Academic journal article Journal of Real Estate Literature

A Review of Recent Empirical Studies on Property Price Gradients

Academic journal article Journal of Real Estate Literature

A Review of Recent Empirical Studies on Property Price Gradients

Article excerpt


This paper reviews the empirical studies that have been conducted on property price gradients. There are basically two different methodologies of estimation and three different assumptions on the spatial structure of an urban area. The two methods are: (1) the hedonic pricing model and (2) the repeat-sales model; and the three assumptions are: (1) the monocentric assumption; (2) the non-monocentric/polycentric assumption; and (3) no a priori assumption about the urban spatial structure. The methodologies and the findings of these previous studies are compared and analyzed to clarify the path for further studies.


The value of residential housing has long been shown to be dependent on location. The importance of location on the value of housing was first explained by von Thünen (1826), who stated that if there is a central marketplace in a farming area, where transactions of agricultural products are concentrated, then it saves commuting costs for the farmers to travel between their farmland and the central marketplace. This early spatial theory of agriculture was then modernized by Alonso (1964), who proposed the bid-rent curve. Since then, a large literature, including a vast number of empirical studies, has emerged to test the predictability of urban spatial theory by applying hedonic regressions and repeat sales models.

The classical model, put forward by von Thünen, can be found in many textbooks on urban economics (e.g., McCann, 2001). In von Thünen's model, there is a specific point in the market at which all transactions in agricultural products take place. Farmers need to bring their yields to this marketplace to conduct transactions for profit; and the cost of transportation increases with the distance to the marketplace. As such, von Thünen's model predicts a negative land-rent gradient, by which the rent for farmland will decrease as the distance to the marketplace increases, to compensate for higher transportation costs.

Instead of examining the market for farmland, Alonso's bid-rent model discusses the housing market in terms of a monocentric assumption. A central business district (CBD) is the marketplace at which jobs and other economic activities are concentrated; and each of the households should have at least one member who needs to travel frequently between the residential location and the CBD. The level of utility of each household was expressed as a function of the amount of land (housing) the household occupies, the amount of all other goods they own and the distance from which they must travel to reach the CBD. Alonso denned the bid-rent function as the set of prices (the bids) the household would offer for land at different distances to the CBD, holding the utility at a constant level. As such, this model predicts that the bid-rent curve should have a negative slope.

The analysis of the effect of location on the value of housing property was then largely dominated by the bid-rent function. It is common to include the distance to the CBD as an independent variable in empirical studies of property prices. The inclusion is based on the assumption that there is a CBD in the area concerned (i.e., the monocentric assumption). Many studies, both theoretical and empirical, such as those by Muth (1969), Mills (1972), Burrell (1985), and Atack and Margo (1998), have been carried out on the basis of this assumption. However, some empirical studies have failed to show a significant negative coefficient for the distance-to-CBD variable (e.g., Heikkila, et al., 1989). This raises concerns about the validity of this assumption, as well as about using distance-to-CBD as the proxy. There have been numerous attempts to deal with this concern, as listed in Exhibit 1.

One of the explanations for why a non-negative property price gradient is obtained is that the monocentric hypothesis has been wrongly applied in estimating polycentric gradients. For example, Dubin and Sung (1987) showed that if subcenters do exist, then each of them would exert a certain degree of influence on housing values; and that the distancc-to-CBD exerts a similar influence on housing values as any one of the subcenters. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.