How do you manage a portfolio through a real estate cycle? Nick Tyrrell looks at ways to effect a transition with a focus on downside limitation
European real estate has enjoyed a period of exceptional returns in recent years, driven initially by yield compression and, more recently, a pick-up in rental growth. At the time of writing, however, the future looks appreciably less rosy than the recent past.
At best, returns must at some point come off their recent highs; yields cannot continue to compress forever, and rental growth has to revert to mean at some point. At worst, real estate as an asset class could be facing a significant downturn, driven by the direct (yield corrections, higher borrowing costs) and indirect (lower service employment and GDP growth) effects of the "credit crunch".
Whether returns merely subside to trend, or go into reverse, and what the timing of such a change will be, is unclear. Nonetheless, real estate cannot be bought and sold on a sixpence; and a real estate manager who does not think now about how to defend his or her portfolio against a slowdown will find it too late when that slowdown comes.
In an ideal world, a real estate portfolio would contain high-volatility assets during upswings, and low-volatility assets during downswings. Figure 1 shows this ideal situation; the resulting portfolio (in green) has both a higher mean return and a lower volatility than the high-volatility portfolio.
Of course, such a theoretical ideal is not achievable in the real world; it would require both perfect foresight and an ability to switch in and out of different assets instantly, neither of which is (unfortunately) possible in the slow-moving and informationally inefficient world of real estate. Nonetheless, it is clearly desirable that managers should tailor their portfolios so as to achieve a profile as close as possible to that shown in Figure 1.
How can this be done? First, we need to be more precise about what we mean by a real estate "cycle". Real estate prices are effectively set in two markets: the occupier market, in which rents are determined, and the investor market, in which the price paid per unit of rental income (or yield) is determined. These are separate markets with similar (but not identical) supply drivers and very different demand drivers, and it is therefore possible for a downturn in real estate values to be generated by either, potentially requiring different strategy responses.
Usually, this is not the case. Under normal circumstances the main reason for the investor market to weaken is a weakening occupier market, or at least an expected weakening, since lower rental growth expectations will lead investors to require a higher current yield to compensate them for lower capital growth. This said, circumstances are often not normal and in fact the two markets have been moving in opposite directions during the past few years, with investor markets booming and occupier markets - at least until recently - in abeyance.
Clearly, some of the drivers of investment demand over recent years have been structural forces - such as improving transparency and changes in institutional investment strategy - rather than cyclical. If these structural factors had not been present then the weakness of rents in the earlier years of this decade would probably have generated a rise, rather than a decline, in yields. Nonetheless, we need to be aware that this disconnect could continue; and indeed the "soft-landing" scenario favoured by many commentators at present sees a reversal of this pattern, with real estate investment pricing pushed down by the credit crunch but the occupier markets surviving more or less unscathed. In what follows we focus primarily on rent-driven cycles, but where investor market upsets require a different strategy response this is noted.
With this in mind, what are the steps a portfolio manager should take to adjust his portfolio as the cycle matures? We believe there are four principal tools.
The most obvious strategy is to shift out of more volatile markets and into less volatile ones. However, this is also the least easy to achieve, since it can only really be brought about by trading real estate, with all the attendant costs, lags and uncertainties. It may be possible to achieve a similar effect using property derivatives; we return to this theme later in the article.
In fact to execute this strategy successfully probably requires not only an ability to trade real assets rapidly but also a greater ability to foresee trends than the market (which most analysts would claim but only a minority can, by definition, achieve). If the market already expects a real estate slowdown then it will be making the same trades and pushing relative yields down in less cyclical markets and up in more cyclical ones.
Nonetheless it is worth considering which markets, generally speaking, are likely to be less or more volatile than the average. By property type, it is well documented that offices and (to the extent it follows the commercial property cycle at all) the residential sector are more volatile than the retail and warehouse sectors.
Figure 2 shows the standard deviation of annual prime office rental growth over the period 1991Q1-2007Q1 for a selection of European cities. The data is quarterly so we have 65 observations covering approximately two real estate cycles - probably enough to place reasonable confidence in these results (unless there has been a significant structural break during the period). It's worth noting that the standard deviation measure tends to make the differences look less dramatic than they really are; consider that for Brussels (stdev = 5%) the probability of rental growth in any one quarter being more than 10% above or below the mean is less than 5%, while for Madrid (stdev = 21%), it is greater than 60%.
The markets with the lowest volatility here tend to be cities whose occupier base has a high government component and where supply is relatively responsive. By contrast cities with inelastic supply, or where supply is slow to respond, and which are dependent on cyclical industries such as telecommunications or financial services, tend to be more volatile. The results for Berlin, which fits neither of these patterns, we believe to be an aberration caused by the huge structural changes that city has undergone over the past 15 years.
Of course, the results for the retail sector would be different from those shown in Figure 2. It seems unlikely that there would be much difference in the cyclicality of different European warehouse markets, although older-style stock located close to manufacturing hubs, as opposed to distribution nodes, is likely to suffer more in a downturn since in a weak market occupiers will have the luxury of being able to ignore second-best solutions.
