Houshold consumption expenditures and demand estimation in Turkey

In this study, theory of consumer demand and the evolution of complete demand models (systems) consistent with the theory and its use in empirical studies are examined. Complete demand models can not applied in this study since time series data on consumption expenditure and price are not available in Turkey. In this study, the household consumption expenditure and socio-demographic data obtained from the TUIK-?2003 budget survey? is used for empirical demand analysis. The demand studies with cross-section budget survey data generally uses household expenditure and sociodemographic data since prices for commodities and services are not usually obtained with this survey. In the empirical model, household demand is modeled using Engel curve called as Workin-Leser since this model yields better results than the other functional form and at least satisfies Engel aggregation condition dictated by the theory. The dependent variable of the model is the budget share of goods and services within total household expenditure and independent variables are total household expenditure and household socio-economic variables. The model is applied to broad based commodities and services group and sub-food groups in Turkey. Socio-demographic variables include education level of household head (9), age composition of household (2 or 3), location (urban and rural) and gender which are different from the socio-demographic variables (location and household size) used in previous similar demand studies in Turkey. The results of the model estimation indicate that budget shares are very responsive to total expenditure (income proxy) and socio-demographic variables. The estimated expenditure coefficient and most of the estimates coefficients for socio-demographic variable are statistically significant at 1 or 5 percent confidence level. It is proved that expenditure elasticity of commodities, particularly food-sub-groups, exhibit significant differences between urban and rural in Turkey. It is also proved that the level of education has different effects on different product groups, particularly food subgroups. It can be considered that the important findings of the study are the results obtained with aggregate and disaggregate data yields parallel but significantly different results. It is found that the demand estimates with aggregate data yield consistently higher expenditure elasticity (overestimates) although expenditure elasticity calculated from the house level data gives systematically similar results.