My view on Subprime crisis

With the second write-down of US$ 19 billion Subprime loss by UBS AG this week, the long-lasting subprime crisis from Feb 07 refuses to be out of the spot light. Up till Mar 08, the total write-down by the financial institutions across the globe was estimated to be around US$ 285 billion, with the global banks of UBS (U$37.4 billion write-down), Merrill Lynch (US$ 24.4 billion write-down) and Citigroup (US$ 18.1 billion write-down) top on the casualty list.

Investors and bankers start to be immune to the write-down numbers. It is "good" time to announce bad figures since the market gets used to billions of write-downs or losses. Bankers in Wall Street cry that it is the worst case scenario in banking and black days in history. But on retrospective (a bit on hind sight), who else caused this problems, other than the bankers themselves?

It would be good to understand the issues, before the figures can be pointed.

What is Subprime loans?
  

Traditional lending
  

The banks lend the money to the mortgagors who agree to let banks to seize their houses if they are not able to pay off the borrowed amount (thus the collateral for the bank loan is the market value of the house). In order to ensure a reasonable high repayment ratio, credit risk management departments of the banks assess the repayment ability (such as Probability of Default and Loss Given Default) for each loan application and take only those with high repayment potential.

In the time of Alan Greenspan as Feb Reserve Chairman in the States, banks in the states were "encouraged" to lend to the borrowers who may not meet the credit risk management criteria according to usual criteria, probably because of their instable income or unsatisfactory financial situation. In order to cover the additional credit risk caused by these loans, the banks raised the loan lending rates over than the common bank prime lending rate. Thus, here comes "Subprime loan".

The Greenspan's policy came out with good intention to bring home to Americans. Loose credit enabled more people to buy houses and therefore drove up the demand  (and thus price) for housing market in the States. The lofty house price that didn't reflect economic fundamentals caused higher amount of Subprime loan borrowed from the banks and also "bubbles" in housing market. It is a vicious circle.

When the fundamental ceased to support the market expectation, the house price dropped. The drop of valuation of Subprime loans' collateral (i.e. houses' market value) triggered "margin call" to the borrowers to either top up with cash payment or repay the loan earlier than scheduled. If the borrowers can't fulfill the margin calls, they are "in default". The banks may seize the house (whose value might be lower than the outstanding loan amount) and suffer the loss of difference between outstanding loan amount and the current market value of the house.

Financial market
  

In the past few decades, the primary and secondary market has never been more active. The people in "common street" view secondary market as stock exchanges where stocks, securities and other simple derivatives change hands, while bankers in Wall Street trade exotic products with their elite counterpart bankers in secondary market. Collateral Debt Obligation ("CDO") is one example of "not recent but very popular" exotic products.

CDOs are in essence securities issued based on a pool of assets (called underlying assets). The value of underlying assets is split into tens of thousands of securities (stocks or bonds etc) with each security given a split proportion of the value. CDOs are available for investors across the globe, thanks to great efforts by investment bankers. The underlying assets can be any type of assets, ranging from a pool of credit card loans to a few blocks of office buildings.

Before the innovation of the CDOs, the banks have to hold the loan portfolio (such as Subprime Loan etc) as assets in their balance sheet and valuate these assets on a regular basis, as required by the accounting standards and tight banking regulations. If the value is estimated to be lower than the value recorded in balance sheet, the bank has to make "provisions" (an accounting term to reduce the assets in balance sheet as well as increase expense in profit and loss statement) and justify the amount of provisions in front of numerous internal auditors, external auditors and regulators.

Obviously, it is very convenient to package the loan portfolios in balance sheet into CDOs and sells them to investors outside the bank who know much less about the quality of underlying assets (e.g. Subprime loans). No more valuation (since no more loan assets on balance sheet), no more auditors but more resources for other business. In order to ensure the loan portfolio is "cleanly" sold off and taken off from banks' balance sheet, very often a special purpose vehicle ("SPV"), independent from the originating bank (that sold the loan portfolio), is established as a registered company to take over and manage the sold loan portfolio.

But why investors are so "stupid" to buy these CDOs? I think it is because of information asymmetry. At the point of CDOs sale, the originating bank knows the borrowers' repayment ability best because they used to assess their customers (i.e. borrowers) regularly, while the investors are not able to know the information well enough to challenge the CDO selling price, although investors heavily rely on the credit rating given by rating agencies (such as Standard & Poors, Moody's etc) before they make investment decision. In addition, the repayment ability known to the originating bank is based on its best estimation for the future until the maturity of loans (usually more than 20 years for housing loans). You know, estimation is subject to change! Therefore, each side of the deal believe that they have obtained the best price and here comes the deal.

