Exchequer

The UK OBR macroeconomic model, rewritten in Python.

A full Python translation of the Office for Budget Responsibility's ~370-equation simultaneous macroeconomic model — the engine behind every UK Budget forecast. Originally written in EViews, now open and hackable.

What is this?

The OBR model is the forecasting tool used by the UK's independent fiscal watchdog to produce economic and fiscal projections at each Budget. It covers:

• GDP and expenditure — consumption, investment, trade, inventories
• Labour market — employment, wages, unemployment, productivity
• Prices — CPI, RPI, import/export deflators, input-output cost structure
• Public finances — tax revenues, spending, borrowing (PSNB), debt (PSND)
• Balance of payments — current account, capital flows, exchange rates
• Financial balance sheets — household wealth, corporate liabilities, international investment position
• North Sea oil — production, trade, and fiscal revenues

Quick Start

# Install dependencies
pip install -r requirements.txt

# Run with synthetic placeholder data
python run_model.py

Usage

from obr_model.model import OBRModel

# Create and run the model
model = OBRModel()
model.generate_synthetic_data()       # or model.load_data(your_dataframe)
model.solve('2025Q1', '2030Q4')

# Inspect results
model.summary('2026Q4')
gdp = model.get_variable('GDPM', start='2025Q1')
print(gdp)

# Export
model.export_results('results.xlsx')  # or .csv

Architecture

obr_model/
    config.py                # Solver parameters and constants
    eviews_functions.py      # EViews-to-Python translation helpers
    data_generator.py        # Synthetic placeholder data (~600 variables)
    solver.py                # Gauss-Seidel iterative solver
    model.py                 # Main OBRModel orchestrator
    equations/
        group01_consumption.py    # Household consumption and durables
        group02_inventories.py    # Stockbuilding
        group03_investment.py     # Business, housing, and public investment
        group04_labour.py         # Employment, unemployment, participation
        group05_exports.py        # Goods and services exports
        group06_imports.py        # Goods and services imports
        group07_prices.py         # Wages, costs, CPI, RPI, deflators
        group08_northsea.py       # North Sea oil production and trade
        group09_public_exp.py     # Government spending
        group10_receipts.py       # Tax revenues
        group11_bop.py            # Balance of payments and investment income
        group12_public_totals.py  # Fiscal aggregates (PSNB, PSND, debt)
        group14_financial.py      # Interest rates, money, equity prices
        group15_income.py         # Household income, saving, sectoral balances
        group16_gdp.py            # GDP identities and market sector GVA
        group18_balance_sheet.py  # Household, corporate, and RoW balance sheets

How the Solver Works

The model is a system of ~370 simultaneous equations. For each forecast quarter:

1. Initialise endogenous variables from the previous quarter
2. Pre-solve the SCOST/CCOST/UTCOST input-output cost block (3×3 linear system)
3. Iterate through all equations (Gauss-Seidel with damping), updating each variable in place
4. Converge when max relative change < 1e-8 (typically 50-60 iterations)
5. Apply additive adjustments for forecast alignment

Using Real Data

The synthetic data lets you test the structure. To use real data:

1. Build a pandas DataFrame with a quarterly PeriodIndex (1970Q1 onwards)
2. Populate all ~150 exogenous variables (interest rates, oil prices, exchange rates, demographics, policy settings)
3. Populate historical values for all endogenous variables
4. Set additive adjustment variables (_A suffix) for near-term forecast alignment
5. Call model.load_data(df) instead of model.generate_synthetic_data()

Key data sources: ONS (national accounts), Bank of England (rates), OBR databank (fiscal), HMRC (tax receipts).

Key Variables

Variable	Description
GDPM	Real GDP
GDPMPS	Nominal GDP
CONS	Real household consumption
IF	Real gross fixed capital formation
CPI	Consumer Price Index
LFSUR	ILO unemployment rate (%)
PSNBCY	Public sector net borrowing
PSND	Public sector net debt
CB	Current account balance

Acknowledgements

Based on the OBR's published model equations. The original EViews model is published by the OBR for transparency purposes.