Shokhrukh Komilov
PhD candidate, (Mamun University)
E-mail: shokhrukhkomilov1@gmail.com
ORCID: 0009-0001-2739-1852
Abstract. Aggregate liquidity indicators for Uzbekistan’s banking sector are strong — the liquidity coverage ratio (LCR) stood at 208% at the start of 2026 — yet aggregates can conceal bank-level fragility. This paper develops a Deposits-at-Risk (DaR) framework for the 30-day run-off horizon used in liquidity regulation. Using 88 monthly observations (January 2019 – April 2026) of household and corporate deposits in national and foreign currency compiled from Central Bank of Uzbekistan statistical bulletins, one-month DaR is estimated with four models and 204 out-of-sample forecasts are backtested with the Kupiec and Christoffersen tests. Deposit-flow shocks are skewed towards inflows and have a thin lower tail (GPD ξ = −0.36); consequently, symmetric normal models overstate run-off risk and fail coverage tests, whereas filtered historical simulation with an extreme-value tail delivers accurate 99% coverage (0.98% against 1%), and historical simulation passes all fourteen robustness tests. Corporate foreign-currency deposits are the most volatile segment, having fallen by 40% in US-dollar terms in the first half of 2023. A Monte Carlo simulation of a synthetic 34-bank system calibrated to published concentration and liquidity data shows that, in the system’s worst 1% of months, at least one bank exhausts its usable liquid assets with a probability of 18%, rising to 88% if digital channels triple the speed of household withdrawals. At bank level, household run-off can exceed the Basel III assumption of 10%.
Keywords: Deposits-at-Risk; liquidity risk; bank runs; extreme value theory; backtesting; Monte Carlo simulation; Uzbekistan.
Skip to PDF content