4.1.1 Which investors have we considered?

The world’s largest asset owners include pension funds (AUM $41.4 trillion at end of 2017)Footnote ⁴ and insurance firms (AUM $23 trillion in 2015)Footnote ⁵ .

Long-term investors also include sovereign wealth funds, endowments and foundations, Venture Capital/Private Equity funds, mutual funds, corporations, and high net worth individuals, but for the purposes of this paper, the subset of investors considered will span insurers with long-term liabilities and mostly private sector defined benefits pension funds. Although the discussion following (with respect to industry practices and regulatory aspects) originates mainly from a UK perspective, we believe that most of the economic principles are applicable to these investor classes globally.

In the UK, institutional clients continue to account for the majority (79%) of total assets under management (all asset classes, as at the end of 2017). The largest client group remains UK and overseas pension funds, accounting for 44% of total assets. Assets managed on behalf of insurance represented approximately 15% of all assetsFootnote ⁶ .

4.3.1 Structural risk capacity and investment strategy

An entity’s economic risk capacity should not be confused with its risk appetite or mood. The former is the investor’s ability to bear some amount of volatility risk, whilst the latter the investor’s willingness or unwillingness to take that risk. The long-term investor has a natural ability/capacity to bear the shorter term pricing fluctuations, owing to the extended future time period over which it becomes liable to pay out cash. An example of this is illustrated in Table 1, where the CPPIB takes a genuinely long-term view. However, an investor’s ability to take risk not only depends on the economics, in terms of the nature of their liabilities, but also on practical or external considerations, for example, in terms of regulation.

Bearing in mind their risk capacity and appetite, an entity will choose an investment strategy which is consistent with both their strategic risk capacity and their risk appetite at a given point in the cycle.

Table 1. CPPIB case study

This topic of structural risk absorptive capacity is covered in greater detail in PersaudFootnote ⁸ , in which the author describes how ‘The different requirements of short and long-term funding provide a genuine capacity to price an asset differently,’ and notes that not only does this make economic sense for the long-term investor, but it also mitigates systemic risk since “The right place for a risk is where there is a capacity to absorb that type of risk. One critical advantage of placing risk where it can be best absorbed if it erupts is that it becomes less dependent on its size being measured correctly. Incorrect measurement of risk is at the heart of financial crises… If risks in the financial system are in the wrong place, then no reasonable amount of capital will save the system”. Persaud favours reinventing financial regulatory systems to change the focus away from risk-sensitivity towards measuring sizes of mismatches between market volatility risk taken and the investor’s structural risk absorptive capacity.

4.3.2 The importance of, and defining, investment shortfall risk

Although different types of investor will be faced with the same array of risks, the liabilities faced by a particular investor will heavily influence the appropriateness of different investment approaches. In particular, the liabilities affect:

• ○ The term of investment (short term versus long term).

• ○ The need for inflation protection from inflation linked assets (e.g. RPI/CPI)

• ○ The nature of the liability (including size of reserves, certainty of cash outgo, whether there is any new money from premiums/contributions, treating customers fairly)

The liability structure is thus a core influence on the entity’s risk capacity.

To emphasise the importance of the liabilities to the investment approach, in addition to the different categories of risks faced by investors outlined in Appendix 2, the concept of investment shortfall risk is of critical importance.

As previously stated, for the purposes of this paper investment shortfall risk is understood and defined in terms of economic reality over time: and is defined in Table 2.

Table 2. Definition of investment shortfall risk

4.3.3 Understanding Investment shortfall risk

Different types of investor take different approaches and may therefore warrant the use of different risk measurement approaches. By definition, this paper will not be able to fully address the bespoke needs or goals of any specifically identified business entity but will focus on the most important economic fundamentals which should drive strategy decisions.

For the purposes of this paper, the measure of investment horizon shortfall risk will be the extent to which the actual results achieved fall short of these targeted outcomes. The size of the shortfall risk will depend critically on the extent of the match between investment horizon and funding horizon.

Figure 1 shows the concept of shortfall risk, for a single cashflow at a point in the future.

Figure 1. Shortfall/surplus and the concept of shortfall risk – Cash flow (point in time)

4.3.4 Shortfall risk: a global case illustration

Consider a case where the targeted outcome is investment returns which at least preserve the investor’s purchasing power, i.e. returns in excess of inflation. For this study, a constructed S&P 500 Total Returns index was used as a sample proxy for a well-diversified institutional investment portfolio.

Figure 2 shows a comparison of annual S&P 500 nominal total return performance against US inflation over differing periods of investment, providing the average annual excess returns over annual inflation for varying time periods of investment along with the number of instances of underperformance. This chart uses data for over a century from 1913-2017 and considers overlapping periods of investment.

Figure 2. Chart of real annual returns

The data for these returns charts was sourced from the following:

“Board of Governors of The Federal Reserve System of the United States”; “Bureau of Labor Statistics (“BLS”) of the U.S. Department of Labor”; “Federal Reserve Bank of St. Louis Economic Research Division – Federal Reserve Economic Data (“FRED”)”; “Multpl.com, S&P 500 dividend yield by month”; “Multpl.com, S&P 500 historical prices by month”; The World Bank, Databank – Global Financial Development”.

Generally, we would expect that shortfall risk is diversified by time and decreases over increasing time horizons. The chart in Figure 2, which is based on actual market data over the past century, appears to support this hypothesis.

Over the past century, nominal annual equity returns have fallen short of the targeted outcome of earning returns in excess of the inflation in at least 25% of the instances for short-term investment of less than 5 years but it steadily declines to less than 5% for investment terms of 15 years or longer.

The returns profile of long-term investment, particularly that of publicly listed equities, is a noisy one, with short-term variability often obscuring and distorting the path of the long-term trend line.

A number of studies have shown that, with increasing time horizon, the long-term trend tends to reach a point beyond which the effects of the short-term trading noise become far less relevant.Footnote ⁹ This is shown in Figure 3. Asset valuations tend to oscillate over a multi-year horizon around a fundamental value of future cash streams. Further, for assets which are bought and held, the transient capital value fluctuation does not matter as much.

