The distribution of urban land values: Evidence from market transactions

Land values are possibly the most fundamental price in urban economics, yet data on urban land values have remained rare and are typically collected for a single area at a time. This paper provides a detailed decription of land values using a novel data-set of land values taken directly from market transactions. It is the first of its kind to use market transactions to compare land values cross-sectionally across a wide range of U.S. cities. 

The analysis illustrates several striking features of land values. First, the statistical distribution of land values per acre appears to be roughly log-normal, both nationally and within metropolitan areas. Second, the standard deviation of log prices within metro areas is greater than one and roughly similar across different cities. Therefore, differences between cities in the log price of land are roughly constant across quantiles of the within-city price distribution.

Lot size is the strongest predictor of price per acre at the parcel level, explaining over 40 percent of the variation in a simple regression. Controlling for metropolitan area increases the explanatory power of the regression to 57 percent. Additional controls such as quarter of sale, travel time from city center, and proposed use give modest improvements in explanatory power. Even over the boom and bust period of our sample, space is a much more important determinant of land values than time. When comparing various measures of distance from the city center, distance measured in driving time appears to predominate versus straight-line distance and driving distance.

At the county level, we find that land values for urban uses at the 10th percentile are strongly and positively related to values for agricultural uses. Non-linearity in the relationship is likely to reflect costs of converting agricultural land to urban use. Estimates using distance from center within metros are consistent with estimates using population, possible arc of development, and agricultural values across metros. Our estimates provide novel support for the standard monocentric city model of urban economics. They suggest land receives 7 percent of income and the cost elasticity of urban population is 4 percent of income.