Francisco Javier Parra Rodríguez
Universidad de Cantabria (UNICAN), España
The objective of this article is to estimate a marginal propensity to consume ( MPC) for the American economy through a classic Keynesian consumption function with microdata from household surveys . The MPC value are influenced by alternative definitions of spending units or consumption expenditure and income, at real or a per capita level, usually the data are grouped by arbitrary but seemingly reasonable income intervals. The form to data are collected to describe expenditure patterns, organized by income interval, and highly aggregated, determine the different MPC in consumption function. We propose an estimation method that uses income percentiles to grouping survey sample. These estimates usually have normal problems in error term due to outlier, dependence, heteroscedasticity or the existence of nonlinear relationships in the variables. By Regression Band Spectrum (RBS) , and Generalized Linear Models (GLM), we can solve this problems.
Classification JEL (Journal of Economic Literature): E12, E21,R21