Figure 3. Investment returns over increasing time horizons

This study focuses on review of equity investments especially in the context of a pension scheme with continuous investment inflows requiring sustainable long-term value creation with limited risk of underperformance.

Figure 3 considers the constructed S&P 500 annual Total returns over the past century from 1913 to 2017 for overlapping time period of investment ranging from a year to up to 30 years. This chart shows that returns have been highest over shorter investment horizons before stabilising to marginally above 10% p.a. over longer investment horizons. High volatility (standard deviation in returns) has been experienced over shorter investment horizons but this reduces significantly as the time period of investment increases while the returns are broadly stable around the 10% p.a. mark. These trends appear to imply that equities have provided strong and stable returns over the long term and assist in long-term value creation with relatively low volatility.

Additionally, to more comprehensively assess the risk of underperformance in equities, equity returns were evaluated against alternative low risk investments. In this regard, the constructed equity proxy returns were compared against returns on US 10 year Treasuries for overlapping time periods of investment ranging from a year to up to 30 years during the last century (1913 to 2017). Average annual excess returns over a fixed period of investment along with the number of instances of underperformance with respect to long-term interest rates are shown in Figure 4:

Figure 4. Investment returns of equities versus bonds

The sample equities portfolio has provided higher returns than 10 year Treasury Bond but, historically, the risk of underperformance is large over the short term to the near medium term (five years). Later, the historical proportion of lower equity returns than long-term interest rate reduces significantly and over the far end of the long term (greater than 25 years), reduces to 0%. Thus, historically, these equities have provided approximately 5% p.a. higher returns than the long-term Treasury Bond whenever the period of investment was more than 25 years.

This analysis above examined sample equity performance over the past century including time periods influenced by major economic and geopolitical events like the 1929 Great Depression, the two world wars, and so on. However, past performance may not necessarily provide an indication for the future prediction. (For a fuller description of limitations relating to use of past data, please see the Disclaimer).

4.3.5 Shortfall risk: capturing both sides of the trade off

This feature (the shortfall risk varying by time) drives fundamental differences between different entities – for example, banks versus insurers and pension funds.

The driving principles are well expressed in the IMF Working Paper (Reference Jones2016) which stated that long-term shortfall investment risk could be broken down into

1. 1. The risk of large negative returns causing the fund to lose value, as well as

2. 2. The risk of forgoing large positive returns, thus failing to achieve the targeted long-term return objective.

In other words, shortfall investment risk should be concerned about taking too much and too little risk – i.e. consideration of a risk management approach that takes into account both the left and right tail of the return distribution.

This is in contrast with many market risk management approaches which only focus on the left tail of the return distribution, for example, the use of the value at risk (‘VaR’) measure – which is generally the focus of shorter term investors (referred to as ‘Type 2’ investors in the IMF paper – see Table 3).

Table 3. Distinction of different types of investor. Source: Jones, B.A., (Reference Jones2016). Institutionalizing Countercyclical Investment: A Framework for Long-term Asset Owners, IMF Working Paper WP/16/38

The distinction made in the IMF paper reinforces the points made in this paper; namely that not all investors are ‘Type 2’ investors and the risks faced by longer term investors (‘Type 1’) are of a different nature.

4.4.1 Liquidity measures and asset class

Liquidity risk assessment especially for evaluating performance of long-term investment must adopt measures that assess investments taking into account the underlying business risk. Therefore, these measures must:

1. (i) consider the objectives for the investment and the nature of the liabilities required to be serviced; and

2. (ii) ensure risks are assessed over a period of time that corresponds with the tenor of risk being held.

The measures that are used to evaluate the liquidity of various assets classes may include evaluation of:

• Marketability – reflects the depth of the market, which is dependent upon the level of activity in the market. This can be reviewed through measures such as turnover ratio or trading volumes etc.

• Resilience – indicates that assets prices are not significantly impacted by the level of market activity.

• Immediacy – indicates that the time to complete transactions is low.

Further consideration of the marketability aspect of liquidity is considered in Appendix 3.

4.4.2 Liquidity considerations specifically for UK insurers and pension funds

In general, insurance firms do not explicitly hold capital specifically for liquidity. In SII however there are limitations on the matching adjustment (‘MA’) where the liabilities have variable cashflows, and this is similar to holding extra capital. The need to restore MA compliance within a 2 month time frame in a stress could incentivise firms to invest in lower return, more liquid assets such as corporate bonds rather than higher return assets such as illiquid assets.

Not requiring capital to be held in respect of liquidity risk could be viewed to be sensible because, in a liquidity event, holding capital is not necessarily viewed as a suitable mitigant. However, this does not mean that insurers should ignore the effects of liquidity risk.

Liquidity Risk is primarily a part of the Pillar 2 of the SII Directive. Organisations must have a robust governance system and demonstrate a proper risk management with respect to various risks including liquidity risk. In particular, the underlying objective is to ensure:

• identification and proper assessment of all risks they are (or could be) exposed to;

• maintenance of sufficient capital to face these risks; and

• development and usage of better risk management techniques in monitoring and managing such risks.

It has been noted that the points mentioned here can be mutually contradictory, since identification and assessment of the risks in the economic context of the (long term) business do not always imply that more capital is the optimal approach.

Also, in accordance with SII directives, national competent authorities should ensure that in its risk management policy, the undertaking covers at least the following items with regard to liquidity risk:

• the procedure for determining the level of mismatch between the cash inflows and the cash outflows of both assets and liabilities, including expected cash flows of direct insurance and reinsurance such as claims, lapses or surrenders;

• consideration of total liquidity needs in the short and medium term, including an appropriate liquidity buffer to guard against a liquidity shortfall;

• consideration of the level and monitoring of liquid assets, including a quantification of potential costs or financial losses arising from an enforced realisation;

• identification and costs of alternative financing tools;

• consideration of the effect on the liquidity situation of expected new business.

In terms of various assets classes, the SII directive specifically focuses on the level of mismatch in the payoffs of the assets and the liabilities and thus the management of the liquidity shortfall. Also, the level of liquid assets must be monitored and the potential costs or financial losses arising from an enforced realisation in the form of forced sale, repo etc. must be considered.

