The following sections will focus on the components that can make up a robust performance analytics function.
8.3.1 Performance data: Dynamic information
One of the key requirements to undertake efficient and effective performance analysis is the possession of deal-related performance data with which to conduct the analysis. Five years ago most of this information was predominantly available both to a selected audience (i.e., password-protected web
sites) and in a format that would not easily permit automatic feeding and processing of an investor’s portfolio management systems (i.e., printed hardcopies, faxes, PDF files, etc.). This changed when specialist data providers such as Lewtan Technologies, Trepp, Intex—to name a few—discovered how they could translate bulk data from investor reports into more digestible and usable data formats such as comma-separated files (CSV) or extensible markup language (XML).
Automatic processing has many major advantages compared with manual entry of performance data. It is typically quicker and timelier and also avoids data errors caused by simple typographical errors. In addition, automatic document-processing methods combined with intelligent character recognition (ICR) allow bulk processing of reports in various file formats as soon as they become available. Subsequent automatic data validation (e.g., automatic checks of data formats, types, ranges, etc.) can further increase the overall data quality of such automated processing.
After receipt from the data provider, ‘‘scrubbed’’ or cleansed data can easily be loaded into the investor’s internal portfolio management and performance analytics platform for further analysis.
8.3.2 Key performance indicators (KPIs)
Performance data in isolation are not automatically useful. They need to be interpreted and should undergo a careful thought process to identify which items out of the available data universe of deal
specific information allow meaningful analysis. For instance, 90þ-day delinquencies as a predictor for expected losses (EL) or typical recovery rate (RR) as an estimator for loss-given defaults (LGDs) are such performance data items that, if used and interpreted appropriately, can serve as early warning for deteriorating performance.
When using KPIs supplied by data providers acting as intermediaries, particular care should be taken to understand which of these performance items are directly taken from the transaction’s performance report or whether they are calculated values. For instance, if you are investing in a portfolio of 25 U.S. RMBS bonds and you were to compare KPIs for these individual deals, the calculation of some of them could follow different formulas for different bonds. However, the actual performance parameter name (i.e., CPR) may be called the same for each of these bonds. There is no shortcut to identifying which KPIs are calculated and subsequently used to analyze the relevant formulas. You may then conclude that the discrepancies in calculating these KPIs are far too great to justify usage of the data provided to you and you might eventually end up calculating them yourself.
8.3.3 Credit ratings Rating actions
Long-term vs. short-term ratings
Most structured finance transactions carry long-term ratings, although short-term or financial strength ratings are used for vehicles with a short-term funding nature: for instance, asset-backed commercial paper conduits (ABCPs)—which essentially roll over and renew every 364 days—and some short/
medium-term notes.
Rating actions
Rating actions are any changes to the transaction’s rating. They include actual changes in the rating itself. These changes can be ‘‘affirmation’’/‘‘confirmation’’, ‘‘upgrade’’, and ‘‘downgrade’’. Some agencies differ between affirmation and confirmation, whereby the first action is an agency-initiated action and the latter a reiteration of an agency’s rating following a request from a relevant party, such as the originator or trustee.
Furthermore, ratings can also be withdrawn (WDR) for a variety of reasons. Formal withdrawal of ratings are usually either requested by the issuer or because the agency chose to withdraw the ratings.
CRAs will remove a bond rating and flag this on their websites as paid in full (PIF) when a bond has been redeemed as expected and there are no remaining outstanding balances for the particular bond tranche left.
Rating watch
In addition to the rating itself, rating modifiers may be assigned to ratings which typically either indicate some uncertainty (‘‘evolving’’, with a 50% rating upward probability and a 50% rating downward probability) or some uncertainty about the likelihood of an upgrade (‘‘rating watch positive’’) or of a downgrade (‘‘rating watch negative’’). The definition of rating watch states differs by agency in terms of likelihood and timeframe: The agency that placed a particular bond on rating watch will be undertaking further analysis to clarify the uncertainty around the affected ratings. Once the agency has concluded its analysis, it will then action the rating itself and usually remove the watch status.
Outlook
Rating watch status covers a short-term period of time, typically between 1 month and 6 months.
