03 January 2019
What is the role of investments in insurance companies' capital optimisation? Participants discuss in part one of this two-part roundtable from Insurance Asset Risk and Aon the opportunities and challenges for insurers' investments and how it fits into life and P&C insurers' business models.
Atanas Christev, head of investments, Direct Line
Christian Bird, group head of economic capital modelling, Validus
Emily Penn, head of capital efficiency, LV
Eric Paire, capital advisory, Aon
Gerard-Jan Van Berckel, head of delegated solutions for European insurers, Aon
Guillermo Donadini, chief investment officer for international operations, AIG
Kris McCullough, capital actuary, Chubb
Martin Pfutzner, head of strategy analytics, Aspen Insurance Group
Peter Helt, treasury manager, Pacific Life
Ronan McCaughey, deputy editor and commercial editor, InsuranceERM
Chaired by: Vincent Huck, editor, Insurance Asset Risk
Vincent Huck: From a chief investment officer’s perspective what are the challenges and opportunities for insurers’ investments at the moment and how does it contribute to adding value from a capital optimisation perspective?
Guillermo Donadini: For P&C insurers what is happening these days, compared to the 1980s and 1990s, is that real interest rates are really low or even negative. Through a very simple risk-free investment strategy insurers could make 4% RoE per annum in the past. Since the 2008 crisis we are in a negative interest rate arena, so the question is, how do we make money?
This is very interesting particularly for P&C underwriters because in life firms, investments are part of the product, and management has much more awareness about the investment side. However, in P&C the investment side acts more like an source to provide liquidity to pay out the claims.
So for me, it is about how we can add value although we are never going to catch up with the level of return on equity (RoE) from the 1980s and 1990s, because that would mean a massive amount of risk premium, a massive amount of capital and a lot of volatility that maybe we are not ready to show on our balance sheet. There comes the concept of ‘capital efficiency’.
Often capital efficiency is understood as minimising the ‘E’ to maximise the ‘R’ over ‘E’.
There is no clear understanding that, through diversification of investments and liability risks, you can increase equity, and consequently returns increase even more. Therefore the challenge is to make people aware of that, particularly the senior leadership, when accounting is not designed to show the numbers in that way.
Eric Paire: On the liability side the margins have been getting lower and lower as well, so there is an interesting convergence of assets and liabilities.
Secondly, I quite like your point about looking holistically at the assets and liabilities. Because you may want to accept that through diversification you increase the capital charge on the asset side, but you can decrease it on the liability side, especially at a time where it is relatively cheap to buy reinsurance in P&C.
Finally, you are right to illustrate a major cultural difference between life and P&C. Because where for life industry practitioners would think about both sides of the balance sheet in conjunction with one another, P&C tends to separate them.
Christian Bird: Some practitioners are trying to do that, like the so-called ‘total return reinsurers’ in Bermuda. I suppose even Warren Buffett is another example. In general most non-life companies are trying to keep the investment side safe, take little risk there and concentrate on the underwriting side.
However, there are opportunities where underwriting margins are low to make a meaningful return on the asset side of things, which can make the difference.
Martin Pfutzner: We are in a situation where 80% of the effort of a non-life company typically goes into generating underwriting returns and 20% goes into the investments. However, historically the money has probably come from 80% investments and 20% underwriting.
On a risk-adjusted basis, you will also generally find that assets tend to have much higher returns on capital than liabilities, especially if you go to the short-duration, higher-quality assets where the capital charge can be close to zero or sometimes even negative.
If you combine those two facts, I find it really surprising that most of the non-life insurance companies focus so heavily on the underwriting side.
Christian Bird: It is where people are most comfortable. Insurers are expecting to take some liability volatility: you lose $100m from a hurricane hitting Florida, that is understood. You lose $100m from your investment portfolio, that could spook the market. Is the world going to change, and are people going to be just as comfortable with asset risk, asset exposures, as liabilities? I do not know.
Guillermo Donadini: The question is, can the industry afford not to pay attention because it does not feel comfortable dealing with balance sheet volatility generated by investments to have proper return on equity?
I can understand that it is a legacy that we do not do that, or that our peers do not do that, or that the market would not understand it, and so on. However, can we reach RoEs of between 10% and 15% without the asset side?
Gerard-Jan Van Berckel: To the point on the ‘80/20 rule’ mentioned earlier by Martin, sometimes you see in the same company approx.150 underwriters and then three or four people on the investment side running billions of dollars.
If you think about creating, implementing and managing the investment portfolio, Solvency II, regulations, internal models, the reporting requirements, and everything that needs to be done by this relatively small investment team, it is a very challenging job.
Having a more holistic approach would help. And finding other capital efficient asset classes that can be incorporated in the investment portfolio to create better diversification within the portfolio can help in creating better outcomes
Emily Penn: Introducing other alternative asset classes is not easy, although it depends whether you are an internal model, or a standard formula firm.
