Macroeconomic Shocks, Credit-Risk Persistence, and the Limits of Nonlinear Transmission in Emerging Europe
Abstract
This paper examines how economic downturns and currency movements affect the quality of bank loans in Central, Eastern, and Southeastern Europe, using annual data for 14 national banking systems over 2008–2023. We estimate a bias-corrected dynamic fixed-effects model, verify inference with Driscoll–Kraay, cluster-robust, and wild cluster bootstrap procedures, run formal threshold tests, and conduct scenario simulations. Credit risk is highly persistent. The bias-corrected autoregressive coefficient of 0.944 implies a half-life of 12.0 years, although the bootstrap confidence interval of 0.601 to 1.048 does not rule out near-unit-root behavior. Exchange-rate depreciation predicts higher non-performing loan (NPL) ratios and survives both the strictest few-cluster test (p = 0.028) and a correction for euro-adoption breaks, while lower real GDP per capita growth is marginal under the same test (p = 0.060). Threshold tests that re-estimate the threshold in every bootstrap replication do not reject linearity in any of 14 configurations (minimum p-value of 0.071). Institutional quality does not measurably moderate the exchange-rate channel. A severe combined adverse scenario raises the projected NPL ratio from 6.54 to 12.32 percent over five years (90 percent interval: 8.2 to 24.6 percent). Together, the surviving channels and the disciplined null results delimit nonlinear transmission in emerging Europe.