It is one thing to have a good investment idea, but quite another to extract the maximum potential from that idea and combine it with others to create an optimal portfolio. This is where portfolio construction comes in.
13 minute read
Diversification may be the only free lunch in finance, yet even seemingly well-diversified portfolios can harbour hidden biases and correlations. These can lead to excessive weightings of certain strategies or a portfolio that expresses too many similar ideas, overexposing it to particular risks. In a difficult market environment, when traditional correlations break down, it can come as a nasty surprise. This is where good portfolio construction shows its worth.
One of the most important roles of an asset manager is to construct portfolios for a variety of futures
“One of the most important roles of an asset manager is to construct portfolios for a variety of futures, even those that never materialise,” explains Euan Munro, chief executive officer at Aviva Investors. “During good times, positioning strategies for the very worst-case scenarios might appear wasteful, and some might argue over-elaborate. But for events like the global financial crisis or COVID-19, the effort to make portfolios as watertight as possible bears out, especially when there is so much we don’t know.”
“When you have the right combination of strategies that deliver returns in a smart way, it gives you a portfolio that is more resilient to changes in correlations and to market events,” adds Iain MacCormick, head of multi-strategy implementation at Aviva Investors.
Letting the best ideas shine
So, we know portfolio construction is important, but what differentiates an effective process from one that fails when you need it most?
Portfolio construction does not preclude the ability to increase or reduce the overall risk in a portfolio, or to add or exclude specific asset classes – these decisions remain dependent on investors’ objectives, guidelines and preferences. Instead, it aims to optimise risk-adjusted returns for a given set of constraints and to embed diversification and resilience into the investment process. The quality of investment ideas and strategies remain paramount, but portfolio construction identifies the best way to put them into practice.
Josh Lohmeier, head of US investment-grade credit at Aviva Investors, likens this to creating a framework that lets your best ideas shine. “Any great portfolio construction process needs strong idiosyncratic ideas. Portfolio construction is how you build a portfolio around those great ideas to provide resilience and downside protection.”
A robust, repeatable portfolio construction process can uncover and remove biases and correlations, allowing investors to add return without taking additional risk. This is as true for funds that invest in publicly traded assets such as bonds and equities as it is for real assets like infrastructure and real estate. Instead of blindly following an inefficient benchmark index, investors can construct portfolios of well-understood assets that allow their expertise to add value – identifying and optimising return drivers, ensuring diversification and taking a forward-looking view to managing risk.
Addressing inefficiencies in tracking error and indices
Active managers will often follow a bottom up and simplistic approach to portfolio construction; purchasing the securities they like, avoiding those they dislike and determining whether they are happy with the resulting overall risk and tracking error. It is a good start, but more specific risk allocation techniques can help deliver more resilient investment returns.
In a paper published last year, Lohmeier illustrated what tracking error can and – perhaps more importantly – what it cannot do through a simple example.1 As shown in the table, it is possible to create a variety of portfolios from a sample benchmark index. There can be differences in tracking error, volatility and performance in different market conditions; even in a simple universe where every portfolio owns exactly ten per cent of each issuer.2
What is surprising in this example is that both the risky and defensive portfolios have the same tracking error of 50 basis points, despite significant differences in how they deviate from the benchmark and how they might perform through an investment cycle.
Figure 1: Tracking error can be misleading
“Tracking error is a very important tool for understanding how you are deviating from your preferred benchmark, but it is not necessarily a great tool for measuring risk,” explains Lohmeier. “When we think about portfolio construction, we always want to acknowledge what our tracking error is and where it is coming from. Those deviations need to be value-creating and risk-reducing; it is really about defining how you are attacking that inefficient benchmark.”
Simply aggregating the best investment ideas does not mitigate basic index inefficiencies
Sunil Shah, head of Canadian fixed income at Aviva Investors, argues that simply aggregating the best investment ideas does not mitigate basic index inefficiencies – forfeiting an opportunity to generate excess return repeatably.
In a recent research paper, he explained that using tracking error can increase investors’ exposure to risks embedded in an index by maintaining inefficient index concentrations or discouraging the inclusion of non-index ideas. He illustrates this by reference to the Canadian bond index, which is particularly concentrated.3
Figure 2: Top ten Canadian non-government issuers by market value
Similar inefficiencies can be found in real assets indices, as Chris Urwin, director of real assets research at Aviva Investors, highlighted in his analysis of MSCI commercial real estate indices.4
Even the most comprehensive real estate indices provide incomplete coverage in terms of sectors and geographies. Like all indices, the coverage and weightings evolve as a function of holdings rather than real economic activity. Just as an equity index will tend to overweight stocks that have been bought the most and gone up in price, a real estate index will change as assets trade between investors who contribute to the index and those who do not.
