Capital Market Assumptions

Transcription

1 Aon Hewitt Consulting Investment Consulting Practice Capital Market Assumptions As of December 31, 2014 Risk. Reinsurance. Human Resources.

2 Table of contents Modeling risks...3 Inflation...5 Government bonds...6 Inflation-linked government bonds...7 Investment grade corporate bonds...8 US high yield debt and emerging market debt...8 Equities...9 Private equity...10 Real estate...10 Hedge funds...11 Volatility...12 Correlations...13 Capital Market Assumptions methodology...14 Contacts Capital Market Assumptions

3 Modeling risks In our quarterly Capital Market Assumptions we publish returns and volatilities for a variety of asset classes. These assumptions lie at the heart of much of the advice that we give to our clients. For example, as well as helping inform our investment advice, they are also used when advising on assumptions used for actuarial valuations and funding advice. However, the headline figures that we publish only tell part of the story. When setting an investment strategy, it is also important to capture the full range of outcomes that can occur in order to properly assess the risk-reward characteristics of that strategy. A simple model of asset returns such as the so-called lognormal distribution has the benefit that it can be relatively easy to understand and the process of updating the model is relatively painless. While simple models such as this can be perfectly appropriate for tasks such as calculating an expected return on assets, they are less useful when trying to understand the full extent of risks associated with an investment strategy, a key part of an asset liability modeling exercise. For example, fitting the lognormal distribution to data covering the past 100 years would suggest that US equities could have been expected to lose as much as 35% only once in every 200 years. However, investors actually lost this amount in three separate calendar years in the past 100 years (1931, 1937 and 2008). Similarly, this model would suggest that annual losses of around 29% or worse would only occur twice in the past 100 years but investors suffered this fate six times. Other simple models also fail to pass certain criteria. One of the most wellknown statistical distributions is the Normal distribution. However this assumes that returns are symmetrical with the implication that you could earn a return worse than -100%. Unless an investor is using gearing/leverage, this is impossible as they cannot lose more money than they originally invested. While simplicity may be a desirable feature of a model, it can clearly also have its shortcomings. At best, all models are approximations of reality. However, some are better approximations than others. We think that there are some key features of markets which should be incorporated when assessing the appropriateness of an investment strategy as part of an asset or asset-liability modeling exercise. In the detailed modeling work we carry out for clients when we are advising on investment strategy, we use a highly sophisticated model which attempts to capture some of the important features of markets. If we did not allow for these features, we could misstate the risks of different asset classes and could recommend strategies that, while looking attractive in the model, did not represent real world opportunities. Using equities as an example, our models reflect the following characteristics: Although equities are expected to generate a positive return in the long run, there are many instances of negative returns. These can occur in individual years and also over relatively long periods. For example, US equities delivered a negative return in the first ten years of this millennium. Equities can go up and down, but although they can go up by more than 100% they cannot go down by more than 100%. Very large rises and falls occur more often than you might expect (in technical terms, the distribution of returns has fat tails ) but unfortunately for investors, the large falls occur more than the large rises (in technical terms, there is a negative skew to returns). Bad things happen more often than simple models predict. Markets do not always progress in a nice, smooth, manner, and they sometimes suddenly jump either up or, more frequently, down. In other words, the market is vulnerable to shocks which sometimes seem to come out of nowhere. Equities sometimes experience periods of high volatility and sometimes low volatility depending on the prevailing economic and financial market conditions. Their correlations with other asset classes do not remain stable. They change according to the type of economic environment we are experiencing. For example, in periods of heightened risk aversion, there is often a flight to quality and all risk seeking assets will tend to fall together. Ignoring this observable feature would overstate the benefits of diversification. Aon Hewitt 3

