Residential land prices prior to development

Journal of Real Estate Research, The, 1997 by Karl L Guntermann

Abstract. This paper tests various hypotheses related to expectations and the value of undeveloped land. Evidence is found to support the hypothesis by Capozza and Helsley (1989) that the price of land in rapidly growing cities reflects a significant premium based upon expectations about future growth. There is also evidence that this premium varies from less than 40% of land value during down times to over 70% during boom times. Additional hypotheses tested related to development expectations for smaller geographic areas within the market. Land values reflect forecasts of employment up to five or six years into the future for nine square mile planning areas. The level of residential development activity from two to three miles around individual parcels is also capitalized into value. Much of the value of urban land may be explained by the growth rate of the metropolitan area and micro-geographic factors related to individual parcels.

Introduction

The price of land at the urban fringe has been the subject of considerable research over the last two decades. One line of research focused on explaining the timing of residential development and when it is profitable to convert rural land to an urban use. Important theoretical work by Markusen and Scheffman (1978) and Arnott and Lewis (1979) identified the expected appreciation rate in raw land relative to such factors as interest rates, development costs and agricultural use opportunity costs to explain when land is converted to an urban use.

It is well known that land prices can be extremely volatile, often rising and falling dramatically over very short periods of time.' It is clear that expectations about the future, as well as many other factors, contribute to changes in land prices. In early research, Hushak (1975) identified parcel size, accessibility, zoning, and property taxes as factors explaining variations in the price per acre of urban-rural fringe land surrounding Columbus, Ohio. A follow-up study by Hushak and Sadr (1979) expanded the model and tested it on other land markets with similar results. Chicoine (1981) confirmed the importance of parcel size, accessibility and zoning as well as neighboring land uses on the price of land south of Chicago, Illinois. He also found that prices per acre were lower for individual buyers and sellers than for other types of market participants (corporations, partnerships, etc.).

While previous research has identified many factors that explain variations in land prices, the process of land value determination prior to development is still not fully understood. The relationship between expectations and land values has been studied in some previous research but it has not always been the central focus of those papers. This paper extends previous research on urban land prices by testing various hypotheses related to expectations and land values, both market wide and for small geographic areas around individual parcels.

The Phoenix, Arizona metropolitan area provides an ideal laboratory for studying land prices prior to development because of its long history of high average growth rates. The Phoenix economy is also quite cyclical and several pronounced real estate cycles have occurred over the past several decades. The 1980s are a particularly interesting time to study land prices in Phoenix. A period of substantial growth and development was followed by several years of slower growth and eventually a prolonged period of stagnation. Real estate activity accelerated rapidly beginning in 1982 or 1983 and peaked in the 1985-86 period, resulting in severe overbuilding in every sector of the market. The availability of financing and the perception that population and employment growth would continue at very high rates by historic standards encouraged residential, commercial and industrial development to continue long after it became apparent from published data that population growth and the local economy were slowing down. This greatly aggravated the problem of excess supply and led to the perception by the late 1980s that land prices would be depressed from earlier peaks for many years. The depression in the residential real estate market ended by 1992.

Previous Research

The research presented in this paper builds directly upon the findings published in several previous articles. In an early important paper, Adams, Milgram, Green, and Mansfield (1968) developed models to explain residential, commercial and industrial land values (price per acre) as development occurred in a line extending outward in Northeast Philadelphia. They tested hypotheses relating to variables for accessibility, state of land (crude proxies for time to expected development), zoning, and plot or property characteristics (size, waterfront location, railroad siding, etc.). A time variable was included in the models to measure the appreciation in land values over the 1945-62 time period. They found evidence that specific variables in each category influenced price in separate models estimated using residential, commercial and industrial data. Several accessibility variables were statistically significant in the residential model, such as distance and travel time to the CBD (negative), distance from a major commercial boulevard (negative) as well as a premium for a location on the commercial boulevard. Distance from public transportation was negatively related to land value. Higher density residential zoning increased land prices relative to single-family zoning. The "state of land variables" tested in the various regressions were simple proxies for expected time to development. Among those, specific variables for "raw land" and "interior streets in" were significant but had negative signs, indicating that land without subdivision approval sold for a lower price.2