In The American Economic Review, Vol 68, No. 2 (May 1978): pp. 256-260.

Voters, Legislators and Bureaucracy: Institutional Design in the Public Sector

By MORRIS P. FIORl!'IA and RoGER G. NOLL*

The purpose of this paper is to outline a sidization, and expenditures on the produc­ theory of representative democracy which tion of public goods. Third, government bu­ explains why rational actors construct an reaucracies, in carrying out public policies, excessively bureaucratized government. impose costs on citizens by ensnarling them We define excessive bureaucratization as in red tape. the selection of an inefficient production We assume that government activities technology for the public sector, can be characterized by a production func­ characterized by relative factor proportions tion, the arguments of which can be use­ that entail more bureaucracy than the pro­ fully classifiedinto bureaucratic and nonbu­ portions that would minimize total costs. reaucratic inputs. Bureaucratic activities Thus, the question of excessive bu­ include keeping formal records, developing reaucracy is related to but conceptually dif­ and enfordng procedures to govern rela­ ferent from whether a particular policy is tions between the bureau and its clients, worthwhile. Furthermore, it presumes a communicating among parts' of the organi­ concern ... more fundamental than the zation, and controlling and evaluating observation that implementing a public personnel. These activities impose an policy inevitably requires the expenditure external cost on citizens because their com­ of scarce resources. plexity creates an informational problem Section I describes a theory formally .. for citizens who seek services, because the presented in the authors' forthcoming data required by the bureaucracy come in article. Section II develops the predictions part from the agency's clients who incur of the theory, most of which have not been . some expense providing them, and because tested. Section III outlines reforms that bureaucratic processes are time consum- might und o some of the effects that the ing. theory predicts. A legislator serves the home consti­ tuency in various ways (see Fiorina, pp. I. The Theory: A Voter's Dilemma 41-49). Legislators collectively decide general issues of public policy by majority We assume that in choosing among al­ rule votes. (We assume that tlJ,e distribution ternative political actions, voters, bu­ of voter preferences supports a majority reaucrats, and politicians pursue their self­ rule equilibrium. While unnecessary, this interests. For voters, this means casting assumption simplifies our argument.) Each votes in a manner that maximizes expee.ted legislator also is a near-monopolistic sup­ utility, given the platforms of compe't�ng plier of unpriced facilitation services to candidates. For bureaucrats, this means constituents. Facilitation services take maximizing some measure of the size of the several forms: intervening in bureaucratic bureaucracy. For politicians, this means processes to aid citizens ensnarled in red maximizing the probability of electioq. tape, providing information to citizens who Politicians can affect the welfare of indi­ want to know how and where to approach vidual voters in three ways. First, some of the bureaucracy, and acquiring for the arguments of utility functions are constituents a share of "d istributive" government activities. Second, government activities. A government activity is dis­ redistributes income through taxation, sub- tributive if it is divisible into subactivities, each of which is evaluated and decided *Califot�ia Institute of Technology. upon separately and is beneficial to a rela-