The consideration of stressed scenarios for insurers assumes that the banking sector will remain robust. As demonstrated in the financial crisis, such an assumption does not always hold – particularly under instances of extreme financial stress and so such scenarios may be considered as unrealistic in extreme environments. A full discussion on systemic risk is beyond the scope of this paper.

For pension funds, there is no explicit (regulatory) requirement for trustees to manage or quantify liquidity risks but it is one of the key considerations that trustees should consider when setting the investment strategy:

4.5.1 Human emotions and investor trading behaviour

In particular, risk is not just an objectively quantifiable number – it also depends heavily on people’s perceptions, emotions and expectations. As such, understanding the economics of the investment approach is not identical to the unambiguous logic of mathematical equations. The literature on behavioural economics is comprehensive and outlines how real life decisions made by individuals can vary to those implied by classical theory. For example, due to:

• Heuristics: the human mind desires to understand the world, and seeks a few simple, elegant ‘rules’ to ‘elegantly’ explain as much of the world as possible and the basis on which to make decisions (‘rules of thumb’).

• Framing: whereby anecdotes and stereotypes might influence the way in which individuals understand and respond to events.

• Market inefficiencies: for example, as demonstrated in momentum effects for stocks or how social paradigms change over time and cycles of fashion exist (e.g. whether companies are private/public, sources of funding, role of government/social protection).

Some of these theories can be used to explain some of the issues outlined above – for example, whilst some fear is essential for survival, excessive fear can be damaging (e.g. paralysis, stagnation) especially if socially transmitted, amplified and reinforced over time.

The role of surrogate decision making, whereby decisions are taken on behalf of others (e.g. pension Trustees) amongst long-term investors might also influence the level of risk being taken. Surrogates often display more muted behaviour (in investment decision making) converging towards more socially accepted choices. As surrogates want to be socially seen as making the right public decisions on behalf of others, their decisions may inter alia be influenced by self-image preservation and may create a bias towards more conservative choicesFootnote ¹² .

4.5.2 Behavioural incentives of different stakeholders

Behavioural aspects also play a most influential part in the socioeconomic context. One crucial consideration is that there are asymmetric incentives impacting on some important decision makers. The regulator itself faces substantial risk from loss-causing events but has scant incentives to optimise the economic growth in the country’s economy. This impact is compounded by the fact that the preferences of public policymakers tend to lean strongly towards conservatism to avoid any need to support financial institutions under any circumstances.

There is a trade-off between the level of security and the price the consumer has to pay for that insurance… The price the consumer has to pay for that is something that they will not get a choice in. That choice is made by the regulatory system.Footnote ¹³

Whilst the topic of behavioural economics is vast and thus a full discussion is beyond the scope of this paper, its relevance to the theme of investment risk should not be ignored.

4.5.3 Risks are heterogeneous

Notwithstanding the human inclination to want to simplify, treat things uniformly and with a common base, and assign an arithmetic value to quantities, it should be recognised that there is no one homogeneous ‘substance’ called ‘risk.’ In Section 4.3, we highlighted how risk was inherently subjective and ‘in the hands of the holder.’

Further, different risks might need to be mitigated in different ways. Whilst there is some degree of interdependency, liquidity risk and credit risk should not be confused – they are hedged substantially differently.

Whilst capital can act as a buffer against credit risk, it is not necessarily appropriate to hedge liquidity risk. An investment portfolio’s economic exposure to credit risks is mitigated by instrument holdings diversification, for example, across the layers of an issuer’s capital structure, across different industrial sectors and geographies. In sharp contrast, liquidity risk cannot be similarly hedged by increasing the number of distinct portfolio holdings, it is a risk which varies with the investment markets’ supply/demand reactions and trading over time. Holding capital is not the only, and indeed often not even the best, way to mitigate risk.

4.6.1 Life insurers

Insurers have structurally disinvested in equities over many years, and as mentioned above now account for only a small portion of total growth assets invested in the UK. This is partly driven by consumers losing confidence in with-profit products and by the rise of unit-linked products where the choice of investment is not within the control of the insurer.

Insurers are very heavily regulated – this, along with accounting rules and conventions, is a key driver of asset allocation. The situation is exacerbated by the introduction of SII where liabilities are valued on a market consistent basis.

The EU regulatory capital standard under SII is a one-year value at risk measure. By this we mean that risk is measured over a one year period based on a certain probability of loss. For the SII regulatory capital requirement, the Solvency Capital Requirement (‘SCR’) is set as the 99.5^th percentile loss over the one-year period. This could potentially be viewed as putting too much emphasis on short-term volatility instead of focusing more on long-term volatility. SII makes some allowance to reduce pro-cyclical effects and recognising the long-term nature of liabilities (for example, through the equity dampener, use of volatility adjustment (‘VA’), matching adjustment (‘MA’), etc) but these measures are often rigid and may be viewed as too prescriptive.

The Own Risk and Solvency Assessment (‘ORSA’) under SII encourages firms to perform their own bespoke analyses, including carrying out long-term run-off modelling on their own assumptions and use that modelling to set their investment strategy.

Liquidity considerations are typically managed through internal management controls of liquidity planning, for example, assessing the liquidity needs in various liquidity stress conditions. The balance of liquidity depends on the mix of business written where, for example, a closed with-profit fund running-off is likely to have a proportionately higher liquidity requirement compared to an annuity fund which is actively writing new business.

4.6.2 General insurers

With shorter tailed liabilities, the investment focus for General Insurers tends to be over shorter time horizons. However, the provision of PPOs are increasing the investment outlook for some General Insurers – particularly in respect of their larger claims for which PPOs are more prevalent.

A large number of PPOs are provided from, or near, the public sector (for example, through the NHS Resolution in respect of clinical negligence) which means that the principles of investment are either not applicable (because the pay-as-you-go nature means there are no funds held) or the principles are distorted.

