The Moving Face of Valuation: Transparency, Independence and Consistency

One of the interesting areas to receive more scrutiny in the past 18 months is that of valuation governance. Specifically, the process of choosing a final price for a particular asset or liability on the books is being questioned. The increased scrutiny has been driven by two major events – the implementation of the FASB ASC Topic 820 Fair Value Measurements & Disclosures (“the new FAS 157”) and the increased probability of default driven by the Bear Stearns and Lehman closings. The new FAS 157 did not change any definitions of fair value, but it did change the application guidelines. Whereas the original practice was often to focus on the original transaction price, or entry point, the new guidelines specify closeout levels or exit prices. This requires new procedures, new data sources, and new modeling assumptions about liquidity for Level II and Level III assets.

The market volatility and headlines of the past 18 months played their part as well, of course. The large defaults made market participants double check their assumptions. Also, assumptions based on historical inputs proved to be much less reliable.

In the pre-turmoil arena, the process was fairly basic, even in the over-the counter markets – find between three and five prices, throw out the high and low, and take the average of the remaining. However, this simplistic process entails three assumptions: 1) it presumes that prices are available, 2) it presumes that they will cluster together in some obvious fashion and 3) it presumes that all price givers – and thus all prices – are equal. Those assumptions no longer hold in all cases.

The availability of pricing sources seems to be directly correlated to the liquidity of the instrument! Even interbank dealers who make a market in a particular financial instrument are more cautious about providing levels these days. Partly their reluctance is due to a heightened sense of legal risk. Client levels and bank books and records levels used to be considered separately. Now there are concerns about liability if a client pricing level is significantly different than the bank’s own price. Difficult questions are raised when a salesperson provides a price level significantly below the client’s original price. Neither the salesperson nor the trader may be able to provide sufficient information to clarify the discrepancy. The reasons may range from market illiquidity to modeling assumptions.

In today’s market, much wider levels of bid/offer spreads are common with the less liquid instruments. One of the challenges these spreads presents is that the valuations obtained are likely to show more dispersion or be materially significant. Whereas pre-2007, a choice between a high price and a low price might be 25 basis points (equivalent to a quarter of a percent), now it might be over 100 basis points (or one percent). For CDOs, the difference can be more than 25%! Admittedly, part of that difference is purely arithmetic. As prices drop to lower levels, the translation from dollars to interest rates increases.

The last point, determining the equality of price givers and final prices, is the most interesting from a process viewpoint. How is it best to distinguish between providers? Are internal prices best? Are interbank dealer prices materially better than vendor prices? And amongst vendors, are evaluated pricing services significantly better than open source providers? Even more interesting, can you assume that the relative rankings are stable? Will a dealer price always be better than a vendor price, or do you have to justify their ranking every time?

Having laid out the problem, the following section presents various solutions.

The greatest challenge is to be able to find valuations in the first place. Secondly, these valuations should be sufficiently transparent in their inputs and methodology so as to provide a basis of comparison. The good news is that there are many services to provide modeling, if your firm does not have such resources internally. It is a good practice to have at least one model brought in-house so that a member of the firm – who has a vested interest in the correct valuation –is familiar with it. There are many firms that provide such tools. This can be the most cost effective solution when you have a book of hard-to-value instruments – 15 or more individual items. There are also custom valuation services that will provide a white paper on appropriate methodologies for specific instruments. However, this approach can be expensive unless the firm has only a few such items on the books. There are also several academic websites with papers that are either free or low cost. However, that decision requires an internal resource who can evaluate their utility. Despite the fact that interbank dealers are becoming more reluctant to provide levels, this option should still be explored. However, it is best to confirm with your trader before going down this route with multiple dealers. He or she may not want certain dealers to know of the position. Anonymity in trading has a value. There are also more general pricing services that provide levels for many OTC instruments, and even structured deals.

After the data is obtained, one of the cleanest ways to provide a defensible price is to create a price scoring system. This method ranks the provider’s prices by date and type. The ranking is then mapped to a weight for the price. The final settlement price is the weighted average of the price inputs. For example, interbank dealer prices are generally considered the highest quality, but they can be the oldest by the time they are delivered to a client. Creating a matrix of weights based on those two variables, and using them to determine the final price can have two benefits. First, the method is consistent so that the books and records valuations are reproducible by an external source. Second, in cases where all the source providers are equal, the resulting price can reflect the end price that would have been found using previous methodologies. The latter is helpful, in that significant price breaks from prior practice raise red flags. Below is an example.

This example is fairly simplistic, with transaction levels getting ranked first, then interbank dealers, then custom price assessments, then evaluated pricing services, then general pricing services. The end result is lower than an average in this example, because the “best” price is the lowest one obtained. If the prices were simply averaged, the settlement price would have been 18.25. The influence of a poorer quality, stale price would have been significant. If the high and low were thrown out as outliers, the final settlement price would have been 15. The influence from the best price would have been reduced.

Internal models can be inserted based on the expertise and credibility of the individual firm’s modeling capability. As the ranking declines, so does the weight given to the associated price. The relative weights can be subjectively decided and then should be consistently applied. Equally, current date prices are weighted fully, but the weighting declines as the age of the price increases. This approach can be a useful way of incorporating transaction prices from a prior quarter with model prices that are current. The final settlement price in our example of 14.35 is skewed towards the best quality price, and yet is still completely defensible.

Another way to improve governance is to rank, or score the end prices. Scores can then be aggregated across portfolios. Low scoring portfolios might be subject to a reserve process or trigger a signal for further analysis. Scores are set up using the same inputs as above. In this case, it might be conceptually easier to use an inverse ordering system so that scores are higher for end prices obtained using all dealer inputs as opposed to a level obtained using general vendor levels. Scoring a portfolio can help particularly when transaction volumes are not homogeneous – that is, when a few larger transactions can dominate a book of smaller transactions.

Valuation questions have been known to dominate credit or risk committee discussions and take up valuable management time. Creating some objectivity on these crucial questions can permit senior management to discuss the business implications of a position – rather than whether it should be marked at 6 or 16. Simultaneously, risk to the total enterprise is reduced when books do not need to be revised over disputes about valuation on a single security or a portfolio. In order to follow the new FAS 157 guidelines and comply with “best practices” in valuation governance, the key is the consistency of the process chosen and the transparency of the reasoning behind it.