256 VOL.68 NO. 2 DECENTRALIZATION, BUREAUCRACY. AND GOVERNMENT 257 tively small proportion of the electorate. fective incumbent, and consequently Examples are federal construction projects, provides a lower expected payoff in this categorical grant programs, and com­ role. This tendency will be accentuated if a modity-specifictariffs. legislator becomes a more effective facilita­ A bureaucracy can assist a legislator in tor over time. Because part of facilitation is carrying out facilitation. It can accom­ the possession and use of information modate inquiries by the legislator on behalf which is acquired through experience, and of constituents by providing information because seniority enhances the influenceof about certain services or by expediting a a legislator in determining the fate of an decision. It can propose, and try to justify, agency, incumbents can be more effective distributive activities in a legislator's home facilitators than their challengers. district. A rational bureaucrat will use these possibilities to serve the objectives of the II. Applications: Evidence and Predictions bureau by rewarding legislators who sup-· port its programs and appropriations. Di rect observation of the production If a bureaucracy responds favorably to function for a government activity is espe­ facilitation activities, it makes legislators cially difficult, and a test of the primary im­ more attractive to their home consti­ plication of the theory-that public tuencies. Effective facilitation lowers the activities are excessively bureaucratized­ external costs of bureaucracy and raises the is beyond us at present. We offer some in­ share of government distributive activities direct evidence, and pres�11tother predic­ to the constituency. The latter is attractive tions of the theory. because the taxes used to finance a project in one district are imposed on everyone, A. Facilitation and whereas the benefits are concentrated. Congressional £/ecrions Moreover, performance as a facilitator de­ pends on the personal actions of the legisla­ During the early postwar period tor, enabling a legislator to claim credit for congressional elections were low in­ it (see David Mayhew, pp. 52-59). In formation affairs in which most citizens contrast the public policy decisions of the voted according to traditional partisan legislature are unlikely to be affected by the affiliations that reflected a generalized vote of a single legislator. preference for one party (see Donald One consequence of the preceding argu­ Stokes and Warren Miller). Since the early ment is that legislators and bureaucrats l 960's, the impact of partisan affiliations on have an incentive to provide government congressional votes has been declining. services in an excessively bureaucratized The most important influence to take up the manner. To do so raises the demand for slack is incumbency. The '"incumbency ad­ facilitation services. The electoral process vantage'' now appears to be 5 to I 0 percent does not check this tendency because in House elections (see Robert Erikson; voters face a prisoner's dilemma in choos­ John Ferejohn) and a few points higher in ing among candidates. If voters disapprove the Senate (see Warren Kostroski). The in­ of excessive bureaucratization, electing a cumbency advantage is not a result of post­ legislator who attacks bureaucratic ineffi­ Wesberry redistrictings favorable to incum­ ciency will be unlikely to alter the outcome bents, nor to increased knowledge of in­ of a majority-rule legislature, but will cumbents-despite their greatly increased produce a less effective facilitator. advertising before a more educated, less As the public bureaucracy grows larger, partisan electorate (see Fiorina, ch. 3). the importance of the performance of facili­ Elsewhere (see the authors, 1977) we tation will grow, and a legislator who is a show that rational incumbents should base good facilitator will be increasingly likely to reelection efforts on facilitative activities be reelected. A challenger who is unproven rather than programmatic advocacy. This as a facilitator is a riskier choice than an ef- conclusion follows from the relatively un- 258 AMERICAN ECONOMIC ASSOCIATION MAY 1978 controversial nature of the former and the TABLE I-THE PERSONAL STAFFS OF CONGRESSMEN greater personal effectiveness of the legisla­ tor in facilitative activities. 1960 1967 1974

