McCubbins Admin Procedures as Inst of Poli Control

8/2/2019 McCubbins Admin Procedures as Inst of Poli Control

1/36

Administrative Procedures as Instruments of Political ControlAuthor(s): Mathew D. McCubbins, Roger G. Noll, Barry R. WeingastReviewed work(s):Source: Journal of Law, Economics, & Organization, Vol. 3, No. 2 (Autumn, 1987), pp. 243-277Published by: Oxford University PressStable URL: http://www.jstor.org/stable/764829 .

Accessed: 05/03/2012 15:21

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access toJournal of Law,

Economics, & Organization.

http://www.jstor.org
http://www.jstor.org/action/showPublisher?publisherCode=ouphttp://www.jstor.org/stable/764829?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/764829?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=oup


2/36

Administrative Procedures as Instruments ofPolitical ControlMATIIEWVD. Mc(UBBINSUnlivrrsitytof Californlia, Saln DiegoROG()(ER. NOLLStanfoIrd t/Univcrs'itYBARRY R. WEIN(ASTStanlfoIrdt/niivcrs'ity!

A cen,tral l)rolein ot rel)resenltative (lemoI(cracds h(ow to eIistir( tlhat l)olicyldecisions are respolsive to the interests o(r plre'frences of citizens. TheU.S. (Conlstitiltion deals with thle electoral side of this lpr()oblemb con-structing institutional safteguards and illcentive structlures designedl to mnakeelecte(l replresletatives respl)oi(sive to citizens. B3tt mnaking1policyi,nvolvesmore than dlecisions 1, elected legislators and tlhe Iprsidenlt. lexvital)ly,elected officials delegate c(onsid(erab,leI)olicy'makilngautlhot! to uielec'tedlburieauicrats.Becaulse elect(d (oficials have liillited( resouilrc'eso)r1monlitoringagency' plt0ertformanceaC,he possibility arises tlhat bl)lreaulcratswill not complylwit- tlieir lo(licy lrelferenc, s. Th'is givs riseoi tohe (uestion l)ow-or, in-deed, whetlh-er-electe(l political officials ca'l reasolnably effectively assurethat their policy intentiMonswill lb carried ot.

TIlis paper explo)res the principles of the political c of)lt )l ' bulreaucraticdlecisions. In so (hoiln, we seek to develolp a unifying co)'el)ttial frLameworkfor tw() (eneral typeso)'f coIntrols: 'oversight"-t monlitoring, rewarding, andIullisling l ulre-acratic )behavior-alld ad(llillistrative plroc) llures.

\\e )begill with the pl-remniseJlat tlh political contro)l of agencies is aprinc'ipal-ag-^let lpr)olem. ll general, prinlcilal-agenlt pro(l)leis ()o llt have

'The authors thank lMorris Fiorina, lionLevinl, Jrrvy Mashaw, and Matthew Spitzer forlwelpl to()nlnlo(ts. Mr. NlcCuib! ins acknowledges the supp)ort of the National Science Found(a-tion uncder grant SES-8421 16.IJournal(of Law,. Ecolnomic.s, and Organiz.stion \vo. :3, no. 2 Fall 1987? 198(7 b Yale University. All rights reserved(. ISSN 8756-6222

24:3


3/36

244 / JOURNALOF LAW,ECONOMICS,AN ()ORGANIZATIONII:2, 1987first-best solutions that guarantee perfect compliance. Moreover, the bestavailable solution typically consists of a method for altering the incentivesof the agent (here, the agency). Usually this involves some mechanism for(costly) monitoring of the agent, combined with a system of rewards andpunishments. Standard political oversight-hearings, investigations, bud-get reviews, legislative sanctions-corresponds nicely with this form of so-lution to a principal-agent problem.Administrative procedures are another mechanism for inducing compli-ance. Procedural requirements affect the institutional environment inwhich agencies make decisions and therel)y limit an agency's range of feasi-ble policy actions. In recognition of this, elected officials can dlesignproce-dures to solve two prototypical problems of political control. First, proce-dures can be used to mitigate informationaldisadvantages 'acedby politiciansin dealing with agencies. Second, procedulrescan be used to enfianchiseimportant constituents in agency decisionmnaking processes, thereby assur-ing that agencies are responsive to their interests.The most subtle and, in our view, most interesting aspect of proceduralcontrols is that they enable political leaders to assure compliance withoutspecifying, or even necessarily knowing, what substantive outcome is mostin their interest. By controlling plrocesses, political leaders assign relativedegrees of importance to the constituents whose interests are at stake in anadministrative proceeding and thereby chlanlel an agency's decisions to-ward the substantive outcolitmesthat are ,(most favored b)' those who areintended to be benefited by the policy. Tlihls, )political leaders can be re-sponsive to their constituencies witho()t kno(wing, (r neelding to know(, thedetails of the policy outcomes that these constituents want.The idea that administrative law plays an important role in political con-trol is not novel. Nevertheless, scholarly analysis hlas left manl controver-sies and puzzles. The traditional study(lof administrative law, reviewed indetail 1by Stewart, views administrative 1roce(lres as llealns of iasstrinigfairness and legitimacy in decisions by adlmninistrators.Its fomindations areconstitutional and common law principles of non1delegation, separation ofpowers, due process, and( other procedures tha tprotec tgainst auttcraticand capriciotus decisions by government officials. A major 1)ptzzle thatemerges firom this view is why administrative law is as complex as it is, andespecially why legislation often specifies admiliistrative pr)ocedutres that gobeyond the requirements folr assuring c'onfoirmity with tilese principles.Among thes epuzzles are the (liflt'rences amiong re,gulatory agencies ill p)'-cedures foi p)olicy (lecisions, enfo'Mrcemlent,nd jud(licial review. Fomrxam-ple, the Environmental Prtotection Agency (ElPA) and the OccupationalSafety and Health Administration (OSHA) often regulate the same 1()olll-tants, OSHA iinside the workplace and EPA everywhere else, yet theirprocedures-firom~ setting priorities to methods fir evaluating regulations,to the timling, scope, and nature of judicial review- are quite (lifferent.


4/36

ADMINISTRATIVE PROCEDURES AND POLITICALCONTROL / 245Obviously, neither the Constitution nor common law requires that thesame problem be attacked in different ways by different agencies to achievefairness and political legitimacy.Positive political theory has virtually ignored administrative proceduresin analyzing how political actors-the president and the Congress-can re-tain control of policymaking when dealing with bureaucracies. The theoryholds that political actors secure and retain office in part on the basis ofpolicy outcomes, including the overall performance or "output"of a pro-gram and the distribution of its benefits and costs among politically rele-vant categories of citizens. This theoretical superstructure rests on the as-sumption of effective political control of the bureaucracy.Yet the literatureon political oversight provides good reasons to believe that the traditionalmeans of political control are unlikely to be very effective. The puzzle isthat if political actors cannot control administrative decisions and, there-fore, policy outcomes, their preferences, and those of their constituents,are irrelevant, and it is difficult to imagine why performance in officeshould matter at all to voters in evaluating candidates.The traditional view of administrative law provides a partial solution tothe puzzle in political theory. Judicial review of agency decisions includesan examination of the conformity of an agency's decision to its mandate(which is derived not only from the actual legislation, but from committeereports, floor debates, veto messages, and other detritti of the legislativeprocess). It also considers the conformityof the legislation and agency deci-sion processes with individual rights and democratic values. In reviewingagency decisions, then, an impartial court can veto agency choices whichdo not conform with legislative intent and democratic procedural values.Nevertheless, the mechanism of judicial review is insufficient for assur-ing political control. First, legislative mandates are often vague and broad,thereby placing only loose boundaries on agency decisions. Second, evenwhen legislation is relatively specific, it is unlikely to foresee completely allcontingent circumstances that might confront an agency, which inevitablyleaves some degree of discretion to administrativeofficials. Hence, judicialreview can only hold administrative discretion within reasonable bounds.Third, to the extent that administrative law does more than simply protectdemocratic values, the nature of political forces that give rise to administra-tive law becomes a puzzle in the problem of political control of agencies.Fourth, as argued by Shapiro (1986), traditional views of judicial reviewrest on an analyticallyweak foundation, for they assume that the objectivesof the judiciary (unlike those of elected officials and bureaucrats)are purelyto pursue principles of fairness and legitimacy. To the extent the courtspursue policy objectives that do not conform to the wishes of elected offi-cials, administrative law (through legislation or executive order) may be inpart a means for controlling the judiciary as well as for assuring adherenceto democratic values.


5/36

246 / JOURNALOF LAW,ECO()N()MICS,NI)()IR(ANIZATI()N11:2, 1987This paper examilies the role of adininistrative law in assisting politicalactors in controlling the lbueaucracv. While it lborrows fioml tie discussion

of administrative law as assuring the political legitimacy of'agency deci-sions, our discussion is l)roader aind is positive, not normative, in its objec-tives. Specifically, the hypothesis we put ftrth is tllat mnuchof' administra-tive law-indeed, mnostadministrative law that is not derived fi-om judicialinterpretation of the Constitution and commion law principles of adminis-trative fairness-is written for the purpose of helping elected politiciansretain control of policymaking. Our hypothesis primarily concerns Con-gress, but it also applies to the Executive, which often helps develop andalways approves legislative changes in administrative law and which di-rectly alters administrative )procedlures tihrough the issuance of executiveorders.