For regulated private sector General Insurers that provide PPOs, the extent to which the investment strategy is more focused towards the longer term depends on the number of settled PPOs. Whilst there are external factors and trends influencing PPO propensity, the insurer will have their own views and some influence on the extent to which claims settle as PPOs. The key issue in this regard is the relative attractiveness of a PPO against a lump sum settlement, with claimants weighing up the size and risks of different award types and insurers weighing up the capital consequences.

Although an insurer may have some influence over whether a claim settles as a PPO or not (through the settlement process), ultimately a PPO can be imposed by the Court and so this may present a significantly different investment challenge to what a General Insurer is used to. In particular the main challenges to a PPO provider are:

• The liabilities are often significantly longer than a General Insurer’s other claims risk.

• The insurer is exposed to the longevity risk of the claimant living longer than expected.

• PPOs are normally linked to an earnings index (‘ASHE’) which cannot be readily hedged.

• There is the possibility of a step increase in the amount payable if the PPO is reopened as the claimant’s circumstances change.

As such, the ‘interest rate risk’ that is usually a small component of the overall risk for a General Insurer (relative to Life Insurers and Pension Schemes) can become more significant. There are though potential benefits to the insurer in that PPOs have a lower liquidity requirement – due to the immediate cashflow requirement being lower.

Many insurers make use of reinsurance to share the risks of PPO claims. Whilst to some extent this may reduce the insurers’ overall level of risk, some PPO reinsurance contracts allow for the reinsurer’s share to be capitalised – hence meaning the General Insurer bears all of the investment risk.

4.6.3 Pension funds

In the pension landscape, there has been an obvious shift towards Defined Contribution (‘DC’) schemes where the investment and longevity risk is borne by the individual. Whilst funds are still managed by institutions, and the investment choices made by individuals may be subject to the behavioural biases discussed in this paper, we have not considered DC pensions as part of our core focus – as they relate to individual investments.

For Defined Benefit schemes, their investment purpose is to provide retirement security in the form of real purchasing power for retirees. Risks to a pension fund include longevity, domestic economic performance and wage growth. Essentially pension funds want investment returns to hedge against the risk these factors pose to the level of contribution rates and benefits paid.

As benefits payable to members are defined, shortfall risks are borne by sponsors. In reality, if at all, ultimately only a State can afford to offer guarantees because of the taxation power. Not only do most private sector entities not possess such a source of finance but also few private sector directors truly understand the nature of an open-ended financial guarantee.

Broadly speaking, before 1997, most pension increases were either discretionary or guaranteed at minimal levels. That changed for all future benefit accruals and for some schemes for prior benefit accruals.

Over the past 20 years or so, financial economics has become generally accepted by UK pension actuaries as the dominant approach to assessing funding requirements. Whilst this provided the basis for funding approaches, some highlight that there is no evidence that this can be relevant to long-term entities. The theoretical academic evidence related to perfect markets, but perfect information is not generally available, and it cannot be (Grossman & Stiglitz, Reference Grossman and Stiglitz1980).

As a separate rationale, it was vigorously stated that using bond yields was, in any case, prudent, so not understating the liability was seen to be in the members’ best interests. However, prudence can only be identified from the best estimate, which is rarely, if ever, disclosed to stakeholders.

Further, discounting cashflows to a single liability value on which to make funding and investment decisions suggests a greater degree of certainty about the future than can be justified. Using capital values fails to allow the communication of the many risks and rewards that may lie ahead and their relative likelihoods.

Since 2003, the vast majority of private sector defined benefit schemes have been closed. This has had the effect that more of the benefits than before have become guaranteed. That has led to sponsors and trustees being advised to switch very largely from real assets to bonds. During the last decade since the crisis, bond yields have been exceptionally low, especially for index-linked gilt-edged stocks. The value is hard to discern. If anything, given that the cashflows stretch far into the future, a pension fund should be a natural home for ‘patient capital’ (see FCA consultation for DC plans, announced 12, December 2018).

4.7.1 Insurance companies

Currently, rules for the amount of capital that UK insurance companies must hold are specified in SII regulatory regime, Directive 2009/138/EC of the European Parliament and of the Council (2009), which came into force with effect from 1 January 2016. Since then, there have been a number of ongoing consultations and discussions addressing the real and perceived shortcomings of, and potential improvements to, the existing specifications.

Article 101 under the Level 1 Directive sets out the methodology for calculating the Solvency Capital Requirement under the SII framework. Paragraph 3 of Article 101 sets out the time horizon for this purpose:

Under SII’s Standard Formula, the base equity shock is 39% for Type I equities (equities listed in countries which are members of the EEA or the OECD) and 49% for Type II equities (all other equity types).

There are various measures within the SII framework which allow some mitigating offsets to the severity of the specified shocks, and these are touched on briefly later in the notes that follow.

In late 2016, the UK Treasury Select Committee (‘TSC’) launched an inquiry into the SII regime. A number of pertinent points were raised by several key respondents:

Impact on equity investment and other real assets

The IFoA was of the view that ‘market consistent approaches have discouraged investment in equities other than where the policyholder bears the risk, such as on unit-linked business.’ There are both positive and negative perspective to consider:

1. 1. From a positive perspective, ‘it discourages firms from giving over-generous guarantees and ensures de-risking can take place at short notice.’

2. 2. However, this could disincentivise investments in ‘long term risky assets’. ‘For example, if the objective is to generate strong real returns with say a 20 year investment horizon then equities are a good asset as over this time period their returns are likely to be highly correlated with long term inflation impacts as indicated through the long term equities value creation review under Section 4.3. However, SII imposes a one-year horizon for capital and over that time period equity returns may be strongly negatively correlated with inflation.’

(By way of background, ‘Actuarial risk principles’ Section 6. on ‘Measure the risk’ states that one should allow for possible effects over the full time horizon of interest: different factors may be more prominent over different time scales. It is important to clarify which time horizon is most important for the stakeholder, so that we can focus on considering the right system drivers for that period. For full details, see Institute and Faculty of Actuaries (“IFoA”) (2017). Risk management – an actuarial approach.)

