Fiscal Illusions

In June​ , Philip Hammond, in his last few weeks as chancellor of the Exchequer, wrote to the candidates vying to succeed Theresa May as leader of the Conservative Party and asked them to pledge that, if elected, they would retain his target of bringing down national debt as a percentage of GDP. ‘If we do not commit to getting our debt down after a nine-year run of uninterrupted economic growth,’ he demanded, ‘how can we demonstrate a dividing line between the fiscal responsibility of our party and the reckless promises of John McDonnell and Jeremy Corbyn?’ Four years ago in these pages I warned that the government’s plans to bring down the headline debt figure through asset sales, including the sale of part of the student loan book, would mean a loss of millions of pounds to the Treasury (LRB, 5 March 2015).1 We have since seen two student loans sales. The most recent accounts from the Department for Education show that loans worth £5.6 billion were sold for £3.6 billion, entailing a loss of £2 billion – which doesn’t exactly fit most people’s definition of fiscal responsibility.

It is easy to become inured to the large sums involved in the student loan scheme. More than £18 billion of loans were issued to UK and EU students to study at British institutions for the last academic year alone. Those figures will only grow. From September, students will be able to borrow up to £9250 a year towards tuition, and additional sums towards living costs: as much as £11,764 for those studying in London and £8944 for those elsewhere. The sum of outstanding balances on the loan book now stands at more than £100 billion: twice what it was four years ago.

The government began its programme of annual sales of parts of the loan book in December 2017, with the aim being to raise £12-15 billion over five years. Those figures fail to match even a single year’s issue of new loans, so at first glance it seems to make no sense that the government should put in so much effort – £15 million spent on consultants, £23 million in total on getting the programme off the ground – for such meagre short-term gains, and such large long-term losses, when the immediate proceeds are negligible in light of the growth of the scheme overall. ‘It is not immediately obvious,’ as the Office for Budget Responsibility says, why selling the loans ‘at such a loss is of net benefit to the taxpayer’.

The Treasury is unsettled by the current level of debt. Under the last Labour government the target for public sector net debt (PSND), the official measure, was 40 per cent of GDP; currently, net debt is just over 80 per cent of GDP. So the loan sales have been presented as part of the government’s ‘plan to repair the public finances’: it will look to sell assets such as student loans ‘where value for money to the taxpayer is assured, and where there is no policy reason to continue to own them’. The chair of the Public Accounts Committee, the Labour MP Meg Hillier, sees instead ‘another example of the government selling off assets for short-term capital gain. The government prioritised removing the loan book from the balance sheet, even though some other options might have generated higher proceeds.’ First among these options would be keeping the loans on the books and waiting for the repayments to come in. But from the government’s point of view, the risks associated with waiting probably offer the strongest justification for short-term thinking. Student loan debt is evaluated by making projections of loan repayments over future years. These are quite uncertain. With standard debt, repayments are fixed and known, even though the effect of inflation on the value of the repayments isn’t. But student loan repayments vary according to borrowers’ incomes, so modelling what they will be in future involves trying to predict the variation in people’s earnings – along with other macroeconomic indices – over many years to come.

In the sale of 2017, the government raised £1.7 billion by ‘securitising’ 400,000 loan accounts dating from the late 1990s and early 2000s. Securitisation, in contrast to an outright sale, involves selling the rights to a share of the income stream generated by loan repayments. Most of what was on offer was snapped up by pension funds and insurance companies, though a variety of investment funds also made purchases. The Department for Education assessed the worth of the loans sold off that year at roughly £2.6 billion (based on estimates of future repayments), meaning that it booked a loss of £865 million on the deal. The second sale raised £1.9 billion, had a similar structure, a similar size (370,000 borrowers) and, based on an estimate of what the loans were worth, entailed a loss of £1.1 billion.

The £3.6 billion the government got from the two sales was fixed and known, unlike the £5.6 billion of the DfE’s official valuation: a smaller, determinate sum today was preferred to the uncertain prospect of a larger sum accumulating over the decades to come. The government was able to represent the sales as a ‘derisking’ of the public sector balance sheet. (Notice, in passing, the mixed messages coming from a government that tries to persuade us higher education is an investment while setting up a sale programme on the assumption that the ‘graduate premium’ isn’t to be relied on.)

