«DiscoVeriNg HomelessNess Volume 13, Number 1 • 2011 U.S. Department of Housing and Urban Development | Office of Policy Development and Research ...»
Life-Cycle Pattern Davidoff (2006) found that maintenance and additions spending peaks around age 40 to 45.
Exhibit 7 suggests that the life-cycle profiles of maintenance and additions expenditures are not identical. Our definition of additions expenditures follows a similar pattern as Davidoff’s definition of maintenance and additions expenditures, but our measure of maintenance spending is much flatter after age 50. Maintenance spending appears to be constant or slightly declines after age 50, but additions expenditures fall dramatically after age 45. Exhibit 7 reinforces the necessity to run separate models for maintenance and additions expenditures.
The question remains as to whether this shape will be evident in a multivariate regression analysis when controlling for other variables, such as income, that might also have this hump shape. Exhibit 5 suggests that the probability of maintenance spending increases with the age of the homeowner after age 60, and a similar pattern is seen in the level of maintenance expenditures. The likelihood of additions spending exhibits the more familiar life-cycle pattern, with a decrease seen after age 40.
Exhibit 7 Mean Maintenance and Additions by Age of Household Head
Model Fit The last common thread in the literature is that the explanatory power of the maintenance and additions models is low, and our household-level results are no exception; our R2 values are.06 and.02 (exhibit 6). These results should not come as a surprise; the idiosyncratic factors that drive a particular household to spend on their home are very likely to dominate the effects of any of the variables we observed. The data set does not include, for instance, a variable for leaky roofs.
Given that the cross-sectional models have low explanatory power, the question becomes—Do we care about explaining this cross-sectional variation? In the aggregate, home maintenance and additions expenditures are important to the economy. Thus, the dynamics that explain aggregate housing maintenance and investment spending decisions are important to the capital stock. Therefore, it might be more useful to determine whether our household-level cross-section models explain the variation in aggregate-level home maintenance and investment spending. This approach avoids attempting to answer the impossible question of whether the Smiths decided to buy a new water heater, but instead addresses the question of why the likelihood of additions expenditures and the level of additions expenditures have increased dramatically since 2000.
To determine the quality of fit at the aggregate level, we found the cross-sectional fitted values for each household and calculated the quarterly mean to match the time-series patterns seen in the earlier exhibits. The goodness-of-fit statistic then is equivalent to an R2 value, which is defined as 1 minus the ratio of the residual (time-series) variance to the data’s (time-series) variance. In contrast to the cross-sectional results, we found that our household-level model has fairly strong explanatory power for the aggregate series of data. Exhibits 8, 9, 10, and 11 present the quarterly average of our models’ fitted values along with the data they model.
The model in exhibit 8 provides information that can be used to explain the long-run downward trend in the percentage of households with nonzero maintenance expenditures. Overall, our model explains 82.3 percent of the variance in the probability of a nonzero maintenance expenditures series.9 For the probability of nonzero additions expenditures, shown in exhibit 11, the model fares well in the long run, but it misses some of the shorter run dynamics, most notably the apparent turning point in additions in 2001. The R2 for additions expenditures is 66.4 percent.
Exhibits 10 and 11 display the fitted values for the seemingly unrelated regressions for maintenance and additions expenditures, respectively. The maintenance expenditure series is less challenging, and the model captures the slow downward trend well, with an R2 of 84.1 percent. As was the case with the probit models, the reduced form demand model for additions expenditures does relatively well in the long run and relatively poorly in the short run, missing the 1991-to-1992 and 2001 drops. The R2 for additions expenditures is 72.6 percent.
Both series exhibit strong downward trends, which may explain the high explanatory power of the model. When differencing the data, the explanatory power does not diminish.
Additional Household-Level Results Using a data set new to the literature, our results reinforce many of the basic facts from the existing literature that uses the AHS. In addition to the results already described, we also included other variables in the model. For household characteristics, we found that larger families are less likely to spend on maintenance but more likely to spend on additions, which suggests that larger families focus more on increasing the stock of housing and foregoing some upkeep. The same is true for the level of spending.
