Recent disclosures of Volkswagen’s deliberate frauds to foil U.S. emissions standards on its diesel cars may be mind-boggling in their audacity, but they are a gift to econ teachers like me. What better example could one find of the particular set of market failures known as “principal-agent problems”?
In this case, the “principals,” those with fundamental interests at stake, are those who own stock in Volkswagen. The “agents,” those who are supposed to work to further the best interests of the principals. This would be the engineers, managers and executives who came up with, or signed off on, the idea of using software in the cars’ electronics to only lower emissions when it sensed that the car was being tested for such emissions — the apparent benefit being that the cars’ performance would suffer on the road if the emissions were kept low.
VW has excellent engineers, and German culture has a strong element of moral rectitude in business affairs. So this enormous deception has gotten wide attention. The affair is likely to drag on for some time and may even pose an existential threat to the company itself.
The agents in this case did not set out to harm the principals. Nor did they necessarily want to harm the environment or the purchasers of VWs and Audis with diesel engines. They merely responded to the incentives they faced, some created by VW management and some by outside circumstances.
That is both why principal-agent problems can seem intractable and why they often differ from simple criminal acts.
When it comes to corporate crime, let’s consider the difference.
Recently, a former executive at medical device maker St. Jude Medical was charged criminally for misdeeds dating back to last year at the Twin Cities-based company. These included charging personal purchases and otherwise falsifying company expense accounts, and stealing company secrets. Yes, an employee stealing from his employer is a case of the objectives of the individual — personal enrichment — differing from those of the company’s — maximizing earnings for shareholders over the long run.
But the core of this case is simple theft. It is not greatly different from someone breaking into a company facility and stealing a computer. Only in this case, the perpetrator was an employee. And theft is a crime, which is not true in all principal-agent cases.
Knowingly selling dangerously tainted food also is a crime. That got three executives of the defunct Peanut Corporation of America jail sentences of five to 28 years on 72 counts of fraud and selling toxic food in a 2008 case that killed nine people and sickened at least 714 others. They probably did not intend to harm anyone, but they did knowingly ship contaminated product and falsified sanitation and quality-control records.
However, because two of the execs sentenced are members of the family holding majority ownership of the company, this crime was largely not an example of a principal-agent problem. The customers they harmed in their own perceived interest are not principals, and the executives were not agents of anyone else.
Many observers commented on the VW scandal emerging only days after the settlement of a case against General Motors for selling cars with defective ignition switches. These resulted in at least 120 deaths, many more injuries and millions in property damage. Unlike the VW case, no one at GM started out with a clear objective of deceiving anyone. The defective switches simply were badly engineered.
If GM had recognized the problem and acted promptly to fix it, there might have been civil suits for damages but no basis for a federal fine. GM did take a series of actions to deny the problem, delay its solution and eventually compensate those harmed. Like the initial faulty design, these were actions by agents, engineers and managers, who, in the long run, harmed GM stockholders. However, those making the initial decisions were not out to feather their own nests at the expense of the company’s shareholders. And they were not out to protect shareholders: The cost of fixing the defect probably would not have made a difference in GM’s share price at the time. They largely were just were bad decisions.
I include the qualifier “largely” because at some levels or junctures, individual managers or engineers took actions to please bosses, win bonuses, meet budgets or otherwise further their own careers. They took these actions without weighing how the results, combined with the actions of colleagues, might harm the overall long-term profitability of their employer and its owners.
And that does get to the heart of principal-agent problems. How do top managers, who, in theory at least, report to stockholders through the corporation’s board of directors and executives, structure incentives for their employees in such a way that most benefits owners?
The profit motive is key in business; indeed, politically, this is often cited as a key advantage of the private sector over government to get things done. Controlling costs is crucial. Motivating engineers to design components that do some function at a lower cost is important, and designing such incentives is not hard. But someone has to consider potential harms.
It is much harder to incorporate incentives so that someone weighs possible greater damage — death or corporate harm — from a design flaw that is a nickel or a dollar cheaper than the alternative. If a new design saves a dollar in manufacturing costs but has an additional one-in-a-thousand chance of starting a fire, should one go ahead? What if one in a hundred of such fires will hurt a person? What if one in a hundred of such injured people dies?
The problem is that in the real world it is difficult even for a skilled engineer of great integrity to estimate such risks ahead of time. It is little different than a sober, careful driver estimating the risk of taking her eyes off of rush hour traffic for the split second necessary to hit a radio button. We all do it but in some minute fraction of events, it can cause tragedy.
The cumulative results of human behaviors in large organizations is a large part of principal-agent losses. No one in GM set out to harm buyers or shareholders. But their aggregated decisions resulted in just that.
The mind-boggling aspect of the VW scandal is that a substantial group of people set out to criminally defy the law. It was not just one rogue. There obviously had to have been strong incentives for these engineers and program managers to design a diesel engine that had high performance and measurably low emissions — but not really low emissions. No one involved would have participated in such a brazen endeavor casually. So the carrots dangled by senior managers had to have been enormous.
That exposes the fundamental dishonesty of statements by departed VW CEO Martin Winterkorn of his company’s value of integrity. You can fill a mission statement with statements of honesty as a core value, but if you then structure incentives to key employees that motivate them to commit fraud in this way, your pious declarations are hollow. A fish rots from its head.
Clearly, some engineers, code writers and program managers were under enormous pressure to deliver. And clearly, they had little incentive to fully consider the effects on company finances and reputation if caught in their deception. Now, shareholder equity is being devastated.
And many others — dealers and owners especially — are suffering serious losses. But more broadly, so is society. The core of principal-agent problems is that they undermine the efficient use of resources in an economy as a whole. This waste of resources is not readily visible, but it is real, nevertheless, and it means households get fewer of their needs and wants satisfied that might be possible. The cumulative magnitude of this often outweighs the direct losses to principals themselves.