Sunday, February 15, 2009

Analyzing a Bank - business quality, credit discipline and resilience to losses

If you intend to invest in banks, you'll need to get a feel for how good a bank will be as in investment. I'm going to describe one way you can get a feel for a bank's business strength and its credit quality, and by proxy, whether the bank's management is making good lending decisions. This is especially important in light of the recent reckless lending behavior of some banks, which has already caused a number of "big banks" (like Washington Mutual) to fail.

Assessing the strength of a Bank's business
and investment merits

The value of a bank as an investment is not based on its credit quality alone, though it is an important factor. For example, a bank could be making excellent lending decisions, but be continually losing customers because of lousy service or because of irrational competitors who underprice their loans. Under-pricing to gain market share is a dangerous trap that businesses with long-tail risks and short-term management can fall into. A holistic picture of a bank's value can only come from looking at:
  1. the factors and quality of managerial judgment, seen through:
    (a) the bank's propensity to loosen its standards and make bad credit decisions, and
    (b) the bank's willing to prepare for tough times - banking is an inherently cyclical business that tracks underlying economic cycles and cycles of excessive credit and credit shortages caused by competitive behavior - high earnings cause competitors to flood the market with cheap credit, until the credit becomes too cheap to handle normal default rates/or default rates go up, which then causes credit to dry up. The dynamic is similar to what happens in the insurance industry, and also similar to the boom-bust cycles in commodity prices.

  2. competitive position: the bank's ability to cling on to its customers / attract more customers, in the face of rabid and sometimes irrational competition.

  3. macro economics: whether the economic footprint underlying the bank's customers is a deep, broad and self-sustaining one (banks are creatures that depend on the underlying economy).

The latter two are qualitative assessments you need to make. The bank's public financial statements aren't likely to give you enough information to make a proper assessment. A bank's public financial statements can only help you judge a bank's credit discipline, and by proxy, the quality of its managers' judgments. Specifically, you'd have to look at the bank's balance sheet and its accompanying notes for clues on how a bank's loans have performed during the year.

Nonetheless, you should be aware this this is not a perfect indicator of a bank's current management discipline, because:

  1. bad loans tend to default only after a year or two. So a bank could have really loosened its credit quality over the last 12 months, and you probably wouldn't see any indication of this in the balance sheet.

  2. a low default rate doesn't mean anything if the economy is doing well and asset prices are rising. In such times, you'll probably see all banks experiencing low loan losses as the rising tide lifts all banks. This is particularly true in a credit bubble, when a borrower who is having difficulty paying off a loan can usually refinance it with another bank - this merry go round will make it look like the loan has never defaulted.

    Examining a bank's balance sheet during an economic downturn tends to be more useful, because we we can compare a bank's loan performance against its competitors. As loans start to default during a downturn, and borrowers have no way refinance their loans, you will be able to see which bank has been reckless and which has been prudent in its lending decisions. As they say, it's only "when the tide goes out that you can see who's been swimming naked".

Understanding the Loan Accounts,
in order to assess credit quality

The loan amounts that are shown in the balance sheet refer to (a) the outstanding principal and (b) the accrued interest that the bank's customers have yet to pay off. The accrued interest refers to the interest which the bank has "earned" or "charged" the customer up to the current date. For example, if this is the 5th year of a 10 year loan, then the amount of accrued interest reflected in the balance sheet would only include interest charged up to the 5th year and not yet paid. It would not include the interest yet to be charged for years 6 to 10, since the bank hasn't actually earned the interest. The interest is only earned if the bank continues holding the loan the next year; and there is a possibility the borrower may just repay the whole loan next year.

Loans generally start out as "Current loans". This means that the borrower is paying the installments regularly, and the bank has no reason to believe that this will stop. In the loan books, such loans are treated as current and accruing interest - the bank assumes that it is continuing to earn interest on the loan and hence add this interest, as it is accured, to the loan books. Such loans are carried on the loan books and are a part of the total loan amount shown on the balance sheet.

