FA news - Digging deep to find value

FA news - Digging deep to find value

by Malcolm Holmes

The face of the South African listed property market has changed markedly over the past 16 years.

Not only has the number of collective investment schemes investing in property risen dramatically (to the current 43), but the key drivers of the listed property market have changed as well. This has made it harder for investors to understand what they are buying.


In 2002, South Africa’s property sector consisted of a few locallyfocused, illiquid shares with a total market cap of about R10bn and an income yield of around 16%. Today, the market cap of the sector is more than R500 billion, there are over 40 different property shares, liquidity has increased tenfold and the income yield of the sector is closer to 9%.

Today, many listed property companies own a mix of retail, commercial and office exposure. Further, they have expanded into Europe, the UK, US and Australia. Several offshore property companies like Hammerson PLC have secondary listings in South Africa. Cross-ownership adds to the complexity, as shown in the Resilient stable of companies.

This complexity makes it difficult for the average investor to differentiate between the companies and the collective investment scheme (CIS) products that invest in them. Unlike the broader equity market, there are no sector classifications to help distinguish between size, sector, sub economic industries and geographic exposure in the property space.

Using only the quarterly fund factsheet will not help investors to understand how or why one fund will perform relative to its competitors if the rand strengthens, Brexit is concluded successfully, local interest rates rise, or the issues relating to the Resilient group are positively resolved.


In 2018, in response to the introduction of the FTSE / JSE All Property Index, STANLIB Multi-Manager completed an exercise to classify the different property offerings. The results were extremely interesting.

We defined a series of categories, which we call buckets, into which we put property shares with similar qualitative characteristics. These similarities were validated using quantitative methods, giving us a high degree of confidence, not only in the primary characteristic of each bucket but also in the constituents.

We also separated the benchmark into the same buckets which allowed us to compare it with a property CIS.

Using this process, we have identified the following buckets or groupings of shares.

Based on these buckets, we highlight the market cap-weighted split of the benchmark FTSE / JSE All Property Index at the end of December 2018.


Creating an equally-weighted index of the companies in each of these buckets, we tracked the historic performance of the buckets and compared it with the market. The following chart shows how they performed against the All Property Index between September 2017 and 31 December 2018.

This performance data and the holdings per bucket of a property CIS were compared with the benchmark. A manager’s alpha can be analysed by breaking down the contribution from each bucket. Positive alpha can come from being overweight in a better performing bucket or through good stock selection within that bucket.

In the chart below, we highlight the relative positioning of one of the STANLIB Multi-Manager Property funds, together with the resulting attribution for the quarter ended 31 December 2018.

This analysis shows the collective wisdom of the selected underlying managers of this fund was to underweight the large South African hybrid property companies and overweight the small and mid-South African hybrids.

The fund was underweight South African Inc – possibly worried about a weak local economy and its impact on property fundamentals – and underweight the UK PLC’s, which possibly reflects managers’ concerns about the effect of Brexit on UK property.

These positions either excelled or lagged relative to the markets, reflected in the alpha column. It shows the largest contributor to the fund’s 0.97% positive alpha during the quarter was its underweighting in the UK PLC’s, which were down 16% during the period – adding 0.5% to the fund’s total alpha.


Given the increased diversity of the listed property market in South Africa and the lack of neat sector breakdowns, we need better techniques to understand how managers build property portfolios and what drives their returns.

Refining these categories is part of the multi-manager process and we believe they will provide unusual and valuable insights in building multi-manager solutions for investors as the sector continues to evolve.

Goal-based portfolio construction

Goal-based portfolio construction

“Your client asks you to build a goal-based portfolio that can achieve CPI+5% per annum over-rolling six year terms (which is part of a range), how do you build it?”. – Malcolm Holmes

This is often the question I ask the new portfolio managers in the team as a way of opening their minds to various possibilities. It is indeed a tricky question, because there is no one right way, but rather there are several construction methodologies that could be used, each with their own pro’s and con’s, but they all generally follow the same principle – get your client invested in an optimal combination of different asset classes that over time have the greatest probability of achieving your objective.

This edition of Mindset covers many interesting topics related to goal-based investing. Our focus in this particular article is the portfolio construction process and the considerations that go into building a goal-based portfolio aiming to achieve CPI+5% p.a. over the long-term. 

Investor understanding

Financial advisers and their clients need to understand that this does not imply CPI+5% p.a. every year. Some years will be better than others depending on the level of inflation and the returns from various capital markets, but a well-constructed portfolio should produce CPI+5% p.a. on average over the long-term. It is critical to understand that this is not merely wishful thinking, but fundamental to how assets are priced and hence how returns are derived. The reason that returns don not progress neatly from day to day, month to month, and year to year, is that many factors will influence their price over the short to medium term (and even over the longer-term).

This creates a potentially big issue for investors who happen to be experiencing a particularly bad patch of capital market returns, especially when this persists for longer periods of time as they are less likely to achieve the CPI+5% p.a. objective. At the other extreme, you will have an investor who is fortunate enough to have invested before a patch of significantly higher returns and therefore achieves more than the CPI+5% p.a. objective.

