Leverage - that is, borrowing to finance part of the acquisition cost of an asset and repaying the interest and capital on the loan out of the returns from the asset - is a commonly-used means of increasing potential returns to equity investments.
Leverage (also referred to as gearing) is especially common in real estate investment for two reasons. Firstly, real estate is a physical asset which generally retains some value even in worst-possible outcomes, and which can be used as security on loans.
Secondly, real estate investments often generate covenanted fixed or semi-fixed income streams which can be used as a reliable source of cash for making interest payments and, potentially, amortising loans. For this reason banks are often prepared to lend at relatively low margins on real estate investments, while investors often perceive little additional risk in taking on loans that can be repaid out of covenanted income.
The use of leverage in real estate investing amplifies both profits and losses, and thus increases risk as well as expected return. Since leverage increases both risks and returns, it can be directly compared with other means at a real estate investor's disposal which have the same effect.
These means could take many forms but one of the most important and fundamental alternatives to leverage is to buy riskier assets - these could be assets in less liquid or transparent markets, or in non-prime locations, or with weak tenant covenants, or with asset management challenges such as vacancy, short leases or refurbishment/redevelopment requirements.
Faced with a choice between leveraging and buying riskier assets, a rational investor who wishes to raise returns above that expected for core unleveraged investments would choose the course that generated less risk per unit of extra return. In order to see how this will affect investor behaviour, we need to consider how this marginal trade-off between risk and return is likely to behave as investors rise up the risk/return curve, either by leverage or by increasing exposure to underlying real estate risk.
It can be shown that if financing costs are independent of leverage, then the trade-off between risk and return from leverage is linear - ie as leverage employed rises, so the proportion between additional return and additional risk is constant. In fact this ratio can be calculated from the following formula:
ΔR/ΔS = (Ru - C)/Su
where Ru and Su represent the return and standard deviation respectively of an unleveraged investment and C is the cost of finance.
However, in the real world financing costs are very unlikely to be independent of leverage. Higher levels of leverage entail a higher loan to value ratio and a lower debt service coverage ratio. Lenders will compensate for this by charging higher margins on loans. They are also likely to charge higher arrangement fees and to specify more restrictive covenants (such as amortisation via automatic cash sweeps) which reduce borrowers' margin of manoeuvre and effectively increase costs further. As leverage starts to approach opportunistic levels, extra debt will likely be offered on a junior or mezzanine basis which will increasingly strip out the returns to equity.
As a result, the trade-off between risk and return through leverage is not, in fact, likely to be linear. As leverage rises, so the units of extra return that can be bought with a unit of extra risk will decline. The situation is represented graphically by line RuL in chart 2: the trade-off flattens as risk and leverage rise, and in theory at least could eventually turn downwards.
While it is relatively straightforward to compute the risks and returns associated with leverage, the risks and returns associated with buying riskier real estate are less easy to uncover. Organisations such as IPD collect data on returns for large numbers of individual properties, but this property-level data is not made public for confidentiality reasons.
However, at least some idea of the trade-off can be gleaned from higher-level data. Chart 1 shows historical risks (measured by standard deviation) and annual returns for offices, retail and industrial sectors in the four European countries for which there is a sufficiently long run of historical data to make it possible to put any faith in history as a guide to the future, plus the US and Australia. Although the eighteen data points seem to show only a weak upward trend, within each sector the results are a little more encouraging.
For offices, the trade-off is strongly positive - the slope of the line is 0.48 (i.e. for each additional percentage of standard deviation added to risk, returns rise by 0.48%) - and also significant (R2 = 0.62, t-statistic on the slope = 2.6). For retail the slope is 0.27 but barely significant. For warehousing, the slope is positive but insignificant. These revealed risk/return trade-offs are not out of line with the trade-off generally observed in all asset space.
Eagle-eyed readers may notice that within countries, the trade-off between sectors seems to be negative however; for example, a trend drawn through the three solid triangles representing the office, retail and industrial sectors in the Netherlands would clearly slope downwards. This is a surprising result, but it is beyond the scope of this article to delve further. Suffice to say that, on currently available data, much mystery still surrounds the nature of the trade-off between risk and return in real estate space. In what follows we assume it is an upward-sloping straight line (RuT in chart 2).
If an investor has a target rate of return that is greater than that obtainable from unleveraged core real estate (Ru), then whether he achieves this through leverage or through buying non-core real estate depends on which option gives him less risk per unit of extra return. If leverage is more expensive in terms of risk, then a rational investor will not use leverage and will achieve his target by taking on real estate risk.
If on the other hand the initial trade-off between risk and return from leverage is favourable, then initially an investor would raise returns by borrowing. However, as leverage increases so the trade-off to leverage deteriorates because of rising loan costs. When the trade-off is equal to that available in real estate - represented by the dotted line in chart 2 - any further additional returns should be obtained by buying riskier real estate. This point is shown by the red dot in chart 2.
What can this tell us? Unfortunately, our lack of knowledge about the true risk/return trade off in real estate means that it can tell us far less than we would like it to. In particular, we cannot at present say with any certainty what the optimal level of real estate leverage actually is for any given return target and set of financing costs. Nonetheless, we can draw a few tentative conclusions.
Firstly, chart 2 suggests that all investors should leverage core investments up to the optimal leverage point, only taking on non-core investments when their return target exceeds that achieved at this point. This is almost certainly inconsistent with observed practice, since under relatively loose financing conditions (such as is presently the case in continental Europe) investors with relatively low return targets often buy non-core assets instead of relying solely on leverage.
The model also tells us that it is not the case, as might be thought, that the amount of leverage that should be employed is positive so long as it has a positive effect. Leverage is suboptimal so long as the risk/return trade-off in real estate is superior to that from leverage; this will certainly be the case if there is negative leverage (ie, leverage reduces returns and RuL slopes downwards from the axis); but it could also be the case for positive but low leverage (ie RuL slopes upwards from the axis but at a shallower gradient that the real estate trade-off line). Again, this is probably inconsistent with observed practice since even under very tight financing conditions many investors still choose to use significant amounts of leverage.
Thirdly, and uncontroversially, it tells us that as financing costs rise, so the proportion of total returns that investors obtain from leverage will decline. However, it does not follow that leverage will decline as financing costs rise , since that part of returns that is still accounted for by leverage will require higher levels of leverage to achieve. Therefore the relationship between optimal leverage levels and financing costs might be highly non-linear.
Finally, although we do not know with any great accuracy what the risk/return is in real estate space, we can surmise that it is likely to be relatively stable over time.
By contrast, while we can calculate precisely the risk/return trade-off from leverage, we also know that it is likely to be highly unstable as financing costs change.
For example, if core real estate is expected to return 8% per annum with a standard deviation of 10% and borrowing costs are 7%, then the trade off between return and risk is just 0.1 and it is very likely the case that real estate risk is a more efficient way of buying returns and optimal leverage is zero. If however, financing rates fall to 4% then the trade-off is 0.4 and a significant amount of leverage is likely to be employed.
Consequently the optimal amount of leverage is likely to move around significantly as financing conditions vary. The fact that observed leverage does not vary anything like as much may well reflect uncertainty - quite sensible, as it turns out - as to what the optimal level of leverage actually is at any given time.
This article is a condensed version of Tyrrell N and Bostwick J Leverage in Real Estate Investments - an Optimisation Approach, in which these results are derived formally. Copies of this paper are available from the author.
Nick Tyrrell is head of research and strategy of the European real estate group at JP Morgan Asset Management