If a borrower fails to pay an installment within 30 days of when it is due, then the loan becomes classified as a "delinquent loan, 30-89 days past due, still accruing interest". If the borrower still fails to pay the late installment after 90 days, then the loan becomes classified as a "delinquent loan, 90 days or more past due, still accruing interest". The bank generally believes that it won't incur losses on these loans because either (a) the borrower will resume payment on the loan soon, or (b) the bank can seize the underlying collateral (e.g. the house is collateral for a mortgage) and sell it at a price higher than the outstanding balance of the loan. Such loans are still carried on the loan books, and are a part of the total loan amount shown on the balance sheet.

If at any time, the bank believes that the delinquent loan will not be recovered in full, then the loan becomes a "non performing loan, not accruing interest", also known as a "non performing asset". The bank believes that it will incur losses on the loan because (a) the borrow cannot continue paying the installments, for example because he/she is bankrupt, and (b) the market value of the underlying collateral is less than the outstanding balance of the loan. The bank also assumes that it isn't earning any interest on-top-of the principal, since the loan has effectively "stopped working". Even if interest is collected, it will be written off against the principal. Such loans are still carried on the loan books, and are a part of the total loan amount shown on the balance sheet, even though it is likely some part of the value will be lost in future.

During the year, if the bank believes that the borrower won't be able to repay a loan (Delinquent Loans or Non-performing), then the bank will begin proceedings to recover from the loan by seizing and selling the collateral for the loan (e.g. seizing a house and selling it). When it does this, the bank may incur a loss if the sale price of the collateral is less than the amount of the loan that is shown on the bank's loan books. For example, if the outstanding principal and accrued interest on a mortgage is $100,000, but the bank only managed to sell the house at $80,000, it would incur a loss of $20,000. Such losses are charged to the "loan loss reserves"/"credit loss reserves", so that they can be written off the books.

  • As an accounting matter, these net charge-offs do not impact the current year's income statement; they only reduce the bank's loan loss reserves. The income statement was affected in previous years when the bank "added/built-up" its loan loss reserves. The reserves are in a sense, off the balance sheet (strictly speaking they are shown in the assets side. You will see it in the "total loans - loan reserves = net loans" item. However, they are conceptually off the balance sheet). When the bank builds up its "off-balance sheet" loan reserves, on the balance sheet what happens is that the net loan amount drops on the asssets side, and the shareholder's equity drops on the equity/liabilities side.

  • The Credit Loss Reserves (Allowance for Loan losses) are reduced by "Net Charge-offs", and increased when the bank makes additional "Provisions for credit losses" (which hit the income statement).

Ways to Assess a bank's credit quality,
and management's credit discipline

There are 2 ways you can observe the banks credit quality, and by proxy, its management's credit discipline:

  1. Point in time snapshot: One way to see the bank's credit judgment is to see the non-performing loans as a percentage of the total loans made by the bank. This gives you a snapshot of the credit quality of the loan book at a point in time.

  2. Over the year: It is also important to review the net-charge-offs made over the year, as a percentage of the average loan balance during the year. This shows you the total amount of loan losses charged off during the year, as opposed to a snapshot of the state of the loans currently on the books.

While these are useful indicators, it is also important to look at these numbers with reference to (a) the loan mix, and (b) the overall interest rates charged for the loans. For example, a bank specializing in high-interest loans to low-credit-score borrowers would naturally see a higher rate of loan losses. But this is ok if the bank is able to charge interest rates high enough so that over the large base of customers, the loan losses are more than made up for by the higher net interest income. Just like in insurance, it is ok to take on higher risk business as long as you can price for it. Generally speaking, the most risky to the least risky (in terms of net charge-offs expected) loans are:

  1. Credit card and Unsecured loans - these are unsecured loans, and experience higher rates of default. However among credit cards, it appears that affinity branded cards (e.g. alumni, associations etc) tend to have slightly lower loss rates.

