Chapter 1: Background to the Review
1.1 The Atkinson Review (2005)
In the UK, around 20% of Gross Domestic Product (GDP) is accounted for by the outputs of public services, comparable to most other western economies. How best to measure this substantial section of the economy has been a key question in the UK National Accounts for a long time, but the recent experience of the coronavirus (COVID-19) pandemic, and the desire to find innovative ways to improve public services without increasing spend has brought the focus back to public services, both in terms of measurement and improving delivery itself.
In 2023 the then Chancellor of the Exchequer asked the National Statistician, Sir Ian Diamond, to review the measurement of public service productivity.
Previous in-depth investigation took place in 2003 when the then National Statistician, Len Cook, asked Sir Tony Atkinson to conduct an independent review of the measurement of government output in the UK National Accounts, with a Final Report produced in 2005. This informed the development of the System of National Accounts (SNA) 2008 in how to conceptualise and empirically measure output of the public services in GDP. Importantly, it established a set of key principles which informed future work. The UK, alongside several other countries, pressed ahead with implementing these methods, particularly in relation to healthcare and education. This work addressed the largest parts of public services, but gaps remained.
The Bean Review (2016) commended the Office for National Statistics (ONS) for this work but identified that renewed efforts were needed to update methods where quality adjustments were in place, and to create new adjustments where these were not. The ONS then extended the landscape by developing measures related to Adult Social Care, Children’s Social Care and Public Order and Safety (excluding Police and Immigration). This work led to 55% of public services by volume being quality adjusted.
The coronavirus pandemic highlighted that public services can be subject to significant changes which measurement systems designed for more standard times can struggle to accommodate. In particular, ONS and the Organisation for Economic Co-operation and Development joint research evidenced that when different countries apply different methodologies in their national accounts international comparisons can be hard to interpret.
The potential for Artificial Intelligence and other innovations to transform public service in future years was an added driver to ensure the data presented by the ONS is as robust and useful as possible, and measures are future-proofed and able to account for innovation and technological changes.
This report presents the key challenges identified, key cross-cutting methodological improvements developed and consequent sector specific improvements.
1.2 The Atkinson approach
This Review is very much in the tradition of the Atkinson Review (2005) and should be seen as a periodic update of that approach and methodology. However, the first recommendation reflects the longevity and intellectual power behind Professor Atkinson’s proposals.
Recommendation 1:
The principles outlined by the Atkinson Review should remain the underpinning intellectual methodology behind the measurement of public services, both in UK National Accounts and public service productivity statistics.
Central to these principles is that the objective is to deliver a dataset which is useful for analytical purposes. The Review experienced the desire from policy-makers to understand these data, often at granular levels. Whilst the Review has attempted to develop metrics which meet this need, it has also been aware of data burdens where, as with Atkinson, the ONS has looked to use only the data departments should already be using to manage their services.
1.3 The national accounts roots of the Atkinson approach
The Atkinson Review derived its principles from core measurement principles contained in the national accounts, particularly those articulated in the SNA2008. It took valuation methodologies used in the market sector and considered their application to the non-market sector, specifically public services. Importantly, in some services national accounts concepts may differ from how policymakers would normally view their service: for example in Tax Administration, whilst the ONS measures the number of taxpayers being processed, the HMRC notes the importance of the total tax-take gathered. This issue is discussed further in Chapter 13.
Underpinning the Atkinson Principles is the fact that, in a competitive market, the value society places on a good, service, input or asset is reflected by the market price. This market price should reflect and balance two alternative perspectives on value:
- The costs of production, including an appropriate margin.
- The discounted sum of benefits the consumer believes at the time of purchase they will accrue from acquisition of the product.
