A woman in a village in Jharkhand collects rice at a ration shop after her thumbprint is checked against a national database. A child in Tamil Nadu eats a cooked meal at school. A shopkeeper in Delhi stops handing out thin plastic carry bags. Each of these moments is the end of a long chain of public decisions: a law, a set of rules, a budget line, a court order, an officer's instruction. Public policy is the name for that chain, and this course follows it from the Cabinet room to the ration shop.

The most quoted definition comes from the American political scientist Thomas Dye: public policy is "whatever governments choose to do or not to do" (Dye, Understanding Public Policy, 1972, p. 2). It is short on purpose. It includes laws and spending, and it also includes decisions to leave a problem alone. Critics call it too broad to be useful, because almost anything a government does then counts as policy.

In South Asia a single policy usually lives in several documents at once, each with a different legal force. Practitioners who read only the scheme guidelines often miss the Act that creates the entitlement, or the court order that set the minimum standard. The table below shows the main forms with Indian examples checked for this course.

Economists and political scientists give several reasons why a problem becomes a public responsibility. These are reasons to consider public action. None of them proves that a particular government programme will work better than doing nothing, which is a separate question this course returns to in Sections 6 and 7.

The second half of Dye's definition matters in practice. When a government keeps an issue off its agenda, leaves a law without rules, or lets a scheme run without evaluation, that is a choice with consequences. Political scientists call issues that never reach a formal decision non-decisions. They are hard to study because nothing is written down.

Policy is made by many actors with different powers. Formal authority sits with the executive, the legislature and the courts. Real influence also comes from the bureaucracy, state and local governments, parties, donors, business, civil society and the media. Section 9 shows how to map them for a specific issue.

Policy choices rarely pit good against bad. They pit good against good: speed against care, efficiency against equity, a tight target against wide coverage. The same word can also mean different things to different sides: fairness to a taxpayer may mean paying only for the needy, while fairness to a farm worker may mean a benefit nobody has to prove they deserve.

In 1956 the political scientist Harold Lasswell divided the decision process into seven functions: intelligence, recommendation, prescription, invocation, application, appraisal and termination. Later textbooks condensed his list into the familiar policy cycle. The stages give a common vocabulary for asking where a policy is, and what kind of work it needs next.

Thousands of problems compete for a government's attention and only a few get it. Scholars distinguish the broad public agenda, the issues people talk about, from the institutional agenda, the issues a Cabinet, legislature or court is actually considering. Moving an issue from the first to the second is the purpose of most advocacy.

Formulation turns a problem into a proposal. In India the formal route runs through the ministry that owns the subject, which must consult other departments before a proposal goes further. Rule 4 of the Government of India (Transaction of Business) Rules 1961, as amended to 13 January 2025 sets out the main consultations, and they explain why a draft often changes before anyone outside government sees it.

A proposal becomes policy only when an authority with legal power adopts it. Under Article 74(1) of the Constitution the Council of Ministers, headed by the Prime Minister, aids and advises the President, who acts on that advice. The Second Schedule to the Transaction of Business Rules lists the cases that must go to the full Cabinet, starting with "cases involving legislation including the issue of Ordinances".

Adoption fixes what should happen. Implementation is what does happen. In India most social policy is delivered by state governments and local bodies even when the Union designs and funds it. The VB-G RAM G Act 2025 shows the division of labour written into law: the Union sets the framework and the normative allocation, each State must notify its own Scheme within six months of commencement (s3), and Gram Panchayats prepare Viksit Gram Panchayat Plans from which works originate (s4).

Evaluation asks whether a policy did what it promised, at what cost, and for whom. India's main government evaluation body is the Development Monitoring and Evaluation Office (DMEO), an attached office of NITI Aayog constituted in September 2015 by merging the Programme Evaluation Office and the Independent Evaluation Office. To inform the Fifteenth Finance Commission's deliberations, the Ministry of Finance entrusted DMEO with evaluations of Centrally Sponsored Schemes, reported sector by sector; the Women and Child Development report is dated March 2021.

