|
Article Excerpt
Time series methods are applied to study monthly inventory and sales dynamics in the U.S. market for new single-family homes. The inventory measure used is the inventory of unsold, new single-family homes, regardless of the stage of construction. Stylized facts regarding inventory, sales, the inventory-sales ratio and the implied series of speculative housing starts are produced. Implications for structural models of housing supply and the relationship of inventory investment in this market to economy-wide inventory investment are considered. Finally, innovation accounting methods are applied to a structural VAR to study the responses of inventories and sales to permanent and transitory shocks.
**********
The short-run and long-run dynamics of the production and sales of new single-family homes have been actively studied by economists. Interest in short-run dynamics arises largely because of a desire to understand how residential construction activity fits into the business cycle, while interest in long-run dynamics arises from a desire to explain the economy's housing stock and its pricing.
However, these studies have largely ignored the dynamic behavior of the inventory of unsold new homes. These inventories are quite sizable relative to sales and exhibit cyclical fluctuations at business-cycle-like frequencies. We will present data that will indicate the average inventory-sales ratio in the market for built-for-sale single-family homes is approximately six. In comparison, Blanchard (1983) and others have found the inventory-sales ratio in the market for new automobiles produced in the United States to vary from around 1.5 to 3.2. The monthly inventory-sales ratio in the U.S. economy's durable goods sector is currently about 1.5. Therefore, the implicit assumption that the commercial homebuilder's production and sales decisions are equivalent is not even approximately correct. Producers in this market commit to the production of output, most of which will be sold at an uncertain future time and at an uncertain price. Ignoring the role of inventories of unsold new homes can provide an incomplete and misleading picture of the producers' decision-making problem and, consequently, its solution, that is, housing supply dynamics. Indeed, reduced-form regression evidence presented in papers by DiPasquale and Wheaton (1994), Topel and Rosen (1988) and, to some extent, Somerville (2001), suggests that the size of commercial homebuilders' inventories of unsold homes may have significant effects on production decisions.
The overall objective of this paper is to begin to fill this gap by introducing aspects of the time series behavior of the monthly inventory of unsold new single-family homes in the United States into the dialogue on housing market dynamics. (1) To accomplish this objective, we will use time series tools to produce a set of "stylized facts" that can help guide the future development of structural models of housing inventory investment.
Note that our focus on inventories is not meant to suggest that housing research ought to be directed at studying inventories instead of starts. Nor is it meant to suggest that it is more important to understand inventory dynamics than production dynamics. It is meant to suggest that micro-based research on starts that ignores the role of inventories on production decisions may be missing an important element of the production decision; thus, a better understanding of inventory dynamics may contribute to a better understanding of production dynamics.
In addition to the contributions we hope to make to the housing economics literature, we anticipate that our work will be a useful addition to the macroeconomics literature on inventory investment. Although aggregate inventory investment is a relatively small share of GDP, during recessions declining inventory investment makes up a relatively large share of the decrease in GDP. Thus, theoretical and empirical research on inventory investment has become an important component of modern business cycle research. (2) The development of micro-based models that can explain the "facts" of inventory dynamics is one of the key components of the macroeconomic inventory research program. Another key component is determining what the "facts" are. Fair (1989) and others have argued that facts based on physical units may look quite different from facts based on deflated dollar figures. However, as Blinder and Maccini (1991) point out, "data in physical units are available for precious few industries" (pp. 77-78). We have such data for built-for-sale single-family homes. Thus, one way in which our paper may be a useful addition to the macroeconomics literature is by providing additional facts regarding inventories based on data for physical units in a relatively large and cyclically sensitive industry. At the same time, we recognize that the relatively long production period for new homes makes this industry somewhat unusual. Whether this is simply a matter of degree or whether it reflects some deeper difference is unclear.
The remainder of the paper is organized as follows. The next section provides background detail and a literature review to establish how the paper fits into the housing and macroeconomics literatures. The following section develops a number of stylized facts about new home inventories and considers some of the implications of these facts for modeling the production and sale of single-family homes. In the fourth section, the dynamic responses of inventories and sales (and implicitly starts) to transitory and permanent structural shocks are developed using a structural VAR. The final section presents a summary of the main results and concluding comments.
Background and Literature Review
The Production and Sales Processes
To motivate our research and place it within the existing literature, it will be useful to begin with a brief description of the supply side of the market for new single-family homes. The U.S. Census Bureau's data on the construction of single-family homes distinguishes between "owner-initiated" and "built-for-sale" single-family housing starts. An owner-initiated housing start occurs when a household purchases a tract of land, then either builds the house itself or hires a general contractor to produce the home, with the household occupying the house upon its completion. A built-for-sale housing start occurs when a commercial homebuilder begins to build a home upon a tract of land it has purchased, with the house and land to be sold to a household, which will subsequently reside in the house. These commercial homebuilders are frequently referred to as speculative homebuilders. Over the period 1997-2000, for example, the average annual number of single-family housing starts was 1,235 million units a year. About 70% of these units were built-for-sale. (3)
Our interest is in the built-for-sale segment of the residential construction market. The supply process works essentially as follows. The producer, who owns the land, obtains a permit at time [t.sub.0] to build a house on that land. At time [t.sub.1], [t.sub.1] [greater than or equal to] [t.sub.0], the producer begins construction of the home, which is completed at time [t.sub.2], [t.sub.2] > [t.sub.1]. The producer sells the home (and land) at time [t.sub.3], which can be greater than, less than or equal to [t.sub.1] or [t.sub.2]. Thus, the home can be sold at any point before, during or after the construction process. In addition, the time between receipt of the permit and the start of construction, [t.sub.1] - [t.sub.0], and the length of the construction process, [t.sub.2] - [t.sub.1], are not fixed.
Literature Review
Most of the empirical literature on the supply of new single-family homes does not distinguish between housing starts and housing sales, essentially assuming that [t.sub.3] = [t.sub.1], so that there are no inventories of unsold or incomplete homes to be concerned with. This literature has focused on price and cost elasticities of housing starts as well as on the efficiency of the market for existing single-family homes, emphasizing the responsiveness of homebuilders to price signals. Examples of studies that emphasize the estimation of price and cost elasticities include Topel and Rosen (1988), DiPasquale and Wheaton (1994), Malpezzi and Maclennan (2001), Blackley (1999), Chan (1999), Fergus (1999), Somerville (1999) and Mayer and Somerville (2000). Examples emphasizing the efficiency of the market include Poterba (1984), Meese and Wallace (1994) and Rosenthal (1999).
However, within this literature, papers by Topel and Rosen (1988) and DiPasquale and Wheaton (1994) provide indirect evidence that single-family housing starts are sensitive to homebuilders' stocks of unsold inventory. More specifically, these authors fit regressions of single-family housing starts that include the median number of months since the start of construction for the stock of unsold new homes (Topel and Rosen 1988) or the average number of months on the market for newly sold new homes (DiPasquale and Wheaton 1994) among the explanatory variables. In both cases, the authors find that the "inventory" variable has a significant negative effect on housing starts.
There are papers, for example, Boorah (1979), Hendry (1986), Lee (1992), Coulson and Richard (1996) and Coulson (1999), which explicitly recognize that homes take...
|
