In HMRC Commrs v Tooth  EWCA Civ 826 (heard on 2 April 2019), the Court of Appeal reaffirmed that, in order for there to be a ‘discovery’ under TMA 1970 s 29, HMRC had to reach a conclusion it had not previously reached about the insufficiency of an assessment. However, the Court’s findings about the scope of ‘deliberate’ behaviour, though not part of the main decision, broke new and disturbing ground by illustrating how an inaccuracy can be ‘deliberate’, and therefore potentially lead to longer time limits for investigation and higher penalties, even if there was no dishonesty and the taxpayer had explained fully in the ‘white space’ what he had done and why.
Essentially, the case involved a tax avoidance scheme (the ‘Romangate’ scheme) to utilise employment-related losses to relieve liability on other income. The losses were incurred in 2008/09 and were set off against 2007/08 tax under ITA 2007 s 128. However, because of a technical issue with the software, the employment loss was inserted, not in the correct place on the return, but in the partnership pages using a made up partnership UTR, a method sometimes used by the promoters of the scheme to work around the software problem. The taxpayer explained in the white space that these were not partnership losses but employment losses, that his interpretation of the law probably differed from HMRC’s, and that he expected HMRC would raise an enquiry.
The Romangate scheme was defeated by anti-avoidance legislation in 2009 but the taxpayer claimed that his self-assessment for 2007/08 should stand because it had not been validly challenged. Even though HMRC had opened an enquiry in 2009 and amended the taxpayer’s self-assessment to remove the loss relief claim, it was later forced to admit that it had raised the enquiry under the wrong provision. It had used TMA 1970 sch 1A – which is used to enquire into a claim that is not included in a return – when the claim had in fact been included in the return, so HMRC should have proceeded under s 9A. That HMRC had followed the wrong course had just become clear from the Supreme Court decision in HMRC Commrs v Cotter  UKSC 69;  1 WLR 3514 (‘Cotter’).
Since the enquiry had turned out to be invalid, in 2014 HMRC raised a discovery assessment on the grounds that the 2007/08 self-assessment to tax was insufficient and the taxpayer had made a deliberate inaccuracy. The taxpayer appealed successfully to the First-tier Tribunal and HMRC’s appeal to the Upper Tribunal was dismissed.
HMRC’s onward appeal to the Court of Appeal raised two questions:
- Did HMRC ‘discover’ that an assessment to tax was insufficient under TMA 1970 s 29(1)(b)?
- Did the taxpayer (or taxpayer’s agent) ‘deliberately’ bring about a situation in which an assessment to tax was or had become insufficient? If so, HMRC could then raise a discovery assessment under TMA 1970 s 29(3), or go back 20 years under TMA 1970 s 36(1A)(a).
Had there been a discovery?
Floyd LJ said that, for a discovery assessment to be valid, HMRC had to have newly discovered that an assessment was insufficient, not that the wrong mechanism had been used to address the insufficiency. The HMRC officer’s conclusion in 2014 that the self-assessment was insufficient was not new – HMRC had reached that view back in 2009. The officer had merely realised, following Cotter, that HMRC’s position was no longer protected by the open enquiry because HMRC had proceeded under the wrong statute. Therefore there were no grounds for a discovery assessment.
Given that decision, the judge said it was unnecessary to consider whether there was an inaccuracy, and if so whether it was deliberate, but since the matter had been argued before the Court he would consider it. He said there was no inaccuracy in the return merely because the employment losses had been inserted in the wrong box, given that a full explanation had been made in the white space. However if there had been an inaccuracy, it would have been deliberate, and would have led to the insufficiency of tax: the partnership losses had been put in the wrong box deliberately, and once they were there, the software automatically fed them into the calculation of tax due.
On discovery, the other two judges, Patten LJ and Males LJ, agreed with Floyd LJ. But they differed from him on the deliberateness issue. They considered that there was a deliberate inaccuracy even though there was no dishonesty on the part of the taxpayer, and the inaccuracy did cause the loss of tax.
For the future, the judges’ analysis of deliberate inaccuracy might mean that HMRC could go back 20 years rather than the usual 4 years or 6 years for careless inaccuracy, even where there is no intent to deceive or defraud, and even if the inaccuracy has arisen solely because of the need to work around a problem with the software and is fully explained elsewhere on the return. It can also mean a higher level of penalties. Neither of those results was, I believe, foreseen or intended by those responsible for the policy on inaccuracy penalties back in 2006 and 2007. Given the state of the software, the taxpayer in Tooth had no real option but to act as he did, unless he was prepared to file an incomplete return which clearly he was not, and in an age when the use of HMRC-approved software, with all its potential for malfunction, is gradually being made compulsory, that is a worrying reflection.
Robin Williamson MBE CTA (Fellow) is an author and commentator on tax, welfare and public policy, and a part-time senior policy adviser at the Office of Tax Simplification. He was technical director of the CIOT’s Low Incomes Tax Reform Group from 2003 to 2018.