By Hamlin Lovell, NordicInvestor
Jyske Bank’s independent asset manager, Jyske Capital, is the first manager we have seen applying quantitative style factors as part of its process for selecting Danish mortgage bonds. It is also unleveraged.
NordicInvestor interviewed Head of Mortgage Bonds, Allan Willy Larsen (pictured above), who leads a team of four other senior portfolio managers, one quantitative analyst and two backup quants.
Callables
Some Danish mortgage strategies only trade callables while others avoid them altogether. Jyske varies the mix between bullets and callables. The Jyske strategy can invest in bullets without prepayment risk, up to 5 year, and callables, with prepayment risk, up to 30 years.
“The allocation to callables can range between 35% and 75%, and was 60% as of April 2024, which in practice is probably as high as it will go since we would not want to go to the outer limits of the range,” says Larsen. (A separate 100% callables strategy is managed for pension funds).
The reference benchmark is 20% government bonds and 80% mortgages, which is split 50/50 between bullets and callables.
No duration bets
The benchmark is also constant maturity, so when rates rise, its duration increases and often changes the share and composition of callables. Benchmark duration is around five years, and both the benchmark and the fund rebalance to this level every month. There is not much portfolio turnover in fact.
The outperformance target is relatively modest, because the strategy cannot take interest rate bets. Its main tools for making active bets are varying the allocation between callables, and bullets, and issue selection. “The strategy first determines the proportions per segment, and then selects the ISINs, i.e., securities within each segment,” explains Larsen.
The outperformance target of 0.50% has been exceeded: over the past ten years the fund has beaten the benchmark reference index by 0.79% (in common with many funds in Denmark it also has an official index of government bonds, but clearly the reference benchmark of 80% mortgages is more relevant).
Quantitative style factors and discretionary judgment
Analysing mortgages naturally involves quantitative analysis and even managers who describe themselves as “discretionary fundamental” will have to use a considerable amount of quantitative analysis.
Jyske is distinguished by how its process applies two specific quantitative factors, carry and value, to select bonds. Drawing inspiration from corporate bonds and emerging market bonds, Jyske started applying the factors in 2011 and most recently reviewed the approach in 2021.
The calculation of value and carry for mortgage bonds involves special analysis. Value needs to consider option adjusted spreads, and also adjusts for yield, volatility, and prepayment forecast. Carry needs to consider simulation analysis that also takes into account the convexity of the return profile.
There are also quant signals used top down to pick attractive segments, and bottom up to select issues within segments.
(A 31 page, 8,000 word, White Paper available on request from Jyske provides more detail on the quantitative thinking behind these factors – as well as the discretionary parts of the process).
Discretionary and quantitative
Jyske are of the opinion that the market is not inefficient enough for a pure factor approach to work, but intuitively sensible quantitative signals provide some signals and also lay the foundation for building more qualitative analysis on top.
The split between quantitative and discretionary analysis averages 50/50, but can sometimes go to 60% quant, 40% discretionary. “Issuance, buybacks and momentum can be harder to apply quantitative methods to,” says Larsen.
Interestingly, both the quantitative and discretionary parts are both top down and bottom up.
“The discretionary analysis involves a combination of model signals, including proprietary prepayment models, complemented by real world market experience of when homeowners prepay and buy back bonds,” says Larsen.
Performance attribution
The performance drivers depend on the framework used for performance attribution. “Quantitative analysis of performance attribution has decomposed alpha into mainly the value and carry factors, while more traditional performance attribution would identify industry allocation and security selection,” says Larsen.
Not only residential mortgages
The portfolio also needs to diversify amongst at least four of the five mortgage banks, and is expected to have 80% in them. Each mortgage bank issues from different cover pools that also vary in their composition.
The strategy will be exposed to residential, commercial, industry, agriculture, solar and wind farms across the bonds issued from the pools, so a Danish mortgage bond portfolio not only has exposure to many different types of mortgages but also to many different borrower segments.
Jyske can drill down into data on each mortgage pool, including ISIN codes, splits between private, corporate and commercial loans, sizes of loans, and cashflows due for the next quarter – and in some cases, carbon data on the individual bond issues.
Defining “Green bonds”
The number of green bonds in any market partly depends on definitions and labelling.
Only 3.5% of Danish mortgage bond issuance meets the strict definition of Green bonds labelled as such. A broader way to proxy “potential” green finance within the broader pool of mortgage bonds is to use the Energy Efficient Mortgages (EEM) definition, which are loans securitized on buildings that meet the sustainability criteria for a green bond. The associated bonds cannot be formally classified as green bonds but because of the match funding system their individual share of EEM loans can be counted as “quasi green finance”. The percentage is an additional 19.2% of issuance taking the total share of sustainable finance to 22.7% in the Danish mortgage market.
Whichever definitions are used and preferred, Jyske Capital is currently the only Danish mortgage bond manager that actively use data on both green bonds and EEM loans, broken down for every ISIN code or capital center, and integrate this data to their portfolio construction.
