Life after attribution
Your managers may be investing in new risk and attribution capabilities, but are they ready for the knock-on effects?
As an investor in the fixed income funds, you may not be too concerned with the relationship between your manager’s front, middle and back offices. As long as your funds are showing steady returns and you continue to receive timely and accurate reports on the fund’s performance, is it reasonable to ask for more?
Maybe not in the past, but the situation is changing. A new risk measurement technique is appearing that allows detailed, in-depth examination of all parts of your manager’s investment process. Known as fixed income attribution, it’s the ability to decompose returns by source of risk, rather than market sector. Because attribution gives feedback on which risk decisions were profitable and which were not, it is a hugely valuable tool, and one that is beginning to be widely used in the debt markets.
Attribution has the potential to bring substantial and often unexpected benefits that will affect all parts of the fund – and hence your returns. But it will also, inevitably, require a substantial rethink of the role of all parts of the investment organisation.
This article shows what the attribution means for you, the investor, and why you can expect much more from managers that have this capability.
Consider the state of fixed income funds management, say, five years ago. A front office manager talked to economists, assessed the markets, made forecasts about expected movements in the yield curve and credit spreads, and adjusted the portfolio accordingly.
At the end of each month the fund’s return came in from the back office. Hopefully, there weren’t any nasty surprises. But if there were, it was seldom clear why they had occurred. There was little or no feedback from results to risks.
Let us now fast-forward to about six months after attribution goes live at a fictional fund manager. Two representative (but plausible) problems have cropped up with our imaginary fund and we see how it addresses them.
Firstly, the fund’s monthly, quarterly and annual performance reports contain far more detail than pre-attribution days. The marketing and sales teams are delighted. For the first time, they can show hard figures to back up their claims about the skills of their credit analysts.
But the results also show two major issues.
The firm’s country asset allocation decisions consistently lost money. The fix, of course, is to stop taking asset allocation decisions. But this is a difficult political issue. The fund’s asset allocations are decided by the asset allocation committee, which includes several of the most senior individuals in the organisation. It is going to be difficult to get them to admit that they should shut up shop and take the fund’s asset allocation back to benchmark. The discussion goes up to the CEO and the board.
Detailed attribution reports show that unhedged, non-parallel yield curve movements are also losing money. In some months, this was more than the fund made on correct forecasts of parallel curve shifts.
This isn’t such a hard problem to fix, but it requires some changes in the way the fund handles risk. So first, it spends some money on more sophisticated risk analysis and hedging software.
But spreadsheets and sector-based analysis just isn’t enough, so next, it appoints one of its quants as ‘interim performance manager’. The job involves liaising with the back office, keeping performance figures up to date, and producing daily attribution reports for the trading desks. These reports provide continuous monitoring and feedback of returns on risk.
The contents seldom come as a surprise to most traders, who keep a close eye on the value of the investments they trade. But when there is a problem, it can be fixed right away. For the first time, the fund is completely on top of its risks. The asset consultants take note, and business increases.
What will attribution bring in its wake?
I’ve identified two common problems in arising with our fictional fund. In
general, there are several knock-on effects that are likely to come out of the ability to run fixed income attribution. These are:
o Increasing pressure to get more frequent performance reports out of the back office, hence allowing more frequent attribution. Why? Because one of the biggest advantages of attribution is thrown away if you stick with monthly-based reporting. By the time end-of-month has come round and you find you’ve lost heavily on that unhedged curve flattening, it’s already too late.
o Increased data storage requirements. Whoever is custodian of the fund’s data warehouse (typically, the back office) will have to understand more about fixed income risks. This is because they won’t just be looking after portfolio holdings and returns; they will also have to record information such as yield curves, security-level duration and convexity, and detailed benchmark data. This will involve people moving outside their comfort zone, and perhaps a merging of middle and back office, or a blurring of roles between the two.
Including attribution as part of the investment process is undoubtedly a good thing.
