Estimating light-vehicle sales in Turkey

This paper is motivated by the surprising rapid growth of new light-vehicle sales in Turkey in 2015. Domestic sales grew 25%, dramatically surpassing the industry estimates of around 8%. Our approach is to inform the sales trend estimate with the information obtained from the light-vehicle stock (the number of cars and light trucks officially registered in the country), and the scrappage data. More specifically, we improve the sales trend estimate by estimating the trend of its stock. Using household data, we show that an important reason for the rapid sales growth is that an increasing share of household budgets is spent on automobile purchases. The elasticity of light-vehicle sales to cyclical changes in aggregate demand is high and robust; its estimates are around 6 with a standard deviation of about 0.5. The price elasticity of light-vehicle sales is estimated to be about 0.8, but the estimates are imprecise and not robust. We estimate the trend level of light-vehicle sales to be roughly 7 percent of the existing stock. A remarkable out-of-sample forecast performance is obtained for horizons up to nearly a decade by a regression equation using only a cyclical gap measure, the time trend and obvious policy dummies. Various specifications suggest that the strong 2015 growth of light-vehicle sales was predictable in late 2014