However, the universe of purely green mortgage bonds is not yet well diversified enough to build a portfolio that would meet Jyske’s criteria on liquidity, duration and non-callable features. “For instance, the labelled “green bonds” are currently only floating rate notes up to 5 years,” says Larsen. Including the EEM loans expands the “quasi green” opportunity set into the 30-year callable segment.
Measuring and estimating carbon emissions
One mortgage bank, Jyske Realkredit, has also released carbon data on every ISIN code. Larsen calculates that, “about 8% of loans (including agriculture) emit 57% of carbon emissions, but this is partly based on estimates”.
The portfolio’s C02 emissions have come down by 65% between September 2021 and December 2023. This reflects a combination of factors: changing allocations between issuers with less carbon emission from agriculture, changing adaptation of renewables in energy consumption, and finally improvements in data quality.
The carbon footprint has dropped considerably. Since 2019, the carbon footprint has dropped from 3.5 tonnes to 1.4 tonnes (per million DKK invested) on average. “But there is a wide spectrum. Certain ISIN codes could have as much as 38% “quasi green” finance, with carbon dioxide emissions of only 0.37 tonnes per million DKK invested,” says Larsen.
Renewables growing
A significant driver of lower carbon is simply renewable power generation. “Data on carbon emissions from the energy mix in heating is falling, and that percolates into carbon emissions data from issuers,” says Larsen. “Growing use of renewables in the energy mix used for electricity generation explains part of the drop. There is less coal, natural gas, oil and other fossil fuels being used, and more renewables,” says Larsen.
Improving carbon data quality
Part of the reported drop also reflects better data and Jyske has played an important role in raising standards throughout the industry. Jyske started issuer engagement in 2021. The other mortgage banks are listening, improving and watching the good example set by Jyske. “The industry association, Finance Denmark, has also developed a better carbon dioxide model and introduced standards,” says Larsen, who admits that there is more room for improvement in this ongoing process: “some of the data is still not up to date. It can be based on an old energy split between fossil fuels and other sorts of fuels. This is because when the data did not exist, it was simply estimated based on common assumptions”.
However, all of the data still involves some assumptions about averaging across properties, rather than digging up data for each house. “Currently the level of granularity stops with bond series level mortgage bond data and does not drill into ISINs (outside Jyske mortgages). It is not possible to drill into individual properties, though banks may have (some of) this data. Denmark Statistics can provide data on each sector and household’s carbon dioxide emissions,” points out Larsen. “Energy Performance Certificates (EPCs) are valid for 7 years so some of them may be out of date. The banks however can estimate an updated shadow EPC figure for each house.,” he adds.
This should improve in future. “Data from Statistics Denmark can be helpful for a sense check. As part of our active ownership engagement process, we are asking banks to update old or obsolete data,” says Larsen.
For instance, Jyske can triangulate between mortgage issuer data on outstanding loans, and Denmark Statistics data, to estimate C02e per million LTV. “Sometimes it matches and sometimes it does not. When there is a mismatch we engage with the issuer,” says Larsen.
Jyske are signatories of the Net Zero Asset Manager initiative and mortgage bond investments form part of its interim C02 reduction targets for 2030. Since many other asset managers and asset owners in the Nordics are also NZAM signatories, they should follow the trail being blazed by Jyske.
Wider ESG
Thus the ‘G’ in ESG, governance, is being pursued through engagement with issuers to improve data quality. The ‘S’ in ESG is also satisfied with data on social housing.
SFDR
“All funds report under SFDR 8 and the Luxembourg SICAV is also in the process of obtaining article 8,” Larsen concludes.
Yield pickup versus Sweden
Beyond the distinctive quant process and improving ESG data, there is a more simple reason why Swedish investors are looking south for bond investments. Jyske Capital’s strategy has been backed by a leading Swedish asset manager, and one attraction is potential yield pickup – though there are different opinions on how to measure it. Comparing yields between Swedish and Danish bonds is not a simple exercise.
“Sweden has no callable mortgages, so there is no exact comparison for Danish callables. An Option Adjusted Spread (OAS) could in theory be used but is hard to calculate. ARM (Adjustable Rate Mortgages) can be compared, and they trade at nearly the same levels.” says Larsen
The largest yield pickup is apparent when comparing Danish mortgages to Swedish government bonds, according to Jyske’s data. The mortgages pay more and their triple A credit rating is also higher than the double A or single A rating on the corporate bonds. The yield hedge to SEK can be 5% on a hedged basis. This benefits from a foreign currency hedging premium, based on 1 month FX forward yields, which in April 2024 annualised at 0.33% or 33 basis points, on top of the Danish Krona yield.
Beyond that, Jyske are of the opinion that Danish mortgages have less credit risk because leverage is lower, affordability criteria are stricter, and more loans are fixed rate whereas all are floating rate in Sweden.
Danish mortgages provide some yield pickup and may now provide an adequate return without using any leverage. The market also has limited potential for alpha generation through quantitative and discretionary analysis. Meanwhile the increasingly better quality carbon data reporting, and declining carbon footprints, should also become an important input into asset owners’ and asset manager’s own ESG ambitions and targets.
