New Home Sales and Foreclosures

The drum beat of economic bad news continued on Thursday. In addition to record levels of continuing claims and further falls in new manufacturing orders, new home sales fell 14.7 percent in December, with large declines in every census region. The drop in sales pushed month’s supply up to a record 12.9. And, the median time for sale since completion rose to 9.3 months: over half of the new homes for sale in December were completed in March or earlier.

The report did not, however, contain only bad news; the number of homes for sale dropped a record ten percent. Inventories are continuing to adjust in the housing sector and apparently the rate of adjustment is accelerating (see my post on the topic here). As I noted previously, but for the record number of foreclosures, I would have expected the housing market to begin its recovery in early summer.

I have been wondering about how to measure the impact of foreclosures on the housing market. Very well done academic studies have failed to find a substantial effect of foreclosures on starts, new home sales, or house prices. (See for instance the recent study by Calomiris et al.) Although these studies control for many variables, by necessity they study foreclosures during relatively good times, times when the economic outlook is fairly benign and when the number of foreclosures is relatively low.

My intuition has always been that foreclosures are an independent negative factor for housing markets. In my view, foreclosures act as inventory in ways that are very similar to new housing: the houses are empty and need to sell. Proving this empirically is difficult as foreclosures do not happen in a vacuum; think of the housing market in a mill town when the mill shuts down, falling prices and high foreclosures.

With this release of new home sales this month, it occurred to me that the differences in behavior between new home sales and existing home sales might give us an indication of the magnitude of the effect of foreclosures. Foreclosure sales appear in the data as an existing home sale. Since both new homes and existing homes are affected by the same macro factors (employment, interest rates, financing availability) a change in the relative sales rate should be informative. An increase in the sales rate of new homes would indicate a relative oversupply of new housing and vice versa.

Take a look at the following picture. In the graph, I plot the ratio of existing home sales to new home sales from 1967 through December 2008. Amazingly, the ratio is relatively constant from the early 1970s through January 2006. This period is not one of stability. There are at least four recessions and as many housing market swings over this interval. Yet, the ratio remained solidly between 4 and 6 for more than 35 years.


It was not until 2006, when the first large wave of subprime foreclosures hit the market, that the ratio began to move upward, that the number of existing home sales began to outpace the number of new home sales. As foreclosures became a more acute problem, the ratio swung rapidly upward, reaching almost 13 in December.

If this increase owes to foreclosures rather than some other factor, I would expect that regions hardest hit by the foreclosure crisis would experience the greatest change in the ratio. The graph below compares the ratio for the West census region to the total shown above. While the ratio for the country has less than tripled, the ratio for the West, which includes the hard-hit states of California and Nevada, has more than quadrupled.

The sudden movement in the ratio is strong evidence that foreclosures themselves are having a large impact on the housing market over and above what would be expected from the downturn alone. However, I cannot use this data to give a numerical estimate of the impact. With essentially only one observation, I cannot map this swing into changes in house prices. I know that foreclosures are an important contributing factor in California and I know that house prices in California are falling. I cannot separately identify the shift in demand from the shift in supply. I can only say that the supply effect (foreclosures) is larger than at any time since 1967.