Investing in real estate has traditionally been popular for three reasons: its inflation hedging characteristics; as a diversification asset with regard to the time variation in the proceeds on stocks and bonds and; its relatively stable returns.
But nowadays, with risk being key in every investment decision, real estate risk has been under greater scrutiny. The global financial crisis has made investors wary of the unknown and unexpected. This means more weight is given to simple metrics of historic volatility whenever new capital is allocated.
In the June 2014 edition of sister-magazine IPE, Stefan Dunatov called for an alternative view on the true risks in the allocation strategy. We offer, apply and illustrate a simple but powerful alternative framework to make allocation decisions while accounting for all relevant risk dimensions in adequate proportion. The basic assumption underlying our view is that true real estate investment risk is not a simple and uniformly applied metric like a standard deviation of 6.83%. Proper investment risk assessment is forward-looking in nature, takes the cyclicality that characterises real estate markets into account and can be tailored to specific investment strategies, both in investment style and duration.
Investment decisions require due diligence efforts, a data-driven process. Therefore, allocating funds to real estate assets means that the available information needs to be analysed thoroughly before the point of no return. In the case of real estate, investors face two distinct challenges in this phase of deliberation – time series and metrics.
We have long passed the point that only three things mattered in real estate decisions – location, location, location. Real estate markets around the world have collected and disseminated time series for investor analyses. This allows an ex-ante view on expected returns and accompanying volatility. In financial theory, volatility has become a synonym for risk, but what this measure shows is just a view on deviations from observed returns around a mean which, we argue, is not the one-size-fits-all measure of risk in illiquid and generally long-term private property investments.
Moreover, unlike those for stocks and bonds, real estate time series are often limited in duration, frequency, coverage and quality. Due to the lack of asset liquidity – the high transaction costs force investors to take a long-term perspective on their real estate investments – most real estate time series are based on appraisals instead of market-tested pricing information of transactions. The underlying appraisal process, the human overlay, tends to smooth out the intertemporal fluctuations, as bad news is often reported later and in small portions at a time.
Furthermore, these time series offer only a limited history as they typically stretch less than two decades of annual data for most markets – an era which includes a full investment cycle but offers only limited degrees of freedom for the investment analysis. In other words, the available real estate time series are rather short, and need to be handled with care in order not to mistake the appraisal signals for low risk evidence. While the emergence of transaction-based indices has started to alleviate the issues with historical time series for private real estate markets, these tend to be relatively short, limited in market coverage and will fail to show proper market movements when it matters most – times of crisis characterised by illiquidity, not uncommon in direct real estate markets.
Once the apparent data issues are acknowledged, investors also need to be careful about the metrics they apply to analyse risk appropriately. To echo Dunatov, we agree that the commonly used standard deviation is a poor measure for relevant risk. It is a metric that captures a history that is rapidly losing relevance. The past decade has demonstrated that there is little standard in the most striking deviations in investment returns. Furthermore, a simple standard deviation is unable to disentangle the different layers of uncertainty that matter in varying degrees to investors with different ambitions and endowments.
We would like to offer an alternative view on the allocation analysis without requiring a novel set of mathematical skills to apply increasingly complex formulas on constrained data sets. In order to better incorporate relevant risk assessment in private real estate investment, we propose a multi-factor framework that can gauge investment risk while taking investor specifics into account.
As traditional risk models focus on the volatility in returns – in essence, a structural factor – they fail to incorporate the current status of highly cyclical real estate markets. After all, strong cycles have become the proxy for risk for many investors but offer a source of return for others and are arguably irrelevant for the likes of sovereign wealth funds with 30-year investment horizons. A broader set of consistently measured global risk metrics allows for a model that is capable of satisfying the needs of long-term and shorter-term investors without building on anecdotal evidence or losing international comparability.
The model that we propose – which has been incorporated into CBRE Global Investors’ forward-looking global Risk Adjusted Real Estate (RARE) model – builds upon long-term (structural) and short-term (cyclical) economic and real estate specific variables.
Each segment in this four-quadrant overview incorporates a couple of factors derived from reputable data sources with global coverage which can be combined into one risk measure. An example of a structural economic risk factor is the economic base of a country. While a focus on one industry is beneficial, if that industry performs it naturally increases the risk profile of a market. Cyclical economic risk is measured by factors like credit default swap spreads and perceived uncertainty in economic growth outlook. Structural property markets risk relates to factors such as transparency of a market and the important ease of building permit issuance.
The last quadrant not only incorporates the cross-sectional volatility of property markets but also measures the current position of each market within its own history, as not only the expected return is affected by entering into volatile and historically expensive markets.
The structured yet flexible nature of this model makes it not only capable of adequately calculating relativities in risk profiles for investors with different characteristics but also allows for a deeper understanding of major risk factors that drive these relativities, as illustrated in figure 2. An investment industry traditionally focused on expected returns still greatly benefits from a better understanding of what drives true investment risk before being satisfied by advances in index measurement.
Dirk Brounen is professor of real estate economics at TIAS School for Business and Society at Tilburg University. Maarten Jennen is associate professor at TIAS School and director of investment solutions at CBRE Global Investors