Recognising the cost of loans upfront – Eurostat’s intervention

As a follow-up to yesterday’s post I want to look in more detail at the written evidence Eurostat submitted to the Economic Affairs Committee at the beginning of February.

Eurostat have clearly seen the evidence submitted ten days earlier by the Office for National Statistics. And they have some differences to register: what ONS say is in line with the accounting ‘for standard loans’ but in its opening summary paragraph, Eurostat sets out its position (my emphasis in bold):

The recording rules summarised by the ONS do not cover the cases where loans are not expected to be repaid: [European System of Accounts paragraph] 20.121 prescribes in this case to record a capital transfer at inception … .
Thus, statisticians should formulate an initial assessment regarding the probability of recovery of loans at the moment they are granted. For nonstandard loans, Eurostat recommends to assess the expected losses on the loan portfolio at inception, and to record a capital transfer for the portion that will probably not be recovered, and only the remainder as a loan.

The relevant ESA paragraph is omitted from the submission, I give it in full for reference:

20.121 Loans include, in addition to loans to other government units, lending to foreign governments, public corporations, and students. Loan cancellations are also reflected here with a counterpart entry under capital transfer expenditure. Loans granted by government not likely to be repaid are recorded in the ESA as capital transfers, and are not reported here.

Eurostat is saying that the government should be declaring the expected losses on student loans upfront not delaying them until accounts are closed. Eurostat indicates that this treatment of loan write-offs is suitable for unexpected bad debt, not on deliberate subsidies. 

The significance of the reference to “capital transfer” is that such transactions are recorded as spending today. Grants (or gifts) are capital transfers.

Eurostat goes on to reiterate its main point but underscores how murky a resolution might be:

The ESA 2010 accounting rules for loans are presumably applicable to student loans. There is however some flexibility, to a certain extent, depending on the expected future losses on the loan portfolio in question. The document provided by the ONS summarises the impact of the different transaction in public sector accounts according to ESA 2010 in relation to ‘standard’ loans, without taking into account the expected future losses on the loan portfolio.

 Eurostat are emphasising that UK student loans are “nonstandard”, not by virtue of their income contingent structure, but because of the subsidy knowingly built into the scheme; what Eurostat elsewhere refers to as an “original intention to convey a benefit“. Now most people don’t consider the current fee-loan regime to be very beneficial, but the terms of student loans are “generous”, when compared to commercial loans.

It’s precisely this comparison that is pertinent: the decision to treat student loans using conventions for standard loans (“financial transactions”) gives a huge presentational advantage to government, as the recording of long-run costs bound up in the deliberate write-off subsidy are deferred for decades.

Given that the government only expects part of the initial cohort loan outlay to be repaid (official figures state that it loses the equivalent of 45% in net present value terms), ONS needs to look again at the conventions in use.

Tellingly, Eurostat outlines a specific method, whereas ONS had told the Economic Affairs and Treasury Committees that there was no sensible alternative.

Our problem is that the Eurostat submission is a pithy four-pages and a number of crucial issues need unpicking.

Eurostat’s method involves taking the initial outlay and splitting it into two components:

• the portion likely to be repaid can be treated as a loan, in the way that ONS does now;
• the position unlikely to be repaid, the expected loss, is treated as “a capital transfer at inception (ESA 20.121), impacting the deficit”.

Those who’ve been following the student loan debate closely will immediately think about the Resource Allocation and Budgeting[RAB] allocation and charge. This is used in departmental accounts but not in national accounts. RAB cannot simply be plugged in to national accounts, because it is a net present value impairment not a nominal cash figure: net present values are not used in national accounts, which do not recognise any form of discounting.

Some other method would have to be devised and that involves a number of decisions, which are presumably now the focus of the ONS’s collaborative review.

Firstly, loan repayments are not differentiated as payments against interest or payments against principal (the original loan outlay), but lumped together. Eurostat may require ONS to make a determination here.

Secondly, the related point: if you don’t know whether repayments are covering interest or principal, which is predominantly written off? We are in a particularly perverse situation in that if you treated all repayments as principal, then just about enough repayments come back across three decades to match the cash value of the original outlay. But this leaves billion upon billion of interest written off. If you were to treat all repayments as servicing interest (and annual repayments are always below interest accruing), then no principal would be repaid and the principal would seem to be gifted … and the accounts would then have to recognise that as a capital transfer. In sum, are losses determined by original loan outlay or should we also consider interest accrued but foregone?

