Banks should not be able to game accounting rules
A stack of ten euro banknotes is arranged with fifty and twenty euro banknotes for a photograph inside a Travelex store, operated by Travelex Holdings Ltd., in London, U.K., on Monday, Jan. 12, 2015. The euro approached a nine-year low against the dollar as European Central Bank officials fueled speculation the institution will start a program of government-bond buying as early as next week to stave off deflation. Photographer: Simon Dawson/Bloomberg©BloombergW
hen it comes to banks, bad accounting practices can contribute to financial instability. Booms flatter their measured profitability, which encourages them to take more assets on to their balance sheets. Thus leverage begets more leverage throughout the banking system, until asset prices can rise no longer and the whole edifice comes crashing down.
Accounting should aim off for this natural tendency to spiral into debt-fuelled expansion. But far from doing so before the financial crisis, international rules leaned the other way.
For instance, they gave banks the ability to shift assets between the trading book, which was marked to market, and the banking book, where they were assumed to be held to maturity and not sold at market prices.
In good times, this allowed the banks to mark ever more assets to market and book the profits that emerged — even if these were unrealised. Bad times brought the opposite incentive: banks squirrelled assets away into loan books. That allowed them to avoid realising ugly losses or accounting for the higher risk that they might occur.
Nor was that the only snag with the old regime. A second problem concerned conventional loans that were held in the banking book. Provisioning against the value of these assets was only required when there was evidence of impairment, such as a missed payment. At best this made banks slow to react to emerging problems when the financial crisis struck. At worst they allowed outright concealment of economic realities; fair value gave way to the fictions of “extend and pretend”.
Granted, the authorities have acknowledged the need for significant changes. Since 2009, the rule setters have sought to eliminate some of the procyclicality and opacity to which the old system was clearly prone. New rules — in the shape of IFRS 9 — will be introduced in much of the world outside the US in 2018. Banks will have less scope for moving assets from one book to the other. As for the loan book, the directors will have to take a provision for an advance as soon as it is made, and more later on if the chances of a problem have grown.
Nonetheless the planned reform has only softened the old system’s perversity. It has not been done away with. Not only will asset-juggling from one side of the house to the other still be possible, the rules on provisioning are both unnecessarily complex and too weak. They do not for instance require a bank to take a provision against the lifetime expected loss on the loan, but just an estimate of the likely loss in the next 12-month period. On a book of 25-year mortgages, that is akin not just to guessing the outcome of a football match, but the precise times at which the goals will be scored.
Accounting should be tougher and less prone to idiosyncratic judgments. A simple prudential rule could be introduced that would reduce the leeway to put assets beyond mark-to-market valuation. All trading assets could be valued at the lower of cost and net realisable value. Further, in the loan book, provisions should take into account not just the next 12 months but the lifetime outcome of the loan. This would give investors a far truer and fairer view of the value of those assets the balance sheet contains.
Bankers complain that a tougher regime might force them to realise more losses in the short term. Tough. The special nature of banking demands a sound accounting regime. The existing plans go some of the way but not far enough. Investors should monitor more closely what financial institutions do. But if they are to perform this task, they need the tools to do the job.