Behavioural Nudges for Water Conservation in Unequal Settings

In the period 2015-2018, the City of Cape Town experienced a sustained drought that eventually reduced the city’s usable reserves to 10% of capacity. Fortunately, there were individual water meters in many homes, and as part of the city’s response to the water crisis, it was able to implement increasingly stringent water restrictions, using both price (higher water tariffs) and quantity (lower water pressure and legislated constraints on daily use by households).

In addition to these restrictions, the city attempted to reduce residential water consumption by using eight behavioural messages; this study evaluates their impacts.1 These eight nudges were implemented at the start of the drought, just as the water crisis began to unfold. Behavioural messages were sent to domestic water users over a six-month period, starting in November 2015. A unique message was inserted into the utility bill of households in each of eight treatment groups. Since each household’s water use is recorded, this intervention also tested the ability of behavioural messages to encourage water conservation in times of water austerity, and to reinforce the effectiveness of more conventional instruments.

South Africa is characterised by extreme levels of income inequality. This inequality makes green nudges especially attractive. Unlike more traditional demand-side-management (DSM) tools, green nudges are not punitive and regressive for poor households who are especially vulnerable to higher tariffs and physical restrictions (Datta et al. 2015). Non-monetary incentives can thus be introduced across the income spectrum. Moreover, where low-income and indigent households have their water consumption subsidised (as happens in Cape Town) and are therefore safeguarded against financial instruments, they might nevertheless respond to non-monetary interventions.

Against this background, and using a very rich dataset of residential users, we were able to examine both the long-term impacts of these nudges as the drought escalated, and also the differences in treatment effects across income groups. 

The intervention involved roughly 360,000 residential households that participated in eight behavioural treatments over the period November 2015 to April 2016. The nudges took the form of messages sent as inserts with the monthly utility bill. These messages were split into two distinct groups:

The first group promoted water conservation by addressing informational failures around price and usage of water. Expecting households to optimise their water use first requires that they understand it. Few consumers know the amount of water used by common activities such as toilet flushing, washing machine use, garden irrigation, and leaking taps. Even where usage is visible, it is not always easily quantified. For example, it is not easy to estimate total water usage during a shower. Informational failures as described here are therefore likely to play a role in inefficient resource usage. Since quantifying the water used by different activities can be both complex and costly, it becomes de facto unobservable, and as such receives less weighting than other preferences (adapted from Ramos et al. 2015).

With respect to price, consumers are less responsive to price-based signals when price information is unclear and the pricing system is complex (Ramos et al. 2015; Chetty et al. 2009; Gaudin 2006). This is particularly true of water where (i) conventionally metered consumers pay for water ex-post and not at the moment of usage, and (ii) municipal tariff structures are often complex and nonlinear – such as the inclining block tariff system used by the local municipality in which marginal prices increase with consumption. Although the price elasticity of demand for water has typically been described as inelastic (Olmstead et al. 2007), recent literature suggests that consumers are simply inattentive to price changes and thus fail to respond appropriately (Chetty et al. 2009; Datta et al. 2015; Gaudin 2006). In this respect, Gaudin (2006) found that elasticity of demand increased by at least 30% when price information was provided on the bill.

In contrast, focus groups held in Cape Town prior to the roll-out of the behavioural interventions suggested that consumers from across the income spectrum were generally unaware of the quantity of water they used, the stepped tariff structure and the tariff rates. 

Given this background, the first group of messages was designed to address these informational failures around usage and price. Tips suggested methods of reducing water use and gave the water savings associated with each action. The tariff graph treatment provided a visual breakdown of the nonlinear tariff structure and situated the household’s consumption within the six tariff blocks. The financial gain treatment replicated the graphic in the tariff graph mailer and projected the monthly and annual savings from reducing consumption and moving into the preceding tariff block. Given the growing traction of loss aversion (Kahneman & Tversky 1979;​Tversky & Kahneman 1992), we also consider whether a loss or a gain framing would have a greater impact on water conservation. The loss treatment is identical to the gain treatment, but quantifies the financial dissaving of not reducing consumption and moving to a lower tariff.

The second group of messages promoted water conservation via social incentives and appeals to the public good. 

Social-norm messaging is increasingly seen as a useful mechanism for promoting proenvironmental behaviour. Allcott (2011) and Ferraro and Price (2013) find that social-norm-based appeals reduce US energy and water consumption by 2% and 4.8%, respectively, while Andor et al. (2017) find that similar social-norm nudges in Germany yielded more modest reductions (0.7%) in terms of energy consumption. The literature indicates that the social comparisons inherent in social-norm messaging facilitate social (observational) learning about the households’ privately optimal level of water usage (Allcott 2011). In the social norm treatment, a household’s consumption is compared to the average household in its neighbourhood.

Social pressure can be leveraged to drive pro-social behaviour – particularly if social recognition is provided for positive (do-gooding) behaviour. For example, in a laboratory setting, revealing identity increases public-good contributions (Andreoni and Petrie 2004; Rege and Telle​ 2004). Randomized field experiments have used social recognition to increase public good contributions in varying public-good contexts, including charity contributions, voter turnout and blood donations (Gerber et al. 2008; Lacetera et al. 2011). In the context of pro-environment conservation, Yoeli (2009) examines take-up of energy-saving technology in California. Customers whose decisions are made public are 1.5% more likely to sign up for the technology than those customers whose decisions remain anonymous. In the intrinsic motivation treatment, households were asked to help the city save water by supporting a city-led water-saving initiative and reducing consumption by 10% over the study period. In the social recognition treatment, households that succeeded in reducing their usage by 10% were publicly recognized on the city’s website.

The final treatment was framed around the public-good dilemma synonymous with the water crisis. Since the benefits of water saving are shared equally by all households irrespective of their individual contributions, there is an incentive to free ride (Hasson et al. 2010; Brekke et al. 2008). However, laboratory evidence suggests that while the dominant strategy in linear publicgood games is for each player to contribute nothing, participants do make positive but suboptimal contributions to public goods (Cherry et al. 2005). A public good appeal might alter the moral cost of water usage or, alternatively, generate conditional cooperation whereby households alter their usage in the belief that others will do the same (Allcott 2011; Gächter and Herrmann 2009). Accordingly, the public good treatment appealed to all households to conserve as much water as possible. 

In terms of the results, the social recognition and public good framings outperformed the other messages. We thus find public recognition of water conservation efforts and appeals to the public good to be the most effective motivators of pro-environmental behaviour. More broadly, over the intervention period, all treatments successfully induced a reduction in household consumption. Reductions ranged from 0.6% for the tips treatment to 1.3% for the social recognition and public good treatments. Income distribution was also a factor: social incentives were particularly effective among wealthier households. In the fifth quintile (the wealthiest households), the social norm treatment induced a reduction of 1.2%, with the treatment effect increasing to 1.9% for the public good message and 2.3% for the social recognition treatment.

These findings indicate that behavioural messages can promote water saving even when used on top of more conventional DSM tools. In fact, the five-month nudge campaign of early 2016 led to sufficient sustained savings over the subsequent two-year period to delay the city’s final crisis point by almost three weeks – enough time for the first rainfall to arrive.