[LiangShi369]

This update adds age-specific log-income slopes and a reference-adult student indicator, retaining log broad expenditure, the existing 14 age groups, common quadratic curvature and other controls.

The updated input/regression_files/reg_wealth_financial.xlsx, sheet FW1a, contains all 65 coefficients and the full covariance matrix. The matching RMSE is 0.6457928727473985 in input/regression_files/reg_RMSE.xlsx, sheet UK, entry FW1a. Other regression processes and the cash-income and donor-matching algorithms are unchanged.

We also revised the LCFS dataset used for estimation and validation:

1. Benefit-unit own income now sums members’ personal net incomes, replacing the previous allocation of household income by expenditure shares.
2. The same allocated Council Tax—or domestic rates in Northern Ireland—is deducted from both benefit-unit income and broad expenditure, aligning their local-tax boundaries with the disposable-income measure used by SimPaths.

@justin-ven: The UKHLS initial-population scripts already aggregate constructed personal incomes within benefit units. Could you advise whether their disposable-income measure also needs adjustment for Council Tax/domestic rates, and whether this would require regenerating the initial populations?

The preparation and estimation scripts, validation materials and detailed documentation are maintained separately on our Box drive:

• SimPathsUK\modules\wealth\03 consumption\collect_lcfs.do: revised LCFS dataset construction.
• SimPathsUK\modules\wealth\03 consumption\consumption_modelling_lcfs.do: consumption estimation, exporting model inputs and reporting outputs to the corresponding output_ folders.

I am completing the wealth-module documentation, including the revised consumption-modelling section and a new consumption-validation section.

[Copilot]

🔵 Needs a closer look

Synthetic prediction parity is supported, but end-to-end simulation effects of the recalibrated consumption model remain unverified.

No open issues 🎉

What changed in this PR

Updates SimPaths’ consumption regression support for the revised model, adding age-specific income slopes and a reference-adult student indicator while retaining common quadratic curvature.

Changes:

• Adds age–income interactions, student status and a centred year trend that preserves the existing freeze.
• Adds 700 synthetic Stata fixtures and a script to regenerate them.
• Adds regression parity tests and checks for age boundaries, income flooring and trend freezing.

File	Description
src/​test/​stata/​export_consumption_fixtures.do	Exports deterministic predictions from saved Stata estimates.
src/​test/​resources/​simpaths/​model/​consumption_stata_fixtures.csv	Supplies 700 reference prediction cases.
src/​test/​java/​simpaths/​model/​ConsumptionRegressionTest.java	Checks prediction parity and regressor boundary behavior.
src/​main/​java/​simpaths/​model/​BenefitUnit.java	Exposes the revised consumption regressors.

_{Review effort: Balanced}

[dav-sonn]

@LiangShi369: the check on the integration tests is failing. If the implemented changes are meant to change the outputs, then the integration tests should be updated accordingly.