Phillips curve with spatial effects based on Russian regional data

The paper tests the hypothesis of the presence of spatial effects for quarterly CPI in the Russian regions over the period 2015-2021. Contiguity, distance and migration matrices were used for spatial Phillips curve modelling. Due to spatial non-stationarity of the model for the whole Russia, the model was used for estimations separately for western and eastern regions. Panel data testing showed insignificance of the spatial lag of the dependent variable, which casts doubt on the hypothesis of "instant" (within the same period) inflation spillover. Perhaps the key factor here is the frequency of time series data: quarterly or monthly CPI better suit for spatial analysis than annual ones (for which the spatial lag will be significant). Spatial Durbin error model (SDEM) estimation showed that the inflation expectations in neighboring regions negatively impact on inflation in the region in this period. The estimations of the direct effects contribution for pi( t - 1), pi( t + 1) and indirect effect contribution for pi ( t - 1) expectedly have positive signs. The sum of estimated coefficients for inflation lags in spatial hybrid Phillips curve is close to 1. The use of a migration matrix for the western regions was unsuccessful, perhaps due to strong distortions introduced by Moscow and the Moscow region into interregional interactions.