In an analysis, if cross-sections consist of geographical regions, the existence of spatial interactions of contiguous units should be investigated. Because it is thought that close areas affect each other. In this study, it is investigated that how government expenditures affect the GDP growth per capita in Europe for the period 1997-2016, via models considering spatial dependency. Eventually, increases in government expenditures increase economic growth in all models. Fixed effects panel data, random effects panel data, fixed effects spatial lag, random effects spatial lag, fixed effects spatial Durbin, random effects spatial Durbin, fixed effects spatial error, random effects spatial error, generalized random effects spatial error and general spatial models are examined. As results of Hausman tests, it was seen that random effects model is efficient in all models. Lagrange Multiplayer (LM) test showed that not only spatial lag and spatial error models are significant, but also the general model is significant. Between these models, the general model which contains also spatial lag and spatial error model either, has the maximum Lagrange Multiplayer (LM) test statistic. As for spatial Durbin model, the coefficient estimates of explanatory variables multiplied by spatial weight matrix are found statistically insignificant. Consequently, the politicians who want to achieve full employment in Europe, can increase the government current expenditures. Besides, they should take into account the situation in neighbor countries. Because, the coefficients indicating spatial autocorrelations are positive. As areas approach to each other, they more affect each other.