Even hardened cynics were shocked by the revelations in September that Volkswagen had programmed its diesel cars to defeat emissions tests. Whilst it was no secret that manufacturers attempted to ‘game’ these tests, what was astonishing about VW’s behaviour was the scale and method of their deceit. The technology VW used for cleansing Nitrous Oxide emissions entailed a trade-off against fuel economy and performance, which VW’s engineers and programmers had decided to manage by introducing two distinct driving ‘modes’ into the car’s software: a ‘best behaviour’ mode that complied with US Nitrous Oxide emissions requirements, and another for all other circumstances. The cars’ systems were smart enough to detect whether they were undergoing the highly predictable and stylised requirements of the US Environmental Protection Agency (EPA)’s tests, and to put themselves on best behaviour accordingly. The trade-offs were significant, however: the cars emitted up to 40 times more Nitrous Oxide during normal circumstances than under test conditions. About 500,000 vehicles were affected in the US, spanning the model years 2009-16. The tactic was used in other markets as well, raising the total number of affected vehicles to approximately 11 million worldwide.

Nitrous Oxides and particulates, a by-product of diesel combustion, are harmful for human health. Some analysts suggest that the additional pollution from VW’s cars would have led to up to 100 additional deaths in the US alone from respiratory-related illness. To bring this problem home, note that 1.2 million affected cars were sold in the UK—more than twice as many as across the entire US. It’s worth noting that annual average levels of Nitrogen Dioxide in Oxford High Street are 25% above the maximum levels permitted by EU law and considered ‘safe’ according to WHO guidelines.

At this point, it remains unclear precisely how and why VW came to do this. US law firm Jones Day have been retained by VW to conduct an internal investigation, the results of which are due in the fourth quarter of 2016. VW’s official line is that this was the result of the actions of a few engineers and programmers. Let’s call this the ‘rogue technicians’ theory. However, with such a large number of vehicles, over a period of about seven years, it seems—as was argued in a recent post by J.S. Nelson—more plausible that senior management might have been aware of the scheme (or warning signs of it) at some point prior to its revelation. Let’s call this the ‘management inaction’ theory.

Whether it was rogue technicians or management inaction, it appears likely that cost savings were a key motivation. VW’s competitors in selling diesel cars into the US market, BMW and Mercedes, developed a more expensive technology for cleansing diesel fumes of particulates and Nitrous Oxide, known as urea filtering. By choosing not to license this technology from Mercedes, analysts estimate VW could have saved up to $4.8bn worldwide.

Although this may seem a large sum, it is far smaller than VW’s likely losses as a result of the scandal. VW faces a bevy of lawsuits from all over the world. Details of their expected quantum emerged in late April from VW’s 2015 Annual Report and the announcement of an agreement in principle regarding a settlement with US authorities and class-action lawyers, under which VW reportedly agrees to repair or buy back the cars sold in the US with the illegal software, and to pay additional compensation. There are still many uncertainties regarding liability, as the agreement apparently only covers civil litigation in the US. In addition, the US Department of Justice has opened a criminal investigation, and civil, criminal and/or administrative actions have been commenced in many other countries.

VW’s 2015 Annual Report announced provisions totalling €16.4 billion ($18.4 billion) for the clean-up and legal costs, including €7.0 billion ($8.0 billion) dollars for “legal risks” and a further €1 billion ($1.2 billion) for contingent liabilities. VW also began a product recall in the EU of cars sold with affected engines.

An interesting question is the extent to which, over and above the costs of liability, the recall and lost sales of non-compliant vehicles, VW has suffered further reputational losses. In the finance literature, ‘reputational losses’ are often taken to mean the cost to a company of people being less willing to trade with it—for example, having to lower its prices to shift product, or raise promised returns in order to raise capital. Such a shift can be triggered by the revelation of information that shows the company has a propensity not to keep its promises. But prior literature (Karpoff et al, 2005; Armour et al, 2010) finds that reputational effects of this sort are only triggered where the firm is revealed to have harmed parties who trade with it: where the harm imposes costs on third parties (e.g. the environment) then there is no stock price movement beyond the amount of the expected legal liability.

VW’s stock price fell rapidly by 40% from its pre-scandal position, although it subsequently regained ground, now wavering  at just under 20% down. This suggests the market’s estimate of VW’s likely losses is similar to the firm’s provisions, at around $18 billion.  As a result, it doesn’t seem that VW’s case bucks the trend of purely environmental harms (as opposed to uncompensated harms to customers) not triggering reputational losses.

Given these losses, cheating was clearly a very bad thing for VW to do from an ex post perspective. But what about ex ante? This really depends on expectations about the probability of getting caught, and attitudes to risk. Consider first the (highly implausible) idea that the decision to cheat was in fact deliberately taken by management, assuming that the benefit was $5bn and the costs were $18bn. This would have a positive expected return for VW provided that the probability of getting caught was less than approximately 25%. Now consider the (much more plausible) position of a CEO who has a whiff that something may be amiss amongst junior engineers. He can either pursue an internal investigation, which will alienate engineers and may reveal wrongdoing. Or he can do nothing and press on regardless. The problem is that an internal investigation will surely increase the likelihood that the firm will get caught for any prior misconduct. If the CEO judges the initial probability of getting caught to be low, it is easy to see that it may maximise expected profits to turn a blind eye, rather than draw attention to a potential problem.

Of course, executives are risk averse, and so the option of cheating will seem less attractive. Nobody wants to lose their job under a cloud, or what is worse, face potential personal liability for criminal or negligent misconduct. This provides a nice introduction to the challenging questions the VW case poses for corporate governance. Why was it that risk-averse individuals amongst the management might have failed to investigate potential smoking guns? My second post on VW will further comment on the corporate governance implications of the emissions scandal and provide my answer to this question. 


John Armour is the Hogan Lovells Professor of Law and Finance at the University of Oxford.