Dieselgate’s Big Data Trail: What Analyzing Innovation Data Says About Diesel Engines’ Spectacular Fall

In late 2015, intrepid emissions testing researchers sent shockwaves through the automotive world by catching Volkswagen in one of the biggest corporate scandals ever: Hidden software allowed its diesel engines to cheat on emissions testing. The repercussions are still being felt today, as more and more countries are turning against the technology. The common narrative is that this was a surprising, unforeseen event, and in many ways, it was. However, newly developed, specialized, big data analysis allows us to investigate diesel innovation – or rather, the lack of diesel innovation – and uncover some interesting trends. By applying our Lux Tech Signal software tool to the topic of diesel engines, we see a suspicious decline in interest in diesel innovation since 2010, as shown in the figure below:

For context, the Lux Tech Signal works by algorithmically doing high-throughput parallel analysis of various early-stage datasets about innovation, resulting in a revealing picture about any technology’s progress – or in this case, lack thereof. Intriguingly, this decline in diesel engine innovation interest matches up very well with the timing of when diesel-powered cars incorporating the cheating software hit the market. Our analysis shows the decline happening from 2010-2014; meanwhile, the cheating vehicles hit the market around the same timeframe, as highlighted in the figure below. This presents an interesting contrast: During a time of ever-tightening emissions regulations, and of more and more automakers rolling out supposedly advanced diesel engines, developers appeared to have been pulling back on their interests in further innovation. While it would be over-reaching to state that this is a causal and direct relationship, it does hint at the importance of tracking whether innovation is or is not proceeding apace.

Could this trend be simply due to slowing innovation in conventional engines as a whole? To account for this possibility, we also looked at whether innovation within the internal combustion engine (ICE) as a whole is fading: It is not. As the figure below shows, combustion engines in general continue to receive steady interest in innovation. Diesel engines, however, have not. (Since diesel engines are also ICEs, the disparity between the two curves would be even higher if we were to segment out non-diesel ICEs versus diesel ICEs.)

We also used our Lux Tech Signal to look at how other areas of automotive drivetrain innovation performed: We find that while diesel engines’ innovation interest was spiraling downwards, electric vehicles were seeing a genuine uptick. Ironically, in its post-Dieselgate scramble to pivot, VW has jumped between the two curves, shifting priorities from diesel engines towards a mad dash to catch up in electric vehicles instead. (Fuel cell vehicles, on the other hand, performed very poorly in our data-driven analysis, foreshadowing their abysmally low sales on the market today.)

Digging even deeper, we can compare and contrast different corporations’ innovation activity within diesel: The key parties implicated in the Dieselgate scandal were Volkswagen and automotive supplier Bosch. Analyzing the innovation activity of these companies within diesel – by using one of the key datasets behind the Lux Tech Signal – we see a telling trend: Diesel innovation activity at VW and Bosch appears to have declined by more than 50% between 2001 and 2015 (specifically, their averaged diesel innovation fell from 2.0% to 0.8%, as a percent of worldwide diesel innovation). Meanwhile, VW’s mainstream global automotive competitors – Toyota, GM, Ford – doubled their averaged diesel innovation activity during that same timeframe, as shown in the figure below: Each rises from about 1.2% to 2.4% on average, again as percent of overall global diesel innovation. (A note on methodology: The reason we choose to combine VW and its supplier Bosch here is because the Dieselgate scandal obscured each party’s role; it appears both were involved, but the exact nature of their collaboration is opaque, so we lump them together. This should have the affect of boosting the resulting combined innovation activity, so the fact that it declines even when lumped together is telling.)

It is striking to see in the above figure that VW’s and Bosch’s diesel innovation decline – according to this data set – began about eight years before the automaker sold its first cheating diesel vehicles. The decline then continued for another six years before they were discovered. Although only the hands-on investigative work from vehicle emissions researchers could have unearthed Dieselgate, the data around the industry’s lack of diesel engine innovation interest – and specifically, VW’s and Bosch’s – was there all along, buried in early-stage data sets about technology. With advances in data science methodologies, such innovation trends can now be analyzed rapidly and consistently, for any technology-related topics. As this case study shows, there is no escaping the need for corporations to innovate: Cutting corners shows up in the data, and then shows in the marketplace.

By: Cosmin Laslau