Most development programmes in South Asia measure poverty by income, consumption, land or schooling. Each of those is shaped by caste. Who owns irrigated land in a village, who gets called for a job interview in a city, who cleans a septic tank and who sits on the panchayat are questions with caste answers. A practitioner who treats caste as background culture will misread the evidence in front of them.

The Census of India counts two groups defined by the Constitution: Scheduled Castes (SCs) and Scheduled Tribes (STs). In 2011 they made up a quarter of the population. These are counts of people in notified lists, which is why later sections treat them as legal categories as much as social ones.

You do not need to be a sociologist to work responsibly on caste. You need three questions and the habit of asking them at design, delivery and evaluation. Each one turns an abstract concern into something a programme team can check in the field and record in its monitoring data.

Programmes often reach caste only through poverty targeting: if the poorest are served, caste will take care of itself. The evidence reviewed in Section 7 shows why that fails. Dalit graduates with the same qualifications as upper-caste graduates receive fewer interview calls. SC and ST entrepreneurs own far fewer enterprises than their population share. Caste affects people at every income level.

Language about caste carries history. Several words in common use are slurs, and official categories differ from the names communities choose for themselves. Use the legal category when you mean the legal category, and the self-chosen name when you are describing people and movements.

Discussions of caste often slide between two Sanskrit terms. Keeping them apart removes a great deal of confusion, because most of the inequality a practitioner meets operates through jati, while most popular explanations of caste talk about varna.

In 1916, as a doctoral student at Columbia University, B. R. Ambedkar argued that the defining feature of caste is endogamy: marriage only within the group. Occupation, ritual rank and food rules all vary across regions. The rule that a person marries inside the jati is what reproduces the group, generation after generation, and keeps its boundaries closed.

Sociologists describe caste through a set of linked features. Not every region shows all of them with equal force, and each has weakened or changed form in cities. Together they explain why caste outlasts changes in law and income.

Untouchability is the practice of treating some castes as ritually polluting, so that their touch, shadow, food or water is avoided. It shows up as refused entry to temples and homes, separate vessels in tea shops and exclusion from common wells. The India Human Development Survey asked whether anyone in the family practised it and, if not, whether an SC person could enter the kitchen or use the family's utensils. A household counts as practising if it said yes to the first question or no to the second.

Averages hide a wide spread. In the IHDS-II data, admitted practice of untouchability was highest in the Hindi-speaking north and centre and lowest in West Bengal and Kerala. Practice was also reported within groups that are themselves discriminated against: 15% of SC and 22% of ST respondents admitted to it, which shows how deeply graded hierarchy runs.

Ambedkar described caste as a system of graded inequality: each group is subordinated to those above it and placed above others below it. This gives almost every group a stake in the order, which is one reason caste is hard to dismantle and why solidarity across oppressed castes has been difficult to build.

Urbanisation, schooling, migration and the market have loosened many caste rules. Few people now enforce occupational rules strictly, and inter-dining is common in cities. Yet the features that matter most for inequality, endogamy, unequal land ownership and network-based hiring, have proved durable. Caste also became a basis for political mobilisation and for claims on the state after 1950.

Every caste category in current data has a history. The lists of Scheduled Castes descend from colonial lists of 'depressed classes'. The idea that caste can be ranked on a single national ladder was pushed hard by the colonial census. Knowing this history makes you a more careful user of caste data, and helps explain why communities contest their classification so fiercely.

As Census Commissioner for 1901, Herbert Hope Risley directed that a scheme be drawn up classifying Hindu castes under the four varnas and assigning each a place by social precedence. Risley also used anthropometry, including the nasal index, to argue that caste was based on race. Modern historians treat that racial theory as pseudo-science.

Nicholas Dirks, in Castes of Mind: Colonialism and the Making of Modern India (Princeton University Press, 2001), argues that caste as a single, all-India system is a modern product of the encounter with British rule. Dirks accepts that caste existed before British rule. He argues that census-taking, ethnography and law made caste the master category for describing and governing Indian society.

