Week 3 - Congressional Testimony Regarding the Impacts of Executive Orders on Public Agencies

Article

The Trouble w ith Counting: Cutting Through the R hetoric o f Red Tape C utting

Jodi L. Shortt

“You know that I am called the Count Because I really love to count I could sit and count all day Sometimes I get carried away I count slowly, slowly, slowly getting faster Once I’ve started counting it’s really hard to stop Faster, faster. It is so exciting! I could count forever, count until I drop.”

Sesame Street - The Count’s Counting Song Lyrics | Metro Lyrics

INTRODUCTION On January 30, 2017, President Donald J. Trump signed Ex­

ecutive Order (EO) 13,771, “Reducing Regulation and Control­ ling Regulatory Costs. To promote deregulatory goals, EO 13,771 requires administrative agencies to repeal two regula­ tions for every one they propose or issue, leading many to refer to it as the 2-for-l * 1 Order. The Office of Management and Budget (OMB) has published detailed guidance instructing agencies how to implement this Executive Order, which con­ strains their ability to promulgate new regulations under their

t Professor of Law, Honorable Roger J. Traynor Chair, UC Hastings Col­ lege of the Law. For insightful comments and discussions, I am grateful to Scott Dodson, Jared Ellias, Dan Farber, Erik Gerding, Sarah Light, Dave Owen, Zach 1 rice, Dont Reiss, Reuel Schiller, David Zaring, and participants in the Dereg­ ulatory Frontiers Conference at UC Hastings College of the Law. I am deeply indebted to Tiffanie Ellis for her heroic research assistance. Copyright © 2018 by Jodi L. Short.

1. This Article will use either 2-for-l, the Order, or EO 13,771 to refer to Executive Order 13,771 that President Trump signed on January 30 2017 re­ quiring regulation counting.

93

94 MINNESOTA LAW REVIEW [103:93

statutory m andates.2 While many com m entators have derided EO 13,771 as silly and irrational,3 it would be a m istake to dis­ miss 2-for-l as a one-off political stunt. In fact, the idea m otivat­ ing the Order emerges from a larger intellectual project arguing th a t economic growth is being ham pered by the “sheer quantity of regulation[s] .”4 In a string of studies,5 researchers have a t­ tem pted to establish this relationship by counting regulations

2. Of f ic e o f M g m t. & Bu d g e t , E x e c . Of f ic e o f t h e P r e s id e n t , OMB M e m o . N o . M-17-21, Gu id a n c e Im p l e m e n t in g E x ec u tiv e Or d er 13,771, Ti ­ tled “R ed u c in g R eg u la tio n and Co n tr o l lin g R egulatory Co sts” (2017) [h erein after OMB GUIDANCE] (explaining policy reg a rd in g rep ealin g two ex ist­ in g reg u latio n s w henever a n agency p ro m u lg ates one new regulation).

3 See, e.g., C aroline Cecot & M ichael Liverm ore, The One-In, Two-Out Executive Order Is a Zero, 166 U. PA. L. REV. ONLINE 1, 11 (2017) (questioning th e ra tio n a lity of th e O rd er’s m ethodology for d ereg u latin g in lig h t of th e OMB’s s trin g e n t co unting req u irem en ts, w hich could slow d ereg u lato ry in itiativ es, and th e fact th a t “[i]f agencies issu e no reg u latio n s, th e n th e O rd e r’s req u irem en ts to rep eal a t le a s t two ex istin g reg u la tio n s w ould n o t get trig g e re d a t all”); S u san D udley, Putting a Cap on Regulation, 42 ADMIN. & REG. L. NEWS 4, 5 (2017) [h erein a fter Capping Regulation] (suggesting th a t it would be irra tio n a l no t to consider n e t b en efits to “d istin g u ish a good ru le from a b ad r u le ”); N icholas A sh­ ford Trump Rejects Science, Technology, Economics, and the Constitution With His Two-for-One Executive Order, H U FFPO ST: BLOG (Feb. 1, 2017, 12:35 PM), http://w w w .huffingtonpost.com /nicholas-ashford/trum p-rejects-science-tec_b_ 14552756.htm l (describing “th e folly of ab an d o n in g reg u la tio n as a m ean s of stim u la tin g technology an d em ploym ent”); M ichael C. Dorf, Trump s Cut-Regu­ lations Order Is Plain Stupid, NEWSWEEK (Feb. 5, 2017), http://w w w .new sw eek.com /m ichael-dorf-trum ps-cut-regulations-order-plain-stupid -551293 (pointing o u t th e illogical elem en ts an d th e m ean s of evading th e O r­ der); D an F arb er, T ru m p ’s 2-for-l Order: Legal Issues, LEGALPLANET (Feb. 6, 2017), http://legal-planet.org/2017/02/06/trum ps-2-for-l-order-legal-issues (ex­ p lain in g five different ad m in istra tiv e law issu es arisin g u n d er th e O rder), D an F arb er. Trum p’s Anti-Regulation Executive Order, LEGALPLANET (Jan . 30, 2017), http://legal-planet.org/2017/01/30/trum ps-anti-regulation-executive -order (“W h a t i t will do is cause chaos an d u n c e rta in ty .”); Sally K atzen, Cutting Federal Regulations? Let’s Be Sm art About It, THE HILL: CONGRESS BLOG (Dec. 12, 2016, 12:55 PM), http://thehill.com /blogs/congress-blog/econom y-budget/ 309941-cu ttin g -fed eral-reg u latio n s-lets-b e-sm art-ab o u t-it (pointing o u t t h a t a two-for-one re q u ire m e n t could affect u n co n tro v ersial reg u latio n s th a t “enable th e governm ent to function an d A m ericans to go ab o u t th e ir lives ).

4. See O m ar Al-Ubaydli & P atric k A. M cLaughlin, RegData: A Numerical Database on Industry-Specific Regulations for All United States Industries and Federal Regulations, 1997-2012,11 REG. & GOVERNANCE 109, 110 (2017) [here­ in a fte r RegData]; see also infra n o te 7.

5. Clyde Wayne Cr e w s , J r ., Co m p e t it iv e E n t e r . I n s t ., T e n T h ou san d COMMANDMENTS: AN ANNUAL SNAPSHOT OF THE FEDERAL REGULATORY STAIE (2017), https://cei.org/sites/default/files/T en% 20T housand% 20C om m andm ents % 202017.pdf [h erein after TEN THOUSAND COMMANDMENTS] (providing a n over­ view of P re sid e n t T ru m p ’s executive actions an d arg u in g for th e need for a tr a n s p a r e n t reg u la to ry state); RegData, supra note 4, a t 109-10; B re n t Coffey, P a tric k A. M cLaughlin & R obert D. Tollison, Regulators and Redskins, 153 PUB.

2018] THE TROUBLE WITH COUNTING 95

and correlating these counts to various macroeconomic outcomes of interest, like U.S. employment, productivity, and competitive­ ness.

This intellectual project appears to have two principal aims. The first is political. Regulation counting studies radically sim ­ plify complex regulatory phenomena to make criticisms of regu­ lation salient to law m akers and the general public and to bolster political support for deregulatory policies. The second is em piri­ cal and, ultim ately, legal. By producing scholarly literatu re doc­ um enting a correlation between the num ber of regulations and negative economic outcomes, anti-regulatory scholars and advo­ cates generate a body of empirical research th a t agencies and courts can rely on in im plem enting and upholding the legality of deregulatory counting policies like 2-for-l.* * * * * 6

The roll-out and m arketing of EO 13,771 plainly demon­ stra te the political value of counting studies. President Trump and his supporters have repeatedly blamed the num ber of fed­ eral regulations for negative economic outcomes and claimed th a t they m ust reduce the num ber of regulations to spur benefi­ cial economic outcomes. On the campaign trail, President Trump

Ch o ic e 191, 202 (2012) [hereinafter Regulators and Redskins]-, John W. Daw­ son & John J. Seater, Federal Regulation and Aggregate Economic Growth, 18 J. ECON. GROWTH 137, 137 (2013) (describing the relationship between regula­ tion and macroeconomic indicators and variables); Peter L. Strauss, Publication Rules in the Rulemaking Spectrum: Assuring Proper Respect for an Essential Elemen t, 53 ADMIN. L. R ev. 803, 808 (2001). See generally Casey B. Mulligan & Andrei Shleifer, The Extent of the Market and the Supply of Regulation, 120 Q.J. ECON. 1445 (2005) (providing analysis regarding the extent and limit of regula­ tion through a fixed-cost theory of regulation); J.B. Ruhl & James Salzman, Mo­ zart and the Red Queen: The Problem of Regulatory Accretion in the Adminis­ trative State, 91 Ge o . L.J. 757 (2003) (detailing a systems-based model to understand regulatory burdens).

6. A pending lawsuit challenges EO 13,771 directly as an unconstitutional exercise of executive power because it “revise[s] statutes to condition issuance of new regulations on repeal of two or more existing regulations that offset the new costs and also challenges the OMB Guidance Memorandum implementing EO 13,771 as arbitrary and capricious in violation of the Administrative Proce­ dure Act. Plaintiffs’ Motion for Summary Judgment at 38, Pub. Citizen, Inc v Trump, 297 F. Supp. 3d 6 (D.D.C. 2018) (No. 17-253). The court ruled that the plaintiffs lacked both organizational and individual standing to pursue these claims, though final judgment has not yet been entered as of the date of this publication. Pub. Citizen, Inc. v. Trump, 297 F. Supp. 3d 6, 40 (D.D.C. 2018). Future challenges to EO 13,771 are likely to arise in challenges to agency deci­ sions to rescind particular regulations that are not justified by traditional reg­ ulatory analysis and that seem to be motivated primarily to the requirement to rescind two regulations before promulgating another one. Challenges are also likely to arise to agency decisions to delay rulemaking or deny rulemaking pe­ titions based on their need to first repeal two other regulations.

96 MINNESOTA LAW REVIEW [103:93

promised to “remove bureaucrats who only know how to kill jobs.”7 Upon signing EO 13,771, he said th a t the purpose of the Order was to elim inate “the repetitive, horrible regulations th a t h u rt companies, h u rt jobs.”8 At a briefing on the im plem entation of EO 13,771, President Trum p stood between two piles of p a ­ p e r - o n e more th a n six feet tall and comprised of multiple stacks, the other, a single stack about six inches high.9 “This is today,” the President said, gesturing toward the larger pile, tow­ ering over him .10 “This is 1960,” he said, indicating the small stack of p aper.* 11 He th en cut a piece of ceremonial red tape con­ necting the two piles, and went on to discuss how his ad m in istra­ tion’s regulation-cutting project had already resulted in in ­ creased wages, lower unemployment, rising economic growth, and “the stock m a r k e t. . . soaring to new record levels.”12 These claims have been echoed by supportive in terest groups. The U.S. Cham ber of Commerce, for instance, has been vocal in its sup­ port of 2-for-l on the grounds th a t it will “drive growth and jobs.”13 The Competitive Enterprise In stitu te has argued in sup­ port of the Order as well, stating th a t the removal of regulations “would provide genuine economic stim ulus.”14

7. Tamara Keith et al., Fact Check: Donald Trump Unveils His Economic Plan in Major Detroit Speech, NPR ONLINE (Aug. 8, 2016), http://www.npr.org/ 2016/08/08/488816816/donald-trump-looks-to-turn-the-page-on-bad-week-with -economic-speech; see also Office of Budget & Spending, Remarks by President Trump on Deregulation, WHITEHOUSE.GOV (Dec. 14, 2017), https://www. whitehouse.gov/briefings-statements/remarks-president-trump-deregulation (commenting that regulations have cost the “country trillions and trillions of dollars, millions of jobs, countless American factories, and devastated many in­ dustries”).

8. U PI, Trump Signs Executive Order on Regulatory Reform, BREITBART (Feb. 24, 2017), http://www.breitbart.com/news/trump-signs-executive-order-on -regulatory-reform.

9. CNBC, President Donald Trump: Regulation Is a Stealth Taxation, Let’s Cut the Red Tape, YOUTUBE (Dec. 14, 2017), https://w w w .youtube.com /w atch? v= B tkuy W T phm Q .

10. Id. 11. Id. 12. Office of Budget & Spending, Remarks by President Trump on Deregu­

lation, WHITEHOUSE.GOV (Dec. 14, 2017), https://www.whitehouse.gov/ briefmgs-statements/remarks-president-trump-deregulation.

13. Robb M andelbaum , Trump’s Regulation Order: What It Means For Small Businesses, FORBES (Jan . 31, 2017), https://w w w .forbes.com /sites/ robbmandelbaum/2017/01/31/trumps-regulation-order-take-it-seriously-maybe -not-literally/#f2d0c895d375 (describing the organizations and interest groups that support the Order).

14. T e n T h o u sa n d Co m m a n d m en ts, supra note 5, a t 54.

2018] THE TROUBLE WITH COUNTING 97

In this Article, I evaluate the empirical and legal value of regulation counting studies in light of prevailing standards in social science research. I argue th a t regulation counting studies do not, and cannot, rationalize deregulatory policies like 2-for-l because they are, themselves, irrational and empirically u n ­ sound. The trouble with counting is th a t it does not m easure v ar­ iables capable of yielding valid causal statistical inferences. The validity of empirical claims arising out of regulation counting studies depends on th eir proponents’ ability to dem onstrate th a t regulation counts are an accurate m easure of some relevant con­ struct, and to theorize and empirically establish a causal rela­ tionship between th a t construct and the economic outcomes of interest. In the social science literature, this is referred to as “construct validity.”15 It is commonly claimed th a t regulation counts are a proxy for the construct of costs or burdens on reg u ­ lated en tities.16 While these claims have undeniable political ap ­ peal, as I elaborate below, there are no good reasons to believe th a t counting the num ber of regulations is a useful proxy for the costs or burdens of regulation. If regulation counts are not a good proxy for mechanisms like cost or burden th a t are purported to affect economic outcomes, they have lim ited ability to support causal claims about the relationship between regulation and eco­ nomic outcomes.

Why should anyone care about the empirical validity of a relatively small body of poorly (or cynically) conducted research? First, it is im portant to promote integrity in empirical research on regulation, because such research shapes regulatory policy.17

15. See generally Vladimir Atanasov & Bernard Black, Shock-Based. Causal Inference in Corporate Finance Research, 5 CRITICAL Fin. Rev. 207 (discussing construct validity” and showing that the usefulness of a count variable requires

theoretical construct validity); Michael Klausner, Fact and Fiction in Corporate Law and Governance, 65 STAN. L. REV. 1325 (2013) (demonstrating that count­ ing studies in corporate governance do not actually measure anything meaning­ ful).

16. These constructs tend to be vaguely defined in the literature, but they generally refer to the direct costs regulated entities incur to operate in compli­ ance with legal obligations or the opportunity costs of lost efficiency or produc­ tivity attributable to regulations. See infra note 223.

17. The use of unsound empirical studies is nothing new. Cost benefit anal­ ysis (CBA) was initially sold based on a suite of widely circulated regulatory scorecards that purported to demonstrate that the costs of government regula­ tions vastly outweigh their benefits. These studies have been debunked. See Frank Ackerman & Lisa Heinzerling, Priceless: On Knowing the Price of Everything and the Value of Nothing 35-40 (2004); Richard W. Parker, Grading the Government, 70 U. CHI. L. Rev. 1345, 1347-48 (2003). The practice

98 MINNESOTA LAW REVIEW [103:93

Administrators are policy-makers who engage in fact-based de­ cision-making, and the facts upon which they rely often come from empirical research. Administrative decisions are only as sound as the research upon which they are based, thus there is a strong interest in promoting high quality research on regula­ tion.

Second, empirical research is used by agencies to support their policies on judicial review. The Administrative Procedure Act (APA) 18 requires agencies to articulate a rational connection between the facts before the agency and the policy choices made by the agency. 19 Administrative decisions that lack a rational foundation are deemed arbitrary and capricious and must be struck down by reviewing courts.20 Agencies often cite empirical studies to demonstrate the rationality of their decisions, and re­ viewing courts will find agency decisions arbitrary and capri­ cious if they are not sufficiently justified by the empirical evi­ dence before the agency.21

The logic of regulation counting studies that empirically tie regulation counts to macroeconomic outcomes appears calcu­ lated to resonate with widely accepted efficiency rationales for cost-benefit analysis (CBA). CBA, instituted through a series of executive orders issued by Presidents dating back to Jimmy Carter,22 rests on the premise that when agencies have the stat­ utory discretion to do so, they should make policy decisions in a

of CBA has improved as it has assimilated critiques of cruder versions. See gen­ erally Richard L. Revesz & Michael A. Livermore, Retaking Rationality: How Cost-Benefit Analysis Can Better Protect the Environment and Our Health (2008) (describing the need for cost-benefit analysis in the regula­ tory toolbox and how CBA’s flaws can be remedied).

18. 5 U.S.C.ch. 5(2011). 19. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 416—17

(1971). 20. See 5 U.S.C. § 706(2) (1966); Motor Vehicle Mfrs. Ass’n of U.S., Inc. v.

State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983); Citizens to Preserve Overton Park, 401 U.S. at 416.

21. See, e.g., State Farm, 463 U.S. at 46 (rejecting Department of Transpor­ tation’s decision to rescind rule requiring passive restraints despite studies sug­ gesting th at these requirements would increase seatbelt usage and save lives); Bus. Roundtable v. SEC, 647 F.3d 1144, 1148-49 (D.C. Cir. 2011) (rejecting the Securities and Exchange Commission’s proxy access rule for failing to include appropriately rigorous economic analysis regarding several of the issues before the agency).

22. See Exec. Order No. 12,044, 43 Fed. Reg. 12,661 (Mar. 23, 1978) (Carter EO directing all executive agencies to improve existing and future regulations); Exec. Order No. 12,291, 46 Fed. Reg. 13,193 (Feb. 17, 1981) (Reagan EO reduc­ ing burdens for existing and future regulations); Exec. Order No. 12,866, 58 Fed. Reg. 51,735 (Sept. 30, 1993) (Clinton EO reforming and improving regulations

2018] THE TROUBLE WITH COUNTING 99

way th a t maximizes aggregate welfare, meaning the benefits of regulation net its costs.23

There currently is broad scholarly and political consensus th a t “Presidents can legitim ately influence the development of regulations—a t least when the goal is to ensure th a t the regula­ tions promote societal welfare to the extent perm itted by law.”24 The requirem ent to maximize efficiency through adm inistrative policy has been described colloquially as nothing more th a n the proposition “th a t agencies should attem pt to produce more good th a n harm .”25 In fact, CBA has become so deeply entrenched in federal adm inistrative policymaking th a t Cass Sunstein, a prom inent former skeptic of CBA,26 now argues th a t it might be a rb itrary and capricious if agencies fail to justify th eir decisions based on an analysis of quantified benefits and costs (absent statu to ry m andates prohibiting them from doing so).27 Some have in terpreted recent U.S. Supreme Court case law to endorse this position.28

It is not surprising, then, th a t supporters of EO 13,771 have suggested th a t 2-for-l is merely an extension of cost-benefit analysis,29 and thus can be rationalized on sim ilar grounds. The

to maximize economic growth); Exec. Order No. 13,422, 72 Fed. Reg. 2,763 (Jan. 18, 2007) (Bush EO inserting CBA policies and preferences into regulatory pro­ cess); Exec. Order No. 13,563, 76 Fed. Reg. 3,821 (Jan. 18, 2011) (Obama EO directing agencies to stream line promulgation processes to b etter protect citi­ zens and to maximize economic benefits of the regulatory process).

