Is regulation good for you?

• Author: Hahn, Robert W.

Will all federal regulations soon pass a benefit-cost test? If the OMB's 2003 report is any indicator, the answer may be yes-at least for some categories of regulations. Applying the midpoint of OMB's estimates for quantified costs and benefits of agency rules, we find that 100 percent of regulations studied would pass a benefit-cost test Jot several agencies, and about 80 percent would pass for all agencies considered. Moreover, these regulations would confer at least $100 billion annually in net benefits, again using OMB's numbers. Sound too good to be true? That's probably because it is.

We argue that OMB's numbers are plausible, given the methodology that OMB uses. Whether they are reasonable is less clear. Some work by economists on related sets of regulations suggests that the percentage could be lower. A survey of experts in the field also casts doubt on the estimates of the number of regulations that would pass a benefit-cost test derived from OMB's report. The experts also suggest, in line with academic research, that there is considerable room .for improvement in regulations that pass a benefit-cost test. We, conclude with several suggestions for improving the regulatory process.

1. INTRODUCTION

   The Office of Management and Budget (OMB) recently issued its sixth report on the costs and benefits of regulation. Normally not for prime time, this report made the front page of the Washington Post. (1) The punch line was that the benefits of clean air regulations "during the past decade were five to seven times greater in economic terms than were the costs of complying with the rules." (2)

   Will almost all federal regulations soon pass a benefit-cost test? If the OMB's 2003 report (3) is any indicator, the answer may be yes. Using OMB's numbers, we find that 100% of regulations already pass a benefit-cost test for several agencies. (4) Furthermore, the aggregate net benefits of regulation could be substantial. For example, the OMB 2003 Report estimates that for the ten year period from October 1, 1992, to September 30, 2002, the estimated total annual quantified benefits for major federal rules were between $146 billion and $230 billion, and the total annual quantified costs ranged from $36 billion to $42 billion. (5) That yields a minimum of over $100 billion annually in aggregate net benefits.

   In this paper, we examine the OMB's numbers in detail to assess their plausibility and implications. Part II summarizes data on the fraction of regulations that pass a benefit-cost test, using OMB's data as a starting point. Part III examines whether OMB's numbers are reasonable. We conclude that OMB's numbers are plausible, given the methodology that OMB uses. Whether they are reasonable is less clear. My suspicion is that they are not, and we present some evidence and new survey research that supports this view. Part IV considers whether regulations and regulatory analysis are getting better or worse with time. Some evidence suggests that there is no obvious trend--either in the fraction of regulations passing a benefit-cost test or the quality of regulatory analysis. (6) But a new survey suggests that experts think regulations may be getting worse over time, at least as measured by the fraction likely to pass a benefit-cost test over the last two decades. (7) Finally, Part V summarizes my findings and makes several suggestions for improving the regulatory process.

2. HOW MANY REGULATIONS PASS A BENEFIT-COST TEST?

   Before asking how many regulations pass a benefit-cost test, a word is needed about the nature of the test. We will use a test that focuses on quantifiable benefits and costs in the interest of simplicity. It is not because we think unquantifiable benefits and costs are unimportant in making decisions, but rather because we do not have a simple way to address them here.

   OMB basically takes the agency's analyses of the expected economic impacts of regulation as given and monetizes benefits where it can. (8) For example, if the agency computes the tons of pollution reduced or the number of lives saved, OMB will monetize those numbers. Further, OMB only counts regulations for which a substantial portion of costs and benefits was quantified and monetized by the agency or, in some cases, monetized by OMB. (9) This is important because there are many regulations for which agencies do not quantify any benefits. (10) The question naturally arises as to whether there are really benefits to those regulations. In addition, OMB does not revisit any of the assumptions or numbers in the agency's analyses. Whatever the agency says is gospel for purposes of OMB's analysis.

   One reason OMB may take the agency numbers as given is that it has already reviewed the benefit and cost numbers as part of the regulatory review process. During that review process, OMB can ask the agency to make changes and use different assumptions at both the proposed and final stages during both informal and formal reviews. Even though OMB can provide such input, this process is a negotiation. And OMB has very limited resources to review these regulations. Thus, while OMB has input, it does not provide a detailed review of many aspects of these regulations. OMB neither does its own analysis from scratch because of resource limitations, nor does it effectively enforce its own regulatory guidelines. (11) Thus, it is not clear the extent to which OMB technocrats would actually approve of the agencies' analyses if they were in a more academic environment. We speculate that the reason that OMB takes the agency numbers as given is because it would be politically difficult for an Administration to deal with two sets of benefit-cost numbers--one from the regulatory agency and a second from OMB.

   Interestingly, OMB reports on the aggregate net benefits of regulation, but it does not report on the number of regulations that are likely to pass a benefit-cost test by category. We believe information on the number of regulations that could pass a benefit-cost test is potentially useful to decision makers. If a large fraction of regulations failed a benefit-cost test in particular categories, there might be more need for oversight. (12) Of course, if the oversight process resulted in substantial improvements in net benefits, then it could still be quite useful. (13) The problem is that it is very difficult to ascertain the impact of OMB oversight. (14)

   Some insight into the number of regulations that pass a benefit-cost test using OMB's numbers is shown in Table 1. The table presents various estimates of benefits and costs using the low and high endpoints for benefits and costs along with the midpoint. (15) The range of regulations that pass a benefit-cost test appears to be between 60% and 90%, depending on one's assumptions. (16) Furthermore, some agencies have perfect batting averages despite some controversial regulations. The Department of Energy, for example, has several regulations requiring energy efficiency standards that supposedly pass muster. (17) The Department of Health and Human Services ("HHS"), similarly, has estimated benefits that exceed costs for all rules, including privacy standards for health records a regulation that has been quite controversial in the medical community. (18) The Department of Education and the Department of Housing and Urban Development ("HUD") also have perfect records, but have finalized only a few major regulations. (19)

   Recall that OMB's analysis does not include regulations with benefits that were not estimated. In order to measure the extent of the bias caused by the omission of these regulations, we calculated the number of regulations that would pass a benefit-cost test for each agency if these regulations were included in the totals. There were a total of thirty-four more regulations to consider that were included in the OMB 2003 Report as major regulations, but excluded from their totals because the benefits or costs were not estimated. (20)

   We assign zero benefits to these regulations for lack of a better assumption. Some would argue that zero is a lower bound and can lead to misleading results. We would argue that this is a matter of interpretation. For example, if a regulation had some quantified costs, and benefits were assigned a zero value, then quantifiable net benefits would be negative, and the regulation would not pass a benefit-cost test based on quantifiable net benefits. But the regulation could still pass a more broadly defined benefit-cost test if non-quantifiable benefits were included in the final decision.

   Homeland security regulations provide a good example. So far, it has been very difficult to quantify the benefits of those regulations. At the same time, we would argue that it is useful to put pressure on the agency to try to quantify the benefits of those regulations to the extent feasible to avoid wasteful social expenditures.

   Note that if some costs and all benefits are left out of the calculation, the direction of bias in the net benefits estimate is unclear. That is, the estimate could be high or low relative to some objective estimate of net benefits. This is so precisely because we do not know how the unquantified costs and the unquantified benefits compare without further information.

   In short, we think it is not unreasonable to assign a zero dollar value to unquantified benefit and cost categories for three reasons. First, it would provide regulatory agencies with an incentive to provide more information on quantifiable benefits and costs. Second, any other assumption seems totally arbitrary since we do not have information on the actual...