I'm kindly invited to the British Academy to discuss a report https://www.thebritishacademy.ac.uk/publications/mapping-the-economic-returns-of-r-and-d-in-the-uk/ Mapping the Economic Returns to R&D.
Here are some notes.
1. A puzzle. “changes to the Office for National Statistics (ONS) Business Enterprise R&D (BERD) survey methodology suggests that business investment in R&D is higher than previously thought. However, the UK is still seeing stagnating productivity and economic growth.”
How can this be? There is a confusion here on levels and growth. Remember that the level of productivity depends on the level of the capital/labour ratio. The level of capital this year depends on the net of deprecation last year capital plus investment. If we have discovered more investment then
a. The level of GDP rises since there is less intermediate and more investment
b. The level of capital rises with more investment.
c. What about growth? proportional capital growth is I/K – deprecation. So discovering more I only raises capital growth if there is no more K, but the more I means more K. So in the steady state, capital growth is the same.
2. Is the “UK good at research but bad at commercialisiation”. Commercialisation costs money. So if we were bad at it, then the returns to UK investment would be very low. Below is a picture of the return on UK market sector capital.
The UK is the thick black line. No sign that it’s particularly low.
3. The report rightly points out that there are many pathways to innovation. Using the latest ONS data on intangible investment, https://www.ons.gov.uk/economy/economicoutputandproductivity/productivitymeasures/bulletins/investmentinintangibleassetsintheuk/latest#investment-by-industry-section-in-2022, we have the following
a. R&D GFCF as a proportion of GVA is largest in manufacturing
b. But viewed besides other intangibles, it’s relatively small: notice the investment in software.
c. There’s an important point here: in the national accounts, ESA10, software made in the process of conducting R&D is counted as GFCF in software and not R&D.
4. The report has an interesting take on government innovation around it’s discussion of “moats”. I think the idea is this
a. There is a wedge between the private value of spending on innovation and social value if spenders cannot appropriate returns from spending e.g. rivals copy their ideas.
b. The usual economics answer is that if returns can be appropriated = a “wide moat” e.g. watertight IP, then firms need no support and vice versa for a “narrow moat”.
c. They then distinguish between “tradeable” and “non-tradeable” technologies and processes. They say “For non-tradeable technologies and processes, or those that primarily operate in a domestic market, research, commercialisation, and adoption tend to occur in domestic pathways, with commercialisation and adoption more tightly bound, often with a single actor or firm. There is a high proportion of process-led innovation and direct implementation within firms.” I think the terms are unfortunate since tradeable and non-tradeable mean something specific in economics, but I think what is implied is whether commercialisation occurs, for reasons of technology, secrecy etc., either largely within firms or could also occur outside.
d. Their examples are in their Table 1:
Where for example mRNA is tradable, by which I think they mean potentially developed anywhere, but has patent protection. Innovative galleries, are, non-tradable ie. Found in a particular country, but have a wide moat, which means they are hard to duplicate.
e. This leads to policy proposals
5. Some thoughts on this
a. The point that knowledge is non-rival is often used to justify that innovation will not earn returns. But this depends if the knowledge is also excludable, which is the moat point. This is valuable. The UK Patent box costs £2bn with 145 companies accounting for 92% of relief. They probably have IP protection, so this feels like money that can be reassigned (https://haskelecon.blogspot.com/2025/10/how-much-is-uk-r-tax-credit-scheme.html).
b. The R&D tax credit costs not £7.6bn, with around 40% going to large firms. Perhaps we have to live with this since we have a relatively high corporate tax rate (https://obr.uk/box/corporation-tax-in-historical-and-international-context/) , 2020 data from OBR below.
c. What is the problem we want to solve here? In my work with Hans Neilson we have some of the following: a.
i.
The UK post GFC slowdown is mostly a slowdown in
TFP in non-ICT service industries.
ii.
The UK productivity growth story since the pandemic
is this. Excluding sector L, GVA per hour grew 0.34pppa 2011-18 and 0.24pppa
2019-24. (Notice how important sector L
is, Whole econ data are 0.50 and 0.35).
iii.
The Marimekko diagram is this
Where notice that health and education have growth and shares of -3.5%pa and 1.8%pa and 10% and 7% (pre-pandemic, 2011-18 growth rates were 0.4%pa and -0.3%pa). Thus the 2019-24 contributions to labour productivity growth were -0.31%pa and 0.12%pa.
So to a first approximation, the answer to the question " we are doing all this R&D and what is happening to labour productivity growth" is, since 2019, the drag from the health service.