“A desire to retain portfolio exposures in line with the benchmark can result in investors buying into markets as they get more expensive,” says Urwin. “A prominent example of this is the increased ownership of central London offices by overseas investors who do not typically contribute to MSCI. This has led to a big decline in the share of central London offices in the index over time – a market that we feel holds a lot of value.”
The rise of passive investing has led to many investors now getting exposure to big-picture trends through ETFs
Changes to the way markets function can cause distortions, something that has been evident in recent years in the stock market. According to Mikhail Zverev, head of global equities at Aviva Investors, the rise of passive investing has led to many investors now getting exposure to big-picture trends through exchange-traded funds (ETFs), rather than rigorous research of company fundamentals.
When something dramatic happens, such as a sudden change in the macroeconomic environment, ETFs reallocate at a large scale, which affects a lot of companies, sectors and investors.
“The volatility those relative moves cause can be substantial. The magnitude of the sell-off and the dichotomy of safe versus risky assets during the COVID-19 crisis has illustrated this well,” says Zverev. “The result is that whatever exposure you think you have to residual factor risk, and which you thought was safe ten years ago, is probably no longer safe because those winds are blowing harder.”
For investors who do not want to be at the mercy of volatility, the next step is to understand how to make their portfolio construction process more robust and capable of delivering resilient outcomes.
Diversification: Easier said than done
One benefit of not being too closely wedded to tracking error is the freedom to allocate to a variety of assets or asset classes. However, diversification is more than the sum of different assets; it requires an understanding and careful calibration of the underlying drivers of risk.
Figure 3: The risk drivers of income in European real estate
Professor John Kay, who recently co-authored ‘Radical Uncertainty: Decision Making for an Unknowable Future’ with former Bank of England governor Mervyn King, sums it up nicely: “Widespread diversification is not something you approach by calculating betas in the way portfolio models typically do, but by asking the “what is going on here” question, and by understanding the underlying determinants of asset price returns.”
In real assets, a study by the Investment Property Forum looked at the volatility of returns on over 1,000 properties in the UK between 2002 and 2013. As the specific risks are so different from property to property, it found that diversification can be achieved rapidly – portfolios of 15 to 20 assets would, on average, have recorded volatility of returns close to that of the overall market.5
Figure 4: 10-year standard deviation of simulated real estate portfolios
In credit, there are multiple return drivers such as carry – how much yield investors are getting – and forward-looking views on how spreads could change in various markets or sectors.
“You are trying to isolate the most efficient places to generate yield and carry relative to the benchmark, while simultaneously efficiently placing your idiosyncratic ideas, which are more focused on spread compression,” explains Lohmeier.
A good risk approach can disaggregate the portfolio’s risks into these distinct categories
Equity risk can also be broken down by the level of exposure to regions, sectors or factors; whether style factors like value, growth, quality and momentum, or macroeconomic factors like energy prices or interest rates. There is a degree of overlap, but a good risk approach can disaggregate the portfolio’s risks into these distinct categories. Investors need to assess whether they are exposed to these risks by design, via a non-consensus view and well-argued investment case, or unintentionally.
“As you build portfolios from the bottom up, you accumulate unintended risks. We look to monitor and control these unintended risks and cross-check them with the views of our macroeconomic colleagues,” says Zverev. The goal is for most of the risk budget to be dedicated to intended and diversified stock-specific risks.
True diversification comes through constructing a portfolio of assets with low correlation to each other. Success or failure in achieving this will only really become apparent in a crisis.
“Diversification is a cornerstone in investing, but that only works when correlations work for you; meaning when you have low correlations between your ideas to reduce the total amount of risk, or positions that offset the risk of other positions,” says Wei-Jin Tan, investment risk and portfolio construction specialist at Aviva Investors.
“Correlation is a huge factor in this, so we closely monitor the correlation structure of portfolios – in other words, how correlated all the ideas are to each other – and by using metrics like marginal risk contribution. It tells us which positions are increasing risk in the portfolio, which are offsetting it, and which don’t add much risk but do enhance diversification,” he adds.
Figure 5: Correlation schematic in AIMS portfolios: Market, opportunistic and risk-reducing strategies
It pays to constantly monitor correlations between assets, however, as these can change over time and in different conditions.