23. Cass R. Sunstein, Cost-Benefit Analysis and Arbitrariness Review, 41 HARV. ENVTL. L. r e v . 1, 9-10 (2017) [hereinafter CBA & Arbitrariness Review],

24. Cecot & Livermore, supra note 3, a t 3. 25. CBA & Arbitrariness Review, supra note 23, a t 9. 26. See generally Cass R. Sunstein, Cost-Benefit A nalysis and the Separa­

tion of Powers, 23 ARK. L. R ev. 1267 (1981) (critiquing the executive branch’s use of cost-benefit analysis as a tool to control agency actions).

27. See CBA & Arbitrariness Review, supra note 23, a t 9. 28. See Capping Regulation, supra note 3, at 5 (“Indeed, recent Supreme

Court decisions have indicated th a t it would be unreasonable for agencies not to consider benefits and costs in m aking regulatory decisions.” (citing Michigan v. EPA, 135 S. Ct. 2699, 2707 (2015))).

29. See R e p ly in Support of Defendant’s Motion to Dismiss a t 20, Pub. Cit­ izen, Inc. v. Trump, 297 F. Supp. 3d 6 (D.D.C. 2018) (No. 17-253) (“Executive Order 13,771, like its predecessors, recognizes th a t agencies m ust inherently weigh conflicting goals, priorities, and associated costs as a necessary p a rt of reasoned decisionmaking under the APA.”); see also TED GAYER, ROBERT LlTAN & P hilip Wallach, Brookings Ctr. on Regulation & Mkts., Evaluating the Trump Administration’s Regulatory Reform Program 3 (2017), https://www.brookings.edu/wp-content/uploads/2017/10/evaluatingtrumpreg reform_gayerlitanwallach_102017.pdf [hereinafter BROOKINGS EVALUATION] (situating EO 13,771 as p art of a broader bipartisan effort to “require agencies

100 M IN N ESO T A L A W R E V IE W [103:93

difficulty w ith this gambit is th a t 2-for-l explicitly rejects the well-established efficiency maximization lodestar on which CBA rests. C ontrary to CBA, 2-for-l addresses only the costs (and not the benefits) of regulations, and the OMB Guidance M em oran­ dum im plem enting the Order expressly prohibits agencies from considering the benefits of regulation in deciding which regula­ tions m ust go and which may stay.30 This is, quite simply, u n ju s­ tifiable under fundam ental principles of welfare economics: “At least as far as economic theory is concerned, any new regulation that offers more benefits than costs should be undertaken, reg ard ­ less of its contribution to the aggregate regulatory cost to soci­ ety.”31 If EO 13,771 cannot be justified by the efficiency rationale underlying CBA, it m ust rest on some other principled basis. This is where regulation counting studies come in. They promise to provide empirical support for the policy of reducing regulation counts in the service of promoting desired (and ostensibly more efficient) macro-economic outcomes. This Article explains why they fail to deliver on th a t promise and, thus, why such studies cannot rationalize EO 13,771 or adm inistrative decisions based on it.32

Finally, it is im portant to u n derstand the regulation count­ ing project on its own term s because it reveals deeper insights into the broader project of deregulation. Specifically, the practice

to pay greater heed to analyzing the costs and benefits of major new regula­ tions ___ ”); Benjam in M. Miller et al., Rand Corp., Inching Toward R e­ form: Trump’s Deregulation and Its Implementation (2017) (characteriz­ ing EO 13,771 as a way of forcing agencies to more accurately account for costs in their cost-benefit analyses); Susan Dudley et al., Consumer’s Guide to Regu­ latory Impact Analysis: Ten Tips for Being an Informed Policymaker, 8 J. BEN- EFIT-COST Analysis 2, 187, 190 (2017) (“The OMB’s guidance on [EO 13,771] . . . may lead to an increased use of RIAs to examine the effects of mod­ ifying existing regulations, as well as prospective regulations.”).

30. See OMB GUIDANCE, supra n o t e 2, a t 2. 31. BROOKINGS EVALUATION, supra n o t e 29, a t 5 ( e m p h a s i s in o r ig in a l) . 32. Decisions made pursuant to EO 13,771 cannot be justified based solely

on the President’s power to direct executive agency policymaking. While the President has broad discretion to direct executive agencies to implement stat­ utes consistent with the administration’s policy preferences and priorities, the President does not have the power to direct agencies to make policy decisions that are arbitrary and capricious. See Pub. Citizen Health Res. Grp. v. Tyson, 796 F.2d 1479, 1495 (D.C. Cir. 1986). See generally Steven G. Calabresi & Saikrishna B. Prakash, The President’s Power To Execute the Laws, 104 YALE L.J. 541 (1994) (explaining that the President’s reach of power is broad when executing laws that are consistent with the administration’s policy); Elena Ka­ gan, Presidential Administration, 114 HARV. L. REV. 2245 (2001) (same); Strauss, supra note 5, at 4 (explaining that if the rules are appropriately adopted and do not question other authorities, they will generally be valid).

2018] THE TROUBLE WITH COUNTING 101

of regulation counting suggests th a t the tru e foundation of the deregulation movement lies not in concerns about economic effi­ ciency, but ra th e r in fervent feelings and fears about regulation’s restrictions on liberty, 33 particularly the freedom to conduct business. It is vital to recognize these feelings and fears as the wellspring of the deregulatory impulse if we ever are to advance legal and political debates about regulation. In the meantime, however, it is urgent to stress th a t inflamed regulatory passions cannot serve as a rational basis supporting regulatory reform policies im plemented by agencies.

The Article proceeds as follows. P a rt I describes the m eth­ odology of counting projects, including w hat counts and how it is counted. P a rt II lays out my critique of counting methodologies. I explain the necessity of construct and m easurem ent validity for generating valid statistical inferences and then discuss why reg­ ulation counts are not a valid m easure of any of the constructs th a t have been proffered by regulation counters, including the costs, burdens or constraints of regulation on regulated entities. P a rt III explores the question of what, if anything, regulation counts could signify even if they do not m easure costs, burdens, or constraints on regulated entities. I suggest in this P a rt th a t regulation counts may be an attem p t to represent w hat I call the “unquantifiable costs” of regulation, including regulation’s re ­ strictions on liberty and the psychic burdens it imposes on cer­ tain segments of the business community. I argue th at, standing alone, neither psychic burdens nor the desire to feel more free provide rational support for deregulatory counting policies. How­ ever, I argue th a t recognizing these unquantifiable costs of reg­ ulation could help advance dialogue about regulatory reform. Fi­ nally, P a rt IV discusses the very real harm s and costs of the counting project and explains why it is time to stop counting and s ta r t engaging in meaningful dialogue about specific societal problems and appropriate regulatory responses.

I. WHAT COUNTS? The first and critical step in any counting project is to define

the universe of things to be counted. T hat universe may be de­ fined capaciously, as it is by Count von Count, the Transylvanian expatriate on Sesame Street, whose sampling heuristic is: “there

33. See Jodi L. Short, The Paranoid Style in Regulatory Reform, 63 HAS­ TINGS L.J. 633, 636-37 (2012) (demonstrating that concerns about regulation’s restrictions on liberty dominated debates about regulatory reform between 1980 and 2005).

102 MINNESOTA LAW REVIEW [103:93

is always something to count.”34 Indeed, the Count will count w hatever he encounters in his im m ediate sensory field, be it hot dogs, oranges, building floors, telephone rings, or even the sheep in his friends’ dream s.35 Below, I describe the variety of different approaches regulation counters have tak en to define the things to be counted in th e ir universe. Section A describes the counting methodology of EO 13,771. Section B discusses the largely dis­ credited methodologies th a t have traditionally been used in counting studies. Section C lays out the counting methodology used by RegData, which to date has not been scrutinized. This Article’s critique of counting focuses on RegData because it p u r­ ports to be the state-of-the-art in regulation counting and has been m arketed to government officials as a tool for deciding which regulations to cut.

A. Counting Methodology of EO 13,771 Unlike many academic counting projects, EO 13,771 re ­

quires agencies to count both regulations and costs.36 In addition to the 2-for-l requirem ent, the Order im plem ents a type of reg­ ulatory budget37 th a t requires agencies to offset fully the costs of

34. Sesame Street: Ernie Takes on the Count’s Job (PBS television broadcast May 4, 2000), https://www.youtube.com/watch?v=xkfUIpziqwg.

35. See, e.g., id. 36. EO 13,771, § 2(a) provides that “[ujnless prohibited by law, whenever

an executive department or agency (agency) publicly proposes for notice and comment or otherwise promulgates a new regulation, it shall identify at least two existing regulations to be repealed.” Exec. Order No. 13,771, 82 Fed. Reg. 9339, 9339 (Jan. 30, 2017). In furtherance of this requirement, EO 13,771 also requires that “any new incremental costs associated with new regulations shall, to the extent permitted by law, be offset by the elimination of existing costs associated with at least two prior regulations.” Id.

37. A regulatory budget is a shadow budget that caps the costs an agency can “require private agents to consume in the pursuit of the regulatory goal.” Robert W. Crandall, Federal Government Initiatives To Reduce the Price Level, 1978 BROOKINGS P a p e r s o n E c o n . A c t i v i t y 401, 429 (1978). The idea of a reg­ ulatory budget was introduced by Robert Crandall, who saw it as a way of “con­ fronting regulators with the costs of their actions.” Id. A variety of different models for implementing a regulatory budget have been proposed. See, e.g., Christopher C. DeMuth, The Regulatory Budget, REGULATION 29, 30-31 (1980) (proposing that the federal government “establish an upper limit on the costs of [federal] regulatory activities to the economy and . . . apportion this sum among the individual regulatory agencies”); Susan E. Dudley, Can Fiscal Budget Con­ cepts Improve Regulation? 19 N.Y.U. J. LEGIS. & PUB. PO L’Y 259, 265 (2016) (“Operationally, a regulatory budget would share similarities with the fiscal budget.”); Eric A. Posner, Using Net Benefit Accounts to Discipline Agencies: A Thought Experiment, 150 U. PA. L . REV. 1473, 1477-84 (2002) (proposing a net benefit account budget, in which agencies would be required to keep positive

2018] THE TROUBLE WITH COUNTING 103

new regulations through the repeal of existing regulations.38 Ac­ cording to the OMB Guidance Memorandum, to comply w ith EO 13,771, agencies m ust issue two deregulatory actions39 for each regulatory action.40 This requirem ent is m eant to ensure th a t the increm ental costs associated w ith the total num ber of regu­ latory actions are fully offset by the cost savings of deregulatory actions.41

balances on ledgers that tally the benefits and costs of every regulation prom­ ulgated by the agency); Jeffrey A. Rosen & Brian Callanan, The Regulatory Budget Revisited, 66 ADMIN. L. REV. 835, 837 (2014) (suggesting that regulatory costs should be considered as the equivalent of tax dollars). Notably, most pre­ viously proposed regulatory budget designs envisioned involvement from Con­ gress as well as the executive branch. See BROOKINGS EVALUATION, supra note 29, at 6-11.

38. Exec. Order No. 13,771 § 2(c), 82 Fed. Reg. 9339 (Jan. 30, 2017). 39. The OMB Guidance Memorandum defines an “EO [13,771] deregula­

tory action” as an action that has been finalized and has total costs less than zero. OMB GUIDANCE, supra note 2, at 4. EO 13,771 deregulatory actions are not limited to those actions that would be defined as significant under EO 12,866, and they may be issued in a wide range of forms, including rulemaking; guidance or interpretive documents; or streamlining of information requests like recordkeeping, reporting, or disclosure requirements. Id. The OMB Memo­ randum explains that EO 13,771 regulatory actions subsequently overturned by Congress, for instance under the Congressional Review Act, count as “EO [13,771] deregulatory actions.” Id. at 7. By contrast, it takes a case-by-case ap­ proach to regulatory actions vacated by judicial order. See id.

40. OMB GUIDANCE defines “EO [13,771] regulatory action” as a “signifi­ cant regulatory action” that has been finalized, and that imposes total costs greater than zero. Id. at 3. “Significant regulatory action,” as defined by refer­ ence to EO 12,866, § 3(f), is defined in th at EO as:

any regulatory action that is likely to result in a rule th at may: (1) Have an annual effect on the economy of $100 million or more or adversely affect in a material way the economy, a sector of the economy, produc­ tivity, competition, jobs, the environment, public health or safety, or State, local, or tribal governments or communities; (2) Create a serious inconsistency or otherwise interfere with an action taken or planned by another agency; (3) Materially alter the budgetary impact of entitle­ ments, grants, user fees, or loan programs or the rights and obligations of recipients thereof; or (4) Raise novel legal or policy issues arising out of legal mandates, the President’s priorities, or the principles set forth in this Executive Order.

Exec. Order No. 12,866, 58 Fed. Reg. 190 (Oct. 4, 1993). The Memorandum ex­ plicitly includes significant guidance and interpretive documents in the defini­ tion of EO 13,771 regulatory actions. OMB GUIDANCE, supra note 2, at 3. It specifies that the Order’s requirements apply to all EO 13,771 regulatory ac­ tions issued after noon on January 20, 2017. See id. at 1-2.

41. According to the Memorandum, only those regulatory impacts that have traditionally been counted as costs when taking a regulatory action should be counted as cost savings when taking an EO 13,771 deregulatory action. Id. at 9. In other words, cost savings that an agency has traditionally counted as benefits when taking regulatory action cannot be counted as reductions in costs for pur­ poses of EO 13,771 deregulatory actions. Thus, the guidance makes clear that

104 MINNESOTA LAW REVIEW [103:93

This methodology requires determ ining w hat counts as a regulatory or deregulatory action, and OMB has decided th a t the two will be distinguished by focusing on a particu lar set of costs. The OMB Guidance M emorandum provides th a t, generally, costs should be counted using the methods and concepts articu­ lated in OMB Circular A-4, which was released in 2003, to assist agencies in conducting cost-benefit analyses required under EO 12866.42 However, costs are defined much more narrowly under EO 13,771. As has been widely observed and criticized,43 the only costs th a t count for purposes of EO 13,771 are compliance costs to regulated entities.44 For instance, lost benefits to the public or to regulated entities are not to be counted as costs of deregula­ tory actions.45 So, for instance, if an energy conservation regula­ tion produces m easurable cost savings to regulated entities in

agencies should not count net costs of deregulatory actions, but only those fi­ nancial impacts that would typically appear on the cost side of the cost-benefit analysis ledger in the initial regulatory analysis. See id. at 9.

42. Id. 43. See Cecot & Livermore, supra note 3, at 3 (“We conclude that the Order

is not calibrated to maximize social welfare because it narrowly focuses on the costs of regulation to regulated entities and fails to acknowledge the benefits of regulation.”); Capping Regulation, supra note 3, at 5 (suggesting that it would be irrational not to consider net benefits to “distinguish a good rule from a bad rule.”); Jeffrey S. Lubbers, Comments on OMB’s Interim Guidance Implement­ ing Section 2 of Executive Order 13,771 “Reducing Regulation and Controlling Regulatory Costs", 42 ADMIN. & REG. L. NEWS 7, 8 (2017) (asserting th at the “overall main shortcoming” of EO 13,771 is “th at it does not account for the benefit of regulations at all”); David A. Dana & Michael R. Barsa, The High Cost of Cutting Regulatory Costs, CHI. TRIB. (Feb. 24, 2017), http://www .chicagotribune.com/news/opinion/commentary/ct-epa-anti-regulatory-trump -epa-executive-order-perspec-0227-jm-20170223-story.html (“By focusing on only the costs of complying with the regulation, and not on the costs th at the regulation is trying to prevent, Trump’s order puts us all at grave risk.”); Jody Freeman, Trump’s “2 for 1” Executive Order, ENVTL. L. HARV. (Jan. 30, 2017), http://environment.law.harvard.edu/2017/01/freemanstatement (arguing that EO 13,771 would “strangle even the most beneficial rules under the guise of cutting red tape”); Amit Narang, The Stunning Triumph of Cost-Cost Analysis, R e g . R e v . (Feb. 19, 2017), http://www.regblog.org/2017/02/19/narang-stunning -triumph-cost-cost-analysis (“[T]he [Order] spells the end of cost-benefit analy­ sis and the rise of a new form of analysis that focuses only on regulatory costs while ignoring benefits.”).

44. If an agency is unsure whether an ambiguous item counts as a cost or a benefit, the Memorandum provides that it should be categorized to conform to accounting conventions the agency has followed in past analyses. So, for in­ stance, if the agency has historically categorized fuel savings associated with energy efficiency investments as benefits, it should continue to do so and should not count them as negative cost savings when deregulating. OMB GUIDANCE, supra note 2, at 9.

45. Id.

2018] THE TROUBLE WITH COUNTING 105

the form of lower energy costs, the loss of this benefit cannot be counted as a cost of repealing this regulation.

Although EO 13,771 contains a cost param eter, straight-up regulation counting rem ains fundam ental to its operation. Even if the repeal of a single existing regulation would fully offset the costs of a proposed regulation, the proposing agency still m ust identify a second regulation for repeal before proposing the new regulation. Any rule on the books is a candidate to be the second regulation repealed, so long as it imposes present costs greater th a n zero on regulated entities.46

B. Counting Methodology of Regulation-Counting Studies

In contrast to EO 13,771, counting studies typically count only regulations and not costs. This choice of object is odd in light of the fact, discussed below, th a t most regulation counters see the num ber of regulations as a proxy for the costs to regulated entities. Yet, instead of attem pting to count those costs, they count regulations instead.

