Speech by Hans Hoogervorst at 3rd ECB Conference on Accounting, Financial Reporting and Corporate Governance for Central Banks

04 June 2012

On 4 June 2012 Hans Hoogervorst, Chairman of the IASB, addressed the 3rd ECB Conference on Accounting, Financial Reporting and Corporate Governance for Central Banks in Frankfurt.

"In debates on accounting, one particular issue keeps on popping up. This hotly debated question is whether the primary purpose of financial reporting should be to provide transparency, or if it should also serve the goal of stability.

In this debate, transparency and stability are often juxtaposed as if they were conflicting goals. I think that this is essentially a false and counterproductive contradiction. In my view, it is clear that transparency is a necessary precondition of stability. Indeed, a lack of transparency significantly contributed to the credit crisis. Huge risks were allowed to build up both on and off balance sheets without being noticed. Without proper transparency about risks, stability is bound to collapse in the end.

In short, stability is not the same as transparency, but there can be no durable stability without transparency. So accounting standards/financial reporting can contribute to stability by enhancing transparency. Before I make clear how we intend to do so, let me make also perfectly clear what we cannot do.

Stability should be a consequence of greater transparency, but stability cannot be a primary goal of accounting standard-setters. It is not our remit and we simply lack the tools for fostering stability. For example, we cannot set capital requirements for the banking industry. This instrument belongs to the prudential regulators and central banks which do have stability as their main mission.

What accounting standard setters can also not do is to develop standards that make items appear to be stable when they are not. And, quite frankly, we are sometimes suspicious that we are being asked to put a veneer of stability on instruments that are inherently volatile in value. Our standards should not create volatility that is not already there economically. But, if volatility exists, our standards should certainly not mask it.

That being said, there are plenty of ways in which we are trying to make a contribution to greater transparency in the financial industry, often in close consultation with the prudential community and regulators, such as the Basel Committee and the Financial Stability Board.

First, the accounting standard-setters have improved consolidation requirements to prevent undesirable off-balance-sheet financing. In particular, US GAAP was tightened up in this respect. While the broad consolidation principles of IFRS held up reasonably well during the financial crisis, in the United States off-balance-sheet financing through special purpose vehicles and repo transactions was more of a problem. With tighter consolidation requirements and better disclosures, we can reasonably hope that this problem will now be a matter of the past.

The use of fair value accounting has been the biggest bone of contention between accounting standard-setters on the one hand and prudential and central banking authorities on the other hand. Opponents of fair value accounting state that too much reliance on market prices exacerbates the economic cycle in both upturns and downturns. These critics believe that fair value accounting strengthens pro-cyclicality and thus leads to artificial volatility, which threatens stability.

This line of reasoning was greatly reinforced by the fact that the efficient market hypothesis was heavily discredited by the financial crisis. The ECB and the Basel Committee asked that we limit the use of fair value to address this pro-cyclicality.

As a former Minister of Finance, as well as a former regulator, I have always been sceptical of the efficient market hypothesis. Too often, I have witnessed that markets can go very crazy indeed, especially in the short run. However, if you operate in a market environment, you had better be prepared for markets to go loony every now and then. I have always been amazed by bankers telling me that market information cannot be relied upon, while they themselves are major players in that very market, or even market-makers!

Moreover, it is hard to imagine an industry that is as prone to volatility as the financial sector. Both sides of a bank's balance sheet are vulnerable. Its assets can be very sensitive to the economic cycle, whether they are derivatives or loans backed by bricks and mortar. Even gold-plated, triple-A government bonds can turn sour very quickly, as we have seen in the case of Ireland.

The banking industry's liability side is also notoriously vulnerable. Funding, whether it is wholesale or retail, can evaporate with the speed of a mouse-click. As if this is not risky enough, the banking industry has been allowed to operate on the flimsiest of capital margins. The capital cushion of the banking industry has shrunk dramatically in the last century. Just before the crisis, tangible common equity of many banks was negligible. It was generally only 1 to 3 per cent of the balance sheet, and sometimes even below that.