Rating agencies also indicate their medium-term view on transactions they rate. This medium-term view is called ‘‘rating outlook’’ and covers a period between 12 months to 18 months. Similar to the rating watch, this can either be ‘‘outlook positive’’, ‘‘outlook negative’’, or ‘‘evolving’’.
Intra-agency rating nuances
As if these scales were not enough, I am aware of cases where an agency used rating modifiers (i.e., watches and outlooks) differently, depending on which regional office was undertaking the analysis. This particular CRA had rating outlooks available for the EMEA region for some time, but outlooks did not exist for structured finance transactions in the U.S. and were only introduced in mid-2008. Until the introduction of the outlook rating modifier to the U.S., rating watches were used in some instances instead of the outlook.
New developments for additional rating qualifiers
The credit crisis put rating agencies in the limelight, questioning the transparency, clarity, and meaning of what a rating should and should not express. Subsequent calls for more regulation of CRAs and introduction of additional measures to increase the transparency of agencies’ ratings triggered a major revisit of the currently used rating classifications. The agencies’ own initiatives proposed various changes and amendments to their commonly used structured finance ratings, which included suggestions to include additional information on liquidity, volatility, etc.
Different ways to report rating actions
There are different ways of looking at rating changes over a given period of time and this can lead to different ways of reporting them. This became particularly apparent during the first half of 2008 when literally thousands of structured finance ratings suddenly became subjected to—in many cases—
multiple rating changes by all three agencies. The way in which such rating changes can be measured depends on the particular dimension we would like to report on:
(1) Deal dimension (2) Tranche dimension (3) Rating action dimension (4) Time dimension.
Although it was quite unusual to see changes in transaction ratings more than once by the same CRA in a short period of time (i.e., a month) in the pre-2007 market, this changed completely with the onset of the credit crisis and liquidity crisis. There were periods in late 2007 and the first two quarters of 2008 when the ratings of many deals were changed more than once within a month by the same agency (and by the other agencies at the same time). Most of these actions were driven by the downgrade of the monoline insurers AMBAC, CIFG, FIGC, MBIA, RADIAN, and XLCA. In addition, as part of the vicious circle of downgrades, complex instruments such as CDO of ABS and CDO2s became subject to blanket rating watches and downgrades in short succession by multiple agencies. Consequently, investors holding large portfolios of these instruments were challenged as to how to report the impact of these changes to their portfolios. In most cases, such actions require reporting not only to senior management, but also to external third parties such as the investor’s external auditors, the regulator, rating agencies, and external investors/shareholders.
These actions on their own posed a major challenge for some institutions regarding their reporting capabilities and management information systems. In addition, there was the more generic question of how to report these changes. Whilst there is no right or wrong answer, it clearly depends on the audience and the message that is to be conveyed when reporting a portfolio’s rating migration. The remainder of this section explains differences in the identification and presentation of rating migration over a given period of time (e.g., the second quarter of 2008).
The deal dimension means simply reporting on the number and type of rating actions impacting direct holdings (i.e., the tranche which a particular investor holds on her book) and related tranches
within the capital structure of this particular deal, which could either be senior or subordinated tranches relative to the tranche held by the investor. The different questions that could be asked are
or
(Q1) How many deals have seen rating actions in the selected period (i.e., 2Q08)?
(Q2) How many rating actions have happened in the reporting period for the selected deal(s)?
On a tranche level, it could well be that, whilst our holding has not directly been impacted by any rating action, subordinated tranches below ours may have experienced changes to their ratings.
(Q3) How many subordinated tranches have been impacted by any changes to their ratings?
This could mean that the credit enhancement provided by subordination, overcollateralization, and excess spread could slowly be eroding over time and, if so, our holding may become more susceptible to changes of the rating in the near future. Alternatively, we could also ask
(Q4) How many tranches of direct holdings have seen rating actions in the selected period?
From a rating actions dimension viewpoint, an investor could be interested in any kind of rating change (upgrade, downgrade, affirmation, withdrawal, rating watch, outlook) or any subset of these iterations. If an investor is capable of querying this information for his portfolio to understand, for instance, the rating differential between his internal rating and the respective external ratings, then he could take a conservative stance from a risk management perspective. Such analysis can, for instance, help in identifying bonds where the internal rating is currently higher than the external rating(s) and the internal rating could subsequently be adjusted to the lower external rating—unless there is a concrete reason or the internal rating is considered to be more adequate than the external rating.
