income IOp at 43%, while Himachal Pradesh has the lowest at 16% (Map 2.2 and Table 2.9). High levels of income IOp are particularly seen in the eastern (Jharkhand and Odisha), northeastern (Assam), and central (Madhya Pradesh and Chhattisgarh) regions of India. In the eastern region, Jharkhand tops the list with an income IOp of 43%, followed by Odisha at 36%. West Bengal has an income IOp of 26%, while Bihar has the lowest among the eastern states at 20%. The high income IOp in Jharkhand and Odisha is largely due to their rich natural resources, such as coal and minerals. These resources often lead to income disparities as the wealth tends to benefit a small, elite group, leaving the majority with fewer economic opportunities (Deaton & Dreze, 2002). In central India, Chhattisgarh leads with an income IOp of 36%, followed by Madhya Pradesh at 33%. Uttarakhand and Uttar Pradesh have lower income IOp at 21% and 18%, respectively. The high IOp in Chhattisgarh and Madhya Pradesh is also attributed to resource concentration issues, similar to those in Jharkhand and Odisha (World Bank, 2016).
Map 2.2
Regional income IOp (MLD).
Source Authors calculations from PLFS, 2022–23
In Northern India, Delhi and Punjab have high income IOp of 34% and 32%, respectively. This is due to their diverse economies and large urban populations, which contribute to greater income disparities (Singh, 2012). Haryana and Jammu and Kashmir also have high income IOp at 29% and 27%. Rajasthan has a lower income IOp of 24%, and Himachal Pradesh has the lowest in the north at 16%. In Western India, Gujarat has an income IOp of 26%, and Maharashtra has 23%. These states show relatively high income IOp due to their advanced economic development and industrialization. In Southern India, Karnataka and Telangana have higher income IOp at 28% and 26%, respectively, followed by Andhra Pradesh at 22%. Tamil Nadu and Kerala have lower income IOp at 19% and 18%, respectively. Kerala’s lower income IOp is attributed to its progressive social policies and better distribution of resources (World Bank, 2021). Overall, the patterns of income IOp across states generally mirror those of consumption IOp, reflecting similar underlying regional disparities (Das & Biswas, 2022).
2.6 Summary and Conclusion
The analysis reveals that unequal circumstances account for a substantial portion of inequality in consumption and income in India. Specifically, about 26.2% of consumption inequality, 27.8% of income inequality, and 28.2% of wage earnings inequality can be attributed to these unequal circumstances. Similarly, using the conditional inference tree, around 24% of consumption inequality and 28% of income inequality can be attributed to circumstances. Despite a significant decline in overall inequality in consumption, income and wages in recent years, the relative contribution of IOp has increased significantly. This suggests that while the outcomes have improved, the disparities in opportunities themselves have persisted or even widened in recent years.
The findings highlight the need for a policy shift towards addressing IOp rather than focusing solely on outcome-based metrics. This shift is crucial given that IOp in regular salaried employment is notably higher and has increased significantly over the past two decades. Conversely, IOp in casual wage employment and self-employment has almost stable or decline, reflecting differing dynamics across employment types. Machine learning (ML) algorithms have provided deeper insights into the factors contributing to IOp, revealing that parental background—specifically, education and occupation—plays a pivotal role in determining unequal opportunities in labour income as well as for regular salaried jobs. Gender disparities are more pronounced in casual wage employment and self-employment.
Regional analyses of consumption and income IOp reveal significant differences across Indian states. Chhattisgarh and Jharkhand have the highest consumption IOp, while Kerala, Himachal Pradesh, and also Bihar has the lowest IOp. In both cases, states with rich natural resources, like Jharkhand and Odisha, exhibit higher IOp due to uneven wealth benefits, while some underdeveloped states like Uttar Pradesh and Bihar show lower IOp attributed to high poverty with low consumption and income. On the other hand, the developed states such as Kerala and Tamil Nadua have lower IOp, while Maharashtra and Gujarat show higher IOp. These contradictory results indicate uneven distribution of resources in industrialised states, and progressive social policy and better resource distribution in other developed states.
To mitigate these deep-rooted inequalities, policies must be customized to meet the unique needs of different geographical regions and groups. It is important to focus on improving access to quality education for children from disadvantaged backgrounds, as equitable educational opportunities can significantly reduce IOp and lead to better overall outcomes. Additionally, regional policies should aim to improve infrastructure and implement progressive social measures to ensure fair distribution of resources and result in lower income, and consumption IOp.
Notes
1.
“Types” is defined as a group of individuals sharing the same circumstances.
2.