In conclusion, the recent volatility was inherent to the financial sector's business model. If accounting requirements had a role to play, that role was at most only as a minor actor. Indeed, many independent studies have concluded that fair value accounting played at most a very minor part in the turmoil of the financial crisis. That conclusion was only to be expected, given that the bulk of traditional banking assets (e.g. loans) are still valued at amortised cost.

The IASB has decided to continue with a mixed measurement model in IFRS 9. In IFRS 9, financial instruments that have basic loan features and that are managed on a contractual yield basis are measured at amortised cost. For such instruments, amortised cost is deemed to provide more relevant information than short-term market fluctuations.

The IASB is currently reconsidering limited parts of IFRS9.

We recently decided to re-establish a fair value through OCI category for debt instruments that are managed with the objective of both collecting the contractual cash flows and selling the assets. This can be the case for assets that are held for liquidity management. But assets that are solely held to collect contractual cash flows — among which vanilla debt instruments — will continue to be measured at amortized cost. In this respect there is no fundamental difference from our previous proposals.

The last area of transparency I would like to discuss today relates to impairment. A well-functioning impairment model is of paramount importance for an amortised cost measurement to be reliable and credible. After the outbreak of the crisis, our current impairment model, which was based on incurred losses, was criticised for being too little, too late.

We think that this criticism was partially justified. The fact that the market capitalisation of many banks is far below their book value is an indication that market participants do not believe that their current level of provisions reflect economic reality.

I say partially justified, because I am convinced that the incurred loss model could have been applied much more vigorously in the last couple of years. In current circumstances, I do not think that there is a lack of triggers to start writing off certain assets. There has been simply too much hesitancy to do so or political pressure not to do so.

The very late write-downs of Greek government bonds are a case in point. Despite severe market dislocation, repeated downgrading and steep discounts of Greek debt, most banks only started provisioning when a restructuring decision had been taken. And even then, some banks thought an allowance of 21 per cent was enough. In sum, we are convinced that even the current impairment rules allow for much more decisive measures.

Nonetheless, both the IASB and the FASB are convinced that we need a more forward-looking impairment model. In fact we are well on our way to completing an expected loss model.

The basic principles of this model are as follows. From day 1, for all new financial assets, an allowance balance needs to be built up that captures the expected losses in the next 12 months. If credit quality deteriorates subsequently to such an extent that it becomes at least reasonably possible that contractual cash flows may not be recoverable, lifetime losses need to be recognised. We will not try to define exactly what ‘reasonably possible' means, but it primarily refers to the inflection point when the likelihood of cash shortfalls begins to increase at an accelerated rate as an asset deteriorates.

To some degree, this expected loss model will rely on judgement, because it is not possible to predict with precision when the probability of default starts to accelerate. To arrive at this judgement, market indicators can and should play an important role. So even for assets that are measured at amortised cost, fair value can contain very important information.

Take the example of sovereign debt securities. If a sovereign's debt is faced with clear sustainability issues, sinks below investment grade and suffers from double-digit market discounts, clearly there is a serious possibility that contractual cash flows will not be paid in full. A lifetime loss will probably need to be recognised, even if the securities in question are still being serviced. One just has to look at current market conditions to realize this model would lead to a much more timely recognition of losses than is currently the case.

I know that many prudential regulators hope that an expected loss model can serve to dampen the cycle of credit booms and busts. As a showcase, they have often pointed at the dynamic provisioning model that Spanish banks were using before the crisis broke out. While dynamic provisioning contained some elements of an expected loss model, it clearly was not able to adequately counter the cycle.

For the following reasons, I believe we have to keep our expectations realistic about the anti-cyclical effects of accounting rules. First of all, accounting standards are not an instrument of economic policy; they merely serve to depict financial and economic reality as reliably as possible. Dampening the economic cycle is neither our task nor within our area of expertise.

