Gordon Brown - The Deferred Bill

The charge usually laid against Gordon Brown is that he was a fool. This is both untrue and analytically useless. He was manifestly not a fool. He read more Treasury papers than any Chancellor of the modern era, understood the arithmetic better than most of his officials, and out-argued a Cabinet full of people who had been to better schools. The problem was never intelligence.

The problem was a specific and repeated pathology: an inability to distinguish between how a liability is recorded and whether a liability exists. Brown's decade at the Treasury produced a remarkable body of structures that banked a political benefit immediately and pushed the corresponding cost beyond the electoral horizon, or into an accounting treatment where it did not have to be shown. He was not incompetent at government. He was extraordinarily competent at a particular kind of presentation, and that competence was the mechanism of the damage.

The distinction matters, because the "Brown was an idiot" version of this argument is easy to refute and therefore lets the real critique escape unexamined.

Twenty years of arguments

It is worth establishing what he had done before May 1997, because the pathology is not mysterious once you look at it.

Brown went up to Edinburgh at sixteen, took a first in History, and stayed for a doctorate on the Independent Labour Party in Scotland in the 1920s, later worked up into a biography of James Maxton. From 1976 he lectured in politics at Glasgow College of Technology. From 1980 he was current affairs editor at Scottish Television. In 1983 he won Dunfermline East, and spent the next fourteen years on the opposition benches, the last five as Shadow Chancellor. He became Chancellor of the Exchequer at forty-six.

Consider what is absent from that sequence. A doctoral thesis is late until it is finished, and nobody's capital is impaired in the interval. A lectureship has no outturn to compare against an estimate. Television imposes the one genuine constraint in the whole record — a programme goes out at a fixed hour on a fixed budget — but the failure mode of a bad broadcast is a weaker broadcast, and it lasted three years. Opposition front-bench work produces speeches, papers and positions; it produces nothing that is built, commissioned, delivered or paid for, and therefore nothing whose costing is ever tested against what the thing actually cost.

In roughly twenty years of working life, the deliverable was never a thing. It was always an account of a thing, judged on whether the account persuaded.

The obvious objection is that this proves too much. Alistair Darling was a solicitor and had the better crisis. Nigel Lawson was a financial journalist and is generally reckoned a serious Chancellor. Geoffrey Howe was a barrister. There is no reliable correlation between having run a business and running the public finances well, and Britain's record of importing businessmen into government does not support one.

So the claim here is narrower, and it is not about credentials. It is that a career of that shape never once closes a feedback loop while the person responsible is still in the room. Nothing Brown produced before 1997 had a mechanism by which a plausible account and an accurate one could be forced apart. He arrived at the Treasury having never encountered the distinction under conditions where it cost him anything — and then, for ten years, built structures whose costs were arranged to surface long after he had been credited with the benefit.

Whether that formation should have disqualified him is a judgement each reader can make. What it does is explain why the same error recurs across seven policy areas that otherwise have nothing to do with one another.

The tell: 395 tonnes

Start with the smallest item, because it diagnoses everything else.

Between July 1999 and March 2002 the Treasury sold 395 tonnes of the United Kingdom's gold reserves — roughly half the national holding — across seventeen auctions, realising in the region of $3.5bn. The proceeds went into dollar, euro and yen instruments.

Selling gold is not, in itself, indefensible. Gold yields nothing, and a reserve manager can construct a coherent argument for diversification. The Swiss National Bank sold considerably more over a comparable period.

The indefensible part was the announcement of 7 May 1999, which set out the schedule in advance. The market was told exactly how much would be sold and approximately when. It responded as any market told the size and timing of a forced seller responds. Gold reached $252.80 on 20 July 1999 — the low that the trade still calls the Brown Bottom — and the Treasury proceeded to sell into a hole it had dug itself.

The Swiss achieved a materially better execution over the same window by working through the Central Bank Gold Agreement and declining to pre-commit. The difference was not luck. It was the difference between a seller who understands that announcing your hand moves the price against you, and one who was more concerned with demonstrating transparency than with the proceeds.

The sums involved are small in the context of the public finances. As a diagnostic they are perfect: a transaction structured for how it would look, executed in a way that guaranteed a worse outcome, defended afterwards on grounds of process.

