In another post John Short draws attention to the stonger focus on outcomes in Scotland. What's the right balance between outcomes and outputs in performance budgeting?
I find Norman Flynn's paper on outcome budgeting (commissioned by the Scottish Executive) very useful.
Amongst the principles he identifies are:
i. Outcomes should be defined, measured and assessed in all areas of government activity. However, because of a continuing interest in efficiency, outputs and unit costs will remain a part of the budget process.
ii. For all areas there should be defined a set of high level outcomes which are:
- Measurable and supported by good data and information systems
- Small in number and readily understandable
- Maintained over a long period...
iv. It should recognise that there will not necessarily be completely reliable links between all activities and all outcomes, but that activities that do not contribute to desired outcomes will be strongly questioned.
The last point is important and some would argue calls into question the practicality of 'budgeting by outcomes', at least in its purest sense, i.e., classifying and appropriating expenditure by outcomes. See for example Graham Scott's paper (also attached). The following selected quotes give a good flavour of Graham's arguments, which I find convincing:
‘In my observations the lessons from that experience have been that firstly, outcomes are a difficult concept of performance to make operational in many important areas of government and that outputs have made a comeback. A close examination of the actual performance specifications used in countries that are committed to managing for outcomes reveals that amongst the outcome indicators are smatterings of outputs, so called ‘intermediate outcomes’, process indicators and even inputs as a response to the practical difficulties of using outcomes as performance indicators…
Secondly the solutions to the problem of managing for outcomes are very demanding, and largely unattainable in low capacity governments.
…the state of the art appears to me to be that managing and budgeting for outcomes in practice occurs as special cases but the holy grail of applying it across a whole government is just that – an unattainable myth. Firstly, this is because there are a lot of activities that can be presented as contributing to multiple outcomes, which is not helpful for budgeting or performance management.
Secondly a lot of high level outcomes statements, and the objectives they derive from, are not of much operational value…
Thirdly, there are many important services provided for which the connection to an outcome is intuitively obvious, but incapable of attribution or measurement…
It is for this reason that I and my colleagues have found that most productive way to get started in building a performance management system is to identify the outputs or services being produced rather than begin with trying to define high level outcomes and associated objectives.
This alternative approach begins with high level outcome statements, then dissects them into contributory outcomes in several layers. It then allocates them to ministries and finally to subordinate service delivery units as performance targets. Then budgets are attached to them. This has some superficial intellectual appeal to some, although to me it seems like the ghost of twentieth century central planning – particularly when the outcomes encompass the private sector. More importantly in my experience it does not work as a methodology, because the outcomes are not mutually exclusive and the set of outcomes does not exhaust all the activities that government spends money on. So it is no use as a budget classification. What happens next is that designers invent outcomes for everything money is spent on, which reverses the desired causality between outcomes and budgets.’