Due to the stringent requirements underlying SII and in particular the matching adjustment constraints, companies have had to jump through many ‘hoops’ before including suitable asset classes within the matching adjustment (‘MA’) portfolio. These include the need to create ‘“matching adjustment friendly” (e.g. decomposing property assets into income strips and residual property values due to the need for asset cashflows to be fixed unconditionally to be matching adjustment eligible). This type of activity does not lead to any reduction in investment risk. Indeed, the additional complexity adds operational risk, creates asset valuation challenges, reduces transparency and adds cost.’ Footnote ¹⁴

Criticism of market consistent principles and 1 year value at risk (‘VaR’)

Some criticism from the industry highlighted that these principles could encourage pro-cyclical behaviors which causes systemic risks. Market consistent techniques are neither stable nor verifiable for the full range of assets and insurance liabilities. Embedded within market-consistent valuation methodologies is the fundamental belief in the efficient markets hypothesis (i.e., that at any given time and in a liquid market security prices reflect all available information; it is not possible to know that you could know more than the current price).Footnote ¹⁵

The 1-year VaR framework is not necessarily considered to be compatible with insurer’s long-term liabilities where there this method may ‘fail to capture the path dependency of risk factors beyond the 1 Year period.’

Furthermore, valuing a long term business on a 1 Year basis is at odds with the time horizon of a going concern enterprise or the new requirements for directors to provide a long term viability statement in the annual report, since it says little about the ability of an insurer to meet its liability-related cash flows as they fall due.Footnote ¹⁶

Some SII mitigants

The long-term guarantee packages in the form of MA and volatility adjustment (‘VA’) helps balance things somewhat but these measures are still viewed to be overly prescriptive and perhaps opportunities for a principles-based approach may be available post-Brexit. While MA has a generally high take-up in the UK, on VA, there appears to be general consensus that the rules are too rigid. Recently, there has been discussion between major industry players and the regulator on possible ways to evolve a more dynamic VA.

Table 4. Regulatory versus accounting buffers. Source: Legal & General Group plc written response to Treasury Committee enquiry SII (SOL0009 of November 2016)

Further unintended consequences of stringent time-sensitive capital regulations could include wider macroeconomic impacts: there is greater systemic risk. The market-consistent elements of SII encourage behaviours which tend to move all market participants in the same direction simultaneously, thereby increasing the risk to the financial system as a whole on a much larger scale. SII’s rigidity is illustrated in Table 4, where it is contrasted to the accounting approach to the balance sheet.

In summary, notes from the EU High-Level Expert Group on Sustainable Finance (‘HLEG’) (2018) seem to express the issues concisely:

4.7.2 Pension funds

Introduced in 1997, the Minimum Funding Requirement (‘MFR’) was discontinued in 2006, with the advent of the Scheme Specific Funding Requirement (‘SSFR’, see below). It was intended to provide a simple uniform approach for determining whether or not a discontinuing scheme would have sufficient assets to secure the accrued benefits with a specialist pension provider. Originally, it was intended to be the Minimum Solvency Requirement but that was thought to be too ambitious an objective. In reality, the MFR was insufficiently tied to the actual market cost and was seen to have failed. The idea of uniformity was also held to be out of place and the SSFR was introduced in 2006.

Sponsors had already had to take account of revised pension accounting standards, with FRS17 Retirement Benefits introduced in 2003Footnote ¹⁷ and IAS19 Employee Benefits introduced in 2005. The differences between the standards were not that major but they were applied to different types of organisation. These accounting regulations, along with falling bond yields with which the accounting basis is set, led to large accounting pension deficits being shown on the sponsors’ balance sheets, which led to some pension schemes being discontinued.

In June 2003, the Debt Upon Employer regulations were enacted without any consultation. This new legislation made solvent sponsors responsible for pension scheme discontinuance benefits.

Tied to each pension scheme’s circumstances, the trustees were placed in control of funding requirement assessments. Taking advice from their own actuary and from no others in relation to benefits funding, they also had to agree their Statement of Funding Principles with the sponsors and comply with the Pension Regulator’s strictures. Trustees were specifically permitted to set a discount rate which took account of returns on the assets held, but most trustees elected to opt for gilt-based yields. Since 2007, especially after the June 2016 Brexit referendum, those yields became increasingly lower, with correspondingly higher capital values of the perceived liabilities. With actual investment returns having failed to keep pace, the assessed deficits (or shortfalls) became very much higher, with sponsors becoming even less willing to continue the pension scheme.

Because trustees were being told that they needed to match their benefit cash flows, a greater emphasis was placed upon gilts and bonds than real assets. That such asset transfers were made when gilt yields were very low led to real substantial losses.

Indeed research by OECD and Oliver Wyman showed that between 1980 and the mid-1990s, UK pension funds steadily invested between 60-80% of their investments in equities. Since then, UK pension fund’s exposure to equities has steadily declined such that by 2008 the exposure to UK equities was below 40%. In their report, the OECD and Oliver Wyman highlight:

• The timing of changes in the regulatory regime and how this coincides with changes to funding regulation – in particular, the introduction of MFR in the mid-1990s coincided with the start of the decline of equity exposure.

• Trends in the United States and Netherlands and how exposure to equities in those countries has actually risen slightly over the same period.

Despite the clear trends, there are some schemes bucking them. For example, the BAE Systems Pension Scheme recently noted:

4.7.3 Insurance interactions with pension funds

Throughout this paper we have generally considered insurance companies and pension funds separately to reflect the different regulatory regimes that apply and the different investment approaches taken.

The authors recognise that in practice the two industries do not operate independently – in particular as pension funds continue to look for ways to secure their liabilities, for example, via buy outs. As such, there are occasions where pension funds in particular are required to quantify and asses their liabilities through an insurance lens.

As such, the linkage between the two industries is subtle but nonetheless real – and any regulatory differences between the two sectors (for example, regulation that presents barriers to insurers to accept long-term pension liabilities) impacts indirectly/directly on the ability of pension funds to de-risk without harming members.