M. N. Srinivas, working on Coorg in Karnataka, introduced the term sanskritisation in Religion and Society among the Coorgs of South India (1952). In Social Change in Modern India (1966) he defined it as the process by which a low caste, a tribe or another group changes its customs, ritual, ideology and way of life in the direction of a high, frequently twice-born caste, usually followed by a claim to a higher position.

Srinivas also coined the idea of the dominant caste from his fieldwork in Rampura, a village near Mysore. A dominant caste is one that is numerically strong in a locality and holds preponderant economic and political power, above all through land. Srinivas first defined the term in 1955 and set it out fully in 'The Dominant Caste in Rampura', American Anthropologist 61(1), 1959.

No single theory explains caste. Each lens below draws attention to something real and misses something else. Practitioners can use them as a checklist of mechanisms to look for, without having to settle the debates.

In August 1932 the British government's Communal Award offered the 'depressed classes' separate electorates. Gandhi began a fast unto death against it. On 24 September 1932 the Poona Pact, signed by Ambedkar, Gandhi's associates and other Hindu leaders, replaced separate electorates with seats reserved for the depressed classes within a joint electorate.

History is a working tool here. The categories in a 2026 baseline survey come from 1901 enumeration methods, 1950 presidential orders and decades of litigation. Knowing this changes how you design questions, how you code answers and how you interpret change over time.

Jotirao Phule (1827-1890), born into a Mali farming family in Maharashtra, built the first sustained modern movement against caste. In 1848 he and Savitribai Phule began a school for girls from lower castes, with Savitribai teaching. In 1873 he published Gulamgiri (Slavery), which compared the condition of lower castes to slavery in America, and founded the Satyashodhak Samaj (Society of Truth Seekers).

Savitribai Phule taught in the 1848 school and in the schools the couple went on to open, facing fierce resistance from orthodox Brahmins. The Phules also supported widow remarriage and opened a refuge for widows and their children. Their work shows that caste and gender cannot be separated: control over women's marriage and sexuality is how endogamy is enforced.

Bhimrao Ramji Ambedkar trained as an economist at Columbia University and the London School of Economics and as a barrister. He chaired the Drafting Committee of the Constituent Assembly. His writing on caste combines anthropology, economics and law, and it remains the most practical theory of caste for anyone designing policy, because it names mechanisms that can be changed.

In 1935 the Jat-Pat-Todak Mandal of Lahore, a caste Hindu reform group, invited Ambedkar to preside over its 1936 annual conference. When the organisers read his printed address they asked him to cut it, and he refused. The conference was cancelled, and Ambedkar published the speech himself in 1936 as Annihilation of Caste. The first edition of 1,500 copies sold out within two months.

Defenders of caste called it a division of labour. Ambedkar's reply is a piece of labour economics. A division of labour is efficient when people choose occupations by aptitude and can move when conditions change. Caste fixes occupation by birth, so labour cannot move to where it is most productive and talent is wasted.

Ambedkar set out what should replace caste. His ideal was a society based on liberty, equality and fraternity, and he gave fraternity a precise meaning: many shared interests, free contact between groups and a habit of respect. He called this democracy as a way of living together, wider than a form of government.

Phule and Ambedkar belong to a larger tradition. In the south, Narayana Guru in Kerala and E. V. Ramasamy (Periyar) in Tamil Nadu led movements against Brahmin dominance and untouchability. Ambedkar also weighed the bhakti saints and judged them ineffective: in his 1937 reply to Gandhi, printed with the second edition of Annihilation of Caste, he wrote that none of the saints ever attacked the Caste System.

Ambedkar's analysis gives a practitioner four mechanisms to look for, each of which can be observed and each of which a programme can affect. That makes it more useful for design than theories built around purity alone.

Article 17 is one of the few provisions of the Constitution that speaks directly to private conduct. It abolishes untouchability, forbids its practice in any form, and makes enforcing any disability arising from it an offence. Parliament gave it effect through two statutes, examined in the next slides.