Effective facilitation requires resources, ·�-----·-·-�-- and in the past two decades these resources Total staff 2,344 3,276 5, 109 have increased dramatically. Congressional Percent assigned to district 14 26 34 employment is one example. Several sup­ Percent district offices open port organizations have been established only when congressman is home or after adjournment 29 II 2 (Legislative Reference Service, Office of Percent congressmen with Technology Assessment, Congressional multiple district offices 4 18 47 Budget Office). Committee staffs have expanded, and the personal staffs of Source: Annual Congressional Staff Directories, congressmen have grown from six to a compiled by Charles B. Brownson. baker's dozen (see Table l). The ostensible reason for these developments is the grow­ Great Society programs through Congress ing complexity of governance. was his willingness to use distributive Congressmen also argue that this growth politics to purchase congressional support. offsets an imbalance of expertise between In I 964, over one-third of all federal grant the executive and legislative branches. programs allocated funds by formula. Dur­ Perhaps, but another reason may be growth ing 1965 and 1966, the number of grant pro­ in facilitation activities. grams increased by 70 p.ercent, 'and only Few solid data exist, but congressmen about one-sixth of the new pr,ograms used have greatly expanded their district staff allocation formulas (see Advisory Com­ presence (see Table I), and presumably the m1ss10n on Intergovernmental Relations, staff in Peoria has less to do with legislation p. 151). than with facilitation. Moreover, rough Congress can encourage facilitation in estimates indicate that even the Wash­ ways other than distributive policy making. ington staff spends more than half its time For example, regulatory legislation can be on facilitation (see Fiorina, p. 59). Other so vague that an agency must make nu­ innovations that serve the facilitation role merous detailed decisions before imple­ are mobile district offices, allowances for menting it (see Theodore-Lowi). Later, computerized records about constituents, oversight subcommittees find it easier to and more paid trips home (representatives use budgetary review as a lever for affect­ had three in 1960 and thirty-two in 1978). ing regulations than to change a law. A second implication concerns the B. Policy Trends increasing acceptance of new programs after they are enacted. While legislative Three general implications about the proposals may be controversial, opposition construction and implementation of public will decline once the program is es- policy follow from the theoretical dis- , tablished. Before enactment no facilitation cussion in the first section. takes place. If constituents care about a First, because legislators profit from ef- proposed program, they evaluate fective facilitation and bureaucrats profit politicians according to their positions on from accommodating legislators, both have the issue. Once the program is established, an incentive to structure programs so that both supporters and opponents need facili- facilitation is important. A principal way to tation. A legislator's policy position be- do this is to inject distributive elements into comes less important than facilitative a program. Expenditure programs, for abilities. Die-hard legislators who oppose example, can employ project-by-project de- an agency unsuccessfully only succeed in cisions or automatic distribution according penalizing their districts, perhaps by overt to a fixed formula. One explanation for agency actions, but more likely by forego- Lyndon Jo�nson's wizardry in steering ing what would be their "due" under the VOL. 68 NO. 2 DECENTRALIZATION, BUREAUCRACY. AND GOVERNMENT 259

program (see Barry Weingast). Republicans The scope of the federal government has who vote en bloc against new social pro­ expanded during the past two decades, grams quietly go along with reauthorization creating greater opportunities for citizens (see David Stockman). We suspect that to profit from bureaucratically administered new programs gradually become altered in programs, and numerous occasions for the direction of distributive politics, which citizens to run afoul of bureaucratically would provide a further reason why pro­ promulgated regulations. An incumbent's grams that initially trigger major political experience and seniority are an important battles gradually become the object of a resource which disappears upon election of political consensus. a challenger. In this brave new world, A third policy implication is the loosen­ citizens have come to attach more im­ ing of policy ties between representatives portance to the facilitation activities of and their constituencies. Because facilita­ congressmen (see Roger Davidson). tion gives incumbents an electoral ad­ vantage, their policy positions are less im­ III. Prospects For Change portant in their constituents' evaluation of them. For any given incumbent advantage The foregoing theory of the legislative arising from facilitation, an incumbent may process does not work in quite the way that deviate from the position of the median constitutional theory postulates. Our voter by some amount and still expect to theory simply isolates the incentives facing defeat a challenger who adopts the median modern voters, bureaucrats and legislators, voter's ideal position (see the authors, incentives created by the institutions within forthcoming). which legislators and bureaucrats act. To The model does not imply that the change the system requires changing the amount of public goods provided by the institutions. Several possibilities are government is greater than would be the imaginable; none appear likely. From most case in the absence of facilitating activities. to least drastic they include: Because the public sector adopts inefficient I. Remove incumbents' facilitative production technologies, the amount of powers. This involves slashing staffs and public goods that the median voter desires removing the constitutional basis for will be less than if production were at congressional power over the bureaucracy, minimum cost; however, the incumbency making the latter more responsible to the advantage may offset this effect if the in­ president, who has a national constituency. cumbent's policy position moves in the di­ Of course, this possibility is fraught with rection of more public goods than the the dangers of the imperial presidency. median voter prefers. The incumbent has a 2. Change the electoral system for generalized incentive to move in this direc­ congressmen. If legislators increasingly are tion because a larger public sector implies a elected for facilitative efforts, legislators greater demand for facilitation and a cor­ are less responsive for their policy posi­ respondingly greater incumbency ad­ tions and no one is responsible for legisla­ vantage. Additionally, to move outside the tive policy. Altering this situation probably model for a moment, as the policy ties requires abandoning the single member dis­ between incumbents and their districts trict. Proportional representation makes a weaken, the former are increasingly at candidate's election depend on the liberty to cooperate with special interests percentage of the vote received by parties that desire some particular government nationally rather than each candidate's per­ activity or expenditure. sonal percentage in one district. This makes To summarize, we attribute the increase legislators more dependent on the policy in the incumbency advantage in congres­ position of the parties, but eliminates sional elections to the gradual transforma­ representation of particular constituencies. tion of congressmen from makers of na­ 3. Lessen incumbents' facilitation tional policy to ombudsmen and grantsmen. responsibilitil:'s. In other countries the 260 AMERICAN ECONOMIC ASSOCIATION MAY 1978

