Strategic asset allocation

Document Sample
Strategic asset allocation Powered By Docstoc
					International Social Security Association
International Conference on the Investment
of Social Security Funds
Merida, Mexico, 27-28 September 2005

Strategic asset allocation

Christopher D. Daykin
Government Actuary
Government Actuary's Department
United Kingdom
Strategic asset allocation

Christopher D. Daykin
Government Actuary
Government Actuary's Department
United Kingdom

Investment decision-making is usually considered to be made up of two distinct sets of
activities: strategic asset allocation and tactical investment selection. Strategic asset
allocation is about "the big picture", and sets the direction for the portfolio in terms of its
broad subdivision into asset classes. It is concerned with achieving the overall balance of
types of asset that is appropriate for the broad structure of the liabilities and consistent with
the fulfilment of the objectives of the investment strategy. Tactical investment selection is
concerned with the implementation of an agreed strategic asset allocation through the
selection of individual stocks, financial instruments or other relevant assets within each of
the asset classes in order to meet the objectives set for each asset class.

Strategic asset allocation is an important and integral part of the overall financial governance
of an institution, which should be a major preoccupation of the Investment Committee, where
one has been established, and requires the approval of the Governing Body, at least in
broad terms, and in greater detail if there is no Investment Committee to look after the
detailed articulation of the policy. Whilst value can certainly be added by making good
decisions regarding individual investments at the tactical level, more often than not it is the
strategic asset allocation decisions that make a real difference to the overall performance of
a portfolio and to the ability of the portfolio to perform well in different economic conditions.

Strategic asset allocation can be defined as "the efficient optimization of investment
allocation to major classes of investment, in order to meet the overall investment objectives
of the institution and to achieve an acceptable balance between risk and return".

The overall investment objectives should be agreed by the Governing Body. What is the
purpose of holding investments? Are the investments held to back particular liabilities? Is
there a rate of return objective? What is the attitude to risk? What criteria are to be used for
judging the suitability of individual investments or of the portfolio of investments held in any
particular class or category? The Governing Body should also determine the governance
structure for setting the investment policy and for managing, monitoring and controlling its
implementation. There needs to be complete clarity about who (either an individual or
perhaps a committee) is responsible for what, who is to be held accountable for what and to
whom. Who monitors their performance and how are the results and the evaluation of
performance to be reported?

                                    Christopher D. Daykin

Good corporate governance practice would suggest that the principles of the strategic asset
allocation should be made public to stakeholders, through publishing some sort of
"statement of investment principles".

The purpose of investment
Investments may be held for a variety of different purposes. An insurance company holds
assets to back a specific identifiable portfolio of liabilities. It will have received premiums
from policy-holders in return for a promise of future benefits or to provide an indemnity
against certain risks. The company must therefore set up provisions in respect of the
liabilities and hold appropriate assets to back those liabilities. An insurance company will
also hold further assets as a solvency margin, derived from capital invested by shareholders
and from profits retained from past operations. These assets are not held to back specific
liabilities, although some of them may be regarded as held against the possibility that the
assets held to back the provisions may turn out to be insufficient when the liabilities
eventually mature. Otherwise they can be regarded as assets which should be invested to
give a good return to the shareholders, either through investment in the development of the
business, or through investing to maximize returns.

Some individual assets may be held in order to hedge specific liabilities, since the
characteristics of the asset are the same as or similar to those of the liability and this may
enable the investor to "hedge" the liability and thereby reduce the risks associated with the
liability, since the hedging asset will behave in a similar way to the liability in changing
market conditions. A similar principle may apply in respect of a portfolio of assets which is
constructed to match a particular portfolio of liabilities.

In other cases a portfolio of investments may be held without any particular regard to
liabilities, perhaps because the liabilities are as yet unknown, or are somewhat ill-defined.
For example, an educational institution may seek to build an endowment consisting of a
portfolio of investments. This may be simply putting money aside "for a rainy day" or may be
with a view to being able at some future date to embark on a new initiative, to construct new
facilities or to provide scholarships for needy students. The investment objective might be to
maximize the return, subject to not putting the capital of the endowment at too much risk.