  2. 2nd Lien Secured Loans (Home equity and 2nd lien mortgages). In these loans, the bank only has 2nd claim to the underlying collateral, because the 1st right of claim goes to someone else (the holder of the 1st lien).

  3. 1st-Lien Secured Loans (Residential Mortgages, Commercial Real Estate Loans, Lease financing). For such loans, the bank has the right to seize the collateral and sell it, if the borrower fails to make payments on the loan.

What we're look at is past credit experience,
it's no guarantee of future management behavior

The bank's credit experience gives you an idea of what the bank's management has been doing in the past. However, it says nothing about the bank's credit decisions over the past 12 months, because bad lending decisions often don't show up immediately:

  1. The bank may have written a set of bad loans with a low teaser interest rate which only resets after 1-2 years. Credit losses will spike only when the resets hit.

  2. The bank may have written a recent set of loans with higher Loan-To-Value ratios than in the past. So even if the bank continues lending to the same type of credit worthy borrowers, the amount of charge-offs will increase even if the proporation of people who default remains the same. Each individual default will on average, result in more losses to the bank because the loan outstanding is very close to the initial valuation of the collateral. A smaller drop in collateral price will result in losses to the bank.

  3. The bank may have failed to increase its LTV guidelines in the face of a bubble in collateral prices. Simply maintaining a 80% LTV ratio may be insufficient if you are in a housing bubble and you expect housing prices to drop by 30%.

  4. And in general, loans usually don't default until 12-24 months have passed (unless a really really bad lending decision was made and the borrower can't even service his/her first installment)

In other words, the numbers in the accounts give you a clue as to the bank's past practices, but you need to make a qualitative judgment as to whether the bank's management has changed its behavior in the recent past.

Another indicator of management judgment: a Bank's Resilience to Losses
Is the bank conservatively managed?

Generally, you'd want a bank to be managed conservatively, so that it can weather economic downturns and financial crises. The bank's balance sheet should be robust enough to withstand the booms and busts of free market capitalism. So how much losses can a bank take before it goes bankrupt? And how should the balance sheet be managed to weather such storms?

Banks must always have more assets than liabilities, otherwise they wouldn't be able to pay depositors and bondholders/creditors back their money. So conceptually, the maximum amount of money a bank can lose and still remain solvent is equal to the amount in shareholder's equity. Anything more and the bank would be insolvent (bankrupt).

In practice, regulators impose minimum capital adequacy requirements on banks. For example, a country's regulator may require banks to have a minimum equity to asset ratio (leverage ratio) of 4%. Many countries impose capital requirements on banks based on Basel 2 guidelines. For example, in the U.S., banks are required to have a Tier 1 capital ratio of at least 4%, and a Tier1 capital + Tier 2 capital ratio of at least 8%.

The very most a bank can tolerate in losses in one year is a combination of (1) that year's net income and (2) an amount of equity that would not cause the bank to breach its capital adequacy requirements. Banks can incur losses in many ways, such as:

  1. Through credit losses when borrowers default, and the underlying collateral cannot be sold at a price to cover the outstanding loan amount.

  2. Through losses in the value of securities it holds as assets. Banks typically invest some of their cash into treasury bills, bonds and other securities. If the securities default or experience a permanent loss in value and the bank is unable to hold them to maturity, then the bank will incur losses. (Note that some securities that are on a bank's books may not be marked to market. Accounting rules allow some securities, known as Level 2 and 3 assets, to be booked at a value determined by the bank's valuation models. Assets carried at mark-to-market values are classified as Level 1 assets)

  3. When business costs rise too fast, for example because of unexpected litigation, a failure to control expenses, pension costs, and so on.

