Institutional and Digital Predictors of Private Sector Credit in Azerbaijan: An Elastic Net Approach

Private sector credit is central to economic diversification in resource-rich economies, yet the joint informational role of institutional quality and digital connectivity remains insufficiently examined. This study analyzes the predictors of domestic credit to the private sector in Azerbaijan over 1995–2025 using elastic net regression, which selects informative predictors among correlated candidates. Seven institutional, digital, external-sector, and macro-financial predictors are considered. Internet usage and regulatory quality carry most of the predictive information, while government effectiveness and net FDI inflows are largely redundant given these two variables. The baseline model explains 79.8% of the in-sample variation in credit (RMSE = 4.35 percentage points of GDP). Time-respecting cross-validation selects a sparser lasso-type specification that retains the same two leading predictors. The results hold after linear detrending but weaken under quadratic detrending and in first differences, and persistence-based benchmarks outperform the model in one-year-ahead forecasting. Bounds tests indicate a long-run relationship between credit, internet usage, and regulatory quality. The findings therefore describe long-run co-movement in credit depth rather than short-run dynamics or causal effects. Results also suggest that digital and institutional conditions are jointly informative about credit development in Azerbaijan.