Summary
+PolicyEngine.py (PolicyEngine Contributors 2026) is +an open-source, multi-country microsimulation framework for tax-benefit +policy analysis, implemented in Python. The package provides a unified +interface for running policy simulations, analyzing distributional +impacts, and visualizing results across the US and the UK. It delegates +country-specific tax-benefit calculations to dedicated country packages +(policyengine-us and policyengine-uk) while providing shared +abstractions for simulations, datasets, parametric reforms, and output +analysis. The framework supports both individual household simulations +and population-wide microsimulations using representative survey +microdata with calibrated weights. PolicyEngine powers an interactive +web application at policyengine.org that enables +non-technical users to explore policy reforms in both countries.
+Statement of Need
+Tax-benefit microsimulation models are essential tools for evaluating +the distributional impacts of fiscal policy. Governments, think tanks, +and researchers rely on such models to estimate how policy reforms +affect household incomes, poverty rates, and government budgets. +Existing microsimulation models face significant access barriers. TAXSIM +(Feenberg and Coutts +1993) at NBER computes only tax liabilities and omits the benefit +side of the ledger entirely. The models maintained by the Congressional +Budget Office and the Tax Policy Center are fully proprietary and +unavailable to external researchers. In the UK, UKMOD (Sutherland and +Figari 2013), maintained by the University of Essex, requires a +formal application and institutional affiliation to access, and the +models maintained by HM Treasury and the Institute for Fiscal Studies +are similarly proprietary. PolicyEngine addresses these gaps by +providing an open-source Python microsimulation framework that spans +multiple countries under a consistent API. Users can supply their own +microdata or use built-in datasets, and compute the impact of current +law or hypothetical policy reforms on any household or a national +population. The Simulation class supports individual household analysis, +while population-level aggregate analysis uses representative survey +datasets with calibrated weights. Because existing proprietary models +cannot be independently verified, PolicyEngine enables reproducible and +transparent policy analysis. The framework’s open development on GitHub +enables external validation, community contributions, and reproducible +policy analysis across countries.
+State of the Field
+Tax-benefit microsimulation — pioneered by Orcutt (1957) and surveyed by Bourguignon and +Spadaro (2006) — underpins much of modern +fiscal policy evaluation. The primary UK microsimulation models include +UKMOD, maintained by the Institute for Social and Economic Research +(ISER), University of Essex, as part of the EUROMOD family (Sutherland and +Figari 2013), and proprietary models maintained by HM Treasury +and the Institute for Fiscal Studies. OpenFisca (OpenFisca Contributors 2024) pioneered +the open-source approach to tax-benefit microsimulation in France. +PolicyEngine originated from OpenFisca and builds on this foundation +through the PolicyEngine Core framework (Woodruff et al. 2024).
+Rather than contributing these features directly to OpenFisca, +PolicyEngine introduced a separate analyst-facing layer because the +project required capabilities that cut across countries and sit +downstream of legislative modeling: harmonized dataset handling, a +stable reform API, standardized distributional outputs, and integration +with a public-facing web application. This design lets country model +packages focus on statutory rules while shared analysis workflows evolve +independently.
+PolicyEngine differentiates itself in several ways:
+-
+
- Open-source, multi-country framework: a single +Python package supports the US and UK tax-benefit systems under a +consistent API, with no institutional access or license fees +required. +
- Comprehensive program coverage: the US model covers +over 11 programs including federal income tax, payroll taxes, state +income taxes, SNAP, SSI, Social Security, Medicare, Medicaid, EITC, CTC, +and TANF; the UK model covers over 37 programs spanning income tax, +National Insurance, Universal Credit, Child Benefit, Council Tax, and +devolved policies in Scotland and Wales. +
- Separated modeling, analysis, and data layers: the +project splits reusable engine logic into PolicyEngine Core, +country-agnostic analysis workflows into PolicyEngine.py, country +legislation into policyengine-us and policyengine-uk, and enhanced +survey microdata into companion repositories (Woodruff et al. +2024; Woodruff and Ghenis 2024). This separation allows +each layer to be versioned and updated independently as legislation, +methodology, and microdata change. +
- Programmatic reform and economic analysis: users
+can define hypothetical policy reforms as date-bound parameter values,
+compose multiple reforms with the
+operator, or implement +structural changes via simulation modifiers, then evaluate impacts on +households, poverty, inequality, government budgets, and subnational +regions. Behavioral response modules model both intensive-margin (hours +adjustment) and extensive-margin (participation) labor supply responses +to policy changes.