  4. By failing to manage their asset and liability interest rate and forex sensitivity. For example, during the Asian financial crisis when several Asian currencies experienced sudden devaluation, some banks ran into trouble because they had liabilities in foreign currencies.

One way of seeing when a bank is in danger of becoming insolvent from bad lending decisions is to make an assessment of what a probable credit loss rate would be, then see if the bank can withstand that.

Credit Risk Modeling: (Stress testing)
how bad can credit losses get?

It really depends. For example:

  • In the 1997 Asian Financial Crisis, some banks in Singapore had up to 7% of their total loans become non-performing (in some countries, up to 25% of total loans were non-performing). Even up to 2001, Indonesia's banks NPLs were estimated at 48% of total loans.

  • In 1976 (the aftermath of the 1974-75 recession) in the US, up to 5.3% of loans (according to the New York Times) were problem loans.

  • In 2005, 8.6% of China's loans were reported to be NPLs (according to Xinhua)

So a large part of estimating NPLs is based on an analysis of the strength and resilience of the underlying economy. This is a macro-economic assessment that you need to make. The long term health of a bank is greatly dependent on the strength of the economy that underlies it.

If you have a view of the underlying economy, then you can make an educated estimate for the credit losses a bank might sustain. For example, if we assume:

  • The bank has an average outstanding loan to value ratio of 80% (e.g. mortgaged houses are valued at $100 in the market, and the borrower's outstanding loan is $80). In other words, the bank won't suffer any losses even if the borrower defaults, unless the selling price of the property (or other collateral such as machinery) drops more than 20% i.e. to below $80.

  • Collateral (or housing) prices will drop by 35%. We can make this estimate based on a back-of-envelope analysis of the Case Shiller index (Composite-10 CSXR overall price index). If we assume it will drop back to year 2000 levels, then it will drop 35% from Oct 2008 levels. (October 2008 index value=169.78, Dec 2000 value = 113.56). This means that houses drop from $100 to $65.

Scenario 1: Now, if we assume that all customers will default the moment their house re-sale prices drop below their outstanding loan value (entirely possible if the house was purchased as an "investment", and not for living in), then all customers would default and the bank would have to seize and sell the underlying (mortgaged) property. The bank is owed $80 for each loan, but is only able to sell the house for $65. This implies the bank will suffer a loss of $15 on each loan. In other words, credit losses will be ($15/$80) = 18.75% of the total loan value.

This scenario assumes that all loans that are in negative equity will default. However, this usually isn't the case if the bank has only lent to people who borrowed to buy the house to live in. People who buy their houses to live in tend to continue paying off their loans as long as they are employed and have the means to. This is in part because of the need for a place to stay, a desire to keep a good credit rating, and because of a psychological escalation of commitment to the house. If they default on their loan and give up the house, they will lose the $20 they paid for the house. (From their point of view, even if they buy a new house for $65, they will have paid in total $20+$65=$85 for this new house, excluding interest costs. i.e. they will only save $15)

Scenario 2: If the bank has been lending to people who will only default when they are unable to continue paying because of unemployment, then the bank's losses will be lower. (A similar principle applies for commercial and other loans, even though our example is based on residential mortgages). Let's assume a worst case scenario of 20% unemployment, which we roughly take to mean that that 20% of borrowers will default. Of these customers who default, the bank will, after seizing and selling the houses, lose 18.75% of the loan value because of the lower market prices of the houses. This translates into an overall loan loss of 3.75% of the total loan book.


This in a nutshell, is how you can approach the task of assessing a bank as a investment. Much of any analysis will involve qualitative factors, as I have described here. Be careful not to ascribe too much importance to the quantitative results of your analysis, without the looking at the qualitative factors that form the backdrop of your analysis.

The current dour sentiment for banks in general ignore the fact that there are a number of good quality banks in the industry. Of course this is a relative statement, no bank will be viable if the entire economic or financial system collapses. But barring such a systemic event, you will be able to find some banks which will weather this storm reasonably well.