Hoen's Critique of the Renewable Energy Policy Project (REPP) Study
Sterzinger et al., (2003) analyses roughly 24,000 transactions near 11 windfarms in the U.S., and compared average transaction values for houses in a control area outside the viewshed of the windfarm with transactions occurring within the viewshed (a 5-mile radius). The study comes to the conclusion that, "There is no support for the claim that wind development will harm property values.” (p. 9), and even declares, “For the great majority of projects [windfarms] the property values rose more quickly in the viewshed than they did in the comparable community.” (p. 2). Although this study is often quoted,28 its methods have been criticized (e.g. ECW, 2004) for four reasons. First, the authors attempt to calculate a value for the variable “view of windmills,” without properly controlling for it. There is no attempt to discern which properties within the ten different 5-mile viewsheds can see the windfarm or not. In effect, the study makes the erroneous assumption that all properties in the 5-mile radii can see the windfarm, when many houses’ views in fact are obstructed by geological features, trees, and other houses (RBA, 1998a; Poletti, 2005).29
Secondly, the analysis does not control for distance to the turbines, thereby making the assumption that the “viewshed” effect is the same, on average, for homes five miles from the windfarm and those in immediate proximity to the turbines. Third, there are problems with how the study validates its results. The report provides readers with only R2 (or goodness-of-fit) values for its outcomes, and this is problematic, since, by itself, the R2 statistic is a poor indicator of explanatory power (Halcoussis, 2005). Compounding this problem, the report gives R2 values which are very low, for instance 0.02 for some odels, which is saying in essence the model describes only 2% of the actual movement of property values. Despite this somewhat flagrant disregard for rigor it treats these models as it does models where the statistic is high, for example 0.85. This inconsistency is not addressed by the report. The last reason this research is often criticized is that no attempt is made to sort out inappropriate transactions. Sales that are not arms-length (divorce, sales between family members, estate sales etc.) are included. By doing so the report includes transactions that do not represent the agreement between a willing buyer and a willing seller, a requirement for accurate analysis. Combined, these four omissions in rigor render the results of the report extremely weak, if not entirely misleading.
28 A “Google” internet search using all of the following words, “REPP”, “wind” and “property” generates 18,600 results. [tested 2-20-06]
29 Sterzinger et al analyze the community surrounding the Madison County windfarm, which is the subject of this report. We found 66% of the homes sampled in the 5 mile radius could not see the windfarm at all.