The replacement of India's rural employment guarantee shows every stage of the cycle. The table maps the steps that can be checked in public documents. It does not try to explain why the change happened, which needs the theories in Section 3 and the political economy tools in Section 9.

The stage model is a teaching device, and it is useful for organising a policy analysis or a monitoring plan. It misleads when it is read as a description of how policy really moves. Stages overlap, decisions are revisited, and implementation often reshapes the policy so much that the officials who carry it out are, in effect, still formulating it.

The policy cycle carries a hidden assumption: that decision makers define a goal, list every option, estimate the consequences of each, and choose the best. This is the rational-comprehensive model. It is a useful benchmark for an analyst. As a description of how governments decide, it fails, because no ministry has the time, information or agreement on goals that the model needs.

Charles Lindblom's article "The Science of 'Muddling Through'" (Public Administration Review 19(2), 1959) contrasted the rational-comprehensive, or root, method with the method of successive limited comparisons, the branch method. Administrators, he argued, look mainly at options that differ only slightly from current policy, and judge a policy partly by whether the people involved can agree on it.

Incremental thinking is easy to see in public budgets. Departments usually start from the previous year's allocation and argue over the change, because rebuilding every line from zero each year is beyond any finance ministry. The example below is invented for teaching; it shows the logic, and its numbers come from no real budget.

John Kingdon's Agendas, Alternatives, and Public Policies (first published 1984) asked why some issues reach the agenda and others do not. His answer is that three largely separate streams flow through government. Change becomes possible when they meet in a policy window, and someone is ready to join them.

The 1991 reforms are often used to illustrate a policy window. The facts below are documented; the mapping onto Kingdon's streams is an interpretation offered for teaching, and other readings are possible. The point is the method: separate the problem, the available ideas and the politics, then ask what joined them.

Frank Baumgartner and Bryan Jones, in Agendas and Instability in American Politics (1993), noticed that most policies stay almost unchanged for long periods and then shift sharply. Their explanation combines limited attention with power. A small policy community keeps an issue quiet by defining it as technical or settled. Change comes when the issue's policy image is redefined and it moves to a new venue where different people decide.

A clear South Asian example of venue shift is the right to food case. In 2001 the People's Union for Civil Liberties filed a writ petition (WP(C) 196 of 2001). On 28 November 2001 the Supreme Court directed governments to provide every child in government and government-assisted primary schools a prepared midday meal with at least 300 calories and 8–12 grams of protein, for at least 200 days a year. States serving dry rations were told to switch to cooked meals in phases.

None of these theories is right on its own. Each draws attention to a different force, and a good analysis uses more than one. The table summarises what each explains well and what it misses, with the question it suggests for your own work.

Once a government decides to act, it must choose a tool. Christopher Hood's The Tools of Government (1983) grouped the tools by the resource each one uses. The scheme is often remembered as NATO: nodality, authority, treasure and organisation. Every instrument in this section draws on at least one of the four.

India's Economic Survey 2018-19 placed policies on a spectrum by how strongly they push behaviour: laissez faire, nudge, incentivise and mandate. Its Figure 2 draws each policy as a bar running from laissez faire to the furthest level it reaches. The table follows those bars; whether each placement is right is a good discussion question.

Regulation sets rules and punishes breaches. It looks cheap, because the government pays little directly. Its real cost sits in inspection and enforcement. India's ban on identified single-use plastic items is a good example. The Ministry of Environment, Forest and Climate Change notified the Plastic Waste Management Amendment Rules 2021 on 12 August 2021, prohibiting identified items from 1 July 2022.

Price subsidies are popular because they look universal. Their benefits often go to people who consume more, or leak before reaching anyone. The Economic Survey 2014-15 put numbers on this using the NSS 68th round (2011-12) and its own estimates. The chart shows the share of the public distribution system (PDS) allocation estimated as lost to leakage, by commodity.

Transfers give households resources directly. In-kind transfers, such as grain through ration shops, guarantee a specific good but need a physical supply chain. Cash transfers are cheaper to deliver and let households choose, but their value falls with inflation and they need bank access. India's Economic Survey 2014-15 argued that the JAM trinity, Jan Dhan accounts, Aadhaar and mobile phones, could make direct cash transfers possible at scale.

