The consumption function revisited : an error-correction model for Finnish consumption

Abstract

The permanent income hypothesis asserts that consumption depends on current labour income, expected present value of earnings and cumulated net wealth.Based on this idea, a three variable cointegrationsystem, including consumption, income and net wealth, is tested and approved.Net wealth is included into the system in market values to allow capital gains in real estate wealth, which means that weal this not simply accumulated savings.Wealth is also disaggregated into financial assets, real estate wealth and debt, in order to confirm the existence of one cointegration relationship and to test the proper wealth concept for the consumption function.The tests suggest that a broad concept of net wealth is preferred for the consumption function.It is also found that the average propensities to consume are different for financial wealth and real estate wealth. The results from the cointegration system show that neither disposable income nor net wealth can be regarded as weakly exogenous in the system nor can either one be excluded from the system of endogenous variables.So, in fact, there does not exist a separate statistically meaningful consumption function.Despite this, effort was made to explain which variables in the past had the greatest effect on consumption, and a so-called error-correction consumption function was estimated.In addition to the I(1) core of the consumption model including income, net wealth, few weakly exogenous stationary variables; the real interest rate, inflation, unemployment rate and relative prices of consumption subgroups were found to be significant.