4 While capturing these features adds complexity to our model, we believe this additional complexity is worthwhile as it allows us to more accurately assess the risk-reward characteristics of an investment in equities and avoid unexpected risks. Investors can go forward with their eyes open rather than complacent to the risks to which they are exposed. The graph below shows the distribution of historic annual price returns on the US equity market over the past 100 years compared to a simple model (lognormal used as an example) and the model used by Aon Hewitt. These are backward looking and differ from our forward looking assumptions but have been shown in this way in order to compare the distribution of realized experience with the distribution imposed by these two models. History vs Statistical Distributions (models calibrated to experience) % -50% -40% -30% -20% -10% 0% 10% 20% 30% 40% 60% History Simple model (lognormal) Aon Hewitt model 60% 70% 80% 90% 100% As we can see, although not perfect, the distribution of returns using the Aon Hewitt model gives a much better match to the shape of the historical returns than the simple model. Also, the probabilities of poor (and good) returns are a closer match to history: Probability below Return Return History Lognormal Aon Hewitt model -35% 3.0% 0.5% 1.6% -20% 8.0% 7.2% 7.8% -5% 24.0% 28.7% 24.6% Probability above Return Return History Lognormal Aon Hewitt model 20% 26.0% 25.5% 25.0% 40% 3.0% 7.1% 3.9% 55% 0.0% 2.3% 0.6% Although we believe that the models we use are a useful representation of how assets can be expected to behave over time, one should never forget that at the end of the day they can never be more than an approximation of reality. Models should never be allowed to make decisions for you and professional oversight should always be incorporated. Hopefully this short piece has given you a flavor of some of the complexities involved in our stochastic modeling. If you have any questions about this, or any other aspect of the modeling, please direct any questions to the Economic Scenario Generator Team in the first instance: DG-AH-UK-EconomicScenariosTeam@aon.com 4 Capital Market Assumptions

5 Inflation USD GBP EUR CHF CAD JPY CPI Inflation (10yr assumption) 2.1% 1.9% 1.6% 1.1% 1.8% 1.6% RPI Inflation (10yr assumption) 3.0% Realized inflation has been falling on a global basis and near term inflation expectations have continued to decline. This is largely due to the fall in the oil price although movements in currencies have also impacted some individual markets. However, despite near term deflationary concerns, consensus expectations for longer term inflation remain relatively well grounded. CPI inflation expectations have been downgraded slightly but inflation is expected to be relatively close to central banks 2% target level over the next 10 years in the US, UK and Canada. Even in Europe, where near term inflation expectations are very low, inflation is expected to pick up in later years taking the 10 year inflation assumption to 1.6%. The Japanese authorities have undertaken huge amounts of monetary stimulus with an objective to generate growth and break the hold deflation has had over the economy, attaining a new higher inflation target of 2%. However, consensus remains unconvinced that the 2% inflation target will be achieved on a sustainable basis. Japanese inflation is expected to be around 1.6% per annum over the next 10 years in Japan. Though below the 2% target, this is well above the experience of the past decade. Aon Hewitt 5

