A case for convergence?

With Basel II regulatory capital becoming increasingly more risk-sensitive, the value of a separate economic capital system must be questioned. Christopher Hall of Risk Advisors considers the differences between economic capital and regulatory capital, and asks whether they will ever converge

Economic capital was born about 15 years ago out of the desire to reflect risk in the decision-making processes of a bank. The two key questions an economic capital system was designed to answer were: how much shareholders' funds should the bank hold to achieve its desired credit rating, and how should the cost of capital be allocated to create appropriate risk-adjusted measures for strategic and tactical business decisions?

Basel II reflects these developments in banks by making regulatory capital more risk-sensitive and by reducing the regulatory capital arbitrages that were apparent under the Basel I rules. For example, Basel I gave regulatory capital relief for securitisations even though many of them transferred little or no risk.

Many banks now produce two risk-based capital figures: regulatory capital and economic capital. This article looks at some of the differences between a regulatory capital system designed to protect depositors and an economic capital system designed to enhance shareholder value, and asks if they are fundamentally different systems.

Regulator's objectives

The regulator must protect depositors and maintain the stability of the financial sector. This is a very clear debt-holder view of capital – that is, how much total capital (Tier 1 and Tier II debt) is required to ensure there is little chance of the depositors losing money from extreme downside risks.

Having sufficient capital buys time for the regulator to either influence banks to take action or to take direct action itself if the bank starts suffering financial difficulties and breaches its minimum regulatory capital levels.

The regulator does not have a duty of care to the shareholders or any direct desire to maximise shareholder value. However, the regulator does have a general interest in banks being profitable so that they are less likely to breach their minimum capital ratios. Hence, the regulator is only interested in one of the two original questions required from economic capital: how much capital should a bank hold?

Bank's objectives

A bank's view is rather different. Banks need to ensure that the level of shareholder funds is consistent with its credit rating in order to give confidence to its depositors and debt-holders. However, it is the bank's profitability and risk levels that are the primary drivers behind its credit rating, not its equity levels. Reviewing the factors in any credit grading tool will show how relatively little capital levels affects grade, a view that is also supported by the credit rating agencies. Although there is a relationship between Tier I levels and credit rating, this relationship is low. This is illustrated in figure 1, which shows that only 16% of the variability in probability of default (PD) can be explained by Tier I capital levels. Indeed, by just removing the two outlying observations for Commerzbank and HVB, the regression line would show a slight inverse relationship between Tier I levels and PDs.

Pillar II rules make it clear that 'supervisors should expect banks to operate above minimum regulatory capital ratios'. Hence, no matter how low a bank's economic capital figure may be, the regulatory rules will still be the binding constraint on total capital levels. In addition, the bank will need some capital capacity to ensure that it is unlikely to breach its minimum regulatory capital levels in a downside scenario.

Banks do need to assess the debt-holder perspective of risk to compare their downside risks against the regulatory capital rules for the Pillar II supervisory review; help determine the mix of equity versus debt; and assess how much capital capacity (such as profits, dividend cuts, Tier II raising, etc) is required to ensure that the bank does not breach its minimum regulatory capital in a severe recession.

But this is where banks' objectives can start to materially diverge from those of the regulator. The key question for a bank is how capital should be allocated and charged to business units and transactions. This is most important as it drives the performance measures, bonuses, transaction costing, motivational incentives, and hence the strategic/tactical business decisions within a bank.

A bank, like any commercial organisation, should be trying to maximise shareholder value, not debt-holder value. Indeed, under normal circumstances, maximising shareholder value is also consistent with maximising the value of all stakeholders in the company, including the debt-holders, which benefit from the increased profitability by maintaining their high credit rating.

Does allocating capital using the regulatory capital rules, or based on a very high confidence level, really create the correct management incentives to maximise shareholder value? Allocating capital to a very high confidence level does reflect the drivers of the extreme tail and hence, the drivers behind total capital requirements as defined by Pillar I of Basel II. However, allocations to extreme points in the tail have many issues, including:

• allocations are not very robust, particularly as they are very sensitive to changes in assumptions;

• allocations are very sensitive to risk concentrations – for example, exposures to a country, industry, individual large loans or material operational risk concentration;

• and noticeably more capital is allocated to the better-quality large corporates, which often makes them appear uneconomic. This is unintuitive and out of line with many valuation models.

