Green Ambition, Allocation Friction: A Simulation-Calibrated Study of Overinvestment and Underinvestment in Vietnamese Banks

<jats:p>Investment inefficiency is commonly measured as the absolute residual from an expected-investment model, but absolute values can conceal whether a financing variable alleviates underinvestment while simultaneously encouraging overinvestment. This study develops an asymmetric investment- efficiency analysis for banking institutions operating in Vietnam. The source workbook contains a strongly balanced panel of 49 institutions over 2008-2024; the complete-case regressions use 503 observations from 39 institutions. Investment inefficiency is derived from deviations of asset growth from expected investment based on lagged operating-income growth. We decompose the signed residual into overinvestment and underinvestment severity and estimate bank and year fixed-effects models with bank-clustered standard errors. Operating cash flow is insignificant in the conventional absolute-inefficiency equation, but the decomposition reveals large offsetting effects: cash flow increases overinvestment severity (coefficient 0.694, p &amp;lt; 0.001) and reduces underinvestment severity (coefficient -0.624, p &amp;lt; 0.001). A green-patent proxy is associated with overinvestment (0.027, p = 0.017) but not underinvestment, whereas the ESG proxy is insignificant in both equations. These sustainability variables are explicitly labelled as simulation-calibrated in the supplied workbook, so their coefficients are exploratory rather than evidence about observed patents or commercial ESG ratings. Estimator comparisons further show that the green-innovation coefficient disappears under FGLS and system GMM, whose large instrument count limits causal interpretation. The results demonstrate that aggregation can produce a misleading null and that bank investment-efficiency studies should separate overinvestment from underinvestment, discipline internal liquidity, and use verified sustainability data.</jats:p>