Parliament first enacted the Untouchability (Offences) Act 1955 (Act 22 of 1955), notified on 8 May 1955. It was amended and renamed in 1976 as the Protection of Civil Rights Act 1955. The Act extends to the whole of India and is implemented by state governments. Offences under it are cognisable and tried summarily (Section 15).

By the 1980s it was clear that the PCR Act did not reach the violence used to enforce caste. Parliament enacted the Scheduled Castes and the Scheduled Tribes (Prevention of Atrocities) Act 1989 on 11 September 1989; it came into force on 30 January 1990, with Rules in 1995. It defines specific atrocities committed by a person outside the SC and ST communities against SC or ST persons.

The SC and ST (Prevention of Atrocities) Amendment Act 2015 (Act 1 of 2016) came into force on 26 January 2016. It responded to evidence from Dalit and Adivasi organisations that many common forms of humiliation and exclusion fell outside the 1989 list, and that trials took years.

On 20 March 2018, in Subhash Kashinath Mahajan v State of Maharashtra, the Supreme Court required a preliminary inquiry before registering an FIR under the Act and approval before arrest, and allowed anticipatory bail. Dalit organisations protested across the country. Parliament responded with the SC and ST (Prevention of Atrocities) Amendment Act 2018, which inserted Section 18A.

Manual scavenging, the cleaning of human excreta by hand, is the clearest surviving case of caste-assigned work. The Prohibition of Employment as Manual Scavengers and their Rehabilitation Act 2013 (No. 25 of 2013, assent 18 September 2013, in force 6 December 2013) replaced a 1993 Act. It widened the definition to cover open drains, pits and railway tracks, banned hazardous manual cleaning of sewers and septic tanks, and required surveys and rehabilitation.

The government distinguishes manual scavenging (handling excreta from insanitary latrines) from hazardous cleaning (entering sewers and septic tanks without safety gear). By that distinction it reports no deaths from manual scavenging. It does report deaths during sewer and septic tank cleaning, drawn from the National Commission for Safai Karamcharis.

Strong statutes do not enforce themselves. Studies by Dalit rights groups and the statutory commissions repeatedly identify the same weak points between an incident and a conviction. Each is a place where a programme or a legal aid partner can help.

The Constitution creates two commissions to watch over these safeguards. Article 338 establishes the National Commission for Scheduled Castes. Article 338A, inserted by the Constitution (Eighty-ninth Amendment) Act 2003, establishes a separate National Commission for Scheduled Tribes. Each has a chairperson, a vice-chairperson and three members appointed by the President.

Article 15(1) bars the State from discriminating on grounds of caste, and Article 16(2) bars discrimination in public employment. The same articles then permit special provisions. This is the constitutional basis for reservation: equal treatment of unequally placed groups would preserve inequality.

Under Article 341 the President, after consulting the Governor, notifies the castes deemed to be Scheduled Castes in each state or union territory. Only Parliament, by law, can add or remove a group afterwards. Article 342 does the same for Scheduled Tribes. Lists are therefore state-specific: a caste can be SC in one state and not in its neighbour.

For direct recruitment to central government posts by all-India open competition, the Department of Personnel and Training reserves 15% for SCs, 7.5% for STs and 27% for OBCs. A further 10% is reserved for economically weaker sections who are not covered by SC, ST or OBC reservation. States set their own percentages for state posts, reflecting their populations.

The Second Backward Classes Commission, chaired by B. P. Mandal, was set up in 1979. Its 1980 report identified 3,743 castes and communities as backward and estimated that they made up 52% of the population, excluding SCs and STs. It recommended reserving 27% of central government posts for OBCs, which kept total reservation below half. The report was shelved until the V. P. Singh government announced its implementation in August 1990.

On 16 November 1992 a nine-judge bench of the Supreme Court decided the challenge to the Mandal orders by six votes to three. The judgment set the rules that still frame reservation law, and later amendments and cases are best read as responses to it.