Section 1 reviews the literature on political control of the bureaucracv.The principal mechanislms fiorinfluencing bureaucratic implementation thatare discussed in the literature are monitoi'ing and salctions. Congress andthe president can reward or punish agencies for their policy choices. Weargue, however, that the costliness of monitoriing and sanctions limits theireffectiveness. Moreover, the magnitude of' sanctions is limited. This leadsto less effective control and ultimately to slippage in compliance.

Section 2 argues that the legal constraints imposed in the AdministrativeProcedures Act (APA) and elsewhere enal)le political officials to overcomecertain informational inelqualities between themselves and administrativeofficers. By requiring agencies to collect and disseminate politically rele-vant information, Congress and the president lmake the tihreat of' sanctionsa more efficacious control device. Moreover, the administrative systeln isdesigned so that some of the costs of enfiorceiment are b)orne not by politi-cians, but by constituents and the courts. Finally, administrative proce-dures affect the costs to agencies of implemlenting policies that are opposedby groups enfranchised by these procedures. This alters the incentive structureof the agencies and thereby shapes their decisions.Section 3 contains specific examples of the phenomena described in sec-tion 2. Section 4 discusses how the administrative system is affected by,and affects, the prerogatives of the courts. Section 5 examines how thesystem creates a form of representative democracy in administrative pro-ceedings.

1. SOLUTIONS TO THE PROBLEM OF BUREAUCRATICCOMPLIANCECreating an agency with discretionary authority to make policy decisionscauses a potentially important problem: the agency mav mlake decisionsthat depart from the policies (including distributive benefits and costs) thatCongress and the president would otherwise have chosen. The mechanics of


6/36

ADMINISTRATIVEPROCEDURES AND POLITICALCONTROL / 247how and why agency decisions would depart firom he policies preferred bymembers of Congress and the president are the subject of a vast literattire,and we review it only briefly here. The fundamentalpremise of this litera-ture is that bureaucrats have personal preferences which conflict withmembers of Congress and the president. The policy choices of the latterare disciplined by the requirement that periodically they seek ratificationoftheir performance in office by their constituents. The choices of agencyofficials are not subjected to electoral discipline. Consequently, in the ab-sence of effective oversight, they are likely to reflect personal preferences,derived from some combination of private political values, personal careerobjectives, and, all else equal, an aversion to effort, especially effolrtthatdoes not serve personal interests.The crime of runaway bureaucracy re(quiresopportunity as well as mo-tive, and this is supplied by asymmetric information. A conse(luence ofdelegating authority to bureaucrats is that they may become more expertabout their policy responsibilities than the elected representatives who cre-ated their bureau. Information about cause-effect relations, the details ofexisting policies and regulations, the pending decision agenda, and the dis-tribution of benefits and costs of agency actions is costly and time-consuming to acquire. As in all agency relationships, it may be possible forthe agency to take advantage of its private information.Several consequences can emerge from this situation. One is simpleshirking:an agency becomes a Club Med for government officials who un-dersupply policy decisions. Another is corruption:agency officials allow thebureau to be "captured"by selling out to an external group. Still another isoligarchy: the peculiar political preferences of the agency override demo-cratic preferences. The challenge for political overseers is to prevent theseoutcomes.

Because monitoring and enforcement are not costless, no method of in-fluencing administrative decisions will be perfect. Rather, elected repre-sentatives face a tradeoff between the extent of compliance they can com-mand and the effort that is expended to assure it, effort which has anopportunity cost because it can also be used for other politically relevantpurposes. Moreover, political actors can be expected to engage in delega-tion even if they find perfect compliance to be excessively costly. Delega-tion confers a benefit by expanding the scope of politically relevant activityavailable to them. Imperfect compliance, then, is simply a cost of delega-tion to be balanced against this benefit.The problem of bureaucratic compliance has long been recognized as aprincipal-agent problem.1 Specifically, members of Congress and the presi-dent are principals in an agency relationship with an executive bureau. As

1. See, for example, Rose-Ackerman;Moe (1984); Weingast (1984); and, of course, Mit-nick (1975, 1980).


7/36

248 / JOURNAL )F LAW, ECON()MI(S, ANI) O()IANIZATION 111:2, 1987

in all agency relationships, tilt' principals will seek methods to cinsuire thatdelegation is more I)encficial tihani ostly.One difference l)etween this agency r-clattionshipnd the olnes inormallyencountered in economiinic heorv is that there are manyil rillc'ials. The

Constitution establislhed a govcrlliental svstlll wllerel)v the authlority tomake and control pub)lic policy is fragmente(l: not only was autlhority di-vided between the president andlC(ongress, l)llt legislative authioritywasdivided into two branches. Il(lee(l, one of t(he virtles Madisonl saw in theConstitution was that authiority was divided still firtiler between the fed-eral and state governments. In order to reduce the olppomrtunitiesor collu-sion among the members of'each instituition,the founders gave each a dif-ferent tenure of office and mad(ethem responsive to (lifierenltconstituen-cies. Because they represent (lifferent interests, tle elected officials in eachof'these institutions, acting as principals with respect to the agency, willlikely seek to influence an agency's policy chloicesin (lifferentdirections.In the twentieth centuiry,p1olicymakingl Congress has become ftirtherfragmented, as committees and(sulcolmmiittecs have bIecome the loci fordecisions. In our analysis, we ignore Imost of' this complexity. We focussolely on the mechanismis lby which elected officials conitrol decisions inagencies, and not on lhow various elected officials cnlforce their own agree-ments and compromises.1.1. REWAIR)S, SANCTI()NS, ANI) MONITI()OIN(; AS C()NTRO()I SITRATEGIESThe literature on political control of the bureaucracy descril)es a variety ofcontrol strategies available to political actors that exelmplifv raditional solu-tions to the principal-agent prollem. For the most part, the scholarly liter-ature emphasizes the use of "reactive" strategies which elected politiciansadopt after some blunder or radical departure fiomi their inten(led policyhas occurred.2 Even the monitoring fiinction is largely reactive, especiallyin Congress, where active oversight of many agencies (especially regulatoryagencies) is infrequent (U.S. Senate, Committee on Goverinment Opera-tions, 1977).Politicians have several means available to reward or punish agencies.Under extreme circumstances, civil servants can be removed fiomi office,and even prosecuted, if their actions stray too far from the grey areas sur-rounding the mandate and power of their agency. The top officials in anagency are usually political appointees and can be impeached by Congressor, with a few exceptions, fired by the president. Appropriations and re-authorizations bills also provide a means for either general or progranmmati-cally targeted rewards and punishments (Kirst). Public hearings and inves-tigations, while part of the monitoring fiinction, also can serve to subject

2. See, for example, Fiorina (1982); McCubbins and Schwartz (1983); Moe (1985); andWeingast and Moran (1983).


8/36

ADMINISTRATIVE PROCEDURES AND POLITICALCONTROL / 249recalcitrantbureaucrats to public humiliation that devastates their careers.Finally, legislation or executive order can reorganize agencies and shufflepolicy responsibilities among them, thereby reallocating policymakingauthority.

Although these actions are reactive in that they take place after a sus-pected impropriety has occurred, their availability affects the incentivesfacing bureaucratic decisionmakers. The value of a political punishment,multiplied by the likelihood that improper behavior will be detected andpunished, enters as a cost in the calculation of net benefits by a bureaucratwho contemplates straying from the preferences of political overseers. Ifdetection and punishment are sufficiently likely, and the magnitude of thepunishment sufficiently great, a noncomplying action can be deterred.Thus, the presence of sanctions, by forcing administratorsto anticipate po-litical reactions to their policy decisions, provides some measure of protec-tion from noncompliance.3Recent studies of bureaucratic decisionmaking have supplied evidencethat agencies are responsive to members of Congress and the president.Weingast and Moran(1983) show how the mix of cases at the Federal TradeCommission (FTC) changes in response to changes on the relevant over-sight committees in Congress; Moe (1985) shows that both the presidentand congressional committees play an important role in the determinationof cases at the National Labor Relations Board (NLRB); Grier and Beckboth show that the Federal Reserve Board responds to its political princi-pals to a far greater extent than was previously thought possible.Nevertheless, by itself, a system of rewards and punishments is unlikelyto be a completely effective solution to the control problem. This is due tothe cost of monitoring, limitations in the range of rewards and punish-ments, and, for the most meaningful forms of rewards and punishments,the cost to the principals of implementing them.

