A comparative assessment of aggregate car ownership model estimation methodologies

This work examines the implications of advances in time series analysis on car ownership modeling in Greece. Variables include adults population ratio, GDP per capita, car occupancy, bus kilometers, inflation and unemployment. We developed and compared (a) a classical regression model estimated on raw levels, (b) an econometric model estimated on data stationarized using graphical and unit root tests and (c) an "atheoretical" ARIMA model. Although significant methodological implications were noted, all models forecast 48 to 49 private cars per 100 inhabitants by the year 2010, a development of momentous energy and environmental implications.