Emulator of the OBR macroeconometric model
OBR emulator
Scores second-round GDP effects of household-income reforms; behaviour checked for income tax and NICs only.
Good for
- Income tax and NICs changes: the only reforms with checked behaviour
- Other household-income reforms: they run, but unchecked
- GDP and consumption by quarter against the official path; investment responds behaviourally only under the corporation-tax closure
Not for
- Who gains and loses: PolicyEngine microsimulation
- Long-run structural change: OG-UK
- Where the economy is today: BoE SVAR
- Bank Rate: the shock runs, but the response is about 3–6× below the Bank of England’s range (limits)
- Corporation tax: raw shock only; refused through the reform bridge; no benchmark
- Spending multipliers, capital spending, borrowing (computed by the model, not returned by
obr_shock) or prices (CPI does not respond to rate or oil shocks): limits
A worked example
Four example shocks from 2025Q1 over 12 quarters. GDP is the change from baseline.
1p on the basic rate of income tax
- The microsimulation’s costing lowers household income, so consumption falls
- Our effect grows; the OBR’s fades
obr_shockreports real GDP, where the q12 ratio is 0.33; in current prices it is 0.37 (why two bases)
Which costing£1,637.5m a quarter, the current costing
Run on the microsimulation’s current costing of £6.55bn a year (£1,637.5m a quarter). The response is linear in shock size, so the ratios do not depend on the costing.
Government consumption up £1.25bn a quarter
By construction GDP rises by about the shock: accounting, not a behavioural multiplier (why). In the model, borrowing rises by the cash cost (+£1,336m at q1, +£1,419m at q12) and receipts do not respond; obr_shock does not return borrowing.
Bank Rate up 1 percentage point, sustained
Only consumption and interest flows respond, so the fall is about 3–6× below the Bank of England’s range for a four-quarter rise (benchmark, why).
Corporation tax main rate up 5 points
Corporation tax: raw shock only; refused through the reform bridge; no benchmark (obr_shock, TCPRO, through the cost of capital); far from its plateau at q12.
How far to trust it
Three checks, weakest included; each opens its evidence.
- Anchored fit, real GDPHeld to the EFO by add-factors, so not skill; drifts to +0.42% by 2027Q40.24% MAPEby construction
- Held add-factors, real GDP2026Q1–2027Q4; 7 of 8 in band, two of them identities on EFO inputs0.63% MAPEweak
- Free-running, real GDPThe emulator’s own dynamics, not OBR forecast error4.24% MAPEweak
Read the direction and scale, not single quarters. These are point estimates with no confidence bands, and several channels do not transmit (all limits).
Run it
The income-tax example as a reform: 1p on the basic rate. The hosted run uses the current five-year window, so its figures differ from the example’s 2025 start, and it takes over ten minutes.
Score a 1p rise in the UK basic rate of income tax with the OBR emulator and show the change in GDP each quarter.The hosted server calls score_reform with model="obr"; no install. Connect
pe-macro score --country uk \
--reform '{"gov.hmrc.income_tax.rates.uk[0].rate":0.21}' \
--model obrfrom policyengine_macro import score_reform
result = score_reform("uk", {"gov.hmrc.income_tax.rates.uk[0].rate": 0.21}, "obr")With the microsimulation
Static costing in, second-round effects out.
A reform is costed first, then fed in as lower household income. Mechanics · every join
Model cardinputs, outputs, horizon, data, access
- Answers
- The second-round effect of an income tax or NICs change on GDP and consumption; investment responds behaviourally only under the corporation-tax closure
- Inputs
- A PolicyEngine microsimulation static costing, or a curated shock to one model variable
- Outputs
- GDP and consumption (investment responds behaviourally only under the corporation-tax closure): deviations from the anchored baseline
- Horizon
- Quarterly, typically 3-5 years
- Data
- October 2026 EFO baseline · October 2025 equation listing · vendored ONS snapshot
- Uncertainty
- Not comprehensive: read direction and scale, not single quarters
- Access
- Hosted MCP, CLI, Python
- Runtime
- About one to five minutes for a raw shock on the hosted server; over ten minutes for a translated reform on the default five-year window
- Joins the microsimulation
- The static costing enters as a household income add-factor (the join above)
- Status
- Anchored to the October 2026 EFO, not validated as a forecaster; predictive and policy checks weak
- Cannot answer
- Borrowing (not returned by the adapter), statutory incidence, capital spending, corporation tax through the bridge (limits)
Measured at model pin 0e44ec8 (obr-macroeconomic-model, October 2026 EFO), every quarter converged. Reproduce on the Use it tab.