Monitoring. By themselves, rewards and punishments do not deal di-rectly with the problem of asymmetric information. If agencies have betterinformation, they have a range of discretion that is undetectable to politicaloverseers, and so, in the absence of monitoring, some noncomplying deci-sions will not be subject to retribution. Thus, if noncompliance is a seriousproblem, one would expect political actors to invest substantialresources inmonitoring; indeed, this appears to be the case (Aberbach).Policy monitoring in Congress takes two forms. The mor-eapparent, but

3. Examples are numerous: the dismantling of the Area Redevelopment Administration n1963, only two vears after its creation, is an extreme (Ripley). In the last twenty years, Con-gress has repeatedly intervened in decisionmaking at the FTC, the most famous case regard-ing proposed FTC regulation on cigarette advertising (Fritschler). More recently, Congress hasintervened to stop the FTC fromiregulating the insurance industry, television advertisingaimed at children, funeral homes, and used car sales. Congressional intervention at the FTCalso sends a clear message to administrators in other agencies: it is necessary to anticipatecongressional reaction to their regulatory policies (Mendeloff).


9/36

250 / JOURNALOF LAW,ECONOMICS,AND ORGANIZATION11:2, 1987probably less important, is ongoing oversight and evaluation by congres-sional subcommittees and agencies that are arms of Congress, such as theCongressional Budget Office (CBO) and the General Accounting Office(GAO). Less apparent, but probably more important (judging from howmembers of Congress allocate their time and staff), is "fire-alarm"monitor-ing (McCubbins and Schwartz, 1984)). This form of monitoring consists ofdisappointed constituents pulling a member's fire alarm whenever anagency harms them. Oversight is then a form of constituency service formembers. Constituency service has become an increasingly important ac-tivity of members of Congress in the postwarera, to the point where it nowaccounts for more than half of the staff effort in Congress and is a criticalfactor in assuring the reelection success of melembersFenno, 1973; Fiorina,1977a; Fiorina and Noll, 1978; Cain, Ferejohn, and Fiorina).Policy monitoring in the executive branch is concentrated in the evalua-tion process. The Executive Office of the President is comprised of numer-ous organizations, most notably the Office of Managemenetand Budget(OMB), that scrutinize budgets, programs,and operations of agencies. Fur-thermore, cabinet officials-the people mnostpolitically responsive to thepresident-also have their own independent evaluation stafls.To facilitate the monitoring process, political actors impose informationcollection and reporting re(quirementson agencies. Both Congress and theOMB receive oceans of data and reports fi-omoffices within agencies aboutongoing programs. And, through GAO and the General Services Adminis-tration (GSA), political actors impose rigid accounting and record-keepingrequirements on agencies that can be used subsequently as the basis forsanctions (Kiewiet and MIcCubbins).Though monitoring is probably far mlore pervasive and effective thanwas once thought, it imposes costs on political actors. First, resources de-voted to monitoring have an opportunity cost, for they presumably couldbe devoted to delivering more governmlent services to constituents orshrinking the burden of the public sector. Second, the time used by politi-cal principals in acq(uiring information, assessing the degree of noncompli-ance, and deciding what punishiiient strategy, if anyl,to lundertakealso hasan opportunity cost. Easy and (tuick complliance is pref-erred, for it enablespolitical actors to provide more service to constitulcnts in a given amount oftime.

In addition to imp)osing significant costs, moniitoring is likely to be onlyan imperfect mechanism for detecting Iloncompliance. First, cause-effectrelations in human affairs oftein are subljcct to an important (legree of irre-ducible uncertainty, so that nio matter how carcfiully the colse(luences of anagency's actions are monitored, political actors will be ullsure about theextent to which an agency undertook best efolrts to comply with its princi-pals' wishes. Second, monitoring consists primarily of information receivedafter an action is taken. One form of' noncompliance is for an agency to


10/36

ADMINISTRATIVE PROCEDURES AND POLITICAL CONTROL / 251make a decision that creates a new political interest which is antithetical tothe existing political structure.4 In this case, sanctions against an agencygenerate political costs that would not have to be faced had the policy deci-sion been anticipated and prevented. The result is more noncompliancethanwould otherwise be the case, despite accurateretrospective monitoring.The final difficulty of any monitoring system pertains primarily to themore traditional oversight function, as practiced by OMB and in congres-sional hearings. It is the ultimate dependence of external monitors on infor-mation supplied by agencies. In a sense, the agency both keeps the booksand performs the audit. If agencies have importantprivate information,notall of which can be obtained by external monitors, they can use this infor-mation to hide noncompliance. This problem cannot be solved by increas-ing the intensity of external monitoring but must be solved by giving theagency an incentive to make all of its informationpublic.We have already observed that noncompliance must be perceived as aserious problem because so much effort is devoted to monitoring. Now weconclude that direct monitoring by elected representatives is likely to be animperfect mechanism for detecting noncompliance. If so, there is a goodchance that some shirking by bureaucrats goes undetected. This can beoffset only if penalties are sufficiently high and sanctions, when applied, arenot commensurately costly to political principals.Limits to sanctions. A key result in principal-agent theory is that insome circumstances-and especially when noncompliance is difficult todetect-the magnitude of the sanctions necessary to effect compliance mustbe very large (indeed, often much larger than the potential rewards to theprincipal from compliance-see Holmstrom). In the case of policy imple-mentation through administrative agencies, the maximal stakes of bureau-crats are criminal penalties, and these are available only when noncompli-ance takes the form of willful violation of the letter of the law. In caseswhere noncompliance is not criminal, the range of sanctions is more lim-ited and of comparable magnitude to the costs an agency can impose onpolitical actors. Specifically, each can undermine the career objectives ofthe other, and each can thwart the other from achieving preferred policyactions. Thus, absent illegal activities by agencies, the sanctions availableto political actors are roughly comparable to the costs agencies can imposeon politicians.If this is the case, two logical consequences follow. First, monitoringeffort should be intensive so that the lilnits to sanctions can be offset tosome degree by higher detection probalilities. Hence, limits to sanctionsof public officials provide an explanation for the elaborate monitoring svs-

4. See Wildavskv concerning the development of agency constituencies, and Noll andOwen (1983) concerning the creation of interest groups by regulatory decisions that definenew property rights.


11/36

252 / JOURNAL OF LAW, ECONOMICS, AND ORGANIZATION 111:2, 1987

tem of the public sector. Second, if sanctions are based prilnarilyon Inoni-toring, then imperfect monitoring implies that substantial noncomplyingbehavior will go undetected.

Political costs of sanctions. Not only is the magnitude of sanctions fornoncompliance limited, but most of the methods for imposing meaningfulsanctions also create costs for political principals. Some forms of sanctionsrequire legislation, which demands the coordinated effort of both houses ofCongress and the president.5 The introduction of legislation creates theadditional problem that it can reopen long settled, but still contentious,aspects of a policy that are unrelated to the compliance problem. To im-pose legislative sanctions, therefore, requires running the risk of other un-desirable legislative outcomes from the perspective of any given electedofficial.Another potential cost of sanctions is the response of the electorate tonew informationabout government wrongdoing. Citizens have a principal-agent relationship with elected political officials that is broadly similar tothe relationship between elected officials and bureaus. Specifically, in theface of imperfect and costly informationof consideralle complexity, constit-uents must assess the extent to which elected officials, not agencies, are

guilty of noncompliance with wishes of the electorate. Hence, a publiclyvisible investigation and punishment of an agency may raise doubts in theminds of the electorate about the attentiveness to business of their electedofficials. They may conclude that Congress or the White House, not theagency, was the Washington branch of Club Med. Or, if citizens vote retro-spectively, they may respond to information about government malfeasanceof any sort by simply voting against incumbents.Finally, the act of imposing sanctions distracts agencies fiom the deliv-ery of public services, focusing their attention on minimizing the damagethey will suffer as a result of detected malfeasance. Investigations, "mid-night massacres"of key officials, and reorganizationsare disruptive of thebusiness of the agency. To the extent the agency is delivering some politi-cally relevant services, this disruption, too, is costly to elected politicians.Of course, if imposing significant sanctions is costly to political actors aswell as to bureaucrats, the ardorfor them among politicianswill be propor-tionately dampened. Specifically, not all acts of noncompliance, once de-tected, will be punished significantly, giving agencies an additional incen-tive to pursue their own preferences at the expense of political principals.

5. Procedures have been established that decrease the collective action costs involved inimposing sanctions. Committee vetoes, as imposed on the FTC in the FTC ReauthorizationAct of 1980, involve only a small subset of the legislature. Procedureshave also been designedthat circumvent the requirement of presidential acceptance for proposed sanctions. The mostobvious example is the legislative veto. Congress adopted legislative vetoes in hundreds ofacts since the 1930s. Another method of effectively bypassing presidential influence is throughthe use of riders on appropriationslegislation. With the loss of the legislative veto, appropria-tions riders have become a favorite tool with which to sanction agencies (Cooper).


