To evaluate fund managers, investors must appreciate how their per-formance is derived. Malcolm Hunt uses attribution analysis to explain

Within real estate, performance measurement comes in many guises. Property fund managers and research teams blend a mix of investment strategies with strategic asset picking to identify the best assets in the best sectors to hold - and for the correct length of time.

The extent to which outperformance can be attributed to a specific investment strategy or stock selection process can be analysed at fund level. This can range from a relatively straightforward comparison of a fund's total return against the portfolio IPD benchmark over a specified time period to more detailed segmentation or ‘attribution' analysis.

Attribution analysis is an attempt to separate market performance drivers in a portfolio from the component derived from individual property selection. The two components of attribution analysis are the variable ‘structure' segments as well as the individual ‘property' constituent, with the latter widely considered to reflect the majority of relative returns. Small changes in the structure segmentation are therefore of considerable importance.

This structural segmentation most commonly relates to the sector or regional allocation a fund is weighted towards, measuring the extent to which capital was employed towards areas of the market which performed above average and avoided those which performed below average. Property measures the extent to which capital was invested in individual assets, within defined segments of the market, which performed above the segment average while avoiding assets below the segment average.

But analysis by IPD suggests that a much broader set of structure-based attribution analyses must be employed to give a more thorough, bespoke picture of a fund's relative sources of returns over different phases of the property cycle. Whichever structural segmentation is used as the comparable, by definition the component that such analysis shows is not attributed to structure returns is ascribed to individual property asset selection.

IPD has carried out an analysis of the main forms of attributions analysis - based on 166 portfolios with a 20-year performance history in the IPD UK Databank - to identify which are the most helpful in explaining the sources of underperformance or outperformance in the various cycles over the past two decades. Conclusions will be drawn to help provide investors with some insight as to which tools may be worth adopting for their own portfolios as the UK commercial real estate market recovers.

Sector-regional versus asset selection
The best-known form of real estate attribution analysis is a sector-regional versus individual asset selection analysis. This technique calculates the proportion of a portfolio's return that is derived from its sector allocation to the strong and weak parts of the wider market. This is compared against the extent to which individual property selection is responsible for the fund's performance.

The structure or property score, indicated on the y axis, is illustrated in figure 1 where the structure segment is sector-regional selection. Essentially, this analysis aims to isolate the component of returns that is not attributable to asset selection. As the figure shows, individual asset selection is a far greater contributor to relative performance throughout the 20-year period, which means returns are largely differentiated by a fund manager's stock picking. There are two periods where that trend is weakened and stock picking mattered less; in 1991 and from 2002-03.

The most pronounced departure, in 1991, produced a structure score of 0.35, where relative returns were attributed to sector-regional strategies with the balance asset selection. This is due to the fact that over the 12 months the spread of market returns within the different sectors of the property market were far broader than usual. Indeed, according to the IPD UK Annual, in 1991 the spread of total returns was 19.9 percentage points - with industrials delivering +9.1% and offices returning -10.8%. Bringing this analysis as up to date as is possible, we can see the return spread at PAS segment level over the first eight months of 2009 is deep into double digits.

Therefore, the variation in returns delivered by the 166 portfolios directly related to the overweight or underweight positions to the industrial and office sectors. The 1991 spike is also an indicator that over the year there was higher diversity of investment strategies among portfolios, that is to say, investors were making widely different market bets on where outperformance would come from. Over the following decade, the significance of sector-regional strategies as a driver of portfolio returns failed to register a score above 0.2, with an average over the period of 0.1. The average spread of returns over this period was just 4.1 percentage points.

The second deviation from trend, from 2002-03, is again accounted for by the spread of returns between the best and worst performing sectors - in these years retail considerably outperformed offices to produce a 10.7 and 12.3 percentage point spread, respectively. This compares with a long-term average returns spread, over the 20-year period, of 6.6 percentage points. In both the early 1990s' recession and the post-dot com crash, it was the relative underperformance in offices that was responsible for the wider than average sector return spread causing the three spikes.

In the more recent property boom years, from 2004 to mid 2007, the tide of capital appreciation sweeping the UK market enabled most investment strategies to benefit, with less emphasis on property selection or strategy expertise. This is corroborated by the environment of ‘hot' retail capital chasing limited stock, coupled with cheap money allowing investors to gear up excessively. As a consequence of heavy capital inflows, yields stretched indiscriminately across sectors prompting capital values to rise, delivering strong returns to investors.

During the recent downturn, this relationship weakened with a return to an environment where sector-regional allocations are more important than during the boom. This is perhaps an indication that the property fundamentals are returning to the market; that is, investment strategies will be more critical in determining underperformance or outperformance.

Indeed, segmentation analysis itself often needs to be widened to more accurately capture bespoke performance attribution. For example, the counterpoint segmentation to individual stock selection could be run to include tenant lease lengths, covenant risks, lot size and floor space analysis. This more detailed attribution analysis is underway at IPD.

Lease length and covenant strength
IPD has drilled into last year's sector returns to show the impact of variations in lease lengths. As figure 2 shows, in both offices and industrials there is a clear incremental insulation from the worst of last year's sector returns the longer the lease was. In offices, the returns spread between the shortest and longest leases was eight percentage points in favour of longer leases, while in Industrials the spread was four percentage points. The trend, however, was not consistent across the retail sector where shorter leases were marginally - by 80 basis points - the best performer in what was a poor year across all sectors and lease lengths.

In an environment in which leases are shortening, with increased flexibility such as break clauses and longer rent-free periods, this kind of analysis will only become more relevant to investors in the future.

Another increasingly important performance driver is the covenant strength of the underlying tenant, particularly in an environment where the economy has unilaterally suffered across the sectors. Tenants' relative financial strength, or lack of it, can be measured in relation to the total return delivered at sector, market segment and even at the individual property level. To do this, a ‘weighted risk score' (WRS) is derived from the sum of all risk scores, - provided by Dun and Bradstreet - of all rated tenants, weighted by current contracted rent based on either a benchmark, segment or investment portfolio.

Figure 3 illustrates the varying weighted risk scores for the 10 main IPD PAS segments over the first two quarters of this year. The closer to 100 the WRS is, the more secure the underlying tenants and, it would be expected, performance would be better. The graph shows the WRS for the PAS segments, relative to the all property average benchmark, for the 12 months to the end of the first and second quarters.

As illustrated, the improvement in covenant strength over the two rolling 12-month periods have been most pronounced in City offices, where the property segment has risen from almost par with non-central London offices and industrials to be the clear leader in covenant strength. The improved weighted risk score among City offices over the second quarter helps underpin improving confidence among investors targeting that part of the market. The positive relative weighted risk scores among standard retails chimes with the outperformance relative to the IPD average, which for south east and rest of UK was 8% and 5.5%, respectively, over the two 12-month rolling periods.

Comparing these weighted risk scores with market segment performance at portfolio level can help investors gauge how influential high covenant strength is in delivering outperformance in the long term. This form of analysis can be rolled up to full attribution analysis with weighted risk score as the structure segment - the counterpoint to individual property selection. Depending on the phase of the property cycle, the segmentation analysis that proves most insightful, can vary. During periods of wider market volatility covenant strength is a critical indicator.

Overall the explanatory power of structure, as one of the two attribution analysis variables, would be greatly improved by additional segment analyses such as lease lengths and covenant strengths. Together these performance-evaluation tools can help investors, as well as portfolio managers, to understand the extent to which returns delivered can be attributed to a fund's investment style, its sector allocation, its asset selection or simply its market timing.

Malcolm Hunt is head of UK client services at IPD