Optimizing to the last digit: How taxes influence vehicle CO2 emission levels
What EU cars’ CO2 emissions have in common with supermarket prices
When you think back to your most recent stop at the supermarket, you will probably recall that most of the price tags you saw ended with the digit “9”. Maybe 99 cents for a chocolate bar and 49 cents for a banana. Numerous consumer research studies have demonstrated the psychological effect of two-figure vs. three-figure numbers, and price tags ending with a “9” in general. In short, they make it more tempting for us to buy.
Curiously, we observe a similar pattern for new cars in the EU—not in their price tags (though the familiar pattern repeats there too), but in their CO2 emission levels. You would expect vehicle CO2 emission levels to be evenly distributed, so that a vehicle would be no more likely to emit 88 g CO2/km than 89 g/km. And, in fact, this is exactly what the new car market in the EU looked like a few years ago. But that has changed.
In a just-released study (see here) we aggregated new car CO2 emission figures into bins according to the last digit of their official CO2 emissions (in g/km). For example, new cars with a CO2 emission level of 89, 99, 109, 119, 129, … g/km are all grouped into the “9-bin” and their total registration numbers for a particular year summed. The graph below summarizes the results for the years 2001 and 2014. When looking at the situation in 2001, it is obvious that the shares of each bin are evenly distributed; i.e., there were about the same number of vehicles with a CO2 emission level ending in “8” as in “9”.
New car registrations in the EU, aggregated into CO2 emission figure last-digit-bins, for 2001 and 2014.
The same analysis for 2014, however, shows that about one third of all new passenger cars fall into the 9-bin. In comparison, only about 2 percent of new cars had a CO2 emission figure ending with a “1”. That is, it is about 15 times more likely that the official CO2 emission level of your new car is something-“9” g/km than that it’s something-“1”.
That change in the market structure first became notable in 2006 and since then has been accelerating. Furthermore, there are significant differences among EU member states and vehicle manufacturers. The member states where clustering in the 9-bin is most pronounced are the UK, Spain, Ireland, Sweden and Portugal.
What accounts for this phenomenon of cars’ CO2 emissions clustering around certain CO2 bins? To better understand this, one needs to look more closely at national tax schemes. Many of the EU member states have changed their vehicle taxation scheme in recent years, and have based it on the CO2 emission levels of the new vehicles. In doing so, quite often they have defined tax thresholds above which the amount of tax suddenly rises to a higher level. What we observe in our data is how vehicle manufacturers and customers have reacted: by optimizing vehicles CO2 emission levels to the tax thresholds, and purchasing those tax-optimized cars.
Spain offers a good illustration of how this works in practice. In Spain, there is a vehicle registration tax threshold at 120 g/km of CO2. New cars with CO2 emissions up to 120 g/km are excluded from the one-time registration tax, while vehicles above 120 g/km have to pay a tax of at least 4.75 percent of the vehicle price (excluding VAT). Assuming a vehicle net price of €20,000, the tax benefit for a vehicle below 120 g/km therefore is €950. Unsurprisingly, more than 13 percent of all new cars registered in Spain have a CO2 emission level of 119 or 120 g/km.
At a first glance, you could say that the findings demonstrate that CO2-based vehicle taxation is an effective way of influencing vehicle characteristics, and hence a great success in terms of fiscal and environmental policy. However, thinking more carefully about it, these developments are problematic in two respects: First, if the tax level of a vehicle is significantly reduced because of a small change in CO2 (making it into the next lower tax category), that CO2 reduction is relatively expensive and not optimal from the government’s perspective (since it forgoes significant tax revenue in exchange for a small reduction in CO2). Second, we know from previous studies that real-world CO2 emission levels in the EU are much higher than claimed by vehicle manufacturers. Tax thresholds provide another incentive to optimize the laboratory test results to artificially low CO2 emission levels, thereby accelerating the growing gap between official and real-world emission figures.
What is the solution then? Our best-practice recommendation – and as a matter of fact also the recommendation of the European automakers – for designing vehicle taxation schemes is to make use of a continuous linear tax rate function, i.e. not building in any tax threshold steps. That way, the incentive to cluster around certain CO2 emission levels is avoided and the temptation to exploit loopholes in the vehicle testing procedure is reduced.