What makes it more "interesting" is that CDOs, the derivatives itself, also change hands, and change very frequently. With each change of hands among bankers  across the globe, the information asymmetry comes into work. Each party is happy with the deal and the rating agency is happy with the fees. The seller is happy to sell off the risk with good price and the buyer is happy to buy the securities with potential good yields. You can image the loss of information along the chains of CDOs about repayment ability of the ultimate underlying Subprime loan portfolio.

The whole process is called "securitization", which reflects the mechanisms to transform illiquid assets (such as long term housing loans in balance sheet) into liquid securities whose sale proceeds can be re-invested by the originating banks for other attractive investment chances.

Why it becomes a problem for banks now?
  

The problem starts from the lending market, with the burst of the housing bubble in the States. With inability to repay the loan or meet the margin calls from the banks, the borrowers may choose to surrender their houses to the banks. It is the least thing the banks want, because the banks have to rely on the market value of the house (which is not reliable at all nowadays) for loan repayments. If the borrowers agree to restructure the loans (such as pay less monthly installment by extending loan tenor), the banks not only have the last resort of the market value but also have better repayment prospective to tap on borrowers' regular income for loan repayment (although it will take much longer to resolve the issue).

The issue is already a big headache for banks with direct exposure to mortgage loans in the States, such as Citigroup, Merrill Lynch etc, because the banks have to invest a lot of resources for loan recovery (often a lot without success). Unsurprisingly, the issue spreads much faster like a fire across the CDO secondary market, where the bankers hold much more rosy picture based on unreliable predication of yields.

The literally financial Frankenstein of subprime loan securitization became too daunting for anyone to understand. With very little information about default statistics (and/or restructure statistics) from mortgage loan market, the banks who unfortunately still held the CDOs started to become panic! These unlucky banks had absolutely little ideas about the value of the CDOs in their balance sheet. The common mark-to-market risk management practice ceased to work, because there was no market at all for these CDOs in this scenario (literally worst case scenario, as called by risk management professionals). The banks were not able to accurately valuate the CDOs with their state-of-art valuation systems, simply because there were not many relevant information feeds into these systems. Worst of all, when the banks started to scramble for relevant information, the "necessary" long chains of securitization, that worked very well for good deals, worked very disadvantage against this information collection exercises.

You may think financial transactions are zero-sum games. Since some banks lose, other banks gain. Well, this argument has a point in CDOs secondary market. If you follow the financial press, you will know Goldman Sachs' significant result of US$ 11.6 billion gains in financial year 2007 and it is nowhere in the subprime casualty list. Exact reasons for the "luck" of Goldman are not known to me, but Goldman short sold CDOs before the Subprime issue turned sour. So Goldman was among few winners of the zero-sum game.

Nevertheless, you will know the Subprime issue is not a zero sum game when you compare the loss of US$ 285 billion loss worldwide, because the borrowers in mortgage loan market started to default and the banks with direct exposure started to write off the Subprime loans in their balance sheet. Even in secondary market, the originating banks have to take the sold loan portfolio back from independent SPVs to their balance sheet for various reasons. One reason may be their legally or morally binding commitment to honor the redemption requirement from the investors. If the originating banks have the "recourse" written in legal documents for the sold loan portfolio (i.e. they have to guarantee the value of the CDOs), they shouldn't remove these loan portfolio out of their balance in the first place despite the sale, because the risk is still with the bank and therefore the assets still on balance, according to account standards. More often, the originating banks have "moral" commitment to honor the sold CDOs, because the reputation loss (and loss of their clients' good will) may be larger than the loss to honor the sold CDOs. This is what happened to HSBC when it took a few CDO funds back to its balance sheet.

You may notice that all loss we talked here are all in accounting terms and it may or may not have immediate impact on the actual cash flow of the banks. Why it is so critical for banks to issue securities or to obtain cash injection immediately after the accounting loss is posted? It started from the liquidity management of the banks and the banking regulators' requirement.

When banks take deposits from their customers, they put deposits as liability in their balance sheet (i.e. they need to honour the redemption of deposit upon maturity together with the agreed interests). In order to fulfill their liability as well as to earn revenue, the banks lend out the depositors' money at a higher interest rate than deposit rate in the form of all types of loans. Obviously, in order to achieve higher returns in loans, the loan tenors should be necessarily much longer than those of short-term deposits, e.g. housing loans for 20 or 30 years. Therefore, there is inherent imbalance of the banks' liquidity (i.e. short-term liability v.s. long term assets). In order to ensure enough cash (or capital) to fulfill their short-term liability, the banks put certain amount of capital in the banks at all times that can be converted into cash quickly to meet the withdrawal demands of the depositors. These capital are called "economic capital".