6 Government bonds 10yr Annualized Nominal Return Assumptions USD GBP EUR CHF CAD JPY US 5yr 2.1% 1.9% 1.6% 1.1% 1.8% 1.6% 15yr 3.0% 2.9% 2.6% 2.0% 2.8% 2.6% UK 5yr 1.7% 1.6% 1.3% 0.7% 1.5% 1.3% 15yr 2.9% 2.8% 2.4% 1.9% 2.7% 2.4% Eurozone 5yr 1.1% 1.0% 0.6% 0.1% 0.9% 0.6% 15yr 2.4% 2.3% 1.9% 1.4% 2.2% 1.9% Switzerland 5yr 1.2% 1.1% 0.7% 0.2% 1.0% 0.7% 15yr 2.0% 1.9% 1.5% 1.0% 1.8% 1.5% Canada 5yr 1.8% 1.7% 1.4% 0.8% 1.6% 1.4% 15yr 3.0% 2.9% 2.5% 2.0% 2.8% 2.5% Japan 5yr 0.7% 0.5% 0.2% -0.3% 0.4% 0.2% 15yr 1.5% 1.4% 1.0% 0.5% 1.3% 1.0% We take French bonds to represent Eurozone bonds, as there is a reasonably liquid market in French inflation-linked bonds and we want to ensure consistency between the nominal and inflationlinked government bond returns. Our calculation of a weighted average Eurozone government bond yield leads to a figure which is slightly higher than the yield on French government bonds. Our analysis therefore supports the use of French bonds as a proxy for Eurozone bond portfolios, where these portfolios do not have a large exposure to the higher yielding periphery. After rising during 2013, government bond yields fell sharply over 2014 and ended the year well below start of year levels. In addition, yield curves flattened which implies that the market was pricing yields to rise by a smaller amount over time than previously. In other words, not only were yields priced to be lower for longer, they were priced to be lower forever. As a result of this, our return assumptions for government bonds are now much lower than at the start of the year in local currency terms, with assumptions for short duration bonds falling by more than long duration bonds. The short duration assumptions have fallen by more because the flattening in yield curves means that the average yield on these bonds over the projection period is a lot lower than at the start of the year. The larger downward moves in European bond yields have led to relatively large reductions in European fixed income bond assumptions over 2014 which are converging on the ever-low Japanese and Swiss assumptions. 6 Capital Market Assumptions

7 Inflation-linked government bonds 10yr Annualized Nominal Return Assumptions USD GBP EUR CHF CAD JPY US 5yr 2.7% 2.6% 2.2% 1.7% 2.5% 2.2% 10yr 2.7% 2.5% 2.2% 1.7% 2.4% 2.2% UK 5yr 2.1% 2.0% 1.6% 1.1% 1.9% 1.6% 15yr 1.7% 1.6% 1.2% 0.7% 1.5% 1.2% Eurozone 5yr 2.1% 1.9% 1.6% 1.1% 1.8% 1.6% 10 yr 2.1% 1.9% 1.6% 1.1% 1.8% 1.6% Canada 5yr yr 2.1% 2.0% 1.6% 1.1% 1.9% 1.6% We have taken French bonds to represent Eurozone bonds, partly because there is a reasonably liquid market in French inflation-linked bonds. Our analysis of nominal government bonds also suggests that French bonds are a reasonable proxy for Eurozone government bonds so we make the same assumption here for consistency. The bonds represented are linked to Eurozone inflation. We formulate return assumptions for 10 year US and Eurozone inflation-linked government bonds rather than 15 year bonds. This is because we think that the absence of inflation-linked bonds at the longest durations in these markets can lead to misleading 15 year bond return assumptions. We also no longer publish a 5 year duration Canadian inflation-linked government bond assumption due to the lack of short duration bonds in this market. A second factor influencing inflation-linked bond return assumptions is inflation expectations. In this respect, returns from UK index-linked gilts benefit in relative terms compared with the other markets by virtue of the fact that returns on these bonds are linked to UK RPI inflation. This has an impact because other regional inflation-linked bond returns are linked to CPI inflation and this is assumed to be much lower than UK RPI inflation. UK real yields remain well below the level of the other markets so if it were not for the difference between RPI and CPI inflation, the return assumption for UK index-linked gilts would be much lower compared with the other regions than shown in the table above. A similar story holds for inflation-linked as for nominal government bonds when, after the large rises seen in 2013, real yields fell back during Real yields are in negative territory in both Europe and the UK even at long durations. Low or negative real yields lead to low return assumptions for inflation-lined government bonds. Aon Hewitt 7

