Policy analysis sits at the heart of good governance. It helps decision-makers figure out whether a scheme is working, what its consequences have been, and what should be done next. But policy analysis is not a single method – it is a family of approaches, each designed to answer a different kind of question. Understanding these types helps students, researchers, and administrators choose the right lens for the right problem.
Table of Contents
- What policy analysis actually means
- Empirical, evaluative, and normative analysis
- Empirical analysis: understanding the facts
- Evaluative analysis: judging worth
- Normative analysis: what should be done
- Retrospective and prospective analysis
- Retrospective analysis: learning from the past
- Prospective analysis: looking ahead
- Descriptive, predictive, and prescriptive analysis
- Descriptive analysis
- Predictive analysis
- Prescriptive analysis
- How the types fit together
- Purpose of the analysis
- Availability of data
- Stakeholder needs
- Political and institutional context
- Why these distinctions matter for practice
What policy analysis actually means
At its core, policy analysis is the systematic examination of public policies – their causes, content, consequences, and worth. Analysts rely on several approaches depending on whether they want to understand a problem, judge the value of a policy, or recommend what should be done. A flagship scheme like MGNREGA, for instance, can be studied in multiple ways – by counting person-days of employment generated, by judging whether it improved rural livelihoods, or by asking whether wage guarantees are the right moral response to rural distress. Each of these is a different type of policy analysis.
Broadly, the literature groups types of policy analysis into three overlapping pairs: empirical vs. normative, retrospective vs. prospective, and descriptive vs. prescriptive (with predictive analysis sitting alongside). Let’s walk through each.
Empirical, evaluative, and normative analysis
This first grouping is organised around the kind of question being asked – is it about facts, worth, or what ought to be done?
Empirical analysis: understanding the facts
Empirical policy analysis focuses on causes and consequences. It asks: what happened, why did it happen, and what effects followed? It is one of the major competing approaches in the field, often contrasted with normative analysis. The method relies heavily on data – household surveys, administrative records, economic indicators, and field studies.
Take the Pradhan Mantri Jan Dhan Yojana. An empirical analyst would examine how many accounts were opened, the share held by women, average balances, and whether account ownership changed borrowing or savings behaviour. The goal is to describe reality accurately, not to pass judgement on whether the scheme is good or bad.
Evaluative analysis: judging worth
Evaluative policy analysis goes one step further. It asks whether a policy was worth the effort – did it deliver value, achieve its goals, and use resources efficiently? What matters in the evaluative process is not just success or failure but what policy actors and institutions learn from the exercise.
Consider the Mid-Day Meal Scheme. Beyond simply counting meals served (empirical data), an evaluative analyst would ask whether the programme actually improved enrolment, attendance, nutrition outcomes, and learning levels. They would weigh the costs against these benefits and identify unintended consequences – say, caste-based discrimination during meal distribution, or a fall in household food spending.
Normative analysis: what should be done
Normative analysis is the most value-laden type. It asks what ought to be rather than what is. Harold Lasswell, writing in 1951, considered the ability to clarify value goals as integral to a policy analyst’s job, and this tradition continues today.
When analysts debate reservation policy in higher education, for instance, they are largely in normative territory. The question is not just whether reservations raise representation (empirical) or whether they are cost-effective (evaluative), but whether they align with constitutional principles of equality, social justice, and dignity. Disagreements in normative analysis often stem from differences in models, empirical estimates, and values – including who should have standing and how different criteria should be weighted.
Retrospective and prospective analysis
The second grouping is about timing. Is the analyst looking backward at a policy already in place, or forward at a policy yet to be adopted?
Retrospective analysis: learning from the past
Retrospective analysis examines policies after they have been implemented. It helps governments understand what worked, what didn’t, and why. Its purpose may be to identify opportunities for policy reform, by evaluating whether existing policies are economically justified and by identifying changes that could lower costs or increase benefits.
A retrospective analysis of demonetisation, for example, would look at its actual effects on cash circulation, digital payment adoption, GDP growth, and informal sector employment. Historical data from the RBI, NSSO, and CMIE would be central. The limitation, of course, is that it cannot rewrite the past – it can only inform future decisions.
Prospective analysis: looking ahead
Prospective analysis does the opposite. It evaluates a policy before it is implemented, forecasting likely impacts and identifying trade-offs. Much policy analysis is retrospective and not well suited to helping policymakers decide what to do; in contrast, prospective policy analysis seeks to assist in formulating responses to challenging public policy questions.
When NITI Aayog prepares a blueprint for electric vehicle adoption, it draws heavily on prospective analysis – projecting charging infrastructure needs, battery supply constraints, fiscal costs of subsidies, and employment effects in the automobile sector. The challenge here is uncertainty: forecasts are only as good as the assumptions behind them.