A defined benefit occupational pension fund holds assets that represent the accumulation of
past contributions by employees and by the employer, over and above the cost of benefits
already paid. Future benefit payments will be met by a combination of the proceeds of the
assets in the fund and future contributions from the employees and employer. The assets
are therefore held as part of the overall funding or financing mechanism. Different
investment objectives might be set, according to whether the priority is to try to maximize
returns in order keep the expected need for future contributions as low as possible, or
whether greater priority is to be given to ensuring a high level of security for the accrued
rights of members, so as to meet regulatory funding requirements. Alternatively the strategy
may be to avoid the risk of the employers having to make emergency contributions in order
to eradicate a deficit or the risk of a significant impact on the company balance sheet as a
result of the requirements for accounting for the costs of the pension scheme.

Defined contribution pension funds, including provident funds and privately managed
individual accounts, hold assets which explicitly represent the interests of the members.
When operated on an investment-linked basis, the performance of the investments may
directly impact on the value of the members' individual account. Other contractual structures
may have more of an indirect impact, but the investment returns achieved will still be of

                                    Christopher D. Daykin

essential interest to the member. In principle the investment strategy should be that which is
preferred by the members.

Social security institutions providing defined pension benefits, or operating across a range of
other social security benefits, usually have assets to invest because there has been an
excess of income over outgo. This may have arisen because of a deliberate policy of
equalizing contributions over a period, whereby surpluses could be expected in the early
years and deficits in the later years, so that the accumulated surplus can then be drawn
down to avoid having to increase contributions. Surpluses may also have arisen through
favourable experience, good economic conditions, or because the contribution rates were
set high to be on the safe side, and the margins proved to be unnecessary. In some cases,
the assets may be little more than a working balance, sufficient to even out the variability of
cash flows arising from seasonal and other fluctuations in the contribution income or the
benefit outgo. In other cases reserves will have been built up with specific contingencies in
mind, such as to cope with variations in experience over an economic cycle, or to be able to
cope with catastrophic events.

General principles
Some general principles can be enunciated to guide strategic asset allocation. The first is
that the assets should be appropriate to the liabilities. The assets should ideally have a
similar time horizon to the liabilities and have similar characteristics, for example being
denominated in the same currency. If the liabilities are linked to the retail prices index, or
some other indicator of economic growth, in other words if they are "real" liabilities, as
opposed to being fixed in nominal terms, then ideally the assets should also have such real
characteristics, rather than being fixed.

A second general principle is the need for diversification. This arises because most forms of
investment inherently involve some risk and diversification is an important way to reduce
risk. In practice not all risk can be controlled through diversification. Some risk is "systemic",
which could mean, for example, that it affects, to a greater or lesser extent, all assets in a
class. However, some risk relates to the individual asset, with the value of the asset
depending, for example, on the underlying performance of a particular company or project,
and on the perception of other investors of its worth. This sort of risk is "diversifiable", since,
by holding smaller amounts of a number of assets of the same type, but with different
specific characteristics, instead of a much larger amount of one asset, the impact on the
portfolio of a possible severe loss of value in respect of one asset becomes of much less
significance than if it is the only asset held. Of course, such diversification also reduces the
possibility of making a large profit if the particular asset held performs extremely well. In
general, diversification through having a larger number of holdings reduces the variability of
the outcome.

Another general principle is that higher levels of return are usually associated with higher
levels of risk. This principle is particularly associated with efficient markets, in which an even
more general principle might be considered to be that of no arbitrage, in other words that you
cannot get something for nothing. If two alternative investments had similar prospective
returns and one had a higher level of risk, then investors would all buy the investment with
the lower risk. So in practice prices in a market adjust so that the asset with a lower level of
risk becomes more expensive and so offers a lower prospective return, whilst that with a
higher level of risk becomes less expensive and offers a higher prospective return. Higher
returns can in principle only be achieved by taking a greater degree of risk. Very high risk
investments may offer very substantial potential returns – but they also, of course, come with

                                     Christopher D. Daykin

the possibility of loss, or of failure to achieve the hoped for high returns, because the future
return is inherently uncertain or risky.

