Bottom Line

A historical analysis of U.S. home prices suggests that many U.S. housing markets are no longer overvalued and are gradually approaching more sustainable valuations. Potential homeowners need not be as cautious today as compared to just a few years ago. Still, when viewed through the lens of long-term investment cycles, home prices in most markets have a bit more downside and are unlikely to experience substantial price increases anytime soon.

Preview

This is the third and final issue of a three-part series on cycles as a long-term investment thesis. Here are links to the first two:

• Long-term stock market cycles, September 2009
• Long-term gold cycles, October 2009

I have written about housing markets many times over the last few years. Here are links to a few of them:

• More on housing: price-to-rent ratios, December 2006
• CME housing futures, April 2007
• Why home prices will continue to fall, February 2008
• Price-to-rent ratios, May 2008
• Slow-motion housing cycles, February 2009
• Tales from the housing trenches, June 2009

Housing markets go through investment cycles just like other asset classes. Sometimes homes are cheap. Other times homes are expensive. A historical analysis of price-to-rent ratios suggests that many U.S. housing markets are approaching sustainable valuations. In most markets, potential homeowners can relax a bit about the risk of buying an overvalued home. That being said, housing as an investment class is still likely to underperform for a few more years.

Real Estate Cycles

The most famous measure of long-term home prices in the United States was constructed by Yale economist Robert Shiller, published in the NY Times in August 2006:

Shiller's index was constructed using a "constant-quality" methodology, meaning that market transactions of the same homes were examined over time to estimate average price appreciation. The time series above represents an average price level for the entire United States. Updating Shiller's graph through today using the S&P/Case-Shiller home price index puts the 2009 data point in the 135-140 range (that's over a 30% decline from 2006).

When I look at the graphic above, I see oscillating periods of rising and falling home prices. Using the data from Shiller's home price index, I take annual price changes and plot 2-year and 5-year moving averages below:

The smoothing makes the up and down cycles more obvious. Perhaps the use of the word "cycle" here is a bit strong in this case - I don't see any obvious pattern in length or amplitude, just a general tendency for rising prices to follow falling prices, and vice versa.

But why should inflation-adjusted home prices fluctuate around a long-term trend? The answer, in my view, is the relationship between prices and rents. In the long-run, prices and rents should grow at roughly equal rates. Inflation as a whole has a pretty high correlation with rents (after all, Housing makes up roughly 40-45% of the CPI, with Rent of primary residence plus Owners' equivalent rent together comprising roughly 30%).

A more accurate way to understand housing cycles over time is to examine price-to-rent ratios. A price-to-rent ratio is the market price of a home divided by its gross annual rent. For example, if a house sells for $360,000 and rents for $2,000 per month, that's a price-to-rent ratio of $360,000 / (12 x $2,000) = 15. When viewed in a historical context, price-to-rent ratios provide a rough estimate of whether housing is currently cheap or expensive.

Consider the following: how was it possible to know that home price in certain cities were "too high" back in 2006? Here is a chart of the Los Angeles price-to-rent ratio that I constructed and wrote about in December 2006:

Notice the divergence between the growth rates of prices and rents from 2000-2006. The same type of divergence occurred in the late 1980s and resulted in a 40% inflation-adjusted price drop. Given the long-term view of investment cycles, it was unsurprising to expect a similar drop in this cycle.

Here is the price-to-rent ratio for Los Angeles with updated data through August 2009 (click here for more detailed information on how these series are constructed):

Consider the same charts for New York City. First, the version from December 2006:

Now with updated data (notice the divergences between prices are rents in the late 1980s and the early 2000s):

New York no longer looks as overvalued relative to history as it did in 2006. Still, the early 1990s experience suggests slowly deflating valuations for a few more years.

The price-to-rent ratio is useful for identifying real estate cycles because you can see home valuations for a given city in a historical context. In theory, each city should have a different equilibrium price-to-rent ratio that depends on long-run factors like population growth, housing supply, etc.. Adjustments could also be made for mortgage rates and changes to tax law. I won't get into these issues here, but suffice it to say that the simple price-to-rent metric was useful, at least, for identifying overvalued housing markets in 2006.

[Technical note: Above and beyond changing variables like mortgage interest rates, there are other data-driven differences between the price series and rent series (most notably that they measure different housing stocks) that might prevent these particular price-to-rent series from having a flat trend over time. Still, a doubling of the price-to-rent ratio over just seven years should have caused even the most serious data skeptic to raise an eyebrow.]

Of the twenty-five cities in my sample, seven had price and rent divergences in the late 80s. Price corrections followed these divergences in all seven cases. The table below shows the rise and subsequent drop in the price-to-rent ratio for these cities:

The price corrections in the early 1990s took an average of 7 years to adjust. Given that we are just 3-4 years into this housing down cycle, it seems unwise to expect rising prices for at least a few more years.

