The Classification of Spatial Quantile Regression Models for Healthy Life Years in European Countries

Abstract

EU countries are heterogeneous in terms of healthy life years (HLY). A quantile spatial autoregression model is employed to identify the factors affecting HLY. Quantile regression allows to measure the impact of covariates on different parts of the conditional distribution of a dependent variable. This is especially useful if the distribution is asymmetric or trimmed. Quantile spatial autoregression models add the spatial dimension to the standard quantile regressions. The aims of the study are as follows: identification of the factors affecting healthy life years of men and women in the EU countries, and clustering of the estimates obtained for the different quantile orders. We analyze to what extent spatial quantile regression can be useful in studying the behavior of HLY. We use exogenous factors belonging to three groups: socio-economic, healthcare and lifestyle. After initial estimation of the models for 19 quantile orders, the insignificant factors are removed and the remaining parameters are re-estimated. The final estimates are then clustered to facilitate the interpretation of the results. Instrumental variables method coupled with bootstrap techniques are employed for estimation and inference while the k-means algorithm is used for clustering. We find that the impact of the factors on HLY varies for different quantile orders. In general, the quantiles of low and medium orders are affected by the factors from all the three groups (socio-economic, healthcare and lifestyle). For the medium and high orders, the socio-economic factors are no longer important. We argue that the spatial quantile autoregression models offer a new perspective for studying the factors affecting HLY in the EU countries.(original abstract)