Archive for August 2009

The Black Swan Test

August 31, 2009

Many commentators have suggested that firms need to do stress tests to examine their vulnerability to adverse situations that are not within the data set used to parameterize their risk models. In the article linked below, I suggest the adoption of a terminology to describe stress tests and also a methodology that can be adopted by any risk model user to test and
communicate a test of the stability of model results. This method can be called a Black Swan test. The terminology would be to set one Black Swan equal to the most adverse data point. A one Black Swan stress test would be a test of a repeat of the worst event in the data set. A two Black Swan stress test would
be a test of experience twice as adverse as the worst data point.

So for credit losses for a certain class of bonds, if the historical period worst loss was 2 percent, then a 1BLS stress test would be a 2 percent loss, a 4 percent loss a 2BLS stress test, etc.


Further, the company could state their resiliency in terms of Black Swans. For example:

Tests show that the company can withstand a 3.5BLS stress test for credit and a 4.2BLS for equity risk and a simultaneous 1.7BLS credit and equity stress.

Similar terminology could be used to describe a test of model stability. A 1BLS model stability test would be performed by adding a single additional point to the data used to parameterize the model. So a 1BLS model stability test would involve adding a single data point equal to the worst point in the data set. A 2BLS test would be adding a data point that is twice as bad as the worst point.

Random Numbers

August 30, 2009

Just a quick thought on random numbers.

Perhaps we have the wrong model for a random number generator with the regular statistical probability distributions.
I am wondering if it wouldn’t be better to think of random numbers as coming from a toss of several dice where the dice are drawn from a barrel where some number of the dice in the barrel are non-standard dice. We do not know how many dots are on those dice or how many of the non-standard dice are in the barrel.

We might go for dozens of tosses without hitting one of the non-standard dice, and then one day we get two of them.

Somehow we need to figure out how to play the game well when we get the regular dice but be ready when the non-standard dice are drawn without warning.

Who wins with leverage?

August 29, 2009

Leverage increases apparent returns in best of times but Increases risk considerably in worst of times. Investors do not benefit from leverage over time. Managers benefit greatly from leverage. Derivatives are highly levered. Traders think that it is silly to spend any time thinking about notional amounts of derivatives. Insurers should learn that they need to pay attention to the notional amount of their insurance contracts. Owners of now highly diluted shares of banks (and AIG) now know that the leverage of those organizations did not in the end create value. Insurers, like banks, are by their fundamental nature highly leveraged with capital a tiny fraction of gross obligations. Insurers should take extreme caution when considering activity that increases leverage. And they should make an analysis of the true amount of leverage in their activities an important activity before entering a new activity and periodically as the world turns.

For example, if one investor puts his money in a 2/20 hedge fund that is 10 for 1 levered and that pays 10% interest on its funds. If the returns for the first four years are 20% per year, After 4 years, the investor is up over 400%! The hedge fund manager has been paid over 150% of the original investment and the debtholder has been paid 400%. But then in year 5, the investment loses 20%, giving back just one of those four years of outsized gains. All of a sudden, the investor is down to an 8% cumulative gain!!! while the manager and lender have slightly higher gains than after the four fat years.

The sister of this investor had the same amount of money to invest, but put it into an unlevered fund with the same types of investments and without the 20% profit share for the manager. After four fat years of 10% gains, the sister is up over 35% and the manager has been paid only 7% of the original fund value. Now the market drop hits sis’ fund with a 10% loss and she ends the five years up a respectable 19%. The fund manager gets about 9% of the original fund for his five years of work.

Leverage Illustration

Leverage Illustration

Of course, the illustration can be manipulated to make anyone the supreme winner. But this scenario seems pretty telling. Leverage primarily benefits the fund manager, not the investor in this scenario. In many scenarios they both benefit, but there are no scenarios where the manager does poorly on a leveraged investment fund.

So when you die, pray to come back as a leveraged hedge fund manager.

Multi dimensional Risk Management

August 28, 2009

Many ERM programs are one dimensional. They look at VaR or they look at Economic Capital. The Multi-dimensional Risk manager consider volatility, ruin, and everything in between. They consider not only types of risk that are readily quantifiable, but also those that may be extremely difficult to measure. The following is a partial listing of the risks that a multidimensional risk manager might examine:
o Type A Risk – Short-term volatility of cash flows in one year
o Type B Risk – Short-term tail risk of cash flows in one year
o Type C Risk – Uncertainty risk (also known as parameter risk)
o Type D Risk – Inexperience risk relative to full multiple market cycles
o Type E Risk – Correlation to a top 10
o Type F Risk – Market value volatility in one year
o Type G Risk – Execution risk regarding difficulty of controlling operational losses
o Type H Risk – Long-term volatility of cash flows over five or more years
o Type J Risk – Long-term tail risk of cash flows over 5 five years or more
o Type K Risk – Pricing risk (cycle risk)
o Type L Risk – Market liquidity risk
o Type M Risk – Instability risk regarding the degree that the risk parameters are stable

Many of these types of risk can be measured using a comprehensive risk model, but several are not necessarily directly measurable. But the muilti dimensional risk manager realizes that you can get hurt by a risk even if you cannot measure it.

