Direct naar de content
Dit artikel wordt u aangeboden door State Street Global Advisors
De redactie van Pensioen Pro draagt voor deze inhoud geen verantwoordelijkheid.

Factor investing is trickling into bond markets

PENSIOEN PRO PARTNER ROUNDTABLE

Factor investing has made substantial inroads into the equity market, gaining traction with research-backed factors such as value, momentum, and volatility. In fixed income, adoption has been lagging. Research on factor credit models is improving rapidly but the lack of live track records remains a key hurdle. More certainty around actual factor premiums in bond markets, transaction costs, and scalability is needed.

Many quantitative models have been proving their value in equity markets, so it is not a strange assumption that they could also work in fixed-income markets. But do they work in a similar way? Do they deliver alpha? And how about the scalability? Is there enough research on this quantitative segment? These were some points of discussion at a recent Pensioen Pro Partner Roundtable dedicated to quantitative credit investing.

Most of the existing quantitative models in corporate and sovereign bond markets still lack sufficient (live) track records, especially in European fixed-income markets, and investors require more proof on (out)performance, transaction costs, and scalability. The main takeaway from the debate is therefore no surprise: it seems unlikely that factor credit models will replace traditional strategies in the foreseeable future nor taking a sizable chunk out of the pockets of traditional fundamental discretionary credit managers. However, there was another important conclusion: with a pragmatic approach from an experienced asset manager, quantitative models can offer substantive value.

A lack of research

The equity market has had decades of factor-based research, which has allowed for a more comprehensive understanding of how factors like value and momentum behave. However, credit markets, remain underexplored. That lack of research on factor credit investing is an important first hurdle for implementation, says Sevinc Acar, Head of Public Markets at NN Group.

Acar: “When I did a Google search on factor investing/credit, the catch was very slim. I did find some research, even one with a track record of 10 years, but it doesn’t compare to what you find on equities. The research is still too limited, making it hard to see how these models will play out in the long run. I wonder if there are fixed-income investors that use these quantitative models.”

NN Group is not one of them. According to Acar, this is not only because of the limited research but also the expectation of a higher turnover (and transaction costs) of these models compared to the traditional, more buy-and-hold strategies.

Acar’s observation about the scarcity of research as a threshold for the implementation of factor credit investing was echoed by other participants at the roundtable. “Some years ago, we started with a small allocation to factor investing, but as I look at our current allocation, fundamental is the main game in credit. There is a threshold for moving further into factor investing, in the sense that you must understand better how these quantitative credit models work, and then feel comfortable advising it to clients”, states Rik Bekkers, Fiduciary Manager Fixed Income, at the largest Dutch pension provider APG.

Is there added value?

Pension provider PGGM has allocated a sizable part of the equity portfolio of its client PFZW towards factor strategies, but for now has not allocated to factor investing strategies in corporate credit. Reason being not so much because of the limited availability of research, says Mark Geene, Senior Investment Consultant Fiduciary Advice. “We have done a lot of quantitative research on credit, in addition to reviewing academic and industry research on factor investing in credits. In addition, we have several systematic models for government bonds, both in developed markets and emerging markets, but they are supplemental to other elements in the respective investment processes.”

Geene lists some of the doubts and hesitations that investors have on this topic as they are not sufficiently convinced of the added value of factor models in corporate credit portfolios. “The relatively short track records, especially in terms of implementation in addition to back-tests, are not yet convincing enough in terms of return and risk. There is not sufficient capital, i.e. that have put billions at risk, in live track records available to convince investors that sizable portfolios can be implemented in all market circumstances. This is especially a concern in less liquid parts of the credit markets like high yield.”

Back-testing of models only is not enough, states Geene. “You cannot only look back with back-testing to what has occurred because the way bond markets are trading nowadays is different from 10 years ago. In addition, during a stress period like Covid there were not many factor strategies alive with sizable allocations. You need several of these stress events to assess how the actual implementation of back-tested portfolios works out. Also in equities, you have a continuous development, because the world is changing. One of the rationales of past factor premiums could be different going forward. In addition, for PFZW it is important to assess how factor investing in credits can be implemented to bring it in line with its 3D investment belief, i.e. how factor investing in credit can equally balance return, risk and sustainability.”

A frontrunner in factor credit

State Street Global Advisors (SSGA) has long been a frontrunner in factor credit strategies. Lead Portfolio Strategist for systematic fixed income strategies, Arkady Ho, agrees that the factor investing landscape for bonds is far less mature than for equities. However, he emphasized the potential benefits of systematically analysing bond portfolios. According to Ho, it can deliver alpha and a low correlation with traditional, active fundamental credit strategies.