An alternative method is to allocate capital in a manner consistent with the valuation models used within the businesses. Although there are a number of approaches, these typically use volatility measures to allocate capital across the group. These measures are consistent with traditional corporate finance valuation theory: beta and how to charge for the cost of equity. One of the main advantages of this approach is consistency between estimating group, business unit and transaction economic profit.

Most senior executives and board members have incentives to maximise shareholder value from their share options. It is therefore logical to allocate the cost of capital in a manner that will identify and motivate shareholder value creation, rather than to protect depositors.

In maximising shareholder value, the general macroeconomic risks are most relevant for decision-making. Conversely, will board directors or managers really pay too much attention to very low-frequency events (one in one thousand a year or less) in their decision-making?

Other practical advantages of allocating capital based on volatility include its simplicity – the fact that it is additive and produces intuitive results if it is applied appropriately. We also have considerably more data to estimate volatility, so the models will be more robust. This is important as it means the decision-makers are more likely to be able to debate the capital cost appropriate for an investment decision. Conversely, a debt-holder view is very complex – how many senior executives will understand copulas or the intricacies of Monte Carlo models in a manner that they could challenge the results?

Choosing the volatility approach to capital allocation does not mean debt-holder/regulatory considerations should be ignored, just that they should be managed in a different way. Regulatory capital limits and risk concentration limits should be set and monitored as part of the normal process of optimising the value of the bank, subject to constraints.

Precision

The UK regulator seems well aware of the difficulty of estimating extreme points in the tail of a distribution, and the many risks that are inappropriate to incorporate as a single economic capital number. Qualitative factors in assessing total capital requirements are just as important to the Financial Services Authority (FSA) as the few risk categories that can be evaluated reasonably well. Factors such as size, diversification, management quality, quality of the capital plan and non-quantified material risks are just a few of the many factors likely to be considered by the FSA when setting a bank's individual capital guidance (ICG) relative to its peers. The ICG represents the uplift from the 8% minimum regulatory capital to reflect the bank's specific risk profile.

In other words, assessing the total capital levels is as much an art as it is a science. This is a problem if a bank is trying to allocate capital from a debt-holder's perspective.

However, if capital is allocated within the group based on contribution to shareholder volatility, the process becomes much more objective and robust, especially if the emphasis is placed on the relative risk of different transactions.

Loss given default (LGD) estimates

PDs and LGDs change throughout the economic cycle, creating a pro-cyclical effect – that is, when profits are reduced in a recession from higher provisions, regulatory capital also increases.

Given that one of the regulatory objectives is to maintain confidence and stability in the financial sector, one would expect that the regulator would want to give banks some regulatory capital relief in a severe recession rather than penalising them. The regulator may agree to a mechanism to give some form of predictable regulatory relief as part of the Pillar II supervisory review.

Possibly in order to try to reduce the pro-cyclical effects, the regulatory definition of LGD is measured on an economic downturn estimate. However, there are issues with this approach for economic capital and valuation models. For investment appraisal decisions, banks should be using the expected cashflows – that is, taking account of expected economic forecasts and the variation around them. Using an economic downturn LGD will systematically overcharge loans, as in most years banks will not be facing a downside scenario.

When making investment decisions from a shareholder perspective, one should consider both the upside and downside scenarios, rather than just the downside as reflected in the regulatory definition. Similarly, for risk management instantaneous estimates of the current risks facing a transaction are required to make appropriate risk decisions about hedging, costing and forecasting actual provision rates conditional on the economic environment.

Probability of default

The ideal type of PD models to use may also differ between regulatory capital and economic capital. Broadly, there are two main classes of PD models, although many grading models fall somewhere between these two extremes. These are:

• Through-the-cycle models. These try to reflect the counterparties' expected state in a downside economic environment. Hence, these models need to take account of the current state of the economy. Through-the-cycle models are considerably more stable over time, and hence may be better for regulatory capital purposes in order to reduce pro-cyclicality or where long-term investment decisions are being made.

• Point-in-time models. These reflect the condition of the counterparty now. These models are more cyclical and need to be used in conjunction with grade transition matrices to predict how the grades may change over time. Linking the grade transition matrices to the current economic state should also give a good estimate of the expected loss and the level of projected risk given the current economic forecasts. This is very suitable for capital planning, scenario analysis and linking into valuation models. However, the potential volatility in PD resulting from this grading system will increase the pro-cyclical effects of regulatory capital, making capital planning harder to predict and manage.

Time horizon

The regulator has set a one-year time horizon from a debt-holder perspective to assess total capital levels. This arbitrary time period may be adequate for the regulator to buy time to resolve the situation if a bank is in financial difficulties, but does it make sense for a bank?