The place of birth representing by States/UTs of India are divided into broad five zone or regions as North: Jammu & Kashmir, Himachal Pradesh, Punjab, Haryana and Uttarakhand; East:Bihar, Jharkhand, Orissa, West Bengal; Central: Uttar Pradesh, Rajasthan, Madhya Pradesh, Chattisgarh; North East: Sikkim, Arunachal Pradesh, Assam, Nagaland, Meghalaya, Manipur, Mizoram, Tripura—Northeast; South: Karnataka, Andhra Pradesh, Tamilnadu, Pondichery, Kerala, Lakshadeep; and West: Gujrat, Daman & Diu, Dadra & Nagar Haveli, Maharashtra, Goa.
3.
The consumer expenditure is converted from nominal to real values by using Consumer Price Index (CPI), CPI-agriculture labour for rural areas, and CPI-industrial workers for urban areas.
4.
The self-employed income, regular salaried income, and casual labour income expenditure are converted from nominal to real values by using Consumer Price Index (CPI), CPI-agriculture labour for rural areas, and CPI-industrial workers for urban areas.
5.
The bootstrap standard errors are based on 100 replication and nearly zero, which suggest the robustness of the estimate.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Appendix 1: Sample Selection and Variable Construction
Variable Selection
From the 2018–19 PLFS survey, six key variables were selected for analysis. These include three variables used in their original form—sector, caste, and gender—and three variables that were modified or created—states, parents’ education, and parents’ occupations.
Sector is categorized into two groups: rural and urban.
Gender is classified as male and female; the transgender category was excluded from this analysis.
Caste is divided into four categories: Scheduled Caste (SC), Scheduled Tribe (ST), and Other Backward Classes (OBC), and General.
For the state variable, which originally covered 36 states and union territories in India, we have consolidated these into six broad geographical regions:
North: Jammu & Kashmir, Himachal Pradesh, Punjab, Haryana, and Rajsthan
East: Bihar, Jharkhand, Odisha, and West Bengal
Central: Uttar Pradesh, Uttarakhand, Madhya Pradesh, and Chhattisgarh
The education variable is categorized into four broad groups:
Illiterate or No Education: Code 1 (Illiterate)
Below Secondary: Codes 2–7 (Literate to up to middle school)
Secondary and Higher Secondary: Codes 8–10 (Secondary to higher secondary)
Graduate and Above: Codes 12–13 (Graduate and post-graduate)
The occupation/skill level variable is classified into four categories based on the National Classification of Occupations (NCO) at one digit, as per the OECD Employment Outlook 2014 and the NCO 2015 from the Ministry of Labour and Employment, Government of India:
Unskilled or Routine Manual Tasks: Involves simple and routine physical or manual tasks (NCO code 9—Elementary Occupations), such as domestic helpers, cleaners, street vendors, and garbage collectors.
Low Skilled or Non-Routine Manual Tasks: Includes tasks like operating machinery, driving vehicles, maintenance, and repair (NCO codes 4–8), such as clerical jobs, service workers, sales workers, and craft and trade workers.
Medium Skill or Non-Routine Cognitive Tasks: Involves complex technical and practical tasks requiring extensive knowledge in a specialized field (NCO code 3—Professional and Technical Associates).
High Skill or Cognitive Tasks: Requires complex problem-solving, decision-making, and creativity based on substantial theoretical and factual knowledge (NCO code 2—Professionals and Technicians).
For NCO code 1 (e.g. legislators, managers), skill levels are not applied due to the wide variation in skills required for these occupations, making classification into the four broad skill levels impractical.
Sample Selection
The sample selection process follows a multi-stage procedure:
Identification of Parents6: Initially, each respondent's parents were identified using the “relation to head” variable. For individuals classified as “self” (code 1), the household member labelled as “Father/Mother/Father-in-Law/Mother-in-Law” (code 7) was considered the parent. This generated the first set of data, linking children with their parents.
Categorization of Children: In the second stage, individuals were categorized as unmarried children (code 5) or married children (code 3). The parents of these children, identified as household heads labelled “self” (code 1), were then used to create the second set of data.
Handling Duplicates: Duplicate records were carefully reviewed and removed. For instance, if records for both males and females existed for the same individual, the duplicates were deleted to ensure accuracy. After cleaning both data files, they were merged based on key variables as described above.
This systematic approach ensures a comprehensive and accurate dataset for analysing inequality and related factors.
Appendix 2: Consumption and Income IOp Across States
Whenever available, information for parents (both father and mother) has been used; however, in few cases, data is available for only either father or mother.
The mean-log deviation (MLD) is chosen as the measure of inequality because it uniquely satisfies all six of the following properties: (1) anonymity or symmetry, (2) population replication or replication invariance, (3) mean independence or scale invariance, (4) the Pigou-Dalton principle of transfers, (5) additive subgroup decomposability, and (6) path independence.
For the analysis IOp, only major states of India with a sufficient sample size (3,600 or more) have been considered to ensure robust state-level estimates.
Whenever available, information for parents (both father and mother) has been used; however, in few cases, data is available for only either father or mother.