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

5.3 Wilson’s typology of organisations 
Wilson defines an output as a performed task which is observable or unobservable depending on 
whether work is programmed and if it is supervised. Outcomes refer to achieving the goals of the 
programme. Observability may be hampered by ambiguity about the mandate of the agency or 
uncertainty about the impact of the agency on targeted problems. For example, there may be a time 
lag between the output and identification of the outcome, thus making a cause and effect link 
difficult to prove. 
Table 7 offers a matrix contrasting the observability of two variables – outputs and outcomes – and 
identifies (following Petrie) a range of managerial approaches contingent to the type of 
organisation. Results based management regimes can be seen as an example of application of 
enforceable contracts which may be appropriate to production and craft organisations. In the case 
of production organisations, where the observability of both outputs and outcomes is high, it may be 
possible to invite competitive bids from external agencies to improve productive efficiency.
However, according to Premchand, where the actions of agents are observable, then the principal 
directly monitors the agent. “Only when there is a distance – in the literal and policy sense – 
between the principal and the agent does principal-agent theory have to be applied” (Premchand, 
1993, pp 86-7). In the case of craft organisations, the use of an internal contracting may ensure 
compliance with desired outcomes. According to Petrie, an enforceable contract has the potential to 
result in a number of advantages: 
•
greater ‘buy in’ from the agent and therefore stronger accountability; 
•
less scope for ambiguity or uncertainty over exactly what is required to be delivered by each 
party; 
•
a greater likelihood that performance under the contract will be monitored; and
•
a greater likelihood that conflicting roles will be highlighted, and fully contractible functions 
identified and contracted out to competing suppliers. (Petrie, 2002, p 138) 
However, he acknowledges that this approach could impose additional transaction costs of hiring 
lawyers, loss of trust and motivation, and reduced flexibility. Premchand is sceptical of the 
applicability of principal-agent theory to public expenditure management: although the available 
literature provides insights into the workings of relatively simple organisations, applying them to 
government as a whole remains problematic (Premchand, 1993, p 88). 
It should be recognised that there is disagreement amongst those writing on principal-agent theory 
about the appropriate situations in which contracting to an agent should be invoked. Molander et al,
disagreeing with Premchand, argue that “P/A relationships based on outsourcing and contracts 
should be expected when outputs are relatively easy to monitor, whereas in-house production by 
employees should be expected in the reverse case” (Molander et al, 2002, p 35). Their argument is 
that where outputs are difficult to specify, then principals should revert to input control rather than 
attempting to measure the extent to which agents are achieving outcomes. It is significant to note 
that Wilson identified education as his example of where both outputs and outcomes are 
unobservable. Since Wilson published his work in 1989, considerable amounts of work have been 

---

39
undertaken to develop performance measurement techniques in education, as political masters have 
demanded greater impacts from education expenditure. The case of the Maori education strategy in 
New Zealand which was discussed earlier is a good illustration of current practice.