Valuation

Are valuations a problem?

If they are I think you will find this worth five minutes of your time.

A few weeks ago I was reviewing some of this year's work. One thing struck me. Product valuation is a problem for many banks. Valuations cause difficulty for support areas. In particular finance, middle office, settlements, audit and regulatory reporting.

You will no doubt remember the old adage "don't do it if you can't price it". Why the increased interest in valuation?

Because accounting and regulatory changes mean there are more support staff involved in producing valuations. This information then feeds into the profit and loss, risk management and regulatory reports. And these reports go wrong because the valuations go wrong.

That's when the disagreements start; a lot of time is wasted and eventually there may need to be an "adjustment".

One of the main contributing factors to this difficulty is the improvements that have been made in systems. Don't get me wrong. These systems have significantly enhanced the ability of banks to manage risk.

But at the same time the consequences of human error are even more costly. Once the trade is booked valuation is made by pressing a button and there are few subsequent questions. That is until we disagree with the trader or the counterparty.

Can this problem be prevented? Not entirely but it can be significantly reduced. From my experience there are 9 points that will help you do this. So here is some practical advice based on quite a few years of experience.

When you read these you may smile. They may seem obvious to you. But it isn't obvious to many people who have to value deals and that's the difficulty.

Some products trade like apples and pears. If you go to the fruit market every day, you see exactly how much apples and pears cost. Despite what dealers tell us, some financial markets work just like this! Foreign exchange, liquid bonds and futures are just like apples and pears. Their price is dictated by supply and demand. If you go to the market, find the price, you can value the position. Are there any complications? Not really although you may wish to apply the difference between buying and selling prices, (bid-offer) spread.
Other products trade as cash flows. When you enter into these deals you pay and receive streams of future cash flows. Forward rate agreements, interest rate swaps and currency swaps fall into this category. Valuation is straight forward. You need to "explode" the products into their cash flows. These are then discounted to give you the present value. Most systems will do a good job on this. It's one of the reasons why you collect yield curve data. Are there any problems? Yes, a few. The maths can be a little tricky. You may need to add credit spreads when discounting some non swap based cash flows. And for currency swaps remember that basis swaps do not always trade flat/flat.
Then there are the "complex" products. Generally these are options. They include caps, floors, swaptions and foreign exchange options. They are valued in the same way as swaps; you discount cash flows. But there is one difficulty; the cash flows are uncertain. You need to work out the probability that they will arise. That's where an option pricing model comes in. There are different models for different products. Are there any problems? Yes. In theory, sold option positions can lead to unlimited losses. They make regulators and managers nervous. That's why valuations need to be robust. The complexity of the models can lead to difficulties. In particular inputs like implied volatility can be misunderstood. It's one area where banks have lost millions.
What about illiquid products? These can really give you problems, (and a few headaches). I know; in one of my jobs I spent five years trading emerging market credits. Getting a price on a good day can be hard enough! There are no simple solutions. Some banks let traders value these positions. I don't really like this; segregation is not being applied. In some markets like credit derivatives, you can buy third party data. For other products there is no single solution that I am aware of. But experience can go a long way; when was the position last revalued? Who did it? Has the price moved? If so, was that move consistent with other markets? For example, did the implied asset swap spread change in a similar way to credit default swap pricing? Answers to questions like these will help you.
Perhaps structured products present the greatest challenge. Frequently these are bonds specially designed for investors. You may see them if you have a medium term note programme. An investment bank structures the note and offers you the swap to hedge. What do you get? Cheap funding. The investment bank makes a profit and the investor? ...who pays for the yachts? You have a back-to-back transaction so why is this a problem? There are two reasons.

First you have reputation risk. Caveat emptor may be the defence but if you can't price the risk you don't know fair value. What's more the note has your bank name on it. Second you have counterparty credit exposure on the swap. How can you quantify this? You need to value the trade. This is not easy. It involves breaking the product into the components. Here is a simple example; a collared floating rate note normally involves a FRN plus a short cap and a long floor. You can get a handle on the value of this FRN; calculate the implied forward rates and margins, discount them and then add the option valuations. It sounds easy but from experience these things can be tricky and sometimes embarrass traders.
I forgot something. The accuracy of your valuation depends on the quality of your data. If the trade is booked incorrectly the valuation will be wrong. This sounds simple but in a recent ISDA survey* some firms were rebooking up to 21% of trades. It is one of the reasons why the use of Financial Products Markup Language is on the increase. Another source of error is the integrity of the market data. Obtaining clean and consistent data is difficult.

*(ISDA 2005 Benchmarking Survey)

Is there any more that you can do?

A lot of people won't tell you this. One of the most sought after attributes for support staff is the ability to recognise when things are going wrong.

Sometimes in the back of your mind there is that voice that says "No... something is wrong here". Call it intuition, call it experience; it must have happened to you.

How does this help? Because the experienced operator will not just accept what the system says. They will have the confidence to challenge it when they think it is wrong. How can you instil this experience? It's not easy but this will help.

Get ownership of responsibility. One of the dangers of using computer valuations is that the process is mechanical. You do the same thing every day, routine takes over and complacency creeps in. Every bank I know has experienced problems as a result. You can improve on this. You need to stress these things; the importance of accuracy, the consequence of assumptions and the significance of escalation procedures. How do you do this? Learn from mistakes. Build your own set of case studies. Include the invalid risk reports, the profit and loss errors, the nostro and reconciliation problems. Show how much management time was wasted. Explain the damage done to morale and credibility. A top business school couldn't offer you better material.
Improve understanding. It is just a lot easier to value transactions if you know how they work. Even a basic understanding of the products will help. Here is an example. Let's suppose you work in collateral management and dispute the value on a portfolio of swaps. After reconciliation you identify just one swap that is causing the problem. What can you do? Replicate the trade on a simple spreadsheet, include all the floating legs and the accrued interest and you will be in a much better situation to determine where the error lies. Furthermore swaps are used for interest rate hedging. Any positive mark-to-market value is a credit risk. And that's why you call for collateral. It can reduce your losses if the counterparty defaults, it may increase return on capital. That's why your job is important and why errors are expensive.
Develop insight. There is a direct link between market prices and valuations. This is helpful. It is used by risk managers for their reports. You can use it to your advantage. It provides you with a test of reasonableness. Let me explain. I was working with a client several months ago. They used a third party to provide trade valuations. Some of the valuations had not changed. Was this right? No, a quick comparison of yield curves and credit spreads showed that the valuation agent had not been doing its job properly. Are there any simple techniques that you can adopt that may help? Yes. "rules of thumb" really can assist. There are some simple techniques that traders use. These are delta values; they provide a rough estimate of risk for particular changes in interest rates or foreign exchange rates. Use them. They give you an estimate of how trade values may alter. What is more the calculations are simple. Your system will already produce them.

First Published by Barbican Consulting 2006