Know your embedded assumptions

August 27, 2009

An implicit assumption in the way that many practitioners use financial models is that their planned activity is marginal to the market. If you ask the manager of a large mutual find about that assumption and they will generally laugh out loud. They are well aware that their trades must be made carefully to avoid moving the market price. Often they will build up a position over a period of time based upon the normal flow of trading in a security. That is a very micro example of non-marginality. What happened with the sub-prime mortgage market was a drastic shift in activity that was clearly not marginal. When the volume of sub prime mortgages rose 10 fold there were two major changes that occurred. First, the sub prime mortgages were no longer going to a marginally more creditworthy subset of the folks who would technically into the sub prime class, they were going to anyone in that class. Any prior experience factors that were observed of the highly select sub prime folks would not apply to the average sub prime folks. So what was true on the margin is not true in general. The second marginal issue is the change in the real estate market that was driven by the non-marginal amount of new sub prime buyers who came into the market. On the way up, this expansion in the number of folks who could buy houses helped to drive the late stages of the price run up because of that increased demand. That increase in price fed into the confidence of the market participants who were feeding money into the market. Risk managers should always be aware that marginal analysis can produce incorrect results. They should follow my mother’s caution “what if everybody did that?”

ERM only has value to those who know that the future is uncertain.

August 26, 2009

Businesses have three key needs.

First they need to have a product or service that people will buy. They need revenues.

Second they need to have the ability to provide that product or service at a cost less than what their customers will pay. They need profits.

Once they have revenues and profits, their business is a valuable asset. So third, they need to have a system to avoid losing that asset because of unforeseen adverse experience. They need risk management.

So Risk Management is the third most important need of a firm.

And there is often a conflict between risk management and the other two goals. Risk management will sometimes say that a business activity that produces revenue is too risky and must be curtailed or modified in such a way that it produces less revenues. Risk Management often costs money or otherwise depresses profits. For example, an insurance policy covering fire of a building owned by the firm will cost money and depress profits.

So Risk Management needs to defend its value to the firm. Many risk management proponents have been asked to tell the value added of their activities. This is difficult to explain. Not because risk management does not have a value, but because the cost of risk management in terms of reduced revenues or increased costs are usually tangible and definite, while the benefits are probabilistic. Often the person asking the question is looking for a traditional spreadsheet answer that shows two columns adding up and perhaps the difference between the two is the benefit of risk management.

It does not work that way. For Risk Management to have value, one must understand that the future is uncertain. The value of risk management comes from the way that it shapes that uncertainty.

The next time you are asked about the value of risk management, ask the questioner what value they would put on the airbags and seat belts in their car. If they have no uncertainty about their ability to avoid accidents, then they will put a zero value on the safety devices – the personal risk management systems. If they resist answering, ask them if they will agree to have them removed for $20? Or for $2000? What value do they place on that risk management?

Most people will agree that the demise of a company is less serious than the demise of a person, but it is not difficult to see that there is some value to activities that increase the chance that a company will not expire in the next business cycle or windstorm.

So risk management decreases the uncertainty about the survival of the firm. There is a way to quantitatively value that reduction in uncertainty and compare it to the reduced revenues or increased costs of the risk management activities.

VaR is not a Bad Risk Measure

August 24, 2009

VaR has taken a lot of heat in the current financial crisis. Some go so far as to blame the financial crisis on VaR.

But VaR is a good risk measure. The problem is with the word RISK. You see, VaR has a precise definition, RISK does not. There is no way that you could possible measure an ill defined idea as RISK with a precise measure.

VaR is a good measure of one aspect of RISK. Is measures volatility of value under the assumption that the future will be like the recent past. If everyone understands that is what VaR does, then there is no problem.

Unfortunately, some people thought that VaR measured RISK period. What I mean is that they were led to believe that VaR was the same as RISK. In that context VaR (and any other single metric) is a failure. VaR is not the same as RISK.