The timing is right. Previously, market inefficiencies in bond markets allowed traders to generate alpha through mispricings. However, with the rise of electronic trading, portfolio trading, and ETFs, these opportunities have diminished. The market has become more transparent. That also means that data about individual bonds – their pricing, liquidity, and risk characteristics – are much more readily available as input for systematic approaches.

SSGA currently boasts two live strategies in U.S. high-quality corporates, with combined assets of almost $700 million, with more than ten months of live track record. The model is based on three factors: value, momentum, and sentiment.

Value, momentum, and sentiment

Ho explains the SSGA model: “Each factor is scored to ten, and we equally weigh them into a combined score. This is for 8000 bonds, US investment grade, daily. Value is bond-specific, and we investigate the issuers in a peer group, zooming in on sector maturity and quality. Then there is also a comparison of fundamentals, like financial leverage and interest coverage, and the excess spread to peers or ESP. The other two, momentum and sentiment, are both cross-assets, so they are based on an issuer’s equity data.”

Factor investing can deliver alpha and a low correlation with traditional, active fundamental credit strategies

Arkady Ho
Lead Portfolio Strategist for Systematic Fixed Income Strategies at State Street Global Advisors

“For momentum, we look at a weighted average of equity momentum in an issuer’s stock prices. For sentiment, we zoom in on an issuer’s equity-short interest. An issuer that has a high degree of equity short interest gets penalized.”

Factor premiums

The major sticking point for participants was the uncertainty surrounding factor premiums in fixed income. Factor premiums – the extra return investors can expect from exposure to certain factors like value or momentum – are well-documented in equities. That is not the case with bonds. Geene: “Do we have sufficient evidence, robust, both historic, but also academic and in terms of economic rationale, that factor credit premiums actually exist? And will they work in the long run? As pension providers you do not want them to last only a few years. Or are they only a pure risk factor without providing an additional return premium?”

One of the first challenges is that there is no consensus on which factor definitions should be used, according to Bart den Boon, portfolio manager Investment Grade Credit at MN. “Within equity markets, the price-to-earnings (P/E) ratio is often used as a value factor. But still, a quant investor needs to decide whether to use a trailing P/E ratio, a forward P/E ratio or a through-the-cycle P/E ratio. The need to make choices also exists in the corporate bond market. The momentum factor applied to a bond can be based on the performance of the company in the equity markets, or it can be directly based on the momentum of the (excess) return of the bond itself. And to go a step further, a manager needs to decide which lookback period is used (three months, six months, etc.). In the end, it is critical that the factor is robust to definitions.” Den Boon also points to the risk of “overfitting” models to historical data, making them less effective in real-world conditions.

Ho: “That is why a quantitative model must be transparent, not a black box. We also did not just take the factors as a given but did a lot of work to validate the research and develop the best approach. So, for example, in US Investment Grade, you end up with a concentrated portfolio of the highest-scoring 100 issuers, but you don’t want just a handful of issuers driving your performance, right? So, we randomly moved from 100 issuers to 200 issuers to 400 issuers. As long as you’re tilting the portfolio to the highest-scoring bonds, no matter what, moving it up from 100 issuers to 400 issuers, you still get that outperformance. The scores are driving performance.”

Bekkers asks if there is a sweet spot in excess return. “Do you get the highest excess return with the top 100 issuers, or with 200 or 400?” Ho admits there is some degree of alpha decay, but “it’s not substantial”.

Ho continues with an example of factors having added value: “We saw some volatility in July and August. Momentum can add value in those volatile periods. Equity momentum turned in Intel. That got us out of Intel before they were downgraded. In times of stress, the momentum signal can play an important role in reducing exposure to issuers experiencing negative returns on their share prices.”

Scalability and complexity

Scalability is an important challenge for factor credit investing. Does the strategy lose some of its ‘magic’ when you flood a particular market with billions that end up being invested in a small number of bonds? Bekkers: “Liquidity can already be a concern in the corporate bond market, and factor investing could exacerbate that by concentrating positions in specific securities. How far can you go with a factor strategy before size becomes too big a threshold to operate effectively on that?”

The size of markets is indeed an important consideration, agrees Acar. US bond markets are much larger than European markets. In addition, euro markets have less depth and history than USD markets, making it potentially more difficult to test and operate quantitative models or to transfer US models to Europe, says Acar. “Go far enough back in history and you see pesetas and guilders pop up in your Excel sheets.”

Acar also finds the complexity of quantitative models worrisome. “From an asset owner’s perspective, why would you want to build extra complexity into your portfolio? At the end of the day, you must be able to explain this to the regulator and your stakeholders.” Those questions will also come from pension participants under the new Dutch pension contract. “We expect they will want much more information about their holdings than under the previous pension regime.”