Many regulation counting studies have taken a broad and undifferentiated approach to regulation counting, not unlike

46. The Memorandum clarifies the Order’s caveat that the 2-for-l deal ap­ plies “unless prohibited by law.” Id. at 5. Regulatory actions that are statutorily or judicially required do not count for purposes of the EO 13,771 regulatory ac­ tion tally. See id. at 2. The Memorandum interprets a statutorily required reg­ ulatory action as one for which Congress has provided by statute both “an ex­ plicit requirement and explicit timeframe.” Id. at 5. For example, the following statute would be considered to statutorily require a regulatory action: The FDA “shall issue nutrition labeling requirements within 10 years” of the statute’s enactment date. Id. However, a statute that required the FDA to issue nutrition labeling requirements when necessary to promote the purposes of the statute would not be considered to statutorily require regulatory action. A judicially required regulatory action is one for which there is a “judicially established binding deadline for rulemaking, including deadlines established by settlement agreement or consent decree.” Id. The Memorandum further explains that EO 13,771 does not change agencies’ obligations under statutes that prohibit the consideration of costs in determining a statutorily required standard. Id. at 8. However, it provides that while agencies may issue such regulations without first identifying offsetting deregulatory actions, they “will generally be required to offset the costs of such regulatory actions through other deregulatory actions taken pursuant to statutes that do not prohibit consideration of costs.” Id.

106 M IN N ESO TA LA W RE VIE W [103:93

Count von Count’s. For instance, studies have counted the num ­ ber of pages in the Federal Register47 or the Code o f Federal Reg­ ulations (C.F.R.) ,48 or the num ber of bits49 in digitized state s ta t­ utory codes.50 Cross-national comparative studies have used indices derived from Organization for Economic Cooperation and Development (OECD) and World Bank d a ta .51 Each of these

47. See, e.g., MAEVE P . CAREY, CONG. RESEARCH SERV., COUNTING REGU­ LATIONS: A n O v e r v i e w o f R u l e m a k i n g , T y p e s o f F e d e r a l R e g u l a t i o n s , AND PAGES IN THE Federal REGISTER (2016); Regulators and Redskins, supra note 5, at 195-201. See generally TEN THOUSAND COMMANDMENTS, supra note 5.

48. See Dawson & Seater, supra note 5; Clyde W. Crews, New Data: Code of Federal Regulations Expanding, Faster Pace Under Obama, COMPETITIVE E n t e r p r i s e INST. (Mar. 17, 2014), https://eei.org/blog/new-data-code-federal -regulations-expanding-faster-pace-under-obama.

49. Bit is short for “binary digit” and is the smallest, most basic unit of data in computing. See Bit, TECHOPEDIA, https://www.techopedia.com/definition/ 23954/bit (last visited Oct. 15, 2018); Bits and Bytes, STAN., https://web .stanford.edu/class/csl01/bits-bytes.html Oast visited Oct. 15, 2018).

50. See Mulligan & Shleifer, supra note 5, at 1469-72. 51. See, e.g., Simeon Djankov, Caralee McLiesh & Rita Ramalho, Regula­

tion & Growth, 92 ECON. LETTERS 395, 395-401 (2006); Norman V . Loayza, Ana Marla Oviedo & Luis Serven, Regulation & Macroeconomic Performance, in B u s i n e s s R e g u l a t i o n a n d E c o n o m i c P e r f o r m a n c e (Norman V. Loayza & Luis Serven eds., 2010) (analyzing both World Bank and OECD data); Alberto Alesina et al., Regulation and Investment 1-3 (N atl Bureau of Econ. Research, Working Paper No. 9560, 2003); Andrea Bassanini & Ekkehard Ernst, Labour Market Institutions, Product Market Regulation, & Innovation: Cross-Country Evidence 5-31 (OECD Econ. Dep’t, Working Paper No. 316, 2002); Daniel Kauf­ man, Aart Kraay & Massimo Mastruzzi, Governance Matters VIII: Aggregate and Individual Governance Indicators 1996-2008 (World Bank Pol’y Research, Working Paper No. 4978, 2009); Norman V . Loayza, Ana Maria Oviedo & Luis Serven, The Impact of Regulation on Growth and Informality: Cross-Country Evidence 3 (World Bank, Working Paper No. S3623, 2005); Giuseppe Nicoletti et al., Product and Labor Market Interactions in OECD Countries 5—108 (OECD Econ. Dep’t, Working Paper No. 312, 2001); Giuseppe Nicoletti & F . L . Pryor, Subjective & Objective Measures of the Extent of Governmental Regulations 4- 22 (OECD & Swarthmore College, Working Paper, 2001); Giuseppe Nicoletti & Stefano Scarpetta, Regulation, Productivity & Growth: OECD Evidence 5-63 (OECD Econ. Dep’t., Working Paper No. 347, 2003); Giuseppe Nicoletti, Stefano Scarpetta & Olivier Boylaud, Summary Indicators of Product Market Regula­ tion with an Extension to Employment Protection Legislation 7-84 (OECD Econ. Dep’t, Working Paper No. 226, 2000). This body of literature is beyond the scope of this article and has been thoroughly criticized elsewhere. See, e.g., TH E QUIET POWER OF INDICATORS: MEASURING GOVERNANCE, CORRUPTION AND THE RULE OF LAW (Sally Engle Merry, Kevin E. Davis & Benedict Kingsbury eds., 2015) (presenting case studies illustrating how global indices measuring regulation mask important issues of context and power); Janine Berg & Sandrine Cazes, Policymaking Gone Awry: The Labor Market Regulations of the Doing Business Indicators, 29 COMP. LAB. L . & POL’Y J. 349, 350 (2008) (arguing that the World Bank’s Doing Business index measures of labor market regulations are “based

2018] THE TROUBLE WITH COUNTING 107

counting methodologies has been roundly criticized for its inac­ curacy and inadequacy.52 The Federal Register contains not only final rules but proposed rules th a t may never become law, as well as public notices, executive orders, proclamations, and other presidential documents.53 In addition, many of its pages are con­ sumed by the extensive justifications th a t accompany final rules to satisfy the dem ands of arb itrary and capricious review.54 The C.F.R. codifies only finalized regulations, b u t “[p]age-count data are subject to the criticism th a t not all pages are equal. A page could be of enormous or trivial consequence to the economy.”55 Digital file size m easures suffer from sim ilar defects in th a t they are totally undifferentiated as to content. These shortcomings introduce m easurem ent error into studies employing these types of counts and underm ine th e ir empirical validity.

C. Counting Methodology of RegData The la te st innovation in regulation counting methodology is

the “Industry-specific Regulatory C onstraint D atabase,” devel­ oped by researchers a t the M ercatus Center a t George Mason University. Recently renam ed “RegData,” presum ably in an a t­ tem pt to m ask its ideological pedigree, the database tallies the num ber of regulatory constraints applicable to different indus­ tries over the fifteen-year period from 1997-2012.56 The m ethod­ ology used to count regulatory constraints is as follows:

on a partial and crude understanding of how labor markets and their institu­ tions function”); Kevin E. Davis & Michael B. Kruse, Taking the Measure of Law: The Case of the Doing Business Project, 32 LAW & SOC. INQUIRY 1095 (2007) (highlighting the challenges of measuring country-level legal variables in the face of legal complexity and uncertainty); Alvaro Santos, Labor Flexibil­ ity, Legal Reform, and Economic Development, 50 VA. J. Int’L L. 43, 45 (2009) (arguing that the World Bank Doing Business indicator “contains a number of very serious omissions that seem to stem from a flawed understanding of regu­ lation”).

52. See RegData, supra note 4, at 111-12; Ruhl & Salzman, supra note 5, at 769-75.

53. See generally Federal Register, About the Federal Register, Nat’L AR­ CHIVES, https://www.archives.gov/federal-register/the-federal-register/about .html (last updated Aug. 8, 2018) (detailing the types of sources included in the Federal Register).

54. Ruhl & Salzman, supra note 5, at 772. 55. RegData, supra note 4, at 112; see also Dawson & Seater, supra note 5,

at 139 (recognizing th at “[a] counting measure obviously is imperfect in that two identical values may comprise regulations of different types and, even within a given type, may represent regulations of different stringency”).

56. RegData, supra note 4, at 109.

108 MINNESOTA LAW REVIEW [103:93

Regulatory texts typically use a relatively standard suite of verbs and adjectives to indicate a binding constraint, such as “shall,” “must,” and “prohibited.” This observation motivated us to search the CFR for key­ words th a t are likely to indicate binding constraints. As a departure point, we search for five strings that are likely to limit choice sets: “shall,” “must,” “may not,” “prohibited,” and “required.” We refer to this set of five strings as “restrictions.”57

These search terms are the sole criteria applied to identify what counts as a regulatory constraint. 58 The creators classify re­ strictions identified by this methodology by the industries to which they apply. 59

RegData is a particularly important development in the reg­ ulation counting project for four reasons. First, RegData claims to be the state-of-the-art in regulation counting, improving upon prior methods by counting only those provisions of the C.F.R. that impose regulatory demands60 or restrictions61 and by map­ ping those restrictions by industry, thus allowing for cross-in­ dustry comparisons of regulatory outcomes. Second, while prior counting methods have been thoroughly critiqued, including by the authors of RegData, the shortcomings of RegData have not yet been addressed. Third, RegData was recently made publicly available, together with a call that it be used to identify causal relationships between regulation counts and various macroeco­ nomic outcome variables of interest, including employment and productivity.62 It is important to clarify the meaning of the vari­ ables it has constructed in order to properly interpret any results produced by future studies. Fourth, RegData is being marketed to government officials as “a research platform that allows users to quickly analyze state regulations and identify the specific in­ dustries most targeted by excessive regulation.” 63 It has been used by the Governor of Iowa to identify “onerous” regulations

57. Id. at 112. 58. Id. 59. The authors employ a complex methodology for industry coding that is

described in Online Appendix B. Id. at app. B (describing their methodology for industry coding). I do not discuss this methodology here because it is outside the scope of the critique developed in this Article. Allocating regulations by in­ dustry does not mitigate any of the methodological problems identified below.

60. RegData, supra note 4, at 110. 61. Id. at 112. 62. Id. 63. Rod Boshart, Iowa Confident Alert System Won’t Send False Alarm, GA­

ZETTE (Jan. 16, 2018), http://www.thegazette.com/subject/news/government/ iowa-confident-alert-system-wont-send-false-alarm-20180116.

2018] THE TROUBLE WITH COUNTING 109

for repeal64 and by officials im plem enting a 2-for-l policy in B rit­ ish Columbia, C anada,65 suggesting th a t RegData is poised to gain more widespread use by governments im plem enting regu­ lation counting policies and thus m ust be taken seriously not only as an intellectual project, but also a policy tool. For these reasons, RegData and its counting methodology will be the focus of this Article’s empirical critique.

II. THE TROUBLE WITH COUNTING This P a rt provides a basic introduction to the fundam ental

elem ents necessary to draw valid causal inferences from em pir­ ical studies and explains why regulation counting studies lack these elements, illustrating key points with concrete examples draw n from the C.F.R. Empirical studies use quantified v aria­ bles to m easure observable phenom ena in the world (for in ­ stance, the costs of regulation or employment levels) and to es­ tablish th eir statistical relationship w ith one another. If certain conditions are met, quantitative studies can indicate causal re ­ lationships between variables, establishing, for instance, th a t one observable phenomena (e.g. regulatory costs) is the cause of another (e.g. employment levels).

The validity of a causal statistical inference depends on many factors, but the most basic and fundam ental is m easure­ m ent validity, or the fit between a variable and the conceptual definition of the construct th a t the variable purports to measure. S tated simply, to produce valid statistical inferences, an explan­ atory variable m ust have a well-defined construct th a t is validly m easured. A construct66 is a hypothetical condition or m echa­ nism explaining or predicting some outcome of interest. For in-

64. Id. 65. Laura J o n es, Mercatus Ctr., Cutting Red Tape in Canada: A Reg­

ulatory Rform Model for the United States? (2015), https://www .mercatus.org/system/files/Jones-Reg-Reform-British-Columbia.pdf. In addi­ tion to the province of British Columbia, various versions of the 2-for-l policy have been adopted in other countries, including: Australia, Canada (federal), France, Germany, the Netherlands, Portugal, and the United Kingdom. Nicho­ las Bellos, Is There an International Case for Trump’s “One-in-Two-Out”Order?, REG. Rev. (Nov. 8, 2017), https://www.theregreview.org/2017/ll/08/bellos -international-trump-order.

66. Different academic disciplines tend to use different statistical terminol­ ogy. I use the term construct as it tends to be used in sociology and psychology literature, to refer to an explanatory concept hypothesized to have a statistical relationship with an outcome variable. Economists tend to use the term con­ struct to refer to the observable measure of the underlying concept.

110 MINNESOTA LAW REVIEW [103:93

stance, intelligence is a construct that researchers might hypoth­ esize explains outcomes like academic achievement, income, or wealth. Constructs (like intelligence) cannot be directly observed or measured, and so they must be operationalized as variables for statistical analysis by measuring some observable and quan­ tifiable behavior or behavioral artifact. For instance, a re­ searcher who wishes to test the effects of the construct of intelli­ gence on particular outcomes might use IQ test scores as an observable proxy by which to measure it.

The validity of a measure depends on the extent to which it accurately quantifies the construct it purports to assess. For in­ stance, it would be clearly invalid to use height as a measure of intelligence, because there is no plausible relationship between the construct (intelligence) and the measure (height). IQ scores present a more difficult case: while many dispute whether IQ scores measure intelligence,67 they are widely used in studies as a proxy for intelligence or similar constructs.68 Suppose the re­ searcher described above finds in her statistical analysis that higher IQ scores (the measure) are causally related to higher in­ comes (the outcome). May she make the claim that intelligence (the construct) causes higher incomes (the outcome)? This claim will be valid only if IQ scores (the measure) accurately measure intelligence (the construct). The onus is on the researcher to demonstrate the relationship between construct and measure that supports the causal claim she wishes to make.

The fundamental problem with regulation counting projects is that they lack this rudimentary precondition of fit between construct and measure. This problem exists for two subsidiary reasons. First, as discussed in Section A below, the construct op­ erationalized through the methodology of regulation counting is ill-defined, making both measurement and causal inference dif­ ficult, if not impossible. Second, as discussed in Section B.l—8

67. See, e.g., Stephen J. Ceci & Jeffrey K. Liker, A Day at the Races: A Study of IQ, Expertise, and Cognitive Complexity, 3 J. EXPERIMENTAL PSYCHOL. 255, 255 (1986) (arguing that “IQ is unrelated to real-world forms of cognitive com­ plexity that would appear to conform to some of those that scientists regard as the hallmarks of intelligent behavior”).

68. See, e.g., LEWIS M. TERMAN & M a u d A. MERRILL, MEASURING INTELLI­ GENCE: A G u id e t o t h e A d m in is t r a t io n o f t h e N e w R e v is e d S t a n f o r d -Bi - NET T e s t s OF INTELLIGENCE (1949) (claiming that IQ tests measure intelli­ gence); William W. Brown & Morgan 0. Reynolds, A Model of IQ, Occupation, and Earnings, 65 AM. ECON. REV. 1002 (1975) (using IQ test scores as a proxy for intelligence); Jay L. Zagorsky, Do You Have to Be Smart to Be Rich? The Impact of IQ on Wealth, Income and Financial Distress, 35 INTELLIGENCE 489 (2007) (using IQ test scores as a proxy for intelligence).

2018] THE TROUBLE WITH COUNTING 111

below, the constructs th a t have been articulated by regulation counters cannot be m easured accurately by counting the num ber of regulations or regulatory requirem ents in the C.F.R.

A. C o n s t r u c t D e f i n i t i o n

A threshold problem with regulation counts is th a t it is not entirely clear w hat construct they are supposed to be measuring, or if they are m eant to m easure any construct a t all. A count is nothing more th a n a num ber—like the num ber of telephone rings69—unless it accurately m easures some validly theorized explanatory construct. Regulation counts are often presum ed to m easure constructs like the costs or burdens of regulation. How­ ever, regulation counters have been relu ctan t to clearly define and theorize a stable, coherent construct th a t is accurately m eas­ ured by regulation counts. This makes it difficult to know exactly w hat regulation counts m easure and underm ines causal claims based on th em . 70

Some regulation counters do not bother to supply an explan­ atory construct, or seem to believe th a t the count of regulations is, itself, an explanatory construct th a t is causally related to o u t­ come variables of in te re st. 71 This is like claiming th a t the num ­ ber of pages in the Yellow Pages causes traffic accidents. Now, it

69. Note that the number of telephone rings could be transformed into a construct if it had some significance independent of the absolute number—for instance, if the number of rings encoded some kind of message from caller to receiver. There is no indication in the Sesame Street sketch referenced above that the Count’s caller is sending him an encoded message.

70. See Bernard S. Black et al., Corporate Governance Indices and Con­ struct Validity, 25 CORP. GOVERNANCE: INT’L REV. 397, 398 (2017) (even if in­ dependent variables are properly measured, causal claims can fail “because the underlying theory that posits a relationship between the general aspect (board structure) and the outcome is wrong”).

71. See, e.g., HOWARD BEALES ET AL., REGULATORY TRANSPARENCY PRO­ JECT, Government Regulation: The Good, the Bad, & the Ugly (2017), https://regproject.org/paper/government-regulation-the-good-the-bad-the-ugly (arguing, among other things, that “regulatory accumulation” is detrimental to the American economy); TEN THOUSAND COMMANDMENTS, supra note 5 (focus­ ing throughout the report on pages in the C.F.R. devoted to final rules to deter­ mine regulatory burdens); JONES, supra note 65, at 22-24 (arguing that British Columbia’s reduction in the number of regulations is the reason for the tu rn ­ around in its economy); Ronald Bailey, Federal Regulations Have Made You 75 Percent Poorer, REASON (June 21, 2013), http://reason.com/archives/2013/06/21/ federal-regulations-have-made-you-75-per (arguing that “six decades of accu­ mulated regulations” are responsible for suppressing the growth of GDP and household income); Michael Mandel, Pebbles in the Stream: Does the FDA Slow Medical Technology Innovation'?, MANDEL ON INNOVATION & GROWTH (Dec. 4, 2010), https://innovationandgrowth.wordpress.com/2010/12/04/pebbles-in-the

112 M IN N ESO T A L A W R E V IE W [103:93

may be the case th a t the num ber of pages in the Yellow Pages is a reasonable proxy for the num ber of people living in a particular locality, which may be a reasonable proxy for traffic volume, which might be causally related to the num ber of traffic acci­ dents. But it cannot be the case th a t the num ber of pages in the phone book causes traffic accidents. In this example, the girth of the Yellow Pages is merely the m easure of the explanatory con­ struct, traffic volume. To draw meaningful inferences from a s ta ­ tistical relationship between the Yellow Pages and traffic acci­ dents, there m ust be some valid theory positing a relationship between the two. 72 Similarly, the num ber of regulations, stan d ­ ing alone, has no causal significance unless it can be shown to be a proxy for some other causal mechanism driving economic out­ comes.