+
Software Design
+PolicyEngine is built as a four-layer system. PolicyEngine Core +extends the OpenFisca engine with reusable simulation abstractions, +versioned parameters, and dataset interfaces shared across countries +(Woodruff et al. +2024). PolicyEngine.py adds country-agnostic analyst workflows, +including baseline-versus-reform comparisons, standardized output types, +and visualization helpers. The policyengine-us and policyengine-uk +packages contain statutory logic, variables, and entity structures +specific to each tax-benefit system. Companion data repositories hold +enhanced survey microdata and calibration pipelines for the CPS (Woodruff and Ghenis 2024) +and Family Resources Survey.
+
+
+This split trades some packaging complexity for clearer ownership and +release independence. Legislative changes in a country package do not +require duplicating shared output logic; methodological changes to +distributional analysis do not require modifying statutory formulas; and +microdata refreshes can be versioned separately from the modeling +libraries. It also supports different contributor workflows, since legal +rules, data calibration, and analyst-facing outputs are maintained by +overlapping but distinct groups.
+As shown in Figure 1, at runtime a simulation combines three inputs: +policies from a country model version, household microdata, and optional +behavioral response parameters. The following example doubles the US +federal standard deduction for single filers and runs a full +distributional analysis:
+import datetime
+from policyengine.core import Parameter, ParameterValue, Policy, Simulation
+from policyengine.tax_benefit_models.us import (
+ economic_impact_analysis, us_latest,
+)
+
+param = Parameter(
+ name="gov.irs.deductions.standard.amount.SINGLE",
+ tax_benefit_model_version=us_latest,
+)
+reform = Policy(
+ name="Double standard deduction",
+ parameter_values=[
+ ParameterValue(
+ parameter=param,
+ start_date=datetime.date(2026, 1, 1),
+ end_date=datetime.date(2026, 12, 31),
+ value=30_950,
+ ),
+ ],
+)
+baseline = Simulation(tax_benefit_model_version=us_latest)
+reformed = Simulation(tax_benefit_model_version=us_latest, policy=reform)
+analysis = economic_impact_analysis(baseline, reformed)The analysis object contains decile impacts,
+program-by-program statistics, poverty rates, and inequality metrics.
+PolicyEngine.py then applies a consistent analysis layer across
+countries, producing these outputs from the resulting entity-level
+data.
PolicyEngine models static fiscal impacts; it does not model +macroeconomic feedback effects or general equilibrium responses.
+Research Impact Statement
+PolicyEngine has demonstrated research impact across government, +academia, and policy research in both the US and UK.
+Government adoption. In the US, PolicyEngine +collaborated with the Better Government Lab — a joint center of the +Georgetown McCourt School of Public Policy and the University of +Michigan Ford School of Public Policy — on benefits eligibility research +(Ghenis 2024b). In the +UK, co-author Nikhil Woodruff served as an Innovation Fellow in +2025–2026 with 10DS — the data science team at 10 Downing Street — +adapting PolicyEngine for government use (HM Government 2025). The 10DS team +used PolicyEngine to rapidly estimate the impacts of policy reforms on +living standards, local area incomes, and distributional outcomes. HM +Treasury has also formally documented PolicyEngine in the UK Algorithmic +Transparency Recording Standard, describing it as a model their Personal +Tax, Welfare and Pensions team is exploring for “advising policymakers +on the impact of tax and welfare measures on households” (HM Treasury 2024).