“When we think about portfolios, we also think about how our correlation assumptions change through time under different futures and different events. We are realistic about our ability to predict the future – we know that we can’t. Instead, we focus on building portfolios that are resilient, so that whatever the future may be, the blend of assets and positions will help us navigate through that uncertainty,” says MacCormick.
One way to measure this is through a method called absorption. Using principal component analysis, investors can assess how much of a portfolio’s volatility can be explained by one factor.
If everything can be explained by one factor, it is dangerous because you are not really diversified
“If everything can be explained by one factor, it is dangerous because you are not really diversified, even though you have all these ideas and correlation may have been low in the past,” explains Tan.
To assess diversification, investors should also pay attention to turbulence within a portfolio. Tan explains turbulence as being a measure of correlation ‘unusualness’.
“If you have a classroom of 30 university students and most people’s height is between five and six foot, an eight-foot person would be an outlier. That is your huge standard deviation event – your black swan,” he says. “But you can see I am measuring this just on height, on one dimension. What turbulence does is to measure it multi-dimensionally. If you have 30 ideas in a portfolio, you are looking on a 30-dimensional plane at how unusual their correlations get.”
Figure 6: Turbulence
Turbulence analysis looks at a portfolio over time, how it has evolved and how it maps onto what the expectations were. Investors can analyse whether individual strategies are exhibiting unexpected behaviour and identify internal stresses within the portfolio from a performance perspective.
“This also feeds into concentration risks. We can aggregate individual assets from across strategies to understand overall concentrations; not just from a pre-trade perspective, but also how those concentrations are moving through time,” says MacCormick.
Building efficient portfolios
These tools allow portfolio managers to target the efficient allocation of return drivers across a strategy; helping them determine which positions to add or ignore, but also providing guidance on how to size and structure the allocations to optimise returns for the same level of risk.
In a benchmarked portfolio, it can mean exploiting the inefficiencies of an index by underweighting some sectors and overweighting others or adding non-benchmark ideas to increase diversification. Returning to the example in Figure 1, portfolio construction enables the creation of distinctly different portfolios, from defensive to risky, all within a simple universe of just ten issuers and a strict set of constraints.
Incorporating investors’ broad views on the market – to introduce more traditional alpha through sector weightings or yield curves, for example – can also add value when maximising portfolio efficiency.
“It helps us understand where global economies are headed, as well as the tone and strategic view of the various asset classes competing against each other for investor demand. It helps you take the blinders off from your respective asset class and prescribe how those things are going to impact your market more granularly,” says Lohmeier.
Just as diversification makes a portfolio more robust, Peter Fitzgerald, Aviva Investors’ chief investment officer, multi-asset and macro, believes fostering an inclusive culture of debate within investment teams aids the construction process.
“Our investment team hosts a firm-wide debate [the House View] on where the economy is likely to go and whether that path is priced in. We always seek to understand where there is consensus, which will form part of our central case, and where there is disagreement,” he explains.
Fitzgerald emphasises the fact people are encouraged to speak up when they disagree. “The whole point of the exercise is to understand what we might have missed and whether there are any risks we have failed to spot. When there is vigorous debate, the person who disagrees strongly may not see their view end up in the central case, but it will feature in one of the risk cases,” he adds.
Factoring in a wide variety of possible futures is another critical component of portfolio construction
Factoring in a wide variety of possible futures is another critical component of portfolio construction. MacCormick explains that the House View process allows the team to create a central scenario that represents the collective judgment of the firm and to analyse how well portfolios may perform against those.
“We would expect the portfolio to perform under the central scenario, but we wouldn’t expect performance to be too limited as we move towards more outlying scenarios. This will feed into our allocation decisions,” he says.
To give an example, in the Aviva Investors Multi-Strategy (AIMS) portfolio construction process, the risk management team uses a parallel approach to look at risk: a longer-term model and a shorter-term model. The longer-term model is based on five years of history and weekly data; the shorter-term model is based on two years of data and daily data.
Tan explains that, while AIMS takes a long-term approach to investments, looking at a shorter-term model allows them to ‘kick the tires’ of the longer-term model and assess whether any short-term trends are worth considering.
“If we look at this year, huge amounts of volatility came through in the short-term model; then the longer-term model started to catch up,” he says. “But recently there has been a fall in risk in the shorter-term model and the longer-term model is still elevated. You are trying to trade the long-term trends and not be susceptible to short-term noise but be conscious of the fact that long-term trend is changing, and anything should be reviewed.”
Beware behavioural bias
Such an approach can help uncover and mitigate unconscious biases. Investment managers tend to be optimistic about their ability to forecast performance, and this creates a bias toward riskier allocations. This is not limited to investors, however.
