Dutch asset manager Robeco is among a handful of factor-investing specialists striving to put their systematic techniques to work in credit. The firm’s quant team has identified four characteristics of corporate bonds – ‘factors’ in industry speak – which it believes deliver a premium to investors. They are borrowed from long-established strategies in equities that tilt portfolios towards value, momentum, low risk or smaller-sized stocks. But in bonds, when the firm tries to put on trades to tilt towards these factors, it struggles to do so about 20% of the time.

That is even after chopping down the whole universe of corporate bonds into a subset of the most liquid, says Patrick Houweling, quantitative researcher and portfolio manager at Robeco Asset Management. Without doing so, the figure would be well below 50%.

“Smaller bonds and older bonds tend to be less liquid and trade less often. Completely ignoring such bonds would shrink our investment universe substantially, resulting in sub-optimal factor exposures. At every point in time, there is a trade-off between getting exposure to the factors on the one hand, but also being able to execute the trade in the first place, and execute it cost efficiently,” Houweling says.

Quant investors have seized on credit as the most promising area of fixed-income markets for the application of factor-based strategies because, like equities, the universe of available securities is large. But many practitioners echo Robeco’s frustrations about liquidity.

Common characteristics

The standard approach to factor investing in equities identifies cross-sections of stocks that share certain characteristics – such as size, value and momentum – that explain their returns, and goes long on the top quintile while shorting the bottom. This simply “isn’t implementable” in fixed income, says one researcher and portfolio manager in a quantitative hedge fund’s global alternative premia group.

“It used to be that investment banks were making markets in corporate bonds and holding a lot of inventory but that’s really dried up since the financial crisis with the Dodd-Frank Act,” says Shane Shepherd, a director at investment manager Research Affiliates, one of the pioneers of smart beta investing in equities.

“Many of these strategies require a lot of rebalancing. That works well in equities. But it’s really difficult in bonds. Something that requires 100% turnover [a year] is really difficult,” Shepherd adds.

Etienne Vincent, head of quantitative management at BNP Paribas Asset Management, describes how efforts to put theory into practice often encounter real-world obstacles. “You’ll be hitting bid/offers that are so wide it will kill your alpha,” he says. “You might want to buy Coca-Cola debt but not be able to find it. Or you might find it at a price that is too wide.”

Such challenges help explain the sector’s slow pace of growth. At the time of writing, only 7% of the 963 ‘smart beta’ exchange-traded funds are fixed-income funds, according to data from ETF.com, and only half of those are corporate bond ETFs. Combined assets under management in credit factor investing ETFs are $4.6 billion.

Academic work on risk premia in fixed income goes back to the early 1970s, but the first credit risk premia strategies became available only in the past few years.

Research continues, though, and is gathering speed. The number of published papers jumped fivefold last year compared with 2010 and practitioners have built a solid case that a number of risk premia established in equities translate to fixed income. AQR, for example, has published papers identifying four styles that work in credit. Robeco adds a further two (see box: Lost in translation?).

It’s when it comes to the practicalities of building investment strategies to harvest those premia that complications stack up.

In academic research, to give a simple example, the profitability of trading strategies is typically analysed by looking at long/short portfolios rather than returns just from the long side. But shorting corporate bonds is beset with operational difficulties and high transaction costs.

Some factors are more challenged than others. The size factor, for example, faces difficulty in that smaller firms, which are seldom included in corporate bond indexes, are typically less liquid. Conversely, low-risk strategies tend to favour larger companies that are more liquid.

Short-term versions of the momentum factor change signal quickly and therefore have higher turnover rates.

In New York, WisdomTree Asset Management has concluded that strategies based on the momentum and size factors are too difficult to implement because of the liquidity challenges. WisdomTree found more potential in the momentum factor in high yield compared with investment grade, but the boost to returns came at the cost of sizable turnover. Meanwhile, the manager’s tests on the size factor have shown less potential in both investment grade and high yield.

