Willingness to pay for reliable supplies of safe drinking water in Guatemala: A referendum contingent valuation study

Abstract

Unreliability of water systems has become a major concern in many developing countries. A referendum-format contingent valuation (CV) survey was implemented to investigate household preferences, in monetary terms, for improved water services in the small town of San Lorenzo, Guatemala. A random sample of 500 households report that they currently adopt a variety of averting measures (e.g. in-home water storage and treatment) to cope with service interruptions and low water quality. Findings also indicate that households are willing to pay an increase of more than 200% in their water bill for reliable supplies of safe drinking water.

Keywords:

• drinking water
• system reliability
• contingent valuation
• household preferences
• Guatemala

Notes

^1. The low quality of piped water has received a lot of attention in the national press. For example, Prensa Libre, a newspaper of national circulation, reported in May 3 2013 that 80% of the country's water sources are contaminated with bacteria and chemicals (see http://www.prensalibre.com/comunitario/Pais-agua-mala-calidad_0_911909098.html, last accessed on 2/11/2014). Also, in February 12 2012, Prensa Libre reported that there are 6,203 water systems contaminated by bacteria (see http://www.prensalibre.com/noticias/COMUNAS_0_644935504.html, last accessed on 2/11/2014). Such unreliable and inferior water services lead to many unintended and unhealthy effects on the inhabitants of Guatemala.

^2. Poverty indicators are based on household consumption. The national poverty and extreme poverty lines are 6574 and 3206 quetzals, respectively (see INE, 2006).

^3. Given that water system revenues and costs are usually aggregated into general accounting records (see IARNA, 2005), it is difficult to estimate how much of the municipality's revenue is spent on providing water services.

^4. From an academic perspective, it may be interesting to investigate some marginal rates of substitution between specific attributes such as system reliability and water quality, as well as the marginal monetary value of those attributes. Choice experiments (CEs) are more suitable to elicit that information than the CV method. However, CEs demand more cognitive skills from respondents, which tends to complicate their application in contexts of low education and lack of experience choosing among policy alternatives such as our study site (see Hanley et al., 2010). Both CE and CV yield similar WTP estimates (Christie & Azevedo, 2009; Santiago & Loomis, 2013). Also, from a technical perspective, it may be difficult to improve one service attribute without affecting others. For instance, water quality may be jeopardized when services are interrupted because water pipes tend to fracture due to changes in pressure thus allowing water recontamination (Boulos et al., 2005). Hence, a scenario where water quality is improved alone (i.e. without improving the system reliability) is not feasible and may lack credibility before respondents. For these reasons, the CV method was chosen over CEs.

^5. Variance inflation factors (VIFs) were estimated to investigate the reliability of estimated coefficients due to potential multicollinearity among covariates (e.g. BOTTLEDWAT and INCOME). VIFs are relatively low varying from 1.04 to 2.51, with an average of 1.46. This suggests that estimates presented in Table 2 are reliable.