Units of output are sometimes defined in terms of the achievement of some pre-defined outcome (for example, a specified level of educational achievement), or alternatively in terms of some quality standard interpreted as a ‘conformance to specifications’ activity test. For most public-sector outputs, these definitions of a unit of output are flawed and may have undesirable behavioral consequences. Output measures cannot, in general, do double duty as outcome measures. Outcomes need to be measured separately. Moreover, the activity content of many types of outputs may legitimately vary both over time (as a result of qualitative rationing arising from the budget constraint), and also between clients (as a result of tailoring to varying client needs). Only for a sub-set of services is it appropriate to define a unit of output as complete only when either a specified proximate outcome has been achieved, or alternatively when a pre-defined minimum set of activities has been carried out.
