Disclaimer

The information contained in the website is solely intended for professional investors. Some funds shown on this website fall outside the scope of the Dutch Act on the Financial Supervision (Wet op het financieel toezicht) and therefore do not (need to) have a license from the Authority for the Financial Markets (AFM).

The funds shown on this website may not be available in your country. Please select your country website (top right corner) to view the products that are available in your country.

Neither information nor any opinion expressed on the website constitutes a solicitation, an offer or a recommendation to buy, sell or dispose of any investment, to engage in any other transaction or to provide any investment advice or service. An investment in a Robeco product should only be made after reading the related legal documents such as management regulations, prospectuses, annual and semi-annual reports, which can be all be obtained free of charge at this website and at the Robeco offices in each country where Robeco has a presence.

By clicking Proceed I confirm that I am a professional investor and/or institutional investor and that I have read, understood and accept the terms of use for this website.

Decline
How quantitative easing impacts the duration model’s performance

How quantitative easing impacts the duration model’s performance

15-01-2014 | Research

With its quantitative easing programs, the Fed has distorted bond markets, especially by reducing volatility. This has negatively affected the performance of the duration model.

  • Johan Duyvesteyn
    Johan
    Duyvesteyn
    Senior Quantitative Researcher and Portfolio Manager
  • Olaf  Penninga
    Olaf
    Penninga
    Senior portfolio manager fixed income
  • Martin Martens
    Martin
    Martens
    Head of Allocation Research

Speed read

  • Quantitative easing has reduced bond market volatility 
  • The duration model’s performance is weaker in times of declining volatility 
  • Tapering will increase volatility, which is favorable for the model

The duration model determines the active positions in the funds Robeco Lux-o-rente and Robeco Flex-o-rente. In 2007 we published sensitivity analyses showing that the model performs well under a variety of market circumstances: in bull and bear markets, with steep and flat yield curves and across different stages of the business cycle. We recently investigated how the Fed’s Quantitative Easing programs have affected the performance of the model.

Stay informed on Quant investing with monthly mail updates
Stay informed on Quant investing with monthly mail updates
Subscribe

How quantitative easing affected the performance of the duration model

We see three potential ways in which the Fed’s bond buying could have influenced the bond market: 1. The buying of government bonds by the Fed may have influenced the direction of the bond market. 2. Movements of the yield curve may have changed as the Fed has effectively anchored the short end of the yield curve. 3. The volatility of the bond market may have changed as the Fed has aimed to keep bond yields down.

Direction of the bond market

To start with the first point, the Fed obviously supported bond markets with its large-scale purchases. Global government bonds returned on average 2% in excess of cash per year over the past 5 years. However, global government bonds also outperformed cash by 2% on average per year in the ten years before QE was introduced. The duration model generally performed well in this decade. This clearly illustrates that a bullish market environment need not impact the model’s performance.

Movements of the yield curve

Regarding the movements of the yield curve, we do indeed see a different pattern in recent years. The Fed keeps the short end of the yield curve anchored near zero, so only the long end of the curve can move meaningfully. As a result we mainly see bull flatteners (yields decline and the curve flattens) and bear steepeners (yields increase and the curve steepens) since the end of 2008. Until 2008 we had more bull steepeners (yields decline and the curve steepens) and bear flatteners (yields increase and the curve flattens) and less bull flatteners and bear steepeners. However, there is no consistent relationship between these curve movements and the model performance.

Negative impact of reduced volatility on model performance

The Fed has anchored short-end yields and pushed down long-dated bond yields. As the Fed also promised to keep this easy policy in place for a considerable period of time, we would expect a dampening effect on bond volatility. Figure 1 shows that volatility has indeed declined strongly since the announcement of QE.

Figure 1. Volatility of US government bonds (as measured by the MOVE index)
Source: BoFa Merrill Lynch

Figure 2. Duration model performance as a function of changes in MOVE
Source: Robeco and BofAMerrill Lynch

Figure 2 shows that the duration model clearly has a stronger performance when the MOVE rises and a weaker performance when the MOVE falls. This relationship holds both before and after 2008. 

The stronger performance in times of rising volatility fits with our earlier anecdotal evidence of strong model performance during equity market crises such as 1998, 2001 and 2008. This is a useful property: the model has tended to provide strong performance just when many investors needed it most. The flipside, however, is that model performance is generally weaker in times of declining volatility. The strong reduction in bond market volatility that the Fed engineered over recent years has thus provided a more difficult environment for our duration model.

Tapering will improve market conditions for the duration model

As the MOVE index is still at relatively low levels and the Fed has just started to taper its asset purchases, a further normalization of bond market volatility might be expected. As the model generally performs better in times of rising volatility, such a rise in volatility should constitute a more favorable backdrop for the model. 

We assign each month since April 1988 (the start of the MOVE index) to one of four buckets based on the level of the MOVE index (figure 3) or based on the 1-month change of the MOVE (figure 4). The graphs show the performance of the model for each bucket.