12/36

ADMINISTRATIVEPROCEDURES AND POLITICALCONTROL / 253The thrust of these arguments is not that rewards, punishment, and

monitoring are unimportant in the public sector. Indeed, monitoring isintense only if it is useful to political principals, and its utility arises in partfrom the ability of political actors to punish or reward agencies on the basisof detected compliance patterns. Extensive monitoring makes detection ofnoncompliance more likely and sharpens the incentive effects of sanctionsby allowing political actors to impose them in more exact proportion to theprobability and magnitude of noncompliance. Consequently, shirking andmalfeasance are going to be less attractive to agencies than they would bein the absence of monitoring and sanctions.Nevertheless, by themselves, monitoring and sanctions do not comprisea perfect solution to the problem of bureaucraticcompliance because theyare costly, inexact, and subject to fundamental limitations. Thus, onewould expect politicians to welcome other measures that may be availablefor altering the incentive structure of agencies, especially if these alterna-tives have relatively low cost. An optimal mix of the measures, where eachmeasure complements the strengths of the other and substitutes for theother's weaknesses, will establish less costly and more effective control ofthe bureaucracy by political principals. Hence, the stage is set for analyzingadministrative law as such a mechanism.

2. PROCEDURAL SOLUTIONS TO COMPLIANCE PROBLEMSAdministrative procedures can be solutions to problems of noncomplianceby agencies only if procedures actually affect the outcomes of decisionmak-ing processes. Such is not necessarily the case, for elaborate procedurescan serve at least two other ends. First, as developed by both Michaelmanand Mashaw (1983; 1985a), procedures may be ends in their own right.Regardless of the outcome, people may derive greater value from processesthat treat them respectfully and give the appearance of rationality thanfrom processes that are perceived to be cruel, unfair, and arbitrary. Sec-ond, procedures may be a ruse aimed at the electorate in that they have noeffect on outcomes but transfer apparent responsibility for decisions fromelected political officials to agencies or courts (Fiorina, 1985). Whereas asusually stated this implies irrationalityor gullibility on the part of the elec-torate (for example, see Edelman), this is not necessarily the case. Admin-istrative procedures in some form may be necessary to protect other values(such as constitutional rights and procedural characteristics referred toabove) and so bound (but not determine) outcomes while simultaneouslyincreasing the complexity of decisionmaking processes and the informa-tional requirements to comprehend them. If so, administrativeprocedurescan simultaneously provide a net benefit to citizens and attenuate the abil-ity of citizens to allocate political responsibility for policy outcomes.While both lines of argument point to plausible features of administra-


13/36

254 / JOURNAL OF LAW, ECONOMICS, AN) ORGANIZATION 111:2, 1987tive processes, we argue that this is not all there is to procedures. Specifi-cally, we assume that the details of administrative law as applied to anygiven decision problem will affect the outcome. The basis for this assump-tion is the presumption that decisions depen(l on the informationthat un-derpins them and on the means for relating that information to decisionsthat are permissible according to the strictures of administrative law. Ifdecisionmakers must take account of all of the relevant informationthat isavailable to them, and if participants in an administrative process can berelied upon to provide information that is, on balance, favorable to theirinterests, then rules of standing and evidence and the allocationof burdensof proof will affect the range of decisions available to an agency. Of coursethis observation is hardly novel; a classic public administrationtext, writtenshortly after the passage of the APA, predicted that the new procedureswould alter the range of interests represented before federal agencies (Si-mon et al.: 521). Indeed, the basis forproposalsto assure broad representa-tion in administrative processes (see Stewart) and for relatively narrow in-terpretation of the extent of legislative delegation to an agency (see Aran-son, Gellhorn and Robinson; and Mashaw, 1985b) is a belief that processmatters.If procedures do affect outcomes, political officials have available tothem another tool for inducing bureaucraticcompliance. Specifically, alter-ations in procedures will change the expected policy outcomes of aldminis-trative agencies by affecting the relative influence of people who are af-fected by the policy. Moreover, because policy is controlled by participantsin administrative processes, political officialscan use procedures to controlpolicy without bearing costs themselves, or even having to know what pol-icy is likely to emerge. The burden of this section is to demonstrate theplausibility of the contention that an important function of administrativeprocedures is to provide a means of inducing bureaucraticcompliance thatdoes not require the time, effort, and resources of political actors.A wide variety of administrative procedures may be applied to an agen-cy's decision processes. This reflects in part the diversity of the interests ofits principals. Some mechanisms may disproportionatelyenhance the abil-ity of some principals to control agency choices. The traditional means ofcontrol, both statutory and nonstatutory, are largely exercised by congres-sional committees. The appropriationscommittees, for example, can legis-late a great deal of policy under the cover of spending bills. But this doesnot necessarily mean that committees are benefiting most from the exerciseof this power. It may be, as Fenno (1966) argued, that the appropriationscommittees are acting as agents for the entire legislature. The game playedout between the committee and its chamber, between the chambers, andbetween Congress and the president determines which of these actors,when placed in its relationship as principal with respect to some agency,will be most advantagedby a given set of procedural constraints. The anal-


14/36

ADMINISTRATIVE PROCEDURES AND POLITICAL CONTROL / 255

ysis of this "meta-game" between principals is beyond the scope of thispaper. Instead, we wish to advance the hypothesis that administrativepro-cedures enhance the ability of political principals in general to solve theiragency control problems.There are two general forms off control problems. First, political princi-pals in both branches of government suffer an informational disadvantagewith respect to the bureaucracy. Because of this ubiquitous problem, thepolitical principals will seek a ubiquitous solution. We argue that many ofthe provisions of the Administrative Procedures Act solve this asymmetricinformation problem. Second, the coalition that forms to create anagency-the committee that drafted the legislation, the chamber majoritiesthat approved it, and the president who signed it into law-will seek toensure that the bargain struck among the membersof the coalition does notunravel once the coalition disbands. Specifically, the coalition will seek tocombine sanctions with an institutional structure to create pressures onagencies that replicate the political pressures applied when the relevantlegislation was enacted. Here, the point of administrativeprocedures is notto preselect specific policy outcomes, but to create a decisionmaking envi-ronment that mirrorsthe political circumstances that gave rise to the estab-lishment of the policy. Whereas political officials may not know what spe-cific policy outcome they will want in the future, they will know whichinterests ought to influence a decision and what distributive outcomes willbe consistent with the originalcoalitionalarrangement. In other words, thecoalition "stacks the deck" in the agency's decisionmaking to enhance thedurability of the bargain struck among members of the coalition. We takeup specific examples of "deck-stacking" n section 3.If these uses of administrativeprocess are effective, the agency, withoutany need for input, guidance, or attention from political principals, is di-rected toward the decisions its principalswould make on their own, even ifthe principals are unaware, ex ante, of what that outcome would be. Bystructuringthe rules of the game for the agency, administrativeproceduressetquenceagency activity, regulate its informationcollection and dissemina-tion, limit its available choices, and define its strategic advantage. More-over, an importantfeature of this system is that constituents, agencies, andthe courts bear much of the costs of ensuring compliance. Indeed, courtsare the key, for without them political actors could not rely on decentral-ized enforcement.

2.1. POLITICALCONSEQUENCESOF TIlE AIMINISTRATIVEPROCEI)URESA(CTThe AdmlinistrativeProcedures Act of 1946 codified over a half-century ofcourt decisions affecting administrative proceedings. The court's rationalefbr establishing the procedural requirements enmbodied n the act was to


15/36

256 / JOURNALOF LAW,ECONOMICS,AND ORGANIZATION11:2, 1987ensure that procedural justice applies in agency decisionmaking. But theAPA did more than this. Indeed, it took Congress and the president adecade to work out the details of an act to which they could both agree.Thus the APA is in part a political document, written to enhance politicalcontrol. The twin goals of proceduraljustice and agency control were notedin the Report of the House Judiciary Committee accompanying the pro-posed act: the act "is designed to provide . . . fairness in administrativeoperation" and "to assure . . . the effectuation of the declared policies ofCongress" (U.S. Congress, 1947: 252).Important ideological and symbolic factorsundoubtedly played a role inthe passage of the APA. But though little in the APA represented a newprocedural innovation not found in previous court decisions, the APA wasnot only symbolic legislation, nor was it simply a statutory recognition ofexisting judicial precedent. Prior to the APA, procedural requirements im-posed by the courts differed across agencies. This included procedures re-lating to informationgathering and disclosure and to those concerning evi-dentiary standards. Two important effects of this act, therefore, were toimpose greater uniformity across agencies and to raise the minimum stan-dards to which an agency must adhere. A majorfocus of the congressionalhearings, for example, concerned differing evidentiary standards. For ex-ample, the court had held the Interstate Commerce Commission (ICC)to asubstantialevidence standard since the early part of this century, but manyof the New Deal agencies were not subject to this constraint (see U.S.Congress, 1947).2.2. INCENTIVESTOGAIN RELEVANTPOLITICALNF(RMATIONPoliticians delegate authority to an agency for a variety of reasons. One isthat the policy is inherently controversial and so politicians may seek todistance themselves from the ultimate policy choice by "shifting responsi-bility" to an agency designed to take blame (Fiorina, 1985). In this case,the principle design criterion for an administrative process is likely to beprocedural fairness, as perceived by the warring interests, and a propen-sity to find compromise, so that in the end the participants will have ablunted incentive to take further political action to alter the policyoutcome.Another motive for broad delegation of authority occurs when politicalleaders are uncertain about what politically is the most desirable policy(McCubbins, 1985). It can then be in their interest to set in motion processesthat will resolve these uncertainties and that will use the newly acquiredinformation to carry out the policy preferences they would have if fullyinformed. To accomplish this, political principalsmust first provide bureau-crats with the means to collect informationabout the consequences of vari-ous policy options (Fiorina, 1982). Second, political leaders must impose