A conditional transfer pays only if the household does something, such as keeping a child in school or attending health check-ups. An unconditional transfer pays regardless. Pakistan's Benazir Income Support Programme (BISP) runs both kinds: the Benazir Kafaalat programme makes unconditional payments, while Benazir Taleemi Wazaif (education stipends) and the Nashonuma programme (maternal and child health) attach conditions.

Governments also act by informing and persuading. India's Give It Up campaign, launched by the Prime Minister, asked LPG consumers who could afford the market price to surrender their cooking gas subsidy voluntarily. Each consumer who gave it up was linked to a BPL household that received an LPG connection in turn. By 28 July 2015, 13.9 lakh consumers had given it up; by 1 April 2023 the Ministry of Petroleum and Natural Gas put the total at nearly 1.13 crore.

Richard Thaler and Cass Sunstein's Nudge (2008) argued that the way a choice is presented changes what people choose, and that governments can design choices to help people without forbidding anything. India's Economic Survey 2018-19 drew three principles from this literature: social norms shape behaviour, people stick with defaults, and repeated reminders sustain good habits.

Instrument choice should follow from a diagnosis of why the problem exists. The table is a starting point for discussion; real choices also depend on law, politics and capacity. Use it to check that a proposed tool addresses the cause you have identified, and to name the main risk before it arrives.

India follows a parliamentary system. Article 74(1) of the Constitution provides for "a Council of Ministers with the Prime Minister at the head to aid and advise the President who shall, in the exercise of his functions, act in accordance with such advice". Executive action is taken in the President's name, but the decisions are the government's. Article 77(3) lets the President make rules for the convenient transaction of business, and the Transaction of Business Rules 1961 are made under that power.

Most policy starts as a note in the owning department. Under rule 3 of the Transaction of Business Rules, business allotted to a department is disposed of by or under the direction of its minister, subject to the consultation rules. Rule 4(2) requires the previous concurrence of the Ministry of Finance for any order with a financial bearing, and rule 4(3) requires consultation with the Ministry of Law on proposals for legislation and on rules of a general character.

Rule 6 of the Transaction of Business Rules provides for Standing Committees of the Cabinet, with members chosen by the Prime Minister, and allows ad hoc committees including Groups of Ministers to investigate and, if authorised, decide. Any decision of a committee may be reviewed by the Cabinet. The Second Schedule lists cases that must go before the full Cabinet.

On 1 January 2015 the Union Cabinet replaced the Planning Commission, set up by a Cabinet resolution on 15 March 1950, with the National Institution for Transforming India (NITI Aayog). The government's press note described it as a "think tank" of the government that would give the Union and the states strategic and technical advice. The Prime Minister chairs it, and its Governing Council includes the Chief Ministers of all states.

In 2014 the Union government adopted a policy on consultation before legislation. According to PRS Legislative Research, it applies to every ministry before a legislative proposal, including subordinate legislation, goes to the Cabinet. A draft bill is to be published for public comment for 30 days, with a justification, its financial implications, an estimated impact assessment and an explanatory note. A summary of comments is to be published on the ministry's website.

After Cabinet approval a bill is introduced in either House, except Money Bills and other financial bills, which start in the Lok Sabha. It can be referred to a Department-related Standing Committee for detailed examination, though referral is not mandatory. It is then debated and voted clause by clause, passed by both Houses, and sent for the President's assent.

Committees are where most detailed scrutiny of bills happens. The share of bills referred to them has fallen sharply. PRS Legislative Research reports that 16% of bills in the 17th Lok Sabha (2019–2024) were referred, lower than in each of the three previous Lok Sabhas.

Article 246 and the Seventh Schedule divide legislative power between the Union and the states into three lists. Much of social policy sits with the states or in the Concurrent List, which is why national schemes depend on state governments. The 42nd Amendment of 1976 moved education from the State List to the Concurrent List.