That is because RISK has many aspects. Here is one partial list of the aspects of risk:

Type A Risk – Short Term Volatility of cash flows in 1 year
Type B Risk – Short Term Tail Risk of cash flows in 1 year
Type C Risk – Uncertainty Risk (also known as parameter risk)
Type D Risk – Inexperience Risk relative to full multiple market cycles
Type E Risk – Correlation to a top 10
Type F Risk – Market value volatility in 1 year
Type G Risk – Execution Risk regarding difficulty of controlling operational losses
Type H Risk – Long Term Volatility of cash flows over 5 or more years
Type J Risk – Long Term Tail Risk of cash flows over 5 years or more
Type K Risk – Pricing Risk (cycle risk)
Type L Risk – Market Liquidity Risk
Type M Risk – Instability Risk regarding the degree that the risk parameters are stable
(excerpted from Risk & Light)

VaR measures Type F risk only.

The Need for Speed

August 23, 2009

One of the causes for the financial crisis is the very high speed trading operations. There are several reasons that is true.

First, the high speed transaction oriented operations needs to eliminate any reflection or analysis on the part of the trading firms. There is simply no time for this. If the transactions need any analysis to support them, then that analysis must be outsourced. This is the driver behind the idea that credit risk, which for most of financial history has been seen to be a risk that required careful analysis and reflection, became a traded commodity. Banks started to merge market and credit risk and do away with their credit research staffs. This is one of the most common issues cited in explaining the crisis. Banks failure to do their own credit analysis on the sub prime loans. The lack of scrutiny led directly to the liar loans – since no one was looking at the loan applications there was no need to take any effort to make them correct.

Second, models of these risks need to be closed form models that can run instantly. More robust and complex monte carlo models that might capture the nuances of the risks were just not practical given the time frame. Monte Carlo models take too long to develop and too long to run. With a few simplifying assumptions, the need for Monte Carlo models can be eliminated and a closed form model that runs in seconds developed. The simplifying assumptions also allow the daily updating of these models. This process makes sense if and only if you believe that market prices generally reflect all information, so a closed form model that mostly just replicates and extends market prices is all that you need anyway. This approach to modeling risk makes it almost completely impossible to detect changes in the underlying risks. All users of this approach will always go over the cliff together – and they did.

Third, the speed of transactions means that there is turnover of the risks held during the year. This turnover may be 5 or 10 or 20 or 50 times. This business can easily be seen to be very, very low profit margin since capital is generally only held on the amount of risk at any one time. However, when the crisis hit, the banks were unable to continue to roll their inventory of risks and they piled up in amounts 5 or 10 times their past average holding.

Maybe MTM isn’t exactly what is needed?

August 22, 2009

Everyone (except corporate boards and managers) seem to agree that short term incentive compensation is one of the key drivers for the excessive risk taking that led to the financial crisis. In an earlier post, it is suggested that one of the reasons is that accounting is less reliable in the short term.

Perhaps the problem is Mark-to-Market accounting. While it is an extremely important discipline to know the market value of positions, MTM has a misleading presumption. In effect, MTM treats a position that has been closed by sale on the day that the financials are set exactly the same as an open position.

Short term compensation based upon such accounting allows traders and managers to take credit for open positions AS IF THEY HAD CLOSED THEM. And I mean truly closed them by Risk Transfer, not simply Risk Offset. This means that the firm settles with the trader for something that the trader has not yet done and that there is no sure indication that the trader could actually accomplish.

That is because the MTM value may or may not be the amount of cash that the trader could get for their position, especially if you include the requirement that the risk is actually really and totally off the books, not simply offset. To know the actual cash equivalent and the difference between that cash equivalent and the MTM value, a firm would need to study each market to understand the trend and liquidity.

This issue is particularly important when valuing the custom non-exchange traded derivatives. Practice is to value those contracts by a replication process, using market traded instruments. There is no attempt to assign any illiquidity premium. This accounting practice is one of the fundamental supports to the practice of trading off market. During the height of the sub prime crisis, it was found that there was no market at all for some of these securities and the MTM process produced completely sham values. Sham because the real clearing value for the securities was much lower than the values that the holders wanted to report.

The difference between the next trade and especially a trade of the size of the position valued and the last trade regardless of the size of the trade is the issue here. And the problem is with treating completely closed positions exactly the same as open positions, by valuing them both as realistic cash equivalents.

Finally, there is the issue of continuing risk. A totally closed (transferred or expired) position has no capital requirement. An open position SHOULD have a capital requirement. Even an OFFSET position should have a capital requirement based upon the basis risk, the counterparty risk.

This discussion reveals an additional risk – the clearing risk.

So the value of the open position needs to reflect one level of clearing risk and the capital needs to reflect a much larger amount of clearing risk.

Bad Label leads to Bad Thinking

August 20, 2009

How many times have you heard the term “Risk Transfer” to refer to a risk mitigation action such as hedging or (re)insurance? It is used in text books and articles about risk management. But, be careful, that labeling is bad and it could just lead to bad decisions.