Eurostat need to be clearer on the difference between writing off interest and writing off principal, but, thirdly, they also need to clarify what they mean by “losses”. Their wording makes it difficult to tell whether they think the test is whether borrowers clear their balances (what proportion of accounts clear), whether the issue is overall level of repayments across the scheme (what portion of principal comes back), and whether the government’s own costs of borrowing should be factored into these calculations. In the determination of the RAB charge, the impairment it covers equals original loan outlay minus discounted expected repayments – interest figures only insofar as it generates higher repayments. That’s not currently the case in national accounts, where interest accruing is currently recorded as income and write-offs as expenditure: the final net position giving the cash loss or surplus on loans.

Fourthly, the government’s cost of borrowing (it borrows to create loans) is not hypothecated in national accounts. How should the assessment of “expected losses” take that into account if we can’t use RAB?

These questions are given further edge by the convention outlined by Eurostat.

The above described apportioning at inception (case of non-standard loans) will lead to no recording of accrued interest on the portion that has been considered as a capital transfer.

If we return to the four considerations above, dizziness ensues.

If all repayments first service interest, then all principal is written off: does that mean it should have been a capital transfer and therefore no interest charged?

But if all repayments first service the principal, interest can be charged and treated as income, but then is all written off at the end … because it was never received.

And what’s not clear in the Eurostat submission is, crucially, what this would mean for borrowers.

The fundamental point of National Accounting is that there is “symmetry” between lender (government) and borrower. If the government cannot record interest against a portion of loan outlay determined to be a capital transfer, what does that mean for the interest accruing on the borrower’s balance?

Loan outlay plus interest accruing minus repayments must equal the borrower’s outstanding balance – that’s an accounting identify. You cannot record different levels of interest accruing for the borrower and for government, without some additional balancing transaction being introduced. But what would that be? I have asked Eurostat whether their submission implies that interest terms on loans would have to change to accommodate their recommended treatment.

It should be noted that recording a portion of the loan as capital transfer does not impact on the ability to receive repayments against that portion: these would be classed as (unanticipated) “recoveries” and recorded as income when they happen. Loan repayments are not currently recorded as income.

Conclusion

It seems clear that the ONS loan accounting review was prompted by this intervention from Eurostat. ONS told parliament in January that its treatment was correct and that the only alternative was the rather poor idea of treating all loan outlay as capital transfer and all repayments as recoveries to commencing work. But it began work on a review in March (which it formally announced at the end of April).

Eurostat is firmly of the opinion that:

“The treatment of schemes where a small proportion of loans is eventually expected to be never repaid should differ from the treatment of student loan schemes where significant repayments seem not likely. In the latter case, a grant at inception should be considered … .”

Does this mean that accounting has to change when policy decisions make losses more significant?

Theresa May’s decision to raise the repayment threshold to £25,000 has moved the estimated write off on loans to 45% from somewhere around 30%; neither of those seems to be a “small proportion”.

In February, the Department for Education was granted £14.7billion of additional resource to cover the downwards revaluation of the existing stock of loans resulting from the repayment threshold change. What’s clear on Eurostat’s proposal is that this has the potential to impact the deficit today as capital transfer!

In sum, things are quite confused, not just for borrowers, but for accountants and statisticians. And I should add that another official body (who I can’t name at this stage) is about to weigh into the debate.

This all presents the current HE Review with some difficulties: it’s meant to make suggestions consistent with deficit reduction, but what constitutes deficit reduction will be revised by the results of this review.

On timing: the ONS review apparently needs to be concluded before the appearance of the next edition of the Government Manual for Deficit and Debt, which should appear before the next accounting year starts in April 2019.

Whatever happens, we now have a specific alternative on the table, but one that needs some work to fit with our current student loan scheme.

In principle what we want is a convention that reflects upfront the expected long-run costs but defines those consistently and maintains something close to the clarity of the nominal transactions currently used: loan outlay, loan repayment, interest accruing, write-offs.

In the short term, this may not be what anyone wants as what seems clear is that HE will have more impact on government spending today, which will probably lead to cuts to loan funding.

More optimistically, presenting the costs of loans properly makes alternatives look less expensive.

The overarching problem here is general ignorance about what “the deficit” is. A focus on headline statistics at the expense of sensible economic and fiscal policy plagues contemporary politics. And the Treasury method of budgeting is a real culprit: if start from the deficit you are prepared to run, then backfill policy spending, you are prone to letting accounting play too large a role in policy-making.