Many national programmes are Centrally Sponsored Schemes: the Union designs and part-funds them, and states implement and share the cost. The funding formula is a policy choice with large effects. Under the VB-G RAM G Act 2025, s22, the sharing pattern is 90:10 for the North Eastern States, the Himalayan States and Jammu and Kashmir, 60:40 for all other states and union territories with a legislature, and fully Union-funded for union territories without one.

Indian courts are policy actors. Article 32 lets a person approach the Supreme Court for enforcement of fundamental rights, and Article 226 gives High Courts similar writ powers. After the Emergency, judges including P.N. Bhagwati and V.R. Krishna Iyer relaxed the rule that only an injured person may sue. In Hussainara Khatoon v State of Bihar (1979) the Court took up the cases of undertrial prisoners, and S.P. Gupta v Union of India (AIR 1982 SC 149) widened standing for public interest petitions.

Jeffrey Pressman and Aaron Wildavsky's Implementation (1973) carried the subtitle How Great Expectations in Washington Are Dashed in Oakland: a federal promise made in the capital and disappointed in one city, as the title itself announces. One way to see the problem is arithmetic: a programme that needs many separate agencies and decisions to agree has many chances to stall, even when each step is likely to go well.

Michael Lipsky's Street-Level Bureaucracy (1980) studied teachers, police officers, social workers and others who deal directly with citizens and have discretion over them. They work with too few resources, rising demand, vague goals and clients who cannot go elsewhere. They cope by building routines, rationing their time and simplifying how they see their clients.

Lucy Gilson's review of Lipsky (Oxford Handbook of Classics of Public Policy, 2015) notes that coping routines can control clients, ration services by imposing time or money costs, and sort clients into more and less deserving. Lipsky himself said coping may "widen the gap between policy as written and policy as performed". The examples below are invented to show the mechanisms; they describe no real office.

In a 2009 Harvard Kennedy School working paper, Lant Pritchett argued that India's capable central institutions were poorly connected to delivery on the ground. He pointed to examples from health, education and routine tasks such as issuing driving licences, where state agents routinely did not do the tasks assigned to them, creating a gap between rules on paper and practice.

Devesh Kapur's article in the Journal of Economic Perspectives (2020, 34(1): 31–54) offers a more mixed picture. The Indian state, he argues, performs well on complex tasks run on a massive scale and on macroeconomic outcomes, and badly on everyday local services. He also disputes the claim that the state is bloated.

Aditya Dasgupta and Devesh Kapur (American Political Science Review 114(4), 2020) surveyed local rural development officials across India, including time-use diaries. They describe bureaucratic overload: local bureaucrats are often heavily under-resourced relative to their responsibilities. Officials with fewer resources were worse at implementing rural development programmes, plausibly because they had too little time for managerial tasks.

Unannounced visits are a direct way to measure implementation. Chaudhury, Hammer, Kremer, Muralidharan and Rogers (Journal of Economic Perspectives, 2006) visited schools and clinics in six countries including Bangladesh and India. A later nationally representative Indian panel by Muralidharan, Das, Holla and Mohpal (Journal of Public Economics, 2017) covered schools in 1,297 villages.

The lesson of this section is to treat implementation as part of design. A policy that assumes perfect officials, perfect data and perfect payment systems will fail in predictable ways. The checklist below can be applied to any draft scheme before it is adopted.

Evidence-based policy means using the best available information about what a policy will do. In practice policy makers draw on several kinds of evidence, each answering a different question. A randomised trial can say whether a programme changed an outcome in one place; administrative data can say how many people it reached; qualitative work can say why it failed for some of them.

Muralidharan, Niehaus and Sukhtankar (American Economic Review 106(10), 2016) evaluated biometrically authenticated payments ("Smartcards") for the employment guarantee and social security pensions in Andhra Pradesh. The rollout was randomised across 157 subdistricts covering 19 million people, so the experiment was built into the order of an actual government rollout.

Banerjee, Duflo, Imbert, Mathew and Pande (AEJ: Applied Economics 12(4), 2020) studied a reform to how funds flowed in India's workfare programme. Advance payments to local officials were replaced by "just-in-time" payments triggered by e-invoicing, which made misreporting easier to detect. The study used a large field experiment and then followed the nationwide scale-up.