The pension raid, and what it actually did

The July 1997 Budget abolished the ability of pension funds to reclaim the tax credit on UK dividends. The immediate yield was around £5bn a year. The cumulative figure is genuinely contested — estimates run from £100bn to well north of £200bn depending on the assumptions about reinvestment — and anyone quoting a precise number is guessing.

What is not contested is the mechanism. Defined-benefit schemes were, in 1997, running surpluses. The credit was part of the return assumption underpinning those surpluses. Removing it did not by itself close a single scheme. What it did was remove a buffer immediately before three other things arrived: the 2000–03 equity bear market, the mortality improvements that pushed longevity assumptions out by years, and FRS17, which dragged the deficit onto the sponsor's balance sheet where the finance director had to look at it every quarter.

Treasury papers released under FOI in March 2007 showed officials had flagged the pensions consequence at the time. This is the part that resists the "nobody could have known" defence. Somebody did know, in writing, and the measure went ahead because the revenue was needed in 1997 and the consequence would land on a successor.

The honest version of the causal claim is not that Brown killed the private-sector DB pension. It is that he removed one leg of a stool at the precise moment the other three were being kicked, and had been told he was doing so.

An architecture with no owner

The 1997 tripartite settlement — formalised in FSMA 2000 — took banking supervision out of the Bank of England and gave it to the new Financial Services Authority, while leaving the Bank with the lender-of-last-resort function and the Treasury with the money.

The structural defect was not that any one of the three was incapable. It was that systemic risk had no owner. The FSA supervised institutions individually against conduct and capital rules. The Bank monitored the system but had no supervisory reach into it. The Treasury held the chequebook and no operational capability at all. A risk that lived in the interaction between institutions — in wholesale funding markets, say — sat precisely in the gaps between three mandates, each of which could reasonably say it was somebody else's.

Northern Rock demonstrated this in September 2007 with a funding model that was entirely visible in its published accounts: a mortgage book financed on wholesale markets that would stop existing the moment those markets closed. Three institutions had looked at that balance sheet. None had been given responsibility for the question of what happens when the funding disappears. Queues formed outside branches — the first run on a British bank in well over a century — while the tripartite committee worked out which of its members was in charge.

The Banking Act 2009 and the 2013 restoration of supervision to the Bank were, in effect, an admission that the 1997 design had been wrong. It took twelve years and a banking crisis to obtain it.

Rules that bind until they bind

The golden rule — borrow only to invest, over the economic cycle — was the centrepiece of Brown's fiscal credibility. It contained one degree of freedom: the Treasury defined the cycle.

In July 2005, with the rule about to be breached, the Treasury redated the start of the cycle from 1999 back to 1997, retrospectively, capturing surplus years that had previously sat outside the measurement period. It simultaneously extended the projected end date. The rule was met. Roughly £12bn of headroom had been created by an act of definition.

A fiscal rule that the constrained party can redefine when it binds is not a fiscal rule. It is a communications strategy with a spreadsheet attached. And the market understood this perfectly well, which is why the credibility the rule was supposed to purchase was never actually delivered.

The consequence showed up in 2008. Headline net debt in 2007 was around 36% of GDP — genuinely low by G7 standards, and the standard defence rests on this figure. But the UK was running a structural deficit of roughly 2.5 to 3% of GDP at the top of a fifteen-year expansion, and the headline debt figure excluded PFI commitments, unfunded public sector pension obligations, and Network Rail. The country entered the deepest recession since the war with materially less fiscal room than its own published numbers implied.

PFI: the pathology in its purest form

Around 700 PFI projects were signed, with a capital value of some £55bn and total repayment obligations in the region of £300bn.

The mechanism was straightforward. A hospital built with government borrowing appears on the public balance sheet. The same hospital built by a consortium, leased back over thirty years on an RPI-linked unitary charge, did not — under the UK GAAP treatment then applied. The building exists either way. The obligation exists either way. The obligation is larger under PFI, because private consortia borrow at a higher rate than the sovereign and require a return on equity.

The government paid a premium for the privilege of not having to write the number down.

The bill arrived, as designed, later and elsewhere: NHS trusts running structural deficits driven substantially by unitary charges on buildings they had no power to renegotiate, schools locked into maintenance contracts at multiples of the market rate, and a series of NAO reports establishing that the risk transfer used to justify the treatment had in many cases not occurred. When IFRS brought most of these obligations onto the balance sheet, the numbers had not changed. Only the visibility had.

The 10p rate: competence, straightforwardly absent

For anyone who wants a single episode of ordinary technical failure rather than structural critique, the 2007 Budget provides it.

