Abstract
This study investigates the willingness of homebuyers to pay for co-location with iconic architecture. Oak Park, Illinois, was chosen as the study area given its unique claim of having 24 residential structures designed by world-famous American architect Frank Lloyd Wright, in addition to dozens of other designated landmarks and three preservation districts. This study adds to the limited body of existing literature on the external price effects of architectural design and is unique in its focus on residential architecture. We find a premium of about 8.5 per cent within 50–100 m of the nearest Wright building and about 5 per cent within 0–50 m. These results indicate that an external premium to iconic architecture does exist, although it may partially be attributable to the prominence of the architect.
Notes
1 With
2 The property transactions considered spread over 403 block groups within the relatively small study area.
3 Road distances are calculated using MS MapPoint.
4 A regression of the floor space index (ratio of floor space over lot area) on the same explanatory variables as in model (4) indicates a significant differential within the 50–100 m area (about 5 per cent), but in none of the other distance rings.
5 Another explanation could be that the results are particularly sensitive to altering model specification because of too few observations in the distance band. However, with 84 and (2.5 per cent of the all) observations in the 50–100 m ring alone, the area seems reasonably populated.
6 Data from ‘Olcott's Land Values Blue Book of Chicago’ have also been used by Bednarz (Citation1975), Berry (Citation1976), McDonald & Bowman (Citation1979), McDonald (Citation1981), McMillen (Citation1979), McDonald & McMillen (Citation1990), McMillen & McDonald (Citation1991), Mills (Citation1969) and Yeates (Citation1965).
7 Our tests are based on an extended Table , column (1) specification introducing an interaction term of distance to the nearest Wright building and a yearly trend variable. We thank an anonymous referee for this suggestion.
8 Using a row standardized inverse exponential weights matrix, standard spatial LM test scores do not point to the presence of such problems (see Table notes). Spatial statistics (p-values) from Table , model (4) are as follows: LM error, 0.107; Robust LM error, 0.225; LM lag, 0.254; Robust LM lag, 0.673. +/*/* * denote statistical significance at the 10/5/1 per cent level, respectively. The SEM model we estimated can be written as follows: y = Xβ+μ; μ = λWε, where y is the dependent variable, X is a vector of independent variables, W is a weight matrix and ε is a random error term satisfying the usual conditions.
9 In the model world, this scenario would correspond to an increase in R.