Evidence can also show harm. Muralidharan, Niehaus and Sukhtankar (NBER Working Paper 26744, 2020, revised 2021) evaluated the introduction of Aadhaar-based biometric authentication to collect benefits in the Public Distribution System in Jharkhand, using randomised and natural experiments. Corruption fell, but with substantial costs to legitimate beneficiaries.

Governments often report results from their own administrative data. On 21 April 2025 the Press Information Bureau (Ministry of Finance) publicised an assessment by the BlueKraft Digital Foundation reporting cumulative savings of Rs 3.48 lakh crore from direct benefit transfer (DBT), with subsidies falling from 16% to 9% of total expenditure.

DMEO describes its mandate as monitoring and evaluating the implementation of schemes, programmes and initiatives of the Government of India, to strengthen delivery, and building the monitoring and evaluation system in the country. Its website lists an Output-Outcome Monitoring Framework (OOMF) and a Data Governance Quality Index (DGQI) among its tools, and it commissions sector evaluations through open tenders.

Policy research increasingly uses administrative records that contain personal data. In India the Digital Personal Data Protection Act 2023 and its 2025 Rules impose their duties from 13 May 2027. From that date section 17(2)(b) exempts processing "necessary for research, archiving or statistical purposes" if the data is not used to take a decision specific to a Data Principal and the processing follows prescribed standards.

A strong evaluation is one input into a political decision. It competes with budgets, promises, interests and timing. Kingdon's framework helps here: evidence sits in the policy stream, and it waits for a problem to be recognised and the politics to allow action. Researchers who want their work used plan for that wait.

The National Rural Employment Guarantee Act 2005 (No. 42 of 2005) made rural work a legal entitlement. Section 3 required each state to provide every rural household whose adult members volunteered for unskilled manual work "not less than one hundred days of such work in a financial year". Wages were to be paid weekly, or no later than a fortnight after the work was done. An applicant not given work within fifteen days was entitled to an unemployment allowance (s7).

The Viksit Bharat Guarantee for Rozgar and Ajeevika Mission (Gramin) Act, Act No. 36 of 2025, dated 20 December 2025, replaced it. The Ministry of Rural Development notified on 11 May 2026 that the Act would come into force on 1 July 2026, with MGNREGA repealed from the same date under section 37. The core guarantee rises to "not less than one hundred and twenty-five days" a year (s5).

Comparing the two Acts section by section shows that the change is about more than the number of days. The funding rules, the planning rules and the seasonal pause change who decides how much work is available, and when. The table uses only the text of the two statutes.

The frameworks from earlier sections suggest what to watch as the new Act is implemented. These are questions for monitoring, and none of them is a finding: as of October 2026 there is no published outcome data for a full year under the new law.

India's move to direct benefit transfer rests on the Aadhaar (Targeted Delivery of Financial and Other Subsidies, Benefits and Services) Act 2016, Act 18 of 2016, which received the President's assent on 25 March 2016. Section 7 lets the Union or a state require proof of an Aadhaar number, or authentication, as a condition for a subsidy, benefit or service paid for from the Consolidated Fund.

The evidence on biometric and digital delivery in India points in more than one direction. The pattern across studies is that the technology's effect depends on how the reform is designed and rolled out, and on what happens to people for whom the system fails.

The Benazir Income Support Programme was launched in late 2008 to cushion poor households against the food, fuel and financial crisis, with food inflation close to 25%. An ODI study by Shanza Khan and Sara Qutub (November 2010) records that a plan to identify beneficiaries from the national identity database was dropped during design. Instead, 8,000 forms were given to each member of the federal legislature to distribute to families they considered eligible. A poverty scorecard based on a proxy means test then replaced selection by legislators.

In 2023 Sri Lanka replaced its long-running Samurdhi benefit with the Aswesuma welfare benefit scheme. The Centre for Poverty Analysis (CEPA) published a rapid appraisal by Hemesiri Kotagama on 27 June 2023. It notes that the replacement was a condition in the government's 2023 loan agreement with the International Monetary Fund, and that households are scored on six dimensions: education, health, economic level, assets, housing condition and family demography.

