1. Unbounded rationality
Decision behavior of legislators has been modeled by Weingast, Shepsle and Johnsen (1981). Their model assumes utility maximizing legisla- tors and distinguishes political efficiency from economic efficiency by explicitly considering specific political benefit and political cost func- tions. The outcome of their model shows that the politically efficient amount of output exceeds the economically efficient amount. The model therefore offers an explanation based on utility maximizing leg- islative behavior for a public sector that is too large, that is, larger than the economically efficient size. Noteworthy in their model is the lack of any bureau through which the output is supplied. The legislative invest- ment decision is assumed to be brought fully to fruition. That is, there is literally no agency problem because there is no agency! Even in models of legislative behavior when there is a bureau presence, the will of legislators prevails (see Weingast and Moran (1983), and Weingast (1984), for example). Bureaus are shown to have no ability to behave independently or counter to legislative preferences.
Legislators make the investment decisions. However, as in the private sector, the supply activities that follow are carried out by individual public organizations of production: bureaus. Once the investment in the bureau has been made, the legislator hopes that the desired out- come will be achieved. The degree to which that happens depends on the bureau manager who decides how resources are allocated within the bureau.
That the bureau manager is a separate decision maker with his or her own interests in the process of public sector production has been well understood in the economics of organizations. Models of bureaucratic decision making have been developed, such as those of Niskanen (1971, 1975) and Migué and Bélangér (1974).4 These models focus almost exclusively on the bureau manager’s decision process. These models also demonstrate economic inefficiency through either overproduction of bureau services or optimal production of services at above minimum cost. The role of legislators in these models was either one of passive behavior due to bureau monopoly power (Niskanen, 1971) or limited monitoring ability or interest due to high political cost (Niskanen, 1975). In either case legislative presence was rendered irrelevant.
The implication of these models of bureau (or bureau manager) behav- ior are that (1) ownership and control of resources in the public sector are fully separated; (2) economic property rights effectively accrue to bureau managers; and (3) bureau managers increase their individual managerial utility by increasing bureau size. This is accomplished by expanding budget, which generates either too much output or too high cost, through, for example, excess staff or equipment. Thus the models offer alternative explanations of bureau managerial behavior that each result in a public sector that is too large.
Property rights clearly are a basis for these models. In these models political property rights drive legislative behavior and economic property rights drive bureau managerial behavior. Each set of the models just dis- cussed is unilateral, however: there is no effective bureaucratic presence or no bureaucratic presence at all in the legislative models; and there is no effective legislative presence in the bureau managerial models.
Some models of unbounded rational public sector decision behavior explicitly include both legislators and bureaucrats, as in Lindsay (1976) and, as noted earlier, Niskanen (1975), Weingast and Moran (1983), and Weingast (1984). Although these models have the appearance of modeling bilateral decision behavior they in fact are unilateral in their approach.
Lindsay focuses on monitoring by the US Congress of what he terms the visible attributes of bureau production. Visible attributes are those that are easily observed and measurable, such as patient-days at a pub- lic hospital. Invisible attributes are those characteristics of service that are less easily observed, such as the level of quality of patient care. To ensure budget approval in the Lindsay model bureaus respond by over- supplying visible attributes and undersupplying invisible attributes. The implication of this behavior is that bureau mangers behave as perfect agents to legislators. Bureau managers are subject to legislative interests and have little or no discretionary ability in their decisions regarding the allocation of bureau resources once they understand legislators’ preferences.
Both Weingast and Moran (1983) and Weingast (1984) demonstrate the strength of legislative preferences and legislative leverage in their models that focus on the US Federal Trade Commission and the US Securities and Exchange Commission, respectively. Indeed, Weingast’s model of ostensible principal–agent behavior in the public sector is similar to Lindsay’s in a number of ways. His theory is built on the assumptions that (1) bureaus depend on Congress (as firms do on con- sumers); (2) that high-demand legislative review committees act as a link between bureaus as producers and (vocal) interest groups as consumers; and (3) constituents (voters) are effective monitors. These circumstances effectively reduce the requirement for monitoring by legislators. The nature of the political system mitigates problems due to agency shirk- ing. The implication then of these models is that there is no agency problem. Bureau managers make decisions on resource allocation that fully carry out the intent and preferences of legislators.
Niskanen, of course, modeled the opposite situation, where all the leverage was positioned in the bureau manager rather than in the legislator. In response to criticism of the unilateral approach of his orig- inal (1971) model, Niskanen developed a model of bureau managers and politicians that effectively limits any legislative monitoring because the cost of doing so is too high. The resulting model is still, therefore, in the unilateral spirit. This is clear from the outcome, which is the same as that of his original unilateral model of bureaucratic behavior under conditions of purely passive legislators.