ombudsman role is performed by a special R. Erikson, "The Advantage of Incumbency

office, rather than by individual legislators. in Congressional Electio ns, " Polity, As rational actors American congressmen Spring 1971, 3, 395-405.

, " steadfastly resist suggestions to establish a ___ Malapportionment, Gerryman­ federal ombudsman. Even mild proposals dering, and Party Fortunes in Congres­ such as Henry Reuss's Administrative sional Elections," Amer. Polit. Sci. Rev., Counsel of the Congress receive quick Dec. 1972, 66, 1234-355. execution. British MPs, whose facilitation J. Ferejohn, "On the Decline in Competition powers and resources do not compare to in Congressional Elections," Amer. Polit. the American congressman's, approved Sci. Rev., Mar. 1977, 71, 166-76. the creation of the Parliamentary Com­ Morris Fiorina, Congress--Keystone of the missioner for Administration (ombudsman) Washington Esrab/ishment, New Haven only after they were made the communica­ 1977.

tion link between constituents and the new ___ and R. Noll, "Voters, Bureaucrats office. and Legislators: A Rational Choice Per­ 4. Cumulative policy failure. A long­ spective on the Growth of Bureaucracy ," run prospect for change is inherent in ever J. Pub/. Econ., forthcoming.

less efficientoli p cies that impose ever ___ and , "A Theory of the Con- heavier external costs on citizens. If the gressional Incumbency Advantage," soc.

situation deteriorates sufficiently, the sci . work. paper no. 158, California voter's prisoner's dilemma might be Instit. Technology, Apr. 1977. broken: the attempt to elect antibureau­ S. Huntington, "Congressional Responses to cratic legislators could become rational the Twentieth Century," in David Tru­

even if failure to elect a majority resulted in man , ed., The Congress and America's · losses to the districts which elect them. Future, New Jersey 1965, 5-31. Despite the preceding emphasis on pros­ W. Kostroski, "Party and Incumbency in pects for changing the legislative system, Post-War Senate Elections: Trends, Pat­ we hasten to emphasize that this system terns, and Models," Amer. Polit. Sci. has positive aspects. Bureaucracy is Rev., Dec. 1973, 67, 1213-34. permanent, and information about it is Theodore Lowi, The End of Liberalism, New valuable. Some argue that facilitation is the York 1969. only role modern legislators can perform David Mayhew, Congress: The Electoral Con­ well (see Samuel Huntington). To reach nection, New Haven 1974. judgments of the costs and benefits of legis­ D. Stockman, "The Social Pork Barrel." lative facilitation, we must recognize that Pub/. Int., Spring 1975, 39, 3-30. facilitation exists and include it in our D. Stokes and W. Miller, "Party Government analysis. and the Saliency of Congress," Pub/. Opinion Quart., Winter 1962, 26, 531--46. B. Weingast, "A Rational Choice Perspective REFERENCES on Congressional Norms," soc. sci. work. paper no. 142, California Instit. Tech­ R. Davidson, "Our Two Congresse� : Where nology, Oct. 1976. Have They Been? Where Are They Advisory Commission on Intergovernmental Rela· Going?," paper presented at the Southern tions, Fiscal Balance in the American Fed·

Political Science Association Meetings, era/ System, Vol . I, Washington, Oct. New Orleans 1977. 1967' 151.