ABSTRACT
Many water utilities offer price caps to reassure consumers about future prices. Researchers have focused largely on the implication of price caps on total cost of ownership or conservation rather than consumers’ preference for such guarantees. In this paper, we examine consumers’ preferences for price caps. We first consider how consumers form expectations about their water prices, and secondly how they evaluate those expected prices against guarantees. Our empirical investigation, using new survey data, confirms that as variability of historic prices increases, consumers use less historic information to form their expectations, and replace it with more subjective information, such as their perceptions of the water supplier’s investment efficiency and the quality of their water. Further, water consumers who are more motivated to protect themselves, as demonstrated by averting behaviours even when they consider service quality to be good, more strongly prefer a guarantee when past prices have been highly variable.
Acknowledgements
We are grateful to the University of South Australia, TD Meloche Monnex, and Opus (WSP) NZ for providing funding for this research.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. A model based on these principles was first proposed by Brock and Hommes (Citation1997).
2. For example, Brock and Hommes (Citation1997) show that an adaptive expectation system based on boundedly rational predictors leads to a cycle between stable and unstable predictions.
3. It is noteworthy that the literature has relied on a limited number of surveys. Most of the empirical literature on expectations formation draws on macroeconomic survey data, such as the U.S. Survey of Economic Expectations; the U.S. Survey of Consumers; the U.S. Health & Retirement Survey; and the Survey of Professional Forecasters. See Pesaran and Weale (Citation2006) for a comprehensive summary as at 2005, although more recent literature has also drawn heavily from these datasets, including Dominitz and Manski (Citation2011), Engelberg, Manski, and Williams (Citation2011), Gouret and Hollard (Citation2011), and Hudomiet, Kézdi, and Willis (Citation2011).
4. Although the survey data are nearly ten years old, our aim is to understand how people form expectations. If we have undertaken our analysis correctly, those factors should stand today and into the future. What is more likely to change is the proportion of respondents falling into different calculating classes depending on their trust in the water supplier and more recent price variability.
5. In New Zealand, water utilities/municipalities may review rates annually as stated on the Local Councils Internal Affairs website (http://www.localcouncils.govt.nz/lgip.nsf/wpg_url/About-Local-Government-Local-Government-In-New-Zealand-Council-funding).
6. Consumers who pay volumetrically for their water are typically more aware of water prices than those that do not (Troy and Randolph Citation2006). Creagh (Citation2010) found that in New Zealand, 80% of volumetrically charged consumers reported that they could recall their last water bill, compared to only 25% of fixed charged consumers. Furthermore, compared to annual household expenditures of $52,500 per year (Statistics New Zealand Citation2010), water bills averaging $300 per household per year means that water constitute less than 1% of households’ budget.
7. The New Zealand Health (Drinking Water) Amendment Act 2007 legislates compliance with the New Zealand Drinking Water Standards, and the Local Government Act 2002 requires community consultation on price and levels of service quality.
8. We also tested models that weighted more recent prices more heavily or that considered only recent shocks, but found these models to behave no better. In the interest of simplicity, we used a straight-line extrapolation.
9. We also evaluated models where standard deviation was used directly, i.e. as a continuous variable, but those models did not perform well.
10. Local Government (Rating) Act 2002.
11. Models with four types of respondents were also evaluated but none of those converged.
12. More complicated models like latest class models offer few other opportunities to test assumptions compared to simpler models such as linear regression models. Standard statistics such as R-squared are unavailable.
13. Respondents from the highest income category responded statistically inconsistently with their income category. We removed these respondents from the analysis, as well as two respondents who claimed to have experienced restrictions for more than 52 weeks in a year, leaving 1183 observations on which to test correlation between risk appetite and individual discount rate.
14. We used 100 imputations of discount rate in both models.