The Constitution (One Hundred and Third Amendment) Act 2019 added Articles 15(6) and 16(6), permitting reservation of up to 10% for economically weaker sections who are not SC, ST or OBC. On 7 November 2022, in Janhit Abhiyan v Union of India, a five-judge bench upheld the amendment by three votes to two.

On 1 August 2024 a seven-judge bench held, by six to one, that states may sub-classify the Scheduled Castes for the purpose of reservation, for example by setting aside part of the SC quota for the most under-represented SC groups. The judgment overruled E. V. Chinnaiah v State of Andhra Pradesh (2005), which had treated SCs as one homogeneous class.

Four of the seven judges in Davinder Singh, B. R. Gavai, Vikram Nath, Pankaj Mithal and S. C. Sharma, wrote that the State should evolve a policy to identify a creamy layer among SCs and STs and exclude it from reservation. That view appears in their separate opinions. It is not part of the operative order, which deals with sub-classification.

Article 16(4A) allows reservation in promotion for SCs and STs where they are not adequately represented, with consequential seniority. The Supreme Court has upheld the clause while attaching conditions. The result is a body of case law that governments must follow when they design promotion rosters.

Articles 330 and 332 reserve seats for SCs and STs in the Lok Sabha and state assemblies in proportion to their population. Article 334, as amended, ends this reservation eighty years from the commencement of the Constitution, that is in January 2030, unless Parliament amends it again. Since the 2008 delimitation, 84 of 543 Lok Sabha seats are reserved for SCs and 47 for STs.

The Constitution (One Hundred and Second Amendment) Act 2018 gave the National Commission for Backward Classes constitutional status in Article 338B and added Article 342A on lists of socially and educationally backward classes. In May 2021, in the Maratha reservation case (Jaishri Laxmanrao Patil), the Supreme Court read it as leaving only the Centre able to notify those lists. Parliament responded with the One Hundred and Fifth Amendment in August 2021, restoring the states' power to keep their own OBC lists.

Reservation is the most argued-over policy in this deck. A practitioner should be able to state each side accurately, and to know which claims have evidence behind them. The table pairs the common arguments with the evidence covered in this course.

In rural South Asia, land brings income, credit, status and political power. Caste shaped who received land under colonial settlements and who was excluded from land reform afterwards. The SECC 2011 rural data found that close to three in ten rural households were landless and earned most of their income from manual casual labour.

Siwan Anderson (American Economic Journal: Applied Economics 3, January 2011, 239-263) compared villages in rural India dominated, by majority land ownership, either by an upper caste or by a lower backward agricultural caste. Low-caste households earned substantially more in villages dominated by a low caste. The mechanism was a breakdown of trade in irrigation water across caste lines.

Sukhadeo Thorat and Paul Attewell ran a correspondence study of hiring in India's urban private sector, published in the Economic and Political Weekly (October 2007). Matched applications that differed only in the applicant's name, signalling a high-caste Hindu, a Dalit or a Muslim, were sent to real job advertisements. Employers called back fewer Dalit and Muslim applicants with the same qualifications.

Surveys show that Dalits earn less, but they cannot separate discrimination from differences in schooling or skills. A correspondence study holds everything on the application constant except the signal of caste or religion, so any gap in callbacks is the employer's response to that signal.

Lakshmi Iyer, Tarun Khanna and Ashutosh Varshney used the Economic Censuses of 1990, 1998 and 2005, covering 19 large states, to measure enterprise ownership by caste (Economic and Political Weekly, 2013). SCs and STs owned far fewer enterprises than their population share. OBCs owned about their share.

Iyer, Khanna and Varshney found the ownership gap was not significantly associated with literacy, secondary schooling or reliance on farming across states and districts. Other studies point to mechanisms that programmes can address directly, though the weight of each varies by region and sector.

