What does ‘risk’ really mean in asset allocation?

Select Asset Management’s Dominic McCormick says it’s time to reduce reliance on simplistic risk categories in asset allocation.

The investment industry has long become conditioned to the view that the “risk” of a fund or portfolio can be readily defined by its broad asset allocation across the major five or six asset classes.

In reality, the process is even more simplistic than this, first putting each of those major asset classes into either the “growth” or income (non-growth) bucket and, depending on the growth/income split, allocating the fund/portfolio into one of four to six risk categories – sometimes more – typically ranging from a cautious/conservative mix to a high growth/aggressive mix.

This is supposed to give investors and financial advisers a good guide to the riskiness of the fund/portfolio. 

A couple of decades ago this sort of simplistic categorisation made sense.

VIEW ALL

View all

Firstly, the investment options available both across and within each asset class were quite limited – and mostly vanilla flavoured – as long only, benchmark-relative exposure dominated the fund offerings available, and was the main exposure of diversified funds and portfolios.

Secondly, a strategic asset allocation (SAA) approach to building portfolios was accepted as the standard, which fitted in well with the need for near constant allocations required by the above definition approach.

Today however, this approach to defining the broad risk of a diversified fund or portfolio is becoming increasingly flawed in a much more complex environment. It has failed to keep up with the expanding universe of strategies and styles “within” asset classes, new asset classes/strategies, and also the more dynamic way asset and strategy allocation is being managed at the portfolio level.

For example, let’s look at the “equities” growth asset class.

Today, one’s equities allocation may not just be a long only, index-relative exposure or a mix of value and growth managers, but could also contain long/short, market neutral, buy write, absolute return, low beta/low volatility, income-focused and quality.

These styles and strategies have very different risk characteristics.

The point is, depending on the mix of these styles and strategies, the “risk” of the equity allocation (in terms of volatility, possible drawdowns etc) can be very different – and often significantly lower – than the broad market index.

All equity allocations certainly should not be considered to be equal.

For example, what is the proper defined “asset allocation” of an unleveraged equity fund that is 85 per cent long equities, has 15 per cent cash and 25 per cent shorts.

Is it 100 per cent equities because it is an equity fund, is it 85 per cent equities 15 per cent cash because of its long equities and cash position, or is it 60 per cent equities 40 per cent cash because that reflects its net market exposure. 

Long/short funds that also borrow – and have widely variable gross and net market exposure – are even more difficult to categorise. Twenty years ago no such debates were necessary.

At the supposed other “income” end of the spectrum, fixed interest used to consist of mainly government and semi-government bonds paying attractive income in nominal and real terms. In the mid-1990s, for example, corporate credit made up less than 5 per cent of the UBS Composite Bond Index.

Today, we have a world where traditional sovereign bonds are paying very low yields (often below inflation and sometimes paying next to no yield at all) and are therefore arguably much riskier in that losses/the chance of not meeting certain objectives has increased.

Meanwhile, in response, investors’ fixed interest component is increasingly ‘juiced up’ with assets up the risk spectrum – high yield, hybrids, emerging market debt etc, many of which have risks equal or even higher than some of the ‘lower risk’ equity strategies and styles mentioned above.

I am certainly not arguing against including this broader range of fixed interest offerings in portfolios but, in some cases, the combination of poor value in government bonds and excessive exposure up the credit risk spectrum is making some fixed interest portfolios look as risky, or even more risky, than some well diversified equity portfolios.

Remember, many investments lumped into the fixed interest asset spectrum – finance company debentures, higher-risk mortgage funds and various high-yield funds – lost between 50-100 per cent of investors’ money over recent years, much more than most equity funds that have since largely recovered.

Then there is property and infrastructure. These typically get lumped 100 per cent in the growth asset category yet can sometimes offer risk closer to or lower than standard fixed interest (especially if fixed interest is overvalued).

Take Australian Real Estate Investment Trusts (AREITs), for example. No doubt pre-2007 they deserved to be classified as growth/risky investments.

Dividends were funded by increasing the already high level of gearing and volatile property development earnings, and the asset quality declined and became more risky, particularly in the case of some of the overseas trusts.

The shares were trading at substantial premiums to NTA (net tangible assets), with yields less than 10-year bonds, meaning that the valuation risk was considerable, although many investors and financial advisers at the time saw them as safe.

In contrast, post-GFC (albeit after dramatic capital losses and dilutionary equity refinancing) gearing levels came down, managers focused again on rent collection, prices were again at significant discounts to NTA and yields moved above bond rates (albeit partly as bond rates fell).

Arguably they became low risk/defensive investments again, and recent low price volatility suggests the market has finally woken up to this.

Of course the “chase for yield” which has led to recent good performance and lower risk of AREITs may eventually result in higher valuations that dramatically increase risk.

Risk is not constant through time and the perception of risk by investors and financial advisers is often the exact opposite of the real risk at a point in time.

Importantly, it is obvious that a static categorisation of assets into a particular risk category makes no sense in today’s complex and dynamic world. 

The growth of asset classes, and particularly the alternatives investment bucket, has created even more challenges.

Despite the large universe and the low volatility and correlation of some alternative offerings, they too usually all are lumped into the “growth” category despite some alternatives being low risk on a standalone basis, or being uncorrelated and thus helping to reduce the risk of a total portfolio.