8 Investment grade corporate bonds 10yr Annualized Nominal Return Assumptions USD GBP EUR CHF CAD JPY US 5yr 2.9% 2.8% 2.4% 1.9% 2.7% 2.4% 10yr 4.2% 4.0% 3.7% 3.2% 3.9% 3.7% UK 5yr 2.6% 2.4% 2.1% 1.6% 2.3% 2.1% 10yr 3.4% 3.3% 2.9% 2.4% 3.2% 2.9% Eurozone 5yr 1.5% 1.4% 1.0% 0.5% 1.3% 1.0% 10yr 2.1% 1.9% 1.6% 1.1% 1.8% 1.6% Switzerland 5yr 1.4% 1.3% 0.9% 0.4% 1.2% 0.9% 10yr 1.7% 1.5% 1.2% 0.7% 1.4% 1.2% Canada 5yr 2.9% 2.8% 2.4% 1.9% 2.7% 2.4% 10yr 4.0% 3.8% 3.5% 2.9% 3.7% 3.5% Japan 5yr 0.7% 0.6% 0.3% -0.2% 0.5% 0.3% 10yr 1.0% 0.9% 0.5% 0.0% 0.8% 0.5% Corporate bond returns depend on both a government yield component and a credit spread component but also take account of losses arising from defaults and bonds being downgraded. Corporate bond return assumptions are noticeably lower than at the start of As with all fixed income assets, the main driver of this has been the large declines in government bond yields although movements in credit spreads have also had an impact on some assumptions. US high yield debt and emerging market debt After having fallen to very low level, the high yield debt credit spread increased by a large 1.5% during the second half of This was partly due to concerns over the impact of the falling oil price on credit risk within the US high yield market the energy sector represents close to 15% of this market and experienced a significant sell-off so this had a material impact on the overall high yield credit spread. This higher credit spread benefits future return expectations for high yield debt although low government yields continue to be a drag on long term prospects. Our high yield debt assumption incorporates an assumption that defaults are, on average, in line with the long term historic average over the course of the next 10 years. This is much higher than recent experience. We therefore have not altered our long term default assumptions in light of the oil price declines as we already encompassed an expectation that defaults would increase over time. Taking these points together, US high yield debt is now assumed to return 4.5% a year over the next 10 years. When the US first mentioned that it may be ready to consider tapering its quantitative easing program, emerging market assets were among the worst affected, suffering very poor performance. This raised the credit spread that could be earned on US dollar denominated debt and a credit premium opened up for emerging market debt relative to a comparably rated developed market corporate bond. With recent widening in US high yield debt credit spreads, this gap has narrowed but our emerging market debt assumption remains slightly higher than our high yield debt return assumption. As with high yield debt, lower Treasury yields have put downward pressure on the US dollar denominated emerging market debt assumption, which stands at 4.8% a year as at 31 December. 8 Capital Market Assumptions