Current U.S. Housing Markets

The following table displays the Mid-1990's Valley, the Mid-2000's Peak, and the current price-to-rent ratios for all 25 cities:

The cities are ordered based on the current price-to-rent ratios relative to the Mid-1990's Valley (the "Current" column). The data suggest that while some markets may have some downside left, most markets are closing in on sustainable valuation levels. For reasons described above, I'm not expecting that all cities will necessarily reach their Mid-1990's Valleys (for example, I am skeptical about the reliability of the Honolulu series, which seems to have an obvious uptrend over time), but more caution is warranted for cities higher up the list.

[Technical note: I calculate price-to-rent ratios over time by dividing the price index by the rent index for each city. Because all indices are reported without units (i.e., they are generally constructed so that the index=100 in a particular year), I calibrate them to their June 2007 levels using an article from Fortunate Magazine: "Where housing is headed. To the extent that the price-to-rent ratios in for June 2007 are accurate, the charts above should be reliable for making cross-sectional comparisons across cities. However, I would put more emphasis on the relative values over time for one city rather than making level comparisons across cities.]

I am especially curious to see how price bottoms play out in cities like Philadelphia, Boston, and New York. These cities still seem modestly overvalued according to the valuation exercise above, yet they are beginning to carving out price bottoms just like the other markets. My guess is that these markets will continue to be weak for a few more years, but that the major price declines have already occurred.

Housing is no longer expensive but it is not especially cheap either. Potential buyers need not rush into the market, as valuations are likely to get even better. That being said, housing values aren't too bad these days. In most markets, I would not discourage my friends from purchasing a home if the timing was right for them.

For investors, the time is approaching to put some money to work in real estate. Not quite yet, but soon. If history is any guide, the next 2-5 years should be a good time to accumulate real estate assets. Personally, I will be looking to scale in over the next few years when the right opportunities present themselves.

Putting It All Together

Todd Harrison of Minyanville writes:

"I'm reminded of the late Bennet Sedacca's favorite chart, the graphical representation of denial, migration and panic:

As I snuck a peek at that psychological continuum, I wondered where we are on the curve. The answer, I suppose, is a function of time; as with most things, there are short-, medium- and longer-term horizons. The trick to the trade—this, or any—is making sure your risk profile and time horizons are in sync."

- Todd Harrison, Where The Market Will Go, October 28, 2009

It is the nature of markets to oscillate from one extreme to another. The most difficult part of investing is to invest when others tell you not to and to refrain when others tell you to buy. Wasn't it Warren Buffett who said to Be fearful when others are greedy, and be greedy when others are fearful?

But why should cycles exist in the first place? At least for housing, Harvard economist Ed Glaeser explains:

"There are two reasonable approaches to predicting the future of housing prices. Approach No. 1 is the housing equivalent of 'technical analysis' for stocks, and it assumes that the key statistical regularities of housing price changes — long-term reversion to the mean and short-term momentum — will persist in the future. [emphasis in original]

...The second approach to predicting housing prices uses economics. It is the housing market equivalent of 'fundamental analysis' in the stock market...The best that can be done is to compare prices with those in the unfettered markets, and ask whether the price gaps between these markets look sensible."

- Ed Glaeser, How to (Sort of) Predict Housing prices, July 14, 2009

In other words, just two simple ingredients produce long-term investment cycles: (1) long-term mean reversion, in which, to paraphrase John Maudlin, "an unsustainable trend will not be sustained," and (2) short-term momentum, in which the direction of prices in the last period tends to predict the direction of prices in the current period. Mix these two ingredients together, sprinkle on a little leverage, and BOOM, there you have investment cycles.

In this three-part series (September 2009, October 2009, November 2009), I have reviewed investment cycles over periods of five to ten years or more. The greatest mispricings, and thus the greatest opportunities, occur over these time horizons.

Taking advantage of long-term investing cycles means rotating out of asset classes which are overvalued and into asset classes which are undervalued. I first covered this concept in June 2006: Asset class rotation. Here is the main chart from that newsletter, with updated data:

Viewed through the lens of asset class rotation and long-term investing cycles, I am still currently leaning towards hard assets like gold and commodities and away from stocks, bonds, and real estate. But the tides are shifting. I no longer think that hard assets represent a value proposition but instead represent more of a momentum proposition. At some point the unsustainable trends will not be sustained. I will be watching.

Conclusion

To summarize, executing an investment thesis based on long-term investment cycles requires two specific action steps:

1. Try to identify investment extremes, when certain asset classes are particularly cheap or expensive based on historical valuation metrics. Discard the notion that prices always reflect fundamental value. Instead, assume that some assets may not be priced correctly and try to identify them. Avoid expensive assets and buy cheap assets.

2. If no asset classes seem to be priced at extreme valuations, then overweight assets that are coming off of previous peaks and underweight assets that are coming off of previous valleys. Price momentum, whether upward or downward, can last for years.

As I've said before, I don't think we are at a market extreme where the investing strategy is obvious. This warrants some degree of diversification, with an orientation towards following the cycles described above.