Wednesday, August 3, 2011

Auditing Derivatives: Think of what can go wrong Middle Office (Valuation)


 This article appeared in the Malaysian Accounting Journal, May-June 2011

In the previous issues, we were introduced to different functions of an investment bank (the Trading Floor, P/L Control Group and Risk Management) with the focus being the auditor’s mindset in auditing derivatives in these areas – thinking of what can go wrong.

In this article I introduce the Valuation Group, also referred to as the “Independent Price Valuation” (IPV), the “Price Verification” or “Price Testing” group. The IPV function can exist within the P/L Control group or it can stand alone to maintain its independence.

The IPV function
The valuation function should not be confused with the Accounting role in the investment bank. The latter operates on a higher level in determining adherence to accounting standards in booking trades into the financial statements.

The IPV team in an investment bank reviews the value of all trades recorded by the P/L Control Group. The duty of this team is to tell management every month, if the trades in the books have “fair values”. In other words, if the bank was forced to liquidate the trades, the bank would be able to realise those values on the books. This assumption is closely connected to liquidity and we will observe later that this is a challenge for bespoke (custom-made) trades.  

Trades are normally categorised in a few groups for price verification to take place:
§  Trades which prices come straight from the market – for example shares, bonds and “flow” derivatives like futures. The only focus here is that the products are recorded at the correct market prices taken at the end of day with appropriate bids or offer rates, depending on the positions – long or short.
§  Trades which are not as liquid – for example illiquid bonds and derivatives (traded over-the counter) written on illiquid credit names where market prices are not always readily available.
§  Bespoke trades – trades that are highly customised for the individual customer and traded over the counter. Most of these are structured derivative deals.

The IPV group will break down each trade by its parameters and assesses if the parameters are accurate and observable in the market.

What can go wrong?

The auditor would be very interested in the IPV group, as it is the front line unit to ensure that trades are booked at fair values. But before the auditor begins his review, it will be advantageous if he just pauses and thinks of what can go wrong to identify the risky spots in this potentially vast and complex unit.

Time Pressure and Lack of analysis
The IPV group reports to management on a monthly basis, the variances between the IPV valuations with the trading desk numbers. If the variances breach the limits, the traders and controllers must adjust their books according to the IPV Group’s numbers.

The variance should come with in-depth analysis to be of value. A basic example: if a portfolio of cash bonds shows negative variance for credit spreads, the IPV group must know which bonds are causing the variance and why is the IPV spread different from the traders. Are the traders not updating their curves or was there a specific event that occurred just before month-end.

Very often, the IPV group is only given about 10 days after the month-end to produce the variance report for the whole bank. The IPV group is frequently under great pressure and tends to leave the analysis part of the variance for the last day. As a result, the variances are just numbers that are not backed by profound explanations to convince management and traders.

It will be useful for the auditor to get a feel for the time pressure constraint that the IPV group may be undergoing, to gauge the potential limitations in the quality of the variances.

Completeness
Depending on the flow of information, it must not be taken for granted that all trades will automatically flow to the IPV group. If a manual process is involved, the process is exposed to errors, especially in the short time frame given to conduct the price testing.

The auditor should be keen to ask about the method of transfer of trade details and review for completeness, to ensure all trades and all parameters are included in the price testing.

Challenging parameters for Illiquid and Bespoke Products
Say we have an illiquid option written on a rare currency. One of the main parameters for options is volatility which is implied from the market. The normal practice is just to use parallel options on other currencies with the same strike and maturity to extract the implied volatility. However thought must be given to the liquidity of those main stream options vs. this rare and illiquid option. If the structure were to trade in the market, it is very likely it would not have the same price.

Bespoke products tend to use parameters that have to be calibrated from the market prices. One example is calibrating the correlation parameter for CDOs (credit debt obligations) from CDO index prices like the CDX and iTraxx. At times during the market panic of 2007, the index prices were so volatile, it made calibration impossible. The calibration models were not built to deal with abnormal prices in periods of market stress.

When a trade is illiquid or bespoke, how is one sure that the model price is acceptable? Some prefer to obtain quotes from more than one trader in the market. The type of dealer and its size in market also matter.  The auditor should look for these issues and ensure that proper procedures are laid out to handle these kinds of trades.

Structured trades – look behind the screen
Structured trades are unique products and often involve a mixture of financial instruments. Ideally they should be reviewed trade by trade with reference to the trade documentation. There have been many instances in my experience abroad where the IPV group only relies on the details booked in the system or spreadsheets.

There are cases when not all of the trade features are captured from the trade contract on to the system. For example, a trade that involves an option for the client to sell back the deal to the bank after a certain period is actually a put option that has a value. Depending on the drivers that influence the client to call back the trade, the option could have different values at different times. Frequently this feature is excluded from the trade parameters under the excuse of the event being remote. We know too well now (especially after the 2007 crisis) that remote events (popularly known as “tail” events) must be priced in.

The auditor should look out for structured trades and enquire if the IPV has reviewed the structures against the trade documents. Next he should be comfortable that all features that construe the structure are being valued and tested in the IPV group.

IPV Variances and Limits viewed as being “soft”
For highly complex and bespoke products (like the bespoke CDO), variances can move erratically from month to month. This occurs especially for products where no standard models exist in the market. Due to the illiquidity in prices, banks subscribe to a pricing service (like “markit”-  http://www.markit.com) to gather prices for similar kind of bespoke CDOs in order to calibrate important parameters like  correlation.

There are times when the variances can go either way and it is almost impossible to analyse the underlying reasons behind calibrated parameters.  As such management tend to assign these kinds of variances as “soft” which means that the variance numbers do not warrant adjustments to the books but is taken more for information purposes.

The above is an example of the extent of subjectivity involved in the CDOs where there was no definitive solution in valuing the product as correlation changed in the market.

If the auditor is faced with a similar situation of “soft variances”, he cannot afford to take his eye off this matter as it may be a sign of a potential mispricing that may blow up in the future.

Variances just below the Limits
At times, variances reported are just below limits and therefore do not warrant any adjustments to the books. This could be just by chance but the inquiring auditor’s mind should not rule out the possibility of human nature when one naturally wants to avoid management’s attention and does not want traders on his back, scrutinising his numbers before adjusting their books.

Auditors must have a second look at variances reported just below limits. Moreover, limit levels need to be reviewed to determine if they are reasonable. The auditor will also want to know if the limits themselves were adjusted and why.

Altering pricing models
The IPV group should be alerted when pricing models are changed in the bank. They must be comfortable with the justifications behind the model change. This is to ensure models are changed to reflect prices more accurately and not to abuse the values to shift in a more favourable manner.

Conclusion

We can appreciate now how analytical and detailed the IPV function is and how crucial it is for this unit to operate well in the investment bank. As the devil is almost always in the details, the auditor must have a clear mind in identifying the danger points within the tangled maze of systems and pricing models.


No comments:

Post a Comment