# Figure 1 R1 assumptions and exclusions

- All four rows use the same reported Table 3 Panel B population (`N = 14,623`).
- Displayed percentages retain the paper's two-decimal rounding. Validation recomputes raw rates from reported counts.
- Percentage-point differences use the issuer-reported realization plus issuer-reported threshold row as the baseline.
- The threshold-only and realization-only differences are not assumed additive. Their sum exceeds the both-proxy difference by 2.65 percentage points after rounding.
- The source pairings are observational measurement alternatives, not randomized treatments or a causal decomposition.
- No microdata, transition matrix, legal adjudication, market-wide inference, EOJI performance result, or professional-benefit claim is supplied.