For example, the existence of a matching adjustment for life companies has likely buoyed the bulk purchase annuity (‘BPA’) market, as pension funds continue to value pension liabilities using bonds yields irrespective of asset holdings. Against this, the existence of a matching adjustment also means that this will drive annuity providers taking on these liabilities, whether immediate or deferred, to invest in assets with fixed income flows (including inflation- linked where relevant). This is relevant for deferred BPAs where while in theory insurers could invest in equity / property during the deferred period and move into fixed income once in-payment, not having matching adjustment eligible assets on day 1 would mean that these liabilities would not be eligible for matching adjustment calculations.

4.7.4 Summary of this chapter

In this chapter, we:

• Defined the scope of investors that we are considering. In particular we are focused on long-term institutional investors (such as life insurance companies, pension funds and general insurance PPO provisions).

• Defined the concept of shortfall risk which fundamentally drives the approach to considering the appropriateness of risk measurement approaches.

• Outlined how ‘riskier’ (i.e. more volatile) asset investment returns can become ‘less risky’ if the returns are considered over appropriate time horizons.

• Outlined some of the other considerations that are taken into account when institutions determine investment strategies.

5.2.1 Discounting for long-term investors

Although the discounting principle has been known for 2 millennia, it was only adopted for longer-term finance in the late 18th century. If it were certain that a return of 3 % pa would be achieved for the next year, 103 due in a year’s time can be financed by an initial 100. By converting future cashflows to the present, it can be seen that the discount rate is the inverse of the investment return. In reality, the financial environment is far less secure, given the lack of certainty, much longer time periods and assets which are far more complex instruments than cash. Because the future is unknowable, no uniquely correct interest rate can be defined in advance. There is, in fact, one uniquely correct view of the future but this is not known in advance so that assumptions are needed.

Risk quantification is very poorly captured by scalars. Specifically, single numbers are not appropriate results for representing many future uncertainties, especially when we do not even specify what the result represents (e.g. the mean, median, mode, or a specified percentile?). Rather, multi-dimensional results with confidence intervals provide information which a deterministic approach based upon discount rates cannot convey.

Risks are only taken because potential rewards may reasonably be anticipated (depending upon the circumstances). If discount rates are used, then they need to be consistent with investment returns. Otherwise, systematic bias is introduced, which is a major problem with accounting numbers. While the stakeholders (including regulators) may want the results to be ‘prudent’, that can only be defined in relation to a ‘best estimate’ benchmark. Where the liabilities are longer than the assets, using bond yields alone may even be imprudent.

While capitalisation was the original actuarial tool, which was all we had, it is no longer the only tool available and it obscures more than it reveals (the ‘scalar impact’). Discount rates are simple and simplistic, readily available to anyone and dangerous in both the ‘wrong hands’ and the ‘right hands’. For long-term projects, the authors recognise that using discount rates alone is limited and it is more optimal to consider cashflows using a robust stochastic process, for which the technology is readily available.

5.2.2 Allowance for investment returns within pension scheme valuations

Although specifically permitted by legislation, pension scheme funding valuations in the UK typically do not make an allowance for expected investment returns in the discount rate used to value the liabilities. If they were to do so, then schemes could take credit within their liabilities for investing in growth assets. This could be seen as encouraging pension schemes to adopt a longer term investment perspective and allowing them to seek the greatest return over the long term. It also reduces the focus on short-term volatility. The drawback of this approach is that the valuations of the liabilities are dependent upon the assets held, which can be difficult to explain but the effort needs to be made. The ‘liabilities’ are really the estimated cashflows rather than the capital amount. Thus the liabilities do not depend upon the discount rate.

The discount rate used in pension scheme accounting valuations is based on the yields of high quality corporate bonds. ‘In 2003, the implementation of FRS17 was intended to quantify the true risk of a corporate pension promise. It was originally based on the premise that the credit spread between an AA rated corporate bond and a gilt represents a “small premium above the risk free rate”’Footnote ¹⁹ . However, the margins can be higher and gilt yields are not risk-free. Crucially, it would take alchemy to make the basic cashflows risk-free.

The risk that the assets are insufficient to meet the liabilities rests with the scheme sponsor(s). If there is a shortfall, a plan has to be agreed and put in place to ensure that the liabilities can be met as they fall due. Again, investment returns are often allowed for within these plans and therefore some of the shortfall is effectively assumed to be met through investment returns. This again encourages a focus on long-term investment returns.

5.2.3 Allowance for investment returns within long-term insurance regulatory valuations

Prior to the introduction of SII and the UK’s ‘twin peaks’ reporting regime, liability cash flows were typically discounted using a valuation discount rate based on the expected returns from assets backing the liabilities, less an allowance for risk.

The twin peaks regime was the first step towards a financial economics valuation approach. In particular, realistic reporting under the twin peaks regime required with-profit liabilities to be valued on a risk neutral basis (‘Peak 2’ basis) and this valuation is compared with liabilities valued on a discount rate based on the risk adjusted returns from assets backing the liabilities (‘Peak 1’ basis). The primary purpose for introducing this regime is to recognise the fact that it was difficult to assess the level of prudence implicit within the Peak 1 basis and the Peak 1 basis does not make allowance for the time value of guarantees of embedded options nor adequately reserve for bonus payments in line with policyholder’s reasonable expectations.

Under SII, the reporting regime requires liabilities to be valued on a market-consistent basis and liabilities to be discounted using the European Insurance and Occupational Pensions Authority’s (EIOPA) risk-free curve. The use of a risk-free discount rate curve can be viewed as a short cut for creating replicating portfolios and modelling the risks inherent from investing in these assets. Following industry lobbying, EIOPA included an allowance for matching adjustment and volatility adjustment to ensure appropriate treatment of insurance products with long-term guarantees as part of the long-term guarantee measures. The inclusion of a matching adjustment recognises the fact that annuity writers follow a buy-to-hold investment strategy and are therefore not exposed to the illiquidity premium part of the spread movement. The purpose of the volatility adjustment on the other hand is to prevent pro-cyclical investment behaviour as a result of short-term bond spreads widening. The use of a matching adjustment and volatility adjustment is subject to regulatory pre-approval (this restrictive handling has been criticised by industry).

