Fixed income attribution ... trends & best practices

A client recently asked me what the trend is regarding "building" fixed income attribution systems and what some of the "best practices" are for these systems.

First, as to the trend, I would say it is to buy not build these systems, as there are many vendors who offer very good ons. That being said, we still see many firms who want to build them. We have and continue to work with firms who wish to do this.

And as for "best practices," I will offer two:

1. Strive to have the system match the investment process; this is a "universal rule" when it comes to attribution.
2. Employ two models: one for the managers (internal use) and one for the clients (external).

You may be curious about the second, so I will touch on this briefly. Many portfolio managers want their attribution sliced up into many effects, so they can fully and completely analyze what took place; however, such details can be overwhelming for the firm's clients, thus we recommend a broader, more intuitive model for them.

There continues to be discussion around the possibility of software vendors including one or two commonly accepted fixed income attribution models. I had posed this question several years ago, in discussions with vendors who participated in one of our surveys, asking if they expected to see a "Brinson-type" fixed income model; not that the model would be structured like the Brinson equity models, but rather models that would be found in every vendor's list. While there seemed to be little support for this, I expect this to occur at some point. As always, your thoughts are invited.

Gaining access to the data

We decided to introduce "guest bloggers," and our first "guest blogger submission" is from Stephen Campisi, CFA. Steve has offered his comments on many of my posts, and for a variety of reasons I thought it fitting to have him provide us with some of his original thinking. As a portfolio manager with a very strong background in performance measurement, attribution, and risk, he provides unique perspectives on a number of topics. Hope you find this post of interest and value.

-----------------------------------

I’m working with the due diligence department of a major investment firm that wants to use my fixed income attribution model to screen bond managers for consideration into their product platform, as well as for performing ongoing due diligence and monitoring of their existing managers. They like the model’s robust analytics and drill down capabilities along with its minimum data requirements.  They expect to implement the model at the sector level, since it’s unnecessary to get issue level data to perform the analysis. So, they began contacting their existing managers, asking only for the characteristics of each sector: coupon, price, duration, weighting and return.  To their surprise (and to mine) they were denied this data from one of the largest bond managers around, citing difficulties in getting the data at first, and then switching their story to the potential legal problems with disclosing this data, since this information is not provided to every investor in their fund. They went back to the bond manager, clarifying that they would not be violating any rules in distributing this data, since the SEC rules only apply to individual bonds and not to the total portfolio or to the major bond sectors.  For example, which investor would be harmed because the manager had noted in a report that the average coupon of the corporate sector was 3.5%?  Seems rather silly, no?  This appears to be more of an unwillingness to provide the data, rather than an inability to do so.

So, let’s consider this in the broader context of the legal concept of “informed consent” which requires providing potential investors with all the information they need to make the best decision that is in their own best interests.  And how can an investor do this when managers are unwilling to provide general information about the major factors and parameters of their funds?  I’m told that one would be hard pressed to get returns by sector from equity managers, so that it’s not simply the bond managers who are so seemingly recalcitrant.  In a post-Madoff age of increased disclosure and due diligence, it seems that fund managers should be required to disclose the major characteristics of their fund so that investors can perform their own analysis and see whether the managers’ claims of the sources of risk and return are justified. It’s not enough to accept a manager’s simple attribution analysis as adequate due diligence into the investment process.  Hopefully we have learned this from the Madoff scandal, and hopefully the fund managers will recognize their responsibility to provide greater transparency into their investment process – starting with providing simple data summaries needed for due diligence staff to corroborate the managers’ claims. This could be a “win-win” for everyone, but right now it’s just a dead end.

Attribution standards

Someone e-mailed me this week asking about the status of attribution standards. He cited my article, "A Case for Attribution Standards" (Journal of Performance Measurement, Winter 2002/2003) and wondered if any progress has been made. Sadly, no.

When the Performance Measurement Forum developed our draft standards close to a decade ago we hoped they would spur some interest and an endorsement, but this hasn't occurred.

Since the release of the article and the draft, a French group (GRAP) released guidelines for fixed income attribution and the European Investment Performance Council also provided guidelines.

There was much confusion when the subject of standards was first broached, with some interpreting the idea as defining what type of model, for example, one must use. But rather the idea is about disclosure: that is, for firms that employ attribution to disclose how they do it (e.g., arithmetic or geometric; holdings- or transaction-based; which model).

Perhaps this inquiry and some additional discussion might inject some renewed interest in this concept.

Linking fixed income attribution

As you're probably aware, arithmetic attribution results in subperiod effects which are linking challenged; that is, if we attempt to link these subperiod (e.g., monthly) results to obtain longer (e.g., annual) period results, using, for example, simple geometric linking (as we use with returns), the results won't reconcile to the longer period excess return: a residual typically results [geez! what a sentence!]. As a result, several individuals (e.g., Jose Menchero, David Cariño, and Andrew Frongello) developed models which eliminate the residuals. These models are typically demonstrated with one of the Brinson (i.e., Brinson-Hood-Beebower and Brinson-Fachler) models.

But what if you're doing fixed income attribution and using, for example, the Campisi model: how do we link these subperiod results?

Answer: the same as we would with equity attribution; that is, use one of the linking models. Special models aren't required to make this work.

TIA III has arrived

The Spaulding Group, Inc. is hosting its 3rd annual Trends in Attribution Conference. This year's program is at the Heldrich Hotel in New Brunswick, NJ. Our turnout is much better than one might have expected, given the economy ... we have more folks here this year than in 2008!!!

Our sponsors for this year's event are:

• The CIPM Program
• DST Global
• Eagle Investment Systems
• RIMES
• SS&C
• StatPro
• Wilshire Analytics.