It is possible to think of this in simple terms as a "trade-off" between risk and return. By
assuming more risk the investor can increase the expected level of return. But the investor
who wants to minimize risk must be satisfied with a lower expected level of return. In a
simplified characterization of this trade-off, each different asset available for investment can
be plotted on a graph according to its mean rate of return and the variance of its return. Here
variance (standard deviation could equally be used) is taken as a measure of risk, i.e. risk is
envisaged in terms of the spread of the distribution of outcomes. Figure 1 shows such a
graph, with a few possible assets (or combinations of assets) entered according to the mean
and variance of their returns.

Figure 1. Trade-off of risk and return

                                      Efficient frontier

   Mean rate of return

                         4.0%                              C                      D

                         2.0%     A


                            0%   1%       2%        3%           4%      5%           6%
                                            Variance of return

In practice, the points for different types of asset are restricted to a particular area of the
graph, since, for any particular mean rate of return there is a minimum achievable variance
of return, and for any particular variance of return, there is a maximum achievable mean rate
of return. The line which encompasses all the possible combinations of mean rate of return
and variance of rate of return is known as the efficient frontier. A good principle for getting
the best from the trade-off of risk and return is to try to identify assets (or combinations of
assets) which result in the portfolio being at, or as close as possible to, the efficient frontier.
At such a point the mean rate of return has been maximized for a given level of variance of
return. So, for example, starting with Portfolio B, it should be possible to change the mix of
the portfolio towards more low risk assets in order to move the position of B along the arrow
shown, as far as the efficient frontier. This would maintain the same level of mean return,
whilst reducing risk. If the level of risk of Portfolio B was acceptable, the portfolio could be
made more efficient, at the same level of risk, by changing the balance of the portfolio to
move it vertically upwards to the efficient frontier.

This representation is very powerful but it is also simplistic. Asset risk cannot be measured
by variance alone. Even if risk was just a function of the distribution of outcomes, variance is

                                          Christopher D. Daykin

only one such measure, and for a proper understanding of most sorts of distribution it is
necessary to consider also the higher moments (e.g. skewness and kurtosis). However,
another more fundamental problem is that neither the mean nor the distribution of the rate of
return is known and, even if it could be deduced for any given past period, the distribution in
future may be quite different. In other words, considerable uncertainty is an inherent
component of the risk, which cannot therefore be fully described by a well-behaved statistical
distribution. Estimates of the characteristics of the distribution from past behaviour (and
changes to those characteristics over time) may be useful indicators but they are not an
adequate guide to the future.

Another important limitation of the approach illustrated in Figure 1 is that it usually focuses
on the return over a short period of measurement (e.g. a year) and measures return on the
basis of change in market value (together with any income received). However, it does not
take into account the first principle of investment, namely that the assets should be
considered in relation to the liabilities for which they are held. Depending on the liabilities
against which the assets are held, volatility in the return over a given short period may or
may not be an issue for the investor. If the assets and the liabilities are perfectly matched,
fluctuations in market value may be of complete irrelevance. The risks that are of interest are
not volatility of market value but more fundamental aspects of the assets, such as default

Moreover the basic theory of optimizing risk and return, as evidenced by Figure 1, assumes
that all investors are the same, with the same utility functions and time horizons. This is
clearly not the case in practice. The vitality of investment markets is maintained by a variety
of different sorts of investors, with different liability and risk profiles, time horizons, tax
positions, hedging requirements and behavioural characteristics.

Liability characteristics
The characteristics of the liabilities are defined in terms of the projected amounts and the
timing of each payment. Liability amounts may be uncertain as to amount or as to timing, or
possibly as to both. However, for the purposes of developing a strategic asset allocation,
estimates will need to be made of the distribution of liabilities over time, in other words the
target cash flow which the assets are intended to meet. The liability amounts may be
expressed in nominal amounts, or they may be in real terms (for example at constant
prices). They will be in a defined currency. A useful metric for characterizing the stream of
liabilities may be the weighted mean duration.

