Publication Details
Abstract
This study empirically examines the relationship between the volume of lending to individuals and the quality of the retail loan portfolio in the Republic of Uzbekistan, using monthly data from January 2017 to December 2025 (n = 108) obtained from the Central Bank of the Republic of Uzbekistan. A Structural Vector Autoregression (SVAR) model was constructed with seven variables: the natural logarithm of household credit outstanding (LHHC), the Central Bank policy rate (CBR), the weighted average interest rate on loans to individuals (WACR), annual inflation (INF), the logarithm of real household income (LRINC), the logarithm of the money supply (LM2) and the share of non-performing loans (NPL). Preliminary tests confirmed that all series are I(1), that 2–3 cointegrating equations exist among them, that the optimal lag is 2 months and that six variables Granger-cause LHHC. The estimated SVAR model shows very high explanatory power (R² = 0.954; Adj. R² = 0.946; F = 112.847, p < 0.001; DW = 1.984). Impulse-response and variance-decomposition analyses reveal that real household income (LRINC) and the money supply (LM2) exert the strongest and most persistent positive effects on retail lending (contributing 15.32% and 15.24%, respectively, to the 12-month variance of LHHC), while the policy rate, the market lending rate, inflation and NPL exert statistically significant negative effects. The bidirectional causality between LHHC and NPL empirically confirms the “credit expansion accumulates risks” hypothesis in the Uzbek context. The findings provide a quantitative basis for calibrating the Central Bank’s inflation-targeting regime, refining commercial banks’ credit-scoring and IFRS 9 expected-loss models, and designing early-warning systems that jointly safeguard portfolio quality and sustainable credit growth.