Events of the past couple of years suggest there is scope for improvement in the way property portfolio risk is managed; one key lesson is that we need to identify it ahead of the curve, as Gerry Blundell reports

Investment managers are paid to take risks. But all too frequently the risk they take on in portfolios and vehicles is measured retrospectively using a wide range of measures based on the past variance in returns. Over the years the measures used have blossomed from the mundane standard deviation through Sharpe and others' ratios to tracking error against either a benchmark or an absolute return target. However they are all backwards looking and all assume risk in its many forms can be captured by return variance.

In commercial property, where the typical correlation between individual stock values is relatively low, this can lead to the bizarre situation where the estimated risk attaching to a portfolio is based in large part on stock no longer in the portfolio, since quite long periods of return data are needed to obtain a reasonable estimate of variance. What is almost as bad from the manager's viewpoint, if not that of his or her critics, is that fresh data about risk only arrive after the related investment decisions have been taken. So not only is the measure of risk likely to be misleading, it is late.

It will not come as a surprise, therefore, that there is a poor correlation between past variance and portfolio returns. Across a 145-portfolio sample measured over the 11 years to end 2009 it was only +0.12; at least adherents to capital market theory can comfort themselves that the correlation was positive.

Despite the manifest shortcomings of the ‘rear-view mirror' approach, most of those concerned with property risk management continue to pore over past measures of variance. Why should this be? A number of drivers are at work: variance is the conventional wisdom, the data are easily derived and frequently audited, property gets to speak the language of asset allocation, and what else is there?

A forward-looking alternative
It was to address the last of these that the Investment Property Forum commissioned an investigation into the characteristics of portfolios that were associated with subsequent risk, the full version of which has just been published.* The study looked at the relative performance of +200 portfolios over 11 years to end 2009. Risk was measured as the difference between a portfolio's return and that of the sample average. The bigger the gap, the bigger the risk. Although most people are more sensitive to downside risk, sometimes risk-taking pays off.

So both positive and negative differences were treated equally. The study then looked at a range of characteristics of the portfolios at the preceding year end, searching for risk factors that were significantly correlated with subsequent risk.

Over the 11 years that this exercise was carried out (1998-2009) nine risk factors proved to be significantly linked with risk:
• Property type concentration;
• Regional concentration;
• Tracking error weighted by exposure to type;
• Remaining lease length concentrations;
• Percentage value of the five largest assets;
• Average lot size;
• Tenant exposure;
• Relative equivalent yield;
• Risk in the previous year.

The majority of the nine relate to various types of asset concentration; more eggs in fewer baskets. An obvious point but one that is not limited to just sector and region; the research found a wide variety of ways in which systemic risk can be generated. Also frequently linked to risk was the level of risk in the previous year, suggesting that style might have had an influence, a suggestion borne out by the importance of relative yield, especially when negative risks only were analysed. These nine factors appeared more often than not over the 11 years and were termed evergreen factors.

They were supplemented by four cyclical factors: development exposure, relative covenant strength, vacancy rate and leverage. The first three of these tended to become significant only after real capital values in the market had fallen for two successive years. On the upside they played little part; but on the downside, arguably when risk management really matters, they cut in. These were therefore described as cyclical factors.

Leverage
The fourth cyclical factor, leverage, was ever-present but was treated as a cyclical factor because the sign of its correlation with relative returns changes from positive on the upside to negative on the downside. Not only was it the most powerful of the factors analysed, but because it changed the portfolio returns, its presence swamped the effect of the other 12 factors when leverage was applied to a portfolio. The figure shows the influence of leverage in 2008.

The figure plots LTV on the horizontal axis and risk vertically. The dots are individual portfolios based on a sample of open and close ended vehicles within the AREF classification. The relationship is strongly positive and is strongly non-linear. It can be seen that the fitted curve approximating the shape of the distribution rises slightly towards the left as well as the right. This reflects portfolios which were net of cash at end 2008, a position that generated a positive difference relative to the average return in 2009.

Interestingly, LTV at end 2008 was also correlated with unleveraged returns on direct assets in 2009, albeit not as strongly. In plain words, funds with relatively risky portfolios were the ones that were tending to add the most debt, as if they were struggling to source stock that met their investors' return aspirations and were turning to leverage to enhance returns.

Future prospects
The study found that the level of risk across portfolios had been rising though the last decade, a rise coincident with rising vacancy rates, falling lease periods and rising LTV ratios. This risk is unlikely to go away and we may well face structurally higher risk premia for commercial property than we have enjoyed in the past.

The danger ahead is that the existing debt in the market will have a negative impact on the sector. Quite apart from the thorny issue of debt refinancing, a low-growth economy might well push managers towards excessive risk aversion.

The art of investment management is to balance the eternal triangle of risk, return and responsibility; and excessive risk aversion is as bad in the long term as excessive risk taking. It is the job of managers to take risk; to do that in a responsible manner they need properly to identify it ahead of the curve.

* ‘Risk Web 2.0; an investigation into the causes of portfolio risk' Investment Property Forum (2011)

Gerry Blundell is strategic adviser to Legal & General Property Limited