Particular attention needs to be paid to any embedded options, in other words liabilities that
may be triggered by particular events or particular combinations of factors. Embedded
options often have potentially dangerous characteristics and may present greater difficulties
for finding appropriate matching assets. However, in recent years a wide variety of derivative
instruments has become available and in some markets it is possible to find assets that are
suitable for hedging many such options.

Asset characteristics
Most assets have a capital value, which can be realized by selling them, or which, in certain
cases, is automatically realized when they reach their maturity date. Most assets also
produce a stream of income, which may be determined in advance or may be uncertain and
depend on factors which will only unfold over time. There are some types of asset which do

                                    Christopher D. Daykin

not generate any income and there are also, of course, occasions when an asset may have
no value. The expected return (and distribution of returns) from any asset will usually be a
combination of expected income stream and the maturity value, or, if the asset has no
maturity date as such, the amount for which the asset can be sold when it comes to be

The simplest types of assets to analyze, and to set against specific liability streams, are
those with pre-defined payments of interest and return of capital, such as fixed interest
bonds and some types of mortgage. Bonds can also be structured with payments in real
terms, linked to the movements in the consumer prices index or some other index of the
economy. Such index-linked bonds can be useful for holding against liabilities which are
linked to prices or earnings growth. Bonds and mortgages have different matching
characteristics according to the level of the interest payments, so that they can have a
heavier weight of their overall return in the interest stream or in the capital at maturity (or on
sale). At the extreme, bonds can be structured (or markets can make available "stripping")
so that they represent only a stream of income, or only a maturity payment (this last type of
bond is known as a zero-coupon bond).

Apart from the payment streams, there are other important characteristics to take into
account. Perhaps foremost of these is the default risk. Are the payments absolutely certain
or is there a possibility that the borrower may default on the promised payments? Bonds
issued by sovereign governments generally have less risk of default than those issued by
corporate entities, and those issued by corporate entities will have a wide range of default
risk, depending on the underlying strength and covenant of the issuer. The market responds
to this by expecting a higher return on securities with a higher risk of default. There are
rating agencies which specialize in analyzing companies' financial strength and issuing
grades as an indicator of the potential for default. Companies with lower rating will have to
pay more to issue debt and the secondary market in corporate debt, where it exists, will also
reflect this requirement of investors for higher return to compensate for higher risk. In the
case of the secondary market, the yields will, of course, reflect the latest available
assessment of default risk, rather than that which was available on original issue. In practice
even sovereign governments can sometimes find themselves unable to repay debt on time
and defaults of government bonds are certainly not unknown. Again, markets respond to this
by expecting higher returns on bonds issued by governments whose security is not seen as
absolute (the term "gilt-edged" is used to describe government bonds in the United

A further characteristic of fixed income securities is their degree of marketability. Many
countries have an active secondary market in bonds, both for those issued by government
and those issued by corporate entities. There are often many sorts of intermediate types of
bond, such as those issued by local municipalities, city or regional governments, those
issued by public utility companies and those issued by other special public or semi-public
entities, such as housing finance corporations. To the extent that these have an underlying
government guarantee, their security may not be very different from that relating to
government bonds. However, there may be much less of an active secondary market in such
instruments. In most countries the corporate bond market is not so deep and is less liquid
than the government bond market. However, even government bonds may not be actively
traded in many jurisdictions, with most investors holding the securities from issue to maturity.
Lack of marketability is normally compensated for by higher returns. Investors who do not
require liquidity and are happy to be locked into particular investments can enjoy higher
returns to compensate. Those who require greater liquidity and want to be able to sell assets

                                     Christopher D. Daykin

when required to meet liability cash-flows, will pay a price for that flexibility and will enjoy
lower yields.

There is a wide range of types of asset which do not guarantee in advance the payments
that are to be made. Some of these do provide a guarantee of the invested capital, which will
be returned on maturity, or, in some cases, can be withdrawn at will, or subject to giving
appropriate notice. Most bank deposits and many mortgages are in this category. The
interest payable will vary from time to time, according to interest rates in the wholesale
interbank market, or rates set by a central monetary authority. Interest rates will usually be
higher if the money is locked in to some extent rather than available on demand. However,
interest rates can also be high on very short term investment, for example overnight, since
no guarantees are being given about what rates will be available in future.

