The asymmetry of arms: why avoidance wins before anyone gets clever

Picture the contest honestly. On one side, a public tax authority staffed to civil-service salary scales, its budget set by politicians, its headcount cut in most of the last fifteen years. On the other, a permanent, well-paid, commercially driven advisory industry of roughly 1.5 million people whose entire job is to design and defend the structures the authority is supposed to challenge. This is not a fair fight, and it was never meant to be. The outcome is not decided by who is cleverer. It is decided, before anyone opens a spreadsheet, by who can afford more people.

The structures we describe elsewhere are the weapons. This page is about the armies. The reason avoidance wins so often is not that the avoiders are smarter than the inspectors. It is that there are far more of them, paid far more, and the side that is supposed to police them keeps getting smaller.

HMRC: cut, then cut again

The UK's tax authority has spent over a decade being hollowed out at the front line. The National Audit Office's May 2024 report on HMRC's customer service was blunt: in 2022-23, callers spent a cumulative 798 years on hold, more than double the figure three years earlier. At the start of 2024-25, HMRC needed to cut its customer-service workforce by 14% in-year simply to live within its budget. Customer-service staff had already fallen from 20,139 in December 2022 to 18,996 a year later, a 5.7% drop in twelve months.

The history is worse than the recent numbers. Between 2010-11 and 2014-15, HMRC cut personal-tax customer-service staff from about 26,000 to 15,000, a reduction the NAO later said the department got "badly wrong". The strategy through this period was explicit: deliver more enforcement through productivity, not more people. HMRC's compliance yield, the revenue it claws back through enforcement, still rose to about £41.8 billion in 2023-24, up 23% on the year. The Spending Review of 2025 partly reversed course, committing to recruit 5,500 extra compliance staff and 2,400 debt-management staff by 2029-30, with HMRC's settlement rising to £7.3 billion for 2026-27. HMRC now employs over 69,000 people in total and was one of the fastest-growing departments in the year to late 2025. The point is not that HMRC is collapsing. It is that the enforcement capacity was deliberately run down for over a decade, and the rebuild is recent, partial and aimed mainly at volume.

The IRS: defunding made explicit

The American case is the clearest in the world, because the politics were open. From a peak of about 102,000 employees, the Internal Revenue Service shrank through the 2010s; by 2018 its workforce was 22% smaller than in 2010, its enforcement staff was down by roughly a third, and the audit rate had fallen 42% to just 0.5% of returns. The 2022 Inflation Reduction Act set out to reverse all of it with an US$80 billion, ten-year investment. The Congressional Budget Office estimated that would generate about US$204 billion in extra revenue over the decade, a net fiscal gain of roughly US$124 billion.

Then the reversal. Between 2023 and 2024, Congress clawed back or froze roughly two-thirds of the US$80 billion. From January 2025 to March 2026, the IRS lost 28,312 employees, the largest personnel contraction in its history, shedding about 25,000 by the end of 2025 from that 102,000 peak, a cut of around 27%. The Budget Lab at Yale estimated that a workforce cut on this scale would cost roughly US$350 billion in net revenue over 2026-2035; the Center for American Progress put a 50% staffing cut at about US$909 billion in foregone revenue over the decade. The proposed FY2026 bill would cut the IRS enforcement budget to US$4.999 billion, about US$439 million below the prior year. In March 2026 the agency's chief said it now had "perfect staffing levels," having shed 25,000 staff. Read those last two sentences together and you have the whole argument: the side that catches avoidance was cut, and the cut was called a success.

The advisory army on the other side

Against those shrinking authorities stands the advisory industry. The four largest accounting and advisory firms together employ around 1.5 million people worldwide and earned combined revenues of roughly US$212 billion in their 2024 financial year, rising towards US$219 billion in 2025. For the 2023 financial year their individual revenues were about US$64.9 billion, US$53.1 billion, US$49.3 billion and US$36.4 billion. Their tax-advisory divisions alone, the parts that build, sell and defend the structures large companies use, rival or exceed entire national tax authorities in headcount.

The dynamic is corrosive in a specific, documented way. The big firms recruit and train staff at salary scales near the public sector, then the tax authorities lose qualified people to private practice at multiples of the public wage. The firms have seconded staff into Treasury and HMRC, giving them an inside view of how rules are being designed. They have provided staff time and consultancy to political parties. None of this is illegal, and we are not alleging it is. The structural problem stands on its own: the same small pool of expertise drafts the rules, advises the government, and then advises the clients on how to navigate what was drafted. The referee and several of the players were trained in the same place, and the players pay better.

Where this leads: under-collection

The arithmetic of starving enforcement is not subtle. The UK's corporation tax gap, the share of corporation tax that should be paid but is not, rose from 6.4% in 2011-12 to 15.8% in 2023-24, now about £18.6 billion and 40% of the entire UK tax gap. The Institute for Fiscal Studies noted in October 2025 that simply returning the corporation tax gap to its 2017-18 level would recover roughly £10 billion a year, a real prize available through enforcement, not higher rates.

And the return on that enforcement is extraordinary. IRS data shows that an extra hour auditing a taxpayer earning over US$5 million yields about US$4,900 in additional tax assessed, against about US$650 for one earning around US$200,000. High-end enforcement has one of the highest returns of any government activity, which is exactly why cutting it is such a costly choice, and exactly why the choice keeps being made anyway. A government that saves £1 by cutting enforcement routinely gives up several pounds in revenue it would otherwise have collected. The asymmetry is not an accident of budgeting. It is the predictable result of fifteen years of deciding the cheaper option was the smart one.