Effects of differentiation in car purchase tax based on carbon-dioxide emissions in Finland Andrey Zhukov University of Helsinki November 14, 2013
Background As of January 2008 new approach to car purchase taxation was introduced in Finland. Pursuing environmental goals, it was decided that from that time on car purchase tax should be based on declared carbon-dioxide emissions, thus creating enormous variation in car prices and making place for interesting research projects. Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 1 / 8
Research goals Investigate an impact of tax reform on consumer choice, i.e. changes in market shares of different brands Car price elasticities Effects on changes in sales of diesel and gasoline cars, engine sizes, carbon-dioxide emissions Long-term vs. short-term effects Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 2 / 8
Immediate response to reform In automobile production it is common that the model year begins approximately in the third quarter of the year. It means that there is no unobserved changes in the model characteristics for a car purchased in the last quarter of 2007 and a car purchased in the first quarter of 2008. It is a plausible assumption that consumer tastes are slow evolving and thus consumer tastes in adjacent quarters can be regarded as equivalent. Therefore a regression of a type: y = α + βx + ε t, (1) where y is a change in the number of the cars (in logarithms) of the same line purchased in the fourth quarter of 2007 and the first quarter of 2008 x is a tax-induced log-price difference is likely to account for potential correlation between sales and unobserved car and consumer characteristics and an estimate of β could be interpreted as a causal effect of a one percent tax-induced change in price on the percentage change in quantity demanded (a.k.a. price elasticity of demand ). Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 3 / 8
Expanding a window of analysis The natural next step would be to compare longer-term results with an immediate short-term impact of the tax reform. A way to perform that is to estimate an equation similar to (1): y = α + βx + ε t, (2) where y is now a change in the number of the cars (in logarithms) of the same line purchased in the third quarter of 2007 and the second quarter of 2008. x is a tax-induced log-price difference which is equivalent to the one in (1). Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 4 / 8
Expanding a window of analysis An interest of comparison of estimates from (1) and (2) is underlined by the fact that news about upcoming reform strongly influenced the markets already starting from October 2007 1. Thus difference in obtained calculations of an impact of the car purchase change on the carbon dioxide emissions performance can be interpreted as an offsetting effect of information about the reform getting in the media three months before the reform actually occurred. 1 Adriaan Perrels Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 5 / 8
Annual sales There is still a chance that people who planned car purchases for, say, the third quarter of 2008 resolved to switch a purchase to 2007 after announcement of a tax reform. Thus my estimates in the previous section would be somewhat misleading about the real effect of the reform. Therefore the next step would be to perform an analysis similar to that above but using annual data instead of quarterly data as follows: y = α + β 1 x 1 + β 2 x 2 + ε t, (3) where y is a change in the number of the cars (in logarithms) of the same line purchased in 2007 and in 2008 x 1 is a tax-induced log-price difference x 2 is a change in price of gasoline per 1000km (based on fuel economy of a car line and gasoline price) Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 6 / 8
Annual sales Then we expand a window of analysis as follows: y = α + β 1 x 1 + β 2 x 2 + ε t, (4) where y is a change in the number of the cars (in logarithms) of the same line purchased in 2006 and in 2008 x 1 is a tax-induced log-price difference x 2 is a change in price of gasoline per 1000km (based on fuel economy of a car line and gasoline price) Differences in β estimates in regressions (3) and (4) would demonstrate an effect of announcement of an upcoming tax-reform in the end of 2007 and, arguably some psychological effects. Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 7 / 8
Impact on gasoline vs. diesel, engine size, etc. Then I perform the same analysis as specified previously defining dependent variable differently. For instance, I would investigate an impact of tax reform on the sales of gasoline cars, diesel cars, different engine sizes. Effects of differentiation in car purchase tax based on carbon-dioxide November emissions 14, 2013 in Finland 8 / 8