The modelling challenges in absolute return strategies

Absolute return strategies offer promising risk-adjusted returns and capital efficiencies, but do you need an internal model to gain the benefit and what other complexities are created? Experts discuss in part two of this InsuranceERM/Insurance Asset Risk and Standard Life Investments roundtable

Hugo Coelho: To what extent does covering these investments in your internal model make them more attractive relative to the standard formula?

Emily Penn: In the worst case, under the standard formula, investments in absolute return funds would be treated quite harshly - as type 2 equities – so that is why we want to look at them in a capital-protected format.

Craig Turnbull: Capital protection is certainly one route to improving the capital treatment, but with the ability to look through to the underlying exposures of the fund, absolute return funds can look very capital efficient without a capital-protected format.

Emily Penn: That is true. One of the challenges of absolute return funds is that every fund is quite unique. So if you are building a model, you are probably building a model for one fund, and then if you want to make subsequent investments it would take additional resources. Equally, in terms of communication with stakeholders, you have got to do quite a lot of due diligence for each individual fund.

Craig Turnbull: Yes, some of those modelling and stakeholder communication issues are most significant for with-profits business. In those cases, you may need to project the fund over 30 or 40 years for the purposes of liability valuation. In that context, some sort of dynamic, long-term fund model would be required, which says: 'This is how the fund would behave over that very long time horizon'. That is a potentially challenging piece of modelling for some absolute return funds.

"For shareholder funds, non-profit life business and non-life business, the modelling challenges are less significant." Craig Turnbull, SLI

But for an SCR calculation, it is much simpler because of the one-year horizon. That does not require a dynamic, long-term fund model, it only requires to look-through to the underlying assets and apply the stresses. So for shareholder funds, non-profit life business and non-life business, the modelling challenges are less significant. And even in the with-profit case, the capital benefits can be so compelling as to justify the additional modelling and communication efforts that may be necessary.

Emily Penn: We would have to be comfortable that we could actually get that information, and not all funds are able to provide that.

Craig Turnbull: Yes, we believe it will be a crucial element of the insurance asset management servicing proposition that they provide complete, granular, timely asset data reporting to support insurers' requirements in this respect.

Emily Penn: From an internal model perspective, we probably want to look at it more like, 'You may at this time have 20% in equities, but the fund could move to be 50% in equities,' and we have to allow for that in the model.

Craig Turnbull: We think it will be important to understand how the particular fund's risk profile is constructed and what risk controls are in place. For example, is it a fund that, like in your example, has carte blanche to increase its equities exposure to 50% next year because of an investment view? Or does it have a constrained risk process that places clear quantitative limits on the risk appetite of the fund? If it is the latter, it can be modelled on that basis.

Gareth Collard: That would be a practical way to model it, but I am not sure whether that is convincing in terms of the regulator's view. With path dependency and frictional costs, in extreme scenarios it could be quite a complicated model.

Bruce Porteous: But, in a sense, these comments are not really specific to absolute return and apply to all sorts of insurer risks that are tricky to model. Catastrophe risks in general insurance and complex investment guarantee risks in life insurance, for example. Moreover, as modelling techniques have improved over the years, insurers have learned how to effectively measure and manage what used to be considered difficult risks.

Gareth Collard: True, but for most of us on the life insurance side – particularly annuity writers – the asset portfolio is static, so it's relatively straightforward to model; we do not have to make assumptions about trading in or out.

Modelling a dynamic fund

Eddie Gimlette: It is very much in the nature of an absolute return fund that we would expect it to evolve as our future expectations about liquidity, correlations and returns change.

Gareth Collard: I assume that you have to be transparent with your clients about those changes?

Eddie Gimlette: Absolutely. We report on all those changes and why we have done it. We report the profit and loss of each strategy, typically each quarter.

Gareth Collard: A firm with a substantial amount invested in the fund would have to change its internal model in response to those changes, would it not?

Craig Turnbull: It depends to a degree on how the internal model is constructed. If the internal model already includes the underlying risk exposures of the fund, the fund's line-by-line asset positions can simply be fed into the internal model. So all you are changing is what you put into the sausage machine, if you like, rather than having to fundamentally change the machine itself.

Gareth Collard: So you model a universe bigger than the actual universe that you have?

Craig Turnbull: Not necessarily. The underlying exposures that the fund takes might not be especially esoteric types of exposure. So you might not require many new modelling components or new correlation assumptions. The materiality of some of the assumptions might increase or decrease, however. It really depends on the particular investment philosophy of the fund and the implications for the type of exposures that it therefore takes in its portfolio.

