|
Article Excerpt
This article establishes a linkage between decadal changes in suburban population and the supply of suburban dwelling units. It then estimates an econometric supply-and-demand model for 317 U.S. suburban areas for the 1970s, 1980s and 1990s using the State of the Cities database. Suburban supply is more elastic than central city supply, with suburban estimates between +1.26 and +1.42. However, separate estimates by geographic region lead to supply elasticities of +0.89 for the northeastern quadrant of the United States and +1.86 for the remainder of the United States.
**********
This article addresses issues of population change and housing supply in U.S. suburbs. Central cities often have only limited opportunities for new construction, while surrounding suburbs "beyond Eight Mile Road" may have considerable vacant land to accommodate new employers and new residents. (1) This generalization, of course, oversimplifies. New Rochelle, NY; Evanston, IL; Brookline, MA; Royal Oak, MI; and Lakewood, OH, for example, were developed 100 or more years go. Many suburbs (Puentes and Orfield 2002) are fully built up, many have stopped growing or have experienced population losses, and some have problems of blight or poverty similar to central cities.
This article establishes a linkage between decadal changes in suburban population and housing supply, differentiating between central cities and inner and outer suburban rings. It then estimates an econometric supply-and-demand model for 317 U.S. suburban areas for the 1970s, 1980s and 1990s using the State of the Cities database.
With almost all suburban areas characterized by increasing housing stock and in general more buildable land than the central cities, one would expect suburban supply price elasticities to exceed those of central cities. Using a similar model. Goodman (2004) estimated dwelling unit price elasticities between +0.03 and +0.13 for central cities with declining housing stocks, and between +1.05 and +1.08 for central cities with increasing housing stocks. The expectation of more elastic suburban supply is borne out, with estimates between +1.26 and +1.42. However, separate estimates by geographic region yield a supply elasticity of about +0.89 for the northeastern quadrant of the United States and +1.86 for the remainder of the United States, with a weighted mean of +1.42.
Metropolitan Structure and Housing Supply
Over the past 30 years, U.S. metropolitan population growth has occurred largely outside the central cities. (2) Most models of urban structure, density and growth refer to land use and land rents, with capital stocks adjusting to the differing rents. The suburbs are distinguished only by greater distance from the central place and lower densities. Housing capital stock has only a passive role in such models, with the results differing little from models that examine only land and land rents.
One could argue that housing stock, and particularly new construction, assumes a critical role in characterizing suburban development. Metropolitan population expansions most often occur in suburban areas where empty lots are developed and previously undeveloped tracts are converted into housing developments. Although such development could occur in central cities as well, the costs of "teardowns and rebuilds" often make it less desirable than building on previously undeveloped land. (3)
Housing stock adjustments generally depend on the flow of new stock. Green and Malpezzi (2003, p. 6) describe a U.S. construction industry with a large number of very small producers, implying close to constant returns to scale for new units and close to an infinitely elastic supply of new units. Using such a theoretical framework, Muth (1968) estimates one of the earliest supply elasticities at approximately +14. DiPasquale (1999) surveys the literature and concludes that (i) new supply appears to be price elastic, with estimates between +3.0 and positive infinity: (ii) higher income households appear more likely to improve their homes than to do nothing, but they are more likely to move than to improve their current units; and (iii) repair and renovation expenditures are inelastic with respect to income and price.
Comparative work between the United States and the United Kingdom shows the United Kingdom to have less elastic supply. For the prewar United States, Malpezzi and Maclennan (2001) calculate implied price elasticity of supply from flow models as between +4 and +10, and for postwar, it is between +6 and +13. In contrast, for the prewar United Kingdom, the implied price elasticity from flow models is between +1 and +4, and for postwar, it is between and less than +1.
Bramley (1993a, 1993b) estimates a U.K. price elasticity of supply of about +0.31. Pryce (1999) uses the data provided by Bramley and finds a back ward-bending supply curve in the 1988 boom period but not in the slump conditions of 1992. He estimates the price elasticity of supply to be 0.58 in 1988 and 1.03 in 1992. (4)
Mayer and Somerville (2000a, 2000b) examine new construction price elasticities. Mayer and Somerville (2000b), for example, characterize housing supply elasticity in terms of the housing stock (rather than new construction), in an empirical model derived from urban growth theory. They describe new housing construction as a function of changes in house prices and costs rather than as a function of the levels of those variables, used in the previous studies. Their estimates using quarterly panel data (Mayer and Somerville 2000b) relate a 10% rise in real prices to a 0.8% increase in the housing stock, which is accomplished by a temporary 60% increase in the annual number of starts, spread over four quarters. With local area supply functions (Mayer and Somerville 2000a), they find that the aggregated national data may slightly overestimate price elasticity of new construction and underestimate the time required to respond to price shocks.
Glaeser and Gyourko (2005) argue that an urban area's housing supply is kinked--highly elastic with respect to positive shocks because additional units can be built if desired, but inelastic when shocks are negative because existing homes are quite durable. A positive demand shock is expected to generate more units and people, but only a moderate increase in housing price. In contrast, a negative demand shock is expected to cause housing prices to fall, but induce little change in the housing stock or population. Glaeser and Gyourko (2005) do not estimate supply elasticities, but their model suggests asymmetric elasticities close to in the negative direction but larger in the positive direction. Goodman (2004) provides separate analyses for contracting and expanding central cities, and validates the hypothesis that supply is much less elastic in the negative than in the positive direction. As suburbs are generally expanding, suburban price elasticities would presumably exceed central city elasticities, as the suburbs have access to large tracts of previously undeveloped land that are most often unavailable in central cities.
Suburban Populations and Dwelling Units
This section introduces descriptive analyses using the State of the Cities data system (SOCDS), which provides census data for metropolitan statistical areas (MSAs), metropolitan cities and suburbs for the years 1970, 1980, 1990 and 2000. (5) The version of the SOCDS here uses the 1990 standard for MSAs and primary metropolitan statistical areas (PMSAs) as established in June 30, 1999. Suburban data comprise the data for the metropolitan area less the sum of the data for all central/principal cities (if any) in the metropolitan area. For New England states, the analysis uses metropolitan areas as defined by the standard MSA/PMSA definition, rather than the New England county metropolitan area definition. This study analyzes 317 suburban areas that provide data for all the four years.
Table 1 describes 1970-2000 growth patterns for the populations and numbers of dwelling units in the 50 largest suburban areas as ranked by 1970 suburban population. Only Pittsburgh's suburban population fell by 6.7%, yet all areas had at least double-digit percentage increase in numbers of dwelling units, with the Atlanta, Dallas, Houston, Denver, Tampa and Fort Lauderdale suburbs showing triple-digit increases. Dwelling unit percentage increase generally exceeded population percentage increase, often by 20 or more points, and only Los Angeles and Riverside, California, saw higher unit increase than population increase.
Demographic Changes and Suburban Housing Supplies
This section links patterns of household formation and household size to the numbers of occupied dwelling units. These patterns changed substantially in the last third of the 20th century, particularly in the 1970s. From 1970 to 1980, the average number of persons per household in the United States fell from 3.14 to 2.75, a decrease of 12.1%. (6) Sweet (1984) lists six reasons for this: (i) young people increasingly delayed marriage; (ii) rates of separation and divorce increased; (iii) remarriage rates began to stabilize and decline, after a period of increase; (iv) mortality of the elderly declined; (v) persons of all ages and marital statuses continued their increased propensities to form their own households rather than to share the...
|
