Results-oriented Budget Practice in oecd countries odi working Papers 209

5.2 Contracting 
This section analyses the use of performance contracting as a mechanism for the delivery of public 
services. It examines how results based management is used in this context. 
New Zealand state official Petrie (2002) identified “an increased focus on results, in terms of 
efficiency, effectiveness and quality of service” as one of two elements in the shift to a more 
performance oriented culture. Deciding which contractual mechanisms to use is, according to 
Petrie, determined by government’s objective to maximise the advantages of delegation of control, 
net of agency costs, though a combination of ex ante incentive alignment and ex post governance.
Therefore design criteria should address the following question: what is the best mix of ex ante 
performance specification and the use of ex post instruments for a particular institutional setting, 
and for a specific type of activity or output (Petrie, 2002, p 141-42). 
Performance contracting is used in a wide variety of situations including: 
•
Agreements between minister and heads of agencies 
•
Agreement between agency heads and lower level staff 
•
Agreements between national government and sub-national government 
•
A contract between a government agency and an independent contractor. 
The simplest form of exchange is through a spot contract with immediate agreement on price, 
quantity and quality. In most cases, exchange is not conducted as a spot contract because time 
delays and the need for one party to invest resources in the other. Hence, non-spot contracts are 
used to ensure compliance and performance. Contracts have costs of negotiation, monitoring and 
enforcement, known as transaction costs. Petrie identifies two types of non-spot contracts:
•
Classic contract law, where parties agree formal legalistic documents in situations where mutual 
adjustment is unlikely. 
•
Relational contracts where parties have a mutual interest in an ongoing relationship. Trust levels 
are high, levels of specification are lower resulting in reduced transaction costs. 

---

37
•
A key question for decision makers is whether to conduct transactions within a hierarchy or, 
alternatively, to contract with an external agency. Petrie argues that hierarchical transactions 
may be preferable when 
—
“each party is highly dependent on the other in a long-term relationship; 
—
transactions are frequent and/or complex; 
—
measuring performance is difficult; or 
—
contracts need frequent changes.” (Petrie, 2002, pp 121-22) 
However, agents can use a range of devices to align incentives. In order to understand this, we need 
to examine the fundamentals of principal-agent theory. Principal-agent theory is based on a number 
of assumptions (based on Barzelay, 2001) 
•
actors behave rationally, seeking to maximise utility, reacting with economic rationality as 
opposed to using political criteria (Premchand, 1993); 
•
there are incentive contracts with a mathematical function linking incentives to reward for the 
agent;
•
there is agreement on what defines an output. For US academic Wilson (1989) an output is a 
performed task, though according to Barzelay (an academic based at the LSE), agency theorists 
define it differently; 
•
that outputs can be specified in terms of quality and that this can be observed – for Wilson this 
means that it can be jointly agreed by both principal and agent. Incentives can be diluted if 
outputs are ambiguously defined; 
•
agents are at risk if output varies according to factors other than their own efforts – i.e. 
independent variables. A rational agent will demand compensation for bearing such risks; 
•
rational principals will define outputs that correspond to public policy; 
•
output based contracts compensate for the lack of observation of agents’ actions. 
Taking account of the case of Sweden discussed above, a key issue concerns the capacity of the 
principal (in this case, the commissioning ministry) to monitor the efforts of the agency through a 
mix of ex post and ex ante controls. Several issues arise. First, there is whether the principal is 
motivated to monitor the activities of the agent. Second, monitoring the activities of agents 
involves substantial information costs. Ambiguity can arise where outputs and outcomes are 
difficult specify. Niskanen’s model of bureau maximisation holds that the agent will always be in a 
position to maximise its benefits because of information asymmetries. However, as Lane points 
out, Niskanen does not hold a monopoly on principal-agent theory. Lane argues that applying 
Williamson’s economic organisation theory suggests
“that a public bureau is a superior solution to the problems of exchange between government 
and those who provide public services when the cost of measuring service output is high, when 
investments in asset-specific skills and capital will make one private contractor and 
indispensable monopolist and when uncertainty makes it impossible to write long-term 
contingent contracts” (Lane, 1993, p 58). 
Barzelay (2001, p 117) draws attention to what Aucoin calls explicit contracts, which are actually 
performance plans not ‘schedules for reward of agents’. He adds “Aucoin needs to explain why the 
public service will be motivated to accomplish performance goals” (2001, p 117). 
Petrie raises the question of the complex relationship between outputs and outcomes, neither of 
which are observable in many situations. Indeed, as he points out, their unobservability often 

---

38
explains why these activities are located in the public sector. However, he lists a number of reasons 
why simple application of principal-agent theory to public services may be inappropriate. These 
include that the principal’s ability to act may be constrained by legislation and that an agent may 
have more than one principal. On the use of performance targets, he counsels against their use 
when outputs or outcomes are hard to specify ex ante or measure ex post and “where there are 
significant information asymmetries” (2002, p 134).