Opinion, Berkeley Blogs

VW's Deepwater Horizon?

By Maximilian Auffhammer

VW factory

Last week one of the biggest environmental scandals since the Deepwater Horizon disaster made its way to somewhere near the bottom of page 11 of most major newspapers. Volkswagen admitted to systematically cheating on emissions tests of its diesel vehicles. This might sound snoozy, until you read up on the details.

Vehicles across the U.S. must satisfy emissions standards for criteria air pollutants (e.g., NOx, SOx, CO2). California, of course, has the most stringent of these standards and enforces them for new and used vehicles. If you have an older car, you need to go get your car smog checked every few years to make sure your clunker is still clean enough to be allowed on California roads.

VW factory

It is for this reason that until recently the share of diesel cars in California was extremely low, since almost no vehicles satisfied these stringent standards. Enter the "clean diesels," pushed mainly by German manufacturers of normal-people (e.g. VW) and luxury (e.g. Mercedes and BMW) vehicles. Diesel was finally salon worthy! "Look! It's fuel efficient and clean!"

Many of my Birkenstock-wearing, dog-owning, El Capitan-summiting colleagues and graduate students ran out and traded in their Prii for the VW TDI wagon. "So much space! So much torque! So much fuel efficiency! So clean!" Well, it turns out what sounded too good to be true was.

In a Lance Armstrongian feat of deception, VW has now admitted to having installed a piece of software called a “defeat device” that turns on the full suite of pollution-control gadgets when cars are being smog tested. As soon as you leave the testing station and head out for your Yosemite adventure with Fluffy barking in the back, your car emits 10-40 times the amount of NOx you just reported on your smog-check card. Just to put this in perspective – this is like that 215 calorie Snickers bar instead having 2,150 to 8,600 calories.

Calculating penalties

The Environmental Protection Agency will almost certainly sue VW. The penalties involved here are significant. The EPA can ask for $37,500 per incident, which amounts to roughly $18 billion in fines. Plus there will likely be criminal charges filed against VW executives. Further, depending on whether these vehicles will continue to be sold in the U.S. after everything is said and done, this is a disaster for VW, as they rely heavily on the high fuel-efficiency ratings of diesels to satisfy CAFE (or Corporate Average Fuel Economy). Vehicles across the U.S. must satisfy emissions standards for criteria air pollutants (e.g., NOx, SOx, CO2).

In my eyes there are two interesting economic points to be made here. The first, maybe more headline worthy, is trying to determine the optimal fine in order to deter other manufacturers from engaging in such behavior. An economist would argue that what we have here is the classic case of an externality. By selling the dirtier vehicles, VW exposed kids, adults and dogs to massive quantities of local air pollutants. VW is responsible and should be liable for this.

Hence VW should correct this market failure by paying the full external costs it caused. This calculation would involve estimating the economic damages from this additional air pollution and passing the bill on to VW. My back-of-the-envelope calculation suggests that the NOx portion is about $232 per vehicle over three years (far from $37,500).

But, there is a large law and economics literature on determining the fines to achieve the optimal and efficient amount of deterrence. The problem with just passing on the external damages is that VW was not going to be caught with certainty. If the executives thought there was a 1 percent chance of getting caught, it might have been more worthwhile to cheat than if they thought that they were going to get caught with certainty.

In this case, the penalty should be approximated by the external costs divided by the probability of getting caught. This, of course, would be significantly more than the external costs alone. Getting the external costs right is hard to do (e.g., you need more pollutants and the damages vary across space), but can be done with standard tools in the talented economists' empirical toolkit.

A policy proposal

The broader question is how did this happen? This is not one student cheating on an intermediate microeconomics exam and thinking (s)he would get away with it. This is the world's largest car manufacturer intentionally deceiving federal and state governments by gaming their enforcement strategy. While some cynic might remark that folks will always cheat when there's a dollar to be made, I think we can rethink how we design regulations by building in evaluation from the get go.

Michael Greenstone, who spends his summers two doors down the hall from me, has thought a lot about this recently. In the U.S., we pass many of our major regulations based on ex ante cost-benefit analyses. Testifying on Capitol Hill, he recently made two suggestions that would significantly improve things.

First, he argues that we should institutionalize the ex post review of economically significant rules "in a public way so that these reviews are automatic in nature." He also argues that rules already in effect should start being reviewed using retrospective analysis. The relevant agencies should commit to changing or abandoning rules based on these evaluations, or possibly create new rules based on these evaluations.

The big issue here, of course, is who should review these policies? Greenstone proposes creating a regulatory-analysis division within the Congressional Budget Office. This division would conduct the regularly scheduled reviews and also conduct reviews at the request of lawmakers. I would go one step further and argue that these reviews should not only be staffed with government employees, but require the review and participation of independent academics. There is precedent for this model.

The certainty of independent review of policies and enforcement strategies significantly drives up the probability of detection, which would diminish the expected profits from cheating, by firms large and small. Plus, we are spending scarce public funds on environmental regulation. We should spend it on what works. And we need to figure out what that is.

Crossposted from the blog of the Energy Institute at Haas.