16/36


procedures that cause the information to be used to make decisions thatserve their interests.Delegation of authority makes it possible for agencies to adjust policiesin directions desired by political leaders as more information is obtained(Mashaw, 1985b). But this is only a possibility, not a certainty. If greaterdelegation allows agencies greater opportunities to pursue their own goals,it only helps the agencies, not the political principals. Hence, greater dele-gation implies a greater need for effective control mechanisms (McCub-bins, 1985).In order to make decisions that serve their own interests, elected offi-cials need to know the following: (1) the precise nature of the policy prob-

lem; (2) the relevant policy options and their likely consequences; (3) theidentities of the politically relevant parties and their interests; and (4) thelikely political reactions of each group to each policy option. In fact, withuncertainties about the technical, economic, and political aspects of a pol-icy, ex ante specification of the best policy outcome is not possible. Admin-istrative procedures, however, can be used to guide agencies to make deci-sions that are broadly consistent with the policy preferences of politicalprincipals. If so, "flexibility" n "vague mandates"is, in fact, more apparentthan real, for it will necessarily be accompanied by more procedural con-trols to assure compliance.6 To illustrate, consider the procedures govern-ing rulemaking.The constraints of due process imposed by the APA and the courts areprimarily procedural. Courts ensure that agency actions are neither "arbi-trary"nor "capricious."The requirements are as follows:1. The agency cannot announce a new policy without warning, hutmust instead give "notice" that it will consider an issue, and do so with-out prejudice or bias in favor or any particularaction.2. Agencies must solicit "comments" and allow all interested parties

to communicate their views.3. Agencies must allow "participation"n the decisionmakingprocess,with the extent often mandated by the organic statute creating theagency as well as by the courts (see McCubbins and Page, 1987). If6. NMashaw's19851) criti(queof scholarswho argueforstrict interpretationof the Constitu-tional prohibition against legislative delegation is instructive in illuminating this point. Non-delegation is defended on the grounds that it prevents the "runawaybureaucracy" or "irontriangle whereby policy drifts fi-om one that coul(l be sustained b a legislative majority.Mlashawmakes he observation that, under con(litions of uncertainty, m1orelexible dlelegatiocan produce a normativel' superior policy outcome fioml the perspective of either welfalreeconomics or individual rights. Our additional observations are two: (1) political actors know avague ldelegation when they create one and thus must be doiing so fo-rrational reasons, takingintto accounti the possibility of attenluated political control, and (2) in any event, vague delega-tion cdoes not necessarily mean loss of control, ts long as it is accompanied 1b a combinationof monitoring systems, reward structuries, and deck-stacking administrative procedures thatcreate incentives for the agency that mirrlor the incentives acting upon elected officials.


17/36


hearings are held, then parties may be allowed to bring forth testimonyand evidence and often to cross-examine other witnesses.4. Agencies must deal explicitly with the evidence presented to themand provide a rationalizable link between the evidence and theirdecisions.

These requirements play an important role in governing informationcollection and dissemination by agencies. Their paramount political implica-tions are fivefold.First, they ensure that agencies cannot secretly conspire against electedofficialsby presenting them with a fait accompli, that is, a new policy withalready mobilized supporters. Rather, the agency must announce its inten-

tions to consider an issue well in advance of any decision.Second, agencies must solicit valuable political information. The noticeand comment provisions assure that the agency learns who are the relevantpolitical interests to the decision and something about the political costsand benefits associated with various actions. That participation is not uni-versal (and may even be stacked) does not entail political costs. Diffusegroups who do not participate, even when their interests are at stake, aremuch less likely to become an electoral force in comparisonwith those thatdo participate.

Third, the entire proceeding is public and the rules against ex partecontact protect against secret deals between the agency and some constitu-ency it might seek to mobilize against Congress or the president.Fourth, the entire sequence of decisionmaking-notice, comment, de-liberation, collection of evidence, and construction of a record in favorof achosen action-afford numerous opportunities for political principals to re-spond when an agency seeks to move in a direction that officials do notlike.7 These procedures also ensure that relevant political information isavailable to form the basis of such action. Neither Congress nor the presi-dent need first undertake costly collection of this information, nor needthey contend with an agency which has substantialprivate information.Animportant consequence is that this allows political reaction prior to agencychoice, and prior to the agency's ability to mobilize a constituency. Thestrategic advantage of agencies is therefore limited.Fifth, administrativeparticipationalso worksas a gauge of political inter-est and controversiality(Noll, 1971). In administrativeprocesses with broadrights of standing and relatively harsh evidentiary standards pertaining tothe agency's basis for its decisions, interested parties have an incentive toburden the record with voluminous evidence supporting a decision favor-able to their interests. But marshalling this evidence and its supporting

7. Even though this may mean, as Ferejohn, McCubbins and Page (1987), and Noll(1971) all argue, that no matter what the agency does, some political principal will bedispleased.


18/36


legal argument is expensive, so that parties will face a tradeoff between thelikely effect of more evidence on the agency's decision and the costs ofsubmitting it. In noncontentious proceedings, a party need not participateintensely to affect outcomes favorably;however, in highly controversialpol-icy decisions, intense participation is an absolute necessity to prevent acatastrophic outcome (assuming that an interest's position is in some mea-sure defensible). Hence, demanding procedural requirements (including a"hard-look" udicial review) have the political side benefit of selectivelycausing the most politically contentious issues-and the ones in which po-litical overseers would be most concerned about the distributive aspects ofthe decision-to generate the most complete information,as well as to pro-vide substantial advance warning about the likely decision that, in the ab-sence of political intervention, the agency is most likely to make.2.3. PUBLIC DISCLOSUREREQUIREMENTSIn the mid-1960s, the Freedom of Information Act (FOIA) was added tothe APA. FOIA limits the ability of an agency to impose a change in policywithout warning by requiring that, with minor exceptions, all records bepublicly available.8The APA also requires that an agency "makeavailable"a record of the final vote of each member in every proceeding (sect. 552 (q)(4)). The Government in the Sunshine Act (GITSA)plays a similar role bylimiting how much of the decisionmaking process can remain hidden.These acts enable interested parties to learn about any attempt by theagency to develop a new constituency or to change policy while it is still onthe drawing board. This disadvantagesthe agency by making political inter-vention possible much earlier in the policymaking process, before theagency can mobilize a new constituency (Spitzer). FOIA contains anotherprocedural device to make these requirements binding: it reversed theburden of proof in FOIA disputes and made it almost impossible for theagency to win a FOIA case in court. The agency must prove that it neednot release the information, and it must do so under extremely short timeconstraints. This causes agency personnel to face sanctions if they do notdisclose informationbut no corresponding threat if it is released (Spitzer).Thus, procedures imposed under the APA, as amended, reduce anagency's information advantage over its political sponsors. But this has afurther, critically important implication. These procedures greatly increasethe efficacy of ex post sanctions. Because they can now be more readilytied to specific decisions, attempts by an agency to alter policy (either onits own, or in response to outside interference) can be spotted earlier byconstituents, giving political officials more time to impose sanctions on er-

8. There are importantexceptions, of course, such as the imminent hazardclauses underFDA or CPSC.


19/36

260 / JOURNAL OF LAW, ECONOMICS, ANI) ORGANIZATION 111:2, 1987rant bureaucrats. Incentive effects of swifter and more accurately appliedsanctions cause bureaucratsto be more reluctant to deviate from the policypreferences of political overseers. Moreover, the enhanced efficacy and,hence, heightened incentive effect of sanctions is achieved without the at-tention of political officials.