If you are a standard formula firm, you are constrained in terms of the additional asset classes that you can add as the standard formula is not designed for ‘non-standard’ asset classes.
And if you are an internal model, firm you can’t be very opportunistic because it may take a year to get a new asset class through model approval. Although all these things are nice-to-haves, the actual reality of it is, it is not easy.
Guillermo Donadini: Also, the market in the last five years was expecting interest rates to recover, and everybody focuses on nominal rates, which is not what we need to focus on. We need to focus on real rates, because our liabilities follow inflation.
From the 1980s, our returns were basically pushed by lower-than-expected inflation on both the assets and liabilities sides. However, in a world where the amount of debt of the economy is at historically high ratios, the economy either default on them or inflate them. It is highly likely the solution would be to inflate them, so what happens then? Inflation will increase across the board.
We need to highlight those things, and again, the accounting and the internal models are not capturing that. So how do we raise those issues?
Peter Helt: Inflation is an interesting issue, considering how political it is these days. It is not just a standard central bank decision anymore. It is very much driven by political outcomes now. You cannot really put it into a 20-year model.
Christian Bird: We have not had to deal with it in P&C for some years now. Thinking back to actuarial exams, you are always taught about assets matching liabilities - maybe it is the life insurance bias in actuarial studies - but you cannot match inflation risk in P&C as easily.
Maybe you can match duration, but that is less of an issue in P&C insurance. You have a risk where your reserves could go out because of inflation, but your liabilities are all in fixed income, it is not going to keep pace with it, so what do we do? Do we put all our investments into equities as the inflation-matching instrument? I think not!
Atanas Christev: As a motor insurer, we face this dilemma because since the separation of Direct Line from RBS, the strategic message to investors has been that as a low-risk investor, the company will look to avoid more volatile assets classes such as equities. Instead we have chosen to back our Periodic Payment Order (PPO) liabilities with a mix of infrastructure loans and commercial property investments.
A large part of the property portfolio is in long-term leases which are linked to retail price inflation, but matching then becomes a macroeconomic argument based on the long-term correlation between rates, price inflation and wage inflation, rather than a duration-matching exercise. In our annual asset-liability management reviews, the PPO liabilities and the assets backing them are treated separately from all the other fixed-rate assets and liabilities.
Christian Bird: Although it is interesting that when PPOs first came about, there was an expectation that there might be a way to transfer the longevity risk onto a life company, but it was so expensive that people didn’t go through with it and kept all that risk themselves.
Emily Penn: In the main, the life insurers have not been willing to take on the ASHE inflation risk and the longevity risk is very different to the longevity risk they already have on their books. Our PPOs have remained in our general insurance book.
There are some specialist vehicles that are in the process of getting authorisation to take on these liabilities.
There has been talk for some time that there is a solution to this PPO problem, but we haven’t yet seen anything that works for us.
Atanas Christev: Having exposure to equities would help us. That is where we get to the capital model and how equities (or any other asset class) are represented in it or whether they get the correct treatment in the model. The only way for the investment team to get some understanding of this treatment is to feed in the model combinations of assets with controlled differences and to compare the capital numbers.
Martin Pfutzner: Are you involved in the calibration of the economic scenario generator (ESG)?
Atanas Christev: Yes. When new updates are released we are asked to give an opinion on whether the update makes economic common sense. However, usually the impact is so miniscule that it is difficult to justify a challenge. On the other hand, when changes in duration of the assets (e.g. changing a material amount of bonds from fixed to floating-rate) leave the capital requirement practically unchanged, one needs to question the usefulness of the model as an asset allocation tool.
Christian Bird: A good example is currency hedging. You hedge your currencies and then suddenly there is an earthquake in New Zealand and if you have got a big New Zealand exposure you can try and hedge it soon after. But what if exchange rates had moved just before that, that event may now be a lot more expensive than if the currency hadn’t moved. That is part of the nature of P&C - it is difficult to match things because the liabilities move too much.
Martin Pfutzner: I am guessing currency rates do not drive very much risk in the model?
Kris McCullough: In our model, currency risk is a big capital component. Our major currencies are USD, GBP and EUR. However, the reason it drives the capital is mostly because of the expectation not the risk. Because in an internal model you get to reduce for profit, so every single time our consultants release an ESG update that expectation changes over the forecast year. In the real world, that does make sense; it is not unexpected. But it means that the capital number moves around relatively a lot.
Martin Pfutzner: So you are incurring a big capital charge because at the mean you’re having a loss?
Kris McCullough: No, at the moment we are making a profit, but that profit can quite easily just reduce. That deduction suddenly goes away. Yes, sometimes we do make losses, but either way it is very volatile from one capital model run to the next; because of that expectation.
Martin Pfutzner: Do you adjust the ESG at all?