I have already suggested that attribution will lead to a complete rethink of the investment process – not just for fixed income, but for equities too. In the light of the above suggestions, how will it change?
Let us look at our typical fund’s situation before attribution. The old process model is a one-way information flow. Ex-ante risk data is available as a snapshot of the portfolio’s interest rate risk; performance feeds back into risk in a most rudimentary way. Communication is in the form of reporting, which isn’t used at all in the investment process.
Now consider the fund’s new investment process. With the tools now available, the process can include simulation, stress testing and sensitivity analysis. Just as importantly, the process model has changed from a linear model into a loop.
As soon as this loop is in place, there will be pressure from the front office to increase the frequency at which data flows around it – in other words, on the frequency of reports. Think of this as contracting the loop, getting the attribution results at the time they’re needed.
The best of all worlds is 360-degree reporting for a fund, which reveals:
o where it has been (performance) and what happened (attribution);
o where it is now (risks and asset allocation);
o and where it is going (what-if reporting, stress testing, scenario simulation).
With this process in place, we’ve moved from passive reporting to active feedback. For the first time, the attribution returns are having an impact on the way investment decisions are made.
Is this such a big change? Well, you wouldn’t drive a car any other way. Driving requires constant feedback from the driver into the car – you can’t just do it every 100 metres. You might get away with it on an empty, straight highway. But what goes wrong if you don’t?
Successful managers won’t just be the ones that have an attribution capability. That’s now a given. What will really mark out the leaders is how they use that capability.
Andrew Colin is fixed income research director for the Statpro Group and managing director of the company’s Australia office and is also adjunct professor in the Faculty of Business, Queensland University of Technology, Brisbane. Some of these ideas are explored further in his book ‘Fixed Income Attribution’.
Route to vital information
As an investor, you expect regular reports on where your managers are making or losing money. For instance, at the last quarterly
meeting you might have been told that the fund did well because a
particular tech stock was the subject of a takeover bid, but that holdings in the national airline held back returns.
But this only gives part of the
picture, as it doesn’t really show where the skills of the manager lie. A much more useful view of where excess return came from is to show returns in terms of the types of
Let’s look at this in more detail. A domestic equity manager really only has two types of decision to make – which stocks to buy, and how many of each. For convenience, we usually group related stocks into buckets (such as resources, banks, or food). The ‘which stocks’ decision is called stock selection, while the decision to over or underweight a particular sector in the portfolio - in other words, to buy more or less of that type of stock than would be expected just by looking at the sector’s holdings in the benchmark - is called asset allocation.
The problem with a stock- based report is that many trading decisions are not made with respect to individual stocks. For instance, your manager may have had a view that the banking sector as a whole was likely to rise, in which case he might have overweighted his portfolio with banking stocks, relative to benchmark. But this view doesn’t say
anything about which specific
banking stocks to buy; that is the stock selection decision. And they are both important. A bad asset
allocation decision can wipe out returns made from a good stock pick, or vice versa.
Breaking down the excess return of your portfolio into asset allocation and stock selection contributions is called attribution analysis. By showing returns in this way, it reveals how skilled the manager is at consistently picking stocks over all sectors, and whether their asset allocation decisions are working correctly. Attribution analysis therefore gives a three-dimensional view of your manager’s abilities in a way that a simple returns report cannot.
The situation is more complex for fixed income portfolios, but the basic idea remains the same. Your bond manager has to manage
several types of curve risk, credit risk, and (depending on the types of instruments held) liquidity,
prepayment risk, and many others. The manager is likely to hold views such as “we see a curve steepening this month, coupled with a blow-out in credit spreads’ rather than ‘The 10-year bond looks like a good bet this week”.
As for equity attribution, fixed income attribution breaks down the fund’s returns and show whether a particular risk decision paid off. Just as importantly, attribution will show if unhedged risks exist and are affecting returns. For an industry in which most investment decisions are made in terms of specific types of risk, rather than in terms of trades in individual instruments, this is vital information.