On Government's Numbers

Cost-Benefit Analysis at the Office of Information and Regulatory Affairs

1. Under Executive Orders 12866 and 13563, agencies must quantify the costs and benefits of their regulations, and proceed only if the benefits justify the costs, to the extent permitted by law.

2. This is a key part of what we might think of as the informal Regulatory Constitution of the United States. It has been in place since the early 1980s.

3. I served as Administrator of the Office of Information and Regulatory Affairs (OIRA) from 2009-2012, OIRA helps oversee the regulatory process, including the Regulatory Impact Analyses (RIA) that include detailed discussions of costs and benefits.

4. Agencies, such as EPA, DOT, USDA, DOJ, HHS, and DHS, are responsible for the production of RIAs, but OIRA reads and assesses them (carefully), and submits them for White House and interagency review. For example, the Council of Economic Advisers might not agree with the agency’s calculations, and there might have to be a lengthy discussion to reach consensus. (When I was at OIRA, CEA was usually in the drivers’ seat on economic issues, because its economists were especially terrific.)

5. The RIA might be relevant in court, but in the first instance, its purposes include (1) helping to inform the crucial judgment about whether to proceed with the regulation at all, (2) helping to inform the crucial judgment about which regulation to issue (eg, which alternative - more stringent or less stringent, or different in some way), and (3) public transparency (the American people deserve to know about the costs and benefits of national regulations).

6. The RIA might inform both regulation and deregulation. If deregulation would save $200 million but cost $500 million, it would not seem to be a good idea.

7. I had colleagues at the University of Chicago who told me, with great confidence, that the government cooked the numbers and that the numbers could not be trusted.

8. My experience was this: My colleagues were wrong. The numbers were produced by economists and others within the agencies who were not political appointees and who did not have a political ax to grind. In general, the numbers were pretty solid. They were also scrutinized internally (a lot) and were subject to public scrutiny as well (through the notice-and-comment process).

9. We tried hard to ensure both internal and external scrutiny.

10. There was not a single occasion on which President Obama said or even hinted that the numbers should be changed.

11. On rare occasions, political officials questioned the numbers. These occasions fell into two categories. (1) Political officials had a substantive point to make, eg, did you consider this cost or that benefit? Isn’t one or another number unreliable? Sometimes the point was not a good one; sometimes the point was at least reasonable to raise. (2) Political officials wanted numbers that would cast the rule in a good light (this was very, very rare; maybe it happened 3 times in 4 years?). They didn’t want agencies to lie, but they did want to cast the rule in a good light.

12. On 11 (2), they never succeeded. I would have resigned before allowing phony numbers to go out. I actually said that internally, more than once.

13. Agencies did not “lawyer the numbers,” in the sense of putting the numbers in a phony way that would make the rule look good.

14. But there were some biases. For example, EPA resolved some issues - eg, the rebound effect, that is, the increase in driving that would come from fuel-efficient cars - in a way that favored environmentally friendly rules. That is, the EPA thought the rebound effect would be relatively small. DOT thought the rebound effect would be larger. Still, both agencies pointed to evidence and were within the bounds of reason.

15. In addition, agencies were not unaware of the preferences of Cabinet heads, and of the president, and on some occasions, they may have resolved some issues, credibly and not unreasonably, but in a way that did not fit poorly, or that fit well enough, with those preferences. Any such resolution would be subject to a reality check.

16. If you read the RIAs accompanying some rules in the first Trump administration, including some of the fuel economy rules, you will be impressed by the care and the integrity of the analysis, especially on the vexed consumer welfare issue. (I omit details.)

17. My own experience, over the course of my four years at OIRA (and this is consistent with my experience in other positions in the federal government, going back to 1981), the RIA process gets high marks for integrity. Agency economists are nonpolitical.

18. OIRA provides a good reality check. Sometimes an agency will float an idea with OIRA, and OIRA economists, or OIRA (nonpolitical) leadership, will say, not unreasonably, “that really will not work.”

19. Something like what I have said here can be said, more or less, across Republican and Democratic administrations. Most political leaders (presidents, Cabinet heads) have empowered agencies to follow the evidence and the best available research on costs and benefits.

20. It is true that sometimes Democratic and Republican administrations make different assessments. But they agree far more often than you might expect, and when they disagree, the disagreements are usually not that large, and the relevant numbers are generally within the bounds of reason.

21. The basic integrity of the RIA process, and of the OIRA process, is essential to the operation of the modern regulatory state. It is an extraordinary achievement, and it is an inspiration to many nations around the world.