Property performance can be measured against inflation but the relationship is far from straightforward, argue Mark Callender, Neil Turner and Jenny Buck
The past few years have seen a shift by some pension funds away from relative benchmarks and towards total return targets. Although the trend is less pronounced in property than in equities, the general direction has been the same, as pension funds have focused increasingly on growing the value of their assets, rather than the relative short-term performance of their fund managers. Initially most targets were specified in nominal terms (eg, 7% per year), but pension funds are now also setting their targets as a margin over inflation (eg, retail price index + 5% per year), in order to preserve the real value of their members' benefits.
The migration to total return targets and in particular, the adoption of retail price index linked targets has brought the relationship between property returns and inflation into sharp relief. The unwritten assumption among many fund managers seems to be that property total returns will move up and down with inflation.
This view also appears to be common currency among many clients and it may be that a second motive for the introduction of retail price index + targets has been to stop fund managers getting a free ride if inflation were to accelerate. But are these assumptions valid?
Historically, there has been little relationship in the short term between property performance and inflation. In most of those countries with long-term data the correlation coefficient between all property total returns and inflation has been insignificant (table 1). The exception is Australia, although even there the relationship is not very strong and inflation only explains around 26% of the annual variation in returns. Repeating the analysis at a sector level in the UK confirms the view that, in the short term, property returns are unrelated to the rate of inflation (table 2).
At first sight, the lack of a relationship might seem odd given that rental values, which represent the open market price of commercial property space, should - all things being equal - be related to the general level of goods and services in the economy. However, although there is a loose relationship, shown by the correlation coefficient of 0.40 in the UK, the major influence in the short term is the balance between the demand and supply of space in occupier markets.
While tenant demand for space is largely a function of economic growth and should be quite strongly positively correlated with general inflation, the supply of space tends to lag both the economic cycle and inflation. This is because of the two to three years taken to build a new office, or shopping centre. For example, whereas inflation peaked at 9.3% in 1990, the amount of new office space completed in central London did not peak until 1991 and the subsequent glut of empty buildings meant that rental values continued to fall over the next two years, even though overall inflation in the UK was positive.
Likewise, although rents on existing leases in Sweden are indexed to inflation, the net income received by investors in central Stockholm offices fell by 12% between 2002 and 2004, due to a combination of lower tenant demand, rising new supply of offices, short leases and a 8% fall in open market rental values. General inflation in Sweden was up by 2% over the same period. Thus, even in countries where rents on existing leases are linked directly to inflation, there is no guarantee that rental income will track inflation, if the underlying occupier market is inherently cyclical.
The other key reason why property returns are unrelated to inflation in the short term is that returns are also heavily influenced by changes in property yields (as distinct from the level of yields). A rise in yields will depress property total returns because property is to some extent a long-duration asset and the negative impact on capital values of discounting future rental payments at a higher rate more than offsets the benefits of being able to re-invest that income at a higher yield. This means that in an environment where inflation and interest rates are expected to rise, the skilful property fund manager will try to maximise their holdings of short duration assets, in particular cash.
Conversely, a fall in yields will boost capital values and returns.
Changes in property yields are in turn driven by two main influences: investors' expectations for future rental growth and changes in interest rates and bond yields. In principle it might appear that investors ought to welcome an acceleration in inflation because it should offer the prospect of faster rental growth and income growth. Hence yields would fall if inflation picked up. By the same logic, a slowdown in inflation might prompt a rise in yields because future rental growth would be lower.
In practice, however, property yields have tended to increase when inflation has accelerated, depressing capital values and total returns and, conversely, yields have fallen when inflation has slowed. This tendency is reflected in the negative correlation in the UK (-0.31) between inflation and the impact which yield movements have on capital values. The main reason why investors dislike faster inflation is that it almost inevitably results in higher interest rates. That has two repercussions. First, it hampers the ability of debt-backed buyers to acquire property. Second, and more important, higher interest rates have a negative impact on economic growth and tenant demand, inhibiting future rental growth.