Kris McCullough: No, but that is exactly what we are considering.
Christian Bird: It is common just to zero-centre it.
Kris McCullough: If we zero it out, how do we justify using the rest of the ESG model? If we do not trust that part, why do we trust the rest of it?
Christian Bird: It depends how much you have to trust it. You do not have to bet your house on it. We use it quite a lot for investment optimisation. For example, we ran an optimiser to find the best portfolio to keep our modelled risk at a particular level and maximise our investment income as far as possible. We run it, and it says, ‘Here’s our portfolio’. Taking a sceptical approach we tried again, but with some of the model assumptions changed. This time the optimal portfolio looked quite different to the first one. This seems concerning - the model is very sensitive and small differences can lead to a very different answer.
But all is not lost. What we did was to construct a portfolio which we thought could work, and tested it under each of these assumption sets. It wasn’t the best in any situation, but it was pretty robust and did reasonably well in most situations particularly when compared to how the existing portfolio looked.
Can we guarantee that it is definitely the best portfolio out there? No. But we were just using our model to broadly take us a step forward, without 100% relying on it. We do not take ‘the number’ out of the model. We take a direction.
Guillermo Donadini: From an economic perspective, what is going to hurt you is volatility, not capital inefficiency.
What we do is we try to find the optimal model - if you can find such a thing - using real-world estimates. We do not use value at risk (Var) numbers. We do not use capital numbers to find the allocation. Then we say, ‘Okay, this is the efficient frontier’. You can use stochastic processes or whatever processes you want to forecast investment returns, volatility and correlations.
We came up with something that tries to clean all those model dependencies. Then we say, which of these two or three alternatives is the one that is more capital-efficient? Then we test it with the Var.
So, we just check the capital efficiency of the model that has been optimised using only mean variance optimisation, using real-world estimates. But many times, there is this tendency to say, ‘Let us use the internal model to have that optimisation’. That is like using Var numbers, which are basically end-of-the-world volatilities, which it is highly unlikely will happen in the future, to run your investment portfolio.
Martin Pfutzner: We actually do it the other way around. We start off with the internal model view on optimisation. You get your solution, you try to understand and make sense of it and then you use this understanding to start building actual strategies that you might want to implement. Then you start refining those strategies using qualitative considerations, and by scenario testing them against historical or hypothetical future events. The model optimisation is the starting point of the process, rather than the end.
Optimisation is almost like looking at a map and what the optimiser does is tell you, ‘Look in this general area of the map’ rather than ‘Look at this particular point’. The fine-tuning is the scenario analysis at the end, which is actually the important bit.
Emily Penn: I come from a slightly different perspective as we are now predominantly a life firm. It seems you are all quite frustrated by the constraints that the capital puts on you. We have many more constraints, which is the liability profile we have to match, and also constraints around matching adjustment (MA) and to a lesser extent volatility adjustment. For a large proportion of our portfolio we have limited flexibility in terms of the universe of assets available to meet our liabilities.
Vincent Huck: What are the constraints around MA portfolios and investment efficiencies?
Emily Penn: MA portfolios have very strict rules around what assets you can invest in. They have to be very fixed in terms of the nature of the cash flows; you cannot actively trade your portfolio; there can be no callable features in there and all assets require pre-approval from the regulatory.
There are a number of investments that we are looking at currently, but in order to be able to put them in our MA portfolio we have to go through approval with the Prudential Regulation Authority (PRA). There is one particular transaction we have been working on for well over a year. We have not even got to the stage of actually being able to submit the application to the PRA, because we are still having the conversations around the pre-application process.
Eric Paire: Playing Devil’s Advocate here, do you think you could have a better portfolio just forgetting about the MA and getting higher capital charge, but having more freedom to invest?
Emily Penn: We do not have that luxury of an infinite amount of capital to be able to do that. We need the MA to manage the volatility in our balance sheet, and to manage the absolute capital we have.
We have another portfolio that we cannot get MA on because of the nature of the liabilities. We have historically invested that portfolio predominantly in gilts because we cannot afford to take the risk of significant credit investments. We have now reinsured this book on an asset and liability basis. The reinsurer is able to earn a liquidity premium and we benefited from that through the pricing of the transaction. This was a good example of our investment and capital teams working closely together.
Guillermo Donadini: This is because the reinsurers can do something that you cannot do? What is the efficiency?
Emily Penn: The reason why we could not get this with MA is because there is optionality in there and it is with-profits. The reinsurance arrangement was quite innovative in terms of its structure - we reinsured it on a non-profit basis, and we have retained the optionality risk while the reinsurer takes a large proportion of the longevity and all the asset risk. They are reinsuring it on a non-profit, fixed cash flow basis. Therefore they can get MA or earn an illiquidity premium and pass that back to us with pricing.
Part II of this roundtable will be published on 10 January.