Targeting the right measure of innovation investment

By Josh Martin (ESCoE)

The UK government has, since 2017, had a target for 2.4% of the economy to be spent on research and development (R&D) by 2027. This would bring the UK in line with the OECD average (as it was at the time of setting the target), and up substantially from the 1.7% of GDP that it currently is in the UK.

But is this the right measure, or the right target? Assuming that the government is targeting R&D as a measure of innovation investment, does the choice of measure make a difference? As usual in economic statistics, paying attention to the definitions, methods and coverage of data is very important.

Statistics on R&D investment (in the UK and abroad) are governed by the internationally-agreed rules on National Accounting and the OECD’s Frascati Manual. This Manual defines what R&D is and how it should be measured. Spending must be on activities that are all of the following to qualify as R&D: novel, creative, uncertain, systematic, and transferable. That’s a high bar, and favours certain types of activity, predominantly in the manufacturing industry. Indeed, while making up just 10% of UK employment, the manufacturing sector accounted for 40% of the UK’s R&D investment in 2019. On a per worker basis, it is the most R&D intensive industry of the economy by far. (I discussed this at the ONS-ESCoE-Productivity Institute special session on Missing Capitals at the Royal Economic Society annual conference 2021 – see this blog for a summary of that event or watch this recording)

One advantage of using this measure is that it is collected in an internationally consistent way across a wide range of countries. One drawback is its relatively narrow focus and manufacturing-tint. Nonetheless, the government’s target is based on this measure – 2.4% of OECD GDP in 2017 was spent on R&D, and that became the target. Intriguingly, this is not the average amongst OECD countries, but the rate in aggregate in OECD countries – that is to say, weighted according to the size of the economies. A simple average would be closer to 2.0%, but with larger OECD countries typically spending a larger share of their economy on R&D than smaller OECD countries, the aggregate is higher. For instance, the US, Japan and Germany all spent over 3% of their GDP on R&D in 2019.

The chart below shows where the UK lies relative to OECD countries, the OECD aggregate, and the EU27 aggregate, in 2019 – in the middle of the pack, and well below target.

But if we are interested in more than just Frascati-defined R&D and the manufacturing industry, and instead want a broader measure of innovation investment, what should we target? As well as R&D, the National Accounts treats other intangibles as capital assets: software and databases; mineral exploration and evaluation; and entertainment, literary and artistic originals. The sum of these four assets (know as intellectual property products, or IPPs) is a broader measure of innovation investment, also measured according to international standards across the world.

This measure is far less manufacturing-focussed – now the ICT industry is more innovative on an investment per worker basis than manufacturing, and the financial services industry looks substantially more innovative too.

The chart below shows where the UK lies on this basis relative to OECD countries in 2019, in the same way as the previous chart. On this measure the UK is still mid-table, but closer to the OECD aggregate.

However, a large literature thinks that the National Accounts don’t go far enough. ESCoE members like Carol Corrado among many others have championed work to develop measures of a broader set of intangible assets, beyond those treated as such in the National Accounts (indeed ESCoE is holding a joint conference on intangible assets with the International Association for Research in Income and Wealth (IARIW), in November 2021). These extra assets include branding, firm-specific training, product and process design, financial product innovation, and investments in organisational structures. These expenditures are all considered current expenses (intermediate consumption) in the National Accounts, if they are measured at all.

This is a very broad measure of innovation investment, which might be desirable if all forms of innovation are considered equally important. But while this approach has many supporters in academic circles, these ‘beyond-National Accounts intangibles’ are not measured routinely by many National Statistical Institutes (the ONS is an exception) and the data are not available for all countries.

To compare the UK internationally on this basis, we can use data from the EU KLEMS project – a research dataset covering EU countries, the US and Japan. This prohibits an OECD aggregate, although over half the 38 OECD member countries are represented, and around two-thirds of OECD GDP is covered. The EU27 aggregate is still possible, and can be compared with the previous charts. While these data are as good as it gets for these broader intangibles, and are constructed using EU-consistent sources, the risk of inconsistency across countries is far higher than in the previous measures.

On this basis (perhaps unsurprisingly if you’ve got this far!) the UK is a world-leader – well above the OECD approximation and the EU27 aggregate, and bettered by only a handful of smaller economies (see chart below). It’s a fair bet that if the other OECD member countries were included, the UK would still be around the top.

So what to conclude? Measurement is important (if you’re reading this, I’m sure you know that already!) and it really matters for policy making. Is the UK a straggler or a world leader in innovation? Is manufacturing the most innovative sector of the UK economy, or is it financial services? Depending on how you measure it you get different answers. Maybe all are true, in a way.

For those making policy and setting targets, I would urge attention to the measures being used – what do they mean, how are they collected, what do they cover? Perhaps Frascati-defined R&D is a better target for some outcomes (like export intensity), and a broader intangible investment is better for others (like labour productivity growth). Therefore, ultimately it is a matter of policy preference, but policy analysts should know that what they measure has consequences for what they get.

Josh Martin is ESCoE Topic Lead for productivity and related topics.

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.