Caste networks are a form of social capital: they find jobs, lend money and settle disputes. Kaivan Munshi and Mark Rosenzweig (American Economic Review, 2006) studied schooling in Mumbai. Lower-caste male networks kept channelling boys into local-language schools that led to traditional working-class jobs, even after returns to English-medium education rose in the 1990s. Lower-caste girls, outside those male job networks, switched rapidly to English schools.

Karla Hoff and Priyanka Pandey ran experiments with 321 high-caste and 321 low-caste junior high school boys in rural India (World Bank Policy Research Working Paper 3351, 2004; AER Papers and Proceedings, 2006). Boys solved mazes for money. When caste was not revealed there was no caste gap in performance. When caste was announced publicly, a large gap appeared.

Evidence on caste in the economy can be organised by market and by stage. The table below gathers the studies used in this section and notes what remains less studied, which is useful when you commission new research.

A practitioner needs to know which source records which categories, at what level and with what known problems. Few sources record individual jatis; most record only SC, ST, OBC and 'others'. Choosing the wrong source is a common reason why caste analysis in programme reports is thin.

The Census Primary Census Abstract for SCs and STs is the backbone of most targeting in India. Between 2001 and 2011 the SC population grew 20.8% and the ST population 23.7%, faster than the population as a whole. The notified lists also changed between the two censuses, so part of the growth reflects additions.

The Socio Economic and Caste Census 2011 collected caste names for every household. Its rural socio-economic data was released on 3 July 2015. The caste data was never released. In a September 2021 affidavit to the Supreme Court, the Union government said the exercise had produced more than 46 lakh different caste names nationally.

Bihar conducted its own caste-based survey and released the findings on 2 October 2023. It counted a population of over 13 crore. It was a rare full count of every caste in a large state, and it showed Other Backward Classes, split into Backward Classes and Extremely Backward Classes, as more than 63% of the population.

On 30 April 2025 the Cabinet Committee on Political Affairs decided to include caste enumeration in the next Census. The intent to conduct the Census was notified in the Gazette of India on 16 June 2025. Until 2011 the Census enumerated only SCs and STs. Census 2027 is also India's first digital Census, with an optional self-enumeration portal.

Most programme and research teams will use broad categories. The way you ask and code caste decides whether your results can be compared with national data and whether respondents trust you. A simple protocol avoids the most common errors.

A programme that reports only an overall average can show success while the groups it most needed to reach see no change. Disaggregating by caste category is the minimum. Where samples allow, disaggregate by caste and sex together, because Dalit and Adivasi women often face the largest gaps.

Caste categories are blunt. SC, ST and OBC each contain hundreds of groups with very different histories and outcomes. Official data also misses those whose caste the law does not recognise, such as Dalit Christians and Muslims. Treat every caste statistic as a lower bound on what you do not know.

Caste does not stop at the boundary of Hinduism. Dalits who became Christians, Muslims, Sikhs or Buddhists carried caste with them, because neighbours, employers and marriage markets kept recognising it. The law recognises this only in part: Sikh and Buddhist Dalits can hold SC status, while Christian and Muslim Dalits cannot. Two national commissions have examined whether that line is justified.

Nepal's Muluki Ain (National Code) of 1854 placed every group in a legal caste hierarchy, with rules on food, marriage and punishment by caste (Hofer, 1979, is the standard study). Today the Constitution of Nepal 2015 guarantees a right against untouchability and discrimination (Article 24) and rights of Dalits, including proportional inclusion in state bodies (Article 40).

Nepal's legal framework is among the strongest in the region on paper, yet commentary in Nepal describes the 2011 Act as ineffective in practice. A Kathmandu Post column of 9 January 2025 noted that caste-based violence continues, with Dalits killed in such violence, despite the constitutional guarantees. Hill and Madheshi Dalits also live in different regions and face different local power structures.

Pakistan inherited the colonial category of Scheduled Castes. In its 2017 Census, Scheduled Castes were reported separately from Hindus in the religion data, as a category of their own. A Dalit who identified as Hindu on the form was counted as Hindu, so the two figures together are the closest official guide to the size of Pakistan's Dalit population.