9 Equities 10yr Annualized Nominal Return Assumptions USD GBP EUR CHF CAD JPY US 6.5% 6.3% 6.0% 5.4% 6.2% 6.0% UK 7.3% 7.1% 6.8% 6.2% 7.0% 6.8% Europe ex UK 6.8% 6.7% 6.3% 5.8% 6.6% 6.3% Switzerland 6.4% 6.3% 5.9% 5.4% 6.2% 5.9% Canada 6.9% 6.8% 6.4% 5.9% 6.7% 6.4% Japan 6.3% 6.2% 5.8% 5.3% 6.1% 5.8% Emerging Markets 8.1% 7.9% 7.6% 7.0% 7.8% 7.6% Our equity return assumptions are driven by market valuations, earnings growth expectations and assumed payouts to investors. The price you pay is one of the single biggest drivers of returns, even over the long term. Looking back over recent experience, strong equity market performance has been driven more by increasing valuations than increasing profits. Therefore, as markets have become more expensive, our equity return assumptions have consequently fallen. As inflation assumptions have also been declining, our nominal return assumptions have also declined by more than our real return assumptions. UK equities have a higher return assumption than the other developed markets. The main reason for this is that this equity market is currently the cheapest of the developed markets in valuation terms. As at 31 st December, UK equities were trading on a multiple of around 14 times our 2015 earnings assumption. In contrast, US equities were valued at over 17 times our 2015 earnings assumption. Investors in UK equities are therefore paying less for expected future earnings, which raises the return assumption for the UK market relative to elsewhere. Emerging market equities have significantly underperformed developed markets over the past few years for a variety of reasons including a reduction in emerging market growth prospects and more recently, concerns over the impact of less accommodative US monetary policy. This cheapened the relative valuation of emerging market equities and, even after allowing for a reduction in emerging market growth prospects, has led to an increased return premium opening up for emerging market equities over developed markets. The earnings growth component of our equity return assumptions comprises both near term and longer term elements. While our Capital Market Assumptions process typically involves using consensus inputs, for some time we have believed that the consensus of analysts forecasts has been unrealistically optimistic regarding near term earnings growth prospects. Unlike analysts, against a backdrop of weak global growth we do not expect company profit margins to increase from their already elevated levels. For this reason, we have developed our own in-house corporate earnings paths which have led to lower growth assumptions than forecast by the consensus. For the major developed markets, we assume low single digit earnings growth over the next few years. Not being influenced by short-term market sentiment, our near term earnings growth assumptions have been relatively stable overall, in contrast to consensus expectations which have varied far more. In the long term, we assume that companies earnings growth is related to GDP growth. Crucially, we do not assume a oneto-one relationship between a country s growth rate and the long term earnings growth potential of companies listed on the stock market within that country. We do this because many companies are international in nature and derive earnings from regions outside of where they have a stock market listing. An implication is that European company earnings have only about a 50% direct exposure to developments in the Eurozone and similarly, investors in non-european equity markets should not consider themselves insulated from events there either. It is also notable that emerging markets are an important driver of profits earned in the developed world. Aon Hewitt 9

10 Private equity We assume that global private equity will return 8.8% per annum over the next 10 years in US dollar terms. The assumption represents a diversified private equity portfolio with allocations to leveraged buyouts (LBOs), venture capital, mezzanine and distressed investments. Return expectations for these different strategies depend on different market factors. For example, distressed investments are influenced by the outlook for high yield debt. Similarly, LBO returns are influenced by the outlook for equity markets as well as the cost of the debt used to finance these LBOs. The current low interest rate environment is therefore beneficial for LBO investors. Notwithstanding this, whereas in the past leverage has been a big driver of private equity returns, particularly for LBOs, in future the ability of managers to add value through operational improvements will become more important. On our analysis, the median private equity fund manager has historically performed in line with the median public equity manager, but high performing private equity managers have performed significantly better. Our assumption incorporates the level of manager skill ( alpha ) associated with such a high performing manager. This contrasts with our other equity return assumptions where no manager alpha is assumed. Real estate 10yr Annualized Nominal Return Assumptions USD GBP EUR CHF CAD JPY US 6.8% 6.6% 6.3% 5.7% 6.5% 6.3% UK 6.4% 6.3% 5.9% 5.4% 6.2% 5.9% Europe ex UK 6.4% 6.3% 5.9% 5.4% 6.2% 5.9% Canadian 5.6% 5.4% 5.1% 4.5% 5.3% 5.1% As with all other asset classes, more expensive valuations have put downward pressure on our real estate assumptions compared with 12 months ago. Similarly, lower inflation assumptions also lead to nominal return assumptions falling by a greater margin than real return assumptions. The US currently offers the highest real estate return assumption of all of the markets covered in our Capital Market Assumptions. Although capital values have risen, this market continues to benefit from a relatively healthy rental yield and a relatively robust outlook for rental growth. In Europe, signs of stabilization have provided some support to real estate capital values and they have been rising in parts of the region. This has put some downward pressure on rental yields although they have been more stable than in the other major markets. This region continues to offer a weak rental growth outlook because, unlike equity markets which benefit from their international exposure, real estate is much more closely tied to the fortunes of the region in question. A weak rental outlook acts as a drag on the return prospects for European real estate. Our assumptions here are in respect of a large fund which is capable of investing directly in real estate. The assumptions relate to the broad real estate market in each region rather than any particular market segment. Our analysis allows for the fact that real estate is an illiquid asset class and revaluations can be infrequent, leading to lags in valuations compared with trends in underlying market value. These assumptions do not include any allowance for active management alpha but do include an allowance for the unavoidable costs associated with investing in a real estate portfolio. These include real estate management costs, trading costs and investment management expenses. 10 Capital Market Assumptions