2.4. EVIDENTIARYTANDARDSRules and standardsof evidence in administrative law serve another impor-tant political function. The key decision is to select the stringency of theevidentiary standard that an agency must satisfy to make its decisions with-stand a court appeal. Weak standardsgive agencies more flexibility. If po-litical principals believe that agencies will otherwise comply with theirwishes, a weak standard is preferable, for it minimizes the chance that afavored policy will be precluded by the evidence.For the most part, administrative agencies face weak rules of evidence.In the 1930s there was no universal evidentiary standardrequiring agenciesto present evidence in support of their decisions. The APA extended thesubstantial evidence requirement to all federal agencies, thereby limitingabsolute discretion.9 The substantial-evidence requirement that "findingsof fact, if supported by substantial evidence shall be conclusive," togetherwith the retreat from substantive due process and the presumption ofagency expertise, are normally regarded as limiting the ways in which thecourt can scrutinize the political judgments and choice of agencies. Butthey are an essential ingredient of political control as well. The flow ofprogram benefits to constituents would be considerably more variable anduncertain if political actors could not avail themselves of this protectionagainst the intrusion of the courts. The rather weak requirement of sub-stantial evidence can be viewed not solely as providing agencies with inde-pendence, but as giving agencies the flexibility to choose policies that re-flect the preferences of political overseers. Independence occurs only ifpolitical control through procedures is lacking.In some cases, legislation has sought to impose stricter evidentiary rulesfor agency decisions (as provided for the regulation of chemicals by theEPA under the Toxic Substances Control Act) or even to impose substan-tive policy outcomes (for example, new source performance standards).This may come about when conflicting interests among the political over-

9. Partof the motivation by members of Congress to iimpose his evidentiary requirementon agencies stems from the early New Deal agencies which in solme cases would not even re-veal during extensive hearings their decisionmaking criteria, evidence, or other inputs intotheir decisions. See, for example, the discussion of the Federal Emergency Relief Administra-tion in Wallis. Since these agencies were part of the president's coalition, this extension ap-pears to be an element of redistributionof power from the executive branch to the legislativebranch.


20/36

ADMINISTRATIVE PROCEDURES AND POLITICALCONTROL / 261seers creates conflicting pressures on the agency. The winning side in sucha conflict will impose tighter procedural constraints on the agency to makesure that its interests predominate (McCubbins, 1985).2.5. DECK-STACKINGBecause much of administrative law is derived from case law, the politicaladvantages of procedural requirements may at first appear to be the fortu-itous gift of the courts, a by-product of decisions made on the basis ofother, more noble purposes than enhancing the political control of agen-cies. But much of administrative law is not derived from the courts; rather,it is embodied in legislation and executive orders. Moreover, political ac-tors control the extent of representation of various interests in administra-tive processes. Through these decisions, political actors assure that the in-fluence accorded to different constituents is not random; indeed, bycontrolling the details of procedures and participation,political actors stackthe deck in favor of constituents who are the intended beneficiaries of thebargain struck by the coalition which created the agency. Because adminis-trative processes, once established, endure far into the future and may dealwith issues in which there is considerable uncertainty over key economicand technical phenomena, elected representatives can be expected to beunsure about the substantive details of their most desired policy, eventhough they are certain about who should benefit and how the costs shouldbe shared. In such a circumstance, political leaders could undertake tobecome sufficiently expert that they could fashion legislation that was richin substantive policy content, as is often the case in tax legislation. Alterna-tively, the organic statute can be vague in policy objectives, seeminglygiving an agency great policy discretion, but the administrativeprocess canbe designed to assure that the outcomes will be responsive to the con-stituents that the policy is intended to favor.For reasons developed by Olson and elaborated in Noll and Owen(1983), the resources available for representation in administrative pro-cesses vary systematically and predictably among interests for reasons otherthan their stakes in the issue. Some constituents are likely to be well repre-sented regardless of the cost and complexity of the processes that affectthem, and still others not at all. Moreover, among the less well-represented constituents there may also be considerable differences in theextent to which they are politically relevant in Congress or the WhiteHouse, owing to their participationin elections or because some are betterrepresented on relevant congressional committees.The tools available to political actors to control administrative outcomesthrough process, rather than through substantive guidance in legislation,are the procedural details, the relationship of the staff resources of anagency to its domain of authority, the amount of subsidy available to fi-


21/36

262 / JOURNALOF LAW,ECONOMICS,ANDORGANIZATION11:2, 1987nance participation by underrepresented interests, and resources devotedto participationby one agency in the processes of another (Noll, 1987). Allelse equal, elaborate procedures with stiff evidentiary burdens for decisionsand numerous opportunities for seeking judicial review before the final pol-icy decision is reached will benefit constituents that have considerable re-sources for representation. Coupled with no budget for sulsidizing otherrepresentation, or for independent staff analysis in the agency or in otheragencies that might participate in its proceedings, cumbersome proceduresexemplify deck-stacking in favor of well-organized, well-financed interests.The assignment of the burden of proof is another mechanism for deck-stacking. If cause-effect relationships are subject to considerable uncer-tainty, the assignment of the burden of proof determines the stringency ofthe policy decision by determining which side will be given the benefit ofthe doubt.

Deck-stacking enables political officials to cause the political environ-ment in which an agency operates to mirror the political fiorcesthat gaverise to the agency's legislative mandate long after the coalition behind thelegislation has disbanded. Thus, the agency is not free to manipulate policyby seeking to create a new coalition that supports its preferred policies.The agency may seek to develop a new clientele for its services, but suchactivity must be undertaken not only in full view of the members of theinitial coalition, but in an administrative process that is lesiglnedto fiavorthem. This increases the chance that policy clhange either will benefitthem, will be based on a comproniise with tlhem, or will reflect a dissipa-tion of their political relevance for other reasons outside the agency'spurview.The rules against ex parte communlication in the APA provide a moregeneral bias in favor of politicians. These rules prohilit inforimalcontactbetween private parties and the agency (Iiuringormalrulelmaking and(ad(ju-dication and impose sever sanctions if such contact takes place. The APAauthorizes the agency to take adverse action against the communicatingparty (sect. 557), and "section 556 makes clear that this includes an adversefinal decision in the proceeding" (Spitzer 76). But this does not imply allinformal contact between the parties and the agency is prohibited, onlydirect contact. Using politicians as intermediaries is still possille. Indeed,the rules practicallyguarantee that using politicians as brokers is the onlyform of informal negotiation that is permitted. As a conse(quence, the exparte rules are almost never invoked-the pulnishmlents re too severe if itis detected, and an effective sulstitute is available.

Spitzer indicates a variety of ways in which politicians can steer a clearpath through the rules against ex parte contacts. First, they may re(quest"status reports" about a particularproceeding. Second, they can indicatetheir judgment in a variety of ways, for example, b)vreacting directly to thestatus report, by speaking out against the agency at the Washinlgton Press


22/36

A)MINISTRATIVE ROCEDURES NI) POLITICAL ONTROL 263Clubi,by inserting items into the Congressional Record, and so forth. Thestatus report, when paired with pullic indication of preferences, providesthe appropriate signal to the agency. Unlike ex parte contacts with privateparties-especially those leing regulated-the courts are unlikely to setaside an agency action just because Congress or the executive has beencalling agency memlers (Spitzer: 81).From our standpoint, the major implications of these arguments are asfollows. First, the rules against ex parte contact (lo not prevent agenciesfrom informally negotiating with interested parties. Instead, they channelthis through other governlment officials. Second, it is obvious that the bro-kerage function is not e(lually available to all interested parties but is Ii-ased in f|avor of constitiuents with access to political officials. According tothe literature on Congress, this is precisely the type of congressional bro-kerage service that has come to be a major comlnponent of the "WashingtonEstablishment" (see Fiorina, 1977a; andiFiorilnaand Noll, 1978).2.6. DECENTRALIZED NFORCENIENT NI)TlE (C()RTSProce(dules will only have their desired effect if' theirire(luirements areenfiorced. If the constraints they impose arebinding, they will establish anautomatic control mechanism fior Con'gress that keeps the agency f'rochoosinlg undesirablleoutcomes. Moreover, they will do so with minimaleffort required on the part of politicians. Administrative procedures havethe advantage that their enfoircement s left to constituents, who file suit firviolations of prescrilbedprocedure, and to the courts.The courts thus play a key role in assuring political control. If the agencyviolates its procedures, judicial remedy must b1ehighly likely. If so, thecourts, and constituents who bring suit, guarantee compliance with proce-(dl1'al onstraints, which in turntguarantees that the agency choice will miir-ror political preferences without any need fior political oversight. Put an-other way, enfoircementof procedures is decenrtralizedn that enfiorcementdoes not depend on the action of political plrincipals.This lowers enforce-ment costs and preserves the influence of politicians without direct partici-pation or explicit knowledge on their part.2.7.. CANC(ING CIR'CUN1STANCES:ilE AUT(OPIIA)T,UNCTIONProcedural constraints have vet another advantage. Firomihe p)erspectiveof elected politicians, one potential problem with dlelegating policymakingautloriity is that the relevant political interests may cihange over tile. Con-se(luently, the political costs and benefits of a policy also change, and sodoes the optimal political choice. One effect of estal)lishing a bias in ,favorof participants in the process is that policy decisiols lma(le by the agencyevolve as tile compl)ositiolnof l)articilpating gmroups hanges. The procedures


23/36


imply that agencies respond to changes in their environment even if thepoliticians have not first spotted these changes. Moreover, this allows com-mittee members and agencies in their jurisdiction a degree of flexibility toadjust policies as political interests change without recourse to newlegislation.

