The perfect match?
Liability driven investment (LDI) is a broad framework to manage pension scheme risk. Under LDI investment policies are clearly aligned with the scheme's objectives, with risk and return both measured in relation to the scheme's liabilities. What role property can play in an LDI world is, of course, a very timely question.
We have reviewed the topic again with a particular emphasis on real estate's duration measure. With bond yields declining to near generational lows and liabilities increasingly being discounted at market rates, the funding ratios of many pension plans have come under extreme pressure. In the face of such developments, the investment industry has sought to identify other assets that can behave like government bonds. This paper addresses the following issues:
n Why do many investors consider that real estate is a bond proxy?
n Is UK real estate a long duration asset?
n What role can property play within an LDI framework?
This paper argues that, although property has a fixed income component, it also has an equity component. The capital market, and the property investment community, has chosen to appreciate these two facets to varying degrees at different moments (although sometimes over extended periods) through time. Property is a "chameleon"-like asset class; when capital preservation and income stability are uppermost in investors' minds, property's bond characteristics are to the fore. When confidence is high and businesses are expanding, property's equity characteristics become the focus of investors' attention.
Why do many investors consider that real estate is a bond proxy? Where many in our industry have started from, when making the claim that property behaves like a bond and is therefore good as a liability driven investment, is that the income growth from property rises in line with inflation and/or earnings.
Indeed it can be easily demonstrated that the income generation from real estate is a very good match for both of these series and therefore in a very simplistic way meets the cash flow requirements of pension funds. This pattern of income growth is, of course, derived from a combination of rental value growth and the transmission mechanism to income growth through lease structures and, particularly, upward-only rent reviews. Much of the income is also extremely good credit and this has only added to the attraction of the income stream.
Case study: UK property. Chart 1 illustrates the net income growth from UK real estate through time. This is a great cash flow stream. This is for all commercial property, but we could break it down into different market sectors. This would demonstrate that some sectors in the UK have been better at delivering this income growth than others. For example, retail and industrial income growth has generally been a better match than office income growth which has tended to be more volatile and lower than earnings growth owing to the more pronounced office development cycle.
The next chart demonstrates the positive skewness apparent in UK property income growth. What does this mean? Market rental value growth of UK real estate approaches a normal distribution. That is to say that the mean observation is centred towards the middle of high and low (sometimes negative observations) around a market cycle. Income growth, however, does not behave like this at all, through the workings of upward-only rent reviews.
When market rents increase, during the typical five-year rent review pattern the landlord can exercise a ‘call option' and implement the rent review process to increase the rent. However, were market rents to have fallen during that period, no such transmission mechanism will take place since actual rental income can never fall in an upward-only world. This is seen in chart 2, where the orange income growth distribution is effectively truncated on the left hand side at around zero, and is formally known as a statistically significant, positively skewed series. This effectively means that under a typical lease the income has a degree of deflation protection, subject to the tenant's solvency.
Is UK real estate a long duration asset? The duration of liabilities of pension funds is very long. The duration of a financial asset is the weighted average maturity of a series of cash flows. A more practical use of duration is to interpret it as a measure of sensitivity of the asset to changes in interest rates. A long duration asset will fall in value when interest rates rise and increase in value when interest rates fall. This is why long duration fixed income bonds have become so popular with pension funds since these assets rise in value when the funds' liabilities increase and vice versa, a phenomenon that has been reinforced by the imposition of Financial Reporting Standard 17 (FRS17)1.
Charles Ward at Reading University produced a proxy for property duration which suggested property is a long duration asset. In theory, this means that property prices should be very sensitive to changes in interest rates. However, as Professor Ward discovered, unfortunately property has not behaved like that2.
Let us look at some evidence and what the implications are for investors.
In chart 3, correlations are plotted on the vertical axis.
An investment that behaved like bonds in total return space would have a correlation approaching one. It would fall in value as bonds fall in value and it would increase in value as bonds increase in value.
This is demonstrably not the case and undermines the argument made by some that UK real estate is a perfect LDI asset. There are considerable periods of time when bonds and real estate are negatively correlated and throughout the whole period they enjoy almost zero correlation. Ironically, it was this lack of correlation that was used extensively by the property industry to demonstrate property's credentials as a good diversifier. Property's zero correlation with gilts helps to maximise returns at the multi-asset level between equities, gilts and real estate for a given level of risk or indeed minimise risk for a given level of return.
Interestingly, the periods when correlations were particularly low coincided (although it was no coincidence) with high levels of market rental value growth. This occurred in periods such as the mid 1980s when investors were focused completely on growth. It happened again in the late 1990s as investors again were buying real estate due to large income growth increases driven by rental value growth movements. These periods are marked on the chart as ‘Equity'.
ut look at the early 1990s' property crash. Rental value growth was a dim and distant memory and many properties were over-rented (when the rent being paid was above the prevailing market rent). Here investors were very sensitive to interest rate movements, as they should have been. These periods are marked on the chart as ‘Fixed Income'.
