
This study examines the relationship between regional financial structure and provincial nonperforming loans (NPLs) in Türkiye using a balanced panel of 81 provinces spanning 2008–2025. Regional financial structure is captured by the loan-to-deposit ratio (LDR), loan-to-GDP ratio (LGDP), and financial sector share (FSS), and total, public bank, and private bank NPLs are examined separately. Dynamic fixed effects estimates provide the nonspatial benchmark, followed by a dynamic spatial Durbin model incorporating geographical contiguity and hybrid geographic–trade matrices. The results reveal substantial persistence and significant spatial dependence in provincial NPLs. LDR and LGDP exhibit the most robust positive direct and spatial effects, whereas FSS has greater heterogeneity across bank ownership and spatial weighting structures. One-standard deviation changes in the financial indicators also reveal economically sizable short-run NPL responses. Although increasing the influence of bilateral trade intensity among geographically contiguous provinces changes the magnitude of some spatial effects, it does not alter the benchmark results. Model-fit comparisons show that geographically oriented specifications perform better for public bank NPLs, whereas stronger trade weighting tends to better fit private bank NPLs. The findings remain broadly robust to alternative spatial specifications and macroeconomic controls, emphasizing the significance of regional financial structure, bank ownership, and economic connectivity for addressing provincial credit risk dynamics in Türkiye.
This study examines whether an economic pillar of the financial system bears the cost of chronic inflation. Using the International Monetary Fund's Financial Development Indexes for Türkiye over the period 1980–2020, we estimate autoregressive distributed lag (ARDL) and nonlinear ARDL (NARDL) models for the financial institution (FI) and financial market (FM) indexes and their components. The results are the opposite of the common expectations. Chronic inflation has a large, detrimental, and long-run impact on the institutional pillar, in particular on efficiency, whereas the market index shows no negative response. Toda-Yamamoto, asymmetric, and frequency-domain causality tests support this finding, in which predictive causality flows only from inflation to institutions. The apparent immunity of the market index reflects a measurement issue, as the turnover-based efficiency sub-index reached its normalization ceiling during the 1994 crisis. These findings should serve as a warning to users of these indexes and suggest that disinflation can act as an institutional development policy.
This study examines the Borsa Istanbul stock market (BIST100) response to monthly consumer price index (CPI) announcements over a 95-month sample (January 2018–December 2025), using ordinary least squares (OLS) regression, event-study methodology, EGARCH(1,1) volatility modeling, and Markov regime-switching. We obtain four findings. First, the one-month real-sector surprise does not significantly explain same-month returns; given limited test power, this null should not be read as evidence of market assimilation. Second, the 12-month market-consensus surprise yields a positive return effect, significant under most corrections but marginal under block bootstrap. Third, in the high-inflation regime (September 2021–December 2025), a positive abnormal return three days post-announcement survives a narrow Bonferroni but not a multiplicity correction, and should be read as suggestive. Fourth, EGARCH(1,1) confirms volatility clustering and a leverage effect but no significant surprise–volatility link or sectoral heterogeneity. These results position Türkiye high-inflation episode as a setting for studying market efficiency and news transmission.