The essential logic is that, under the presence of meaningful competitive forces, if there is a rival producer who can deliver the product at a lower cost, or by accepting a lower margin, this will mean they can bid the price lower. Equally, in that competitive model, a consumer who believes they will secure more value from the product will be willing to bid up the price. Where the price balances, in a competitive framework, it must be the case that:
Sum of Costs = Market Price = Discounted sum of Expected Benefits
Therefore, under conditions where a market (or exchange equivalent) price cannot be observed directly, such as in public services which are provided free-at-the-point-of-consumption, there remain two methods which can be used to understand the value of the final output of such public services:
- A sum of costs methodology (which the Review will refer to as ‘inputs = outputs’) where the value of the output is set equal to the value of the inputs which go into its production.
- A methodology which looks to proxy the discounted sum of benefits approach by splitting the analysis into two computable components: a direct measure of the volume of the relevant output (e.g. number of hospital operations) and a direct measure of the quality of the final output (so an operation where on average more patients see a greater improvement in the quality of life is valued more highly than one where the improvement is less strong).
In both cases, subtracting the volume of inputs from the volume of outputs would deliver an estimate consistent with the concept of Gross Value Added (GVA), as applied in the private sector. Similarly, dividing the volume of output by the volume of inputs would present a productivity measure. However, under the ‘inputs = outputs’ approach to measuring output, dividing output by input always gives a result of one. That is, productivity is implicitly assumed to be constant, unless an assumption is made to apply a ‘margin’ or ‘profit’ to the costs the state incurs in producing the service.
This is clearly circular: productivity would equal whatever assumption was used to define this ‘margin’. Hence this method is considered inferior as without being able to apply an objective method for the margin discussed, it is a weak proxy for the exchange equivalent price, and hence is only used where stronger alternatives are not available.
The stronger method is to directly estimate the volume of output, and an objective quality adjustment, using evidence and data, to reflect how the value of this activity changes over time, as received by citizens. This model rests on two assumptions:
Assumption One: That the quality adjustments act in the same way as a market price, so the better the quality of the product, the higher the quality adjustment factor, in the same way that this would normally be reflected in a higher market price. This has two implications:
- A quality change is implicitly equivalent to an output change in volume terms (that is, if someone was to buy car A, which is twice the price of Car B, this is equivalent to purchasing two Car Bs, so replacing a Car B with a Car A in this instance doubles volume output). As such, a 1% increase in quality is equivalent to a 1% increase in output.
- This is clearly equivalent to market transactions. Imagine a factory makes one good brick in one time period, and then two broken bricks in the next. The broken bricks cannot be sold for a positive price and can only be given away (price = £0.) Simply because there are two bricks does not mean quantity has doubled, instead it has fallen one hundred percent. In the public sector, there can be activities (say hospital operations) where the outcome delivered worsens; so comparing one successful hospital operation with two unsuccessful operations, the Review would not wish to suggest the two unsuccessful operations were worth double the one successful operation. In this instance the quality adjustment would fall to ensure the final value reflected the value of the operations being delivered.
It is this recognition that the ONS need to adjust measures of output to deliver a closer analogy of the methods used in the market sector, that marks Atkinson’s Review as the landmark which it is.
Assumption Two: The Government prioritises social welfare in decision making.
Government would, within this, make decisions relating to the quantities of each public service to deliver, such that costs are spent up until the quantity where citizens would no longer desire additional units of output. This is clearly impacted by budget constraints: whilst with an unlimited budget the social planner may be able to achieve the desired outcome, it may well be that within a fixed budget it may not be feasible to take all optimal decisions and hence:
Sum of Costs ≤ Discount sum of Expected Benefits
On the assumption that the budget available to government is insufficient to deliver all valid opportunities, it could be assumed that under this constraint interventions would be chosen where downstream benefits on average exceed the sum of costs.
Sum of costs would then inherently be expected to under-estimate the value produced, even if accurate assumptions could be made about the margin which the public sector would apply (it would be expected for governments to draw from the top of the distribution). For this reason, applying a quality adjustment is essential to accurately reflect the value created by the public sector in a form comparable to those observed within the market.
Back to top