STUART THEOBALD: New accounting rules make comparing banks’ performance with 2009 hard

Companies must now expense the losses they expect to make as soon as their expectations change.

This column was first published in Business Day. 

Bank profitability has predictably been hit hard by the Covid-19 crisis. Profits have slumped between 43% (Standard Bank) and 82% (Absa) in the first half of the year.

The issue is how to judge this in context. Sure, profits have slumped, but the banks remained profitable and their financial strength has not been compromised. Nevertheless, I do wonder if this has been the worst performance in modern financial history.

It looks worse than during the financial crisis. In 2009, Standard Bank’s headline earnings fell 17%, Nedbank’s 30%, FirstRand’s 30%, and Absa’s 24%. But accounting rules have changed since, making it difficult to tell.

Impairments, the amounts banks must set aside for bad debts, were made according to IAS 39 (International Accounting Standards) then, whereas now they are made according to IFRS 9 (International Financial Reporting Standards). The detail of accounting standards can be mind-numbingly dull, but the differences make comparing financials between the two crises tricky. IAS 39 was based on an “incurred loss” model, but IFRS 9 is based on “expected loss” — bankers must expense the losses they expect to make, rather than the losses they have actually made, as soon as their expectations change.

You can imagine how this affects the figures banks are reporting. Even though the banks are reporting on the financial results to end-June, they have incurred huge costs for bad debt. Typically, a loan only counts as defaulted after a client has missed three months’ payments. With the crisis only sparking in March, the numbers of defaulting clients had thus not materially increased by the end of the financial period. But the banks are seeing what’s coming and ratcheting up impairments to pay for it.

Exactly how much the banks should be putting aside has been a matter of debate worldwide. The problem with IFRS 9 is that it is procyclical, amplifying economic shocks. In response to a bleak economic outlook, banks have to ratchet up provisions, which reduces their profits, leaving less capital to support lending, which decreases investment into the economy, causing a bleaker economic outlook.

Regulators have thus been walking a fine line in telling banks to go easy on hiking IFRS 9 provisions while also telling them to stick to the rules. The Bank of England briefed bank managers to not “kitchen sink” provisions. Locally, the Reserve Bank published several directives and guidance notes for banks on how to apply IFRS 9, telling them that loans they restructured due to the Covid-19 crisis don’t automatically represent higher credit risk, and that banks should “avoid procyclical assumptions in their IFRS 9 expected credit loss modelling”.

There was a remarkable language game at work between the Reserve Bank and the banks, in which the subtext was: we would never tell you not to fully apply accounting standards, but to the extent possible, use your discretion to not tank the economy.

As it happens, the banks that have reported so far took different approaches. Absa went furthest, booking credit impairment expenses equivalent to 2.77% of its loan books (up from 0.79% for the same period a year ago). Standard Bank took the lowest charge, at 1.69% of its book (0.76%) and Nedbank was in the middle at 1.94% (0.70%). These were higher than in 2009 when Standard Bank took 1.60%, Nedbank 1.49% and Absa 1.73%. FirstRand will report this week.

Payment holidays

Included in the provisions were management “adjustments” and “overlays” that reflect the discretion of the banks in putting aside money to cover expected losses. Billions of rand were set aside beyond what the models, which rely substantially on historic data, were saying needed to be incurred. Standard Bank took the least conservative approach, allocating just R500m to a central reserve for future losses, though arguably its loss models are more forward looking than the other banks. Absa added an extra R3.6bn of “management adjustments” to its retail and business banking book alone.

These decisions will now be tested by the performance of bank books. A key area to watch is the customers that have received some form of support during the crisis. It is astounding just how many customers had received some type of relief from their banks in the form of payment holidays or other kinds of loan restructuring.

For example, Absa provided some form of relief to 22% of its R975bn in loans, equivalent to R216.8bn. Nedbank provided relief to 15% of its loans and Standard Bank to 18% of its personal and business banking book. These loans will have to go back to fully performing or much greater provisions may be required in future.

The banks are guardedly optimistic. New business volumes have recovered quite well post the level 5 lockdown. As some of this will be pent-up demand, it will take some time to see what the new baseline for business volumes is. But ultimately it is the bad debt performance that will remain the critical issue for the crisis.

• Theobald is chair of Intellidex.