diff --git a/CHANGELOG.md b/CHANGELOG.md index 761d6822..edbcd6eb 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -10,6 +10,11 @@ heading here; `tests/test_repo_hygiene.py` asserts that they do. ## [Unreleased] +### OBR page: fiscal rules view and conventions + +- **Fiscal rules on `/obr`.** A new "Against the fiscal rules" section sets the October 2026 EFO path for the current budget (Table 6.5) and public sector net financial liabilities (derived from Table 6.14) beside what the model books for the four worked examples (`results/demo_shocks.json` at obr-macroeconomic-model `0e44ec8`). Only the spending example moves the public finances; the section says it is not a rule readout. No existing figure changes. +- **Conventions on `/obr/methodology`.** Step 2 gains a table of the OBR conventions the emulator inherits as EFO Table 1.9 input paths, imposes through an opt-in overlay, or does not reproduce. + ### Model re-pin and site review (#278, #279) - **#278: model pins.** OBR `2965e82d` → `0e44ec87` (main after obr-macroeconomic-model #74 and #75), BoE SVAR `c13a9c91` → `c827ab86`, FRB/US `846eeb5c` → `162cfe6a`, US HANK `551a40d3` → `58353e5f`. After a redeploy the hosted tools run these pins. The FRB/US and US HANK compare rows with stored paths re-run identically; the OBR rows were re-run at `0e44ec8`. diff --git a/DESIGN.md b/DESIGN.md index 696ba827..501d8572 100644 --- a/DESIGN.md +++ b/DESIGN.md @@ -742,7 +742,7 @@ row (entry cards or chips): the next tab, `/models#compare`, and `/connect` for | Tab | Layout | Must carry | |---|---|---| -| Overview | lede (one sentence: what it answers) → `#verdict` strip (§4.14) → `#fit` good for / not for → `#answers` "What it returns": worked examples behind a scenario switch, stat tiles + sparkline (home of those figures) → `#pe-bridge` the join as a pipeline, one line, and the page's **one** caveat callout → `section#card` model card, collapsed (§4.13) → chip row. No StepPanel, no "About this step" expanders. | link to `/models#compose`, Validation caveats; every other caveat lives on Validation | +| Overview | lede (one sentence: what it answers) → `#verdict` strip (§4.14) → `#fit` good for / not for → `#answers` "What it returns": worked examples behind a scenario switch, stat tiles + sparkline (home of those figures) → on `/obr` only, `#fiscal-rules`: the EFO path table, the examples table and one sources expander → `#pe-bridge` the join as a pipeline, one line, and the page's **one** caveat callout → `section#card` model card, collapsed (§4.13) → chip row. No StepPanel, no "About this step" expanders. | link to `/models#compose`, Validation caveats; every other caveat lives on Validation | | Methodology | StepPanel, one step per model block; panel = equations (`.meth-eq`) | the model's source paper citation (link to Paper tab) | | Validation | stat row (the headline fit numbers: their home) + trust card (§4.13) → StepPanel of the tests (OBR is the reference) → limits card (`#caveats`, every failure statement in full, as `.limits` groups of `details.expander.limit` rows with a visible `h3` per group; r3) → "On this page" rail ≥ 1100 px | the verdict, every known limit, links to the code that produced each number | | Code | `.scrolly-code` StepPanel: install → run → reproduce; panel = commands | `/connect` for hosted use, repo paths | diff --git a/obr/index.html b/obr/index.html index f6407ef3..32a8322e 100644 --- a/obr/index.html +++ b/obr/index.html @@ -157,6 +157,49 @@
The October 2026 EFO path for the two rules, beside what the model books for the four examples above. Not a rule test: the examples end in 2027Q4, before the target year.
| Financial year | 2026-27 | 2027-28 | 2028-29 | 2029-30 | 2030-31 | 2031-32 |
|---|---|---|---|---|---|---|
| Current budget, £bn (surplus +) | −40.6 | −17.5 | −7.8 | +14.5 | +21.7 | +20.9 |
| PSNFL, % of GDP (derived) | 83.3 | 83.7 | 83.5 | 82.6 | 81.6 | 80.9 |
| PSNFL change, pp of GDP (derived) | +1.34 | +0.39 | −0.20 | −0.86 | −0.98 | −0.74 |
| Example | Receipts | Current spending | Net borrowing | What is and is not counted |
|---|---|---|---|---|
| Spending up £1.25bn a quarter | no change | +£5.7bn | +£5.7bn | The cash cost only; receipts do not respond to the higher GDP |
| 1p on the basic rate | no change | no change | no change | The £6.55bn a year costing is the microsimulation’s input; it enters as lower household income, not as model receipts |
| Bank Rate up 1 point | no change | no change | no change | Debt interest is an input. Checked on the four-quarter run; the sustained run records no fiscal lines |
| Corporation tax up 5 points | no change | no change | no change | The rate moves the cost of capital; no receipts are booked |
Not a rule readout Only the spending example moves the public finances in the model, by its cash cost, so it lowers the current budget by £5.7bn in 2027. The hosted obr_shock returns none of these lines, and no run reaches 2029-30.
PSCR (receipts), PSCE (current spending) and PSNBNSA (net borrowing) changes in results/demo_shocks.json, rows cgg, household_costing, bank_rate and corporation_tax_rate. The current budget is receipts less current spending less depreciation.Three checks, weakest included; each opens its evidence.
The snapshot ships inside the package, so there is no download step. It gives the circular fiscal and financial sub-blocks real balancing data rather than guessed starting values.
+ +A convention is a rule the OBR applies outside its equations, such as taking interest rates from financial markets. The emulator handles each in one of three ways.
+run_reform, off by default.| Convention | In the emulator | Status |
|---|---|---|
| Bank Rate and gilt yields from market curves | Table 1.9 paths for Bank Rate (R) and the 20-year gilt yield (RL). No curve is built, and neither responds to a fiscal shock | inherited |
| Mortgage and deposit rates | Table 1.9 paths (RMORT, RDEP). monetary_conditioning moves the mortgage rate with Bank Rate (Bank Rate limits); the deposit rate, gilt yields, sterling and house prices stay fixed | inherited opt-in overlay |
| Oil price from futures | Table 1.9 path (PBRENT). Its note: Brent spot for outturn, ICE Brent futures over the forecast window | inherited |
| Gas price from futures | Table 1.9 path, loaded as PGAS but used by no equation in the listing, so it has no effect. Its note: one-month-forward gas price expectations averaged over the forecast window | absent |
| Exchange rate held flat | Table 1.9 path (RX): the sterling index is constant at 85.7 from 2026Q4 to 2032Q1. Sterling moves trade prices only, never volumes | inherited |
| Fiscal multipliers imposed by judgement, fading to zero over five years | published_conventions imposes the OBR’s impact multipliers on government consumption and investment (multiplier, capital spending). The straight-line fade between the two published points is our choice | opt-in overlay |
| Indirect taxes passed through to prices | price_passthrough, for VAT and fuel duty shocks (VAT limits) | opt-in overlay |
| Multiplier fade dated from the announcement | The fade starts at the shock start; the emulator has no announcement date | absent |
| Quarter-by-quarter forecast judgement | Recent add-factors are averaged and held flat (limits) | absent |
| Fresh market data between EFOs | The paths are those of the pinned EFO; no curve date or averaging window is applied or updated | absent |
obr_shock exposes price_passthrough only. published_conventions and monetary_conditioning need the model package.published_conventions.py; monetary_conditioning.md; finding G14 of the published-model review.