Methodology v10.16

Four baseline lines were rewired onto CBO current-law series. Social Security now reads CBO mandatory outlays (OASI plus DI) instead of a linear path; defense and nondefense discretionary read the CBO annual series instead of flat shares of GDP; corporate reads CBO current-law receipts instead of a maturity anchor. The four do not point the same way — Social Security was understated and defense overstated — so this was a correspondence repair, not a thumb on the scale.

Healthcare was being subtracted twice. The obligation was booked net of displaced current-law federal health spending, against a baseline that had already been residualized — so the same displaced dollars came out of both sides. Two independent reconstructions agreed on the size of the error to within $8B. Healthcare is now booked gross, and the general rule is written down: book gross against a residualized parent, net against an unreduced one.

The calibration boundary was made continuous. The frozen calibration tables cover 2030–2089. Past them, one half of the calibration identity kept growing from its formula while the other silently returned zero, which opened an $18.5T single-year gap at 2090 and made the stress lab report both published scenarios as relapsing into debt with no stress applied. All four tables now hold their last covered row flat, and a build guard fails on any discontinuity at the boundary. Years past 2089 are continuous but not calibrated, and no published figure is taken from them.

What is still open. The healthcare cost basis is under bottom-up rederivation and is the largest single line in the model, so no retirement year is published while it is open. The figures on this site are provisional in that specific respect and say so where they appear. See /scoring/detail and the master tables.