Bangladesh has Dalit and Harijan communities, and caste-based exclusion is recognised in national law-making: the government's own anti-discrimination bill lists caste among the prohibited grounds, alongside religion, race, language, age, gender and disability. That bill is the most important recent legal step found in the sources checked for this course.

Sri Lanka has its own caste hierarchies, and Parliament legislated against caste exclusion soon after independence. The Prevention of Social Disabilities Act, No. 21 of 1957 (in force 13 April 1957, amended by Act 18 of 1971), makes it an offence to impose a social disability on anyone by reason of caste, punishable by up to three years' imprisonment and a fine.

Every country in the region has caste, and each has responded through different legal tools. Comparing them helps practitioners working across borders, and shows that strong law is a starting point for change, with enforcement still to follow.

Use this checklist at the design stage of any programme in South Asia, whatever its sector. It turns the history, law and evidence in this course into questions a team can answer in a workshop. Record the answers in the design document so they can be checked at review.

Teams often ask whether to target by caste category, by economic need, or both. The answer depends on the problem and the context. This decision table sets out common situations and the approach the evidence in this course supports.

Illustrative. An NGO plans handpumps in a village of 400 households in eastern Uttar Pradesh: 260 households of a dominant OBC caste in the main settlement, 110 SC households in a separate tola 600 metres away, and 30 others. The panchayat proposes all four pumps in the main settlement, near the temple.

Caste data is sensitive. A list of Dalit households with names and locations could be used to target them. Ethical research protects respondents through consent, minimum collection, secure storage and careful publication. The Digital Personal Data Protection Act 2023 and its 2025 Rules commence in stages: the core duties and rights in sections 3 to 17 apply from 13 May 2027 (G.S.R. 843(E), 13 November 2025).

The caste of enumerators and facilitators affects what respondents say and who agrees to take part. A Dalit respondent may hesitate to report discrimination to an upper-caste interviewer from a nearby village; an upper-caste respondent may answer an untouchability question differently to a Dalit interviewer. Plan teams and protocols with this in mind.

Development organisations reproduce the caste patterns of the society around them. Senior staff in Indian NGOs, research institutes and foundations are often drawn from a narrow set of castes, while drivers, cleaners and field staff come from others. Rohith Vemula, a Dalit PhD scholar at the University of Hyderabad, died by suicide on 17 January 2016; his death prompted national debate about caste discrimination in institutions that see themselves as progressive.

The way a report describes caste affects how readers understand the people in it. Precise, sourced and non-sensational writing protects participants and makes the evidence more credible to policymakers. These rules apply to donor reports, academic papers and social media alike.

Dalit politics has moved through several forms. On 20 March 1927 Ambedkar led the Mahad Satyagraha to drink from the public Chavdar tank. On 14 October 1956 at Nagpur he and several hundred thousand followers converted to Buddhism. The Dalit Panthers formed in Bombay in 1972, led by young writers including Namdeo Dhasal, J. V. Pawar and Raja Dhale. Kanshi Ram founded the Bahujan Samaj Party on 14 April 1984.

Adivasi resistance long predates independence. The Santhal Hul began on 30 June 1855 under Sidho and Kanho Murmu and their siblings. Later tenancy laws, such as the Santhal Pargana Tenancy Act, restrict the transfer of Adivasi land in that region. After 1950, Adivasi movements focused on displacement by dams and mines, forest rights and self-governance in Fifth Schedule areas.

Several questions in this field remain open as of October 2026, and the answers will affect programmes and data for years. Each is worth watching through primary sources: Gazette notifications, judgments and Census releases.

If you remember nothing else from this course, remember these. Each one connects to a design decision, a data choice or a legal duty in programme work.

Caste connects to almost every other course in the ImpactMojo 101 Series. These are the most direct next steps, chosen to deepen the law, the economics and the methods used in this deck.