Thursday 14 October 2010

The Capital Adequacy Ratio – How Accurate a Measure?

Hard to Tell the Time With Just One Hand
In a post on the Booz & Company study on Kuwaiti banks' stress tests, I said that The National Bank of Kuwait was the most creditworthy bank in Kuwait.

Advocatus, one of the frequent readers/commenters on this blog, challenged that statement noting that as of 31 December 2009, NBK had the lowest CAR 15.03% versus 15.9% for Gulf Bank, 17.23% for Ahli Bank of Kuwait and 18.22% for Commercial Bank of Kuwait.

I'd like to address his remarks in a bit more detail than is convenient in a comment reply. So I'm moving the discussion to the "main board".

My intent is to reply with two posts. The first theoretical. The second specific to NBK and a few other Kuwaiti banks.

In this post I will discuss some shortcomings in the CAR which make it an inaccurate as well as incomplete measure of financial strength.

To start off a bit of introduction.

What is the CAR?

It is a ratio which measures equity capital against risk weighted assets ("RWA"). Neither the numerator nor the denominator in this ratio are the same numbers that appear in the financial statements. Rather both are constructed according to certain rules set forth in Pillar 1 of the Basel II Framework. A document of some 192 page. As that suggests, the rules are complex. Rather than delve into too much detail, I'll give a broad overview.

Regulatory Capital

The first step in determining regulatory capital is the allocation of a bank's accounting capital into one of three tiers. Tier 1 is considered the most solid. The others progressively less so. For example, common equity is Tier 1. Subordinated debt Tier 2. The amounts of Tier 2 and Tier 3 capital that a bank can use in calculating its regulatory capital for the CAR are limited. Tier 2 may be no more than 100% of Tier 1. Medium term subordinated debt may not exceed 50% of Tier 2 capital. However, to be counted even at the 50% level it must have an original maturity of five-years. If the maturity is shorter, it does not count. If the subordinated loan is a bullet loan, each year it must be reduced by 20% for CAR purposes. Tier 3 capital consisting of short term subordinated debt can be only used to support market risks and only to the tune of 72.5% of such risks. Any amount over these limits for Tier 2 and Tier 3 capital is not counted as part of regulatory capital for the CAR.

Once these have been determined there are a series of further adjustments. Goodwill is deducted from regulatory capital as are significant minority investments (which are deducted 50% from Tier 1 and 50% from Tier 2) and excesses over the 15% single obligor (Group) credit exposure limit. The full amount of equity in insurance firms may be required to be deducted from regulatory capital at the discretion of the national regulator. There are also a variety of haircuts on revaluation reserves for both property and securities.

Risk Weighted Assets

Basel II classifies risks (for the CAR) as either credit, market or operational.


The Basel Framework allows banks three options for determining Credit Risk. 
  1. A Standardized Approach (the bank is given formulas to apply to the face value of its exposure). 
  2. An Internal Ratings Based Foundation Approach ( the bank determines the Probability of Default for customers but uses the Loss Given Default, Exposure at Default, and Maturity as mandated in Basel II). 
  3. An IRB Advanced Approach (here the banks determines all these factors). 
For Market Risk there are two basic approaches.
  1. The Standardized Approach (banks use the set of factors given in the Basel II Framework to determine risk weighted "assets" for market risk)
  2. A Model Approach (generally VaR). 
 For Operational Risk there are three approaches: 
  1. The Basic Indicator Approach
  2. The Standardized Approach and 
  3. The Advanced Measurement Approach
Most banks in the MENA region use the Standardized Approach for credit, either the Standardized or a Model for Market Risk and the BIA or Standardized Approach for Operational Risks. As a general rule, Central Banks in the area prohibit use of the more advanced models: the two IRB approaches for Credit Risk and the Advanced Measurement Approach for Operational Risk.

One way to get a quick sense of how this system works is to look at the Basel II Pillar 3 disclosures of Kuwaiti banks. For 2009, you'll find NBK here (starting on Page 33), CBK here, and Gulf Bank here (from page 17).

What's the purpose of the CAR?

Regulators are charged with preventing problems at a single bank or group of banks from having a systemic impact on their economies. Accordingly, Basel II is designed not only to measure risks (to give early signs to the regulator of impending problems) but by the  very act of measurement itself (expressed in the CAR) to dissuade banks from certain types of activities which are felt to be risky. This is achieved by raising the risk weights for these activities.

It's important to understand this function. It is not merely a measure of creditworthiness. It is also a control device. And since the stakes are high, the controls are generally set with large safety margins.

What are the major issues with the CAR?

Capital

From the above discussion it should be clear that regulatory capital does not consider all the equity support a bank may have. One simple example: If subordinated debt at Bank A is twice its Tier 1 capital, half of that subordinated debt is not factored into the CAR. Yet, if Bank A encounters problems, that "excess" is still subordinated to other creditors providing an element of protection.

More importantly, one critical aspect of capital the quality of retained earning is not considered. And often is not considered on a consistent basis.

One example are changes in the fair value of financial instruments. If a bank takes these changes to income (Fair Value Through Profit and Loss – FVTPL), they are considered a part of Tier 1 capital dinar for dinar. If the same financial institution were to take the changes in fair value to equity (Fair Value to Equity), this income would be considered part of Tier 2 capital and would be haircut 55%. Calling these earnings FVTPL or FVTE does not change their fundamental economic characteristics. Yet for calculation of the ratio it does. 


One might also argue that changes in fair value are not the same as cash earnings. If these changes are deliberately inflated or markets are in a period of irrational exuberance which reverses, equity can vanish with astounding rapidity. 

