PolicyEngine Macro
US · what the platform produces
Score a US reform. The macro half stops here.
The microsimulation covers the US exactly as it covers the UK: a reform goes in, and household incomes, revenue and a distribution come out. What the UK side adds — a macro second round, and a forecast archived before its outturn — has no US counterpart, and this page says so rather than manufacturing one.
Four outputs, and the two that do not exist.
The engine underneath this section is pe-microsim — the only one that covers both countries, so the two household rows below are the same command as the UK ones with --country us. It reports taxes, benefits, net_income, revenue, distribution over a single policy year.
| What you get | Models, in the order they run | What the chain does | Command |
|---|---|---|---|
| One household, exactly | pe-microsim | arithmetic over the statutory rules; no sampling, no weights | pe-macro household-impact --country us --people '[{"age":35,"employment_income":50000}]' --reform '{"gov.irs.credits.ctc.amount.adult_dependent":1000}' |
| The whole population, one policy year | pe-microsim | revenue and distribution; a static costing | pe-macro score --country us --reform '{"gov.irs.credits.ctc.amount.adult_dependent":1000}' --model microsim |
| A macro shock carried into household earnings | frb-us → pe-microsim | the wage-bill change applied through the automatic stabilisers | pe-macro frbus-shock-incidence --var rffintay_aerr --shock 1.0 --year 2027 |
| The same question in the HANK member | us-hank → pe-microsim | impulse responses around a calibrated steady state | pe-macro hank-shock-incidence --kind monetary --size -0.0025 --year 2026 |
Two rows that exist on the UK side are missing here, and both absences are the registry's, not an oversight:
- No reform with a macro second round. Both US members — frb-us and us-hank — record PolicyEngine reforms in their own cannot-answer field, and us-hank adds detailed tax reforms (only monetary/fiscal-spending/productivity instruments; the labor tax is endogenous). No mapping exists from a US statutory reform to a US macro model, and none is invented here.
- No forecast. No model in this suite forecasts the US economy. The only member whose question types include forecast is boe-svar, and its geography is UK-only. A section called Forecasts that showed a US path anyway would be the one dishonest thing on this site.
The bridge that does exist runs the other way: the incidence rows take a macro shock, not a reform, and push its earnings consequences through the microsimulation. So this section can answer “who bears this shock” for the US, and cannot answer “what would this reform do to output”.
Over MCP the four routes are household_reform_impact, population_reform_impact, frbus_shock_incidence and hank_shock_incidence; recommend_model refuses a US reform needing macro feedback, which is the correct answer. Connect a client →
The US models do not forecast.
Archiving “forecasts” from models that do not claim to forecast would manufacture a record. The UK page exists because boe-svar genuinely forecasts and accepts being scored; neither US model does.
What each US model does claim
frb-us is a validated software replication of the Fed’s model — it answers “what does a shock do”, and its validation makes no predictive claim. us-hank is a calibrated validated replication of Auclert et al. (2021); its responses are first-order and stylized, not forecasts.
The Fed’s path, published for comparison.
The closest thing to a US forward path here is the FRB/US LONGBASE conditioning baseline — the Fed-style near-term path that frb-us shock experiments deviate from. It is published for orientation, clearly not ours: no uncertainty bands, and it will never be scored on this page.
| Quarter | Real GDP, y/y | CPI, y/y | Unemployment |
|---|---|---|---|
| 2026Q2 | 2.5% | 3.2% | 4.4% |
| 2026Q3 | 2.0% | 3.2% | 4.4% |
| 2026Q4 | 2.4% | 3.2% | 4.4% |
| 2027Q1 | 2.4% | 3.0% | 4.4% |
| 2027Q2 | 2.4% | 2.8% | 4.3% |
Extracted from the Fed’s own package data by the adapter
(longbase_baseline_yoy.csv);
the US data page shows the same baseline
beside each outturn.
Current US data and model responses.
The US data page tracks the official data — growth, inflation, unemployment, yields — with a committed vintage behind every number, and frb-us answers shock questions on top of the Fed’s own baseline via the hosted tools and CLI.
The bar does not move.
A US track record starts the day a model here makes a genuine predictive claim: rounds committed to Git before the outturn, append-only under the same CI gate as the UK record, and scored against realised data whatever that shows. Until then this page stays empty rather than pretend.