Clearly, categorisation of assets and risk has failed to keep up with the additional investment options in this area.

Then there are the changes in the way asset allocation is being approached and implemented over recent years.

The standard system of categorisation was built in a world where SAA was universally accepted and ranges around these allocations were small and rarely used.

In today’s more dynamic world, asset class ranges have been expanded, are being more utilised and increasingly SAA has been abandoned.

Yet today’s outdated categorisation approach often assesses such funds by requiring a SAA or looking at what their maximum growth allocation could be (while also incorporating the flawed assumptions in the division of “growth” and “defensive” components discussed above). 

Clearly, this approach is almost certainly overstating the risks of more flexible diversified funds and portfolios.

Most funds and portfolios operating a more dynamic approach to asset allocation are driven primarily from a valuation/contrarian perspective, meaning that they tend to lift exposure to “growth assets” when they are attractive value (and therefore lower risk) in absolute and relative terms, and decrease them when they are expensive (higher risk) from a valuation perspective.

Dynamic asset allocation, practiced appropriately, is therefore largely about reducing risk, something today’s simplistic categorisation totally fails to capture. 

So the problem is that a classification system that made sense 20 years ago makes little sense today.

Simply knowing the percentage currently or potentially in simplistically defined “growth” assets may sometimes be a very poor guide to the actual risk of a portfolio, depending on how each asset component is managed and how overall asset allocation is managed. 

So what are the alternatives to this approach?

Some parts of the institutional investment industry have already moved on from focusing on standard asset allocation in assessing risk and building portfolios.

Instead they focus on various risk factors and risk premiums and attempt to build portfolios that diversify across many of these lowly or uncorrelated risk factors with the aim to build a truly diversified portfolio. Standard asset allocation is largely ignored.

However, this risk factor approach is a complex one to understand and implement and I don’t see it as a replacement for retail investors in particular.

Indeed, one does not need to completely throw out asset allocation to build in some of the benefits of better diversification across risk factors: it just needs better understanding of the different strategies and styles across and within asset classes. 

Further, I am not suggesting the industry move to excessive reliance on some of the simplistic quantitative risk measures being proposed by parts of the industry.

The Australian Prudential Regulation Authority has proposed a simple standard risk measure for superannuation funds, defined as the likely number of negative annual returns over a 20-year period.

To go through the flaws in this approach would require an article on its own. Suffice to say that I see it as no improvement (and perhaps even a regression) on the simplistic asset categorisation approach.

There are no simple alternative approaches, but it doesn’t make sense to stick with a flawed approach simply because there is no easy replacement. This is the key point.

Some aspects of investments are inherently complex and there are always plenty of grey areas - concepts like risk can only be properly defined from a number of perspectives, not just one. 

However, I do believe there are steps one can take to be in a position to better manage and communicate asset allocation and risk: 

• Fewer “asset” or “risk” categories. Having five or six categories creates a high degree of implied “precision” with which those categories are defining risk. Fewer categories goes some way in alerting investors to the flaws in that categorisation approach while still giving investors some broad guidance as to the types of assets and risks that will be targeted.
• For example, I have long believed that the “diversified” portfolios of long-term investors can be put into as few as just two categories - defensive and growth (although the labels could be different). Ideally, these mixes should be dynamically managed from a valuation/contrarian perspective, and with an investment time horizon of at least three-to five-years.
• I do believe that shorter-term investors – or longer-term retirees with cash flow requirements – do need another cash/TD/liquidity “bucket”, and at the other end of the spectrum very aggressive investors could also have some exposure to an “all-equity/possibly leveraged” bucket. Note however that these two “buckets” are by nature not “diversified” portfolios. 
• In practice, some combination of one or both of the two diversified portfolios and some in the “cash” or “all-equity” bucket can meet the needs of the vast majority of investors. 
• Recognise that the risk of even a static asset allocation varies over time, given changing valuation and structure of asset classes – which is why a flexible/dynamic asset allocation process is appropriate. Valuation factors, for example, can be a much more important guide to the current risk of an asset class than backward-looking measures of risk such as volatility and drawdowns.
• By all means still look at these quantitative measures of risk over time – including volatility, drawdowns, etc – but remember they are based on history, and don’t rely excessively on any one “risk number”.
• Advisers and/or investors have to understand better the way asset classes and asset allocation is managed, in order to assess risk in a subjective way. Understand the different styles/strategies of funds that make up the various asset categories, and whether this results in an overall mix that is more or less risky than the asset class overall. 

Incorporating these elements does not offer the degree of simplicity and precision implied by fitting into a series of specific asset allocation categories or a certain numerical measures of risk.

However, it’s time for investors and advisers to recognise that the implied precision in current categories and risk measures is illusory anyway.

The industry needs to fully recognise the severe limitations in a categorisation approach that has failed to keep up with developments in the complexity and breadth of asset classes and more dynamic approaches to asset allocation and portfolio construction. 

Advisers, researchers and investors need to seek greater understanding of a range of aspects of a fund or portfolio to assess risk – rather than seek some sort of catch-all, simplistic category or measure.

This approach is clearly more complex and involves more work from all participants, but that is the world we live in.

As Albert Einstein said, “Things should be made as simple as possible, but not simpler”.

Today’s archaic and “over-relied-upon” fund categorisation system has long since failed the “not simpler” test.

Dominic McCormick is the chief investment officer of Select Asset Management.