Lowest existing home sales in years

Yahoo! Finance is reporting that existing home sales have fallen to a nine year low. This means that sales have returned to the level of 1999. Seasonally adjusted sales remained below 5 million for the first time this decade. You can get the data from the National Association of realtors as a PDF here and as an MS Excel sheet here. One way to look at this is that, after now reaching multi-year lows, housing must be do to reach a bottom soon. The other view is to view 1999 as a pretty good year in the housing market (it was) and to argue that even with the long fall, housing has only just returned to long-term trend sales figures. This view holds that while we may have returned to “normal” conditions, we have not yet begun to work off the excesses of the boom. Thus, housing is likely to continue to head lower and underperform other assets, likely for years with yet lower prices required to clear the market.

Naturally, the irrepressible Lawrence Yun and the National Association of Realtors take the first (more optimistic) view. They have chosen to spin the numbers and call yet another bottom to the housing market. They do this by simultaneously focusing on the (un-annualized) month over month decline, which is less than ½ of 1 percent and simultaneously talking up phantom demand that is simply waiting for a “catalyst” – in this case the increase in the ceiling of conforming mortgages, which will enable homeowners with mortgages larger that $417.000 to obtain a mortgage backed by Fannie Mae or Freddic Mac.


In fact, if anything this report confirms that housing will continue to decline for the foreseeable future. The 0.4% decline over December represents a pretty steep decline year over year. And, in the datasheets provided, it becomes clear that housing is now much slower than last January: the year-over-year comparison (the only one which matters) shows a 23% decline in the number of homes sold and a 4.6% decline in median home prices (with a smaller decline in average home prices).

What the data say in clear terms is that even significant price discounts (a 5% fall is huge: final prices exclude other incentives like closing costs paid by the seller or upgrades to kitchens, bathrooms and the like) buyers are simply unwilling to do deals. Inventory remains extremely high. After dropping over the holidays, units on the market have increased and inventory again surged above 10 months supply. More houses are likely to come on the market in the next few months as the foreclosures mount. More inventory and more sales by banks rather than homeowners, suggest that prices will continue to fall and sales will continue to decline.

Buyers will naturally be reluctant to purchase for the same reason that they were so desperate to purchase during the boom: the effect of leverage. If one is able to put down 10% on a home and experience 5% appreciation per year, the leverage provides the buyer with a 50% gain. During the boom, prices regularly increased by 11-15% per year providing an annual gain of 100% or more. With even an even lower or no down payment returns became magnified to even greater extent.

But let us return to that same buyer today. A buyer who can qualify for a conforming mortgage, and who has 20% down is staring at a first year loss of 25%. If the trend continues for two years that same buyer is out 9.75% of the purchase price and has succeeded in wiping out nearly half his capital. Better to keep renting and hold onto the principal. A buyer with 10% or less to put down would have his entire equity wiped out in the first year. The NAR is hoping that buyers who already own homes and who are looking to trade up to a substantial home, $500k or more, will return to the market and generate more sales at the high end. If this were to happen the owner would likely list his current home, which means that no inventory would be removed. The problem remains without first time homebuyers there is no one to clear the inventory and right now there are no first time buyers.

Going inside of the numbers

Let’s move beyond theory and see what the numbers suggest.

This chart (created by the Center for Economic Policy Research using data from the US Census Bureau) shows the vacancy rate for home (both rentals and owned homes) is a clear indication of the source and magnitude of the current problem: there are too many residences and not enough families to occupy them. With a sharp reduction of interest rates in 2001-2003, rental vacancies soared.


The vacancy rate, which had remained fairly stable between 7.5 and 8% decade, leaped to 11% in two years. The decline was both so strong and so sudden that rents actually declined (lower rents suggested deflation in the CPI calculation, which reinforced the argument to keep rates low!). At the same time, home vacancies remained quite stable, virtually a flat line. With many renters shifting to being homeowners, we might expect home vacancies to be falling. That this did not happen shows how rapidly new home construction was running. By 2004, the rush was on to condo convert every apartment building in America. This finally brought rental vacancies down (by eliminating rental stock from the market - not by finding new renters) and the rise in vacant homes began. Many of these units were purchased in anticipation of future price increases and, as we can now see, nearly 3% of owned homes are vacant. But selling or renting these homes will not fix the basic problem: the surplus of residences. We may see arbitrage back and forth between rental units and owned units (condo conversions or rental of purchased condos). It doesn't matter because vacancy rates in both rented and owned homes are much higher than in 2000. This problem will not be resolved until positive absorption can be restored. Even then, the process will take years.

So how low can housing go?

This is the big question. Let's see if we can get inside some of the numbers to make a sensible projection. Before we do, I want to point out that housing corrections are slow to unfold and resolve. The negative impact of leverge and the stress that the losses put on the banking system require years to resolve and sometimes decades. Japan has yet to recover from the property boom of the 1980s.

I take as a principle that housing sales is a mean reverting series with a bias toward a long term rise. Population and household creation in the US are growing, albeit slowly, which leads to a rise in the need for homes (and generally, though not necessarily, owned homes). If the series is mean reverting however, we should expect that a series that experiences a spike outside of the normal range to experience an offsetting dip. So, where is the market now? That depends on how far back you look.

