Article

August 2013

The relationship between the housing and labor market crises and doubling up: an MSA-level analysis, 2005–2011

(continued...)

In terms of the timing of the housing and labor market crises, figure 1 shows the well-known story for the United States: the start of the housing crisis, as defined by the fall in housing prices, preceded the downturn in the labor market. This is also the sequence of events for a number of the largest, but not the majority of, MSAs.

MSA-level picture: an overview. Figure 2 illustrates annual trends in employment and housing conditions for all 353 MSAs (unweighted) for the period 2005–2010. These trends closely match those for the nation (MSAs weighted by population size) reported in figure 1. Most notably, figure 2 again shows that the housing crisis started first, followed by the labor market crisis. The top panel of table 2 provides annual statistics for 2 specific years shown in figure 2—2007 and 2010. Statistics are presented for 2007 rather than for 2005 because most MSAs had experienced rising employment and housing prices even before 2007.33 Table 2 shows that, between 2007 and 2010, housing prices fell on average by nearly 10 percent across the 353 MSAs. This average decline is slightly lower than that for the nation as a whole (see figure 1), because these estimates are unweighted and thereby reflect the fact that housing prices declined less in smaller MSAs.34

Table 2. Descriptive statistics for all 353 MSAs and by MSA type, 2007 and 2010
Variable20072010Percent change
(2007–2010)
MeanStandard deviationMinimumMedianMaximumMeanStandard deviationMinimumMedianMaximumMeanMedian
All 353 MSAs

Employment (thousands)

325.2573.629.1123.55,304.2312.5549.828.6116.75,153.7–4.07–5.52

House Price Index 
(2005 = 100)

108.28.090.6107.3137.398.416.941.5103.1148.5–9.97–3.92

Foreclosure rate (percent)

.80.50.11.673.072.512.23.491.9317.7168.11190.34

Homeownership rate (percent)

68.005.7340.9668.9282.6866.416.0639.5367.1581.46–2.40–2.58

Household size

3.50.232.633.484.323.56.213.063.554.401.541.98

Nonrelatives in family households (percent)

2.49.71.892.374.752.69.721.072.605.877.379.38

Unmarried households (percent)

7.751.872.167.8014.888.221.891.668.1713.885.724.79
Housing crisis first MSAs (55 MSAs)

Employment (thousands)

576.8850.729.1210.95,304.2559.2828.228.6201.65,153.7–3.15–4.42

House Price Index 
(2005 = 100)

103.48.090.7103.9137.378.017.741.581.5124.7–32.60–21.53

Foreclosure rate (percent)

.87.48.12.802.093.602.57.822.8013.2075.96251.48

Homeownership rate (percent)

65.288.0340.9665.8382.6863.207.7939.5363.4281.46–3.29–3.66

Household size

3.63.263.083.634.193.69.223.253.664.241.43.67

Nonrelatives in family households (percent)

2.82.661.802.804.483.25.781.793.105.8713.3111.05

Unmarried households (percent)

8.121.455.067.9811.088.491.595.258.4011.844.305.30
Labor market crisis first MSAs (67 MSAs)

Employment (thousands)

142.3247.939.174.51,968.8137.9248.837.371.11,987.9–3.19–4.57

House Price Index 
(2005 = 100)

107.93.8102.5107.2122.2107.85.593.8107.8123.3–.10.57

Foreclosure rate (percent)

.74.40.25.622.371.62.64.671.513.9654.29144.68

Homeownership rate (percent)

68.094.5754.5568.5175.2266.475.6348.2167.4176.74–2.43–1.61

Household size

3.44.193.083.434.033.51.193.063.524.202.162.69

Nonrelatives in family households (percent)

2.35.691.222.284.402.47.571.232.384.164.874.50

Unmarried households (percent)

6.972.062.456.9312.307.871.813.498.0312.4511.3715.89
Notes:

Note:  These data are unweighted. "Housing crisis first" MSAs are those where the housing market turned downward four or more quarters before the labor market did; "labor market crisis first" MSAs are those where the labor market turned downward four or more quarters before the housing market did. Omitted MSAs are those where the housing and labor market crises occurred "concurrently" (defined as less than four quarters difference in timing) and those where peaks could not be clearly identified.

Sources:  U.S. Bureau of Labor Statistics, Federal Housing Finance Agency, CoreLogic, American Community Survey, and authors' calculations.

Figure 2 and the top panel of table 2 also provide information on annual trends in homeownership rates, measures of doubling up, and the percentage of unmarried households for the 353 MSAs. Consistent with the research on national trends cited earlier, the figure and the table show that doubling up increased, whether measured by higher average household size or by nonrelatives as a percentage of all individuals living in family households. The fraction of unmarried households also rose over the period 2007–2010, in part because of the rising secular trend, but also because of weak economic conditions.35 Finally, table 2 shows that homeownership rates fell in conjunction with both the decline in housing prices and the rise in foreclosures. Although suggestive, these aggregate data mask substantial subnational variation in the severity, duration, and timing of the crises; these data also obscure the considerable heterogeneity in changes in doubling up and homeownership at the subnational level.

MSA-level analysis: timing of the housing and labor market crises. The analysis identifies the start of each crisis by examining quarterly data from 2005 to 2011. A crisis in the housing market is identified when housing prices, a measure of overall housing conditions, peak in a given quarter and then turn downward.36 Similarly, a crisis in the labor market is identified when employment peaks in a given quarter and then falls. In the case that either the housing price or the employment series has multiple peaks, the peak used is the one that precedes the longest downturn.37

Figure 3 provides a histogram of the relative timing of the crises for the 353 MSAs, where differences in timing are measured in number of quarters. A positive value indicates that the housing market crisis occurred first, whereas a negative value indicates that the labor market crisis occurred first. The distribution is slightly skewed to the left, indicating that in a slight majority of MSAs, it was the labor market, not the housing market, that turned downward first. This pattern is contrary to the one for the nation as a whole. (See figure 1.) The reason for the difference is that the national picture is driven by timing patterns in the largest MSAs.

Notes

33 As emphasized in this article, the timing of each crisis varied considerably by MSA, so the choice of 2007 is only for convenience of presentation.

34 Using weighted and unweighted MSAs, Sinai also points to a similar pattern of findings in his comparison of the severity of housing declines. (See Sinai, “House price moments in boom–bust cycles.”)

35 Kreider, “Increase in opposite-sex cohabiting couples.”

36 More precisely, a crisis is identified when the series growth rate from quarter to quarter changes sign, indicating a switch from growth to decline.

37 An alternative strategy is to identify the primary housing or labor market peak on the basis of the magnitude of the change in prices (or employment) rather than on the basis of duration. Most MSAs experienced multiple housing price peaks after the main downturn in their housing markets. However, because these housing price fluctuations were usually small in magnitude and duration, the first housing price downturn marked the peak of primary interest.

prev page5next page

View full article
About the Author

William H. Rogers
rogerswil@umsl.edu

William H. Rogers is associate professor of economics at the University of Missouri–St. Louis.

Anne E. Winkler
awinkler@umsl.edu

Anne E. Winkler is professor of economics and public policy administration at the University of Missouri–St. Louis and a research fellow at the Institute for the Study of Labor (IZA).