An Empirical Evaluation of Alternative Land Valuation Models

Because of the variety of market participants, differing motivations and financing methods, and relatively thin markets in built-up areas, vacant land valuation has always been one of the more difficult aspects of property valuation. On the other hand, improved residential property sales tend to be sufficiently plentiful and comparatively easy to analyze. This paper evaluates whether modern computer-assisted valuation methods can combine vacant and improved residential sales to develop more stable and reliable vacant land values. An affirmation answer to this question would enhance property assessment in general. It would also facilitate implementation of a site value or split rate tax, in which values were separated into land and building components with the land component taxed at a higher rate than buildings.

Research on the use of modern mass appraisal methods to determine land values began in 2000 under a David C. Lincoln Fellowship in Land Value Taxation. Initial work focused on the use of models that combined vacant and improved sales to decompose total property values between land and buildings and analyzed winners and losers under a shift to a site value tax. The paper analyzed data from three urban areas (Boise, Edmonton, and suburban Denver) and concluded that properly specified and calibrated models could be decomposed into land and building components. It also concluded that taxes on vacant land and land-intensive property in general could be expected to rise considerably, while taxes on building-intensive properties could be expected to fall.

A continuation of the research grant for 2001 permitted the investigation of the following two issues: (1) to what extent does location affect land versus building values, and (2) can vacant residential land be modeled with improved residences, avoiding the need to determine separate valuation models? Analyses and results of the first issue are reported separately. Analyses and results for the second issue are reported here. The research concludes that vacant land can be modeled with improved property, provided that vacant land is similarly serviced and there are sufficient vacant land sales to provide benchmarks to ensure a reliable decomposition of values.