Brown's final Budget as Chancellor cut the basic rate of income tax from 22p to 20p — the headline, delivered to a cheering Commons. The cut was funded by abolishing the 10p starting rate. The arithmetic of this is not difficult: it transfers money from people earning very little to people earning moderately more. Roughly five million households were left worse off.

The measure took effect in April 2008, by which point Brown was Prime Minister. It survived approximately six weeks of contact with reality before a £2.7bn emergency package raised the personal allowance by £600 to patch it, funded by borrowing.

This was not a subtle interaction of policy with unforeseen circumstances. It was a distributional consequence visible on the face of the measure, waved through because the headline was worth having and the effective date was thirteen months away.

IR35, and the shape of the thing

Readers of this site will forgive a brief detour to the measure that gives it its name, because it is the whole pathology in miniature.

Inland Revenue press release 35, issued the day after the March 1999 Budget, addressed a real abuse: resign on Friday, return on Monday through a personal service company, pay yourself in dividends, avoid both employee and employer NICs. Nobody serious disputes that this was happening.

But the drafting was so wide that it captured the entire genuine contracting sector alongside the disguised employees, and it did so by importing employment status case law — a body of doctrine built over decades for entirely different purposes, resolvable only case by case, and therefore incapable of giving anyone a reliable answer in advance. The measure created an obligation that could not be complied with, only litigated.

Twenty-seven years and three rewrites later it is still not fixed. The 2017 and 2021 reforms moved the determination and the liability onto the engager, which produced the wholly predictable result that large engagers stopped taking the risk at all — blanket determinations, contractor bans, and a market that reorganised itself around a rule nobody could apply. The 2024 offset mechanism exists to stop HMRC collecting the same tax twice, an admission of what the previous seven years had been doing.

The tax handbook roughly doubled in length across Brown's chancellorship. IR35 is one entry.

The case for the defence, taken seriously

An argument that cannot survive its counter-argument is not worth making.

Bank of England independence, granted in his first week, is generally regarded as the most successful institutional reform of the post-war British state. It removed a genuine and recurring source of political manipulation of monetary policy, and it worked.

The 2008 recapitalisation. When the crisis broke, Brown's Treasury forced equity into the banks rather than merely buying distressed assets, and the approach was adopted internationally within weeks. The April 2009 G20 summit was a real piece of coordination. His crisis management is better regarded abroad than at home, and the judgement abroad is not obviously wrong.

Distributional outcomes. Pensioner poverty and child poverty both fell substantially. Tax credits were administratively disastrous — overpayment recovery generated years of Ombudsman casework — but they moved a great deal of money to households that needed it, and pretending otherwise is not honest accounting either.

Light-touch regulation was consensus. The Conservative position before 2008 was that the City was over-regulated, not under. Attributing the regulatory settlement to Brown alone requires ignoring what everyone else was saying at the time.

And the crisis was not British. Ireland, Spain, Iceland and the United States operated under materially different regulatory architectures and blew up anyway. This is the strongest point against the tripartite critique: it establishes that the architecture was a contributing weakness rather than the operative cause.

Verdict

"Worst thing that ever happened to the UK economy" does not survive contact with the competition. The 1925 return to gold at pre-war parity destroyed more output. The 1970s were worse in every measurable respect. And the dominant fact of modern British economic life — productivity per hour flatlining after a century of compounding at around 2% — has causes that remain genuinely contested, of which financialisation and capital misallocation in the Brown years is one live candidate among several, alongside chronic under-investment, planning constraint and a slowdown visible across most advanced economies.

The narrower charge holds, and is more damaging for being narrower.

Brown built an unusual number of structures whose costs were designed to appear after he had been credited with the benefit. Sometimes this was accounting treatment: PFI, the redefinition of the cycle. Sometimes it was timing: the 10p rate, effective thirteen months after the applause. Sometimes it was institutional design: a supervisory architecture that was cheap precisely because its failure mode was contingent and therefore invisible on any current-year measure.

This is not stupidity. It is something more specific and, for a Chancellor, considerably worse: the persistent treatment of the presentation of an obligation as though it were the obligation itself. A man who genuinely could not tell the difference would have been less dangerous, because he would have been caught earlier.

The bills all arrived. They simply arrived addressed to somebody else.

C.J. Marsden writes on political economy at ir35andmore.com.