Newsletter No 2 June 2011

Using choice experiments to value reductions in flood risk exposure in the Netherlands: advancing existing methodologies


Thijs Dekker, Roy Brouwer and Marjan Hofkes

ee floodClimate change is expected to increase the probability of flooding in the Netherlands and substantial investments are required to adapt to this changing environment. A choice experiment was conducted to identify the extent to which Dutch citizens are willing to pay for reductions in their exposure to flood risks.  

Benefits from public flood risk reducing policies arise in the form of prevented damage to property, individuals and their environment. In the Netherlands, the Hoogwater Informatie Systeem (HIS)-damage module (Rijkswaterstaat 2005a) is used as the main framework to quantify the monetary costs and benefits of flood events and flood protection. For example, Rijkswaterstaat (2005b;2005c) estimated the damage for a flood occurring in the densely populated provinces of North- and South-Holland in a worst case scenario of up to almost  €350 billion. 

Accounting approaches, such as the HIS-damage module, include monetary values for specific property and infrastructure damage categories. It has been argued that such estimates are incomplete in that they are based on expected property damage costs determined through ‘objective’ measurements of the risks involved, not the theoretically correct ‘subjective’ expected utility values (Brouwer et al. 2009). The benefit-cost ratios based on property damage costs alone may substantially underestimate the total benefits since it does not account for public risk aversion. These latter well-being effects, measured in economics through the concept of society’s willingness to pay to avoid a specific risk, should be added to the expected damage costs. The risk of loss of life is an example of an important element of flood risk reducing policies, but to date it remains unclear which monetary benefits can be assigned to a prevented fatality (Dekker et al. 2011). Non-market valuation techniques, of which choice experiments are a special case, can be used to estimate the economic value for risk aversion based on preferences for flood risk prevention policy.

Choice experiments are a survey based method in which respondents are presented with a series of choice tasks. In each choice task the respondent is shown a card with policy alternatives and requested to select the most preferred alternative, in our case a flood risk reducing policy, from a set of different policy options. The presented policies vary in their characteristics. For example, one policy may provide additional reductions in flood risk probabilities, while another may put more emphasis on reducing the consequences of a flood by providing compensation in the aftermath of a flood. Each policy comes at a cost, for instance through increased general income taxation or local taxes paid to water boards. By making a trade-off between the policy characteristics, including cost, in each choice task, it is possible to derive a willingness to pay (WTP) estimate for changes in a specific policy characteristic or estimate public WTP for an entire flood protection policy.

Value estimates based on stated preference methods such as choice experiments have been criticized for the mismatch between the (hypothetical) predictions based on the underlying economic framework and (real) observed choices by respondents. Due to respondent unfamiliarity with the hypothetical survey questions or the policy involved such as the proposed risk reduction policies investigated in this study, respondents may not behave in the expected rational fashion. Preference uncertainty potentially induces inconsistencies in choice behaviour as respondents do not exactly know their preferences yet or are in the process of discovering them in a sequence of choices. 

In a series of papers we attempt to bring the analytical framework more in line with observed variations in behaviour across respondents and dynamics in preferences over choice sequences. First, we investigate in Dekker, Koster and Brouwer (2011) whether respondents are able to make better informed decisions after answering a sequence of choice cards. Two representative subsamples are presented with exactly the same choice experiment. The only difference is in the first choice card, where exactly the same policies are presented at respectively a set of high and low prices. Contrary to economic theory, WTP estimates appear to be sensitive to arbitrary initial value clues. Applying a novel econometric approach to identify preference dynamics, we also do not observe convergence of preferences within and between the two samples over the choice sequence. However, preference uncertainty reduces due to learning effects. Second, in Dekker and Rose (2011) we take into account that preferences for changes in flood risk exposure vary across respondents by introducing a new probability density function, representing the degree of heterogeneity in preferences across respondents. Its simple, flexible form is able overcome drawbacks of more commonly applied distributions.

Overall, choice experiments provide a suitable framework to obtain a monetary value estimate for flood risk reducing policies, which can serve as an input for cost-benefit analysis. However, much work is still needed to bring the underlying economic framework more in line with variations in preferences across respondents. Research needs to take into account that respondents may suffer from preference uncertainty, due to a lack of experience with the presented trade-offs, and that preferences may change over the course of a choice experiment due to learning effects.

Contact:  Roy Brouwer and Marjan Hofkes


References

Brouwer, R., Akter, S., Brander, L. & Haque, E. (2009). Economic valuation of flood risk exposure and reduction in a severely flood prone developing country. Environment and Development Economics, 14(3), 397-417.

Dekker, T., Brouwer, R., Hofkes, M., & Moeltner, K. (2011), "The Effect of Risk Context on the Value of a Statistical Life: a Bayesian Meta-model", Environmental and Resource Economics pp. 1-28.

Dekker, T., Koster, P., & Brouwer, R. (2011), "Changing with the tide: A semi-parametric estimation approach to analyze preference dynamics in a choice experiment on WTP for flood risk reductions", to be presented at the European Association of Environmental and Resource Economics conference, Rome, June 2011.

Dekker, T. & Rose, J. M. 2011, Shape shifters: simple asymmetric mixing densities for mixed logit models, VU University Amsterdam, Netherlands, IVM 11/01.

Rijkswaterstaat (2005a), HIS - Schade en Slachtoffer Module Versie 2.1: Gebruikershandleiding , Rijkswaterstaat, DWW-2005-004.

Rijkswaterstaat (2005b), Veiligheid Nederland in Kaart: Overstromingsrisico dijkring 13 Noord-Holland, Rijkswaterstaat, DWW-2006-023.

Rijkswaterstaat (2005c), Veiligheid Nederland in Kaart: Overstromingsrisico dijkring 14 Zuid-Holland, Rijkswaterstaat, DWW-2006-024