That is because risks are rarely actually “transferred” to another party. If they were actually transferred, then the consequences would be theirs and theirs alone. However, in most hedging and (re)insurance situations, the risk is usually just “offset”. There are two common terms that are used to refer to the real differences between risk transfer and risk offset.

Those terms are counterparty risk and basis risk. If the risk had been transferred, then there would be no counterparty risk, the risk would BELONG to the other party. In fact, the risk does not belong to them, it still belongs to you. You have paired the risk with an offsetting obligation from the counterparty and you are as safe from loss due to the risk as the counterparty is secure. If the counterparty fails to pay their obligation to you, then you still have the risk and experience the entire loss.

And also, because you have not really transferred the risk, there is a possibility that their payment to you will not be a perfect match to the loss in timing or amount or both. The risk offset might be triggered by a different event than your obligation. For example, the trigger for CDS to actually pay-off are not exactly tied to missed bondholder obligations so the spreads on CDS for a distressed bond may move slightly differently than the actual bond spreads, even though they moved very similarly when the bond was not distressed. A hedging strategy based upon market sensitivities (greeks) is only as good of a fit with the hedged risk as the models used to calculate the greeks.

But this bad terminology is not harmless. If you execute a risk offset and tell others that it is a risk transfer, then they might be quite happy to treat the gains as fully realized. They will not inquire about the offsetting positions, because the bad terminology implies that there are none. When in fact, in any case of risk offset, it should be import to monitor and communicate the risk offsets, highlighting the counterparty risk and tracking the potential emergence of basis risk.

Learnings from the Financial Crisis (3)

August 17, 2009

Gone is not always gone – another of the underpinnings of the market risk business is the constant of trading of risks. However, in the case of sub prime, some of the counterparties in these trades were very intimately related to the banks that sold the risky securities. Sometimes, they were investment funds that were sold by to bank customers; sometimes the banks lent the money to the same party that bought the security. Sometimes, the bank kept the security and bought protection from a counterparty. In each of these types of situations, banks found that they ended up needing to take back some of all of the risks that they thought that they had laid off. Insurers can learn that they need to keep relationships clear. The banking model has long suffered from the idea that they were a relationship business and they would try to do as much business as possible with the customers who they have the best relationships with. Insurers need to be constantly aware of this trap that creates more and sometimes cloudy concentration risk. Both net and gross risks need to be tracked and attended to.

One simple reason for this part of the problem is the terminology that risk managers use. Usually hedging transactions are called Risk Transfers. But in fact, they are almost never a real transfer of risk. Usually they would more appropriately be called Risk Offsets. This sloppy terminology supports sloppy thinking. And the high speed of a trading business left no time for reflection, so the misrepresentation was left totally unchallenged.

Now good risk managers knew the truth and so were concerned with counterparty risk and basis risk as well as contract risk. But in the end, the largest risk turned out to be reputation risk. Banks were usually unable to take the hit to their reputation that walking away from their closely or even semi related counterparties. Especially when those counterparties were funded with customer money.

Lessons from the Financial Crisis (2)

August 16, 2009

It must be ok if everyone else is doing it – Banks were unwilling to miss out and not take part of this lucrative idea. In the past insurers have been caught up in this approach to business as well. The presence of a well respected firm in a market does not make that market good for everyone.

THis factor was so strong that one bank CEO suggested that if he did not allow his bank to continue in the lucrative sub prime business, that he would start to lose key employees and eventually would lose his job to someone who would participate in that business.

That must be one of the last stages of a bubble, when markets become so profitable that many feel that they have to participate.

In this case, the very low interest rates and spreads for almost any other type of risk helped to feed the situation. The sub prime market was one of the only place where there was any spread to be had.

And a major flaw with the “everyone is doing it” motivation is that some things only work if a small fraction of the market is doing it and fall apart when everyone jumps on.

Just like the ferry. A few people can stand on the edge looking at the shore, but if everyone on the ferry stands along that same edge and looks, the boat may dip and might even capsize.

Lessons From the Financial Crisis

August 15, 2009

Short Term Compensation for long tailed risks – encouraged more and more risk taking. Did not hold anyone responsible for the ultimate losses. Solution is longer term compensation. Some insurers have started to make underwriter compensation payout over multiple years.

In some cases, some people were paid with short term compensation while others were paid with long term incentives. Those being paid short term maximized their comp, blew up the company after cashing their check, leaving those with long term incentives with nothing to show for years and years of incentive comp.

This has become a common conclusion from the crisis. But it does not seem to have seeped into the boardroom. This article tells how executive comp is moving in the opposite direction.

But why is long term better than short term? The main reason is because accounting is unreliable in the short term. In the long term, everything is cash, so accounting is not as troublesome. But short term incentive income invites managers to figure out the flaws in the accounting rules that give the best immediate results regardless of the underlying economics (read long term cash).

%d bloggers like this: