Challenges of measuring defence output


Challenges of measuring defence output

Ron Smith

In many countries, defence makes up a significant proportion of the national budget, often reducing funds allocated to education, healthcare or social security. In fact, data from the Stockholm International Peace Research Institute shows that military spending rose by almost 7% to a record 2.4 trillion dollars last year.(i) The share of spending in GDP varies enormously across countries and across time. Regardless, spending on defence is significant enough to matter for other public sector spending.

In this context, measurement of the output, inputs and productivity of defence is crucial to form an accurate cost-benefit analysis. However, there are various challenges and complexities involved in this process.

A new ESCoE discussion paper explores the challenges of measuring defence output, inputs and productivity within the UK national accounts.

The paper’s publication follows ESCoE’s recent event on measuring public sector productivity, in partnership with the Office for National Statistics (ONS).

The Atkinson Review

The 2005 Atkinson Review established principles for measuring public sector output that have stood the test of time. Since the Review’s publication, the ONS has applied these methods in its measures of public service productivity

Following the Review, a 2008 scoping paper looked at work by the Office for National Statistics and the Ministry of Defence to improve measurement of defence in the national accounts (which provide an integrated description of all economic activity within the UK). However, it is unclear if any of the paper’s recommendations were taken forward.

In 2023, the Chancellor of the Exchequer commissioned a Review of Atkinson’s principles by the National Statistician to reflect changes in services over the past twenty years, particularly in light of the COVID-19 pandemic. Atkinson also recommended reviewing the implementation of his principles every few years. This paper is a contribution to that discussion.

Challenges in defence measurement

What should we measure?

In the private sector, output is the quantity and quality of goods or services produced in a given time period, within a given economic network, whether consumed or used for further production. However, there is currently no agreed definition of output in the context of defence. In fact, defence is an extreme example of a non-market activity. There is little reliable data on prices, and it is provided free to the whole population (a classic public good, which in economics is called non-excludable and non-rival). This differentiates it from healthcare and education which are provided to individuals. Therefore, the framework used for private goods (expenditure = quantity x price) seems inappropriate for defence. Currently, defence output is still measured as output = inputs in the national accounts. This means that productivity growth is zero.

For defence, performance measures include success in operations, maintaining readiness, and stopping equipment being delivered late, over budget and not meeting technical requirements. These are difficult to convert into indicators that would match national income accounting criteria. The choice of indicators could also be very politically sensitive and raise difficult questions for national statisticians.

The 2008 scoping paper put forward three potential measures for output: activities, capabilities and objectives, with a preference for capabilities. When thinking about capability, defence budget is spent on forces, troops and equipment to provide the ability to do things, rather than actually doing them. However, it can be difficult to measure output when the desirable outcome is that nothing happens; defence is effective when threats are deterred, police are effective when there is no crime, and doctors are effective when disease is prevented. With preventative medical services, you can estimate the effect of the treatment (e.g. heart attacks avoided). When thinking about defence, estimating things like wars avoided is more speculative. How to measure the threat has also been an issue. For example, how can we accurately quantify threats like terror attacks?

In general, inputs can be easier to measure and there is scope to improve the estimates. However, there are also problems with input measurement within defence, including the treatment of declining value of capital stock and setting asset values for large land holdings.

How can we measure the same thing across time and space?

In the context of national accounts, comparability over time is a major issue for defence. What we define as a defence output changes frequently as threat, technology and strategy evolve. Similarly, activities included or excluded by the defence budget change over time. For example, if we use nuclear capability as a key measure, output may fall if the military are occupied with different capabilities, which may be more relevant in the context – for example, the capability to respond to a cyber attack rather than a nuclear war. Therefore, performance measures chosen for output may provide misleading conclusions.

In spatial terms, there is also the question of how to quantify the impact of alliances. For example, how can we quantify how much smaller countries might benefit from the defence budgets of other countries?

Why does it matter?

This research takes a fresh look at the possibilities of defence output measurement. There is little existing literature on the topic, and it has not been examined in detail since 2008.

It may be that there is no practical way to overcome these challenges and provide a measure that meets the Atkinson principles for the measurement of government output and productivity. These include that corroborative evidence should be sought on government productivity, as part of a process of “triangulation” and that explicit reference should be made to the margins of error surrounding national accounts estimates.

This paper is part of ESCoE’s research programme on national accounts. For more on this topic, listen to our measurement series event recording and view the slides. If you have any thoughts on the current Review, please contact

Ron Smith is Emeritus Professor of Applied Economics at Birkbeck, University of London.

ESCoE blogs are published to further debate.  Any views expressed are solely those of the author(s) and so cannot be taken to represent those of the ESCoE, its partner institutions or the Office for National Statistics.

i Tian, N., Lopes da Silva, D., Liang, X and Scarazzato, L. ‘Trends in World Military Expenditure’, April 2024

About the authors