The many dimensions of historic preservation value: national and local designation, internal and external policy effects

Abstract

This analysis examines the internal and external policy effects of national and local register programmes for historic preservation. Robust hedonic pricing models are crucial to informing policy proposals and understanding how property markets relate to urban heritage. Estimating a repeat-sales hedonic model with neighbourhood trends and spatial mixed models, novel to this literature, offers a marked improvement in terms of jointly identifying internal and external policy effects, comparing national and local designations, separating policy from heritage effects and estimating models robust to spatial dependence and trends in hedonic prices. Historic designation variables, while often individually insignificant in the model, are always jointly significant in explaining varying appreciation rates. Local districts exhibit no consistent price impacts across the models. Being located inside a national district confers a price premium that increases over time in the preferred model specification, while prices fall in national districts’ buffers after designation. The sensitivity of results to model specification raises questions about alternative approaches to spatial dependence in the data in the urban historic preservation context. Evidence of the influence of historic district designation on property turnover and renovation investments is also examined.

Keywords:

• Historic preservation
• national and local register programmes
• external policy effects
• repeat-sales hedonics
• spatial econometrics

Notes

1. DNR also manages a state-level Georgia Register of Historic Places, which uses the same criteria as the NRHP and automatically lists properties listed on the NRHP. Neither the state nor national register restricts rights of private property owners. Owners of properties on the state register may be eligible for some subsidies from the state.

2. This was the longest period of time for which they would provide data. Spanning a decade has advantages in containing many instances of the same property selling more than once, variations in macroeconomic conditions and numerous changes to the inventory of historic landmarks and districts.

3. Recent studies employing a similar buffer approach to measuring external effects include Noonan and Krupka (2011), Lazrak et al. (2014), Ahlfeldt and Mastro (2012), and Zahirovic-Herbert and Chatterjee (2012). Distance to nearest landmark building also compares conveniently to Moro et al. (2013), Ahlfeldt and Mastro (2012), and Zahirovic-Herbert and Chatterjee (2012). Buffers at greater distances than 100-m did not add explanatory power in our models, and smaller buffers risked too few observations.

4. Following Bailey, Muth, and Nourse (1963), the set of time controls ΔY_it takes values of –1 for the period of initial sale, +1 for the period of the final sale and 0 otherwise. The data include 11 years and 4 quarters for time and seasonal controls. The coefficients for the year and seasonal controls reflect price changes relative to the omitted times (2000 and winter, respectively). We keep a constant term in the model to allow for a nonzero intercept, essentially capturing the effect of the final sale.

5. When these (ΔRooms, ΔBedrooms, ΔFamRooms, ΔBaths, ΔHalfBaths) variables have missing values imputed using other independent variables from the model, comparable hedonic models can be estimated. Results for the coefficients of interest (δ) are essentially unchanged.