Why spend billions for no public benef it?

Should a U.S. federal agency issue a regulation that will impose up to US$3.5 billion in costs next year, and billions more in the coming decade — while delivering essentially no benefits?

That sounds crazy. But a few weeks ago, the U.S. Department of Agriculture proposed to do exactly that.

OK, not exactly — but pretty close. The proposal is the outgrowth of the longstanding national battle over whether to require labels for bioengineered (or genetically modified) foods. The USDA’S analysis of costs and benefits deserves careful attention, even if it raises serious questions about its own proposal.

In the summer of 2016, Congress required the USDA to impose such labels. Last month, the department responded by inviting public comments on the rule, called the Proposed National Bioengineered Food Disclosure Standard.

The department was candid about the high costs of its proposal. In the first year, companies — mostly food manufacturers — would have to spend somewhere between US$600 million and US$3.5 billion in compliance costs. (The range reflects uncertainty about some assumptions behind the estimates.) After an initial period of learning and adjustment, the annual costs would range from US$132 million to US$330 million. In the history of federal regulation, that may not qualify as monstrously expensive, but it’s a lot.

What do the American people get in return?

The department notes that there is no evidence that bioengineered food causes health risks. For that reason, its disclosure standard “is not intended to convey safety or health information.” It added that the rule “is not expected to have any benefits to human health or the environment.”

To drive the point home, it found that even if the new labels change the ratio of bioengineered to non-bioengineered food purchases, “there would be no impacts on human health or the environment.” That’s not a partisan conclusion. It is supported by the National Academy of Sciences, and in general it reflects the views of experts in both Republican and Democratic administrations. (To be sure, some experts do not rule out possible environmental risks.)

But that wasn’t the end of the story. The Department of Agriculture noted that many consumers have indicated an interest in the information and, in some surveys, say they would pay something to get it.

But the USDA didn’t trust those studies. People’s answers to survey questions may overstate their actual level of concern. And if people want to know whether food is bioengineered, it might be because they wrongly believe that it is unhealthy. If so, regulators should correct their mistake rather than cater to it.

At the same time, the department was aware that if the federal government did nothing, manufacturers might face a costly system of state-level labelling requirements. In 2014, Vermont enacted mandatory labelling legislation. In fact, that enactment provided the impetus for the federal legislation, which pre-empted state requirements, and so attracted support from manufacturers who favoured federal uniformity (and a less aggressive set of labelling standards).

The upshot of the USDA’S analysis is that if all manufacturers complied with Vermont’s approach, so that it became the de facto national standard, they would far higher costs than the federal proposal — between US$1.9 billion and US$6.8 billion in the first year alone. To that extent, the national standard would have benefits in the form of the cost savings.

It’s possible that manufacturers would not adopt the Vermont standard nationally. If they adopted it only in the Northeast, it would be less costly — in the USDA’S accounting, about the same as the federal proposal. The department also noted that without the federal standard, various states might

PRE-EMPT STATE LAWS THAT WOULD BE MORE COSTLY STILL!

mandate labels. It was unable to specify the resulting costs but said they would be very high.

In short: The federal disclosure standard would have no benefits for anyone — except insofar as it would preempt state laws that would be more costly still!

To be sure, the USDA could have tried to quantify benefits by referring to studies indicating that consumers would be willing to pay for the labels. In addition, some analysts suggest that in the face of uncertain, incompletely understood risks of catastrophe, it makes sense to take precautions; some people, fearing environmental risks, believe that bioengineered food falls in that category.

It might have been productive for the department to engage these points. But it is clear that even if it had done so, it would have been reluctant to conclude that its proposal made much sense (even though it was mandated by law).

Skeptics might expect a federal agency in the Trump administration to highlight evidence against new regulation in any case. But the USDA’S analysis is consistent with the scientific consensus; it does not depart from the views of federal regulators in the Obama administration.

Here’s a simple summary of its conclusion: no government, federal or state, should be mandating labels for bioengineered foods.