Another very large class of investment is made of various types of participation in the
ownership of companies and other corporate entities. Securities known as equities, shares
or stocks enable the investor to invest directly in the fortunes of the issuing entity, through
taking a stake of ownership. In return the investor will be entitled to receive dividends or
other forms of pay-out, depending on the issuer's profitability and financial development.
Such investments carry significant default potential, both because the issuer may not be able
to afford to pay dividends at some stage in the future, but also, more fundamentally, because
the issuer could get into financial difficulties or even into bankruptcy. Investors in shares in
the company usually rank below creditors and bondholders and may find that their shares
are worth little or nothing in the event of the company failing.

Shares of companies are usually traded on secondary exchanges (although it is also
possible to buy shares of many small companies that are held privately and are not traded
on exchanges). The value of the shares depends on market sentiment, both generally with
regard to the economy and its prospects and also specifically in relation to the issuing
company and the market's perception of its future earnings potential, dividend policy,
financial strength, potential as a target for takeover, etc. Equity investment can be seen as
investment in the wealth of the economy. Diversification is usually seen as essential for a
sound strategic investment policy, in order to spread risk not only over a number of different
companies but also over different sectors of the economy and even over different countries.

A further key area of investment is in land and property (real estate in North American
terminology). This can be done through fixed interest or floating interest vehicles such as
mortgages but also directly through the purchase of freehold land and buildings and through
various forms of shared ownership, such as leasehold. Property investment may generate a
stream of income from users of the property or from rents and leases. Some income streams
may be fixed for periods but are usually subject to review, so that they can be expected to
increase broadly with inflation. Property values depend indirectly on the performance of the
economy, but also have trends, cycles and fluctuations of their own, which may vary
according to the location of the property and the type of property. Rental income is
potentially subject to default risk, since the lessee may become unable to meet the
payments, and the property may not be able to be let. Underlying property values can also
be subject to significant adverse shocks, for example from a downturn in the economy in the
area where the property is situated, a disaster affecting particular properties (such as an
earthquake or subsidence) or a severe fall in demand (or excess of supply) of a particular
type of property, or more generally because of recession.

This is not intended to be an exhaustive discussion of types of asset that may be available
for investors, and in particular for social security institutions as investors. However, brief

                                    Christopher D. Daykin

mention should also be made of derivatives, which often have very different risk
characteristics from more traditional investments. Derivatives include options, futures,
warrants, swaps and a variety of "over-the-counter" specially designed structured products
designed to exhibit the particular characteristics required by investors. Many derivative
products can be highly risky as investments, with the potential for losing the whole of the
amount invested (and sometimes much more). Because of their high risk nature, they are
not usually considered suitable as stand-alone investments for institutional investors with
fiduciary responsibilities (such as pension funds and social security institutions), but they
may still play a part in strategic asset allocation. For example, interest rate swaps can be
used effectively to transform investment in floating rate investments into a fixed rate
investment (or the other way round). Currency swaps can be used to facilitate investment in
a wider range of asset classes internationally, but without assuming the associated currency
risk. Options can be used to reduce the risk of investing in equities, or to hedge against
particular types of guarantee inherent in the liability structure of the investing institution.
More detailed consideration of the use of complex financial instruments is beyond the scope
of this paper, but suffice it to say that such investments may have a proper part to play within
the investment strategy of even a quite prudent investing institution.

Other factors affecting asset allocation
In practice the strategic asset allocation choices may be limited by other factors. There may
be regulations governing the investment of the social security institution which limit or
prevent investment in particular types of asset. Whilst this could be intended as an
overarching asset allocation decision, it may also be for political reasons unconnected with
optimizing the risk/return relationship of the portfolio, for example to require investment in
government bonds to assist in funding the public sector deficit, or to forbid investment in
private companies in order to avoid creating a situation of back-door government ownership
of private capital.