If you're joining us ... great! If you couldn't make this year's program, you can purchase a copy of the audio (contact Patrick Fowler (PFowler@SpauldingGrp.com) for details.

Attribution ... without securities?!?!?!?

At last week's Performance Measurement Forum meeting in Rome we briefly discussed the issue of "pricing effects." That is, the effect that can arise when your portfolio's prices don't match what's in the index. Recall that we discussed this on November 4.

What I failed to mention earlier is this: what happens if you don't have the index's constituents? For example, what happens if your bond index doesn't give you the securities and the details (such as prices) they include?

A: you obviously won't know whether or not there IS a pricing effect.
B: if there is one, you're out of luck! Unless you can persuade the index provider to give up these details, you can't report on the effect. MEANING that your selection effect will be less than accurate. How "less than"? We just won't know. Sorry :-(

Can you still have attribution if you're missing security details? YES, of course! As long as you have the market values and weights for sectors or subsectors or other groupings you're interested in, you can run attribution! :-)

Trends in Attribution ... near record attendance

In spite of this year's market downturn, The Spaulding Group's upcoming Trends in Attribution Symposium (TIA) will have a very good turnout. We're obviously quite pleased by this.

With roughly two weeks to go there's still time to be a part of this event. It's a single day that's dedicated to this important topic. We've assembled a great group of speakers to address a variety of issues. We will repeat our popular "Fast Attribution" session, which involves a group of panelists who touch on a host of topics in rapid succession.

To learn more, visit the conference website, contact Patrick Fowler (PFowler@SpauldingGrp.com) or Chris Spaulding (CSpaulding@SpauldingGrp.com), or call our offices (732-873-5700).

"New" attribution effects - I

We recently met with a client for whom we're designing a fixed income attribution system. During our meeting the subject of the "pricing" or "price difference" effect came up. This effect identifies the impact when the portfolio and benchmark have different prices for the same security. This is more likely to happen with bonds, because (a) they're less liquid and (b) for the most part they aren't exchange traded, so we probably won't have market prices for most of them.

The conundrum firms face when they encounter different prices is how to deal with them: should they reprice the benchmark with the portfolio's prices or vice-versa? Neither of these options is very good: if you reprice the benchmark, then its return won't match what's published; if you reprice the portfolio then you're using prices which you don't feel are correct and you won't match the return that may be shown in other reports.

The "pricing effect" is a better way to deal with this as it provides visibility without altering returns. It may, however, raise questions which you'll have to be prepared to answer. And, it can only be done if you have the benchmark's constituents (if you don't, then you won't be able to identify pricing inconsistencies).

This topic deserves more detail then we can provide here, so I'll take it up in our newsletter. Stay tuned!

"New" attribution effects

From an equity perspective most practitioners are familiar with the allocation, selection, and interaction effects, while from a fixed income perspective they're familiar with the treasury (aka duration or yield curve management), spread, income, parallel, non-parallel, twist, shift, and again, selection effects. But there are other effects which are sometimes considered.

What happens, for example, when your index prices securities differently than you do? If, for example, you have a bond priced at 99.25 while the index has it at 99.00? If we ignore these differences, then your selection effect might look better when in reality the only thing you did is price your bonds higher (and obviously the opposite can occur, too). And so, what do you do? Well, you could reprice the index with your prices, but then you'd end up with a different result for the index. Or, you could reprice your portfolio with the index's prices, but now your market value and return would change. An alternative would be to break out this difference and report it as a "price effect." While the math behind this hasn't been clearly stated, and there are no doubt various approaches, it's something to consider and explore with your software vendor or development team, especially if you're in the world of bonds or global equities, where there is a greater possibility of having pricing differences.

Okay, so what if you decide to invest in a sector that isn't in the benchmark? How do you handle this? If you use a standard "Brinson" model (i.e., the Brinson, Hood, Beebower or Brinson, Fachler) as it's written, the results may be somewhat nonsensical. There are ways to adjust these models so that the results are better, by incorporating the sector you're in into the index. Personally I prefer breaking this out completely and showing the results as "off" or "out of benchmark results," to give greater emphasis to the fact that you decided to invest in a sector which isn't in the benchmark. There are alternatives here, too, as to how you'd show these results.

Some like to see a "trading effect," which captures your trading activity during he period. One approach to derive this is to calculate your effects using both a holdings and transaction based approach, with the difference being the trading effect. There again are probably other approaches.

Attribution remains a dynamic area. That's one reason we conduct our one day symposium each Fall to specifically deal with this important topic. To learn more please contact Patrick Fowler (PFowler@SpauldingGrp.com) or Chris Spaulding (CSpaulding@SpauldingGrp.com).

Fixed Income Attribution...still a "hot" topic

This week is dedicated exclusively to fixed income attribution. Okay, maybe not everything about the week will be dedicated to fixed income attribution, but we will have a webinar every day on this topic.

For some time we've said that "attribution is the hottest topic in performance measurement," and we believe this is still fairly accurate. But more specifically, fixed income attribution is critically important and "hot." But why does it need a whole week of webinars?

We want to provide those interested in this topic the opportunity to really delve into it, in a way that isn't overly burdensome but that will also provide some great insights. We will have five days of five different models presented. Unlike equity attribution which is somewhat dominated by the "Brinson" models, there is no "Brinson-equivalent" in the world of fixed income. That is, there is no single model which is available from virtually every software vendor. While some of us expect this to happen at some point, it hasn't yet. And so, it's worth understanding some of the differences between the models. There are more than five models so we will have a similar session in the Fall.

Fixed income needs its own model. Why? Because attribution is supposed to assess how the manager's decisions impact their performance. And since fixed income managers tend to manage quite differently than equity managers, they deserve a model that is sensitive to their approach. And not all fixed income managers manage in the same way, so there's a need for some flexibility here, too.