Type of the article: Research Article
Abstract
This study aims to examine whether the association between lagged loan-to-deposit ratios and bank profitability varies across realized real refinancing-rate environments after accounting for earnings persistence and common annual conditions. A balanced panel of 15 Vietnamese commercial banks over 2015–2025 yields 150 observations for 2016–2025. Bias-corrected fixed-effects moment estimation is combined with a finite-sample covariance adjustment and bank-cluster t-inference. Profitability is persistent, with lagged ROA and ROE coefficients of 0.6572 and 0.6272. The real-rate interaction for ROA is positive (0.00258), but its p-value increases from 0.0207 under asymptotic inference to 0.0516 after the small-sample adjustment. In a directly comparable model, nominal-rate and inflation interactions are individually imprecise, and their equal-and-opposite restriction is not rejected (p = 0.8567). The Johnson–Neyman analysis identifies no statistically non-zero conditional LDR slope within the observed real-rate range at the 5% level. The positive ROA interaction persists in the shorter sample and after excluding VPB, but adjusted p-values rise to 0.1714 and 0.1486. The ROE interaction is unsupported. The evidence is therefore suggestive of a changing LDR–ROA association, with substantial uncertainty, rather than a robust monetary-policy effect. Limited annual variation, strong regressor-exogeneity assumptions, and incomplete asset-quality and repricing information constrain interpretation.
Acknowledgment
The bank-level data used in this study were compiled from the publicly available annual reports, audited financial statements, and official disclosures of 15 Vietnamese commercial banks: ACB, BIDV, VietinBank, Eximbank, HDBank, MB, MSB, OCB, SHB, Sacombank, Techcombank, TPBank, Vietcombank, VIB, and VPBank. The authors also acknowledge the State Bank of Vietnam and official statistical agencies for providing the monetary-policy and macroeconomic data employed in the analysis. The institutions mentioned above were not involved in the research design, data analysis, interpretation of the findings, or preparation of the manuscript.