5.4.1 Mark-to-market (‘MTM’)

Mark-to-market valuation approaches seek to provide a market-based assessment of the value of assets and liabilities – or in other words, the price that a market participant would be expected to pay/be paid to take full ownership of both the institution’s assets and liabilities. The use of MTM valuation approaches is widespread – for example, regulatory, accounting or ongoing or funding purposes.

Whilst perhaps not strictly a risk measurement approach, MTM valuations do give a benchmark assessment of solvency that is of particular interest during periods of market stress or upon particular risk events.

The key advantages of MTM approaches are:

• Objectivity – in that most MVs for liquid listed instruments are based on actual recent marginal trades. Thus, they are not subjective since they can usually be observed.

• Apparent simplicity – the human mind is designed to simplify, and it is easy to digest a valuation as ‘just one number’

• Transparency – for listed assets, MVs are publicly available

• Immediacy – for listed assets, s are ‘live’ and change to reflect live events and announcements.

However, there are a number of drawbacks with the use of MTM approaches:

• Focus on accounting values – changes in the MTM value will not always be representative of changes in the economic substance of one or both of the assets or the liabilities. People’s focus and perceptions of risk might be transiently skewed and this will manifest in volatility in the MTM values.

• Focus on short term – related to this, the focus on live and moving MVs might promote too much focus on the short-term. This is especially a concern for institutions with long-term liabilities and investment horizons.

• Marginal trades – MVs reflect the objectives and investment circumstances of the investor’s marginal trades. As such, the values are still intrinsically subjective, particularly since these circumstances and objectives are likely to differ not only across institutions but also across different portions of each of the investor’s asset portfolios.

• Availability – market prices (or valuation approaches) are only readily available for listed assets. In particular, market prices are not generally available for liabilities or unlisted assets.

Current regulations (in the UK) require pension scheme and insurance valuations to be MTM. However, historically, other approaches have been used – particularly for pension schemes. This has been discussed in Section 5.2 under Discounted cashflow.

5.4.2 Volatility

Volatility is widely used as a risk measure – either directly or indirectly. For example, volatility might be used directly as a risk measure of annual investment risk and potential uncertainty in MVs. Alternatively, volatility is key assumption and feeds through to common risks measures such as Value at Risk (‘VaR’). A description and case study of VaR is given in Table 5.

When considering volatility as a risk measure, it is important to distinguish between:

• Historical or observed volatility – as a backward looking measure of the extent to which prices have been observed to fluctuate over a particular period of time.

• Prospective volatility – as a forward looking measure of the extent to which prices are assumed (or implied by the market) to fluctuate in the future.

Our focus is primarily on the prospective volatility.

The use of volatility as a risk measure has a number of advantages – in particular it is commonly used, relatively straightforward to understand, plentiful data to set assumptions, and it often makes mathematical models simpler to use and more tractable.

However, there are a number of widely regarded shortcomings in using volatility as a risk measure (particularly over the longer term) including:

• Asset returns do not follow a normal distribution.

• Correlations and volatilities are not constant – particularly in times of stress which are often of most interest.

• Volatility approaches often do not capture the risk of failing to meet objectives.

• Volatility measures do not capture the risk of permanent capital loss.

• Illiquidity risks are not captured.

• It might contribute to pro-cyclicality in that it might discourage the buying of risky assets when prices are low, because these typically correspond to periods of high volatility.

Further, volatility (and measures that depend heavily on volatility assumptions) can be exacerbated by behavioural effects – for example, short-term sentiment, momentum effects, herding behaviour and speculation. For a long-term investor, allowing for these effects in risk measurement approaches, and hence decision making, is likely to be sub-optimal.

These are relevant considerations for short-term investors such as banks, but are often different to what a long-term investor should be interested in.

Risk measurement techniques change over time and since the early 1990s a lot of focus has been about the spread of market sensitive risk management systems for banks, and the spillover of this approach to other financial institutions.

For long-term investors, such risk measurement techniques might not be optimal, because MVs (which are subject to a lot of volatility) are not consistent, reliable predictors of the future and thus not appropriate for long-term financial management.

Table 5. Value at risk description

^a Thimann (Reference Thimann2014). How insurers differ from banks: Implications for systemic regulation.

Materials for this table were sourced from the Wikipedia entry on Value at Risk.

5.5.1 Alternative asset calibrations

It should be obvious that the calibration approach has a key impact on the final result – it determines the magnitude of the framework from which risk is measured.

Point in time (‘PIT’) versus Through the Cycle (‘TTC’) versus Other calibrations

Asset calibrations which do not MTM at a single PIT, i.e. on given balance sheet date, can take the form of factor-based approaches where the factors used are calculated based on market information generated over a longer time period. The calibration period of historical data needs to be sufficiently long, likely substantially greater than 5 years, to capture the whole cycle. This will alleviate over-sensitivity to transient influences, although it can give a flawed picture when the underlying risks are changing rapidly.

The terms point-in-time and through-the-cycle have historically been used by credit rating agencies, and have become more widespread since the early 2000s, when Basel Committee on Banking Supervision (‘BCBS’) documents frequently referred to economic and business cycles.

The concept of ‘through the cycle’ was elaborated by Moody’sFootnote ²¹ in relation to credit risk:

In that paper, the authors state that this method is useful in situations where a stable input is desirable but qualify their view on the application of their approach by opining that it should be evaluated by comparing the expected costs of adjusting the economic (credit) exposures versus the expected cost of negative credit events (defaults).

Thus, they consider the investor’s utility function paramount.

They describe their modelling approaches as distinguishable by ‘the degree to which each approach has a PIT or TTC orientation.’

A PIT approach

• utilises all available and pertinent information as of a given date

• reacts immediately to all news that affects a firm’s risk

• is highly volatile

• is pro-cyclical

whereas a TTC approach

• primarily reflects a firm’s long-run, enduring credit risk trend

• is highly stable over the (credit) cycle

• changes smoothly over time, and hence is less volatile and pro-cyclical

• is less useful in terms of timeliness and (credit) predictive accuracy relative to PIT measures

The appropriate actions by a firm’s managers will depend on whether the change (in credit quality) is assessed to be permanent or transitory. The authors hold the view that both PIT and TTC approaches are needed for a complete (credit) risk management system.

