Rethinking public sector productivity

The Spending Review published alongside the 2026-27 Budget commits the Scottish government to a programme of reform ‘designed to protect frontline services while reducing corporate costs and improving productivity’. Presumably reflecting its centrality to the programme, productivity is mentioned 39 times in the document.  

Given the scale of Scotland’s fiscal challenge, it is inevitable that public sector productivity will be an issue in the looming Holyrood election campaign. Indeed, political discourse often seems to imply that easy and significant productivity gains in the delivery of Scotland’s public services are just waiting to be grasped if only the government could summon the requisite political will and managerial competence.  

But there are compelling reasons to be cautious in our expectations of productivity growth – and, concomitantly, lowering costs - in the delivery of public services. Thankfully, this is an area where economics really can help illuminate the nature of the challenge. What does it have to say on the matter?

Baumol’s ‘cost disease’: why do the costs of delivering services such as health and education tend to increase at a rate above the whole economy rate of inflation?  

In his final book, the late economist William Baumol described his theory of ‘unbalanced growth’ or, as it became known, the ‘cost disease of services’ to be the ‘longest valid forecast ever to emerge from economic analysis’. He was right to do so. A formidable body of empirical data now supports Baumol’s key propositions: that the costs of delivering labour intensive personal (or face-to-face) services would continue to rise at a rate above the whole economy rate of inflation; that these services would account for an increasingly large share of output and employment; and, that the whole economy rate of productivity growth would tend to slow over time. Dietrich Vollrath argues that Baumol’s theory “allows us to understand much of the story of economic growth in the fifty years since it was published”.  

Let’s try to distil Baumol’s argument.  

He started by grouping economic activities, ‘not entirely arbitrarily’ into two types:

• A ‘progressive’ sector’ in which “innovations, capital accumulation, and economies of large scale all make for a cumulative rise in output per man hour”. The progressive sector includes most manufacturing and extractive activities and some services. In this sector, it is relatively easy to reduce labour inputs to the production process.  
• A ‘stagnant sector’ in which “activities…by their very nature, permit only sporadic increases in productivity”. This sector is comprised of ‘labour-intensive personal services’ such as health, education, fine dining and hairdressing. In this sector it is relatively hard to reduce labour inputs to the provision of the service.  

A number of assumptions and insights in Baumol’s work then help us understand the mechanism by which rising productivity in the progressive sector inevitably leads – and this is the crux of the matter - to rising costs in the stagnant sector.

The first is that it is the role of human labour in the production of the good or service that distinguishes the progressive from the stagnant sector. Famously, Baumol uses examples from the performing arts to help illustrate the intrinsic difficulty of reducing labour inputs, and hence raising productivity, in the delivery of personal services:

Similarly, we tend to worry when the length of medical consultations is restricted or when class sizes rise or when care workers are only able to spend a few minutes with vulnerable clients.

Next is the crucial assumption that wages in the progressive and stagnant sectors travel in broadly the same direction. The model does not assume that wages are identical in the two sectors or that any change in wage levels in the progressive sector will be immediately translated to the stagnant sector. Rather, the model needs only to assume that a wage increase in one sector puts upward pressure on wages in the other sector, which is what we would expect in economies with high labour mobility like Scotland.  To illustrate: public sector pay will rise over time even if productivity growth is slow because we need teachers and doctors and their pay has to be broadly competitive with other occupations including those in the progressive sector.

The third assumption is that the progressive sector manages to offset rising wages with rising productivity (‘if wages and productivity both rise by 2 per cent costs will not rise at all’) but the stagnant sector is less able to do so. Therefore ‘every rise in wages must yield a corresponding addition to costs’. Therefore, the ratio of wages to output in stagnant services is rising over time – this is the key mechanism driving higher costs.  

The final essential insight is that the demand for services such as health and education is price inelastic; that is, demand doesn’t fall as costs increase. Therefore, the share of stagnant services in aggregate output and employment will tend to increase over time.

The main takeaways here are that slow productivity growth is an intrinsic characteristic of personal services and that rising progressive sector productivity leads to stagnant sector services becoming more expensive over time.  

How has Baumol’s model performed?

It is tempting to simply reference the pressure all OECD nations are experiencing in funding public services but, happily, there have been a number of rigorous academic studies exploring these issues.  

Nobel prize winner William Nordhaus (2006) found that the “hypothesis of a cost-price disease due to slow productivity growth is strongly supported by the historical data. Industries with relatively lower productivity growth show a percentage point for percentage point higher growth in relative prices”. More recently, and coming from a more heterodox perspective, Servaas Storm (2017) found that “the growing segmentation [of the US economy] suggests a Baumol-like pattern of ‘unbalanced growth’”. Tabarrok and Helland (2019) find that “The Baumol effect is the best explanation for rising prices in education, healthcare, and other service sectors”.

It should be acknowledged that there are objections to Baumol and that these mainly centre on the rate of productivity growth in the personal services when adjusted for quality; that is, the productivity of healthcare should be adjusted to account for the technological advances which have significantly improved outcomes. This is a fair and crucial point. But as Baumol points out, if our primary concern is rising costs, quality unadjusted figures are more relevant. (We will return to this issue in the final report).  

Much is also said about the potential of new technologies such as AI to drive significantly faster productivity growth in public services. Again, there is cause to be sceptical. The next blog in this series will address this issue in detail.

Where do we go from here?

Of course, none of the above means that efforts shouldn’t be made to improve efficiency in the delivery of Scotland’s public services, only that we should be cautious in our expectations. Baumol’s theory – and the large and accumulating body of evidence testifying to its veracity – suggests that it would be prudent to assume, at best, modest outcomes. To undertake financial planning on the assumption of large and sustained increases in productivity would be irresponsible.  

Despite his somewhat gloomy terminology Baumol was actually cautiously optimistic about the future. The ‘cost disease’ he identified isn’t some kind of pathology and the ‘stagnant sector’ isn’t stagnant in any true sense. As Tabarrok and Helland stress, “changes in relative prices are an inevitable consequence of growth and not a failure of growth. We can have our healthcare and our smartphones, too”.  

Yet Baumol also recognised that “if governments cannot be led to understand the ideas presented here, then citizens may be denied vital health, education and other benefits because they appear to be unaffordable, when in fact they are not”.  

The first step towards sustainable financing of Scotland’s public services is surely to understand why costs are rising. Baumol’s work is a crucial part of, if not the full, story (there are clearly other pressures on costs). Policymakers should make an effort to understand him.