Set side by side, the four cases show recurring patterns in South Asian social policy. Each lesson below draws on evidence or documents cited earlier in this section, and each points to a design question you can ask of any new scheme.

Many well-designed policies are never adopted, or are adopted and then hollowed out. The usual reason is political: the people who would lose from the change are organised and attentive, while those who would gain are scattered or unaware. Political economy analysis looks at interests, institutions and incentives to explain why a policy problem persists and what might shift it.

The World Bank's good practice framework by Verena Fritz, Kai Kaiser and Brian Levy (2009) starts from a specific problem and works down in three steps. Its value for practitioners is focus: the analysis stops when it explains the problem you need to solve.

A stakeholder map lists everyone who can affect or is affected by a policy, then places them by how much power they hold and how strongly the issue affects them. The grid tells you whom to keep closely engaged and whom to keep informed. The placements below are an invented example for a plastic ban, used to show the method.

A useful rule of thumb in political economy: when a policy's costs fall on a small, organised group and its benefits are spread thinly across many people, the small group usually wins. The Economic Survey 2014-15 found that the bottom half of households consumed only 25% of subsidised LPG, which helps explain why untargeted price subsidies persist: their better-off beneficiaries notice and resist any cut.

In a federation, a single policy has two sets of politicians with different incentives. Under a 60:40 funding split, a state government pays two rupees for every three the Union pays, and the political credit may go to whichever level voters notice. Dasgupta and Kapur (2020) found fewer resources go to administrative units where political responsibility for implementation is less clear.

Before recommending a policy, test it against four kinds of feasibility. A proposal that passes on evidence but fails one of these will need redesign or a longer plan. Write down your judgement for each test and the evidence for it, so others can challenge it.

Kingdon's windows and Baumgartner and Jones's punctuations both say that timing matters as much as content. A practitioner cannot open a window, but can be ready when one opens. Preparation means having evidence, a costed proposal, a draft text and allies in place before the moment arrives.

Policy analysis is the work of advising a decision maker on what to do. The six steps below are a common teaching sequence used in many forms by policy schools. They work for a one-page note to a district collector as well as a long report for a ministry; only the depth changes.

Most mistakes practitioners make about policy come from reading only one document. Before designing a programme around a scheme, or commenting on a draft, assemble the full set of documents and check each of these points.

A district team finds that some eligible families are turned away when fingerprint authentication fails at ration shops. The options and scores below are invented to show the method; real scores need local data. Scores run from 1 (poor) to 3 (good) on each criterion.

Different stages offer different openings, and each has its own form and timing. The table lists formal routes; informal relationships with the officers who hold the file matter at every stage.

Adoption is where many organisations stop paying attention, and implementation is where most policies change shape. A short monitoring plan, agreed before rollout, keeps attention on the points where Sections 6 to 8 show failures usually happen. Pick the rows that match the policy's main risk.

These mistakes recur in policy proposals from NGOs, researchers and governments alike. Each has appeared in this course in some form. Use the list as a final check before a recommendation leaves your desk.

A policy brief is a short document, usually two to four pages, written for a decision maker who has little time and a decision to make. It differs from a research paper in purpose. The paper shows how the authors know something; the brief tells a reader what to do about it, with just enough evidence to be trusted.

The structure below puts the conclusion first, because many readers stop after the first paragraph. Every later section supports that opening. Keep the evidence section to the two or three findings that matter for the decision, each with its source.

Officials read briefs quickly, often on a phone. Plain words, short sentences and numbers with sources make a brief easier to act on and harder to dismiss. The before-and-after example is invented to show the difference.

A brief earns trust by being exact about what is known. Give the source and year of every number, say what kind of study produced it, and state the main limit in one plain sentence. Readers forgive uncertainty that is explained; they stop trusting a brief that overstates.

Each point links to a section of this course and to a decision a practitioner makes. Return to them when you review a scheme, comment on a draft or plan an evaluation.

Public policy connects to most courses in the ImpactMojo 101 Series. These are the most direct next steps, chosen to deepen the law, the politics, the money and the evidence behind the topics in this deck.