Another type of constraint may be effectively imposed by the accounting framework. If
assets are required to be accounted for at historic cost, this may be a disincentive to invest
in equities, since the capital gains would not appear as income in the accounts. If assets are
required to be marked to market, this may also discourage investment in equities if it is
feared that there will be a political backlash if the value of the investments falls, as it can be
expected to do from time to time with a relatively volatile class of asset.

Determining the strategic asset allocation
There are many approaches that can be adopted to determining the strategic asset
allocation. At its crudest, limits or ranges of proportions in different asset categories may be
laid down in the law or in the governing regulations. These may not have any sophisticated
justification but their effect is absolute.

A more scientific approach, where the Governing Body has a wide measure of investment
discretion, is to seek to identify specific objectives for which assets are held, or to quantify
liabilities against which they are held, and to hypothecate, or notionally allocate, blocks of
assets to different objectives or different sets of liabilities. The amount and timing of the
specified liabilities, or the characteristics of the objective, should suggest particular types of
asset as suitable (and other types as unsuitable), so that a broad allocation of asset types
can be determined for each block.

                                     Christopher D. Daykin

Given a set of liabilities, expressed in terms of projected cash out-flows in each successive
period of the projection, it may be possible to determine a set of matching assets, which are
expected to produce exactly corresponding inward cash flows in each period. If the flows are
exactly matched, this would offer a very robust investment strategy, where the balance of
assets and liabilities will not be affected, even if there are significant changes in the market
values of the securities held.

However, in practice it is rarely possible to achieve such a strong degree of matching. The
liability cash-flows are often uncertain, since they depend on many different contingent
events and on economic factors such as inflation, salary increases and economic activity.
The available types of assets may also offer uncertain income streams, because of the
possibility of default, and, in most jurisdictions, will not be available in sufficient variety to
make matching a real possibility. For example, the maximum term to maturity of bonds may
be too short to permit matching of the liability cash-flows, or there may be no index-linked
securities (with income and maturity proceeds linked to the Consumer Prices Index (CPI) or
similar), or very limited availability, whilst the liabilities are substantially linked to prices or
even earnings.

To the extent that the liabilities depend on future wages and salaries, it is unlikely that there
will be any suitable matching assets. However, relatively close matching may sometimes be
achievable for a liability portfolio of pensions in payment, if these are either fixed in money
terms, or are indexed to the CPI and index-linked bonds are available. Even if perfect
matching is difficult or impossible, techniques are available for immunizing such a portfolio,
so as to make the asset/liability relationship reasonably robust over a significant range of
possible market outcomes.

If a close matching or immunized strategy exists, this does not mean that such a strategy is
optimal. It may be low risk, in terms of the chance that the allocated assets will prove
insufficient to meet the relevant liabilities, but it may be a very expensive strategy, if the
appropriate matching assets are in short supply or are very highly priced in the market. It
may be thought that the liabilities can be financed more cheaply by adopting a less closely
matched strategy or by investing in a wider range of types of asset that do not offer
particularly close matching characteristics.

One of the problems of this sort of line of reasoning is that it is seen by some as running
counter to one of the fundamental precepts of financial economics, the principle of no
arbitrage. In simple terms this would maintain that there is no real gain from investing in
more risky asset classes, since the apparent additional return (or cheaper price), is offset by
the additional risk taken on. So, for example, corporate bonds may offer a higher yield than
corresponding government bonds with similar characteristics of coupon and maturity.
However, the higher yield is, broadly speaking, the market's assessment of the
compensation needed to offset the possible risk of default. In practice, of course, the
corporate bonds may not default and the investors will achieve a higher overall return than
they would have obtained from investing in the alternative government bond portfolio. Or the
level of default may prove to be unusually high, if the economy is hit by a severe recession,
in which case the overall return on the corporate bond portfolio may turn out to be much
lower than would have been obtained by investing in government bonds. Only if the default
rate corresponds closely to the assumptions underlying the market pricing at purchase will
the two alternative investment strategies yield broadly the same overall return. However,
whilst the outcomes may differ, it is not possible to know in advance that one choice of
investments will be better than the other – many financial economists would argue that the
only reliable assessment is that provided by the market price and that it is impossible for

                                      Christopher D. Daykin

individual investors to beat the market systematically by believing that they know better, and
hence "enjoy a free lunch".