Classifications of bond holdings as "banking book" versus "trading book" can result in different income and fair value treatments for exactly the same securities. Again the economic substance is the same, yet the CAR will differ. Same asset. Same economic reality. Different income and equity treatments.

Assets


Credit Risk
 
As indicated above, the mere act of classification of an asset can have an impact on its value. The same with its risk weighting. Securities held in the trading book generally have lower risk weights than those held in the banking book.

But the key issue for banks is the Standardized Approach for RWA for loans, the major earning assets of most commercial banks.

Under the Basel II Standardized Method for Credit Risk, risk weights are assigned to loans to private sector companies based on their credit grades (if any). 

  1. AAA to AA- receive a 20% risk weight
  2. A+ to A- a 50% weight
  3. BBB+ to BB- a 100% weight 
  4. Below BB- a 150% weight
  5. If the loan is unrated, a 100% weight. 

You can well imagine that weaker credits avoid ratings like the plague because if one is objectively a B credit but not rated, one gets the same risk weighting as a BBB- to BB+. Again the presence or absence of a rating (but not the underlying economic reality) drives the CAR.

More importantly, the following are not considered in determining the risk weight of a loan: the final maturity, the pattern of principal repayments, and the pattern of interest payments. Thus, a five year loan to Company XYZ has the same risk weight as a one year loan. A five-year bullet loan with quarterly interest payments is the same as a five-year zero coupon loan and is the same as a loan with quarterly principal and interest payments. The purpose of loans is not considered. A loan for real estate speculation is treated the same as a loan for a factory. Many forms of collateral are not recognized for CAR risk mitigation purposes, e.g., real estate, except the latter gives a partial mitigation for residential mortgages. Yet, the bank does have recourse to that collateral in the event of a borrower default and is more protected than one that has no collateral. Yet, the CAR does not reflect this.


Operational Risk
 

For Operational Risk, banks are generally limited to the Basic Indicator Approach or Standardized Approach. Under the BIA Operational Risk is 15% of the average of past three years' revenues (adjusted for certain banking book securities income and other items). If a bank has had zero adjusted revenue for those three years, it has no Operational Risk - at least for the CAR. Under the Standardized Approach a similar formulaic approach is used against revenues for eight defined lines of business with specific assigned LOB risk weight to each of the 8  ranging from 12% to 18%. 

As you might expect, banks generally determine which method will give the lowest Operational Risk charge and select that method as being the most "accurate". Again the mere selection of the measurement method affects risk – at least for the CAR, though not I'd add in the real world.

Market Risk

Market risk calculations are more complicated so I'm just going to wave my hands here with an unsupported assertion.

Sometimes VaR will give a lower market risk number than the Standardized Approach.   Sometimes not.  Depending on the choice of the measurement method, the same portfolio can have quite different risk weights.   No difference in economic reality.  But a difference in CAR.

Examples

Let's take two imaginary banks: Bank A and Bank B. For simplicity sake we'll ignore market and operational risk.

Bank A is a conservative bank that lends only for productive activities (no speculation) and structures its loans to require semi-annual payment of interest and principal. It also diversifies its risk among many customers, has a wide variety of tenors on its loans,  and makes a practice of taking real estate as collateral with borrowing base of 50% of collateral. Its KD100 loan portfolio is all to unrated clients so its RWA are KD100. After adjustments it has KD13 in capital giving it a CAR of 13%.

Bank B is less conservative. It also has a KD100 loan portfolio. Its underwriting standards are much lower than Bank A's. If a borrower can find his way to a Bank B Branch, he can get a loan. As part of its customer friendly approach, Bank B does not take real estate or other collateral. All the loans it makes are for five-year tenors with interest and principal due at maturity. It also is less diversified than Bank A and has numerous customers with loans equal to 14% of its regulatory capital (just under the 15% threshold). After adjustments it has KD 15 in capital.

Bank B's CAR is 15%. Bank A's CAR is 13%.

Based solely on CAR, one might think that Bank B was a stronger more creditworthy bank than A. It is not.

What credit factors are not considered in CAR?

  1. The quality of the Board and senior management. Are they seasoned and prudent bankers? Do they have a good business and strategic sense? Or do they chase the latest fad? As long as the music is playing, are they dancing? 
  2. Corporate culture and ethics.  Loose standards lead to problems.
  3. The quality of credit underwriting and structuring.  Most problems in the portfolio are caused by easy credit standards, weak or sloppy loan terms, etc. 
  4. The quality of loan monitoring and administration.  The earlier a bank finds a problem the earlier it can try to fix it.
  5. The quality and liquidity of assets. If the bank needs to raise cash, can it do so quickly? What sort of discount, if any, will be required on its assets? 
  6. The quality of earnings. Equivalent to cash or accruals of  imaginary promises? 
  7. The structure of liabilities. Diversified across lenders, instruments, and markets? Are maturities  appropriate for the bank's business? Or is the bank dependent on a very limited source of  funding? Most or all of which is short term?
These are critical factors that can impact the ability of a bank to weather a storm. They are an essential element of its financial health.  The CAR does not consider them at all.

For these reasons, one cannot rely on the CAR alone as the single measure of a bank's financial health and strength. A wider view is required.   One that evaluates a range of factors, such as the CAMEL approach used by US regulators.  To its credit the Basel Committee on Banking Supervision has recognized this - which is why it instituted the Pillar 3 disclosures - a first step to giving lenders and investors a look under the "hood" of a bank.


The above also explains why some banks (Bank B in our case above) should be required by the regulator to hold more than 12% capital.  Economic capital is the buffer than absorbs unexpected risks.  The riskier the bank the more capital it should hold.