Secondly, as I said before, the expected loss model relies to some extent on judgement. Before the present crisis, many banks and their supervisors obviously were not able to perfectly anticipate risk. Even where the writing that warned of a full-fledged credit orgy was clearly on the wall, the magnitude of problems to come was not predicted.

Given the fact that economic history is littered with credit bubbles and busts, there is no guarantee that future bankers will do a much better job of anticipating risk than current bankers. So it is not likely that all the risks that are building up during an economic boom will be recognised in time. Even with an expected loss model, many losses will only become apparent when the economic downturn sets in.

Once a credit bust erupts, risks tend to crystallise on a massive scale. The current situation in Spain is a case in point. Since the outbreak of the crisis, Spanish banks have written off assets to the amount of some 18 per cent of GDP. Some think that more is still to come. The dynamic provisioning of the Spanish banks was completely overwhelmed by the magnitude of these losses.

The lesson is that economic cyclicality can be too powerful to be dented significantly by mere accounting. Nevertheless, I am convinced that the introduction of our expected loss model will be a major improvement.

First, it should lead to provisions being made in a more timely and realistic fashion and a heightened, more forward-looking risk awareness in the financial industry. Secondly, a timely clean-up of the banking system should free up resources to viable sectors of the economy instead of exercising forbearance on essentially defunct companies.

Thirdly and perhaps most importantly, there is nothing more damaging to the credibility of the financial sector than serial underestimation of the true magnitude of problematic assets. Partial recognition of inevitable losses may buy time in the short run, but in the end leads to round after round of ‘definitive' rescue programmes and a gradual erosion of confidence in the markets.

It is obvious that for a rigorous and adequate application of the expected loss model, banks need to be properly capitalised. Whether the recent reforms of the Basel regime for capital requirements go far enough in this respect is open to debate.

It is well documented how, before the crisis, the Basel capital ratios had been gamed to increase leverage by exploitation of the risk weights. Banks with a seemingly sound Tier-1 ratio of 10 per cent could in fact be leveraged 40, 50 or 60 times! Instead of being a source of transparency, the Basel ratios had been abused as a scheme for hiding leverage.

Basel III will undoubtedly be a great improvement, because it enhances the capital requirements both quantitatively and qualitatively. Moreover, the introduction of a leverage ratio will give more insight in what the true gearing of a bank is. Yet, under Basel III, a bank is still allowed to be leveraged 33 times. I am not a prudential regulator, but I truly wonder if a bank with leverage of even just 20 times can accommodate a crisis of Spanish or Irish proportions.

In addition, the system of risk weighting of assets is still fraught with risk. It allows banks to assume that sovereign debt has little or no risk, which by now we should know is highly doubtful. Both Spain and Ireland had very low levels of public debt and still they lost their triple-A rating almost overnight.

Not only the risk weighting, but also the absence of a large exposure regime for highly rated sovereigns, greatly encourage banks to load up on sovereign risk. Once a sovereign enters the danger zone, the needs for provisioning may explode dramatically. This is of course not an accounting problem, but a real prudential problem. With an expected loss model, this prudential vulnerability will be exposed sooner than is currently the case. That is in itself a good thing, but banking supervisors had better be prepared.

In conclusion, I believe that the introduction of an expected loss model can lead to a much more timely recognition of losses than is currently the case. The incurred loss model provides too much leeway for procrastination and has to go. But an expected loss model in itself should not be expected to significantly dent the pro-cyclicality of the credit cycle.

Unless bankers and their supervisors become a lot better at containing credit booms and their risks, busts with massive losses will periodically take place. Even then, an expected loss model is preferable to an incurred loss model. But for an expected loss model to be applied rigorously, it is essential that banks are well capitalised. If such is not the case, even law-abiding banking supervisors might be tempted to buy time by condoning some stretching of accounting rules. Obviously, that is a temptation to which nobody should be exposed."