Horizon funding could end up being a data definition problem

Before every fiscal event since we as a buccaneering trading nation finally decided to affiliate with Horizon Europe (and Copernicus for that matter), the research sector has been concerned that the cost of this affiliation (likely to be around a cool €2.6 billion each year, around £2.2bn depending on the exchange rate) might have to be found from existing research funding.

The affiliation costs do flow through the Department of Science Innovation and Technology, but according to the DSIT estimates this spending is recognised on an accruals basis – there’s no requirement for a provision for the full lifetime of the programme.

That said, there is a nominal provision – in 2022-23 DSIT surrendered about £1.5bn of unused capital that was supposed to be used for Horizon Europe but (as it turned out) did not.

The most recent ONS dataset on government research and development spending (released in April this year) presents data from 2022, prior to Horizon affiliation. This shows £16bn of UK research funding across the whole of government (both capital and recurrent) – of which just under £6bn flowed through UKRI and a further £3.5bn related to other higher education allocations (for historic reasons, this line includes Research England spending (stuff like QR and HEIF) alongside devolved equivalent). A further £3.9bn flows through other departments.

We do also get a line for “indicative UK contributions to UK R&D” – the last full year of affiliation (2019) shows £1.3bn, with a residual £568m for 2022. As far as I can tell, this latter relates to support for concluding projects and contingencies against future involvement.

[Full screen]

Now I am aware that 2022 is a while ago, and 2019 is in the almost unimaginably distant past, but what I’m really trying to get across here is that Horizon spend (even Copernicus spend) shows up as research spending. We can’t find Horizon affiliation funding from “existing research funding” because Horizon is existing research funding. A cut – in other words – is going to look like a cut: there’s no governmental slight of hand here.

If you are interested in the picture since 2022, there is some data available from the Treasury public sector statistical analyses (PESA) that takes us up to the 2023-24 financial year. It doesn’t quite match up in terms of definitions, because PESA uses the UN classification of the functions of government (COFOG) with some custom additions (and I’ve manually identified the primarily R&D lines), whereas ONS uses the OECD Frascati Manual to decide what is, and what isn’t, R&D spending.

[Full screen]

Here, you’ll note, we don’t see the EU contributions – that’s not because they don’t feature in PESA, but because it is not possible to disaggregate them into purely R&D activity at the level presented by the Treasury.

With apologies for dragging you all through this to get here, the problem of who pays for Horizon (and from what pot) may well come down to a question of data definition.