Furthermore, rating watches (positive, negative, and evolving) can give an investor a good indication as to whether he may experience potential changes to the rating in the near term. Rating watches would typically cover a period of uncertainty from 1 month to 6 months, in some cases longer—but the agency would normally try to specify this period of uncertainty. If an investor can foresee and pre-empt future rating changes, he may be placed to realize relative value from such instruments on rating watch, either by disposing of them or buying them. This does, of course, require a liquid secondary market. Although the secondary market for structured finance instruments had virtually diminished at the time of writing this chapter, I would hope to see some market activity returning by mid-2011.
The analysis of rating changes for a given portfolio over a different time dimension is probably one of the most interesting questions concerning the subject of rating changes. The two fundamental questions that can be asked are
(Q5) What were the rating changes over the selected observation period?
or
(Q6) How have the ratings moved relatively over a given period?
The answer to the first question is usually simpler. In essence, it involves listing all rating actions for the observed period by any agency.
Tracking rating changes
Ideally, performance analytics for structured finance transactions should be able to measure deal performance in near real time and support the identification of performance outliers compared with initial base case expectations for the individual deal and compared with peers in the same asset class.
Subsequently, such analysis can support pre-emption of forthcoming rating changes once the rating agencies run the performance through their own internal surveillance function and performance analysis models, which takes time. We could now argue that tools to identify and track rating changes if and when they are published by the relevant agency are not necessary if there is a robust enough system in place to support internal analysis.
In reality, however, banks and financial institution are well advised to build and maintain sufficient systems and procedures in order to track rating changes by the larger rating agencies. The key reasons for such a requirement are as follows.
Basel II risk weights
In order to calculate Basel II risk weights, investors in structured finance instruments are required to calculate the capital charges for the bonds they hold and need the rating attributes of these instruments as parameter and key driver of such calculation. Hence, unless the bond’s portfolio size is small (i.e., 25 to 250 bonds), it is advisable to have some automated procedure in place that allows real-time or near
real-time tracking of rating changes.
Investment guidelines for conduits
Conduit managers are bound by the investment guidelines of the structured vehicles they run. These guidelines usually involve calculation rules to determine the credit enhancement levels required for the conduit, which are based on ratings and other factors like various concentration levels for asset classes, countries, and industry. As conduits can contain a larger number of bonds—in some cases more than 1,000—an automated mechanism to monitor rating migration which in turn drives the credit enhance
ment calculation is pivotal. Whilst the market has seen massive downward migration of ratings, there have been occasional upgrades in the same period.
Downgrades—particularly those that involve 10 or more notches (i.e., from AAA to speculative
grade ratings in one single action)—can have a considerable impact on credit enhancement calculation.
In turn, upgrades can reduce the level of credit enhancement required. Any rating migration should be identified in near real time which helps to optimize a conduit’s efficiency.
Central banks’ repurchase agreement guidelines
All central banks—particularly the Federal Reserve Bank in the U.S., the European Central Bank, and the Bank of England in the U.K.—reacted promptly during the credit crisis by establishing new liquidity schemes to enable banks and financial institutions to pledge to a large extent a variety of structured finance instruments and to use them as part of repo agreements. The rules allow highly rated bonds to be pledged in return for cash or government bonds. These repurchase (repo) agreements are subject to strict rules and it is the banks’ or financial institutions’ responsibility to manage the collateral that has been pledged appropriately. This includes notification of the relevant central bank of any change to ratings of assets that are pledged as and when they occur. If there are rating changes rendering pledge collateral ineligible for the relevant central bank repo scheme, then there is a limited window within which the bank has to either replace ineligible collateral with another suitably rated bond or, alternatively, remove the asset from the pledged pool altogether.
Regulatory reporting
In light of the credit crisis, financial regulators have become considerably more aware of banks’ and financial institutions’ structured finance portfolios and their particulars. Some of my clients, for instance, who have large portfolios (>1,000 bonds) have been inundated with questions from the regulators and related authorities with a particular focus on the following aspects of their structured portfolios:
. Bond breakdown by asset class, region, internal rating (if applicable), weighted average life, negative basis trade counterparties, and monoline exposure
. Seniority of the assets
. Suitability of assets for inclusion (or exclusion) of the various government asset protection schemes . Usage of central bank repo facilities
. Tracking of rating changes, etc.
