The slow mark

I spent fifteen years in real estate before I read any of the papers in this chapter, and for most of those years I believed what the industry believes about itself. Property was steady. It did not swing the way stocks did. A building held its value while the market panicked, and the appraisal proved it every year. The five papers here took that belief apart one piece at a time, and the strange thing is that all five still leave property as a reasonable investment. What they show is that the way it was measured hid most of its risk.

The chapter follows a single thread. House prices trend, and nobody can trade against that. Appraisals lag, so the reported risk is a fraction of the true risk. The cap rate is an expected return quote, and it moves. Public and private property are the same asset at two prices. And the long history of returns is real at the level of the index and shrinks by more than half by the time it reaches an owner.

Case and Shiller, 1989

Karl Case and Robert Shiller ran the efficient-markets test on houses, the largest asset most families own, and watched it fail. Using repeat sales of the same houses in four American cities, they found that a rise in real house prices in one year tends to be followed by a further rise the next year, of roughly a quarter to a half the size, and a fall by a further fall. Future returns could be forecast from public information: rents, construction costs, incomes. In stocks, that would be arbitraged away in a week.

In houses it persists, and the authors said why. You cannot sell a house short. Buying and selling one costs several percent each way. So the momentum is forecastable and unexploitable at the same time, an inefficiency without an arbitrage. The index they built to run the test outlived the test. It became the Case-Shiller home price index, and later the futures contracts written on it.

For an owner the lesson is a slow one. Property moves in multi-year tides, and two good years are no signal of a top, nor two bad years of a bottom. The edge is in refusing to buy late in a run, and in refusing to value what you already own as if the last two years of rises will continue. Two numbers tell you where in the tide you are. Divide the price of a house by the yearly rent it would earn, and divide it by the yearly income of the people who buy houses like it. Both ratios have a normal range for each city, and you can look it up. When they sit far above their range, prices have run ahead of what rents and salaries can support, and you are late in the tide. When they sit below, you are early. Compare the ratios, never the price on its own.

Geltner, 1991

David Geltner explained why institutional real estate looked so safe. Its returns were reported from appraisals, and an appraiser, asked what a building is worth, starts from what it was worth last time and adjusts. The reported index is therefore a moving average of the truth. Volatility is understated. Correlation with stocks is understated. Turning points show up quarters late. Geltner built the model of that process and, more usefully, the inversion: treat the reported series as a filtered version of reality and undo the filter.

The follow-on work supplied the numbers. The true volatility of institutional property runs at roughly double the reported figure, and its exposure to the stock market rises from near zero to clearly positive once the lag is removed. Every allocation of the 1980s and 1990s that put money into property because it offered stock returns at bond risk was built on the artifact Geltner diagnosed.

The episode is the one I lived. Listed property companies peaked at the start of 2007 and fell by more than half over the next two years, bottoming in March 2009. The appraisal index of institutional property peaked a year after the listed companies had turned, in early 2008, and kept falling until early 2010, a year after the listed market had already bottomed and begun a recovery of more than a hundred percent. Same buildings. One price was marked daily and the other by appraisers, and the appraisers took two years to write down what the market had already done.

Where this bites hardest is debt. Loans against property are sized on appraised values, and appraised values fall late. So the covenant does not breach when the market turns. It breaches a year later, all at once, when the appraisals catch up and the lenders have already stopped lending. Size property debt to the fall a daily mark would have shown, never to the appraiser’s history.

Plazzi, Torous and Valkanov, 2010

The cap rate, the building’s net income divided by its price, is to property what the dividend yield is to a stock, and Alberto Plazzi and his colleagues did for it what the dividend yield papers had done for equities. They split the cap rate into two things it can be telling you: that expected returns are high, or that rents are expected to grow. Using transaction data across American cities and four property types, they found that for apartments, retail and industrial buildings the cap rate forecasts returns. A high cap rate means a high expected return, and a low one means a low expected return. Offices behave differently: their cap rates say more about future rent growth than about returns. Both the expected return and the expected growth move through time and differ across cities.

The practical reading is short and it corrects a habit the whole industry has. A low cap rate is treated as a mark of quality. Mostly it is a mark of low expected return. The total return on an unlevered building is approximately the cap rate, plus rent growth, minus the capital you have to spend to keep it, plus or minus whatever the cap rate does while you own it. That last term is where cycles live. When cap rates have already fallen a long way before you buy, the seller has collected that gain. You are buying a building whose price has already risen, at a starting yield that is now low.

