Summary: the severity of climate impacts on WASH services is uncertain. “Low-regrets” investments or interventions are those which generate net economic benefits under a range of the most plausible scenarios of climate impact severity. The concept is explored in Figure 1, which illustrates relationships between net benefits and the severity of climate impacts for different types of high/low/no-regrets options. It is also important to explore non-climate uncertainty, ideally in a probabilistic way.
Uncertainty is when we have imperfect information about variables in the present or the future. Even though the effects of climate change are increasingly upon us already, the scale and nature of their economic impacts remain uncertain (Burke et al., 2015). The further into the future the projection, the more this uncertainty increases (IPCC, 2022), because: (i) many variables interact in determining climate impacts; (ii) we can (and must) reduce greenhouse gas emissions to mitigate the worst impacts, and any effect of those actions is also uncertain.
The higher levels of water and sanitation services sought by SDG 6 are characterised by infrastructure assets with useful lives of 20-50 years or more (Hutton and Varughese, 2016). It is particularly important to characterise the vulnerability of such long-lived infrastructure to climate risks, especially since retro-fitting can be more expensive than designing for uncertainty upfront (Chester et al., 2020). I go into some of the climate risks to WASH services in this post.
Benefit-cost analysis (BCA) is the most commonly-used economic evaluation method for appraising WASH investments [a short introductory paragraph on BCA is below this post]*. In appraisal of interventions for adaptation or resilience, “no regrets” interventions are those which generate net benefits under all future climate/impact scenarios (Heltberg et al., 2009). A more achievable principle, endorsed by the IPCC (2012), may be aiming at least for “low”-regrets interventions. These are interventions which generate net benefits under a range of the most plausible scenarios. However, they also account for the risk that we might “regret” additional investment in adaptation/resilience if climate impacts are not as bad as expected. No-regrets options would be first choice, and often they will be available. However, low-regrets options may be important if adaptation/resilience increases costs substantially in relation to benefits in a “no climate change” scenario.
Low-regrets thinking has been applied in identifying opportunities for short/medium-term climate risk reduction within development interventions (Conway and Schipper, 2011). Identifying low-regrets options can also help reduce the risk of maladaptation (Barnett and O’Neill, 2010). This line of thinking can be applied whether what is being evaluated is a whole new investment in WASH services, or options for adapting/upgrading existing WASH services.
A few years ago, I was part of a three-country study looking at risk assessment and economic appraisal for adaptation to climate change in WASH (Oates et al., 2014). In making the economic arguments, we used a diagram which I’ve simplified here (Figure 1), and which I think Kit Nicholson came up with. The x-axis plots the severity of climate impacts (broadly defined) as an uncertain continuous variable. The y-axis plots the benefit-cost ratio (BCR) [see explanation at bottom]* of intervention options. The threshold where benefits equal costs on average over the time horizon (e.g. 20 years) is shown as “1”. In simple terms, we want to be above the green band, but we don’t know where we’ll be on the x-axis.
Plenty of WASH infrastructure constructed in recent decades might be climate risky (blue line A), i.e. in the absence of climate impacts it looks economically attractive, but as climate impacts worsen then BCR<1. Designing for the worst-case scenario may result in investments which are high regrets (orange line B), i.e. over-designed such that climate impacts have to be very severe before BCR>1. No-regrets options (both green lines C) are any interventions for which BCR>1 regardless the severity of climate impacts. Low-regrets options (yellow line D) may have BCR slightly below 1 when climate impacts are small, but gradually appear more attractive as climate impacts worsen. Low-regrets options need not necessarily have BCR<1 in the case of “no climate change”, but at least they would need to have lower BCR in that scenario than an option without investment in adaptation/resilience. While BCAs often present decisions as “doing something” versus “doing nothing”, this framework aims to account for the fact that in the real world there are usually multiple options under consideration.
A simplified WASH example can help illustrate. A team is planning a piped water supply with a treatment plant fed by a river intake, and the risk is identified that turbulent flows resulting from an extreme weather event may damage the intake. A “climate-risky” option might be to design the intake to withstand a flood of a given height with a 25-year return period, which is fairly likely to be exceeded within the useful life of the infrastructure. A “high-regrets” option might be to design for a 200-year return period, which would be more expensive, but increasingly worth doing as the probability of climate change-induced floods increases (Figure 1). A low-regrets option might be somewhere in-between. The reality is more complex than this, and there are many specific options within this scenario, related to, e.g. overflows, intake design, floating booms, early warning systems, etc. (Howard and Bartram, 2010).
There are some qualifications to make regarding this way of framing adaptation options. First, this framework does not make value judgements, e.g. high-regrets options are not necessarily a bad idea. However, since all investments have an opportunity cost (i.e. resources are scarce), high-regrets options may be less desirable from an equity perspective, because more people in a given year could be provided with WASH services under a low-regrets option. Second, while I often refer to these interventions as “adaptation options”, many might comprise what we should be doing anyway given existing climate variability, and the need to be resilient to risks other than the climate.
Third, many non-climate parameters in BCAs are also uncertain (e.g. costs, health effects, uptake, maintenance etc.), but this framework puts the focus on uncertainty about climate impacts. Bands incorporating uncertainty of many other parameters may therefore be more appropriate than lines. The low-regrets option from Figure 1 could be assessed in a probabilistic sensitivity analysis (PSA) per climate scenario. Such a PSA would posit plausible probability distributions for key parameters (Briggs, 2000), then run a Monte Carlo simulation with (say) 1,000 iterations. An uncertainty interval could then be posited by graphing the range of the middle 95% of iterations within a band, such as in Figure 2. This line of thinking is the main thing that is new in this post, as compared to the 2014 work (Oates et al., 2014).
Fourth, one challenge in undertaking such analyses is that, due to “deep uncertainty” in the context of climate change, it is hard to ascribe probabilities to many key variables (Hallegatte et al., 2012). Nonetheless, a Bayesian approach to uncertainty requires that the analyst makes their best estimate at the shapes of probability distributions (Briggs, 1999). Simply leaving variables out of the analysis, or not doing a PSA at all, is the same as assuming they are known with certainty. Assuming a uniform distribution for a given parameter only makes sense if the aim is to explore possible heterogeneity across settings, rather than estimating a realistic mean and uncertainty interval to inform a specific decision in a given setting. Expert opinion, tested in scenario analysis alongside the PSA, is therefore likely to play an important role. Fifth, in the real world, the “severity of climate impacts” is not a single continuous variable as in Figures 1 and 2. The IPCC provides multiple projections, and practically it would make sense to undertake scenario analysis using those.
In conclusion, I suggest that appraisal of investments in WASH infrastructure adaptation or resilience can be informed by a “regrets” perspective focused on climate uncertainty (Figure 1), but also taking account of uncertainty of non-climate parameters (Figure 2). Low-regrets options are those which generate net economic benefits under a range of the most plausible scenarios of climate impact severity.
*BCA combines all the consequences of an intervention (e.g. saved time, reduced disease, quality of life gained) and places a monetary (e.g. US$) value on them. These monetised benefits are then compared to the costs of an intervention over time, with discounting. Metrics for comparing options include the net present value (=benefits–costs) or the benefit-cost ratio (=benefits/costs). The benefit-cost ratio (BCR) is often communicated in terms of US$ X economic returns on US$ 1 invested. If the BCR is greater than 1 (the clearing rate or threshold) then the intervention has net benefits, and if less than 1 it does not. Benefit-cost ratios of different intervention options can be compared to assess their relative efficiency, although other factors should be taken into consideration (equity, feasibility, relative size of net benefits, etc.)
WASHeconomics.com 