+Congressional and parliamentary citation. In the US, +Representatives Morgan McGarvey and Bonnie Watson Coleman cited +PolicyEngine’s analysis in introducing the Young Adult Tax Credit Act +(H.R.7547), stating that “according to the model at PolicyEngine, 22% of +all Americans would see an increase in their household income under this +program, and it would lift over 4 million Americans out of poverty” +(Office of +Representative Morgan McGarvey 2024). In the UK, Baroness Altmann +referenced PolicyEngine and its interactive dashboard during House of +Lords Grand Committee debate on the National Insurance Contributions +(Employer Pensions Contributions) Bill in February 2026, noting that +Commons Library research using PolicyEngine provided “a useful picture +of the distributional effects of raising the contribution limit” across +income deciles (House +of Lords 2026).
+Institutional partnership. PolicyEngine and the +National Bureau of Economic Research (NBER) signed a formal memorandum +of understanding for PolicyEngine to develop an open-source TAXSIM +emulator — a drop-in replacement for TAXSIM-35 powered by PolicyEngine’s +microsimulation engine, with support for Python, R, Stata, SAS, and +Julia (Feenberg +2024). The Federal Reserve Bank of Atlanta independently +validates PolicyEngine’s model through its Policy Rules Database, +conducting three-way comparisons between PolicyEngine, TAXSIM, and the +Fed’s own models (Federal Reserve Bank of Atlanta +2021). Co-author Max Ghenis and Jason DeBacker (University of +South Carolina) presented the Enhanced Current Population Survey +methodology at the 117th Annual Conference on Taxation of the National +Tax Association (Ghenis and DeBacker 2024).
+Academic research. Matt Unrath (University of +Southern California) is using PolicyEngine in a study of effective +marginal and average tax rates facing American families, funded by the +US Department of Health and Human Services through the Institute for +Research on Poverty (Institute for Research on Poverty 2025). The +Beeck Center at Georgetown University featured PolicyEngine in research +on rules-as-code for US public benefits (Kennan et al. 2023, 2025). +Youngman et al. (Youngman et al. 2026) cite +PolicyEngine UK’s microdata methodology in their agent-based +macroeconomic model for the UK’s Seventh Carbon Budget at the Institute +for New Economic Thinking, Oxford.
+Policy research. In the US, the Niskanen Center used +PolicyEngine to estimate the cost and distributional impacts of Child +Tax Credit reform options (McCabe and Sargeant 2024). DC +Councilmember Zachary Parker cited PolicyEngine’s analysis when +introducing the District Child Tax Credit Amendment Act of 2023, the +first local child tax credit in US history (Council of the District of Columbia 2023). +Senator Cory Booker’s office embedded a PolicyEngine-built calculator on +his official Senate website for the Keep Your Pay Act (Office of Senator Cory +Booker 2026). In the UK, the National Institute of Economic and +Social Research (NIESR) used PolicyEngine in their UK Living Standards +Review 2025 (Mosley +et al. 2025), and the Institute of Economic Affairs has published +PolicyEngine-based analyses of employer National Insurance contributions +and 2025–2026 tax changes (Woodruff 2024, +2025).
+Acknowledgements
+This work was supported in the US by Arnold Ventures (Arnold Ventures +2023), NEO Philanthropy (Ghenis 2024a), the Gerald Huff +Fund for Humanity, and the National Science Foundation (NSF POSE Phase +I, Award 2518372) (National +Science Foundation 2025), and in the UK by the Nuffield +Foundation since September 2024 (Nuffield Foundation 2024). These +funders had no involvement in the design, development, or content of +this software or paper.
+We acknowledge contributions from all PolicyEngine contributors, and +thank the OpenFisca community for the foundational microsimulation +framework (OpenFisca +Contributors 2024). We acknowledge the US Census Bureau for +providing access to the Current Population Survey, and the UK Data +Service and the Department for Work and Pensions for providing access to +the Family Resources Survey. We acknowledge the UKMOD team at the +Institute for Social and Economic Research (ISER), University of Essex, +for their contributions to model descriptions (Sutherland and Figari +2013).
+AI Usage Disclosure
+Generative AI tools — Claude Opus 4 by Anthropic (Anthropic 2026) — were +used to assist with code refactoring. All +AI-assisted outputs were reviewed, edited, and validated by human +authors, who made all core design decisions regarding software +architecture, policy modeling, and parameter implementation. The authors +remain fully responsible for the accuracy, originality, and correctness +of all submitted materials.
+