Was any other car company gaming emission standards?

As most of you already know, Volkswagen was recently caught cheating on emission tests. But were they really alone in doing so? The EPA data shows some suspicious patterns.

Without getting too technical (in any case, the specifics are still unclear), there's a tradeoff between fuel efficiency, and NOx emissions. So for a given level of performance, the software that controls a modern diesel engine can be tuned so that it is more fuel efficient, but produces more NOx, or it can be tuned so that it is less fuel efficient, but produces less NOx.

Since Volkswagen had problems meeting the NOx standards in the US, they decided to introduce software that detected when an emission test was being performed, and tuned the engine so as to decrease NOx production, in the process potentially worsening fuel economy relative to regular driving conditions.

One possible outcome of this procedure is that the cars involved might actually have better fuel economy in real life conditions, than during testing. For example the Diesel version of the Jetta was famous for this.

Since I happened to have the EPA mileage data lying around my hard-drive from a different project (you can get your copy here), and since I knew that fuelly.com  logged real world mileage performance, I thought it would be interesting to see if this pattern was systematic. Indeed, all of the affected VW models seemed to get between 10% and 20% more miles to the gallon than what their EPA Combined mileage rating would suggest.

Do any other models show a similar pattern? The Disesel version of the Chevrolet Cruze stands out. In fact it shows 20% better mileage on the road than in the tests, more than any other model. Several Mercedes-Benz models stand out as well, but there a very few observations in the fuelly.com dataset for these, so it is hard to draw inference from these.

Another possible explanation for this pattern is sample selection. Since the fuelly.com is not a random sample (users have to download an app), it is likely that their user base is more fuel conscious than the average driver. Also, the combined rating is calculated by the EPA by taking a 55/45 weighted average of the City and Highway tests. If owners of diesel cars spend much more time on highways than regular drivers, it is entirely possible that the observed mileage is simply a product of their driving habits. Nonetheless, owners of other Diesel models don't seem to show this pattern.

To address this concern I calculated what fraction of their driving the users of each model would have had to do on highways, to obtain the observed mileage. In the case of the diesel Cruze, it is 68%.

I briefly thought of buying some out of the money puts on GM with a friend, but the fees were too expensive relative to our limited investment funds. As a result, I can confirm that I have no financial conflict of interest in the matter.


Popular posts from this blog

Why do bands allow scalpers? (Hint: Dynamic Efficiency)

Nikon AF-S NIKKOR 28-300mm f/3.5-5.6G ED VR Lens Review