Modellering van waardedaling van woningen als gevolg van risico door aardebevingen rond het Groningenveld

George de Kam, Eric Hol

OnderzoeksoutputAcademicpeer review


More than 50 years of extracting natural gas from the “Slochteren field” in Groningen in the northern part of the Netherlands has resulted in a series of induced earthquakes of over time increasing strength. The earthquake with thus far the highest magnitude of 3,6 on the Richter scale occurred in 2012 near Huizinge, causing serious structural damage to buildings. Some 200.000 inhabitants of the afflicted area not only worry about the repair of the damage caused, but are since Huizinge also seriously concerned about earthquake induced depreciation of the value of their properties. According to the Dutch Mining act, damages caused by mining activity should be reimbursed by the party responsible for the exploitation. In order to prepare a liability claim against the exploiter (the NAM), one should have a reliable method to assess the damage suffered.
Whereas the cost of repair of physical damage to houses is relatively easy to estimate, determin- ing the depreciation of the value of houses is far more complicated. Several statistical models to estimate the depreciation of houses have been proposed over the years, but so far none of these models has proven itself to provide an adequate solution for this problem. In this paper

we explore the technical impediments that complicate these efforts. We furthermore propose a statistical model that incorporates practical solutions to manage the three major impediments identified in this paper.
The most important impediment to develop an analytical model is to find a suitable indicator that emulates the consumer’s behavior toward buying and selling property. This indicator is a human-behavior driven variable that should account for lower pricing as a result of the risk of damage due the earthquakes. Latent variables like this can only be measured indirectly by one or more proxies. Physical indicators representing the sum of the ground velocity caused by earthquakes have been used for this proxy, as well as the percentages of houses with damage due to earthquakes. The former is an easy to calculate variable but most likely not recognized by the consumers and its relation to buying behavior is not clear, the latter is more visible in the field but inherently unreliable since the extent of damage to houses and the willingness to report this damage depends on several factors and may therefore be erratic.
The second impediment is to determine the boundaries of the area where depreciation of houses is effectuated. One of the most common ways to determine this boundary is to com- pare transactions of houses located in the suspected area with transaction elsewhere outside the suspected area, and select these locations based on equal (spatial) characteristics. When all characteristics are similar, a price difference between houses from the two locations is then assumed to be caused by the only variant, which is risk of damage caused by earthquakes. One of the important questions raised is if social economic status should also be considered as a relevant spatial characteristic. If we add this, the boundary of the affected area in the Gronin- gen case shifts and more properties would be entitled to compensation. Since social economic status is an important determinant of house prices it seems that omission of socio economic data will produce erratic results.
The third issue considered regarding the method is the assumption that locations only differ on one variant, in this case risk of property damage by earthquakes. Besides the risk of damage by earthquakes, other differences are observed as well. Such differences are for instance the reputation of an area which can be expressed (not measured) as the unwillingness to move or live there. Such reputation is most likely an additional cause for lower prices in the afflicted area. In the paper we illustrate these considerations reviewing a model which has been selected by the Dutch government to work out a compensation scheme for home owners living in the af- fected area in Groningen. We demonstrate that this model is flawed because of not sufficiently accounting for the impediments discussed. We also propose an alternative method and model in which we try to work with and work around the impediments indicated, and we argue that the resulting model is an improvement since it incorporates the best available solutions to the impediments discussed. We propose next steps in the development of a better model by proposing a method to construct a latent combination variable accounting for the consumer’s behavior toward buying and selling property that includes all possible factors that may influ- ence this behavior. Without pretending to have developed the perfect model, our efforts have

extended the array of options for the court to consider in its quest for a balanced verdict on compensation, based on state of the art econometric modelling.
Vertaalde titel van de bijdrageModelling the depreciation of houses due to the risk of induced earthquakes in Groningen
Originele taal-2Dutch
Pagina's (van-tot)12-72
Aantal pagina's61
TijdschriftRuimte en Maatschappij: Vlaams-Nederlands tijdschrift voor ruimtelijke vraagstukken
Nummer van het tijdschrift1
StatusPublished - sep-2020

Citeer dit