11 Hedge funds Our fund of hedge funds return assumption is 4.8% a year in US dollar terms. We formulate this by combining the return assumptions for a number of representative hedge fund strategies. As with private equity, this assumption includes allowances for manager skill and related fees (including the extra layer of fees at the fund of funds level), but unlike private equity, this is for the average fund of funds in the hedge fund universe rather than for a high performing manager. Our analysis allows for the fact that hedge fund managers have been unable to deliver the high levels of alpha that they did in the more distant past and that alpha generation is likely to remain challenging moving forwards. The individual hedge fund strategies we model as components of our fund of hedge funds assumption are equity long/ short, equity market neutral, fixed income arbitrage, event driven, distressed debt, global macro and managed futures. Our modeling of these strategies includes an analysis of the underlying building blocks of these strategies. For example, we take into account the fact that equity long/ short funds are sensitive to equity market movements. In practice the sensitivity of equity long/short funds to equity markets can vary substantially by fund with some behaving almost like substitutes for long only equity managers, while others retain a much lower exposure. Our assumptions are based on our assessment of the average sensitivity across the entire universe of equity long/short managers. Given the nature of the asset class, our hedge fund return assumptions are more stable than, for example, our US equity return assumption. Nonetheless, the strategies are impacted by changes to the other asset class assumptions. For example, most hedge funds are cash+ type investments to a greater or lesser extent so the rise in cash return expectations during 2013 raised the hedge fund assumptions and the decline during 2014 has lowered the hedge fund assumptions. Similarly, changes to our equity and high yield return assumptions influence assumptions for those strategies which are related to these markets, such as equity long-short and distressed debt. Aon Hewitt 11

12 Volatility 15yr Inflation-Linked Government Bonds 9.0% 15yr Government Bonds 11.0% 10yr Investment Grade Corporate Bonds 9.0% Real Estate 12.5% US High Yield 12.0% Emerging Market Debt (USD denominated) 13.0% UK Equities 19.0% US Equities 17.0% Europe ex UK Equities 20.0% Japan Equities 19.0% Canada Equities 19.0% Switzerland Equities 19.0% Emerging Market Equities 30.0% Global Private Equity 25.0% Global Fund of Hedge Funds 9.0% History, forward looking indicators and our view on the economic cycle all enter our volatility assumption setting process and the volatilities in the table above are representative for each asset class over the next 10 years overall. For illiquid asset classes, such as real estate, de-smoothing techniques are employed. All volatilities shown above are in local currency terms. For emerging market equities, global private equity and global fund of hedge funds the local currency is taken to be USD. We have conducted a review of our long term volatility assumptions this quarter. This has involved further detailed analysis of historic volatility levels, including correcting for serial correlation in returns, market implied future volatility levels (taken from option prices) and a re-assessment of our views regarding the outlook. 30% 25% 20% 15% 10% 5% 0% -5% have extended this approach to cover asset classes which are traditionally regarded as liquid, such as equities. We would note that realized volatility has been exceptionally low over the past few years at well below historic average levels. Our revised assumptions continue to assume that volatility will increase significantly from recent levels. However we no longer expect it to be at such above average levels in the long run in the manner we have done previously. To illustrate the impact of these changes, we have also included a floating bar chart below showing a 90% confidence interval for our 10 year equity return assumptions this quarter compared with last quarter i.e. a 5% probability that the 10yr annualized return is above the upper end and 5% probability that it is below the lower end. The decline in median return assumptions has shifted the level of the bar charts slightly whereas the change to the size of the range has been influenced by the revisions to the volatility assumptions. In most cases the impact has only been minor. 90% confidence interval for 10yr annualized return assumptions Median return assumptions The implication of this analysis is that we have amended our assumptions to bring them closer to historic experience. In the majority of cases this has involved a slight reduction in volatility assumptions compared with last quarter though we also applied a technical improvement relating to desmoothing that raised our calculation of historic volatilities 1 and in a few cases, this has actually led to an increase in our volatility assumptions. We had already utilized de-smoothing with regard to illiquid asset classes such as real estate but -10% US-last Q US-this Q Canada-last Q Canada-this Q Europe-last Q Europe-this Q UK-last Q UK-this Q Switzerland-last Q Switzerland-this Q Japan-last Q Japan-this Q EM-last Q Please note that due to the level of yields and shape of the yield curve in Japan and Switzerland, lower volatility assumptions apply to bond investments in these markets. EM-this Q 1 De-smoothing has an impact where asset classes exhibit serial correlation i.e. when price movements are related to each other over time. This is most common in illiquid assets when valuations can be stale but also occurs in liquid traditional asset classes such as equities. Emerging markets suffer from this more than developed markets. The problem with serial correlation is that it dampens volatility, making affected asset classes appear less volatile than they really are. Correcting for this raises measured volatility levels. 12 Capital Market Assumptions

