PE: Bureaucracy theory

We will be looking at the political process in an exogenous political environment. Policies are demanded for by citizens/voters and interest groups whereas it is supplied by delegates/representatives/politicians and public administration. Public administration is claimed to be motived by either rent-seeking or community-engaging.

A central question becomes how to measure the quality of government of public services. Four typical approaches are surveys, input-output comparison, efficiency frontier (how well/efficient do they perform compared to an optimum), and efficiency and effectiveness measures.

Economic Theory of Bureaucracy

A homo oeconomicus optimises his/her own utility. Whereas an entrepreneur simply optimises profit, a bureaucrat is forced to pursue self-interest within the institutional constraints often excluding economic profit. Max Weber (Weber, 2002) argues that a natural object of a bureaucrat is power. Russel (Russell, 2004) offers three subdivisions: direct physical power, rewards and punishment, and influence on opinion. Only under uncertainty a potential arises to exert the last type of power, whereas information creates the opportunity to actually do so.

Power allows for personal advantages and bureaucrats accrue non-monetary benefits, job security, jobs for relatives, additional benefits, work efforts, reputation, and more. To justify/hide these advantages, nonpecuniary goals of a bureaucrat become size of the bureaucracy, slack within the bureaucracy and risk-aversion.

To analyse bureaucracy usually a two actor environment is observed. A sponsor how delegates a task and a bureau that executes and delivers results. Issues arise due to conflicting interests (providing public good versus personal advantages) and information asymmetry (true cost is only known to the bureau).  Subsequently, measuring issues  create a monitoring problem. Usually, only the activity of a bureau can be observed, but not the output (e.g. national defence, education). As the sponsor is monopsonist buyer (on behalf of society) and the bureau is a monopolistic supplier (to circumvent wasteful duplication), efficiency is not required and information cannot be sourced from an alternative.

Observations in reality have shown that bureaucrats wages are unrelated to efficiency and that any form of performance dependent pay would be hard to measure.

Budget-maximising bureaucrat

Proposed by Niskanen (Niskanen, 1971) , it constitutes a simple model that assumes:

  1. Personal interests of bureaucrats are followed by maximising the budget
  2. The bureau has a monopoly position
  3. The cost function is not known by the sponsor
  4. The bureau can make all-or-nothing budget proposals

It follows that their budget B depends on the perceived output level Q (B=P(Q); B'>0; B''\leq 0) and that their costs C  depend on the output level Q (C=C(Q); C'>0; C''\geq 0).

A bureaucrat then has a Lagrangian objective function O_B=B(Q)+\lambda(B(Q)-C(Q)) where the Lagrangian multiplier \lambda represents the marginal utility of an expansion of the budget constraints to the bureau and hence is positive. When differentiated and solved for zero one obtains B'(Q)=\frac{\lambda}{1+\lambda}C'(Q) and B(Q)=C(Q). The optimal outcome for society would be B'(Q)=C'(Q).

The bureaucrat can know the sponsors social surplus and compute and request a budget that reduces the surplus to zero, unless B'\leq0. The additional condition explains a bureau’s infringement into other domains to justify increasing the budget.

Alternative institutional assumptions

Any of the four assumptions can be relaxed. Relaxing assumption 4, a sponsor could require more than one budget proposal with different levels of activities, thereby weakening the agenda-setting role of bureaucrats. Bureaucrats must announce the price P at which  it will supply  a level Q that will be subsequently set by the sponsor. A bureau now chooses P that maximises B. The demand elasticity \eta=\frac{P}{Q}\frac{dQ}{dP} can be used by the bureaucrat to choose the largest budget. Under the assumption of linear demand schedule, constant marginal costs, and known sponsor demand, a bureau will ask for a price P such that \eta=1 as long as this is higher  than its marginal costs.

Other assumptions can be challenged as well. Monitoring can be introduced (relaxing assumption 3) where investigative bodies sift through the expenses of public bodies and eventually unveil oversized budgets. Risk-aversion has been ignored so far. The sponsor could conceal its demand or a market for a service (i.e. competing bureaus) could be introduced.

Alternative behavioural assumptions

On the one hand, a slack-maximising bureaucrat wants to maximise x-inefficiency to increase his/her gain relative to the service provided. The minimally accepted service is provided at the indifference optimum below the social optimum such that a larger share of the budget can be acquired. On the other hand, a risk-averse bureaucrat would choose actions with the lowest potential penalties. Therefore, avoidance of action often occurs.

Other behavioural considerations that influence a bureaucrats behaviour can be crowding-out effects of intrinsic motivation by monitoring, social norms, availability bias (work only on salient risks), and lacking feedback-mechanisms.

Counter arguments

Promotions in bureaucracies are highly competitive and require good “track records” of the respective bureaucrat, so there is a market for bureaucrats. The discretionary power is actually lower than in the private sector and therefore inefficiency may not be properly judged. Lastly, (democratic) government are under (re-election) pressure and therefore are keen on monitoring the efficiency of a bureaucracy.

Power of the agenda setter

The agenda setter can prepare a choice where the non-acceptance of an oversized budget would result in the under-provision of a service. The ability to setup the agenda therefore results in power.

Control of the public sector

To reduce excesses in the public sector, usually, some form of competition is introduced (e.g. between administrative units, by private services), tightening of budget constraints (e.g. designated taxes (earmarking), limited tax base, and auditing), political restrictions (e.g. direct votes, separation of administration and politics), and rewards and punishment.


Niskanen, W. A. (1971). Bureaucracy and representative government. Transaction Publishers.
Russell, B. (2004). Power: A new social analysis. Psychology Press.
Weber, M. (2002). Wirtschaft und gesellschaft: Grundriss der verstehenden Soziologie. Mohr Siebeck.


IAP: Netheads and Bellheads

In the 1990s the great debates on how the Internet should be developed was coined the Netheads versus Bellheads. Netheads originated from the people that developed network technology whereas Bellheads originates from the Bell Laboratories – a research institution of telecommunication companies. At the core was a technical discussion whether packet-switching or circuit-switching is more useful and how big the meta-data overhead should be and how long the setup of a connection takes. However, technological development over the next decade rendered the debates irrelevant. In retrospect, it also seems to have been easier to expand the network under the internet with IP (just assign an address) whereas ATM (establish circuits). Also, the technical specifications of ATM turned very complicated to the point where they were too cumbersome.

The technical debates where just a superficial expression of an underlying discussion. The Netheads viewed the Internet from a data perspective whereas Bellheads viewed the Internet from a (continuous) signal perspective. Another issue was that Netheads came from a young industry with little to no corporate backing (market entrant)  whereas Bellheads had a century of corporate history behind their back (incumbent). The Netheads nearly waged a crusade.  It can be viewed through the US political focus as well where Netheads lean liberal and Bellheads lean conservative.