Our returns based approach gives you the insights you need without the usual data overhead
Fixed income is often put in the too hard box for attribution providers. But it’s where we started building attribution systems when we worked at UBS and we still feel very at home here. Nevertheless, we understand why people think of it as difficult. There is often a desire to attribute every last basis point which can lead to a huge extra data requirement which we think is unnecessary. The same insights can be had with a lot less work and a little more thought on how to present the results.
Our model has been developed over time with a lot of reference to portfolio managers. A full write up will soon be available in the Journal of Portfolio Measurement (Winter 2014). We attribute performance at the security level, so in that sense our model is bottom up. But the attribution itself is essentially top down. With a lot of flexibility for sector design, we have found that it has a very wide applicability.
Within duration we calculate the overall effect from the portfolio active duration. We then attribute each market based on its weighted duration deviation(WDD). Finally we capture where the exposures were on each market curve. Together these give good feedback on whether the overall rates positioning was correct.
Each portfolio manager looks at their portfolio differently. A Corporate manager may focus on the subordination within financials. A Sovereign manager may worry more about their Spanish exposures. By keeping the sector definitions completely flexible we give users the opportunity to attribute portfolios in the way that they think about them.
The attribution of sector management follows the same approach as for duration. We look at the overall active spread duration of the portfolio, the active spread duration of each sector and then the security specific element is analogous to curve. This symmetry makes the model very approachable and easy to understand. We also have a yield management term to complete the main elements of the model.
Unlike with equities, derivatives are part of the bread and butter of fixed income. Our performance based approach makes it simple to handle them. As long as the accounting system recognizes them and can price them then we can handle them too. No additional terms and conditions data is required.
A key component of fixed income attribution is the way that the curves are constructed to calculate government returns. We have experimented with a lot of different methodologies looking for one that is robust and reliable. It handles outliers and gaps well, doesn’t produce results that can shoot off to be positive or negative (something spline fits are notorious for) and works for all markets whether there are four or four hundred points to fit.