More sophisticated regulation counters have recognized the need for an explanatory construct to hypothesize causal relatio n ­ ships. However, the constructs they have supplied are poorly de­ fined, under-theorized, and in tension with one another. It has been claimed th a t regulation counts operationalize constructs

-stream-does-the-fda-slow-medical-technology-innovation (comparing regula­ tions to pebbles in a stream, which dam the stream when there are too many); Jared Meyer, Regulation’s Stranglehold on M illen n ia l’ Futures, FORBES (May 25, 2015), https://www.forbes.com/sites/jaredmeyer/2015/05/25/regulations -stranglehold-on-millennials-futures/#7cd2a62e2f39 (arguing that the “sheer quantity of government red tape” hampers economic growth); Will Yakowicz, Why Regulation Is Ruining the U.S. Economy, INC. (Nov. 19, 2013), https://www .inc.com/will-yakowicz/why-regulation-is-ruining-us-economy.html (suggesting the number of regulations, and removing unnecessary ones, will jumpstart the economy).

72. See Black et al., supra note 70, at 398 (asserting that even if independ­ ent variables are properly measured, causal claims can fail “because the under­ lying theory that posits a relationship between the general aspect (board struc­ ture) and the outcome is wrong”).

2018] THE TROUBLE WITH COUNTING 113

like regulatory costs, 73 regulatory burdens, 74 or regulatory con­ s tra in ts , 75 or th a t they “probably capture [] a t least some of reg­ ulation’s complexity.” 76 Rarely is it clear w hat the focal construct is, because these constructs have been used interchangeably.

This construct shell game severely undercuts the empirical validity of regulation counting studies because the different con­ structs do not necessarily mean the same thing. A burden, for instance, may not come in the form of economic costs. As dis­ cussed below, the costs of regulation may be to liberty or to the psychic well-being of regulated business owners. Costs th a t have big payoffs, like the expense of applying for an offshore drilling lease, are not properly characterized as burdens. C onstraints— like requirem ents th a t everyone drive on the same side of the road—may not be particularly costly or burdensome or complex. Indeed, as this example illustrates, constraints often increase so­ cial welfare by solving collective action problems in ways th a t benefit everyone, including regulated businesses. 77 A world

73. See Exec. Order No. 13,771, 82 Fed. Reg. 9339 (Jan. 30, 2017); OMB Guidance, supra note 2; Brookings Evaluation, supra note 29, at 6 (calling the 2-for-l requirement “a blunt institutional reform to rein in regulatory costs”); Dudley, supra note 37, at 262 (suggesting an association between regu­ lation counts and regulatory costs).

74. Dawson & Seater, supra note 5, at 143 (explaining their construct as follows: “It seems reasonable to suppose that the number of pages required to describe regulatory requirements varies directly with the number of require­ ments, at least on average. Our page count measure therefore should capture whatever regulatory burden is reflected in the number of regulatory require­ ments”); Andrew Hale, David Borys & Mark Adams, Regulatory Overload: A Behavioral Analysis of Regulatory Compliance 2-5 (Mercatus Ctr., Working Pa­ per No. 11-47, 2011) (citing the volume of regulations taken with their complex­ ity and style to argue they create a regulatory overload that must be managed and adjusted); see also Tyler Cowen, More Freedom on the Airplane, if Nowhere Else, N.Y. TIMES (Nov. 17, 2013), https://www.nytimes.com/2013/ll/17/ business/more-freedom-on-the-airplane-if-nowhere-else.html (arguing that when there are more regulations, compliance for regulated entities becomes dif­ ficult and burdensome).

75. RegData, supra note 4, at 112; see also Mandel, supra note 71 (compar­ ing regulations to pebbles in a stream, which, like a dam, constrain the flow when there are too many).

76. Dawson & Seater, supra note 5, at 139-40. Complexity is rarely cited as the focal construct explaining economic outcomes. Rather, complexity tends to be subsidiary to constructs like cost, which are ultimately said to drive eco­ nomic outcomes.

77. See CASS R. SUNSTEIN, AFTER THE RIGHTS REVOLUTION: RECONCEIV- ING THE REGULATORY S t a t e (1990) (a r g u in g t h a t re g u la tio n o fte n in c r e a s e s so­ c ial w e lfa re b y so lv in g collective a c tio n o r c o o rd in a tio n p ro b le m s a n d b y m in i­ m iz in g d ecisio n costs).

114 MINNESOTA LAW REVIEW [103:93

without traffic regulations would be a very costly place to con­ duct business. By contrast, clear-cut regulatory commands may not be especially complex, but might still be quite costly. For in ­ stance, the regulation requiring autom akers to achieve a Corpo­ ra te Average Fuel Economy ratin g of 54.4 miles per gallon by the year 2025 is reasonably straightforw ard as regulatory com­ m ands go, but many have claimed th a t it would be very costly for autom akers and consumers.78

In sum, each of the proffered constructs presents a different set of potential causal relationships with business activity and economic outcomes. Yet regulation counting studies have stre n ­ uously avoided tying regulation counts concretely to a single con­ struct and theorizing the relationship between th a t particular construct and economic outcomes of interest. This may be be­ cause, as some regulation counters have acknowledged, eco­ nomic theory would equally support a positive or a negative a s­ sociation between regulation and economic activity.79 Or it may be because m utable constructs enhance the utility of regulation counts as a political tool, allowing regulation counters to speak to different audiences and deflect criticism by subtly shifting th eir rhetoric in response to challenges. If someone points out th a t regulations often save money, the political regulation coun­ te r can respond th a t the problem is really about stifling innova­ tion or increasing complexity. If someone points out th a t regula­ tions often spur innovation, the political regulation counter can respond th a t the real problem is cost. W hatever the reason, the lack of a well-defined, theoretically justified construct m easured by regulation counts m akes it implausible to draw causal infer­ ences from any statistical associations found between these counts and economic outcome variables.

There is one dimension of regulation counting constructs th a t tends to be reasonably well defined, but it only fu rth er com­ plicates the credible theorization of causal relationships between

78. Press Release, Office of the Press Sec’y, The White House, Obama Ad­ ministration Finalizes Historic 54.5 MPG Fuel Efficiency Standards (Aug. 28, 2012), https://obamawhitehouse.archives.gov/the-press-office/2012/08/28/ obama-administration-finalizes-historic-545-mpg-fuel-efficiency-standard; see also Juliet Eilperin & Steve Overly, Automakers Ask EPA To Overturn Recent Review of Fuel-Efficiency Standards, WASH. PO ST (Feb. 22, 2017), https://www .washingtonpost.com/national/health-science/automakers-ask-epa-to-overturn -recent-review-of-fuel-efficiency-standards/2017/02/22/81adl398-f920-lle6 -9845-576c69081518_story.html.

79. See, e.g., Dawson & Seater, supra note 5, at 145.

2018] THE TROUBLE WITH COUNTING 115

the regulation count and economic outcome variables. Regula­ tion counters tend to agree th a t regulation counts are only m eant to capture the costs or burdens or constraints impacting private entities subject to regulation ra th e r th a n the aggregate net costs (or benefits or burdens or constraints) to society.80 This concep­ tualization flies in the face of fundam ental tenets of welfare eco­ nomics and m arginalism .81 The tally of gross costs on one subset of m arket actors reveals little about prices, behavior, or effi­ ciency in the broader m arket. To date, regulation counters have not explicitly justified the theoretical significance of tallies m easuring gross costs/burdens/constraints on one set of eco­ nomic actors, unm itigated by benefits accruing to those same ac­ tors and divorced from th eir relationship to costs (and benefits) accruing to other m arket actors. Thus, it is not a t all clear w hat construct regulation counters purport to m easure when they m easure costs, burdens, or constraints only on regulated e n ti­ ties, unless th a t construct is a select subset of the costs, burdens, or constraints borne by a favored set of political allies.

Thus, in addition to the more technical issues of m easure­ m ent validity elaborated below, the opacity, indeterm inacy, and outright bias in the overarching constructs purportedly m eas­ ured by regulation counts underm ine the ability of regulation counters to support causal empirical claims about the relation­ ship between regulation counts and economic outcomes.

B. Measure Validity Even if a plausible theory to justify the constructs of cost,

burden, or constraint on regulated entities could be developed, the problem rem ains th a t regulation counts do not validly m eas­ ure said constructs. As I argue below, regulation counting does not and cannot m easure the costs, burdens, or constraints on regulated entities because it does not account for a t least nine im portant features of regulatory law: (1) variation in the weight of regulations; (2) variation in regulations’ scope of coverage; (3) the object of regulatory requirem ents; (4) stru ctu ral relation­ ships between and among regulations; (5) basic gram m ar and punctuation; (6) the fact th a t many regulatory requirem ents re ­ late to the dispensation of government largess to regulated enti­ ties; (7) the fact th a t regulated entities enjoy other monetizable benefits from robust regulation; (8) variations in the rigor with

80. See, e.g., RegData, supra note 4, a t 112 (counting keywords in the C.F.R. th a t are likely to “lim it choice sets” of economic actors).

81. Brookings Evaluation, supra note 29, a t 5.

116 MINNESOTA LAW REVIEW [103:93

which regulations are enforced, up to and including total non- enforcement; and (9) the fact th a t Congress, and not the adm in­ istering agency, is the direct source of many regulatory require­ ments.

To be fair, regulation counters have acknowledged the lim i­ tations of the m easure to a point, but they tend to dismiss them casually. As one study explains, “[a] counting m easure obviously is imperfect in th a t two identical values may comprise regula­ tions of different types and, even w ithin a given type, may rep ­ resen t regulations of different stringency.”82 But the authors dis­ miss this concern by noting th a t “if there are many kinds of regulation . . . , it is reasonable to expect an index to provide a useful overall m easure of regulation.”83 Of course, the validity of this claim depends on the contents of the index, and these a u ­ thors display no understanding of the rudim ents of regulatory law th a t comprise th eir index. Similarly, the authors of RegData allow th a t “ju st as one page [of the C.F.R.] may not be equal to another page [of the C.F.R.], one restriction may carry more con­ sequence th a n another.”84 Nonetheless, they suggest th a t this defect is m itigated by the fact th a t RegData m easures re ­ strictions of different weights at “different levels of granularity” (meaning by title, chapter, part, and parag rap h of the C.F.R.), without explaining how m easurem ent error nets out if m easured a t different levels.85

Some regulation counters have dismissed criticism of th eir m easure by arguing, circularly, th a t regulation counts m ust be a valid m easure because they are correlated w ith macroeconomic outcome variables like growth and productivity, and there would be no correlation if the count variable were ju st noise.86 This ju s­ tification ignores the fact th a t it is trivially easy to generate spu­ rious correlations using regression analysis:87 for instance, the finding th a t between 2000 and 2009, per capita cheese consump­ tion in the U nited States was highly correlated w ith the num ber

82. Dawson & Seater, supra note 5, at 139. 83. Id. 84. RegData, supra note 4, at 112. 85. Id. 86. Dawson & Seater, supra note 5, at 139. 87. See Atanasov & Black, supra note 15 (investigating shock-based causal

inferences); Black et al., supra note 70 (demonstrating how easy it is to generate correlations through regression analysis).

2018] THE TROUBLE WITH COUNTING 117

of people who died by becoming tangled in th eir b edsheets.88 In ­ deed, regulation counters have called into question the facial va­ lidity of th eir own m easure by dem onstrating in a recent study a spurious correlation between the num ber of pages in the Federal Register and the winning percentage of the W ashington Red­ sk in s.89

Despite recognizing the lim itations of regulation counts as a m easure, counters tend to conclude th a t these counts are a good- enough composite of something th a t is very difficult to measure. One of RegData’s creators likens regulation counting to saber- metrics, an approach to baseball team m anagem ent popularized by the movie Moneyball and my hometown team , the Oakland A’s . 90 “When working w ith a complex system, w hether it’s the economy, or even a baseball team , it is im portant to m easure its inputs and components if you w ant to advance the performance of th a t system. RegData is a new database th a t does ju st t h a t .” 91

This Section challenges regulation counters’ blithe dism is­ sals of m easurem ent criticism by dem onstrating three key points. F irst, there is extreme incom m ensurability between dif­ ferent item s th a t count the same in regulation counts. In saber- metrics term s, this would be like counting all hits the same, without respect to w hether they are singles, doubles, triples or home ru n s or w hether they occurred with runners in scoring po­ sition or during other key moments in the game. Sure, it’s nice to know how many hits a player has in a season, but the point of saberm etrics is to predict the value th a t a player’s hits are likely to add to the team over the course of a season and to help m an­ agers make decisions about which players to use in which situ a ­ tions in order to maximize th a t value. Saberm etrics attem pts to achieve this goal by appropriately weighting and discounting dif­ ferent types of hits in different contexts.9̂ Like Sabermetrics, RegData is being m arketed as a decision-making tool th a t can

88. Tyler Vigen, Spurious Correlations, TYLERVIGEN.COM, http://www .tylervigen.com/spurious-correlations (last visited Oct. 15, 2018).

89. See generally Regulators and Redskins, supra note 5. 90. Patrick McLaughlin, The Science of Government Regulation, U.S. NEWS

(Oct. 31, 2012), https://www.usnews.com/opinion/blogs/economic-intelligence/ 2012/10/31/the-science-of-government-regulation.

91. Id. 92. See Zachary D. Rymer, Sabermetrics for Dummies: How-to Guide for

MLB Fans to Learn the Ropes, BLEACHER Re p. (Apr. 25, 2014), https:// bleacherreport.com/articles/2040748-sabermetrics-for-dummies-how-to-guide -for-mlb-fans-to-learn-the-ropes (explaining sabermetrics for evaluating hit­ ting, pitching, and player contributions to the team).

118 MINNESOTA LAW REVIEW [103:93

help government officials make complex calls about how to reg­ ulate;93 but unlike Sabermetrics, RegData does not tell them w hether any given regulation is a single, a double, or a home run, or w hether it is likely to score a run, which is ultim ately w hat m atters at the end of the day.

Second, my analysis reveals ram p an t double (and triple, and more) counting of regulatory requirem ents. This m eans th a t even the absolute num ber counted is wrong. Third, my analysis dem onstrates th a t the regulations counted are not only of differ­ ent weight, but also of different directional effect. As demon­ strated below, many regulations or regulatory commands lessen the costs, burdens, or constraints on regulated entities, and yet regulation counters add them to the tally of costs, burdens, or constraints on regulated entities ra th e r th a n subtracting them from the count. This is more th an mere noise—it affirmatively m isrepresents the n atu re and m agnitude of w hat is purportedly being measured. This type of m easurem ent error does not net out as more d ata gets collected, like outlier games in a pitcher s earned ru n average. R ather, it compounds itself. It is like adding every game a pitcher starts to the “win” column of her record w ithout paying attention to the direction the game went. Regu­ lation counters have never grappled seriously with the biases these defects introduce into th eir m easures and any statistical analyses conducted using them , and the empirical claims arising from those analyses, cannot be taken seriously unless they do.

1. Counting Does Not Account for Weight Regulation counts do not account for the weight of the things

counted. This is a fundam ental flaw if regulation counts are m eant to m easure regulatory costs, burdens, or constraints on regulated entities. The burden of carrying 1,000 feathers is very different from the burden of carrying 1,000 anvils, yet regulation counts make no attem pt to differentiate between regulatory feathers and anvils. So, for example, in RegData the command th a t mine operators “shall’’ supply th eir official address and te l­ ephone num ber on documents subm itted to the Mine Safety and H ealth A dm inistration94 is counted ju st the same as the com­ m and th a t autom akers “must comply w ith”95 the complex and

93. Cf. Patrick McLaughlin et al., RegData 3.0 User’s Guide, QUANTGOV, https://quantgov.org/regdata/users-guide (last visited Oct. 15, 2018) (discussing the potential uses for RegData and the sorts of problems it can help solve).

94. 30 C.F.R. § 41.30 (2017) (emphasis added). 95. 40 C.F.R. § 86.1818-12(a)(1) (2017) (emphasis added).

2018] THE TROUBLE WITH COUNTING 119

dem anding greenhouse gas fleet average requirem ents detailed in EPA regulations. Adding these two wildly incom m ensurate requirem ents together does not provide an accurate m easure of the costs, burdens, or constraints of regulation on regulated e n ­ tities.

2. Counting Does Not Account for the Scope of Coverage Regulation counts do not account for the breadth or narrow ­

ness in applicability of particular regulatory requirem ents. For instance, S ubpart 3430 of the C.F.R. contains several regulatory requirem ents applicable to Preference Right Leases granted by the Bureau of Land M anagem ent for coal prospecting on federal lands.96 However, these requirem ents apply only to leases issued prior to August 4, 1976.97 Such leases represent roughly a third of federal coal prospecting leases.98 Yet, m andatory term s iden­ tified in S ubpart 3430 m easure the same quantity of cost, b u r­ den, or constraint as those identified in P a rt 3470, which applies to all federal coal prospecting leases.99

Some regulations apply to even sm aller populations. For in ­ stance, D epartm ent of Justice (DOJ) regulations provide m an ­ datory criteria th a t m ust be met to receive a financial rew ard for disclosing information relating to the unlawful introduction, m anufacture, acquisition, export, loss, or diversion of atomic weapons and special nuclear m aterials.100 These regulations con­ tain directive language demanding th a t the information “m ust be original, and m ust concern”101 unlawful conduct. However, these m andates apply only to the thimble-full of individuals who possess information about unlawful activities relating to atomic weapons and who wish to obtain a financial rew ard for disclosing it to the DOJ.102 Nonetheless, RegData counts such narrowly ap-

96. See 43 C.F.R. pt. 3430, subpart 3430 (2017). 97. Id. § 3430.0.7. 98. Frequently Asked Questions About the Federal Coal Leasing Program,

U.S. De p’t Interior: Bureau Land Mgmt., https://eplanning.blm.gov/epl -front-office/projects/nepa/64842/78268/88489/CoalFAQ.pdf (last visited Oct 15, 2018).

99. See 43 C.F.R. § 3430.6-1 (“Each preference right lease shall be subject to the term s provided for Federal coal leases established in p art 3470 of this title.”).