The second aim of the sale was to reduce PSND. When new loans are issued, an asset is created, since borrowers owe the government money, and so is a corresponding liability: the government borrows to raise the money it lends to students, and this adds to the national debt. You might expect the loans to improve the government’s balance sheet, since they accrue interest and are of greater nominal value than the new liability. But owing to an accounting nuance, student loans are excluded from the asset column in calculating PSND, because they are considered to be ‘illiquid’ – i.e. hard to sell. By contrast, proceeds from the loan sale are deemed to be liquid, so whatever amount is raised by a sale improves PSND. One asset is swapped for another of lesser value, but the sale proceeds are in cash and so boost the headline figure. So far the sales have reduced PSND by £3.6 billion, even though they have weakened the nation’s real balance sheet.

When ministers and senior civil servants appear before parliamentary committees they have skated over the composition of PSND. The loan sales are presented as a straightforward contribution to ‘debt reduction’. Last September, for instance, Charles Roxburgh, second permanent secretary to the Treasury, told the Public Accounts Committee: ‘We got the 48p in the pound – the £1.7 billion – and we paid down debt by £1.7 billion … The chancellor has set out that it is really important that we continue to have debt falling.’

But not everyone agrees that it is ‘really important’ to have debt falling, or even that it is responsible to chase such an aim. The OBR has followed the International Monetary Fund in using the term ‘fiscal illusions’ to refer to ‘situations where fiscal aggregates (accounting measures of the budget deficit or debt) do not reflect the true fiscal implications of the transaction taking place’. In the IMF’s view, such ‘illusions’ are sustained at the cost of broader – and better – understandings of financial and economic health. PSND insofar as it excludes certain assets is precisely the kind of measure they have in mind. Public balance sheets can be damaged, the IMF has said, when changes in headline statistics are pursued at the expense of a considered approach to investment and asset management. The National Audit Office, too, in its recent report into the sale of student loans, dissented from the government line: ‘Public sector net debt is limited in its scope, providing a narrow view of debt.’

Such doubts have not deterred the UK government, which goes to remarkable lengths to improve headline figures. No sale of income-contingent loans had been attempted before.2 ‘We are establishing a new asset class,’ the former higher education minister Jo Johnson told the Treasury Select Committee. The government laboured through several plans before settling on the current one. As recently as four years ago, it was still hoping to make an outright sale of the loan debt of whole annual cohorts of students. That idea fell away, apparently because the terms on which loans were issued before 2012 made the debt associated with them unattractive to potential buyers: the interest rate on older loans was set at whichever was lower at any given time, the Retail Price Index or the bank base rate plus one percentage point. At the time of writing, RPI is around 3 per cent, and the base rate is 0.75 per cent, so purchasers would be facing sub-inflation interest of 1.75 per cent. Not very attractive. Instead, a scheme was concocted to divide up the cashstreams from repayments, generating a menu of options for potential buyers.

All pre-2012 loans are earmarked for sale. At the end of March this year, the outstanding balances on these loans amounted to just over £34 billion, with the DfE valuing them at £22.75 billion. The first sale was of loans issued to students whose undergraduate education took place between 1998 and 2005; the second sale concentrated on those who left university between 2006 and 2008. In both sales, investors were offered a choice between four different kinds of bond. Each ‘tranche’ had its own terms, setting out the future payments buyers could expect in return for the price they were paying today. The tranches were denominated in £100,000 notes but were sold at different prices and in different amounts. In the first sale, the two most secure tranches – the ones consisting of loans associated with the lowest risk of default – were priced at 99 per cent and 93 per cent of face value respectively, so that individual notes sold for roughly £99,000 and £93,000. Together these two tranches comprised more than 40 per cent of the notes on offer, accounting for about £1.4 billion of the sale’s proceeds. This may appear to have given the Treasury good value for money, but even here, returns were set higher than the interest rate at which the government has been able to borrow in recent years.3 A small third tranche, accounting for only 3 per cent of the notes sold, tied payments to inflation. The fourth, ‘unrated’ tranche comprised more than 50 per cent of the notes on offer. These were sold at 8.5 per cent of face value, i.e. £8500 for a £100,000 note. This junk tranche absorbs all the risk that repayments turn out to be lower than expected; the buyers get the bulk of their payments only once all the other bonds have been paid off. That explains the cheap pricing, yet according to the government’s model, investors in this tranche should even so expect a return of more than 13 per cent. Unsurprisingly, demand for these bonds was high – twice the supply on offer. The sale of this tranche raised just £160 million.