The unilateral models of either bureau managerial behavior or leg- islative behavior have been challenged by Moe (1984, 1989, 1990) on political grounds and by Fiorina and Noll (1978), Spencer (1982) and Carroll (1989, 1990, 1993a,b) on economic grounds. Moe theoretically develops the political rationale of legislators in the design and structure of bureaus. Although his model emphasizes the role of interest groups on legislative decisions, he explicitly considers the differences in objec- tives of legislators and bureau managers in the process of structural choice. It is the presence of these different sets of objectives that results in a structural choice that is inefficient.
Fiorina and Noll (1978) model electoral competition from separate districts and bureau managers who have an expense preference for bureaucracy. Their model is similar in effect to the Williamson (1963) managerial model of a for-profit corporation where the manager has an expense preference for staff. The Fiorina and Noll model, however, incorporates the political preferences of legislators for influence. Thus in their model, the combined effects of increased demand for publicly provided goods, legislative preference for political influence, and a monopoly bureau manager’s preference for size result in larger than optimal bureaus but less than optimal output.
Spencer (1982) develops a principal–agent model of legislative fund- ing and bureau managerial behavior with the additional constraint that replacing a bureau manager is costly. Her model is based on asymmetric information where the bureau manager knows more than the funding legislator-principal about the bureau’s production function. Her model predicts a lower budget and more efficiency due to less organizational slack than a Niskanen bureau would supply under competitive condi- tions. Under conditions of bureau monopoly, managerial replacement is very costly so that the constraint becomes nonbinding. This has the effect of limiting the monitoring ability of the legislator.
Carroll (1989, 1993b) empirically demonstrates the lack of bureau monopoly power that is the basis for the unilateral models of bureau managerial decision behavior and the Fiorina and Noll model. She uses market share data to show that bureaus have less power than had been assumed and are subject to the interests and preferences of legislators to some degree. When objectives of legislators and bureau mangers are explicitly modeled simultaneously three outcomes can be demonstrated (Spencer, 1982 and Carroll 1990, 1993a). First, the results of single objective unilateral models are altered by the explicit consideration of simultaneous multiple objectives. Second, the level of politically effi- cient public sector output is closer to the economically efficient levels than is the level of services produced in accordance with bureau mana- gerial preferences. This is especially the case when no amount of budget residual over cost may be appropriated by the bureau manager. Third, monitoring of bureau managers is necessary for legislators to achieve political efficiency. When legislative objectives act as binding constraints on bureau managers, the amount of services provided falls below the bureau manager’s maximum preferred level. This is particularly true if the bureau manager is unable to appropriate any budget residual for personal use within the bureau, such as for nicer office furniture. This would therefore be a situation where larger amounts of bureau output increase bureau managerial satisfaction.
Outcomes of bilateral models of bureau managerial and legislative behavior depend theoretically on assumptions made regarding moni- toring by legislators and empirically on the nature of bureau output that determines the relative weights on the two sets of objectives. When leg- islative monitoring is assumed to be costly and outputs are difficult to measure, as in educational services or national defense, then bureau managerial objectives carry greater weight in the decision process. These conditions prevail in the Niskanen-type models. When legislative mon- itoring is assumed to be costless (or low cost) and outputs are easily measured then legislative objectives carry greater weight. These condi- tions prevail in Weingast-type models (costless constituent monitoring and easily measurable SEC cases, for example) and in the Lindsay model (visible attributes that are costless to monitor).
What does this indicate regarding the issue of separation of owner- ship and control in the public sector? First, the literature implicitly assumes a lack of ownership and thus implies that all residuals are appropriated by those who have control rights. The common outcome of the literature that is unilateral, that is, focused either on only legisla- tive behavior or on only bureau managerial behavior, is that the public sector is inefficient. The implication is that the inefficiency is a direct result of the lack of effective ownership.
Second, the notion of legislators as having residual rights, particularly residual political rights, can be detected in some of the Weingast-type models, but not in the usual way. In effect, Weingast, Shepsle, and Johnsen (1981), Weingast and Moran (1983), and Weingast (1984) assign legislators the Barzel-type economic property rights. They do this by granting legislators control rights but not residual rights directly. These are presumably the rights of citizens who are so far out of the pic- ture that legislators have a free hand in the control of public sector resources for their own benefit. Legislators are mindful of citizens inter- ests so that they get reelected, but they act according to the economic property rights granted them by the indifference of citizens. It is the expropriation of citizens’ residual rights by legislators who exercise their economic property rights that leads to the inefficiency in the public sector in these models.