Although the simplicity of this argument is appealing, it depends on some fairly strong
assumptions about the uniformity of risk appetites, utility functions and investment horizons
and strategies. In practice the market is made up of many different sorts of investors with
different investment requirements. A simple example is the case of higher returns for more
illiquid investments. Here, assuming similar default risk, the only risk for which the investor is
being compensated is the risk that they will not be able to get their money back when they
want it. For many long term investors with a diversified portfolio this is not a concern and
they are happy to have the higher return and accept that part of their portfolio is held to

Recent developments in behavioural economics have begun to explore the dynamics of this
more complex real investment world. However, the bottom line is that most investors still
believe that for much of the time the market does offer opportunities for those willing to
assume greater risk to make returns which will more than compensate for that risk, as seen
from their particular perspective. This can be particularly the case if they are investing over
long time horizons and have little sensitivity to short term price movements, especially with
larger portfolios where frequent short-term tactical adjustment of the portfolio is not a
practical proposition.

Asset/liability modelling
Given that the institution accepts the need for diversification by asset class and that there
are higher expected returns to be had from investing in some asset classes that are more
risky, a variety of modelling tools have been developed to inform the strategic asset
allocation decision-making process and to help to arrive at an optimal approach. This activity
is known as asset/liability modelling or ALM (which can also be used to stand for the more
generalized concept of asset/liability management). In essence it starts from identifying the
target liability cash-flows. These can be deterministic, i.e. a single "certain" projection of the
future outflows period by period (or maybe a set of scenarios). They can be defined as a
function of an economic index, typically the CPI. Or they can be generated stochastically
with one or more random variables.

On the asset side a limited number of major asset categories will be identified as candidates
for the portfolio. For each of these a model is required to generate future income streams
and capital values. These models could be static distributions, from which a sample can be
drawn for each period of the projection. However, more sophisticated ALM will normally use
time series modelling, with varying degrees of complexity, allowing also for interactions to be
modelled between investment categories and between the investments and the economic
indicators which are thought to influence the liabilities. The investment models must be
capable of reproducing behaviour similar to that observed for each asset class, in terms of
expected returns, variability and higher moments, as well as reflecting the correlations
between the behaviour of different asset classes, which can be a critical element in the
analysis of the effects of diversification. However, this science is still at a relatively early
stage of development and there are no generally accepted models for most asset classes,
even in the most sophisticated markets.

The next stage is to simulate a large number of outcomes from the stochastic distributions
and time series. This produces a distribution of outcome scenarios. Because of the
potentially very large number of items of information, it is usual to focus on some particular

                                     Christopher D. Daykin

indicators and seek to define what are regarded as desirable characteristics to optimize for
these. Starting from a few significantly different strategic asset allocations for the portfolio,
relevant behaviours can be explored and then combinations of portfolios can be developed
which provide more optimal results.

In most cases there is no unique optimal solution to any particular strategic objective. Some
of the theoretically desirable solutions may be difficult to achieve in practice because of
limited availability of certain types of asset. Moreover, the results are usually highly
dependent on the model structure, and particularly on how the model interprets future
variability and uncertainty and the correlations between the returns on different asset
classes. The assumptions are critical and inevitably depend on judgement, since it is not
possible simply to use historic parameters for projecting into the future, both because the
past has not been homogeneous and hence the parameters derived will vary according to
the period chosen, and because the future is unlikely to be exactly like any particular past

All of this may suggest that the exercise is of limited value and that some healthy common
sense may achieve equally satisfactory results in terms of selecting a strategic asset
allocation. However, a well-designed asset-liability modelling exercise will provide some
useful insights into the sorts of combinations of asset classes that will yield good expected
returns for acceptable levels of variability and robustness in the face of more adverse
scenarios. It is particularly difficult to interpret the effect of correlations and combinations of
risk and return without a formal model structure. When adding additional assets (or a class
of assets) to the portfolio, the effect does not necessarily correspond to the riskiness of
those assets standing alone, but on how they interact with the rest of the portfolio in affecting
the overall risk level. Sensitivity analysis should be carried out on the assumptions and even
the fundamentals of the model structure, in order to be satisfied that the asset allocations
that seem attractive are in fact reasonably robust to the assumptions.