The key factor is that property returns are largely driven by changes in market rental growth, which in turn is determined to a large extent by economic and business activity. There is a significant positive correlation (around 0.6) between both property total returns and rental growth and also between GDP growth and total returns. This puts property firmly in the growth/equity/risky assets camp.
We would like to make three observations before we move on.
Please remember that for most of the performance measurement period UK leases were very long; 25 years was the norm, accompanied by upward-only rent reviews: perfect, in theory, for the LDI strategy. By the time we get to the end of the performance measurement period the occupier market has decided that it wants short leases.
Many UK leases are too short to contain a rent review of any description let alone an upward-only mechanism.
The second observation runs counter to this and recognises that there is more debt secured against real estate now than ever before and a case can be made that interest rates are a key determinant to property prices. You will notice that the 2002-05 period is one of increasing correlation between interest rates and property values.
Thirdly, the whole period is characterised by the biggest bull market in bond market history, as shown in the next chart. Yields have fallen from 18% in the mid 1970s, when UK inflation was in excess of 20% to their current level of 4.5%.
The structural shift as inflation has been squeezed out of the economy may indeed be a poor period for us to undertake the analysis. A lower and more stable growth and inflation environment is mildly supportive of real estate being more, not less, sensitive to interest rate movements. This is because under these conditions, rental growth is lower and income makes up a larger proportion of the total return to property; and it is the income yield which is most sensitive to interest rates and gilt yields.
While the arguments outlined above also hold true for continental European property, some have identified that index-linked lease contracts could enhance the duration qualities of real estate. History, unfortunately, does not support these claims.
In chart 5 we again see low correlations between real estate and bond returns.
Indeed, the correlation between pan-European property and pan-European bond returns is negative (-0.29) between 1987 and 2006. Once again, although the income growth component from European property is attractive to pension funds the rate at which these cash flows are capitalised is very different. Many of the attractions of upward-only rent reviews are replaced by the index-linked nature of continental leases - and indeed, they are certainly capable of generating more income than UK leases for certain market segments3.
However, the capital market still has an eye on the substantial equity component of the underlying real estate and property yields reflect the property market fundamentals. The only thing which helps in an accounting sense is long-dated government bonds or swaps. Both are assets which are directly linked to the yields used to discount liabilities, not the vagaries of rental values and levels of new supply of office stock.
What about alternative property sectors? Alternative property sectors include the likes of nursing homes, hotels and student accommodation, a significant proportion of which are let to occupiers on long leases with fixed uplifts, often linked to inflation. Unfortunately, we have no performance data on these sectors with which to test whether they have been more correlated with bond returns. However, we suspect that although the income streams may be more stable and less volatile than other types of commercial property, the capital side of the equation will be subject to the same uncertainties.
So what role can property play within an LDI framework? The investment industry is increasingly turning to a model of LDI that has fixed income and swaps on the one side, and a return portfolio on the other, designed to deliver above market returns. Assets generally found in return portfolios are core equities, hedge funds, private equity, commodities and real estate, where the portfolio's risk budget is used to chase alpha in this diversified range of assets. This is where real estate can clearly fit within an LDI strategy.
Further, although lease structures (either upward only or index-linked) generally create income streams capable of matching inflation measures, the capitalisation of these income streams looks substantially different to the capitalisation of sovereign bond portfolios. This raises serious questions over the blind adoption of the asset class' total return delivery to match pension fund liabilities in continental Europe.
n its purest form, the income generation from direct property has been very capable of matching the growth in pension fund liabilities. However, it has not matched the changes in the rate at which those liabilities are capitalised. In order to be a good LDI investment it would need to do both. Direct real estate, while in theory a long duration asset (sensitive to interest rate movements), has behaved like a short duration asset (insensitive to interest rate movements) over the last 30 years.
In our view, property continues to play a valuable role in well diversified, mixed asset portfolios. It clearly has an important role to play in an LDI world also - in the alpha-generating part of the LDI portfolio. However, we do not subscribe to the view that direct real estate is a ‘perfect' liability matching asset.
1 FRS17 effectively means that pension scheme assets are measured using market values, liabilities are discounted at current bond rates and the surplus or deficit is shown in full in the balance sheet
2 Property Investment Theory, Macleary A, 1988. Appendix A contains the derivation of duration for real estate as developed by Charles WardProperty Investment Theory, Macleary A, 1988. Appendix A contains the derivation of duration for real estate as developed by Charles Ward
3 Turner, N. (2003) Lease styles and income growth - exploring the nexus. Paper presented at the IPD/ IPF Annual Property Investment ConferenceTurner, N. (2003) Lease styles and income growth - exploring the nexus. Paper presented at the IPD/ IPF Annual Property Investment Conference FRS17 effectively means that pension scheme assets are measured using market values, liabilities are discounted at current bond rates and the surplus or deficit is shown in full in the balance sheet Property Investment Theory, Macleary A, 1988. Appendix A contains the derivation of duration for real estate as developed by Charles Ward Turner, N. (2003) Lease styles and income growth - exploring the nexus. Paper presented at the IPD/ IPF Annual Property Investment Conference
Julia Felce is head of UK property research at Schroders PIM
Neil Turner is head of global property investment at Schroders PIM