It is clear that the purpose of the fourth tranche was to shift risk by selling it very cheaply: the vast majority of the £865 million loss on the first year’s sale can be attributed to the unrated notes. Had a value for money test been applied to each tranche individually, not to the deal overall, the sale would have failed the test. Half the notes in the second sale in December 2018 also belonged to an unrated tranche and were similarly priced, at 10 per cent of face value. They were offered to buyers ‘looking for high-yielding assets’.

The government would argue that a scheme of this sort was the only way to achieve a ‘genuine sale’: one that removed the loans entirely from its balance sheet. It didn’t want to be left with a rump of unpaid loans on its books. Had the government shifted only the more secure tranches, but picked up the risk of non-repayment itself, it couldn’t have worked the angles to its presentational advantage.

Through a complex bit of accounting chicanery the government has been able to issue heavily subsidised loans and then dispose of them at a further loss (that £2 billion difference between their book valuation and the price obtained) without having to record the costs of either event in the official figures it presents to the public. A fundamental aspect of book-keeping was sidestepped. As an OBR press release put it, ‘a quirk of the accounting treatment means that selling loans before write-offs have been recorded means the true cost of student loans is never recognised in borrowing.’

But, at last,​ time has been called on these tricks. In 2018, the Office for National Statistics, the body responsible for compiling the national accounts, launched a review of its treatment of students loans and concluded that income-contingent repayment loans did not meet the statistical definition of a commercial loan. They are not ‘unconditional debts repaid at maturity’ since the income contingency means they are conditional, and at maturity the balances can be wiped, not paid off. The ONS has announced that from later this year the cost of loans will be registered in the national accounts at the time they are issued, a change which is likely to see the deficit increase by more than £10 billion a year. Future loan sales which involve a significant difference between sale price and book value will now also trigger a cost to the deficit. What’s more, the ONS recently announced that it is in the process of reviewing the two previous sales to determine whether a retrospective change to the headline statistics is needed. But it isn’t clear whether further hits to the deficit will be enough to deter the government from carrying out future sales. In a brief response to the ONS announcement, the government pivoted away from discussion of the deficit and emphasised that the impact of loan sales on ‘the debt’ would be unaffected.

How did a loss-making loan sale, however structured, pass any sensible value for money test in the first place? In short, the government skewed the test.4 The Treasury sale team was allowed to set a higher value on money today (versus future receipts) than the DfE uses for its evaluations. Placing a higher value on money today drives down the acceptable sale price. One justification the government gives for taking this approach is that it has to factor in the possibility that the cash raised in the sales could be used for other things. A DfE report published in December 2018 on the second loan sale states:

A decision to divest is a decision between keeping the money tied up in an asset and releasing it for other uses which generate a social, economic or financial return. In this case two alternatives are being compared: receiving the cash flows over time (retaining the loan book) or receiving the cash today (selling the loan book) enabling productive use of that cash immediately and the opportunity to earn social, economic or financial returns.

The suggestion here is that the cash raised from the loan sales could be used for ‘productive’ ends, yielding greater benefit than would be gained by holding on to the loans.

But that isn’t what has happened. The stated object of the sales is to reduce PSND. Of course, some money could be used to reduce PSND and some used on other projects; but no such projects have been earmarked. What is satisfied here isn’t the public’s alleged preference for spending today over income in the future, but the preference for a marginally lower debt statistic. According to the IMF, the UK’s public sector balance sheet is in the red by more than £2 trillion (£3 trillion in assets, £5 trillion in liabilities). Student loan sales show us in microcosm the intellectual contortions required by a government driven by a wish to represent its general underinvestment as prudent management of the economy. By selling a simplified story of responsibility (‘We will have debt falling as a percentage of GDP’) and focusing on headline statistics, it has weakened the national balance sheet while claiming to have improved it. This is Treasury dogma in its most unhelpful form.