Descriptive, predictive, and prescriptive analysis
The third grouping is the most closely tied to methodology. It mirrors a well-known hierarchy in data analytics as well.
Descriptive analysis
Descriptive analysis answers the question “what happened?” It summarises historical and current data to give policymakers an accurate picture of the problem landscape. Descriptive analytics summarise past and present trends and patterns in the data to answer the question “what happened”.
Census reports, the Economic Survey, and NFHS rounds are classic examples of descriptive policy material. They don’t recommend what to do – they tell us where we stand on literacy, child nutrition, sanitation coverage, or female labour force participation. Descriptive analysis is the foundation on which other types build.
Predictive analysis
Predictive analysis forecasts what is likely to happen if trends continue or certain actions are taken. It uses statistical modelling, econometric techniques, and increasingly machine learning. Predictive analytics use primarily historical data to create models that answer the question “what will happen”.
Climate projections feeding into India’s Nationally Determined Contributions, fiscal projections in Union Budget documents, and epidemiological models used during the COVID-19 response are all predictive in nature. The quality of predictions depends on data availability, model design, and the stability of underlying relationships – which is why predictive analysis always carries a margin of uncertainty.
Prescriptive analysis
Prescriptive analysis recommends specific courses of action. Prescriptive analytics employ data-driven models and optimisation algorithms to recommend the most effective actions, interventions, or allocation strategies, answering the question “what should I do”.
When a state finance commission recommends a specific formula for devolving funds to local bodies, or when a task force lays out a roadmap for universal health coverage, the analysis is prescriptive. It moves from facts and forecasts into the territory of judgement and advice.
How the types fit together
These categories are not mutually exclusive. A good policy analysis usually blends several. Studying the Goods and Services Tax rollout, for example, benefits from empirical measurement of revenue collection, evaluative judgement on whether it achieved tax simplification, predictive estimates of compliance behaviour, and prescriptive recommendations for rate rationalisation.
The choice of approach depends on several factors:
Purpose of the analysis
If the goal is accountability, evaluative and retrospective analysis dominate. If the goal is to design a new programme, prospective and prescriptive approaches take over. Mainstream research is descriptive or explanatory, while critical policy analysis incorporates normative assessments of policy and policymaking.
Availability of data
Historical data enables retrospective and descriptive work. Where data is thin, analysts lean on normative reasoning, expert consultation, and scenario-based prospective analysis.
Stakeholder needs
Policymakers often want prescriptive recommendations they can act on. Academics lean toward descriptive and explanatory work. Civil society groups often emphasise normative critique. A well-rounded analysis speaks to all these audiences without pretending that facts and values can be cleanly separated.
Political and institutional context
Contested policies – farm laws, data protection, citizenship rules – benefit from transparent empirical work to anchor the debate, even though normative disagreements will persist. No policy analysis can be entirely values-free, but good analysis should be values-agnostic in the sense of being transparent about trade-offs between values.
Why these distinctions matter for practice
For a student or practitioner of public administration, knowing the type of analysis being undertaken prevents two common pitfalls. The first is mistaking description for evaluation – reporting numbers as if they settle the debate about whether a policy is good. The second is smuggling normative judgements into supposedly empirical work without acknowledging the values at play.
Clear categorisation also sharpens research design. An empirical study needs reliable data and careful causal inference. A prospective study needs explicit assumptions and sensitivity analysis. A normative study needs honest engagement with competing values, not a pretence of neutrality. When UPSC aspirants, policy researchers, or government officers know which type of analysis they are doing – and which they are not – their conclusions become more credible and more useful.
What do you think? When you look at a major Indian policy debate – say, the Uniform Civil Code or free electricity schemes – which type of analysis do you think is most missing from public discussion? And can any policy question be answered through empirical analysis alone, or do values always slip in?
References
- https://www.encyclopedia.com/social-sciences/applied-and-social-sciences-magazines/policy-analysis
- https://en.wikipedia.org/wiki/Policy_analysis
- https://www.sciencedirect.com/topics/social-sciences/policy-evaluation
- https://link.springer.com/article/10.1007/s11077-024-09525-w
- https://www.hks.harvard.edu/publications/methodology-normative-policy-analysis
- https://datasmart.hks.harvard.edu/news/article/white-paper-retrospective-benefit-cost-analysis-664
- https://www.researchgate.net/publication/44690252_Prospective_policy_analysis_How_an_epistemic_community_informed_policymaking_on_intentional_self_poisoning_in_Sri_Lanka
- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11549582/
- https://www.tandfonline.com/doi/full/10.1080/13501763.2023.2221282
- https://appext.hks.harvard.edu/publications/getFile.aspx?Id=633
Leave a Reply