There are two basic series of housing data. The first is existing home sales. (These are the numbers reported by the NAR above). Unfortunately, the NAR charges for data beyond 12 months. (If anyone has the data series going back a ways and is willing to share, please contact me). The second series is the New Home Sales, which is available from the US Census Bureau. I used the publicly available dataseries from Economagic. The publicly available information is excellent and there are some free chart generators (gif and pdf) There is also a premium subscription available which allows for more data series and more customization of the charts and a direct export of the data series to MS Excel for analysis.

Back to the data. If I were the realtors, I would want to use this chart. Yes, it paints an extremely grim picture of new home sales since the middle of 2006. However, the chart suggests that we have returned to the level of home construction prevalent before the housing bull market started around 1995. Furthermore, new home sales are well below the long germ trend presented here. Given the weakness of new home building in the early 1990s, it could be reasoned that we are now sopping up excess demand and are poised to make a comeback.


The problem with this chart is that the long-term trend is broken, but there is no comparable data series to use to estimate how far this knife is going to fall. Let's take a longer term view with a goal of establishing a) how long housing markets stay down and b) how much of a decline in sales activity we tend to experience and how much of a decline in prices we tend to experience. With that, let us see if we can figure out where we are in the housing cycle.

The data series are available back as far as 1964. If you import the data to Excel and run an Ordinary Least Squares (OLS) regression, you find that the best fit is a polynomial (i.e. not a linear) trendline. Excel proposes this graph:

I find this graph amazing because it shows two things. First, that the housing boom of the 1980s, which peaks in 1987, only represented a return to trend (the peak falls right on the trend line). The impact of the financing crisis that resulted from the aftermath the rapid surge in sales (an increase of about 62%) led to a 10 year period where sales failed to re-attain the volumes at the peak. The trendline also suggests that we have only just returned to the trend line. To put it another way, the decline from 2005 to late 2007 was only the return to long-term trend activity. As a mean-reverting series, it can be seen that we are only at the beginning of the backing and filling necessary to work off the excess.

In fact this peak seems to show more similarity to the 1970s boom, which was driven primarily by the massive growth in household expansion from the Baby Boomers. There has been some discussion of the potential for new sales from the “Echo Boom”, but as noted above, first time buyers have no incentive to enter the market.

One further point, this trend line is quite distorted because the data series is short. If we had more historical data, we would likely see a flatter and more linear curve. This would serve to reduce the amount that sales in the 1970s were “above trend” but would also INCREASE the amount that recent sales are above trend. It is possible that we have YET to return to trend, though I think we are probably close. That chart would look more like this:

So how long are we likely to wait?

I believe that sales are likely to continue to decline over the next 18-24 months. Prices will likewise continue to decline, but that by 2010-2011 we are likely to reach bottom. Most of the earlier peaks and troughs are 4-5 years apart. This would put the bottom somewhere in 2009-2010. This boom is larger than the previous ones, so the down-cycle could be longer. However, the data seem to suggest that the decline is likely to be much sharper. In other words, the fall will be much greater, but will occur over the same time period so the curve will be much steeper.

What can we expect on prices?

Here the news is even more sobering than on volumes. Home prices soared very, very far about the long-term trend. Ironically, one of the usual appeals of real estate investing is that price levels tend to increase at very consistent rates, about in line with wages (which are the big limiter on the amount people can borrow). Even the surge in prices of the 1980s is a comparable mild and look at how long prices remained below trend thereafter. After 1997, however prices began accelerating.

This chart suggests that prices must continue to decline, at least another 5% just to return to trend, but that prices are unlikely to begin rising again once the trend is reached. In fact, it is quite likely that prices will undershoot similarly to overshooting. If that were to happen, prices would have to fall 20-25% below trend and remain below trend until 2012-2013 at a minimum. From here, then, we should expect a 25%-30% decline from here. This means that unless you can get a deal, or have more than 25% to put down, or you can be quite sure that you will not need to move in the next 6-7 years, or you have the ability to ensure adequate cash flow (including a cushion should rents begin falling again) investing in real estate is a losing proposition.

If you already own, but have adequate cash flow and don’t mind just collecting rent checks, you don’t need to sell. If you need to sell, it might be better to do it now rather than risk braving the market 12-24 months from now.

I know many people think the market will come back soon. I still believe that the only true analog to this housing phenomenon is Japan from 1989-2004. After a similarly foolish bubble in asset prices, asset prices declined for 15 years. It probably won’t be that bad: regulators in Japan were painfully slow to force banks to clean up their balance sheets and most banks in the US seem to be moving with reasonable speed. But it is worth noting that every quarter involves yet another massive charge from UBS, Citi, JP Morgan and now Credit Suisse.

If you have cash and credit, you can start looking, but you MUST focus on cash flow. Capital appreciation will not happen for years. A 10-year IRR calculation should assume NEGATIVE capital gain. Thus, cash flow returns must be quite adequate. This is also necessary to ensure that you will not be forced to sell the property in the event of a decline in rental prices (which may happen if many other investors follow similar bottom-feeding strategies). Finally, consider if you have the resources to upgrade and remodel in the event that you find your neighborhood becomes competitive.