An example from my own experience was a company that was importing clothes and applying logos to them. They thought they could claim because they had to research which supplier to buy from and the best machines to use.

After a little more questioning, my advice was that they weren’t trying to advance technology, they were simply seeking to make good use of technology that already existed. It seemed pretty clear that they didn’t qualify.

I later heard that their claim had been taken on by a large claims management company. This ticked me off mightily – here I was, giving honest advice and earning nothing from it, while the advisory company was earning big bucks by wilfully defrauding the tax system. (Even if the client was ignorant about the rules, I’m pretty sure the advisor knew exactly what they were doing).

Most ethical and compliance-minded accountants and R&D consultants have similar anecdotes, and the frustration with the R&D system has been palpable.

What does HMRC’s auditor say about all this?

This situation has not escaped the attention of the National Audit Office (NAO), which asked HMRC to estimate the level of error or fraud in the R&D tax relief system. HMRC’s estimate was £311m for 2019-2020. Sounds like a lot, but that’s only 3.6% of the expenditure on which R&D relief is claimed. Less than 4% is pretty good, right?

Obviously not entirely happy with this answer, Gareth Davies, the NAO’s Comptroller and Auditor General, assessed how HMRC had made this estimate. After reviewing HMRC’s control processes, risk assessment methodologies, assumptions and judgements, Mr Davies then made some frank and pretty extraordinary comments.

He said:

“…controls implemented by HMRC are not currently adequate in preventing or detecting material error and fraud in R&D reliefs”

“HMRC does not yet have a sufficiently developed understanding of the error and fraud risks arising from the schemes”

“…reasonable variations in [HMRC’s] judgements could have a significant impact on the range of estimated error and fraud.”

In other words, HMRC doesn’t really know the extent of the error and fraud problem, and the true figure could be much, much higher than 3.6%. We’ll loop back to this in a bit.

This concern from their auditors goes a long way to explaining HMRC’s current approach to R&D enquiries, which has been to ramp up resources and haggle hard. 100 new Compliance Checkers are now in post and sinking their teeth into a whole range of claims – both small and large. These new recruits are scrupulously applying the guidance exactly as written.

While an admirable move on HMRC’s part, it’s now exposing a bigger issue – one that cuts to the heart of the whole R&D scheme: 100% compliance with the guidance in its current form is close to impossible.

That’s quite a statement, so let me expand on that.

From the enquiries I’ve been asked to advise on over the past 6 months, even carefully prepared claims in strongly eligible sectors are crumbling around the edges, and in some cases, are collapsing completely under HMRC’s increased scrutiny.

Why is 100% compliance almost impossible to show?

Well, two reasons.

1.    HMRC now want to be seen as tough on R&D claims, to assuage the concerns of the NAO and politicians who’ve been questioning the scheme’s integrity. They seem to be starting enquiries with the view that the whole claim is ineligible, and asking you to prove them wrong (yet nothing you provide is good enough to change their mind).

2.    The definition of R&D for tax purposes is very subjective. (Just take a look at CIRD81900 to see how much wriggle room there is in the various definitions. Advance, uncertainty, Competent Professionals – there’s so much ambiguity and judgement required.)

This combination of an abstract and subjective definition of R&D, combined with political pressure on HMRC to rigorously police the scheme, is having one common result – HMRC saying:

  1. We don’t agree with your interpretation of the guidance’;
  2. Because our opinion differs from your opinion, there’s an error in the claim;
  3. Tell us why that error arose so that we can decide what penalty is appropriate.

My educated guess is that close to all of HMRC’s enquiries result in a difference between what is initially submitted and what is eventually deemed to be eligible (sometimes after many months of discussion).

Soooo…if HMRC estimates that only 3.6% of R&D expenditure contains error or fraud, yet at the same time they’re auditing lots of claims and finding that a very high percentage of those claims contain errors… then that doesn’t add up, does it? That estimate of 3.6% can’t be anywhere close to right.

Let’s throw some basic maths at this.

A statistician would tell us that if a) HMRC’s sample size is big enough and b) close to 100% of HMRC’s enquiries find errors, then c) it’s very likely that close to 100% of the rest of the claims will also contain errors.

So, first question – ‘Is HMRC’s sample size large enough to tell us what’s happening in the claim population with a reasonable degree of confidence?’

Well, according to’s handy statistical calculator, HMRC would only need to do around 400 enquiries a year to get a good idea about what was happening within a pool of 60,000 claims (with a 5% margin of error).

Our next question is therefore – ‘Is HMRC doing more than 400 enquiries a year?’

To ballpark this, let’s assume the 100 new Compliance Checkers are each responsible for 20 R&D enquiries a year. That’s 2,000 enquiries a year in total – much more than the sample size of 400 required to tell us what’s going on.

So yes, given the size of HMRC’s team, the number of enquiries they’re likely to be conducting and the anecdotal results of those, it’s a fairly safe bet that almost all claims don’t meet the standard as it’s currently being applied.

That’s fairly mind-bending, and the NAO’s concerns may be well-placed. Rather than 3.6% of R&D expenditure being erroneous as HMRC estimated, it seems more likely that close to all claims would be deemed to contain ‘error’ if they were subject to enquiry.

Holy crap. There are some big implications to this.

So, where does this leave us? Frankly, I’m not sure. All that’s clear is that HMRC appears to be expecting 100% compliance in an area where that’s almost impossible, and is then imposing penalties on companies with claims found to be lacking.

Looking forward, here are some predictions.

At some point, when this all sinks in, there’s going to be a backlash from advisors and companies against HMRC’s current approach. Many will stop submitting descriptive narratives alongside the CT600, on the basis that voluntarily submitting extra information actually prejudices claims rather than supporting them. That’ll make poor advisors even harder to spot, exacerbating one of the scheme’s worst problems.

Secondly, HMRC will stick to its guns and say: ‘Hey guys, you wanted us to police the scheme and raise standards, and that’s exactly what we’re doing – we’re applying the standard as it’s written.’ (And it’s pretty hard to argue with that.) Their success in reducing R&D relief will pay for even more Compliance Checkers, which will scare away applicants and the growth in the scheme will start to level out.

Finally, the stand-off between the advisory industry and HMRC will get so bad that someone very high up in the food chain will be asked to review the whole system of R&D tax relief (again) to determine whether it makes practical, as well as theoretical, sense. If that’s a No, then we can brace ourselves for a very chaotic transition to a new form of incentive.

Of course, it’s going to take a while for this to come to a head.

In the meantime, as an R&D advisor, it’s vitally important that your team is well-briefed on the common factors that trigger R&D enquiries…because now, if you do get an enquiry, don’t expect to win it (almost 100% of the time).

You can learn more about this in our recent blog on Why HMRC’s enquiries are so hard to win right now.