![Sample our Geography journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5937/TOC-Subject-Sample-2021-Geography.jpg"})
Abstract
The impact of flood-risk on local employment has been almost entirely neglected in the empirical urban economics literature. This omission is particularly anomalous in the context of climate change. We extend the literature in four ways. First, we argue that competition for land between firms and households will generate an endogenous role for house prices, which we estimate using a generalised method of moments two-stage least squares spatial econometric model. Second, we model interaction effects between agglomeration and flood-risk using a gravity-based agglomeration measure. Third, we utilise a high-resolution flood-risk measure which incorporates both flood frequency and severity. Fourth, we use a high-resolution measure of employment to capture local effects. We find that agglomeration economies have a significant mitigating effect on flood-risk. This is potentially important because it suggests that flood-risk may have a more deleterious effect on employment in areas where economic agglomeration is weak. Policy-makers, insurers and planners cannot, therefore, assume a uniform effect of future changes to flood-risk as a result of climate change, and this needs to be taken into account when estimating the costs and benefits of interventions to reduce or underwrite flood-risk at particular locations. Our model offers a robust methodological basis for such estimation.
Acknowledgements
This paper is based on research from the CREW (Community Resilience to Extreme Weather) project, funded by the UK Engineering and Physical Sciences Research Council. We are grateful to Aidan Burton and other CREW colleagues based at the University of Newcastle for providing the input variables necessary for estimation of the flood-risk variable. We are also grateful to Nationwide, Experian and Ordinance Survey for data on house prices and neighbourhood characteristics. Finally, we would like to thank CREW researchers, stakeholders/advisors and participants at the European Real Estate Society (ERES) 2011 conference for their helpful comments (NB: this paper was awarded the ERES Journal of Property Research Prize 2011 for Best Paper in Real Estate Economics).
Notes
1. Endogeneity means that a variable is determined within the system. Such variables are dependent on other variables within the model. Exogenous variables, on the other hand, are determined outside of the system, by forces external to the model.
2. Arauzo-Carod et al. (Citation2010, p. 685).
3. Existing house price research on flood-risk does not take into account endogenous employment effects, though there is a growing hedonic literature, more generally, that incorporates employment effects such as the polycentric model of Osland and Pryce (Citation2012).
4. See Chen, Pryce, and Mackay (Citation2011) for a more detailed survey of the housing economics literature on flood-risk and climate change.
5. The simplest case is distance = 1 denoting contiguity or 0 indicating that locations are not contiguous.
6. Locations cannot be neighbours of themselves, so the main diagonal of W comprises zeros.
7. Ordinary least squares.
8. Two-stage least squares.
9. Not to be confused with dependent variable E.
10. The grid search procedure is based on an employment model without spatial lag and spatial error terms.
11. We also experimented with an alternative approach where the house price variable was estimated at each LSOA centroid (rather than as an average of prices estimated at postcode centroids within each LSOA). In the event, this variable proved to be very similar to the one used in the study and did not affect the results in the two-stage least squares models.
12. Moran’s I is a well-established statistical measure used to test for the presence of spatial autocorrelation.
13. This is a standard statistical procedure used to adjust for autocorrelation in the error term.