100. 28 C.F.R. § 13.6(a) (2017). 101. Id. (emphasis added). 102. Id. §§ 13.1-13.2.

120 M IN N ESO T A LA W RE V IE W [103:93

plicable m andates as com m ensurate with, for instance, Occupa­ tional Safety and H ealth Adm inistration requirem ents th a t a p ­ ply to millions of U.S. workplaces.103

3. Counting Does Not Account for the Object of Regulatory Requirem ents

Baked into the construct definition of most counting projects is the assum ption th a t private regulated entities are the objects of regulatory m andates—th a t they are the actors bearing the costs, burdens, or constraints of regulation. However, large sw aths of the C.F.R. do not even apply to regulated entities, but ra th e r constrain the government’s actions, typically for the b en­ efit and protection of regulated entities.

Title 1 regulates government actions relating to governmen­ ta l functions like enacting and publishing laws. For example, it provides th a t certain specified documents “are required to be filed for public inspection with the Office of the Federal Register and published in the FEDERAL REGISTER”104 and m andates th a t each document so published “shall be keyed to the Code of Fed­ eral Regulations.”105 These regulations impose no burden w h at­ soever on regulated entities and, instead, require government transparency and regularity for the benefit of all citizens, includ­ ing regulated entities and th e ir law yers.106

The federal government is likewise the prim ary object of the regulatory constraints found in Title 2, adm inistering govern­ m ent grants and agreem ents. For instance, Office of M anage­ m ent and Budget regulations require th a t granting agencies “m ust ensure” th e adequacy of th eir information processing sys­ tem s related to aw ards107 and “m ust include” a variety of infor­ m ation about the aw ard as well as instructions for applicants in all program announcem ents.108 These regulations also contain detailed guidelines about the policies and procedures th a t agen­ cies “[m]ust establish”109 to govern debarm ent and suspension,

103. Cf. supra notes 56-59 and accompanying text (describing the counting methodology of RegData).

104. 1 C.F.R. § 5.2 (2017) (emphasis added). 105. Id. § 5.5 (emphasis added). 106. Cf. id. §§ 5.1-5.3 (laying out the policy and requirem ents behind the

publication of documents in the FEDERAL REGISTER). 107. 2 C.F.R. § 25.215 (2017) (emphasis added). 108. Id. § 170.200(a) (emphasis added). 109. Id. § 180.25(a) (emphasis added).

2018] THE TROUBLE WITH COUNTING 121

including prohibiting agencies from making aw ards to sus­ pended or debarred persons.110 S ubpart B of Title 2 codifies reg­ ulations prom ulgated by thirty-tw o different agencies separately adopting the OMB guidance regulating grants and agreem ents, using sim ilar m andatory language multiplied many times over.* 111

Title 3, C hapter I, provides standards of conduct for employ­ ees in the Executive Office of the President.112 Title 4, C hapter I, applies to the Government Accountability Office and contains thirty-six regulations using the word “m ust” and 148 regulations using the word “shall,” including requirem ents about the hours during which the GAO building “shall” be open to the public.113 Title 5 contains extensive regulation of adm inistrative person­ nel, including 1,420 regulations using “shall” and 1,781 using “m ust.”114 Title 11 regulates federal elections.115 Regulations in Title 28, C hapter I, pertain exclusively to the D epartm ent of J u s ­ tice and govern issues like parole, release, supervision and re ­ commitment of prisoners,116 im plem entation of the Equal Access to Justice Act,117 and death sentence procedures.118 These regu­ lations have no applicability whatsoever to regulated businesses, except perhaps to protect those th a t find them selves caught up in the federal crim inal justice system. Title 39 applies exclu­ sively to the Postal Service, and contains 610 regulations with “shall” and 227 with “m ust.”119 The regulations in 41 C.F.R. Sub-

110. Id. § 180.400. 111. See Id. §§ 200.100-.113 112. E.g., 3 C.F.R. § 101.1 (2017) (stating that “[ujntil further regulations

are promulgated, the remainder of the entities within the Executive Office of the President, to the extent that 5 U.S.C. § 552 is applicable, shall follow the procedures set forth in the regulations applicable to the Office of Management and Budget (5 CFR Ch. Ill)” (emphasis added)).

113. 4 C.F.R. § 25.3 (2017) (“During normal working hours, the GAO Build­ ing shall be open to the public unless specific circumstances require it to be closed to the public to ensure the orderly conduct of government business.” (em­ phasis added)).

114. E.g., 5 C.F.R. § 2638.306 (2017) (“The agency must provide each em­ ployee upon initial appointment to a supervisory position with the written in­ formation required under this section.” (emphasis added)).

115. 11 C.F.R. ch. I (2017). 116. 28 C.F.R. pt. 2. 117. Id. pt. 24. 118. Id. pt. 26. 119. E.g., 39 C.F.R. § 447.21(b) (2017) (“No employee shall take sick leave to

enable himself to engage in outside work.”).

122 MINNESOTA LAW REVIEW [103:93

title C relate to the government’s m anagem ent of its own prop­ erty.120 For instance, Section 102 of this subtitle contains m an­ dates relating to the fuel economy standards of government ve­ hicle fleets.121

In addition to government-focused regulations occupying large and discrete p arts or titles of the C.F.R., agency-constrain­ ing regulations are woven throughout the agency-specific titles of the C.F.R. Extensive, agency-specific regulations bind agen­ cies to observe certain procedures in order to comply w ith the A dm inistrative Procedure Act,122 the Freedom of Information Act,123 the Government in the Sunshine Act,124 and due process requirem ents, th u s effectuating im portant statutory and consti­ tutional protections for regulated entities. O ther regulatory m andates directed towards the government constrain agency discretion by specifying the factors an agency m ust consider in deciding certain issues. For instance, regulations im plem enting the Endangered Species Act contain requirem ents governing how the Secretary of the Interior “shall review” applications for exemptions from the s ta tu te ’s take provisions.125

In addition, many federal regulatory m andates th a t arise in the context of cooperative federalism program s and block grant program s apply to state governments, not to private entities. For instance, 40 C.F.R. § 256.20 contains requirem ents applicable to state solid waste disposal program s.126 Medicare and Medicaid regulations likewise impose extensive requirem ents on state governments th a t have elected to receive federal funding under these program s.127

The failure to account for the fact th a t regulated entities are the object of only a fraction of the requirem ents found in the C.F.R. significantly distorts the accuracy of regulation counts as a m easure of the cost, burden, or constraint of regulation on reg­ ulated en tities.128

120. 41 C.F.R. subtitle C (2017). 121. 41 C.F.R. § 102-34 (2017). 122. 5 U.S.C. §§ 500-96 (2017). For instance, in certain matters appealed to

the Secretary of Commerce, the Secretary is required by regulation to provide public notice of the appeal, take comments, and evaluate the comments in spec­ ified ways. 15 C.F.R. § 930.128 (a)-(c) (2017).

123. 5 U.S.C. § 552. 124. Id. § 552b. 125. 50 C.F.R. § 451.02 (2017) (emphasis added). 126. 40 C.F.R. § 256.20 (2017). 127. 42 C.F.R. § 403.304 (2017). 128. While regulation counters may see regulations constraining federal and

2018] THE TROUBLE WITH COUNTING 123

4. Counting Does Not Account for S tructural Relationships Among Regulations

Counting the absolute num ber of regulations or regulatory commands does not account for interrelationship between and among regulations and th eir subparts. Indeed, in many cases, the existence of multiple regulations or regulatory commands can result in fewer burdens and more flexibility for regulated entities. I identify four such cases below: exceptions, a lte rn a ­ tives, reference to other regulations, and elaboration or clarifica­ tion of regulatory requirem ents.

a. Exceptions Many regulations containing regulatory commands are

qualified by explicitly articulated exceptions. Some exceptions are so num erous or so broad th a t they swallow the rule. Yet reg­ ulation counters count these regulations or commands the same way th a t they count unqualified regulations or commands.129

For instance, im porters of agricultural products like dates can unilaterally exempt th eir products from USDA grade, size, quality, and m aturity requirem ents in a variety of ways, includ­ ing: by donating nonconforming products to “needy persons, p ris­ oners, or Native Americans”130 or by designating them for pro­ cessing.131 Taking one of these unilateral actions alleviates date im porters from the multiple m andatory requirem ents of 7 C.F.R. § 999.1.132 Sim ilar exemptions are available for im porters of wal­ n u ts ,133 prunes,13* ra isin s,133 and filberts.130 The Nuclear Regu­ latory Commission exempts from classification as “fissile m ate­ rial and compliance w ith associated regulatory requirem ents th a t fissile m aterial which meets any one of six criteria.137 The Federal Deposit Insurance Corporation exempts ten different

state government agencies as costly and burdensome as well, this relationship is nowhere theorized. Such a claim would need to be explicitly articulated and supported. In the absence of such support, it is impermissible to claim that the number of regulations on government entities accurately measures the costs, burdens, or constraints on regulated entities.

129. See, e.g., supra notes 56—59 and accompanying text. 130. 7 C.F.R. § 999.1(d)(2) (2017). 131. Id. 132. M § 999.1. 133. Id. § 999.100(d)(2). 134. Id. § 999.200(d). 135. Id. § 999.300(e). 136. Id. § 999.400(d). 137. 10 C.F.R. § 71.15 (2017).

124 MINNESOTA LAW REVIEW [103:93

types of advertisements from its requirements governing adver­ tising by federally insured depository institutions. 138 The De­ partment of Health and Human Services exempts several cate­ gories of health care providers from immunization reporting requirements. 139

Federal Trade Commission regulations allow “[a]ny person who believes a particular hazardous substance intended or pack­ aged in a form suitable for use in the household or by children [to] be exempted from full label compliance otherwise applicable under the act” 140 upon a showing that “full compliance is imprac­ ticable or is not necessary for the protection of the public health.” 141 While the Department of Homeland Security and the Coast Guard generally require inspected vessels to maintain life­ saving systems with mandated features, vessels are exempt from these requirements if they can demonstrate that they would be “unreasonable or unnecessary” 142 due to the nature of the vessel and its voyage routes or that they would seriously impede re­ search” 143 into novel vessel designs. Many research and develop­ ment activities are similarly exempted from the disclosure re­ quirements of the Toxic Substances Control Act. 144

Even where regulatory requirements are not explicitly qual­ ified by exceptions in the text of the C.F.R., exceptions may be granted by administrative agencies under statutory authority and principles of “administrative equity.” 145 For example, the Department of Energy Act authorizes the Secretary of Energy to make “adjustments” to duly enacted regulations, including ex­ ceptions, exemptions, modifications, and interpretations. 146 Other statutes and regulations empower agencies to provide

138. 12 C.F.R. § 328.3(d) (2017). 139. 42 C.F.R. § 495.22 (2017). 140. 16 C.F.R. § 1500.82(a) (2017). 141. Id. 142. 46 C.F.R. § 199.20 (2017). 143. Id. 144. E.g., 40 C.F.R. § 725.200 (2017) (exempting certain research and devel­

opment activities from reporting requirements); cf. id. § 745.101 (exempting cer­ tain housing transactions from disclosure of potential lead-based paint expo­ sure).

145. Alfred C. Aman, Jr., Administrative Equity: An Analysis of Exceptions to Administrative Rules, 1982 DUKEL.J. 277, 278 (1982).

146. 42 U.S.C. § 7194(a) (2017).

2018] THE TROUBLE WITH COUNTING 125

w aivers,14' no-action lette rs,148 or variances149 th a t carve out ex­ ceptions to regulatory requirem ents on a case-by-case basis.

Regulation counts take no account of this. While the exist­ ence of exceptions may, admittedly, add complexity to regulatory regimes, this type of complexity ultim ately alleviates costs/bur- dens/constraints on regulated entities, and th u s cannot merely be added to a tally th a t purports to m easure any of these con­ structs.

b. Alternatives Some regulations contain multiple alternative means of

complying with prim ary statutory or regulatory requirem ents. For instance, in the quoted regulation FERC gives covered facil­ ities options as to how they will deliver and price energy:

Each qualifying facility shall have the option either: (1) To provide en­ ergy as the qualifying facility determines such energy to be available for such purchases, in which case the rates for such purchases shall be based on the purchasing utility’s avoided costs calculated at the time of delivery; or (2) To provide energy or capacity pursuant to a legally enforceable obligation for the delivery of energy or capacity over a spec­ ified term, in which case the rates for such purchases shall, at the op­ tion of the qualifying facility exercised prior to the beginning of the specified term, be based on either: (i) The avoided costs calculated at the time of delivery; or (ii) The avoided costs calculated at the time the obligation is incurred.150

While covered facilities need only select one of the options pre­ sented in this regulation, and thus will be subject to only one m andate, the text of the regulation contains three m andatory search terms.

Similarly, regulations sometimes give regulated entities the option to consolidate regulatory requirem ents. For instance, 43 C.F.R. § 3430.3-2 allows those holding preferential leases for

147. See, e.g., 47 U.S.C. § 203(b)(2) (2017) (“The [Federal Communications] Commission may, in its discretion and for good cause shown, modify any re­ quirement made by or under the authority of this section either in particular instances or by general order applicable to special circumstances or condi­ tions . . . .”).

148. See, e.g., 17 C.F.R. § 200.81 (2017) (setting forth procedures for obtain­ ing a no-action letter from the Securities and Exchange Commission).

149. See, e.g., 29 U.S.C. § 665 (2017) (authorizing the Secretary of Labor to grant exceptions to occupational safety and health regulations as follows: “The Secretary, on the record, after notice and opportunity for a hearing may provide such reasonable limitations and may make such rules and regulations allowing reasonable variations, tolerances, and exemptions to and from any or all provi­ sions of this chapter as he may find necessary and proper to avoid serious im­ pairment of the national defense”).

150. 18 C.F.R. § 292.304(d) (2017) (emphasis added).

MINNESOTA LAW REVIEW126 [103:93

coal prospecting on BLM lands to satisfy two environmental im­ pact assessment requirements with a single analysis:

(a) After the applicant has completed the initial showing required un­ der § 3430.2 of this title, the authorized officer shall conduct an envi­ ronmental analysis of the proposed preference right lease area and pre­ pare an environmental assessment or environmental impact statement on the application, (b) The environmental analysis may be conducted in conjunction with and included as part of the environmental impact statement required for coal activity planning under § 3420.3-4 of this title.161

This regulation is meant to convey that two regulatory require­ ments can be satisfied by one action. However, it employs three string terms to communicate that flexibility.

Alternatives like these are meant to provide regulated enti­ ties with choice and flexibility. Regulated entities need only com­ ply with one, not all, of the alternatives. The more alternatives there are, the less onerous the regulation typically is, but a count of regulatory mandates indicates precisely the opposite.

c. Reference to Other Regulations Many regulations containing restrictive language do not

contain any new restrictions, but merely reference other regula­ tions. For instance, regulations governing recordkeeping by rail­ roads provide:

(a) For purposes of compliance with the recordkeeping requirements of this part, except for the daily inspection record maintained on the loco­ motive required by § 229.21, the cab copy of Form FRA F 6180-49-A required by § 229.23, the fragmented air brake maintenance record re­ quired by § 229.27, and records required under § 229.9, a railroad may create, maintain, and transfer any of the records required by this part through electronic transmission, storage, and retrieval provided that all of the requirements contained in this section are met.162

Although the word “required” is used five times in this provision, it does not contain five distinct requirements. Rather, it contains one new requirement and merely references requirements con­ tained in four other sections (presumably already counted there). The provision includes these references to weave together mul­ tiple, related code provisions so that regulated entities have a coherent roadmap to the regulatory scheme. But for purposes of the regulation count, the provision is charged with five mandates rather than one. 153

151. 43 C.F.R. § 3430.3-2 (2017) (emphasis added). 152. 49 C.F.R. § 229.20 (2017) (emphasis added). 153. See supra notes 56-59 and accompanying text.

2018] THE TROUBLE WITH COUNTING 127

d. Clarification of Legal Requirements Many regulations are promulgated to explain with greater

specificity the meaning of broad, vague legal requirements found in statutes or in other regulations. Such elucidation of legal re­ quirements is meant to provide clarity and certainty in applica­ tion of the law so that regulated entities understand whether the law applies to them and what they must do to comply. 154 Nota­ bly, agencies typically are not under any obligation to promul­ gate clarifying regulations. Agencies with delegated enforcement power are at liberty to enforce statutory law against regulated entities without any guidance or elaboration whatsoever. 155 The promulgation of clarifying regulations is premised on the intui­ tion that regulated citizens are better off when they know how the agency interprets and plans to apply the law “than if they are remitted to the discretion of local agents and to ‘secret law. ’” 156 Indeed, regulated entities often request agencies to is­ sue clarifying regulations or other guidance. 157 In a survey of regulated entities and their attorneys, one respondent noted: “It’s not the number of regulations that is the problem; it’s the inability to understand how they apply to a specific situation that is the problem. In that analysis, more regulations—and

154. Regulation counters admit that clarity is valuable. Clyde Wayne Crews, the perennial author of Ten Thousand Commandments, stated in a memo to the House Oversight and Government Reform Committee that unclear regulations prevent businesses from planning. Regulatory Reform Task Force Check-In: Hearing Before the Subcomm. on Health Care, Benefits, & Admin. Rules & Sub- comm. on Gov’t Operations of the H. Comm, on Oversight & Gov’t Reform, 115th Cong. (2017) (statement of Clyde Wayne Crews Jr., Vice President for Policy, Competitive Enterprise Institute) (Oct. 27, 2017), https://oversight.house.gov/ wp-content/uploads/2017/10/Wayne-Crews-Regulatory-Reform-Task-Force -Testimony-10242017.pdf.

155. See NLRB v. Bell Aerospace Co., 416 U.S. 267, 292 (1974); SEC v. Chenery Corp., 332 U.S. 194, 201 (1947) (holding that the SEC had a duty to enforce standards governing management trading during a reorganization “re­ gardless of whether those standards previously had been spelled out in a gen­ eral rule or regulation”).

156. Strauss, supra note 5, at 808. 157. Ruhl & Salzman, supra note 5, at 785; Strauss, supra note 5, at 805

(describing how license applicants routinely asked the Nuclear Regulatory Com­ mission to clarify which technical designs would satisfy the performance stand­ ards in the regulations); cf. Seth D. Rothman & Jessica Studness, FDA Faces Uncertainty Implementing 21st Century Cures Act, LAW360 (Mar. 15, 2017), https://www.law360.com/articles/900882/fda-faces-uncertainty-implementing -21st-century-cures-act (describing how the FDA regularly issues guidance to clarify its regulations).