The banking regulators has been concerned with the amount of economic capital by the banks and issued regulations for banks to follow. The recent international banking requirement (Basel II), effective from 1 Jan 2008 for major global banks, requires that the more risky assets the banks have, the more capital they need to put in place. When the risky Subprime loan portfolios sit in banks' balance sheets in mortgage loan and derivative markets, the required economic capital is increased significantly. The quickest way to raise such huge amount of liquid capital is to issue securities in exchanges and/or receive injection of capital for stocks.

What will happen to banks if they are not able to raise enough capital? First of all, the banking regulations are breached. What is more problematic is that the banks can be in "technically" bankrupt if they are not able to fulfill the liability to meet the withdrawal demands from the depositors, although the banks may have more illiquid assets than the deposits but are not able to convert them into cash immediately. With the information about insufficient cash spreads in a lightening speed among the market, all depositors want to withdraw their cash before the bank's cash dries out. This is called "bank run". In Great Depression in 30s, depositors queued in front of the banks to withdraw their life savings. Nowadays, bank run is displayed in a more subtle but equally fatal way, such as private banking clients to withdraw millions of dollars by a press of button of their Internet Banking.

Other than the worst case scenario of "bank run", the banks without this immediate problem may face another problem, higher borrowing costs in interbank market. Other than deposits taken from the public to fuel their lending, banks also borrow from each other on short-term basis (e.g. overnight) to strike a balance between liquidity and return. London Interbank Borrowing and Offering Rate ("LIBOR") is one of the major rate indicators. Among the Subprime loan mess, the banks started to suspect the accuracy of the Subprime loan losses of their counterpart banks. After the near-melting-down of Northern Rock in UK after the bank run, banks have even more reasons to target their counterpart as next possible Subprime casualty. Since business still needs to go on, the lending banks demand higher lending rates (or credit premium as the technical jargon) in the interbank market to cover the possible loss of the lending amount Higher cost caused reduced borrowing and it is called "credit squeeze". Obviously, it is not good news at all for banks, no matter whether they suffers from Subprime loss or not.

Why it is a problem for economy?
  

With the intermediary role of banks (or financial institutions at large) in the economy, the possible bank failures can be fatal to economy. I will leave this topics in good hands of experts in macro economics but hope to elaborate my view on "moral hazard" of rescue of banks by governments.

It is a long-standing argument whether the government should intervene and rescue the banks when the banks have problems. People that favor this idea argue that the issues loom much bigger to whole economy for banks to fail. The counter argument is that banks, knowing they will be rescued when facing problem, will act more carelessly to manage their risk (or moral hazard in jargon).

I think there is no one answer that answers all questions related to moral hazard. It is really up to the top and well-informed government official (such as Chairman of Fed Reserve or Secretary of Treasury etc) to weight costs of bank failure and the cost of moral hazard. I have to say that I'm quite favorable of the idea to rescue the banks in Subprime loan crisis to prevent melt-down of the banking systems (and economical order at large). But I don't think it should form standard expectation from the banks in the future. Most important of all, the costs of moral hazard should be reduced by further regulating guidelines to the banks to manage their risk, which goes to the next section.

What are the lessons learned from risk management perspectives?
  

Again, I will leave the solutions to macro economics to the experts and will only elaborate my view from banks' risk management on Subprime loan and securitization.

Number 1: transparency of securitization
  

Some people argue that it is necessary for banking regulators to define the maximum numbers of securitization for one underlying asset, to reduce the length of securitization chain and protect the investors. I think the regulation doesn't need to be so prescriptive. To draw an analogy, it is natural to understand what you are buying before you make "buy" decision. Similar logic applies to financial transactions as well. If regulators require banks to document relevant information each round the same underlying asset is securitized, you can image that the documentation will be necessarily long at the first round of securitization with subsequent ones much more complicated. The complication of the documentation can naturally costs significant resources to the originating banks and also reveal critical information about underlying assets to potential buyers. With the critical information, rational investors are in much better position to challenge the risk and pricing but in turn they are subject to the same documentation requirement when these investors want to securitize again. Naturally, the margin of securitization will be reduced by rising documentation costs and better informed investors.