Construction considerations

Unfortunately, there are no asset classes in South Africa, or indeed globally, that will deliver guaranteed real returns of CPI+5% p.a. for extended periods. Even inflation-linked bonds have return uncertainty over shorter periods, and could have return uncertainty to maturity if they paid coupons (due to reinvestment risk). With South African cash producing around CPI+1% p.a. over the long-term, achieving CPI+5% for the total portfolio requires exposure to growth assets, like equity and property that can produce much higher returns over the long-term.

One should try to use as many asset classes as possible to maximize diversification. For collective investment schemes we certainly use all of the traditional asset classes and could use alternatives in a regulatory environment that allows for this. It is important to understand that diversification is considered in the context of the investment horizon, and not over short-term periods of days, weeks or months.

Using long-term historic return data for each asset class, and making some assumptions about what real return each asset class can provide on a forward looking, long-term basis, we can model the optimal exposure to each asset class to achieve CPI+5% over a six year term (the modelling and outcome of this is covered in other articles in this edition – suffice to say this is an important step in the process of building such a goal-based solution).

With this strategic asset allocation (SAA) as a guideline, you can either build the solution using a specialist approach or a balanced approach to portfolio construction.

In a specialist approach, individual asset classes are given to specialist asset managers, and each of these specialist building blocks is then combined at the portfolio level. In this way, skilled equity managers manage the equity component and skilled bond managers manage the bond component and so on. Importantly, under this approach, the solution provider (us as a multi-manager for example) can tilt the exposure to each asset class portfolio depending on the current investment environment and our shorter-term outlook for each asset class (this is referred to as tactical asset allocationTAA). 

In the balanced approach, asset managers are given a multi-asset class mandate and are responsible
for both the asset allocation and security selection within asset classes. In this way, the balanced manager can optimize stock selection within his/her equity component for example, giving consideration to the exact instruments and resulting duration of the bond component. These multi-asset (balanced) managers can take much bigger asset allocation positions based on their valuations, and tend to be more concerned with downside protection. This creates challenges for the solution provider creating a range of portfolios to meet various real return objectives.

In building a range of say five goal-based portfolios along the risk/return spectrum, our preference would be to use the specialist approach because this gives us more granular control over how the portfolios will perform as a range. We must also remember that asset allocation is the most important factor in determining the differences in returns observed from funds/portfolios.

It will be instructive to consider whether the portfolio should change over time to reflect past performance i.e. if you have achieved CPI+10% p.a. over the past three years, do you de-risk the portfolio on the basis that you will achieve your objective over the ensuing three years? There are a number of complications with this which are worth exploring in greater detail.

The first, is that the portfolio is aiming to achieve the objective continuously for many generations of investors, and many generations of their investments. Someone first investing in the portfolio after the three years of CPI+10% p.a. would not have enjoyed this return, and would now be invested in a lower risk portfolio expected to produce a lower return than the objective. The second is more caustic. If the portfolio had instead only achieved CPI+0% p.a. over the previous three years, do you increase the risk in the portfolio to try to make up for the shortfall, exposing investors to much more risk than had been initially assumed? Clearly this would have dire consequences if the risks materialized in large losses.

The above approach lends itself to considering both active and passive investing paradigms within portfolio construction, or a combination as required given other priorities (like costs).

Measuring performance

Once the portfolio has been designed and constructed to deliver the investment objectives, and is being managed to do this, it is important to reflect on how it is performing against the initial specification. There are several ways to do this and one needs to be careful with instinctive reactions of failure when the portfolio doesn’t deliver CPI+5% p.a. in the first complete six year period. If the initial discussions were well understood and documented, the discussion could be easier. This is where initial collateral and time exploring expected portfolio behaviour (not just the single dimension of expected long-term return) will be very well rewarded.

This also provides great insight into what makes CPI+x% such a bad benchmark i.e. the fact that it is not investable. This is why we would not give this objective to asset managers as a benchmark, and would never evaluate their performance against such. We would try to understand the performance of the portfolio in the context of the performance from the underlying asset classes, and appropriate peers. These solutions should therefore be evaluated on the same basis, and this is in fact how we negotiate the portfolio performance evaluation with our clients.

Any or all of these tools could be used in the short or long-term to provide insight into the likely probability of achieving your goal. Some are more effective than others for the client to measure your skill in building a portfolio.


Ultimately there are several ways to build a goal-based portfolio and hopefully which ever you choose is effective in meeting the investor’s goal. It is worth investing time upfront ensuring that both advisers and their clients understand the uncertainty in capital markets and how these uncertainties remain within portfolios even when they are very well designed and constructed around very specific investment objectives.

A portfolio’s performance can deviate significantly from an inflation objective in the very short-term, but this tends to converge over the longer term and the solution provider can monitor progress relative to this. We find that a good measurement tool is to compare the return of the portfolio on a short-term basis relative to the return of the optimal strategic asset allocation that we determined upfront in the design process. Using this, we can attribute the key drivers of out or underperformance into the component parts of manager selection and tactical asset allocation. In some circumstances, it may also be appropriate to compare the return of your portfolio to that of a representative list of peers aiming to achieve the same goal.

Ultimately there are several ways to build a goal-based portfolio and hopefully which ever you choose is effective in meeting the investor’s goal.

By Malcolm Holmes,

Head of Portfolio Management
STANLIB Multi-Manager