AbstractThe purpose of the current study is to investigate the degree of inflation persistence, its geographical variation, sources of cross‐regional variation, and presence of geographical/sectoral aggregation bias in national monetary policy. Our data set covers 26 NUTS‐2 level Turkish regions and monthly CPI inflation over the period 2003–2019. We first estimate the degree of regional inflation persistence by autoregressive regressions, check its robustness against the presence of structural breaks (by Bai–Perron's algorithm) and nonlinearities (by Markovian Regime Switching regressions). Second, we examine the possibility of geographical and sectoral aggregation bias. Third, we investigate the cross‐regional determinants of inflation persistence by panel data analysis, employing hybrid‐effects spatial panel regressions. We analyze the direct and indirect effects of the determinants and test for regional spillover effects. Three main results are obtained. First, estimated persistence degrees are heterogeneous across regions. The geographical pattern is empirically robust against structural breaks and nonlinearities. We find that inflation persistence is distributed in a spatially correlated manner. Second, when sectoral and regional aggregation bias is tested, only sectoral aggregation indicates a considerable level of bias. Third, we find that the presence of large firms in the region and a higher share of agricultural output in GDP leads to lower persistence, while an increased share of industrial output, and increased trade volume leads to higher inflation persistence. Moreover, we find spatial spillovers of price variability evident in regression analysis. From a policy standpoint, it is required that structural policy programs are targeted to maintain flexibility in the regions where persistence is high (i.e., providing market entry/exit, institutional quality, policy credibility, stimulation of SMEs). Moreover, sectors that have high persistence, such as Hotels and Restaurants (persistence degree 0.55) and Health Services (0.39) should be weighted more in CPI calculations.
Read full abstract