Other forms of alternative calibrations

In some instances, variations to the standard base case rules exist within the applicable regulation itself. An example is given here:

Conditional calibration of risks

Under SII’s Standard Formula, the base equity shock is 39% for Type I equities (equities listed in countries which are members of the EEA or the OECD) and 49% for Type II equities (all other equity types). The equity shock includes an adjustment to the base equity shock (the ‘symmetric adjustment’) which is capped and floored at +/10%. This adjustment is calibrated based on how the current value of current index has performed relative to the three year average.

Including this symmetric adjustment thus allows a shock in the range of 29% to 49% for Type I equities and 39% to 59% for Type II equities to be applied.

While the primary objective of the symmetric adjustment is to reduce the pro-cyclical effects in times of stress, this is not necessarily an option which is taken up by all insurers nor included within the equity risk calibrations for internal model firms (possibly owing to the equity shock being calibrated independent of current equity levels).

Nevertheless, a shock which is conditional on current conditions albeit with its moderating adjustment restricted to +/-10% has the attraction that it helps to meet the goals of managing cyclical effects.

5.5.2 Mixed and blended approaches

It is recognised that in the majority of cases, long-term funds will have a portion of short-term cash flows, and short-term methods would be appropriate in respect of these. Long-term methods would be superior for the cash flows at the longer end, with a blend of methods in between.

5.5.3 Cost of capital/funding approaches

A common approach used in pension scheme funding is to set funding discount rates, in particular the level of prudence, with respect to the level of confidence in future asset returns. For example, for a given investment strategy, the trustees and advisors may believe that the expected return on the assets over the term of the liabilities is 3.0%pa above inflation. Trustees can then use confidence levels around that expected return to set an appropriate funding discount rate and level of prudence based on the strength of the sponsor. For example, Trustees for a scheme with a strong sponsor may set the discount rate with returns at a 60% confidence level, where they estimate that returns will be at least 2.25%pa say, whereas Trustees for a weaker scheme may choose a 75% confidence level, which may be equal to a discount rate of 1.5%pa say.

This risk measurement approach is based on using modelled scenarios to determine statistical distributions and quantile methods using percentile/confidence levels. A contrasting approach would be to consider a set of existing liabilities, and ask what level of investment returns would be needed to support these liabilities. In particular to consider the total return that is required to support an acceptable probability of meeting liabilities/objectives, for a given starting level of assets. For example, an entity may require a total return of 3.5%pa above inflation in order to be 90% confident of meeting a particular objective.

Articulating the return in this way may be broadly comparable to an entity’s cost of capital which is based on the amount of return, in addition to the amount earned by the enterprise from its investment of capital, that is required for the total return on the investment to be adequate.

5.5.4 Probability of success/failure

This risk measurement approach attempts to simulate business outcomes under a number of scenarios and quantify whether each scenario ‘succeeds’ or ‘fails’.

The approach is underpinned by having a clearly defined and articulated measure of success and/or failure. Such measures are likely to be linked to the entity’s liabilities or risks and their ability to meet these.

A stochastic model is likely to be needed to quantify outcomes in different scenarios. The simulated outcomes can then be considered to demonstrate the risk profile, any trade-offs and the impact that different levers (e.g. short-term investment decisions or pricing strategies) might have on the chances of success.

The Pension Protection Fund adopt a similar approach, as outlined in the case study in Table 6.

Table 6. Case study of probability of success/failure approach

The key advantages of this approach are:

• The assessment is linked to the entity’s business and objectives.

• As such the time horizon of the investment period reflects the entity’s liabilities.

The key drawbacks of this approach are:

• Developing the modelling approach is likely to be a significant undertaking.

• The results are likely to be highly sensitive to a number of assumptions that are required.

• In particular, it may be challenging to capture assumptions about future management actions that might occur.

• It may promote overreliance on the modelling (i.e. false comfort).

5.5.5 Alternative confidence intervals and alternative shocks parameters

It is immediately clear that these values can have a significant influence on the figures calculated. Whilst long-term institutional investors can freely alter these for internal management information purposes, for external reporting there are only limited instances where the regulations might explicitly allow for variations to parameters such as the confidence interval or the shocks to be applied – subject to stringently restrictive conditions. The following case example illustrates this.

An example: For UK insurers – Duration-based equity risk sub-module

EIOPA’s annual review of long-term guarantee measures report 2017 stated:

This report highlighted that only one undertaking in France has taken up this measure. We believe that this measure has not been taken up in the UK because the main investors in equity holdings are with-profit funds and unit-linked funds.

In respect of retirement benefits, Article 304 of the SII DirectiveFootnote ²³ provides for the application of an equity risk sub-module of the Solvency Capital Requirement, which is calibrated using a Value-at-Risk measure, over a time period, which is consistent with the typical holding period of equity investments for the undertaking concerned. There are a number of stipulations around this, however. In addition to the abovementioned 12 year holding period, all assets and liabilities corresponding to the business must be stringently ring-fenced, without any possibility of transfer.

The SII calibration paper states:

The main reason for EIOPA calibrating a lower stress in the context of a longer holding period is the use of a lower confidence level in the VaR assessment as shown in paragraph 5.5.30 above. The 1 year VaR confidence level for a 12 year holding period is therefore equal to 94.16% (99.5%¹²). There is currently zero take-up in the UK because firms are unable to demonstrate that equities within the portfolio are held on an ongoing basis across the holding period. We believe this is mainly to avoid the risks of lapses or asset switches where the impact of liquidity should be taken into account.

5.5.6 Summary of this chapter

In this chapter we;

• Considered the different aspects of risk measurement approaches

• Discussed how discounting cuts across almost all risk measurement approaches

• Discussed the importance of considering risk measurement approaches that reflect the nature of the liabilities

• Considered and appraised different risk measurement approaches – including both those that are commonly used and those that are less used.