Consequently, even with the best in-house performance analytics function, banks need to have efficient ways and means built into their processes to identify any rating changes in an efficient and timely manner. The following section discusses the advantages and disadvantages of a variety of readily available tools in order to track such rating changes effectively.
CRA client portfolio alerts
A simple way of tracking rating changes is by using the agencies’ own ‘‘portfolio’’ functions. This permits the setup of one or more portfolios based on the ISIN, CUSIP, or bond ticker, which will trigger a specific email notification: any deal-specific or sector-specific rating action would trigger the transmission of an automated email notification for the impacted bonds to a predefined email address.
Advantages:
. Easy to set up and comes as part of the subscription to agency websites at no additional cost . You can use a group’s or department’s inbox so that several people will be alerted at the same time.
Disadvantages:
. Client portfolios regularly require maintenance to reflect changes to the bank’s structured credit portfolio (i.e., bond purchase and asset disposal)
. Client portfolios need to be set up with each CRA individually—some may permit bulk upload of a list of deals, others don’t have this capability yet
. Overwhelming number of emails at peak times of rating changes (I consulted an institution with a portfolio of approximately 1,100 deals which received up to 1,000 emails per week during the credit crisis in 2007—even a dedicated surveillance team of three analysts would find it difficult and time
consuming to work its way through so many emails)
. The email is generic and not tailored to a financial institution’s individual needs as it contains no holding information, etc.
CRA rating data delivery feed
The three main CRAs (and some smaller ones) offer a rating action data delivery feed whereby they provide frequent information on any rating action they have taken for the sector or portfolio that has been subscribed for. Subscribers to this service usually receive the whole data universe of ratings (of the subscribed sector) in electronic file format. Subsequent changes to this rating information are then distributed as ‘‘deltas’’ and subscribers to this service will therefore always have updated rating information available. This can then be further fed into the investor’s proprietary systems.
One of the advantages of using the rating agency delivery feed is that the information comes directly from the source (i.e., the rating agencies themselves). This helps to eliminate the potential for errors in the transmission of this information, which can be due to third parties that are sitting in between the rating agencies and the end-user—such as Bloomberg and Reuters. Although Bloomberg is pretty good, there have been instances where they have gotten the rating wrong on the Bloomberg terminal.
Such errors can and will get rectified by Bloomberg very quickly, but once the information has been corrected, there is no trace on its terminal that would indicate that the information had been wrong.
The rating agencies offer this as a ‘‘real-time’’ service meaning you can get delta updates of such rating changes every 15 minutes or so if you really wanted, but at an additional cost.
Consequently, one of the disadvantages of having a direct rating agency feed is the cost. This is, however, further exacerbated by the fact that if you want to subscribe to such a service from all three rating agencies, you will need to enter into three individual subscription agreements. Some of my clients are paying in excess of $200,000 per year for having such a ‘‘service’’ and it makes me wonder whether or not it is really worth having. I guess to some extent it comes down to the level of information that is provided as part of this service. If you feel that you require, for instance, detailed information on rating outlooks as well as additional rating-related parameters that some of the agencies started developing in order to address some of the rating shortfalls experienced during the credit crisis, then I would say, yes, take a close look at these products. Contact the agencies’ product specialists for such data feeds and request a sample file to give you an impression of what the actual file you would receive as part of this service looks like. Furthermore, having the actual file helps you to look through and understand whether or not the specific datapoints you are after are actually contained in the file. Furthermore, when you talk to agencies, enquire about the data-licencing arrangement for this product and what internal uses of this information are permitted or not. In addition, you may find that other divisions in your area may be interested in a similar feed of rating data (e.g., business areas that would use corporate rating information rather than structured finance rating information). You may discover that agencies can offer you considerable benefits if you increase the sectors covered by such a subscription.
The agencies usually refer to the ‘‘added value’’ of passing (already publicly available) rating information on, but the only value I think they are referring to is the electronic availability of the data—which means data can be fed directly into your proprietary system. If you go down that route, however, you can expect some system-tweaking to be necessary as there is no standard format for these rating extracts: the files you will be receiving from the different agencies all have a slightly different file