Home / Uncategorized / RBC paper points to better attribution model for fixed interest

The post-global financial crisis shift to investing in bonds and other fixed interest securities, notwithstanding record-low interest rates in most major markets, is putting increased demands on analytics to ensure investors know exactly what they’re doing. However, the usual tools for such analytics may be flawed, according to a paper by RBC Investor Services.

The retirement period recently entered by babyboomers will tend to prop up the shift to bonds even when the average interest rate level starts to turn and prices fall. Most asset consultants have been warning clients about a fall in western country bond prices for more than a year. New retirees, of course, will start to benefit from higher rates.

But fixed interest fund managers know that the bond market is much more complex than a simple yield return and predictions of the direction of interest rates. Key drivers of bond fund returns include interest rate levels, credit spread movements and steepening yield curves. There are many micro factors impacting on specific securities. And there are many non-traditional instruments, such as floating rate notes, credit default swaps, convertibles and more.

RBC’s paper contends that fixed interest needs its own bespoke attribution models. “From an oversight and investor perspective, a common and transparent approach to pinpointing all sources of risk and return is crucial. The performance attribution model selected must be sophisticated enough to accommodate the range of market instruments as well as the changing nature of the market. At the same time, it must have the capacity to convey the results in an easily understood manner to facilitate timely and informed investment decisions.

“While these characteristics are familiar and reflect methodologies currently employed by equity markets, they are not as relevant to the fixed income sector. A specific attribution model for the fixed income sector is essential.”

For instance, the Brinson-Fachler model, which is widely used for equities, highlights the allocation effects or tactical positioning and the selection effects or security selection. This is referred to as a ‘sector-based’ attribution method. It shows whether the portfolio is under or over-weight in a specific sector. It also quantifies security selection skills within sectors.

The RBC paper says: “But while sector-based models can be customised to any investment segmentation scheme to pinpoint outperformance, they are not consistent with how fixed income managers make investment decisions.

“Fixed income managers typically do not partition the market into broad segments when outlining a strategy. Instead, they investigate sources of return and risk that will make a positive portfolio impact. For example, key drivers of bond fund returns would include interest rate levels, credit spread movements, steepening yield curves, etc.

“The challenge for a manager is determining how to segment a portfolio that allows for many of the dimensions of risk and return to be captured simultaneously. A manager may be able to partition the portfolio by one dimension (for example, interest rate sensitivity), but how would other dimensions be incorporated?”

The solution lies in an attribution system, which is consistent with the fixed interest investment process.

The paper says: “Fixed income investing requires a different attribution approach that pinpoints the sources of risk and return. It is clear that traditional sector-based attribution alone cannot adequately help investors dissect performance figures. “Factor-based models are an improvement although their linear nature does not capture the nuances associated with fixed income instruments. For example, in a more complex portfolio, it is crucial to capture all the optionality embedded in callable bonds, the effects of convexity and all the implications of effective time-to-maturity for mortgage pools.

“To capture all these nuances, the engine driving a fixed income attribution system requires a robust library of pricing functions. These pricing functions can then compute risk (sensitivity) figures consistently and thoroughly. By applying these sensitivities to actual changes in the market environment, an advanced attribution report can be produced. In this manner, all non-linear effects are captured and the attribution model aligns with the investment process.”