Exchange Rate Volatility and Banking Sector Stability in Türkiye: Quantile Asymmetries, Time-Frequency Co-Movement and the Regulatory Ratio Illusion

Abstract

Banking sectors in emerging economies remain exposed to currency shocks because foreign exchange liabilities sit on the balance sheets of both banks and their borrowers. The applied literature has largely treated this exposure as a single average effect, so an increase in volatility is assumed to act with the same sign and magnitude in tranquil and crisis months alike. That assumption merges the risk accumulating while the currency is calm with the mechanical distortion of reported ratios when it collapses, and it leaves open when, and at which cycle length, the relationship emerges. This study identifies the state-dependent structure of the link between exchange rate volatility and the stability of the Turkish banking sector and asks whether the observed patterns reflect underlying credit risk or the arithmetic of the ratios and the regulatory calendar. Monthly sector aggregates for non-performing loans, the capital adequacy ratio, and return on assets are examined against conditional volatility obtained from a generalised autoregressive conditional heteroskedasticity model with Student-t innovations. Wavelet coherence decomposes the relationship by calendar time and cycle length, quantile-on-quantile regression estimates the effect across pairs of quantiles in the two distributions, and quantile Granger tests establish the dominant direction. The estimates reveal a sign reversal. When the currency is calm, the effect on non-performing loans is positive and strengthens in the upper region of the credit risk distribution, whereas at the highest volatility states it turns negative. The capital ratio responds in opposite directions depending on where the sector stands, since severe shocks lift the reported ratio when capital is weakest and depress it when capital is strongest. Profitability deteriorates even in tranquil states, and its apparent predictive power for volatility can be traced to valuation gains and losses embedded in monthly earnings. A falling non-performing loan ratio and an improving capital ratio during currency stress therefore cannot be interpreted as evidence of resilience, since loan revaluation inflates both denominators while classification windows and valuation permissions defer loss recognition. Supervisory assessment requires organic measures that strip out revaluation and forbearance, together with preventive tools that bind in tranquil periods.