TESTING A PUBLISHED PROVISION MATRIX: EVIDENCE FROM AN UZBEK ELECTRICITY DISTRIBUTOR

Abstract

IFRS 9 permits the use of a provision matrix for trade receivables but requires its loss rates to be reviewed against actual credit loss experience. External users cannot conduct this review because published financial statements disclose balances and loss rates by ageing bucket rather than movements of balances between buckets. This article examines how much information a published provision matrix reveals about its own performance. Using only the IFRS financial statements of Regional Electric Networks JSC, an Uzbek electricity distributor, as of 31 December 2025 and 30 June 2026, the study converts loss rates into implied monthly roll probabilities and compares the proportion of balances aged up to 150 days expected to remain unpaid six months later with the observed proportion. The model implies 34% for legal entities and 40% for individuals, whereas the observed proportions range from 48% to 76% and from 45% to 64%, respectively. Expected transfers into the over-360-day ageing bucket exceed the observed change by UZS 115–182 billion for legal entities and approximately UZS 86–88 billion for individuals. The 2026 revisions increased the loss rates for legal entities and decreased those for individuals; therefore, the direction of the net effect on the loss allowance cannot be determined. The financial statements do not disclose a roll-forward by ageing bucket, write-off criteria, or the reasons for the revisions. The findings represent bounds derived from publicly available aggregate data for a single company over a six-month period and do not constitute evidence of financial misstatement.