In this respect property yields behave similarly to yields on long-dated gilts. However, property is by no means a perfect substitute for bonds, because property yields do not rise and fall in exact parallel with bond yields (the correlation coefficient between the two was -0.11 between 1980-2006) and property returns are also affected by movements in rental values.
Property is something of a chameleon and while it may behave like a fixed income asset at the bottom of the rental cycle, it exhibits strong equity characteristics when rents make a sustained recovery. (For a comprehensive discussion of this issue please refer to ‘Property: Its Role in Liability Driven Investing' by Neil Turner and Julia Felce).
Of course, in time the adjustment process is completed and property yields are left permanently higher in the event of faster inflation (or permanently lower in the event of a deceleration in inflation). This in turn will serve to lift property returns, because the rate of income return will be higher, assuming an unchanged level of voids. Moreover, faster inflation will eventually feed through to faster rental growth, once the economy recovers from the shock of higher interest rates and equilibrium in occupier markets is restored.
The critical question, therefore, is how long is the adjustment process for real estate? The chart shows the correlation coefficient between all property total returns and inflation over different rolling periods, ranging from one year up to 10 years, in Japan, the UK and US. In all three countries the correlation between property returns and inflation is higher over 10-year rolling periods than over one year, although, intriguingly, the relationship never becomes significant in the US (ie, the coefficient remains below 0.5).
The results for the UK and Japan suggest the adjustment process is somewhere between five and eight years. That in turn implies that the minimum appropriate period over which to which to measure property returns against either a nominal target, or an inflation linked target is five years.
We believe that the weak relationship between inflation and property returns over the short term means that clients and fund managers adopting retail price index + mandates have to clearly set out and agree the following:
What are the inflation assumptions that have been adopted?
This will lead to an retail price index + target that both client and fund manager are happy with. It should guide the fund manager in the portfolio construction process. We recommend that the long-term expected return to real estate is used to help arrive at this investment objective. To arrive at that target devoid of the ex-ante expected returns to real estate would appear incongruous given the majority of the asset base will be real estate.
What is the agreed performance measurement period and why?
Our research suggests that this should be at least five years. The analysis is clear. Although the pattern of rental value growth resembles the RPI index, it is not a strong contemporaneous relationship, as some would have us believe. Moreover, when RPI is measured against real estate total returns on a contemporaneous basis the correlation is insignificant.
One has to extend the period of analysis to reflect the fact that it takes a substantial period of time to establish a meaningful relationship between property returns and levels of inflation. This has to influence the performance measurement period.
What flexibility is provided to the fund manager regarding cash holdings?
We think that flexibility here is essential. In agreeing the mandate, client and manager have arrived at a collective view with regard to the level of RPI going forward and the premium required which is commensurate with exposure to the asset class. However, both parties should be aware that the pattern of property returns is likely to be markedly more volatile than the RPI index over time, with annual deviations being substantial.
The fund manager should be able to increase the chances of delivering on the desired return objective by combining traditional real estate portfolio management techniques with additional asset allocation techniques to change the duration of the overall portfolio in anticipation of a rapidly changing interest rate environment.
The manager needs to first, increase cash when appropriate; second, reduce the beta of the real estate portfolio (hold lowly geared retail assets as opposed to highly geared City office assets) as appropriate; and third, reduce cash holdings and increase the beta of the portfolio at the appropriate time.
The skill set required to manage a real estate portfolio on this basis is clearly different to the traditional tools of the real estate manager. A multi-asset class environment combining the skills of fixed income and real estate becomes essential.
Do not rely on index-linked lease contracts to deliver RPI + type returns for clients
This is perhaps the most controversial of all of our conclusions. It might be intellectually seductive to argue that buying properties with index-linked lease contracts will deliver RPI + type returns. We believe that these kinds of properties have a role to play.
However, rises in property income from rises in the retail price index are very likely to accompanied by rises in interest rates and property yields. Higher income is good for a portfolio that is managed to keep pace with the retail price index but the higher discount rates that the market will apply to that income in a rising interest rate environment raise serious doubts over such a simplistic investment strategy.
Jenny Buck is head of multi-manager at Schroder PIM
Neil Turner is head of property investment at Schroder PIM
Mark Callender is head of international property research at Schroder PIM