It is also worth considering how secondary and peripheral markets are likely to play out against prime city-centre locations. Generally speaking, supply is less constrained in secondary markets, resulting in less volatile rental growth (though there are exceptions, since some secondary markets are highly depending on a single industry - Toulouse being a case in point). However, it is also generally the case that yield movements in secondary markets are more cyclical than in prime markets, since investors tend to only seek out these markets when prime markets are already fully invested and appetite for risk is healthy (to put the point in bond-speak, spreads tighten in strong markets). It is therefore unclear prima facie whether secondary markets perform better or worse during downswings; further research is required here. However, generally speaking an occupier market-precipitated downswing should see investors favouring secondary locations, whereas an investment market-driven slump should see them shunned.
The second strategy concerns laggards and leaders. Figure 3 shows average office rental growth in the 21 markets shown above, plus two of those markets, Rome and Copenhagen. It is clear from the figure that, historically at least, Copenhagen has tended to lead the European cycle, while Rome has tended to lag it. One simple way of measuring this is to look at the number of quarters between the peaks and troughs of the overall cycle and those of the individual city cycles. For example, the first trough in rental growth in Rome occurs one quarter after the European trough; the peak occurs four quarters after the European peak; and the second trough lags the European trough by five quarters. The average lag in this case is (1+4+5)/3 = 3.33 quarters.
Figure 4 shows the average historical lag or lead relative to the European cycle for 19 markets (Vienna and Moscow were excluded from this analysis as they apparently bear no relation to the overall cycle). The current upswing in rents is also excluded since (hopefully) we have not yet reached its peak; however, this would probably change the results somewhat, especially in the case of Paris and Madrid which have lagged historically but are clearly leaders in the current cycle. Determining leaders and laggards is as much a question of analysing the specifics of the current cycle as relying on history.
It might seem that an easier way of switching weights between markets in order to take advantage of less cyclical or lagging markets as the cycle matures would be to keep the holding of real assets more or less constant and adjust risk exposure using derivatives. This would certainly be appealing if the portfolio contained an asset located in a "bad" market but to which the manager was convinced that he could add value through asset management.
This said, how real estate derivatives price is still not entirely clear. What is clear is that if more than a few managers call a downturn in a given market, demand for buy-side contracts will fade and supply of sell-side contracts surge, potentially resulting in a margin swing that could cancel out perceived changes in relative returns far more rapidly than pricing changes in the direct market. In fact this is precisely what has happened recently in the UK market, with a number of managers holding derivatives as a substitute for direct holdings finding them to be far more volatile than the index on which they are based. True, anyone who had sold the index short prior to the repricing would have done extremely well out of it, but to execute this strategy successfully would have required an extraordinary ability not only to read the real estate markets ahead of competitors but also to predict the vicissitudes of the swap market. If the UK property market in fact turns out to perform less badly than the swap market currently implies, this gain will rapidly reverse, and anyone who was not in the game from the very beginning could see their short position amplify, rather than mute, the effects of a UK downturn. This is not to say that derivatives cannot be used, with caution, to mitigate the effects of cyclical downturns; but their use is not as simple as their early proponents sometimes suggested.
The third strategy available is to shift from market growth and asset management plays to stabilised assets with long lease terms. In a strong market, it makes sense to buy assets with vacancy (since it should be straightforward to add value by leasing the vacant space), under-rented assets, and assets with near-term lease expiries or reversions. During downturns, however, investors would rather sit on long leases and not be exposed to the market. To put it another way, investors should seek out equity-like investments during upswings and bond-like investments during downswings. Among other things this will result in a greater proportion of total returns being delivered in the form of income as opposed to capital appreciation, which, while reducing the capacity for spectacular gains, will also reduce downside risk.
At first sight this may seem as difficult to achieve as switching between markets; however this need not be the case if the stabilised assets held during the downturn are the same as the non-stable assets held during the upturn. By preferring leasing and reversion plays during an upswing over simply buying into rental growth, a manager can ensure that his portfolio self-stabilises as the top of the cycle approaches.
Finally. leverage is often used by portfolio managers to amplify returns. However, it also magnifies losses and thus risk. A rational manager will use more leverage early in the cycle, and reduce borrowing as the cycle matures. This can be achieved by raising extra equity or by selling assets and using the proceeds to pay down debt, but it need not be; simply holding assets through a rising market should see leverage reduced through the denominator effect of rising gross asset values. Again, therefore, a carefully managed portfolio may well have a degree of self-stabilisation built into it.
All of the above may seem little more than common sense. Yet it is not difficult to find examples of investors and managers who break several of the above rules by being unwilling to switch out of sectors and markets that have served them well in the past; by relying on rental growth, as opposed to leasing, during upturns (and very often getting into markets just as that growth is coming to an end); and by continuing to use maximum leverage as cycles mature or, even worse, increasing leverage in an attempt to shore up returns. Nonetheless, if there is a real estate downturn in Europe there is no doubt that the managers who come out of it with their heads held highest and their reputations most intact will not be those who ride the cycle but those who succeed in baulking it - a paradigm change compared with recent years.
Nick Tyrrell is head of research and strategy, European Real Estate Group, JP Morgan Asset Management