3. DECK-STACKING IN SPECIFIC POLICY CONTEXTSThe APA covers all agencies. Yet, agencies differ as to their political char-acteristics. Each agency was the product of a different coalitionalalignment.As part of the bargain among principals to establish the agency, eachagency is intended to provide benefits to a specific set of constituents.Some agencies operate in a continuing state of conflict among the groupswith which they deal, while others do not. Agencies also differ accordingtothe extent to which there is uncertainty over the costs and benefits of alter-native actions. This includes both economic effects and the political conse-quences of agency decisions. Hence, we expect that proceduralconstraintswill vary among agencies to reflect these differences and, further, will stackthe deck in predictable directions. This section examines several enablingacts to illustrate how this is accomplished.3.1. ENFRANCHISING NEW INTERESTSIf participationby outside interests in administrative processes affects pol-icy decisions, policy can be made more responsive to a politically relevantconstituency by enhancing its role in agency procedures. The National En-vironmental Policy Act (NEPA) of 1969 provides an example of how thisworks. In the 1960s, environmental and conservation groups became sub-stantiallybetter organized and more relevant politically. Though some pro-gramswere created to benefit these new interests, on the whole they werenot represented in the decisionmaking processes of existing agencies. Mostagencies, and the congressional committees responsible for them, resistedefforts to change the interest group environment in which decisions weremade.Environmental and conservation groups, and the congressional commit-tees that represented them, sought to affect the programsof almost everyfederal agency. Agencies as diverse as the U.S. Army Corps of Engineers,the Atomic Energy Commission, and the Fish and Wildlife Service weremaking decisions that affected their interests. Because of the resistance tochange of congressional committees, the prospect was bleak for passinglegislation for each agency that would alter its procedural environment inways favorable to environmentalists. Sweeping procedural change that af-


24/36

ADMINISTRATIVE PROCEDURES AND POLITICAL CONTROL / 265fected the decisionmaking of every federal agency was an attractivealternative.

NEPA imposed procedures that re(quiredall agencies to file environ-mental impact statements on proposed projects. These requirements forceagencies to assess the environmental costs of their proposed activities. Theprocedures gave environmental actors a new, effective avenue of participa-tion in agency decisions and enabled participationat a much earlier junc-ture than previously had been possible (Melnick; Taylor). The require-ments of the act also provided environmental groups with an increasedability to press suits against agencies. The passage of NEPA, therefore,effectively mirrored the new political environment in agency proceedings.Because of its broad enfranchisement of environmental constituents,NEPA has had significant policy effects. Mazmanian and Nienaber provideexamples of successful environmental participation in public hearings bythe U.S. Army Corps of Engineers. In some instances, environmentalistsmanaged to torpedo plans for new construction projects. Similarly, Fere-john describes how the proceduralrequirements of NEPA changed decision-making at the corps. In evaluating projects, the corps was forced to seekout environmental interests. The result has been a significantchange in thetypes of projects that the corps proposed.Cohen provides another example. She shows that in nuclear licensingcases before the Atomic Energy Commission (AEC) and later the NuclearRegulatory Commission (NRC), environmental interventionists almostnever won. That is, almost all of their major contentions were denied bythe AEC-NRC. Nevertheless, environmentalists had a major impact onpolicy outcomes. Their participation forced the agency to open importantrulemaking proceedings and imposed significant delays in licensing cases.Indeed, Cohen estimates that the procedures imposed by NEPA added afull year to the consideration of construction permits at the AEC. Thesedelays significantly increased the time required to complete a nuclearpower facility, and the rulemaking proceedings led to more expensive re-quirements and regulatory reviews. This in turn raised the relative costs ofthe nuclear power option for public utilities and served the ends of theenvironmentalists by effectively stopping nuclear power development inthe United States.NEPA and other environmental, health, and safety acts of the 1970simpose substantial compliance and participation costs on business. Smallbusinesses, however, usually cannot afford to participate in agency deci-sions and were effectively disenfranchised. Congress responded by passing

the Regulatory Flexibility Act of 1980. Similar to NEPA in its implicationsand effect, the act requires analysis of rulemaking(but not adjudication)bypublic agencies on the costs to small business. The effect has been to en-franchise automatically the interests of small business in agency decision-


25/36

266 / JOURNALOF LAW,ECONOMICS,ANI ()OR(ANIZATION11:2, 1987making. This in turn has led to exemptions for small business in the re-quirements of many proposed regulations.3.2. SUBSIDIZEDREPRESENTATIONLike small businesses, consumers rarely have the resources to participatein agency decisionmaking and are effectively disenfranchised. With theemergence of several strong consumer lobbying organizations, and a grow-ing concern among constituents about consumer safety, many congressionalcommittees moved to enfranchise consumer interests in the decision rulesof the agencies they oversee. At one time, eleven agencies had authoriza-tions or appropriations for intervenor programs (FTC, National HighwayTraffic Safety Administration [NHTSA], Department of State, EPA, Fed-eral Energy Regulatory Commission [FERC], Consumer Product SafetyCommission [CPSC], National Oceanic and Atmospheric Authority, Foodand Drug Administration, Department of Agriculture, Civil AeronauticsBoard [CAB], and the NRC).Consider, for example, the case of the FTC. The Magnuson-MossAct of1974 created a public participationprogram. Consumer representation in-creased dramatically. From the start, business lobbied hard to stop finan-cial support for intervenors." Following a recession and a new antiregula-tory public sentiment, in 1980 Congress reduced the FTC's intervenorprogram in the commission's reauthorization act. Congress also placed re-strictions on how the FTC could allocate its funds. The FTC Reauthoriza-tion Act of 1980 required that a minimum of 25 percent of the public par-ticipation funds were to be set aside for small businesses.With the waning of the consumer movement, public participation pro-grams quickly ebbed. In the early 1980s, public participation programswere explicitly disallowed for NHTSA, FERC, CAB, and the NRC, whilethe offeror processes at the CPSC was dramatically altered (Sarasohn,1980). The introduction of these programs, and then their rapid demise,shows that agency decision processes can be periodically restructured tomirrorthe new political environment.

10. In the 1970s the FTC proposed new regulations in the following areas: used car sales,funeral industry practices, food advertising, advertising aimed at children, credit practices,retail installment contracts, garment labeling, vocation and home study schools, hearing aids,mobile homes, protein supplements, antacid advertising, health spas, eyeglasses, standard-setting by trade associations, home insulation, warranties, and the business conduct of doc-tors, dentists, and other professionals. On the FTC's efforts, see Schoenfeld; Singer (1974,1979); Wehr; and Wines. During this period, Congress sought to strengthen the FTC's regu-latory ability (Gardner),largely by expanding its substantive authorityand providingincreasedconsumer representation. See also Weingast and Moran (1983).11. See Singer (1979); Berlow; Sarasohn (1980, 1982); and a Congressional QuarterlyWeekly Report article entitled "House Curbs FTC: Senate Bill Advances," December 1, 1979.


26/36

ADMINISTRATIVEPROCEDURES AND POLITICAL CONTROL / 2673.3. AGENDA CONTROLAs was indicated by the Regulatory Flexibility Act of 1980 and the publicintervenor programs of the 1970s, it is often the case that groups withconflicting interests seek representation in agency decisionmaking pro-cesses. If these groups are represented in the coalitional majoritiescreatedto pass legislation, the design of the agency will mirrorthe conflict. This isevident in the organic legislation that created two of the major new socialregulatory agencies of the early 1970s, OSHA and CPSC (as well as theolder NLRB). Responding to the interests of workers and consumers, theoriginal legislative proposals sought to grant these agencies broad authorityto identify and control a wide variety of hazards according to their owncriteria. Both proposals placed the burden of regulation on business. Forthis reason, both proposals encountered strong opposition.In order to obtain passage, sponsors of the legislation imposed additionalproceduralconstraints, limiting the ability of the agencies to establish theirown agenda in rulemaking. OSHA, for example, does not have completefreedom to identify the health hazards that it seeks to regulate. Rather, itcan promulgate regulations only for health hazards first identified by theNational Institute of Occupational Safety and Health (NIOSH). NIOSH is aseparate agency located in another departlnent (HEW, now HHS, as op-posed to DOL), and is overseen by different sulcommittees in Congress.As anticipated, it has proven only partially cooperative with OSHA. Thishas significantly limited OSHA's ability to set its own regulatory agendaand, in particular, to regulate health hazards according to its own priorities.In the case of CPSC, a cumbersome "offeror" process was imposed, lim-iting both the colmmission'sability to select the consumer product hazardsit sought to regulate and its ability to write regulations in any particularcase. Under the Consumer Products Safety Act, the CPSC can take littleindependent action and is virtually dependent on organized interest groupsto write safety standards. The CPSC merely identifies which productsshould be considered for mandatory standards and which hazards associ-ated with these products are to be addressed. Originally, the writing of thesta(ndard was undertaken by outside contractors through the offeror pro-cess. The CPSC had insufficient funds to pay the costs of developing stan-dards, so offerors were always interested parties, either firms, trade associ-ations, labor groups, or consulller organizations. The CPSC acted only as aregulatory lbroker. This process proved so cumll)ersome that CPSC wrotefevw -egulations and failed to become a lmajor fahctorn regulating pr-oductsafetv. 12

12. In its first fetw years, CPSC generaally ook two-andl-one-halftimies longer than themaxilmumallowal)le time (330 days) i(dcertlhe act to complete a regtllation.