From a bank's point of view, the time horizon is more complex. At least two time horizons are needed in practice, even when considering the debt-holder view of risk.

Very short time horizons are required for specific risk – for example, a large credit exposure, a crystallisation of a material operational risk or rogue trading losses. These events occur suddenly and require sufficient capital to ensure that the bank will not breach its minimum regulatory capital levels over a short time period.

On the other end, losses from an economic downturn typically materialise over roughly two or three years. These types of losses, even though they may be much larger than a single specific risk loss, are more manageable. The loss may be absorbed through expected profits, the balance sheet may be managed down, and all banks are likely to be in the same boat together. There is certainly a degree of protection from the regulator if all banks act as a herd with regard to macroeconomic risks.

Both these time horizons should ideally be incorporated into the individual capital adequacy assessment process (Icaap) for the Pillar II supervisory review and give a good focus on the bank's real risk appetite for use in limit setting and reporting risk concentrations.

The shareholder perspective is once again easy to ascertain because it does not require a time horizon, as it is assessing the relative risk of different transactions through the cycle rather than trying to quantify the absolute loss amount.

Complexity and flexibility

The regulator needs to consider how banks might arbitrage a risk system if it is to be used for regulatory capital purposes. Hence, many angles need to be considered, including governance, testing, full documentation and how the measures are used in practice within the bank. This is clearly needed if the internal risk systems of a bank are to be used for regulatory capital purposes, but it makes the regulatory rules complex and very long.

However, the regulatory capital calculations are still a simplification of the real world. One such area is in credit risk, where little distinction is made between the sizes of counterparties. Large corporates are more diversified and hence have higher asset and default correlations compared with similarly graded smaller corporates. This has a material effect on the level of risk in a well-calibrated economic capital model from both a debt-holder and shareholder perspective, yet this appears to be largely ignored in the regulatory capital calculations.

This inflexibility of the regulatory system is also an issue for banks. Would any bank really want to run its performance measures on a system that does not reflect all the drivers in their risk models?

Hence, one key advantage of economic capital either from a debt-holder or shareholder perspective is that it is under the bank's total control. Enhancements can be added and implemented so the motivational impacts are immediately apparent to the bank. There is no need to wait either to educate the regulator or to update the Basel rules, which could take years.

There are many other potential differences between regulatory and economic capital. Regulatory capital is firmly fixed in accounting definitions of capital. Economic capital is increasingly being based on market values, which is more appropriate for decision-making. Although there are linkages between the two measures, they are generally not well-understood or reflected in banks' systems.

The treatment of liquidity is also an issue between the two perspectives. From a debt-holder/regulatory view, there is clearly a benefit from a liquid asset, as it can be sold to manage down the risk and capital requirements in a downside scenario. This ability to reduce risk and regulatory capital also needs to be reflected in the capital plan within the Icaap. What is not so certain is whether there should be a liquidity discount for liquid assets within the capital allocation systems. If two assets have the same risk characteristics, PDs and cashflows, should they have the same value?

Conclusions

It would be great if one risk measure could be used for all business decisions. Unfortunately, life is not that simple, and there are material differences between the analyses required to assess the group's capital requirement from a debt-holder perspective; reflect risk considerations in the capital plan; make good portfolio management decisions; and allocate capital in order to motivate shareholder value creation.

Estimating the extreme tails of a distribution is an imprecise science. No matter how sophisticated and complex the models, realistic estimates of extreme confidence levels will never be either accurate or reflect all aspects of risk. It is the regulator's primary responsibility to determine minimum capital levels, not the bank's. However, the regulator will need to draw information from a bank's risk systems so that it may make an assessment of a bank's extreme downside risk from a debt-holder perspective compared with its peers.

Conversely, a good allocation system should be relatively simple, intuitive and create incentives to maximise shareholder value. The aim is to identify and cost the main risks in the business from a shareholder perspective. This should help to maximise the shareholder value, boost profits and boost the market value of the shares. As a consequence, the rating agencies should reward the bank with a good credit rating.

The good news is that the data required for both the shareholder and debt-holder perspectives is very similar, although it needs to be processed in different ways. Once systems are created, it should be possible to produce both views from the same data.

Christopher Hall is the director of Risk Advisors, specialising in training, advice and systems on economic capital, risk-adjusted performance measurement, value-based management and operational risk. Email: [e-mail]