Den Boon points to another aspect of complexity: the operational complexity quantitative models introduce in the organization. “Running a quantitative driven factor strategy requires high quality, validated and up-to-date data. The lack of data (for example on non-listed companies) can sometimes pose challenges.”

Liquidity constraints

Ho acknowledges another key challenge in factor credit investing: liquidity constraints, which fundamentally alter how these strategies are implemented.

Ho: “It is reasonable to expect that some of the highest-scoring bonds are out of reach. Or extremely expensive. If a bond scores well in our model, it may not be liquid enough to be included in a portfolio. To ensure that the model is implementable, we chop off the less liquid half of the universe. In reality, we can see every bond that is available at any given point in time. Our portfolio management system is fed by live, bond level liquidity data that is piped into our system, which can notify us when a very high-scoring bond becomes available in the marketplace again, and at what levels it can be bought in real time.”

Managers might want to skip a bond that is difficult to source or retain a bond that is hard to sell due to high spreads. Also, costs are a hugely important consideration. Managers can ignore trades within the model where the expected alpha doesn’t outweigh the expected transaction costs. “You do not want to do every transaction according to the model. If the transaction costs are too high, you don’t,” explains Ho.

The research is still too limited, making it hard to see how these models will play out in the long run

Sevinc Acar
Head of Public Markets at NN Group

Bekkers points out that this implies that managers still can have active decisions, within a model. Ho: “There’s human discretion based on whatever is popping up, based on your model. But it is always in the context of the scores. The scores and costs are driving the decision-making and the alpha. There is a small allowance for human discretion at the implementation phase, so as to be pragmatic.”

Complementary

Most participants agree that factor credit investing will not replace fundamental strategies, at least not soon. What quantitative models do already offer, however, is that fundamental managers can use them to enhance their current investment strategies.

Bekkers: “In the end, it can be complementary. If they enable, let us say, lower volatility on a total investment grade portfolio, then I think that could be beneficial. Investment grade bonds will play a far more important role in the protection portfolio, formerly the matching portfolio. Such a portfolio needs to deliver an additional return to the European swap rate but with limited volatility.”

Geene agrees: “What we see as the value-add of these models is how they can enhance and improve the quality of the investment process of fundamental managers and analysts. Because if a quantitative model looks at all the bonds and issuers out there, it can be helpful in screening. It can be helpful in analysing bonds and assessing their value versus what the analysts are saying, i.e. acting as a challenge. Furthermore, it can indicate if the portfolio of a fundamental manager is biased; tilted too much towards a factor like value, for instance. So, it can challenge every step of the process. We have seen some of the managers already doing this. Besides being used in this way for a fundamental manager, the excess return streams of pure quant managers are mostly uncorrelated to the excess returns of fundamental managers. Therefore, quant investing could potentially have additional complementariness by reducing tracking errors of the entire corporate credit allocation.”

Participants are intrigued by these complimentary possibilities. But how does this complementariness work in practice? Den Boon points to the fact that fundamental corporate bond managers already sometimes implicitly use a factor like value: aiming to buy a bond that screens inexpensive relative to the fundamentals. “Therefore, there is already some natural overlap between quantitative and fundamental investing. That might help the integration of both. One of the challenges would be how you weigh and combine your fundamental view with the quant view. We can definitely see added value in measuring a fundamentally constructed portfolio towards different factor exposures, to become aware of any (undesired) tilts.”

Can investors gain additional insights into how their portfolios are exposed to certain risks and return drivers, by overlaying factor scores on a fundamental portfolio? Ho is certain they can. “Look at the active-passive split: in equities, it’s something like 50-50. In fixed income, the vast majority of assets are sort of fundamentally actively managed. We think there is space for a systematic credit approach, not as a replacement for the active approach, but as an add-on or complement. They could use the daily four-factor scores, which have a real market breadth, in their fundamental process. See it as a ‘quantamental’ approach.”

Conclusion

The roundtable discussion made it clear that while factor investing in bonds holds significant promise, the journey from research to implementation is far from straightforward. While liquidity constraints, uncertainties around factor premiums, and the challenges of scaling up remain significant hurdles, the complementary nature of systematic and fundamental investing may offer a promising path forward.

Contact us

For more on how to incorporate systematic active fixed income into your portfolio.

Sabine van Solingen

Head of Benelux Institutional Client Group

E sabine_vansolingen@ssga.com

T +31 6 50 01 70 99

Participants:

Mark Geene PGGM

Rik Bekkers APG

Bart den Boon MN

Sevinc Acar NN Group

Arkady Ho State Street Global Advisors

Sabine van Solingen State Street Global Advisors

Mariska van der Westen FD Business

Tom Jessen moderator

Note:

https://www.ssga.com/ institutional/insights/ active-index-and-systematic-investing-in-fixed-income