Examining Metropolitan House Price Dynamics:

Does Industrial Structure Affect the Speed of Adjustment?

Theresa DiVenti and Michael Hollar[*]

U.S. Department of Housing and Urban Development

Keywords: Housing Market Dynamics, Error Correction, Cointegration

The determinants of house prices include myriad supply and demand factors, of which income is arguably the strongest. Numerous studies have investigated this relationship, particularly with respect to house price bubbles (e.g. Case and Shiller 2003). Although significant attention has been devoted to determining whether local housing markets exhibit bubble tendencies (e.g. Bourassa, Hendershott and Murphy 2001), the process of returning to equilibrium has received little attention in the literature until recently.

Filling this void, a number of papers have begun to apply a cointegration/error-correction framework in order to shed light on the acceleration back to long-run trends. These studies generally focus on which factors allow some cities to revert back to equilibrium faster than others. The first set of studies, Abraham and Hendershot (1996), Malpezzi (1999) and Harter-Dreiman (2004), focus solely on the effect of supply constraints on the speed of adjustment back to equilibrium while Capozza et al. (2002) incorporates both supply and demand factors. The conclusions of this small but growing literature indicate why some cities may be more prone to house price bubbles than others based on traditional supply and demand factors. Together, they imply that existing supply constraints, which delay the increase in new housing in response to positive demand shocks, slow the return to equilibrium and increase the probability of a house price bubble.

Our paper adds to this literature by examining whether a city’s industrial structure affects its return to long-run equilibrium. Specifically, we measure the effect that a city’s industrial concentration[1] and type of primary export industry has on the equilibrium adjustment process. Controlling for supply constraints and city size, the difference in the speed of adjustment for diversified and concentrated cities depends on the relative strength of spillover effects between industries (urbanization-type agglomeration economies) in diversified cities. For example, if these effects are strong, then a positive shock to an export industry would not only increase employment and wages in the industry experiencing the shock, but also in the other export industries, thus further increasing demand and pushing the short-run deviation further from equilibrium. However, if agglomeration economies are relatively weak, then diversified cities will experience less of a departure from equilibrium and will therefore return quicker simply because the affected industry represents a smaller portion of the city’s total economy. In this case, for a given shock, a diversified city will not deviate as far from equilibrium as a concentrated city.

The impact of the type of primary export industry will be conducted for groups of cities with major industrial complexes, such as the automotive industry in Detroit, MI, Bloomington, IL and Flint, MI, and the computer industry in San Jose, CA, Rochester, MN and Fort Collins, CO. The objective of this part of the analysis is to determine whether some cities are prone to return to equilibrium faster due to the type of industries located there.

Econometrically, our analysis differs from previous studies in that we conduct the tests for each MSA individually in addition to using panel techniques. Employing a cointegration/error-correction framework allows us to model short-run house price dynamics and their adjustment to long-run equilibrium. In particular, we are interested in the error-correction term, which indicates how quickly house prices return to their long-run equilibrium following short-run deviations. Using traditional single cross-section time-series methods provides the ability to model a city’s housing market without imposing restrictions on coefficients, which accompany panel techniques. However, panel techniques contain the advantage of pooling data across MSAs, which is important when using short samples, and are thus more powerful tests. Examining the data using both methods allows us to focus on the group of cities for which a long-run relationship is found and further examine those for which a relationship is rejected. The use of traditional time-series methods is only possible using quarterly data, which provides more information and observations than the annual data used by previous studies.

Contrary to Gallin (2003), we find that a relationship between wages and house prices does exist for many MSAs[2]. The analysis first determines which MSAs exhibit a long-run relationship between house prices and income using quarterly data from 1988 to 2004, controlling for demand factors such as average wage and interest rates and supply factors such as employment density, which increases as the supply of land decreases[3]. For those cities that we find evidence of a cointegrating relationship, we estimate the impact of industrial concentration and type of primary export industry on the acceleration of house prices back to long-run equilibrium. The results will advance the understanding of house price dynamics and the existence of house price bubbles by not only controlling for the traditional supply and demand factors, but also by investigating a new dimension which may affect the speed of adjustment toward equilibrium: industrial structure.

References:

Abraham, Jesse and Patric Hendershott. 1996. “Bubbles in Metropolitan Housing Markets” Journal of Housing Research, 7 (2): 191-207.

Bourassa, Stephen, Patric Hendershott and James Murphy. 2001. “Further evidence on the Existence of Housing market Bubbles” Journal of Property Research, 18: 1-20.

Case, Karl and Robert Shiller. September 2003. “Is There a Bubble in the Housing Market? An Analysis” Prepared for the Brookings Panel on Economic Activity.

Capozza Dennis, Patric Hendershott, Charlotte Mack and Christopher Mayer. October 2002. “Determinants of Real Housre Price Dynamics” NBER Working Paper 9262.

Gallin, Joshua. April 2003 “The Long-Run Relationship between House Prices and Income: Evidence from Local Housing Markets” Federal Reserve Board Working Paper.

Harter-Dreiman, Michelle. 2004. “Drawing Inferences about Housing Supply Elasticity from House Price Responses to Income Shocks” Journal of Urban Economics, 55 (2): 316-337.

Malpezzi, Stephen. 1999. “A Simple Error Correction Model of House Prices” Journal of Housing Economics, 8: 27-62.

[*] Please direct correspondence to: Michael Hollar, U.S. Department of Housing and Urban Development, 451 7th Street, SW, Room 8208, Washington, DC 20410; e-mail: ; phone: 202-708-0421 x5878; fax: 202-708-1159. Theresa DiVenti, U.S. Department of Housing and Urban Development, 451 7th Street, SW, Room 8212, Washington, DC 20410; e-mail: ; phone: 202-708-1464 x5883; fax: 202-708-1159.

[1] Industrial concentration is measured by the sum of the squared shares of export employment.

[2] Statistical rejection of this relationship may occur for some cities simply because their housing market cycles exceed the time span of the available data as opposed a true absence of a long-run relationship.

[3] The house price and average wage data is from OFHEO and BLS, respectively. Interest rates are represented by the 10-year treasury rate as reported by the Federal Reserve. Employment density is constructed from Census land area estimates and BLS employment data.