128 MINNESOTA LAW REVIEW [103:93

more specific examples of how they apply—actually might be a good thing.”158

For instance, agencies often issue regulations clarifying th eir interpretation of vague statu to ry term s. A prom inent ex­ ample is the U.S. Army Corps of Engineers’ (Army Corps) and the EPA’s “W aters of the U.S. Rule” (WOTUS rule).159 The Clean W ater Act (CWA), by its term s, protects “navigable w aters,” which the statu te defines as “the w aters of the U nited States, including the territo rial seas.”160 W aters th a t fall w ithin this statutory definition are protected by extensive statutory req u ire­ ments, including standards, discharge lim itations, perm itting, and enforcem ent.161 W aters th a t fall outside this statu to ry defi­ nition are not protected by these statu to ry requirem ents. U nder­ standing the precise coverage of the statutory term “w aters of the U nited S tates” is critical to knowing which activities will be covered by CWA requirem ents and which will not, and thus, to providing certainty in planning for those w ith potentially cov­ ered projects. Unfortunately, the statutory term s selected by Congress are notoriously am biguous.162 Consequently, the Army Corps and EPA have endeavored through rulem aking to clarify the definition of the statutory term s.163

While there is ongoing controversy about the proper scope of the WOTUS Rule,164 since 1992, all current, former, and pro­ posed versions of the rule have excluded “prior converted

158. Ruhl & Salzman, supra note 5, at 783 n.85. In a 2014 case study of New Brunswick and Maine landowners, findings suggested that property owners from both locales were “comfortable with most regulations and many agreed that a combination of incentives and regulations are in fact useful.” Michael R. Quartuch & Thomas M. Beckley, Carrots and Sticks: New Brunswick and Maine Forest Landowner Perceptions Toward Incentives and Regulations, 53 ENVTL. MGMT. 202, 202 (2014).

159. 33 C .F.R . § 328.3 (2017). 160. 33 U.S.C. § 1362(7) (2017). 161. See 33 U.S.C. ch. 26. 162. See, e.g., Rapanos v. United States, 547 U.S. 715 (2006) (plurality opin­

ion) (defining “navigable waters” differently from Justice Kennedy’s concurring opinion in the same case, setting up a circuit split over the proper definition of the term).

163. Cf. Supplemental Notice: Definition of “Waters of the United States” - Recodification of Preexisting Rule, EPA.GOV, https://www.epa.gov/wotus-rule/ supplemental-notice-definition-waters-united-states-recodification-preexisting -rule (last updated July 12, 2018) (seeking comments on a proposed rulemaking to repeal the current WOTUS definition while agencies continue to work on a new definition).

164. In a 2014 proposed rule issued by the U.S. EPA, over one million com­ ments were received. Dorothy Noble, WOTUS Backstory: Waters of the United States Debate Continues, FARMING MAG. (Feb. 5, 2015), https://web.archive.org/

2018] THE TROUBLE WITH COUNTING 129

cropland’ 1 6 0 from CWA coverage.166 Prior converted croplands are wetlands altered prior to 1985 to make crop production pos­ sible.167 The scope of the exception is defined in m andatory term s, including the criteria by which prior converted croplands “shall be identified.”168 These regulations articulate a deregula- tory interpretation of the CWA th a t is trem endously im portant to the agricultural industry. The adm inistering agencies are u n ­ der no obligation to in terp ret the statu te in this way or to provide codified guidance of this interpretation.

The agencies could simply proceed with CWA enforcement actions against those farm ers they believe fall under some u n ­ stated, possibly dynamic, definition of “w aters of the United S tates” and support application of the statu te on a case-by-case basis in adjudications or court proceedings.169 Enforcement t a r ­ gets might or might not include those farm ing converted w et­ lands, as they are not explicitly excluded from CWA coverage by s ta tu te .1' 0 It is difficult to see how this clarifying regulation—

web/20150905080533/https://www.farmingmagazine.com/voices/wotus -backstory-waters-of-the-united-states-debate-continues. At one point, the EPA’s website allegedly included an “extensive list of organizations and indi­ viduals that have requested clarification o f‘Waters of the U.S.”’ via rulemaking. Id. During the beginning of his presidency, President Trump issued EO 13,778, which orders the EPA to review, and rescind or revise the WOTUS rule, and all executive departments and agencies to review, and rescind or revise all “orders, rules, regulations, guidelines, or policies implementing or enforcing the final rule.” Exec. Order No. 13,778, 82 Fed. Reg. 12,497, 12,497 (Feb. 28, 2017). The Order further explains that a new proposed rule “shall consider” interpreting the term consistent with Justice Scalia’s opinion in Rapanos, 547 U.S. 715. Id. The Order claims that such changes will “[r]estor[e] the [r]ule of [l]aw, [f] eder- alism, and [economic [g]rowth.” Id.

165. See Current Implementation of “Waters of the United States”, EPA.GOV, https://www.epa.gov/wotus-rule/about-waters-united-states (last updated Sept! 18, 2018) (stating pointblank that “‘Waters of the United States’ do not include prior converted cropland”). Compare 7 C.F.R. § 12.33 (2011) (exempting from regulation the production of agricultural commodities on land deemed to be prior converted cropland), with 7 C.F.R. § 12.33 (2017) (same).

166. Kristine A. Tidgren, Prior Converted Cropland: A 2015 Review, IOWA St . U. Ctr. FOR AGRIC. L. & TAX’N (Aug. 27, 2015), https://www.calt.iastate.edu/ article/prior-converted-cropland-2015-review; see also Current Implementation of “Waters of the United States,”supra note 165.

167. 7 C.F.R. § 12.2. 168. Id. § 12.32(a) (emphasis added). 169. See NLRB v. Bell Aerospace Co., 416 U.S. 267 (1974) (addressing argu­

ments made in SEC v. Chenery Corp., 332 U.S. 194 (1947)); Chenery, 332 U.S. at 203 (holding agencies may decide statutory standards and issues on a case- by-case basis).

170. Michelle R. McKown, A Wetland by Any Other Name: Where Does Fed­ eral Jurisdiction Apply?, GPSOLO, May/June 2016, at 48, 52.

130 M IN N ESO T A LA W R E V IE W [103:93

th a t notifies large num bers of regulated entities w hether (or not) the adm inistering agencies consider th eir activities to be subject to CWA requirem ents171—imposes costs, burdens, or constraints on them. Nonetheless, in calculating the costs, burdens, and con­ stra in ts on regulated entities, regulation counters count com­ m ands clarifying existing legal requirem ents the same as com­ m ands imposing new requirem ents.

5. Counting Does Not Account for Basic G ram m ar and Punctuation a. Counting Does Not Account for the Negation o f Mandates

RegData’s m andate counts are tallied without attention to the most basic gram m atical context in which the search term s are embedded. Many m andatory term s in the C.F.R. are explic­ itly negated by words like no or not. For instance, 40 C.F.R. § 232.2, which governs the environm ental impacts of pilings placed in the w aters of the United States, provides: “(2)(i) Place­ m ent of pilings in w aters of the U nited S tates th a t does not have or would not have the effect of a discharge of fill m aterial shall not require a Section 404 perm it.”172 Regulation counters count the shall without respect to the not th a t exempts certain pilings from the Section 404 perm itting process.173 Not only is 40 C.F.R. § 232.2 not a burden on those placing pilings, but, in fact, it re ­ lieves them of the burdens imposed by another regulation.174 Similarly, 18 C.F.R. § 292.309 relieves electric utilities of certain obligations after a specified date, stating th a t if certain condi­ tions are met, they “shall not be required, under this part, to en­ te r into a new contract or obligation to purchase electric energy from a qualifying cogeneration facility or a qualifying small power production facility.”175 Instead of construing this regula­ tion properly as a relief of regulatory burden, RegData counts it as a double-burden, because it contains two m andatory search term s.176

b. Counting Does Not Account for Punctuation Some portions of the C.F.R. are phrased in “Question & An­

swer” format to make regulatory requirem ents more accessible

171. See 40 C.F.R. § 230 (2017). 172. Id. § 232.2 (emphasis added). 173. See supra notes 56-59 and accompanying text. 174. 40 C.F.R. § 232.2. 175. 18 C.F.R. § 292.309(a) (2017) (emphasis added). 176. See supra notes 56-59 and accompanying text.

2018] THE TROUBLE WITH COUNTING 131

to lay readers. As a result, the text of the C.F.R. is littered with questions, many of which include m andatory language.177 At the very least, counting m andatory term s in both questions and a n ­ swers double-counts the num ber of m andates. Worse, however, it ignores the substance of the answer. For instance, 41 C.F.R. § 302-2.17 asks: “Must I sign a service agreem ent for a ‘last move home’ relocation?”178 It tu rn s out th a t the answ er is: “No, you do not need to sign a service agreem ent for a ‘last move home’ relocation.”179 Similarly, 41 C.F.R. § 302-12.9 inquires: “If my agency authorizes me to enter a homesale program, must I ac­ cept a buyout offer from the relocation services company?”180 Again, the answ er is: “No, if your agency authorizes you to enter a homesale program, your agency m ust give you the option to accept or reject an offer from the relocation services company.”181 RegData counts m andatory language embedded in questions without bothering to learn the answers.

6. Counting Counts Conditional Benefits as Burdens Many regulations set forth criteria and procedures for ob­

taining valuable benefits from the federal government, including grants, loans, leases, and en titlem ents.182 When the federal gov­ ernm ent elects to provide scarce resources to citizens, it allocates them among applicants conditioned upon m andatory, rule-based criteria. These rules need not be followed by citizens who do not seek the government benefit. Those who elect to follow the rules so th a t they may receive the benefit cannot properly be charac­ terized as burdened or constrained by th e ir choice.183

M andatory conditions are attached to government funding in the form of loans and grants. For instance, electric utilities wishing to obtain loans from the Rural U tilities Service “shall” meet conditions relating to the construction of transm ission

177. See infra notes 178—81. 178. 41 C.F.R. § 302-2.17 (2018) (emphasis added). 179. Id. (emphasis added). 180. Id. § 302-12.9 (emphasis added). 181. Id. (emphasis added). 182. See, e.g., infra notes 184-86. 183. This assertion should not be taken as an endorsement of the “bitter with

the sweet” doctrine. See Arnett v. Kennedy, 416 U.S. 134, 153-54 (1974) (plu­ rality opinion). Rather, it is an empirical claim that regulations of the type de­ scribed in this paragraph are not properly characterized as costs, burdens, or constraints on regulated entities.

132 MINNESOTA LAW REVIEW [103:93

lines.184 All others may ignore these commands. Similarly, re ­ searchers who wish to apply for a federal grant to study the pro­ tection and conservation of m arine mammals “shall” comply with num erous requirem ents.185 Those who do not study m arine mammals, or who do so without federal funding, need not.

The government also attaches m andatory conditions to the grant of federal land use rights. The B ureau of Ocean Energy M anagem ent (BOEM), for instance, grants m ineral leases for oil and gas exploration and development in federal w aters off the coast of the U nited S tates based on authority delegated by the O uter Continental Shelf (OCS) Lands Act.186 These lease rights are extraordinarily valuable, accounting for seven percent of U.S. n a tu ra l gas production and twenty-four percent of U.S. oil production.187 To obtain OCS lease rights, an applicant m ust meet multiple m andatory conditions and follow multiple m anda­ tory procedures.188 However, not one of these requirem ents ap ­ plies to anyone who is not seeking to obtain a lucrative OCS lease from the federal government. Related regulations provide OCS lease applicants the additional benefit of appealing adverse de­ cisions by state agencies denying them perm its for offshore oil and gas development.189 Prospective OCS licensees m ust satisfy num erous requirem ents to have th eir appeals heard by the In te ­ rior Board of Land Appeals.190 However, the appeals procedures provide applicants a valuable avenue for getting th eir oil and gas projects federally approved despite state opposition. These ap- peals-related requirem ents can hardly be characterized as costs, burdens, or constraints on regulated entities.

The federal organic labeling program sim ilarly provides lu ­ crative business opportunities for participants. In 2015, U.S. or­ ganic sales reached a record-high $43.3 billion.191 This sales total represented growth of “a robust 11 percent from the previous

184. 7 C.F.R. § 1726.77 (2017). 185. 50 C.F.R. § 82.8(2017). 186. 30 C.F.R. §§ 550.101, 550.200-.204 (2017). 187. Bureau of Ocean Energy Mgmt., Oil and Gas Leasing on the

Outer Continental Sh e l f, https://www.boem.gov/uploadedfiles/boem/oil_ and_gas_energy_program/leasing/5boemre_leasingl01.pdf (last visited Oct. 15, 2018).

188. See 30 C.F.R. §§ 550.200-.262. 189. See 15 C.F.R. §§ 930.120-.131 (2017). 190. 30 C.F.R. § 550.235. 191. Maggie McNeil, U.S. Organic Sales Post New Record of $43.3 Billion in

2015, ORGANIC Trade ASS’N (May 19, 2016), https://www.ota.com/news/press -releases/19031.

2018] THE TROUBLE WITH COUNTING 133

year’s record level and far outstripping the overall food m arket’s growth rate of 3 percent.”192 Those wishing to access the organic m arket m ust comply with extensive regulatory requirem ents and prohibitions.193 Those who do not, need not. N either group of businesses can properly be said to be burdened or constrained.

The foregoing examples focus on regulations th a t distribute government benefits to businesses based on m andatory criteria, because they most starkly dem onstrate the distortions created by counting such regulations as costs, burdens, or constraints on regulated businesses. It bears notice th a t large portions of the C.F.R. set forth m andatory criteria for individuals to obtain gov­ ernm ent benefits like Social Security Disability Insurance,194 Medicaid,195 V eterans’ benefits,196 or welfare assistance197— many of them quite onerous. The regulations im plem enting the Temporary Assistance for Needy Families program, for instance, m andate th a t “[a] p aren t or caretaker receiving assistance m ust engage in work activities when the State has determ ined th a t the individual is ready to engage in work or when he or she has received assistance for a total of 24 m onths.”198 Such regulatory m andates do, indeed, “lim it choice sets,”199 in the preferred p a r­ lance of regulation counters, b u t it is difficult to see how they do so in ways th a t contribute to costs, burdens, or constraints on regulated entities or how these individual burdens might be causally related to macroeconomic outcomes.

In each of the above contexts, and many more, a business th a t wishes to obtain valuable benefits th a t the government has voluntarily elected to provide m ust adhere to certain require­ ments. All others are completely free to forego the benefit and ignore the requirem ents. Regulation counts ignore the fact th a t m any regulatory commands are means of providing funding and m arket opportunities to regulated businesses ra th e r th a n impos­ ing costs, burdens, or constraints on them.

192. Id. 193. E.g., 7 C.F.R. § 205.236 (2017). 194. See 20 C.F.R. ch. 111(2017). 195. See 42 C.F.R. ch. IV (2017). 196. See 38 C.F.R. ch. I (2017). 197. 24 C.F.R. § 5.609 (2017). 198. 45 C.F.R. § 261.10(a)(1) (2017). 199. RegData, supra note 4, a t 112.

134 MINNESOTA LAW REVIEW [103:93

7. Counting Does Not Account for the Benefits of Regulation to Regulated Entities

Much has been made of the failure of counting to account for the benefits of regulation to society,200 and certainly this is prob­ lematic from the perspective of economic theory. However, even if one were to accept the dubious premise th a t there is some th e ­ oretically coherent reason to count only costs to regulated en ti­ ties, it is not clear why one would not net-out from those costs the m easurable benefits th a t regulated entities themselves re ­ ceive from regulation. There is empirical evidence from certain contexts th a t regulated entities benefit from robust regulation. For instance, a rigorous experim ental study shows th a t work­ place health and safety inspections conducted by a state agency not only reduced worker injuries, but also reduced the direct and indirect costs of these injuries to employers, including workers’ compensation expenditures and lost workdays, with no detecta­ ble job loss.201 Studies of securities regulation have revealed th a t “higher enforcement intensity gives the U.S. economy a lower cost of capital and higher securities valuations”202 th a n countries where enforcement is more lax, collectively benefiting U.S.-listed companies, particularly sm aller firm s.203

These studies suggest th a t regulated entities may accrue m easurable financial benefits, both individually and collectively, from being regulated.204 In these cases, the costs of regulation m ight be viewed as a kind of investm ent th a t pays re tu rn s di­ rectly to the investor. Admittedly, this raises questions about

200. See, e.g., Cecot & Livermore, supra note 3, at 3 (“We conclude that the Order is not calibrated to maximize social welfare because it narrowly focuses on the costs of regulation to regulated entities and fails to acknowledge the ben­ efits of regulation.”); Capping Regulation, supra note 3, at 5 (suggesting that it would be irrational not to consider net benefits to “distinguish a good rule from a bad rule”); Lubbers, supra note 43, at 8 (asserting that the “overall main short­ coming” of EO 13,771 is “that it does not account for the benefit of regulations at all”); Freeman, supra note 43 (arguing that EO 13,771 would “strangle even the most beneficial rules under the guise of cutting red tape”).

201. See generally David I. Levine, Michael W. Toffel & Matthew S. Johnson, Randomized Government Safety Inspections Reduce Worker Injuries with No De­ tectable Job Loss, 336 SCI. 907 (2012).

202. John C. Coffee, Jr., Law and the Market: The Impact of Enforcement, 156 U. PA. L. rev. 229, 230 (2007).

203. See generally Douglas Cumming, April Knill & Nela Richardson, Firm Size and the Impact of Securities Regulation, 43 J. COMP. ECON. 417 (2015) (dis­ cussing effects of public and private enforcement relative to firm size).

204. See Coffee Jr., supra note 202; Cumming et al., supra note 203; Levine et al., supra note 201.

2018] THE TROUBLE WITH COUNTING 135

w hether regulated entities receive a good re tu rn on th eir invest­ m ents. But these are very different questions th a n regulation counters purport to ask and answer. Their tallies omit regula­ tory re tu rn s entirely and th u s distort the m agnitude and n atu re of the costs, burdens, and constraints of regulation on regulated entities.

