Esti̇mati̇on of household consumpti̇on expendi̇ture by nonparametri̇c regressi̇on method: The case of Turkey

In traditional parametric regression methods which are often used to investigate relationship between variables, the functional form of the relationship is specified in advance and the aim is to estimate the parameters of this functional form. In nonparametric regression methods, in contrast, the aim is to estimate regression function directly without specified its functional form clearly. For this reason nonparametric regression methods provide to more flexibility for researchers in revealing relationship between the variables. In this study, for purpose to reveal the relationship between household expenditure, households' income and size of household defined by OECD is used LOESS regression method which is a nonparametric regression method and data from TUIK Household Budget Survey of 2012. The results are obtained show that in low income group of household, initially household expenditure rapidly increase and then this increase decreases while increase size of household. İn high income group of hosehold, in conrast, household expenditure decreases a little bit while increase size of household.