13 Correlations IL FI CB RE UK Eq US Eq Eur Eq Jap Eq Can Eq CHF Eq EM Eq Gbl PE Gbl FoHF IL FI CB RE UK Eq US Eq Eur Eq Jap Eq Can Eq CHF Eq EM Eq Gbl PE Gbl FoHF 1 Domestic Inflation-Linked Government Bonds Domestic Government Bonds Domestic Investment Grade Corporate Bonds Domestic Real Estate UK Equities US Equities Eurozone Equities Japan Equities Canada Equities Switzerland Equities Emerging Market Equities Global Private Equity Global Fund of Hedge Funds The matrix above sets out representative correlations assumed in our modeling work, shown on a rounded basis. All correlations shown above are in local currency terms and can be used by UK, US, European, Canadian and Swiss investors for the asset classes where return and volatility assumptions exist (e.g. Swiss real estate is not modelled). A different set of correlations apply for Japanese investors. Correlations are highly unstable, varying greatly over time, and this feature is captured in our modeling where we employ a more complex set of correlations involving different scenarios. Our correlations are forward looking and not just historical averages. In particular, we think that in many ways the experience of this millennium has been quite different from the previous 20 years, being more cyclical in nature with less strong secular trends. This has many implications. For example, the equity/government bond correlation in the table above is negative which also incorporates the feature that this correlation is negative in stressed environments. The lead article to the 30 June 2014 Capital Market Assumptions publication included further detail on the drivers of the equity/government bond correlation. Aon Hewitt 13