Third, the notion of legislators as having residual (political) rights may be detected to a lesser degree in the Niskanen (1975) and Lindsay (1976) models, as well as the Breton and Wintrobe (1975) critique of the original Niskanen (1971) model. These models view legislators as mon- itors of bureau managers. Why do legislators monitor? They must have some residual right to motivate them. The legislators in these models essentially are cast in the role similar to corporate shareholders.5
2. Bounded rationality
The economic analysis of public sector decision making in bureau supply of services just discussed is based on optimization models, where legislators and bureau managers maximize their utility. These are mod- els of unbounded rational decision behavior. Limitations on informa- tion process associated with bounded rationality apply to the public sector decision process, however. Indeed, Herbert Simon’s observation of a public sector (local government) decision process concerning a children’s playground served as a basis for his development of the theory of bounded rationality, as he described during his acceptance of the Nobel Prize in Economics (1979). In addition, the separation of ownership and control in the public sector, like that in the private sector, is based on the problems of identifying and processing informa- tion so that residual rights can be clearly understood and adequately enforced. Thus, separation of ownership and control is an issue of bounded rationality.
The demand for bureau services is articulated by legislators. Legislators are representatives of their taxpaying constituents and may also be viewed as investors in bureaus. Under conditions of bounded rationality and limitations on information processing, legislators do not know the true interests of their numerous constituents. Even if they could get this information from each constituent, the diversity of interests of each individual across a political district presents a difficult problem facing the legislator of how to weight these interests. Political action committees and interest groups are one mechanism that has developed as a result of this limitation. Becker (1983) has shown that the ignorance of voters gives rise to the influence of political pressure groups.
In addition, legislators have limited information on the supply capa- bilities (that is, the production function) of each bureau they have to review for the appropriation process. Even if they could get the techni- cal production information from each bureau, the ability to fully under- stand and process this information is limited. Self selection by legislators onto committees that review bureaus who produce services most related to their constituency where they have a high demand has been one response to these limitations (Mackay and Weaver, 1979). The practice of logrolling, that is, vote trading by legislators on bureau appropria- tions is another mechanism that has developed in the face of limited information processing.
Bureau managers are also faced with limitations on information processing. A bureau manager will not know or fully understand the demand function for its services or even its production function, that is, how things actually get accomplished through the various levels of his or her bureau. The hierarchical structure of the ‘bureaucracy’ results in information loss through the levels of the hierarchy to the decision maker, that is, the bureau manager. In addition, the tenure granted to civil service employment in the public sector may promote shirking, something that further limits understanding of the actual production function of the bureau by its manager.
For these reasons the legislator as investor and the bureau manager are likely to be subject to bounded rationality. The implication of bounded rationality on the part of both legislators and bureau managers is that they will narrowly frame their decision problem (Payne, 1997). For legislators, this means that their articulated preferences will reflect those interests which can be more easily determined, even if these do not accurately reflect constituent interests. In addition, Spencer (1982) predicts ‘slack to be greater in those areas of operation which are harder for an outsider to understand’ (p. 199). Thus, bounded rational decision behavior on the part of legislators predicts a greater tendency by them to pay heed to political action committees and special interest groups. This occurs not only because these groups provide sources of funds, as would be predicted under conditions of unbounded rationality, but because these groups solve (to some extent) the legislator’s information processing problem.
For bureau managers, narrow framing supports their tendency to focus more on visible outputs when this is possible, the behavior pro- posed by Lindsay (1976). Focusing on the number of hospital beds or patient-days, or the number and magnitude of weapons systems, are examples of this type of narrow framing behavior consistent with bounded rationality. Such behavior not only reduces the monitoring problem for legislators as he proposed, but it also reduces the monitor- ing problem that is inevitable for bureau managers in a complex organ- ization. Both narrow framing and risk reduction can also be seen in the structural and institutional design of public sector organizations through rigid and detailed rules at each organizational level. The step/grade promotion system that reduces the autonomy of decision making is an example of this.
The effect of bounded rationality on organizational design limits the effective separation of ownership and control in the public sector by lim- iting the ability of bureau managers to use their control rights to expro- priate the residual rights of legislators. The rigidity of the bureau’s organizational and institutional structure reduces the economic property rights of the highest level bureau managers. This effect can be reduced as tenure of the bureau manager increases and as the operation of the organization (that is, its production function) becomes better under- stood (Payne, 1997, p. 356). But this mitigation of reduced economic property rights is not likely because these highest level positions are most often political appointments with limited tenure, and is particu- larly true when term limits of the appointing official exist, such as for the US president, prime ministers, and governors.
Thus, under conditions of bounded rationality, outcomes are likely to be inefficient because social preferences are less likely to be met unless special interests and constituent preferences coincide. Becker (1983) shows that this is likely to be the case, suggesting a more efficient out- come in the presence of bounded rationality. The situation of bounded rationality puts into play procedures that, along with competition among special interest groups, serve to mitigate some of these inefficiencies.
Source: Carroll Kathleen A. (2004), Property Rights and Managerial Decisions in For-Profit, Nonprofit, and Public Organizations: Comparative Theory and Policy, Palgrave Macmillan; 2004th edition.