Two rules follow. Read every cap rate net of vacancy, management and recurring capital costs, never the broker’s gross figure, and put it in the same table as your expected returns for stocks and bonds. And write down at purchase which term the thesis depends on, the yield, the growth, or a further fall in cap rates, because only the first is in your hands.

Pagliari, Scherer and Monopoli, 2005

Over the 1980s and 1990s, listed property companies returned about five percent a year more than private appraised property, and the industry concluded that the listed companies were stocks and private buildings were real estate, two different things. Joseph Pagliari and two colleagues tested the claim by making the two comparable. They reweighted the private index to the same mix of property types as the listed one. They removed the leverage from the listed companies. They unsmoothed the appraisals using Geltner’s logic. After the three adjustments, the returns and volatilities of the two were statistically indistinguishable, and in the later part of the sample the gap had shrunk to about sixty basis points.

Public and private property are one market at two liquidity prices. The listed price moves first, usually by four to eight quarters, and the private mark follows. In a crisis, the difference in drawdown between the two is a difference in timing. The risk is the same. The choice between them is about liquidity, control and fees, and not about which is the real asset.

For a family with buildings, the listed market is the daily mirror of its own balance sheet. Use it to mark the direct holdings in the risk ledger. Use it to get sector and geographic spread without new lockups. And when listed property trades at a deep discount to private values, the cheapest building available is usually the one inside the listed company.

Jordà, Knoll, Kuvshinov, Schularick and Taylor, 2019

Five economists built the first long history of housing returns, rent plus price change, for sixteen countries since 1870, and set it beside stocks, bonds and bills. The headline was striking. Housing and equities had earned about the same real return, around seven percent a year, and housing had done it with much lower measured volatility and much lower correlation across countries. On those numbers property was the better asset.

Then came the correction that makes the chapter complete. Other researchers went to the level of the individual property, using the account books of Oxford and Cambridge colleges that had owned buildings and farmland for most of the twentieth century, with every rent, every repair and every sale recorded. Gross income yields were near five percent. Real rent growth was near zero. After all costs, the net real return was about two percent a year for residential property and about four and a half for agricultural land, with large differences from one property to the next. Work on Paris and Amsterdam found the same.

Both results are right. The seven percent is what a costless, nationwide, perfectly diversified index of housing earned, and nobody can buy that index. The two to four percent is what an owner earns after the roof, the taxes and the empty months, on the handful of properties they actually hold. The measured low volatility is largely the smoothing and aggregation the earlier papers described. Owning property in several countries does spread your risk. Owning property does not make you safe. It only looks that way because nobody prices your building every day.

What I do

I own real estate three ways: buildings I hold directly, listed property companies, and private equity real estate funds run by the large global managers. Across the three I spread the holdings by property type and by location. Everything is held free and clear. There is no debt against any of it, so the problem of the late mark and the late covenant that I described above is one I chose not to have. I compare the performance of the whole book with the listed property market, because that is the price my buildings would fetch if someone priced them every day.

Every building is underwritten on its net cap rate, after vacancy, management and the capital it will need, and I write down at purchase which of the three terms the return depends on. And I measure what the properties actually earn at the level of each property, net of everything, the way the college account books did. After all the costs, my buildings earn a few percent a year above inflation, and in fifteen years I have never seen one of them earn the seven percent the index reports.

Three habits

Do not trust the appraisal. Look at what listed property companies have done in the last year, because your buildings have moved by about the same amount, whatever the appraiser wrote. Use that number when you measure your risk, and if you borrow against property, borrow against that number and not against the appraisal.

Read the cap rate as a return quote. Net of every recurring cost, it is your starting expected return, and compression before you buy is return already paid to someone else.

Work out what each building really earns you. Take the rent, subtract the taxes, the insurance, the management, the repairs and the months it sits empty, and divide what is left by what the building is worth. That is your return. Do it for every property, every year. The index says seven percent; your own number will be lower, and it is the only one that matters.

Gala: the price of a building goes up and down as much as the price of a stock. You just do not see it, because nobody tells you every day. Never mistake not knowing the price for the price not moving.