Prof Michael Lewis is a Professor of Operations & Supply at University of Bath School of Management. His work explored a broad wide range of operations and supply management challenges, ranging from fast fashion retail and senior care to the initiation of mega-projects and the retrofitting of buildings. His latest work, "A Better Way to Pilot Emerging Technologies" (with Netland and Maghazei) has just been published in MIT Sloan Management Review. Michael's long-standing engagement with public policy includes helping to evaluate Best Value legislation, support for 'failing' local government services, PFI contracting and, most recently, the regulation of emerging technologies.
Last week’s Budget provoked a range of reactions. Conservative MPs felt it wasn’t political enough and commentators either declared it “underwhelming” or bemoaned “missed opportunities”. Above all there was a sense that it hadn’t acknowledged the scale of the challenge.Stagnant per capita income and weakening infrastructure; not just roads and railways and rivers but the social infrastructuresthat make life worth living. No real mention of the financial crisis in local government, the transformed context for defence and the existential challenges of climate change. For all the talk of fiscal rules, the budget seemed profoundly Micawberish. Conservative party happiness or miseryreduced to the unreliable fiction of ‘headroom’, a forecast of the gap between two vast numbers, spending, and revenue, over a multi-year timeframe. Reinforcing that feeling of unreality was the silence around future spending, the 1% above inflation uplift means £20B in cuts outside protected areas. Instead, the Chancellor emphasised the Office for Budget Responsibility (OBR) claim that a 5% increase in public sector productivity could equal around £20 billion in extra funding. It is this ‘productivity will save us’ part of the budget is what I want to reflect on in this blog post.
In recent years, the only "productivity puzzle" the UK seems to have solved is in the production of plans for productivity.We have had Osborne’s Productivity plan (2015), May’s Industrial Strategy (2017) and then Johnson’s Building back better (2021) with its diagnosis of historic low levels of investment in physical capital and lower levels of basic and technical skills. We even had Hunt’s ‘Autumn Statement for Growth’, only 105 days before this budget, with its 110 measures including many things to, you guessed it, boost productivity.
Productivity can seem simple. If outputs go up more than associated inputs, then the system is getting more productive. Simple? Well not really. Even measuring it is difficult, think of the vast array of inputs and outputs in a setting like healthcare or education. How do you measure productivity in an area like policing? The Chancellor highlighted thecomplexity of its outputs when he talked about how preventing knife crimes helped avoid hospital admissions.
At a national level productivity is fundamental because higher productivity means more goods and services produced with the same amount of labour and capital and this enables higher wages, lower prices, and funding for public services. Essentially, boosting national productivity is crucial for enhancing quality of life, reducing poverty, and ensuring sustainable economic development. With this as a backdrop, understanding why we have, for example, lower financial services productivity, really matters. In this case it is largely a question of reduced demand. For example, between 2018-21, the value of financial service exports to the EU fell by 19%and smaller output from this critical sector contributed to lower national productivity. But the Chancellor also spent a lot of time addressing the related but distinct notion of system-level productivity, especially in his emphasis on £3.4 billion for the NHS productivity plan. This kind of productivity needs to be understood in a different way. For example, given how important waiting times and responsiveness are in the NHS, let’s use a bit of queueing theory to explore budget proposals to improve NHS productivity.
Waiting time in a queue can be approximated as the product of three factors: variability, utilization, and service time.Variability is unpredictability in demand for, and provision of, services. It complicates planning, as the system must accommodate a wide range of patient needs and treatment times. Influencing productivity by reducing variability wasimplied in talk of new appointment systems, via an enhanced NHS app, that minimise missed appointments. Utilization is a measure of both how much resource you have and how theseresources are being used. The NHS has high utilization for example, and this is a balance between too few resources and potentially wasted efforts on non ‘value adding’ work. The chancellor was referring to the latter in talk of 13 million hours lost annually to paperwork and antiquated IT systems. Service time is about the absolute duration of each consultation or treatment pathway and was implied in talk of investments to streamline MRI and CT scanning processes and digitising operating theatre workflows to facilitate an additional 200,000 operations per year. These variables interact of course, although there was no real sense of this. Increasing capacity at one point in a system doesn’t help if you have a constraint elsewhere – for the NHS some of the key capacity limitations lie beyond the hospital and here the missing national care strategy is a significant brake on overall performance. Moreover, treating more patients doesn’t help overall if you have even greater increase on demand becauseof long-term health system problems.
The focus of the NHS plan also raises the role of technology. In a manufacturing business, it’s quite easy to see how investment in machinery increases output per unit of input. Determining its impact in a public service like policing is more difficult. It might reduce utilization (“Police officers waste around 8 hours a week on unnecessary admin”) but the nature of policing outputs means that the impact of this ‘free time’ is determined by numerous other factors such as training, community engagement, and effective policy. The same logic applies to the rather more eye-catching intent to make “appropriate use drones as first responders.” So, while it is likely true, as suggested, that Large Language Models have the potential to improve public sector productivity, the purpose of their use needs to be much clearer. In general, ever more grandiose promises to transform things with ever more sophisticated technology should be tempered by three basic questions:
Moreover, in a world of crumbling RAAC roofing (nb., itself a technological innovation in the 1950s), outdated IT infrastructure, poor data quality, etc. – not to mention AI-specific concerns over transparency, accountability, and energy consumption – is the shiniest technology the place to start spending scare capital resources?
One final thought on this budget – and indeed all budgets. Their tendency towards political theatre, to producing an endless array of politically astute but short-term interventions, many of them disseminated from Whitehall into the specifics of a local service, is itself a large part of the productivity problem. Perhaps it’s time we stopped seeking quick solutions, recommending yet another technology or tax break or even alternative institutional arrangement, but instead embracing a set of productivity fundamentals, akin to the Nolan principles perhaps, aimed at maintaining and enhancing this most critical facet of national performance over time.
All articles posted on this blog give the views of the author(s), and not the position of the IPR, nor of the University of Bath.
Respond