Regarding home characteristics, the CE Survey asks for the year the home was built, and we converted that date to a series of dummy variables representing the home’s age. Our results indicate that owners in new homes are less likely to spend on maintenance than those that have owned the home for 2 to 40 years, and the same pattern is seen for the level of maintenance expenditures. For additions expenditures, owners of new homes are the most likely of all the respondents to spend on additions. Families that live in larger homes spend more on maintenance and additions, as do those that live in an MSA. Those in urban areas spend more on maintenance but less on additions.
The last variable of interest is the OFHEO state home price index. The results shown in exhibit 5 suggest that home price appreciation affects the likelihood of maintenance expenditures but not the likelihood of additions expenditures, indicating that appreciation leads to households being less likely to spend on maintenance. Homeowners may believe that if property values are rising that they do not need to do as much maintenance.
Conclusion The study described in this article has documented some of the stylized facts about home maintenance and investment decisions using a new data set to this literature, the Consumer Expenditure Survey. A replication study on this topic is important because of the importance of the home in the household’s financial portfolio and because of the importance of residential investment in the macroeconomy.
Further, this study has documented new findings, in part, because the CE Survey data has different strengths than the AHS. The quarterly CE Survey data can be better used to highlight the timeseries patterns in maintenance and additions expenditures. For example, the data show a constant decline in the percentage of households spending per quarter between 1984 and 2000. Since 2000, there has been an increase in the percentage of households spending for additions. These time series patterns have not been previously documented, in part, because the AHS is not designed to capture these trends.
Acknowledgments The authors thank Tom Davidoff and seminar participants at the Census Bureau for helpful comments and suggestions.
Authors Jonathan D. Fisher is a research data center administrator and economist in the New York Census Research Data Center at Baruch College. (The research in this article was undertaken while Mr. Fisher was at the Bureau of Labor Statistics.) Elliot D. Williams is a research economist at the Bureau of Labor Statistics.
References Baker, Kermit, and Bulbul Kaul. 2002. “Using Multiperiod Variables in the Analysis of Home Improvement Decisions by Homeowners,” Real Estate Economics 30: 551–566.
Davidoff, Thomas. 2006. Maintenance and the Home Equity of the Elderly. Working paper.
Vancouver, British Columbia: University of British Columbia.
Gyourko, Joseph, and Joseph Tracy. 2006. “Using Home Maintenance and Repairs To Smooth Variable Earnings,” Review of Economics and Statistics 88 (4): 736–747.
Montgomery, Claire. 1992. “Explaining Home Improvement in the Context of Household Investment in Residential Housing,” Journal of Urban Economics 32: 326–350.
Potepan, Michael J. 1989. “Interest Rate, Income, and Home Improvement Decisions,” Journal of Urban Economics 25: 282–294.
Rappaport, Barry A., and Tamara A. Cole. 2003. Research Into the Differences in Home Remodeling Data: American Housing Survey and Consumer Expenditure Survey/C50 Report.
Working paper. Suitland, MD: Census Bureau.
Reschovsky, James. 1992. “An Empirical Investigation Into Homeowner Demand for Home Upkeep and Improvement,” Journal of Real Estate Finance and Economics 5: 55–71.
164 Data Shop Data Shop Data Shop, a department of Cityscape, presents short articles or notes on the uses of data in housing and urban research. Through this department, PD&R introduces readers to new and overlooked data sources and to improved techniques in using well-known data. The emphasis is on sources and methods that analysts can use in their own work. Researchers often run into knotty data problems involving data interpretation or manipulation that must be solved before a project can proceed, but they seldom get to focus in detail on the solutions to such problems. If you have an idea for an applied, data-centric note of no more than 3,000 words, please send a one-paragraph abstract to firstname.lastname@example.org for consideration.
