The Unidad de Fomento as its own currency in the standardised measures of interest rate risk in the banking book: what each component adds

The Chilean text on banking-book interest rate risk treats flows indexed to the Unidad de Fomento (UF) as a currency of its own: its own zero curve, shock sizes and floor, a charge of 2.5 percent of the net UF position, and a pair rule offsetting half of one leg's gain. We implement it as five switches on an open engine and attribute its difference from a single-currency treatment by the exact Shapley decomposition on 30 synthetic books per variant over the Banco Central de Chile curves of 2026-09-03. With a net UF position of 30 percent of level 1 capital, treating the UF as pesos gives a delta NII 4.9, 5.5 and 5.7 times smaller (ratios of means; bootstrap 95 percent intervals within 4.5 to 6.6) and a delta EVE 5.7±0.8, 15.4±0.8 and 24.2±1.1 percent smaller. The charge supplies 52 to 53 percent of the delta NII; for the delta EVE the own curve and the own sizes nearly cancel, refuting that the pair rule dominates. A UF floor made to bind raised the delta NII in 208 of 960 comparisons, refuting that it never does. The treatments classify more than half of the books differently only at 110 to 150 percent of level 1 capital (the top of the range tested), and single books from 80 percent; the prediction of a position flagged by Annex 1 in most books and by the single-currency treatment in none held at 110 percent only. Repository and reproducible pipeline (MIT): https://github.com/fsantibanezleal/CAOS_Contraste .