Residential water demand responds asymmetrically to rising and falling prices

ABSTRACT

This paper econometrically estimates residential water consumption in Germany between 2007 and 2013 based on a panel of almost 3000 supply areas. In particular, the analysis distinguishes periods of rising and falling water and sewage water prices. The short-run (long-run) price elasticity is estimated at around 4.2% (13%), but water demand appears to respond asymmetrically to rising and falling prices. When prices are rising, the short-run (long-run) price elasticity is around 6.5% (17%). When prices are falling, the short-run price elasticity is not statistically different from zero, and the long-run price elasticity is estimated at around 12%. Additional results illustrate that employing average prices instead of marginal prices results in substantially overestimating the price elasticity. These findings are particularly relevant for utilities and regulators planning to alter the tariff structure towards a higher fixed fee and a lower volumetric fee.

KEYWORDS:

• Water consumption
• econometrics
• tariff
• price elasticity
• rebound
• panel data

JEL CLASSIFICATION:

• Q21
• Q25

Acknowledgments

Part of this research was supported by a research grant from the Fraunhofer Institute for Systems and Innovation Research, Karlsruhe, Germany. The authors further thank Franziska Willenbacher for collecting the data, and Gillian Bowman-Köhler for proofreading the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Results from estimating residential energy demand provide mixed results. The findings from Gately and Huntington (2002), Haas and Schipper (1998) or Ryan, Wang, and Plourde (1996) suggest that – in absolute terms – the price elasticity of the demand for oil and gas is larger when prices are rising than when they are falling. For electricity, however, Miller and Alberini (2016) find no evidence of such an asymmetric response.

² Similarly, it is not clear whether household electricity demand responds to marginal or average prices (e.g. Frondel and Kussel 2019).

³ To illustrate, suppose there is no volumetric fee at all, so the utility has to cover its total costs (= total revenues, assuming zero-profits) from the fixed fee only. Keeping profits at zero, a one percent increase in water consumption is then matched by a one percent decrease in the fixed fee and hence in average revenues.

⁴ Espey and Espey (2004) come to a similar conclusion for electricity demand.

⁵ See Reynaud and Romano (2018) for a recent overview.

⁶ Under this price setting, endogeneity should not be a problem. Employing instrumental variable methods, Schleich and Hillenbrand (2009) found no evidence that water prices in Germany are endogenous.

⁷ German water statistics do not distinguish between private households and small businesses such as bakeries, butchers, or hair dressers. Thus, the data on water consumption also include the consumption of these small businesses. Our econometric analysis attempts to control for this.

⁸ Except for the study by Müller (2015), all other analyses of German water demand relied on average prices to estimate the price elasticity.

⁹ The recent meta-analysis by Havranek, Irsova, and Vlach (2018) suggests that, accounting for publication and endogeneity bias, the income elasticity is approximately 0.15 or less. Hence, finding low or even negative values in any particular sample should not be surprising. Also, richer households may be more likely to purchase water-saving devices and more water-efficient appliances, suggesting a negative relation between water consumption and income.

¹⁰ In alternative specifications, we also included rainfall and average temperature during summer months (April to September), and the number of wells. However, the coefficients associated with these variables were far from being statistically significant. Also, including these variables lowered the number of observations because of missing data. To save degrees of freedom, these variables were therefore not included in the final specification.

¹¹ However, the samples are not identical, since data on the fixed fee was missing for a few supply areas for some years in the average price model. Using identical samples, we also find that short-run and long-run price elasticities are substantially larger for the average price model than for the marginal price model.