Tracking the Housing Bubble Across Metropolitan Areas–– A Spatio-Temporal Comparison of House Price Indices Laurie Schintler Emilia Istrate George Mason University Abstract This article presents an analysis of five available house price indices that are used to track house prices at the metropolitan area level. These five indices are (1) the Federal Housing Finance Agency (FHFA) House Price Index (HPI), (2) the Standard & Poor’s/ Case-Shiller® Home Price Indices, (3) an adjusted version of the FHFA House Price Index, (4) the Zillow Home Value Index, and (5) the NATIONAL ASSOCIATION OF REALTORS® Median Home Price. This study first discusses the strengths and weaknesses of these indices for use in a spatio-temporal analysis. Then, it provides a comparative analysis of their change rate for 10 metropolitan statistical areas (MSAs) for two time periods: the third quarter of 2006 through the third quarter of 2007 and the first quarter of 2007 through the first quarter of 2008. In addition, this research constructs a series of spatio-temporal indicators based on time and spatial lags of the HPI for 302 MSAs for the 2000-to-2007 period. The results of this data brief could help researchers interested in spatio-temporal analyses of the latest housing bubble and of house price indices at large.
Introduction Most of the United States has witnessed sharp changes in house prices during the past 5 years. Along with other factors, the contracting housing market has slowed the growth rate of the U.S. economy since 2007 (Federal Reserve, 2008; Joint Center for Housing Studies, 2008). Acknowledging the influence of the housing market on the economy, the federal government took several measures during the past 2 years, such as the Housing and Economic Recovery Act of 2008, the conservatorship of Fannie Mae and Freddie Mac, and additional funding for housing in the Recovery Act of 2009.1 Before the major tightening of the housing market in 2008, experts had no agreement on the existence of a “housing bubble” in the United States. Stiglitz (1990: 13) defined an asset bubble the following way: “if the reason that the price is high today is only because investors believe that the selling price will be high tomorrow—when ‘fundamental’ factors do not seem to justify such a high price—then a bubble exists.” Some researchers concluded that the U.S. house price changes in the first half of the 2000s fit the trend (Himmelberg, Mayer, and Sinai, 2005; Smith and Smith, 2006), with no bubble and especially no burst in sight. Others (Case and Shiller, 2003; Krugman, 2005) pointed to irrational overpricing and speculative investments (Shiller, 2008).
During this debate about the housing bubble, it was unclear which house price index was the most appropriate for tracking the price changes across time and space. This data brief presents an analysis of five available house price indices that are used to track single-family house prices at the metropolitan area level. This study first discusses their strengths and weaknesses for use in a spatio-temporal analysis. Then, it provides a comparative analysis of home price change rates for 10 metropolitan statistical areas (MSAs) during two separate time periods: the third quarter of 2006 through the third quarter of 2007 and the first quarter of 2007 through the first quarter of 2008. In addition, this research constructs a series of spatio-temporal indicators based on time and spatial lags of the HPI for 302 MSAs during the 2000-to-2007 period.
House Price Indices
Five major house price indices are available at the metropolitan area level in the United States:2
(1) the Federal Housing Finance Agency (FHFA) House Price Index (HPI), (2) the Standard & Poor’s/Case-Shiller® Home Price Indices (S&P/Case-Shiller®), (3) an adjusted version of the FHFA House Price Index (adjusted HPI), (4) the Zillow Home Value Index, and (5) the NATIONAL ASSOCIATION OF REALTORS® (NAR) Median Sales Price of existing homes.3 For more information about the conservatorship of Fannie Mae and Freddie Mac, see Goldfarb, Cho, and Appelbaum (2008). The American Recovery and Reinvestment Act of 2009 provided an additional $2 billion for the Neighborhood Stabilization Program (NSP) on top of the $4 billion from the Housing and Economic Recovery Act (HUD, 2010).
This data brief uses the Office of Management and Budget (OMB) definition of a metropolitan statistical area, with “at least one urbanized area of 50,000 or more population, plus adjacent territory that has a high degree of social and economic integration with the core as measured by commuting ties” (OMB, 2009: 8). This study focuses on the metropolitan areas because they concentrate most of the U.S. population. For more information about the demographics and role of the metropolitan areas in the United States, see Berube et al (2010).