I am not saying that current securitization documentation is simple and naive but I think the key relevant information about underlying assets is not well documented when the transaction is first entered and is poorly documented & updated during the tenor of the underlying assets, especially for subsequent rounds of securitization.

Number 2: Independence of rating agents
  

When the Enron failed, the independence of external auditors were challenged and are heavily regulated now. With this Subprime crisis, it is good time to challenge the independence of rating agents, such as S&, Moody's etc. The rating agents play a critical role in securitization to reveal the credit risk levels for each "tranche" of the CDOs. Tens of thousands of CDO securities backed by underlying loan portfolio are divided into different tranches (or categories), some of which will be given higher priority when it comes to repayment or redemption of securities. These tranches are given higher credit rating by the rating agents to reflect lower credit risk. So investors can based on their own risk appetite to decide which tranches of CDOs to purchase.

Who will hire rating agencies to rate the CDO tranches? Now, the originating banks. Obviously, it has some elements of conflict of interest. Though rating agencies should be independent and responsible for the investors, it is the originating banks who pay them for the services. Now it depends on the market mechanism to control the service quality of these agencies. Anyway, those with inaccurate rating won't be trusted by investors and therefore won't be hired by the originating banks (since it defeats the purpose to convince the investors). But with very limited choice of three major rating agencies (Standard & Poors, Moody's and Fitch Rating), the market of rating agencies turn into an "Oligopoly", where a few major market player can set the price and quality of the services.

Undeniably, it is difficult to implement the direct fee payment from investors to the rating agencies, in consideration of the diversified investors across the globe. One way can be explored that fee payments can be collected from investors by the originating bank as the trust and will be paid to rating agencies to ensure the line of responsibilities.

Other than the payment of services, it is worth asking who can have the technical expertise and reasonable authority to challenge (or even audit/inspect) the service quality given by the rating agencies.  

Number 3: banks' internal control mechanism
  

When banks lend out, it is the credit risk management department who challenge the lending decision for new clients and review the repayment ability for existing clients. To ensure the independence,  credit risk management shouldn't report to any executive who has marketing responsibilities. Ideally, all risk managers report to Chief Risk Officer who is part of senior management committee and directly reports to board of directors.

Derivatives (such as CDOs) historically enjoys the "privilege" to be monitored by sophisticated systems and mechanisms. One of the popular mechanisms is to back to back all transactions from branches across the globe to the Head Office, which enjoys economy of scale and scope to monitor the risk centrally. That is to say, every transaction entered by the branch (say in Singapore) with its clients will be squared off in risk terms by an identical transaction placed between the branch and the head office, so that the risk with bank's clients is passed from the branch to Head Office (i.e. zero risk from the perspective of the branch). The associated large amount of information and accounting entries is made possible by connecting the systems from Head Office to all branches worldwide with decreasing costs of computer processing cost. For exotic transactions (including CDOs), one transaction may be split into two or more back-to-back transactions with Head Office according to their "risk elements" (such as credit, rates etc by allocation key).

In back-to-back arrangement, Head Office can be benefited from possible diversification effects (i.e. deals in some branch can be mitigated partially by those in others) and economy of scale for risk selling (i.e. reduced processing costs by pooling all risk selling activities together) etc. The risk management expertise in Head Office can be better utilized and risk management approach is consistent worldwide.

Despite the above benefits, I think there are two major shortcomings for this arrangement. First of all, no more holistic and individual picture for exotic products. I think the risk elements in one exotic product are naturally interlinked. If the link is broken by subjective allocations, the correlation among the risk elements are broken and no longer visible for Head Office risk managers. Neither is it for risk manager in local branches, since it is no more their responsibility to monitor the risk in totality, other than regular feeds of data into systems. On the surface, the risk is well and efficiently managed in Head Office but in fact the bank as a whole is still exposed to risk, especially to systematic risk. It leads to second shortcomings, management of systematic risk. In financial theories, systematic risk is not able to be diversified away and thus has to be taken by the bank or sold off to insurance companies. With the back-to-back arrangements, systematic risk may be under-estimated due to insufficient knowledge of correlation among risk elements for each exotic product. I think the arrangement may amplify the exposure to systematic risk (such as the subprime loan we are experiencing now).

There is no easy solution about this issue. If there is any immediate solution, it would be a decision balancing risk and return. If the risk exposure is deemed too high, more decentralized risk management approach should be adopted at least for exotic products. Dedicated systems shall be developed to cater for the new issues raised by the decentralized risk management. One thing is clear, complicated and exotic financial products should never be part of "mass production" of risk management for plain vanilla products.

P.S. solely personal opinions with figures taken from Reuters.com.