We tend to have a more optimistic view of the future, partly because we imagine we have more control over the outcome than we do
“We tend to have a more optimistic view of the future, partly because we imagine we have more control over the outcome than we do. In other words, when it comes to our decision making, we tend to ignore the downside – it’s called the illusion of control,” says Annie Duke, World Series of Poker champion and author of Thinking in Bets: Making Smarter Decisions When You Don’t Have All the Facts.
To illustrate this, Aviva Investors conducted a study evaluating every US investment-grade corporate bond recommendation from five sell-side research departments, capturing “outperform”, “market perform” and “underperform” recommendations across each rating category.6
Figure 7: Outperform recommendations show optimism
The data shows there is a clear low-quality bias to the outperform recommendations: the number of outperform recommendations for triple-B bonds is six times higher than for single-A bonds. If portfolio managers are looking to research analysts for investment ideas, they will naturally be biased toward riskier securities, on top of their own optimism.
Not only is this important before implementing a new idea; it is also essential to monitor invested ideas on an ongoing basis, to track whether they are behaving in line with expectations and, if not, to understand why not.
“We formally review all the trades in the AIMS portfolios on a quarterly basis and ask whether our thesis still holds, whether the drivers are still the same, if the strategy has hit its return target or review levels, and whether we want to stay in it or not,” says Fitzgerald. “There are cases where, even if a strategy makes money, when it is not behaving as we would expect, we might decide to remove it if the drivers aren’t right.”
Once again, understanding and efficiently allocating the return drivers across a portfolio is key. However, more can be done to bolster resilience.
Resilience targeting is about choosing the ‘efficient’ portfolio that best leverages the central investment thesis but will not be materially affected should the thesis fail to deliver.
For instance, focusing on volatility as a primary measure of risk helps add alpha rather than beta to a portfolio.7 “You are dialling in risk-adjusted returns to build a portfolio in a way that matches the risk of the benchmark prescribed by the investor and delivers better outcomes for that same level of volatility,” explains Lohmeier.
These scenario analyses illustrate the level of risk taken in pursuit of returns
This can be further enhanced by incorporating sensitivity analysis of investment ideas or portfolios under multiple scenarios. These scenario analyses illustrate the level of risk taken in pursuit of returns and what can happen to a portfolio when the central thesis does not play out.
As shown in the hypothetical example in Figure 8, not only is the efficient frontier immediately visible in the central scenario (blue line), but so are the trade-offs between added return and the risk of a different scenario playing out. This helps dispel the “illusion of control” and implicitly drives decisions that will embed downside protection.
Figure 8: Scenario analysis: targeting the optimal risk level for performance and resilience
There are two kinds of scenario analysis. The first, described above, assesses portfolios’ sensitivity to volatility under different market conditions. The second is run by risk teams and also known as stress testing.
Stress testing helps you achieve a reallocation or a rescaling of investment ideas
“What stress testing helps you achieve is a reallocation or a rescaling of investment ideas in a way that enables the capture of the inherent alpha while acknowledging there will be periods of volatility, exogenous shocks to the market that cannot be predicted. Portfolio managers need to prepare for those every day,” says Lohmeier.
In the AIMS process, portfolios are tested on how they would have performed during particularly volatile periods, such as the financial crisis, dot-com bubble and Russian devaluation crisis.
“We look at how the portfolio behaves and whether we see resilience, and if it can weather the storm. As well as looking at past periods, we also think about forward-looking scenarios”, says Tan.
While it is possible to be more granular at a sector level, Lohmeier thinks systemic stress-tests are the first order of business.
“We all know that, in a true stress environment, correlation goes to one and everything gets hit,” he says. “The more defensive assets still get hit, but less than risky ones. Stress testing allows us to ensure that, when everything blows out simultaneously, the portfolio holds enough of the defensive and not too much of the risky assets.”
“Cinderella, you will go to the ball”
For much of the past decade, preaching the virtues of portfolio construction would have fallen on deaf ears as ultra-loose monetary policy has made it easy to generate decent returns through simple (and cheap) exposure to a variety of asset classes. But COVID-19 - the impact of which has severely damaged companies, sectors and entire economies - has shifted the debate once more.
Building portfolios that can hold up in the most testing of circumstances is a matter of design, not good fortune; it results from sound investment and risk management processes as much as good, old-fashioned skill and judgement. Unfairly miscast as an expensive ‘Cinderella science’ during the bull years, portfolio construction now has its chance to shine.