8. Counting Does Not Account for Enforcement Levels It has long been recognized th a t “it is legal rules and th eir

enforcement th a t together shape the incentives”205—i.e. costs— th a t regulated entities face. Regulations are enforced with widely varying degrees of stringency and frequency: some are enforced vigorously and regularly, others are enforced w ith mod­ erate stringency or only sporadically, and many are not enforced a t all.206 At the far end of this continuum are regulations like the detailed licensing requirem ents found in “Public H ealth and E n­ vironm ental Radiation Protection Standards for Yucca Moun­ tain, Nevada,”207 which have laid dorm ant since this designated nuclear waste disposal site was sh u t down in 2010.208

Even in the normal course, adm inistrative agencies have se­ vere resource constraints th a t make it impossible to vigorously enforce all regulations at all times. While environm ental, health, and safety regulations are often cited as among the most b urden­ some, the budget and staffing levels of health and safety agen­ cies substantially inhibit th eir ability to enforce these require­ m ents.209 It has been reported, for instance, th a t it would take the Occupational Safety and H ealth A dm inistration’s staff of around 2400 inspectors more th an ninety years to conduct even cursory inspections of all eligible workplaces in the state of Texas alone.210 Such extraordinary resource constraints require

205. Cento G. Veljanovski, The Economics of Regulatory Enforcement, in EN­ FORCING REGULATION, 171, 171 (Keith Hawkins & John M. Thomas eds., 1984) (emphasis added).

206. See, e.g., id. at 172 (describing different forms of regulation enforce­ ment).

207. 40 C.F.R. pt. 197 (2017). 208. Mike M. Ahlers, Yucca Mountain Project Setup Took Years; Shutdown

Taking Only Months, CNN (May 10, 2011), http://www.cnn.com/2011/US/05/10/ yucca. mountain. shutdo wn/index. htm l.

2 0 9 . See Bridget M. Hutter & Sally Lloyd-Bostock, Risk, Interest Groups and the Definition of Crisis: The Case of Volcanic Ash, 64 B r i t . J. SOC. 3 8 3 , 3 9 9 - 4 0 0 (2 0 1 3 ); Bridget M. Hutter, Variations in Regulatory Enforcement Styles, 1 1 LAW & POL Y 153, 1 6 4 - 6 5 (1 9 8 9 ).

210. Peter Dreier & Donald Cohen, The Texas Fertilizer Plant Explosion Wasn’t an Accident, H U FFPOST: BLOG (June 4, 2013, 2:34 P M ), https://www

MINNESOTA LAW REVIEW136 [103:93

agencies to prioritize enforcement in particular areas and depri- oritize enforcement in other areas.211

In addition to resource-based prioritization, enforcement priorities change with presidential administrations as different chief executives with different policy priorities take office. 212 It has been reported, for instance, that the Securities and Ex­ change Commission sharply curtailed enforcement activities against publicly traded companies and their subsidiaries during the first six months of SEC Chairman Jay Clayton’s tenure.213 In the first year of the Trump administration, several agencies announced that they would suspend enforcement of certain reg­ ulations.214

.huffmgtonpost.com/peter-dreier/texas-fertilizer-plant-explosion_b_3384739

.html. 211. See id. 212. For instance, in M arch 2017, the Federal Motor C arrier Safety Admin­

istration announced th a t it was suspending enforcement of m andated break and hours-of-service rules for commercial truck drivers designed to prevent driver fatigue and associated accidents. See N at 1 Safety Council, FMCSA Scraps Re­ quirement for Overnight Rest Breaks for CMV Drivers, SAFETY + HEALTH MAG. (Mar. 14, 2017), https://www.safetyandhealthmagazine.com/articles/15419 -fmcsa-scraps-requirement-for-overnight-rest-breaks-for-cmv-drivers; see also Hours of Service of Drivers: Notice of Suspension of Enforcement, 79 Fed. Reg. 76,241 (Dec. 22, 2014) (providing notice of suspension of enforcement in 2014).

213. U r s k a V e l i k o n j a , B e h in d t h e An n u a l SEC E n f o r c e m e n t R e p o r t : 2017 AND BEYOND 9 (Nov. 19, 2017), https://ssrn.com/abstract=3074073.

214. For instance, in August of 2017, Neomi Rao, A dm inistrator of OIRA is­ sued a memo to the Chair of the Equal Employment Opportunity Commission expressing the review and immediate stay of EEO-1. Memorandum from Neomi Rao, Adm’r, Office of Info. & Regulatory Affairs to Victoria Lipnic, Acting Chair, Equal Emp’t Opportunity Comm’n (Aug. 29, 2017), https://www.reginfo.gov/ public/jsp/Utilities/Review_and_Stay_Memo_for_EEOC.pdf. This suspends the requirem ent to report salary data by gender, ethnicity, and race. Id. Rao argues th a t such reporting lacks “practical utility, [is] unnecessarily burdensome, and do[es] not adequately address privacy and confidentiality issues.” Id.; see also Stephen Miller, White House Suspends Pay-Data Reporting on Revised EEO-1 Form, SOC’Y FOR HUM. RESOURCE MGMT. (Aug. 31, 2017), https://www.shrm .org/resourcesandtools/hr-topics/compensation/pages/revised-eeo-l-form -suspended.aspx (reporting the stay of the revised provisions of the EEO-1 form). The EPA attem pted to impose a two-year moratorium on the Obama-era fugitive emission of m ethane rule, which restricts m ethane emissions from oil and gas industries. See Clean Air Council v. P ruitt, 862 F.3d 1, 5 (D.C. Cir. 2017). However, the effort was disrupted when the D.C. Circuit held th a t the delay is “tan tam ount to amending or revoking a rule” and the CAA does not authorize the stay of any of the provisions. Id. a t 6; see also Lisa Friedman, Court Blocks E.P.A. Effort to Suspend Obama-Era Methane Rule, N.Y. TIMES (July 3, 2017), https://www.nytimes.com/2017/07/03/climate/court-blocks-epa -effort-to-suspend-obama-era-methane-rule.html (reporting the D.C. Circuit Court’s holding in Pruitt).

2018] THE TROUBLE WITH COUNTING 137

Agencies also sometimes suspend enforcement of regula­ tions while they reconsider them in light of changed circum­ stances or new evidence. For instance, the Federal Motor C arrier Safety A dm inistration announced th a t it was suspending en ­ forcement of a rule th a t m andated breaks and hours-of-service restrictions for commercial truck drivers after receiving the re ­ sults of a study showing th a t the rule did nothing to prevent driver fatigue and associated accidents.215

The upshot of all this is th at, at any given moment in time, only a subset of regulations on the books is being actively en ­ forced against regulated entities. Regulation counts make no ef­ fort to ascertain which regulatory m andates are lying dorm ant due to non-enforcement, and thus not imposing costs, burdens, or constraints on anyone—except, perhaps, the beneficiaries of unenforced legal rig h ts216 or society at large.217 Thus, counts th a t include unenforced regulations, as they all do, overstate the cost, burden, and constraint of regulation on regulated entities.

9. Counting Does Not Account for the Source of Regulatory Requirem ents

Many regulations repeat verbatim (or nearly so) language from the statu te authorizing them. Agencies draft such regula­ tions for convenience, so th a t all the key rules in a regulatory scheme can be found in one place and to avoid the interpretive confusion and inconsistencies th a t might be created by p a ra ­ phrasing ra th e r th a n reproducing statutory requirem ents. The

215. See Notice of Suspension of Enforcement, 79 Fed. Reg. 76,241 (Dec. 22, 2014); James Jaillet, Current 34-Hour Restart Regs to Stay Put Following Issu­ ance of Long-Awaited FMCSA Report, CCG (Mar. 6, 2017), https://www. ccjdigital.com/34-hour-restart-regs-to-stay-put-following-issuance-of-long -awaited-fmcsa-report (describing the Department of Transportation study prompting the removal of the rules).

216. See, e.g., Ben Depoorter & Stephan Tontrup, The Costs of Unenforced Laws: A Field Experiment (N.Y. Univ. Pub. Law & Legal Theory, Working Paper No. 557, 2016) (finding that individuals granted a right to be free from second­ hand smoke suffer a psychological cost when these rights are unenforced even when they are indifferent to the material consequences of second-hand smoke).

217. See, e.g., Utpal Bhattacharya & Hazem Daouk, The World Price of In­ sider Trading, 57 J. FIN. 75 (2002) (finding that non-enforcement of insider trad ­ ing laws results in higher costs of capital in developing countries); Raymond Fisman & Edward Miguel, Corruption, Norms, and Legal Enforcement: Evi­ dence from Diplomatic Parking Tickets, 115 J. POL. ECON. 1020 (2007) (finding a correlation between unenforced parking laws and corruption norms); Ryan Goodman, Beyond the Enforcement Principle: Sodomy Laws, Social Norms, and Social Panoptics, 89 CAL. L. Rev. 643 (2001) (reporting that unenforced sodomy laws in South Africa created a climate of suspicion and surveillance).

138 M IN N ESO T A LA W RE VIE W [103:93

U.S. Supreme Court has term ed this practice “parroting” and de­ nies agencies the heightened Auer deference they are usually ac­ corded for interpretations of their own regulations when they in ­ terp ret a regulation th a t merely parrots statutory language enacted by Congress.218 U nder such circumstances, the Court has said th a t the question presented is not the m eaning of a reg­ ulation th a t the agency itself has crafted, b u t ra th e r the meaning of the statu te.219

Sometimes the parroted statutory language found in regu­ lations contains m andatory term s. For instance, 29 U.S.C. § 1103(a) states th a t “all assets of an employee benefit plan shall be held in tru s t by one or more tru stees.”220 C hapter 29 of the C.F.R. Section 2550.403a-l states, identically, th a t “all assets of an employee benefit plan shall be held in tru s t by one or more tru stees.”221 To be sure, commands found in parroting regula­ tions like this one may constrain the behavior of regulated en ti­ ties. In this sense, they could be said to m easure the construct of costs, burdens, or constraints. However, the commands con­ tained in these regulations are not agency-created costs, b u r­ dens, or constraints on regulated entities. R ather, they are re ­ quirem ents imposed by Congress and the President through duly enacted statutes. Agencies are not a t liberty to modify or elim inate such requirem ents even if doing so would relieve reg­ ulatory costs, burdens, or constraints on regulated entities. Only Congress and the President, acting together u nder Article I, Sec­ tion 7 of the U.S. Constitution, can am end or repeal a duly en ­ acted statutory requirem ent.222 Thus, counting such require­ m ents as costs, burdens, or constraints of adm inistrative regulation misperceives the subject imposing them as well as the mechanism for alleviating them.

In light of the foregoing, it is untenable to m aintain th a t the num ber of regulations or regulatory m andates on the books is an

218. Gonzales v. Oregon, 546 U.S. 243, 257 (2006). 219. Id. 220. 29 U.S.C. § 1103(a) (2011). 221. 29 C.F.R. § 2550.403a-l (2017). Compare 12 U.S.C. § 2607(a) (2011)

(discussing the anti-kickback provision of RESPA, which states that “[n]o per­ son shall give and no person shall accept any fee, kickback, or thing of value for referrals of mortgage loan business), with 24 C.F.R. § 3500.14(b) (2017) (stat­ ing identically in regulations issued by the Department of Housing and Urban Development (HUD) to implement RESPA that “[n]o person shall give and no person shall accept any fee, kickback, or other thing of value” for referrals of mortgage loan business).

222. U.S. CONST, art. I, § 7.

2018] THE TROUBLE WITH COUNTING 139

accurate measure of the costs, the burdens, or the constraints of regulation on regulated entities. If regulation counts do not measure these constructs, is it possible that they serve as a proxy for something else that has not been explicitly articulated? The following Part explores this possibility.

III. THE UN QUANTIFIABLE COSTS OF REGULATION Although regulation counts do not measure actual costs,

burdens, or constraints on regulated entities, they might be an attempt to capture what I characterize here as the unquantifia- ble costs of regulation. Unquantifiable costs are different than costs that are merely unquantified, meaning that agencies have declined to monetize them for one reason or another.223 Unquan­ tifiable costs are costs that defy quantification. This may strike some as an oxymoron. In the long-running debate over CBA, costs are not typically portrayed as unquantifiable in this sense. Costs are the easy part of the calculation. 224 By contrast, CBA has long been criticized for its inability to account for a litany of unquantifiable benefits, from maintaining clear skies to preserv­ ing habitat for polar bears to promoting equity and justice. Ac­ cording to CBA critics, “[t]he basic problem with narrow eco­ nomic analysis of health and environmental protection is that human life, health, and nature cannot be described meaningfully in monetary terms; they are priceless.” 225

Traditionally, costs have not been portrayed in these terms. “[C]osts are typically easier to measure than benefits . . . . Costs often take the form of goods that are priced on markets, while benefits often do not. In addition, regulated entities themselves are often the source for information regarding regulatory costs, and they have incentives to produce information about those costs.” 226 But could it be that some costs are priceless too? Regu­ lation counting may be motivated by the intuition that the true costs of regulation, like the true benefits of regulation, are not fully accounted for in the ledger of monetized costs that agencies

223. Jonathan S. Masur & Eric A. Posner, Unquantified Benefits and the Problem of Regulation Under Uncertainty, 102 CORNELL L. REV. 87, 87-91 (2016).

224. Id. at 116 (“[C]osts are typically easier to measure than bene­ fits . . . . Costs often take the form of goods that are priced on markets, while benefits often do not. In addition, regulated entities themselves are often the source for information regarding regulatory costs, and they have incentives to produce information about those costs”).

225. Ackerman & H einzerling, supra note 17, a t 8. 226. Masur & Posner, supra note 223, at 116.

140 MINNESOTA LAW REVIEW [103:93

regularly include in th eir cost-benefit analyses. If this is correct, w hat might the unquantifiable costs of regulation be? I leave a full elaboration of the answ er to those concerned about account­ ing for them . Prelim inarily, I suspect th a t they encompass a set of psychic burdens, ranging from the stress business owners ex­ perience in trying to comply with extensive and complex regula­ tory requirem ents, to the felt experience of some regulated en ti­ ties and citizens th a t regulation impinges on th eir personal freedom and diminishes liberty in society more broadly. J u s t as proponents of regulation developed metrics to m easure seem ­ ingly unquantifiable benefits of regulation,227 regulation counts might be an attem pt to capture sim ilarly difficult-to-quantify psychic and liberty costs of regulation.

Indeed, much of the political rhetoric around regulation counts links the sheer num ber of regulations to the psychological well-being of business owners and to broader liberty interests. Upon signing EO 13,771, President Trump rem arked sym pa­ thetically th a t it was an attem pt to remedy the fact th a t Ameri­ can businesses “have been treated very badly.”228 In subsequent rem arks, President Trum p lamented:

Unchecked regulation undermines our freedoms and saps our spirit, destroys our companies. . . . We are a nation of explorers and pioneers and innovators and inventors, and regulations have been hurting that and hurting it badly.. . . So together, let’s cut the red tape. Let’s set free our dreams. And, yes, let’s make America great again. And one of the ways we are going to do that is by getting rid of a lot of unnecessary regulation.229

A scertaining no economically justifiable purpose for EO 13,771, economist Robert Shiller surm ised th a t it was likely adopted to salve the feelings of individuals who “have strong business con­ nections and seem to take regulation as a personal affront, as if it stands as a b arrier to th e ir self-actualization and personal ful­ fillm ent.”230

The proposition th a t regulation counts might be a proxy for these types of unquantifiable regulatory burdens opens in te re st­ ing new empirical, theoretical, and political terrain. Empirically,

227. REVESZ & L i v e r m o r e , supra note 17, at 47-48. 228. Bourree Lam, Trump’s ‘Two-for-One’ Regulation Executive Order, A T ­

LANTIC ONLINE (Jan. 30, 2017), https:/Avww.theatlantic.com/business/archive/ 2017/0 l/trumps-regulation-eo/515007.

229. Transcripts, CNN (Dec. 14, 2017), http://transcripts.cnn.com/ transcripts/1712/14/cnr. 0 6 .html.

230. Robert J. Shiller, Why Trump’s 2-for-l Rule on Regulations Is No Quick Fix, N.Y. TIMES (Feb. 17, 2017), https://www.nytimes.com/2017/02/17/upshot/ why-trumps-2-for-l-rule-on-regulations-is-no-quick-fix.html.

2018] THE TROUBLE WITH COUNTING 141

it suggests the contours of a well-defined construct th a t could be used to develop testable hypotheses. For instance, regulation counters could theorize psychic burden as the construct m eas­ ured by regulation counts and could test empirical relationships between the psychic burdens of regulation and macroeconomic outcome variables. To te st these relationships, researchers would have to establish, first, th a t the num ber of regulations or regulatory m andates does, indeed, meaningfully m easure psy­ chic burdens on owners of regulated business and th a t psychic burdens vary with the quantity of regulations. It is not clear th a t regulation counts would be the best m easure of psychic burdens or th a t the psychic burdens of regulation vary continuously, in ­ creasing increm entally w ith each new regulation added.

If such a relationship could be shown, regulation counters could theorize mechanisms by which regulation-induced psychic burdens influence macroeconomic outcomes like productivity and employment. For instance: Do psychically burdened busi­ ness owners forbear from producing or hiring, even if doing so is economically irrational from an efficiency standpoint? Do the psychic burdens of regulation shape business owners’ percep­ tions of the business climate in ways th a t make them less pro­ ductive or less likely to hire? Do citizens in general put forth less productive effort if they believe th e ir liberty to be unduly con­ strained? Such hypotheses are theoretically supportable given w hat we know about how psychology shapes economic action, 231 and exploring them empirically could produce valuable insights.

If such hypothesized relationships were to be established by empirical evidence, the next set of questions would concern w hether removing regulations from the books is the most effec­ tive response. It would be necessary to dem onstrate, for instance, th a t reducing the num ber of regulations or regulatory m andates would actually reduce the psychic burdens of regulation, and th a t this is a more effective method of alleviating psychic b u r­ dens th a n other mechanisms. If the burdens of regulation are, indeed, psychic, then perhaps it would be more effective to shift

231. See, e.g., RICHARD H. THALER, MISBEHAVING: TH E MAKING OF BEHAV­ IORAL ECONOMICS (2015) (explaining how behavioral economics integrates in­ sights from psychology to understand human behavior); Christine Jolls, Cass R . Sunstein & Richard Thaler, A Behavioral Approach to Law and Economics, 50 STAN. L. Rev. 1471 (1998) (explaining how bounded rationality, bounded self- interest, and bounded willpower influence decision making); Jeffrey J. Rach- linski, The Psychological Foundations of Behavioral Law and Economics, 2011 U. ILL. L. REV. 1675 (2011) (highlighting the importance of psychology in under­ standing economic behavior).

142 M IN N ESO TA LA W R E V IE W [103:93

perceptions of regulatory burdens ra th e r th a n to change the ac­ tual num ber of regulations on the books. Such a shift in percep­ tion might be achieved by moderating hysterical political rh eto ­ ric th a t incites regulatory passions by endlessly decrying the massive quantity of regulation crushing the populace.