For some sets of liabilities it may be possible to identify fairly clear optimal asset allocations
in terms of minimizing the risk of failure to meet the objectives. As indicated earlier, this may
turn out to be a relatively expensive strategy and, if it produces a narrow range of outcomes,
will reduce any chance of out-performance as well as reducing the possibility of failure. In all
cases judgement will be required to decide on where to pitch the selected asset allocation

Having developed a first cut of the overall asset allocation strategy, separate modelling
exercises might be carried out for particular asset classes, in order to develop optimal
strategies within the class, for example in relation to duration, marketability and default risk
in the portfolio of bonds or in relation to different types of land and property.

Once a strategic asset allocation decision has been taken, procedures need to be put in
place to implement the strategy and to give mandates to one or more investment managers
to handle the tactical asset allocation within each asset class. Since asset managers are
typically skilled only in one particular asset class, it is normal for a number of different
mandates to be awarded to different investment managers, each being allocated their
portion of the overall portfolio and being given the task of investing within the criteria for the
asset class.

                                     Christopher D. Daykin

There needs to be a good process for monitoring the performance of the different asset
managers, usually by the Investment Committee, or by the Governing Body if no separate
Investment Committee has been established. An appropriate benchmark needs to be
decided for each asset class and a formal performance objective given to the asset
manager. This might be in terms of meeting or beating the benchmark, and may also include
risk criteria, so as to discourage the asset manager from taking too much risk, or to give
some guidance as to how much risk it is acceptable for them to take in order to seek to
achieve the performance objective. The objectives need to be set with a clear idea of what is
to be expected of the asset manager. Are they expected to invest so as to replicate the index
which forms the benchmark, for example, and to run their part of the portfolio as a "tracker
fund"? In that case they are unlikely to be able to beat the index, but nor should they fall
behind to any significant extent. If the intention is that they should adopt more risky positions
in order to have some chance of significantly outperforming the benchmark index, then it
may be desirable to make clear how aggressive a strategy would be acceptable.

Investment decision-making is made up of two distinct sets of activities: strategic asset
allocation and tactical investment selection. Strategic asset allocation is about "the big
picture", and sets the direction for the portfolio in terms of its broad subdivision into asset
classes. Strategic asset allocation should be a major preoccupation of the Investment
Committee or the Governing Body. Strategic asset allocation decisions often make a real
difference to the overall performance of a portfolio and to the ability of the portfolio to
perform well in different economic conditions.

Strategic asset allocation needs to be based on a thorough understanding of the purpose of
investment and an appreciation of the characteristics of the liabilities and of the various
types of assets that might be held in the portfolio. A basic objective is to invest in assets that
are appropriate for the liabilities and optimize the trade-off of risk and return in order to
achieve satisfactory return outcomes whilst avoiding undue risk in relation to the liabilities.

Simple approaches to strategic asset allocation may involve notionally segmenting the
liabilities and considering appropriate assets for each segment. More sophisticated
asset/liability modelling techniques may also be used in order to understand better the
correlations between different asset classes and the benefits to be gained from investing in
non-correlated assets, as well as to model the interaction between the assets and the
liabilities and obtain a better fit overall, which will also be robust to significantly different
possible outcomes.

Once the overall strategic asset allocation decisions have been taken, more detailed asset
allocation structures can be developed within particular asset classes. It is good corporate
governance practice to publish the "statement of investment principles".

Within the overall framework of the strategic asset allocation, investment managers should
be appointed with a mandate to invest in each of the major asset classes. There should be a
good process for monitoring the performance of the different asset managers, with
appropriate benchmarks being decided for each asset class and a formal performance
objective given to the asset manager.

Getting the strategic asset allocation right is a key challenge for the governance of a social
security institution that has been entrusted with significant assets to help in the creation and
maintenance of a sustainable social security system in the country.

                                     Christopher D. Daykin

Shared By: