On the contingent valuation of mortality risk reduction in developing countries

Abstract

Using the contingent valuation method in developing countries to value mortality risk reduction is particularly challenging because of the low level education of the respondents. In this article, we investigate whether some brief training regarding probability and risk concepts has any significant effect on the willingness-to-pay (WTP) responses. We elicit individuals’ risk perceptions by providing information on age specific mortality risks and find that people on average overestimate the mortality risk at younger ages and underestimate it at older ages. Our results indicate a significantly higher WTP for the trained sub-sample and WTP is sensitive to the magnitude of risk reduction for both the sub-samples.

Acknowledgements

The author thanks Fredrik Carlsson, Håkan Holm, Olof Johansson-Stenman, Peter Martinsson and an anonymous referee for very useful comments and suggestions. Research grant from the Swedish International Development Cooperation Agency (Sida) is highly acknowledged.

Notes

¹ There are some studies focusing on valuing health risks in developing countries but not particularly focusing on obtaining the VSL estimates (Cropper et al., 2004).

² This means e.g. that WTP should be twice as high for a two-fold reduction in risk. For substantially large risk changes, however, theoretical prediction of proportionality would not hold, i.e. WTP would increase but less than proportionally to the risk changes and hence the resulting VSL can be smaller compared to the case when small risk changes are valued.

³ For experimental results suggesting disparities between hypothetical and actual (real) WTP, also see Johannesson et al . (1998), Frykblom (2000) and Bothelho and Pinto (2002).

⁴ Alberini (2005) discusses about the robustness of CV estimates and different types of outliers e.g. WTP outliers in relation to income.

⁵ See Siegel and Castellan (2000), for a description of the test.

⁶ An appropriate econometric model for analysing the WTP data that includes zeros would be a two-equation sample selection model that allows for modelling zero and positive WTP separately (Carlsson and Johansson-Stenman, 2000; Yoo et al ., 2001; Strazzera et al ., 2003). However, with only 13 (nonprotest) zero WTP observations in our case, it is doubtful whether a sample selection model can be justified; it is also unclear whether there is any true negative WTP.

⁷ As the distribution of income in our sample is highly skewed we estimated separate models excluding relatively high income. However, as the coefficient of income remains roughly the same, we decided to keep them all in our final model presented here.