Spending cuts and public sector productivity
The coalition government plan to implement substantial cuts in spending on public services, with most departments having been asked to find budget cuts of 25%. The government hopes to achieve these cuts without such radical effects on what it delivers. The up-coming Spending Review aims to make efficiency savings and “to consider how to deliver a step change in public sector productivity”(i.e. the ratio of public sector outputs to inputs).
Public service reforms may deliver genuine efficiency improvements, but even in their absence the spending cuts are is likely to be successful in increasing measured productivity in the public sector. The graph shows the National Statistics measure of public sector productivity from 1997 to 2008. Productivity falls as the increase in inputs outstrips the growth in measured output. As inputs decrease this is likely to be reversed. Measuring the quality of public sector outputs is extremely difficult – just as the numbers may not include all quality improvements as inputs rose, the output measure may miss any deterioration in service quality as inputs fall, leading to an increase in measured productivity.
With the prospect of public sector pay freezes, quality decreases may arise through talented staff leaving for the private sector. Research suggests that remuneration over the lifetime is roughly similar in the public and private sector with a small public sector premium. Hence there is a risk, with public sector pay frozen, and if jobs become available, that quality will decrease if some of the most able employees go private.
Shifting a substantial fraction of activities out of the public sector, for example to be delivered by charitable organisations, could also increase public sector productivity if those services were relatively resource intensive and low productivity. But then comparing the performance of total publicly-provided services over time, and inherently the performance of successive governments, would not be meaningful as the measure would not be comparing like with like.