The theoretical and political implications of surfacing the psychic burdens of regulation are more radical, for they gesture toward the unraveling of the reigning consensus around CBA. As discussed above, regulation counting eschews foundational principles of welfare economics underlying CBA by ignoring the benefits of regulation .232 This deprives CBA of a principled ju s ­ tification. But more subtly, and perhaps more treacherously, reg­ ulation counting suggests th a t even the costs of regulation have not been captured adequately by CBA. It becomes exceedingly difficult to sustain the quantification enterprise if the costs as well as the benefits of regulation include significant unquantifi- able feelings, values, and moral commitments. Understood in this way, the widespread embrace of regulation counting sug­ gests the need to develop (or to resurrect) alternative decision­ making structures to CBA th a t better account for emotional and norm ative considerations.233 Such decision-making structures might allow regulators and the public to weigh more openly and frankly the values a t stake in regulation. For instance, w hat are the relative values of, on the one hand, protecting citizens from preventable harm s like mine collapses, plant explosions, and toxic pollution and, on the other hand, preserving th e ir cher­ ished experience of living in a free society? Ideally, fram ing ques­ tions in this way would lead to more honest political dialogue about regulatory methods, priorities, and trade-offs.

Setting aside for the moment the empirical, theoretical, and political possibilities presented by the psychic burden construct, it is im portant to stress three points. First, to date, none of the above empirical claims about the relationship between psychic burdens and economic outcomes has been established em piri­ cally—or, for th a t m atter, seriously theorized. Regulation coun­ ters have not explicitly claimed psychic burden as th eir con­ struct. It is not entirely clear why not. Perhaps it connotes weakness or victimhood on the p a rt of business owners in ways

232. See supra Part II.B. 233. See, e.g., Gregory C. Keating, Is Cost-Benefit Analysis the Only Game in

Town?, 91 S. CAL. L. Rev. 195, 198 (2018) (exploring the application of “stand­ ards of precaution other than cost-benefit analysis [that] are common in our law”).

2018] THE TROUBLE WITH COUNTING 143

th a t are politically unpalatable. Perhaps regulatory anxiety is too valuable a trigger for political mobilization to acknowledge as the cause of the problem. A ttem pts to alleviate it would blunt this political weapon. In any event, the speculative propositions suggested in this Section are, for now, ju st that.

Second, inflamed regulatory passions, standing alone, do not provide a sufficiently rational basis for deregulatory policies imposed by the executive. A policy calculated to lift the down­ trodden spirits of the P resident’s supporters in the business com­ m unity surely falls into the realm of political justifications for agency action th a t courts have soundly rejected on judicial re ­ view.234 Moreover, it seriously th reaten s the constitutional b al­ ance of powers among the branches in the adm inistrative state. Policies like 2-for-l are not merely political in the sense th a t they relate to the strategy or ideology of a particular party; they are baldly distributional. EO 13,771 is a policy choice to redistribute w ealth to a select category of individuals favored by the P resi­ dent, whom the President sees as disfavored by certain statutory schemes m andating regulation. In many cases, the regulation counting (and cutting) required by EO 13,771 would force agen­ cies to redistribute wealth th at, according to principles of CBA, has been efficiently distributed by already-enacted or proposed regulations.

This is deeply troubling from a separation of powers p er­ spective. D istributive choices are a t the very h eart of the legisla­ tive prerogative.235 Article I, Section I, of the U.S. Constitution delegates legislative power exclusively to Congress.236 Although a long line of cases perm its Congress to delegate the im plem en­ tation of federal statu tes to the executive branch,237 equally well-

234. E.g., Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 51—56 (1983); see also Jodi L. Short, The Political Turn in American Administrative Law: Power, Rationality, and Reasons, 61 DUKE L.J. 1811, 1823-28 (2012); Kathryn A. Watts, Proposing a Place for Politics in Arbi­ trary and Capricious Review, 119 Yale L.J. 2, 83 (2009).

235. Indus. Union Dep’t, AFL-CIO v. Am. Petroleum Inst., 448 U.S. 607, 663 (1980) (Burger, J., concurring).

236. U.S. CONST, art. I, § 1. 237. Whitman v. Am. Trucking Ass’n, 531 U.S. 457 (2001) (holding that the

Clean Air Act properly delegated legislative power to the Environmental Pro­ tection Agency); INS v. Chadha, 462 U.S. 919 (1985) (holding that Congress must abide by its delegation of authority to the attorney general and cannot veto the delegation by the vote of one House); J.W. Hampton, Jr. & Co. v. United States, 276 U.S. 394 (1928) (holding that the delegation of power to the Presi­ dent to increase or decrease the duties imposed by the Tariff Act of September 21, 1992 was constitutional); Cargo of the Brig Aurora v. United States, 11 U.S. 382 (1813) (upholding a conditional law that reinstated a trade prohibition

144 MINNESOTA LAW REVIEW [103:93

established principles of adm inistrative law seek to ensure the rational and non-arbitrary execution of th a t power by agen­ cies.238 The principle of non-arbitrariness serves not only rule of law values, but also separation of powers principles, by making sure th a t agencies do not stray too far from th eir statutory a u ­ thority into the realm of presidential patronage. These founda­ tional constitutional principles and values should not be compro­ mised by m asking baldly distributive political motivations with seemingly innocuous and falsely objective regulation counts.

Third, if regulation does, indeed, impose unquantifiable psy­ chic burdens, this has strategic implications for proponents of regulation as they th in k about how to combat deregulatory in i­ tiatives. In eschewing the pretense of welfare maximization, the regulation counting project suggests th a t the deregulatory im ­ pulse is driven as much by a deep-seated emotional antipathy toward regulation as by economic concerns about costs and effi­ ciency.239 If this is the case, then a generation of regulatory re ­ form efforts aimed a t bolstering support for regulation by grounding it in m arket mechanisms have missed the mark. To date, the prevailing strategy of regulation proponents has been to embrace the CBA discourse of efficiency maximization and to dem onstrate the welfare-enhancing properties of regulation .240 The regulation counting project exposes the limits of this s tr a t­ egy by revealing th a t the tru e aim of the deregulatory project is not more efficient regulation, but simply less.

This insight suggests th a t proponents of regulation m ust do a b etter job of addressing—or neutralizing—negative feelings about regulation if they are to salvage it. In fact, an influential stran d of the regulatory reform project has included sustained

against Great Britain unless the President proclaimed Great Britain was no longer violating the United States’ neutrality).

238. See FCC v. Fox Television Stations, Inc., 556 U.S. 502, 536 (2009) (Ken­ nedy, J., concurring) (stressing that agencies must ground their decisions in “principles that are rational, neutral, and in accord with the agency’s proper understanding of its authority”); State Farm, 463 U.S. at 52; Citizens to Pres. Overton Park v. Volpe, 401 U.S. 402 (1971) (holding that agencies must produce a contemporaneously developed record to facilitate judicial review); SEC v. Chenery Corp., 332 U.S. 194 (1947) (holding that agencies must provide con­ temporaneous justifications for their actions).

239. See Short, supra note 33, at 634-38 (demonstrating that concerns about regulation’s restrictions on liberty dominated debates about regulatory reform between 1980 and 2005).

240. Revesz & Livermore, supra note 17, a t 12-13; Cass R. Su n stein, The Cost Ben efit State: The Future of Regulatory Protection 6 - io (2002).

2018] THE TROUBLE WITH COUNTING 145

attem p ts to do ju st th a t through programs th a t cultivate cooper­ ative relationships with regulated entities. These programs avoid punitive enforcement practices in favor of dialogue with regulated entities about w hat compliance entails, flexibility in the methods required to achieve compliance, and a preference for forbearance and compliance assistance over punishm ent for non- compliance.241 It appears, however, th a t such cooperative ap ­ proaches to regulation have failed to bolster the feelings of the business community towards the regulatory en terp rise . 242 J u st as regulation counting represents a new chapter in deregulatory policy, it should prompt a wholesale reassessm ent of progressive regulatory reform efforts.

IV. WHAT’S THE HARM IN COUNTING? These are demanding prescriptions th a t are not easily re a l­

ized in the short term . Which begs the question, why not ju st go ahead and count in the meantime? Regulation counters appear to enjoy this activity, and while they recognize its empirical shortcomings, they broadly m aintain th a t it is better to have counts th a n not to quantify regulation a t all.243 I argue in this Section th a t there are three reasons to stop the counting now. First, it underm ines the achievement of statutory goals. Second, it is costly and wasteful. Third, it crowds out meaningful dia­ logue and research about the real and difficult problems of reg­ ulation.

First, counting for counting’s sake underm ines the achieve­ m ent of statutory goals. Duly enacted legislation, passed

241. See, e.g., Ian Ayres & J ohn Braithwaite, Responsive Regulation: Transcending the Deregulation Debate 118-20 (Donald R. Harris et al. eds., 1992); Eugene Bardach & Robert A. Kagan, Going by the Book: The Problem of Regulatory Unreasonableness 75-76 (1982); Michael C. Dorf & Charles F. Sabel, A Constitution of Democratic Experimentalism, 98 COLUM. L. REV. 267, 371-73 (1998); Orly Lobel, The Renew Deal: The Fall of Regulation and the Rise of Governance in Contemporary Legal Thought, 89 MINN. L. Rev. 342, 418-19 (2004); Ruhl & Salzman, supra note 5, at 844 n.274.

242. BARDACH & Kagan, supra note 241, at 84. 243. See, e.g., TEN THOUSAND COMMANDMENTS, supra note 5, at 16 (conclud­

ing that “it is worthwhile to track the Federal Register’s page counts” after ac­ knowledging that “there are problems with relying on page counts” and describ­ ing them); Dawson & Seater, supra note 5, at 139-40 (recognizing that “[a] counting measure obviously is imperfect in that two identical values may com­ prise regulations of different types and, even within a given type, may represent regulations of different stringency” but deciding to count the pages of the C.F.R. because this count “probably captures at least some of regulation’s complexity”).

146 MINNESOTA LAW REVIEW [103:93

through the “single, finely wrought and exhaustively consid­ ered”244 procedures set forth in Article I, Sections 1 and 7 of the U.S. Constitution, has charged agencies w ith accomplishing spe­ cific tasks for the purpose of achieving specific social and eco­ nomic goals. These statu tes are intended to benefit or protect many different constituencies, including workers, consumers, in ­ vestors, businesses, children, the infirm, the elderly, and the public a t large. Denying or delaying the enactm ent of im portant new rules th a t are statutorily m andated as “necessary to protect public health and welfare can have severe, even disastrous, con­ sequences: think of tain ted food, toxic spills, unavailable medi­ cines, unsafe train s and planes.”245 Counting thus puts protected statu to ry rights holders a t grave risk of harm .

Second, counting is, itself, costly and wasteful. EO 13,771 requires agencies to devote substantial new resources to analyze the costs of regulations and to propose and support the repeal of regulations through notice and comment procedures.246 Support­ ers of the Order have suggested th a t Congress will have to ap ­ propriate funds for agency staff positions dedicated to these task s.247 In other words, this purported cost-cutting effort will be extraordinarily costly.248 And it provides little in the way of b en­ efits to offset these costs.

244. INS v. Chadha, 462 U.S. 919, 951 (1983). 245. Freeman, supra note 43. 246. Ruhl & Salzman, supra note 5, at 787 (“Initiating and defending regu­

latory erosion also diverts agency resources and public attention from new ini­ tiatives more likely at any moment to be deemed popular and positive.”).

2 4 7 . BROOKINGS EVALUATION, supra n o t e 2 9 , a t 15. 248. Notably, it is much less costly to implement 2-for-l type policies in the

other countries that have adopted them, because these countries are parliamen­ tary democracies. Administrative agencies play a very different role in parlia­ mentary systems than in presidential systems like that prevailing in the United States, and they are governed by very different procedural rules. First, it is eas­ ier to enact regulatory rules legislatively rather than administratively in a par­ liamentary system than in a presidential system because of the identity be­ tween the legislature and the executive. Consequently, there is less delegation of rulemaking functions to administrative agencies and greater legislative con­ trol and prerogative in parliamentary systems than in the U.S. administrative state. Second, because administrative agencies are not entirely independent of the legislature in a parliamentary system, they are not subject to the same kinds of procedural safeguards imposed on U.S. agencies by the Administrative Procedure Act. In particular, parliamentary systems have no equivalent to no­ tice and comment rulemaking, the time consuming and costly procedure by which U.S. agencies must enact and repeal regulations. There are minimal pro­ cedures constraining agencies’ ability to enact or repeal administrative regula­ tions in parliamentary democracies. Thus, removing regulations from the books

2018] THE TROUBLE WITH COUNTING 147

Finally, counting distracts from serious political dialogue and research th a t could produce a more fair and effective regu­ latory system. Citizens look to government for protection from a host of risks arising out of a complex, technologically advanced society, from toxic pollution, to consumer fraud, to discrim ina­ tion, to terrorism . Reducing political dialogue about these issues to the num ber of regulations on the books crowds out meaningful discussion of when and how we should regulate. These conversa­ tions are vital to surface and elaborate collective values about w hat kind of society we w ant to live in and how we should be governed, but it is difficult to hear them above the din of count­ ing.

Research on regulation has an im portant role to play in shaping these conversations and in helping adm inistrators im­ plem ent policy arising out of them. The im portant empirical questions about regulation concern not how much of it there is, but under w hat conditions particular types of regulation are likely to be more or less effective. There are also im portant em ­ pirical questions about the spillover effects regulation might have on various economic outcomes, positive or negative. A handful of studies have sought to ascertain rigorously the rela­ tionship between regulation and employment in specific regula­ tory contexts. These studies, tak en together, suggest th a t this relationship is complex and contingent:249 some regulations are associated with job losses, 250 others are associated with job

does not require the same investment and diversion of administrative resources in these systems that it does in the United States.

249. David M. Driesen, Does Regulation Kill Jobs?: The Lim its o f Quantifi­ cation, 9 Reg. & Governance 193,193-94 (2015). See generally Does Regula­ tion Kill J obs? (Cary Coglianese et al. eds., 2013) (discussing the various em­ pirical research on the effects of regulation on employment).

250. See, e.g., Joseph E. Aldy & William A. Pizer, The Employment and Com­ petitiveness Impacts o f Power-Sector Regulations, in DOES REGULATION Kill JOBS?, supra note 249, at 70 (finding negative employment effects in the most energy intensive sectors); cf. Richard D. Morgenstern et al., Jobs Versus the E n ­ vironment: A n Industry-Level Perspective, 43 J. ENVTL. ECON. & MGMT. 412 (2002) (predicting negative employment effects in industries where pollution abatement activities are not labor intensive and product demand is elastic).

148 MINNESOTA LAW REVIEW [103:93

gains,251 others with the shifting of jobs among regions or sec­ tors252 or even within the same plant, 253 and still other regula­ tions have no statistically significant association with economic outcomes at all.254 In addition, research finds that the relation­ ship between regulation and employment is moderated by non- regulatory factors, for instance, demand levels and industry con­ centration ratios.255 The bottom line is that no credible blanket statement can be made about the aggregate economic effects of regulation. Regulation counting distracts and diverts resources from serious research on economic and other consequences of regulation. Allowing regulation counts to drive policy increases the prospect that we will get these critical empirical questions wrong in specific instances, resulting in more economic harm than good.

CONCLUSION Drawing on prevailing standards in social science research

and a detailed analysis of federal regulations, this Article

251. See, e.g., Morgenstern et al., supra note 250, at 429 (finding positive employment effects in industries like petroleum and plastics where environ­ mental compliance is labor intensive and product demand is relatively inelas­ tic).

252. See, e.g., Michael Greenstone, The Impacts of Environmental Regula­ tions on Industrial Activity: Evidence from the 1970 and 1977 Clean Air Act Amendments and the Census of Manufacturers, 110 J. POL. ECON. 1175 (2002) (finding that in the first fifteen years of the Clean Air Act, jobs and capital in­ vestment likely shifted back and forth from nonattainment areas to attainment areas but th at there was no net reduction in aggregate economic activity).

253. See, e.g., Richard D. Morgenstern, Analyzing the Employment Impacts of Regulation, in DOES REGULATION KILL JOBS?, supra note 249, at 33, 46 (re­ porting that “there is abundant anecdotal evidence th at a large proportion of workers ‘displaced’ by environmental regulation move to other jobs in the same plant or firm”).

254. See, e.g., Aldy & Pizer, supra note 250, at 81 (finding no employment impact of environmental regulations for eighty percent of manufacturers); Eli Berman & Linda T. M. Bui, Environmental Regulation and Labor Demand: Ev­ idence from the South Coast Air Basin, 79 J. PUB. ECON. 265, 269 (2001) (finding negligible employment impacts of Clean Air Act regulation); Nathan Goldschlag & Alex Tabarrok, Is Regulation to Blame for the Decline in American Entrepre­ neurship?, 33 ECON. POL’Y 5 (2018) (finding no statistically significant relation­ ship between RegData’s regulation counts and trends in economic dynamism); Morgenstern et al., supra note 250, at 429 (finding little evidence of employment consequences of regulation in the steel and pulp and paper industries, where labor represents a large share of production costs and demand is relatively elas­ tic).

255. Wayne B. Gray & Ronald J. Shadbegian, Do the Job Effects of Regula­ tion Differ with the Competitive Environment?, in DOES REGULATION KILL JOBS?, supra note 249, at 51, 59-61.

2018] THE TROUBLE WITH COUNTING 149

dem onstrates th a t regulation counting is an irrational and em ­ pirically unsound method of m easuring constructs like regula­ tory costs or burdens and empirically assessing th eir relation­ ship to economic outcomes. Because regulation counts are not a good proxy for mechanisms like cost or burden th a t are p u r­ ported to affect economic outcomes, they have lim ited ability to support causal claims about the relationship between regulation and economic outcomes. Two critical implications flow from this. F irst, agencies and courts should not rely on empirical studies th a t employ regulation counts to im plem ent or uphold the legal­ ity of deregulatory counting policies like 2-for-l. Second, a lt­ hough the num ber of regulations on the books is not an accurate m easure of the costs, the burdens, or the constraints of regula­ tion on regulated entities, it may serve as a proxy for the un- quantifiable costs of regulation. This possibility opens up new ways of thinking about the deregulation and regulatory reform projects and suggests the need for a collective dialogue about the values served by regulation and the way regulation helps and h arm s different groups of citizens. It is tim e to stop counting and s ta r t engaging in meaningful conversations about specific socie­ ta l problems and appropriate regulatory responses.

Copyright of Minnesota Law Review is the property of Minnesota Law Review and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.