14 Capital Market Assumptions Methodology Overview Aon Hewitt s Capital Market Assumptions are our asset class return, volatility and correlation assumptions. The return assumptions are best estimates of annualized returns. By this we mean median annualized returns that is, there is a 50/50 chance that actual returns will be above or below the assumptions. The assumptions are long term assumptions, based on a 10 year projection period and are updated on a quarterly basis. Material uncertainty Given that the future is uncertain, there is material uncertainty in all aspects of the Capital Market Assumptions and the use of judgement is required at all stages in both their formulation and application. Allowance for active management The asset class assumptions are assumptions for market returns, that is we make no allowance for managers outperforming the market. The exceptions to this are the private equity and hedge fund assumptions where, due to the nature of the asset classes, manager performance needs to be incorporated in our Capital Market Assumptions. In the case of hedge funds we assume average manager performance and for private equity we assume a high performing manager. Inflation When formulating assumptions for inflation, we consider consensus forecasts as well as the inflation risk premium implied by market break-even inflation rates. Government bonds The government bond assumptions are for portfolios of bonds which are annually rebalanced (to maintain constant duration). This is formulated by stochastic modeling of future yield curves. Inflation-linked government bonds We follow a similar process to that for nominal government bonds, but with projected real (after inflation) yields. We incorporate our inflation profiles to construct nominal returns for inflation-linked government bonds. Corporate bonds Corporate bonds are modelled in a similar manner to government bonds but with additional modeling of credit spreads and projected losses from defaults and downgrades. Other fixed income Equities Equity return assumptions are built using a discounted cashflow analysis. Forecast real (after inflation) cashflows payable to investors are discounted and their aggregated value is equated to the current level of each equity market to give forecast real (after inflation) returns. These returns are then converted to nominal returns using our 10 year inflation assumptions. Private equity We model a diversified private equity portfolio with allocations to leveraged buyouts, venture capital, and mezzanine and distressed investments. Return assumptions are formulated for each strategy based on an analysis of the exposure of each strategy to various market factors with associated risk premia. Real estate Real estate returns are constructed using a discounted cashflow analysis similar to that used for equities, but allowing for the specific features of these investments such as rental growth. Hedge funds We construct assumptions for a range of hedge fund strategies (e.g. equity long/short, equity market neutral, fixed income arbitrage, event driven, distressed debt, global macro, managed futures) based on an analysis of the underlying building blocks of these strategies. We use these individual strategies to formulate a fund of hedge funds assumption which is quoted in the Capital Market Assumptions. Currency movements Assumptions regarding currency movements are related to inflation differentials. Volatility Assumed volatilities are formulated with reference to implied volatilities priced into option contracts of various terms, historical volatility levels and expected volatility trends in future. Correlations Our correlation assumptions are forward looking and result from inhouse research which looks at historical correlations over different time periods and during differing economic/investment conditions, including periods of market stress. Correlations are highly unstable, varying greatly over time. This feature is captured in our modeling. Emerging market debt and high yield debt are modelled in a similar fashion to corporate bonds by considering expected returns after allowing for losses from defaults and downgrades. 14 Capital Market Assumptions

15 Contacts Duncan Lamont T: +44 (0) Disclaimer This document has been produced by Global Investment Consulting of Aon Corporation. Nothing in this document should be treated as an authoritative statement of the law on any particular aspect or in any specific case. It should not be taken as financial advice and action should not be taken as a result of this document alone. Consultants will be pleased to answer questions on its contents but cannot give individual financial advice. Individuals are recommended to seek independent financial advice in respect of their own personal circumstances. Aon Corporation 200 E. Randolph Street Chicago Illinois 60601, USA Copyright 2015 Aon Corporation

16 About Aon Hewitt Aon Hewitt is the global leader in human resource consulting and outsourcing solutions. The company partners with organizations to solve their most complex benefits, talent and related financial challenges, and improve business performance. Aon Hewitt designs, implements, communicates and administers a wide range of human capital, retirement, investment management, health care, compensation and talent management strategies. With more than 29,000 professionals in 90 countries, Aon Hewitt makes the world a better place to work for clients and their employees. For more information on Aon Hewitt, please visit Aon plc All rights reserved. The information contained herein and the statements expressed are of a general nature and are not intended to address the circumstances of any particular individual or entity. Although we endeavor to provide accurate and timely information and use sources we consider reliable, there can be no guarantee that such information is accurate as of the date it is received or that it will continue to be accurate in the future. No one should act on such information without appropriate professional advice after a thorough examination of the particular situation. Aon Hewitt Limited is authorised and regulated by the Financial Conduct Authority. Registered in England & Wales. Registered No: Risk. Reinsurance. Human Resources.