“We are very much in agreement with the potential applicability of factor investing in the corporate bond space, particularly the quality and value factors. We see less value in momentum and size,” says Rick Harper, head of fixed income and currency at WisdomTree.

High turnover is the reason many managers have scrapped short-term momentum as a factor in corporates – instead focusing on longer-term momentum, based on six- to 12-month historical returns, to reach a strategy with much less frequent rebalancing.

“One of the best signals for credit would be a one-day reversal,” says one researcher and portfolio manager in a quantitative hedge fund. “But that’s a super-high turnover strategy. And that gets destroyed once you account for transaction costs. That one is not implementable.”

Robeco and Research Affiliates both think momentum calculated on a one-month basis cannot be profitably traded in credit. German asset manager Deka Investment also found that the premium from ultra-short-term momentum is cannibalised by turnover.

Demir Bektić, a fund manager in the firm’s quantitative portfolio management division, says: “You can’t trade such a portfolio on a monthly basis in a corporate bond market. That would be insane turnover of about 1,000% a year. After transaction costs, there would be nothing left.”

Meanwhile, to get round illiquidity and turnover costs, asset managers are establishing their own rules around rebalancing frequency, weighting schemes and – especially – factor definition, Bektić says.

We are very much in agreement with the potential applicability of factor investing in the corporate bond space

Rick Harper, WisdomTree

This occurs in equities where, for example, there are multiple definitions of value and momentum, but it means credit portfolios based on the same factor may perform completely differently. “Everyone is going in their own direction to some extent, which I think is a little unfortunate because it makes it difficult for investors to sort through everything out there,” Research Affiliates’ Shepherd says.

WisdomTree manages liquidity by removing the most illiquid bonds from its filters and imposing issue-size minimums on the bonds it will trade. Within high yield, the fund screens out the bottom 5% of the least liquid securities determined by size and seasoning, for example.

BlackRock recently launched two fixed-income factor-based ETFs in the credit space that aim to screen for value and tilt towards quality: the iShares Edge Investment Grade Enhanced Bond ETF and iShares Edge High Yield Defensive Bond ETF.

The firm says it carries out screening during the portfolio management process for its smart beta indexes to identify liquidity conditions while minimising tracking error to the benchmark. As a result, its smart beta indexes are subsets of much broader indexes and are not any less investable, argues Steve Laipply, head of fixed income at iShares.

“When you are creating rules-based outcomes you have to take care you’re not creating rules that would result in excessive turnover,” he says.

Yielding results

Many practitioners remain optimistic that factor investing in credit can take off, in spite of these challenges.

BNP Paribas’s Vincent describes fixed-income factor investing as the “new frontier” in smart beta because of its potential to provide additional income to yield-hungry investors without taking on further liquidity, duration or credit risk.

“If you believe in behavioural finance and you believe the premia are there for behavioural reasons, you can pretty much apply the ideas [found in equities] to fixed income – even if you don’t have a very large database, even if you don’t have much liquidity,” he says.

BNP Paribas navigates the liquidity problem in corporates by using a potential-buy and potential-sell list with limit prices instead of defining a fixed target portfolio, as is usual in equity strategies. In practice, this would mean that if Coca-Cola is not offered or is too high, the fund could buy Pepsi instead.

Deka Investment takes a similar approach, constructing portfolios with a large number of bonds that do not need to take any single large positions. In the case that a single bond is not tradeable at all, the firm trades the “nearest neighbour” bond in terms of similar factor exposure.

Meanwhile, at Robeco, quants are busy collating information from various platforms – including trading platforms, where brokers make a market in the more liquid bonds, and pre-trade transparency platforms, supplying information on brokers’ axes and inventory.

“Whenever we need to trade for a factor portfolio, we select a mixture of liquid bonds, which will ‘always’ be available, and illiquid bonds, which are only temporarily available at the time of trading,” Houweling says.

In this way, the firm can strike the right balance between exposure to the factors, and optimal trade execution.