27/36

268 / JOURNAL OF LAW, ECONOMICS, ANI) ORGANIZATION 111:2, 1987The changes imposed in the 1981 reauthorizationof the Consumer Prod-uct Safety Act reveal fulrtherdeck-stacking in the CPSC's procedures. In

order to issue new rules, CPSC must now invite proposals for voluntarystandards from the industry to be regulated. If a feasible voluntarystandardis proposed, the CPSC must adopt it and end its own process. CPSC canproduce mandatory industry standards only if it finds that voluntary stan-dards are unlikely to reduce risk or would not result in compliance. Thisfinding, however, places the burden of proofon the commission to produce"substantialevidence" to support this conclusion. But Coligress so limitedCPSC's funds for this purpose that the commission cannot undertake theneeded investigations to participate adequately in most of its own proceed-ings. These changes make the commission even more reliant on the firmls tregulates, for now it cannot even choose anyone other than the industry towrite the standard. Additionally, the lackof research budget, together withthe deck-stacking, increases the risk that voluntary "safetv standards"arelikely to be collusive mechanisms, such as the intfamous anticompetitivedesign features once proposed by the bicycle induistry Cornell, Noll, andWeingast). :3Like the procedures adopted for OSLHA, ile 1981 CPSC reauthorizationintegrates the coimmission'sdecisionmlakingwith that of another agency.The CPSC may not issue a rule onichronic health hazardsuntil an expertadvisory panel nominated by the National Academy of Sciences has made areport on the available evidence.3.4. BURDEN ()F P()()FAnother way that procedulrescan stack the (leck il agency decisionmallkinglis by determining who bears the burden of lroof. The re(lliremllentsof'TSCA provide an example. The bill, as originallyproposed )y Senator Wil-liam Spong in 1971, was a Federal Food, Drlug, and Cosmletic Act(FFDCA) for chemnicals, n that it cointainedprovisions which would llhverequired safety tests and use regulations prior to intro(luction on themarket.The FFDCA and its amenidments had been supl)orted strongly bv con-sumers, professionals such as the America.nMedical Association, and theindustry itself because it created(entry barriers. Spong's original proposalwas only mildly supported bv environllentalists, who were mlolrecon-cerned with the clean air and water bills, but was strongly opposed bychemical manufaicturers nd users. Spolg, therlefore,had little hope but fora coinpromise bill. In fact, his provisions were replaced lbya section allow-

13. These exalmplesalso illustrate a princilpleartitculateld ! McCI(bt,biiis1985). (.reaterdegcreesof controversv in tiht legislature lead to botli broaldgIra;tts l authorityt tllt is. s0opeof delegation) coml)ined with greater c'trol through procedutralcoistraintis oII the aglency.These two cases illustrate this princil)e.


28/36

ADMINISTRATIVEROCEDURESAND POLITICAL ONTROL 269ing the production and distribution of a new chemical unless, during a briefnotification period, the EPA moved to promulgate a test rule or to ban orrestrict it.The certification provision of the FFDCA places the burden of proof onthe pharmaceutical manufacturers. Consequently, drugs are more expen-sive to introduce, thereby creating a disincentive for new product innova-tions. While the effect on consumers is ambiguous in that both good andbad drugs are inhibited, the law serves to enhance the value of existingdrugs to pharmaceutical companies by protecting them against entry.Meanwhile, TSCA places the burden on the EPA to prove that a chemicalis a risk to human health or the environment. Consequently, few chemicalsare regulated prior to marketing. This makes it difficult to protect againstthe introduction of toxic chemicals but benefits users of toxics by minimiz-ing the costs of regulation.The Airline Deregulation Act of 1978 provides another example. Prior toits passage, the CAB had interpreted the rate-setting provisions of the CivilAeronautics Act to the effect that air carriersproposing price changes wererequired to show that their proposed prices were neither "too high, nor toolow," but that they were "just, reasonable, and nondiscriminatory"(Breyer). Competitors of the filing air carrier merely had to petition theCAB, arguing that the proposed price change was discriminatory(indeed,that it discriminated against them). The board would then postpone theprice change pending a hearing. In the hearing, the burden of proof fell onthe carrierapplying for the rate change. This, of course, provided a benefitfor existing air carriers by inhibiting price competition.The Airline Deregulation Act, in seeking to increase competition in theairline industry, includes a provision that changed the burden of proof fromthe applying air carrier to those seeking injunctions against the pricechange. The amendment required the CAB to authorize a change unless itwas not consistent "with public convenience and necessity." This provisionbenefited new entrants and ultimately consumers.3.5. AUTOPILOTThe response of the FCC to the development of cable television illustrateshow the administrative process keeps an agency in compliance with thepreferences of political officialsdespite a vague mandate. The Communica-tions Act of 1934 (as well as its predecessor statutes) establishes the author-ity of the FCC to regulate in a classically vague, broad way. Indeed, soflexible is the statute that, although it was intended to regulate AM radio,it has successfully been used to assert FCC authority over FM radio, tele-vision broadcasting, cable television, and direct broadcasts from satellites.The cable television case is especially instructive. Obviously, it repre-sents the most interesting stretch of the initial legislative mandate, for ca-


29/36

270 / JOURNAL OF LAW, ECONOMICS, ANI) OR(ANIZATI(N 111:2, 1987

ble television systems are technically more similar to local telephone net-works than to over-the-air broadcasts. Indeed, in recent years cablesystems have begun to offer two-way telecommunications that compete atthe fringe of the telephone business. As with local telephones, states and/orlocalities would appear to be the naturaljurisdictions for cable regulation.No necessity to allocate scarce electromagnetic spectrum underpins FCCregulation. Instead, the basis of the FCC's assertion of jurisdiction, upheldin U.S. v. Southwestern Cable Co. (392 U.S. 157, 1968), was that uncon-trolled cable development could interfere with the FCC's policies regard-ing over-the-air television. Once jurisdiction had been established, theFCC's principal broadcast constituency-cominercial broadcasters-couldthen influence the development of regulatory policy toward cable.Cable television began as a means for delivering signals to localities withpoor reception, but eventually cable entrepreneurs discovered that cus-tomers in areas with good local reception nevertheless would pay foriaccessto still more television. Southwestern Cable in San Diego was one of thefirst cable operations in a majorcity to attract customers by importing tele-vision signals from other cities (in this case, Los Angeles). Broadcastersimmediately complained about this practice to the FCC, which then as-serted jurisdiction and declared a moratorium on fiurtherdevelopment ofcable systems that did more than bring cal)le to areas with poor receptionor little or no local broadcast outlets. During the moratorium, the FCCbegan a long investigation into appropriatepolicy toward cable, and sevenyears later made public through conlgressionaloversight hearings the gist ofits findings and proposals. Not only Congress )ut, a few weeks later, theWhite House (acting through the Office of Telecomlnunications Policy inthe Executive Office of the President) played a majorrole in altering theseproposals so as to reach a politically acceptable compromise among theinterested groups (Noll, Peck and McGowan, 1974). The result was a rathercomplex and highly restrictive set of rules on signal importation, programduplication, original cable programming, and carriageof local stations. Ca-ble development was slowed by at least a decade (first the moratorium,then the programming restrictions and requirements), largely to benefitthe television industry.Eventually, two events made cable television a politically significant in-dustry. First, a large enough market for restricted cable existed in smallcities, rural areas, and poor reception areas in large cities that a base ofsubscribers emerged that could support programsdirected largely (WTBS)or solely (HBO) at cable. Second, domestic communications satellites dra-matically reduced the cost of distributing programs to the thousands ofsmall cable systems that had been put in place despite the rules. By thelate 1970s, a cable industry with significant customers and importantcom-mercial backers (such as the motion picture industry and professionalsports) had finally emerged. The FCC first created a separate bureau for


30/36

ADMINISTRATIVE PROCEDURES AND POLITICAL CONTROL / 271dealing with cable (previously it had been handled by the Broadcast Bu-reau) and then proceeded to reopen the lquestionof cable regulation. By1980, this led to virtual deregulation of the industry.The history of cable regulation illustrates the autopilot properties of ad-ministrative procedures. First, in the early 1960s, the FCC provided amechanism whereby broadcasters could slow a threatening technology be-fore it really became a serious threat. Second, the process of investigatingthe relationship between cable and over-the-air television provided Con-gress and the president with the means for forging a political compromisewithout recourse to legislative intervention: the agency simply did what itspolitical overseers wanted, after the principalcommercial interests affectedby the issue had made their case. Third, when cable eventually became apotent political force, the FCC institutionalized its representation in theagency and largely overturned its previous policies, again without the ne-cessity for legislative intervention. Legislation was needed for only onecomponent of the last step-the establishment of the Copyright RoyaltyTribunal to collect l

McCubbins Admin Procedures as Inst of Poli Control

Documents