Emily Penn: Even if you take that approach, the core benefit still comes from the fact that you are diversified across individual asset classes. So how would we work with the regulator to convince them that that diversification is valid in a one-in-200-year event, and therefore that we can get a realistic capital charge?

Craig Turnbull: Yes, we certainly believe that a main source of capital efficiency in well-diversified absolute return funds arises from the ability in Solvency II to recognise the realistic diversification benefits that accrue from the breadth of strategies that the fund runs. In our experience, the regulatory dialogue would be similar to other internal model (IM) approval discussions – additional analysis such as sensitivity testing of results to key IM assumptions, stress testing of the underlying portfolio data independently of the IM, can help to generate confidence and transparency in the robustness of the results.

"Firms are having these conversations with supervisors and, actually, have been doing so for some time." Bruce Porteous, SLI

Bruce Porteous: The biggest investor in the Global Absolute Return Strategies (GARS) is a with-profits fund, so that firm has been talking to the supervisor for a while. Also, Standard Life's staff pension scheme runs these multi-asset strategies in a very large way, and obviously they feed into the internal model of the insurance company. So firms are having these conversations with supervisors and, actually, have been doing so for some time.

Emily Penn: Do you know how well-educated the regulators are on these type of investments?

Bruce Porteous: There is definitely work involved in understanding these funds, just as there is in understanding how equity release mortgages can be restructured to fit into the Matching Adjustment framework. In fact, equity release mortgages looks like a harder problem to me for supervisors. We are beginning to engage with supervisors to share our knowledge about these funds and, based on our experience, they are open to learn more.

Solvency II reporting

Chris Cundy: Solvency II reporting requires investors to look-through funds to the underlying investments. Is that a practical problem for absolute return funds?

Gareth Collard: How many positions would you have in a typical fund?

Eddie Gimlette: The number of strategies is about 25–30, but you could have 1,000 positions.

Gareth Collard: That is more positions than we have in our entire portfolio today. So that gives you a feel for the added complexity of the administration.

Craig Turnbull: Yes, that is an interesting point. To manage that, the insurance asset managers will need to invest significantly in their asset data reporting capabilities so as to efficiently support client needs in that area, especially in the new Solvency II world.

Victoria Gibson: We have found legal issues around who owns the data feeds, with some managers unwilling to pass on data to third parties, which is a problem when you use an external data provider.

Bruce Porteous: We agree and these issues are being resolved. Credit ratings is an example: can the asset manager pass a credit rating to the insurance company? If the insurance company does not have a licence, we are not allowed to. That problem will, we believe, get solved over the course of the next year or so, however. We are also hearing about asset managers losing insurance mandates because they were not able to provide the data.

Craig Turnbull: More generally, this could be viewed as an example of a wider trend: insurance asset management is an increasingly specialised and demanding discipline, both in terms of investment strategy and refined mandates and in the areas of client servicing and reporting.

Risk-adjusted returns

Emily Penn: In your experience, how do the risk-adjusted returns and the return per unit of capital of absolute return funds compare with a standard bond or equity type investment?

Craig Turnbull: Our quantitative look-through analysis suggests that, in general, well-diversified absolute return funds can offer a highly capital-efficient route to investment return enhancement.

Taking our analysis of GARS as an example, our look-through modelling produced a solvency capital ratio (SCR) of 15–23%. The fund has a performance objective of cash plus 500 basis points. Five-hundred basis points is somewhere in the ballpark of an equity risk premium, and a diversified equity portfolio will attract an equity SCR of 39–49%. So the comparative expected return per unit of capital is quite compelling in that example.

Similar comparisons can be made for other funds in the Standard Life Investment absolute return suite. For example, when ARGBS, the absolute return bond fund, is compared with conventional credit strategies, similarly persuasive expected return per unit of capital figures emerge. Those metrics suggest absolute return funds can be very capital-efficient vehicles for return enhancement.

Chris Cundy: If interest rates were higher, would absolute return funds still be attractive?

Emily Penn: Yes. I think it comes down to looking at the portfolio on a risk/return basis and if it fits into the portfolio on that strategy. It is not just dependent on where yields are; it is really about taking a long-term view of our portfolio and looking at a balanced approach going forward.

Part 1 of this roundtable can be read here.