Robust Decision Making (RDM)
Best options vs. robust options
Under deep uncertainty, decision-makers need to select robust options rather than best options. A focus on best options is more appropriate when it is possible to predict particular future states. However, when the future is characterised by uncertainty, a focus on best options may carry significant risks. In this context, Rosenhead et al. (1972) suggested the concept of robust decision making.
Robustness is the flexibilities in decision making strategies against multiple future possibilities. This robustness concept distinguishes plans from decisions. A plan is a set of prospective decisions which designs a desired future and methods to bring it about, and a decision is a commitment of resources to one of the prospective decisions (Rosenhead, Elton et al. 1972 p.418). In this respect, robustness is similar to the adaptive planning, adaptive resources management, and decision-pathway approach.
The approach depicts a planning process as a series of decision nodes over time – a pathway of evaluation of risks, identifying options, choosing options, monitoring the outcome and then iterating the process at the next decision node. The approach also focuses on dealing with high levels of uncertainty and possible thresholds in the system being managed. Included in several chapters in the IPCC AR4, it is a more appropriate paradigm than optimal approaches that assume a high ability to predict future risks or outcomes of decisions.
A robust decision making process is required to be flexible to unknown future states. Therefore, instead of making the “one big plan”, the process has a series of “small plans” and a decision maker makes a decision on the first plan and the rest of the chain of plans remain as plans. These remaining plans are open to revision as future states become more clear. In short, Rosenhead (1972) defined robustness value as below:
ri = n(Si)/n(S),
where S is a subset of alternative choices which are currently considered as good or acceptable and Si is a subset of S and attainable after the initial decision making. The robustness r is the proportion of decision sets, which can be used after the decision, so that as the number is higher, the decision maker is likely to have more options in the future or avoid the bottleneck of decision selections. This concept indicated that the robust decision making is not judged by only one uniform measurement. For example, robust decision making can be judged by both the openness to the future as well as the success of strategies. This means that inaction, which keeps the future options widest, does not necessarily prove to be a robust strategy if an inactive strategy is not also a successful one.
Multiple methods to make decisions more robust
Using multiple methods or approaches will be useful to consider multiple future options and measurements of success as these are a limit what one approach can do. For example, Rayner (1998 p.30) suggested multiple dimension of measurements decision makers need to consider:
- Time scale such as short, medium, and long
- Spatial scale such as local, national, regional, and global
- Institutional level such as markets, governments, and civil society.
Only in an extreme case will a decision-maker need to consider adaptation options to fit all three dimensions; usually however, s/he needs to consider more than one dimension to make her/his decision robust. Some analysis, such as GIS, is suitable for spatial analysis whereas another approach, such as text mining, performs well with contextualised institutional analysis. In this way, multiple methods can be viewed as the surrogates of multiple criteria. It should be emphasised that multiple methods, and in consequence multiple criteria approaches, should not be confused with finding a distribution from the outputs of models. Such tools show only the range, but not the likelihood of possible outputs.
Robustness of decisions to future climate conditions
Robust decision-making concepts have been taken up by those working in climate adaptation, as a promising approach to integrate climate science information with other types of knowledge – in view of the attendant uncertainties. In this context, robust decision-making looks for strategies that perform well under a range of plausible socio-economic and climatic scenarios, rather than looking for the optimal strategy for a given climate future (Lempert et al., 2004; 2006). Future climate change scenarios are developed as well as narratives of socio-economic change, covering the range of uncertainty between different projections. Several identified adaptation options of interest are then assessed against these combined scenarios.
The important point is that strategies which are robust and expected to perform well across the range of possible future conditions, are preferable to those that are optimal for one scenario but sensitive to changes and may perform badly under a different (but equally probable) scenario (Wilby & Dessai, 2010).
Futher reading
Rayner, S. & E.L. Malone (eds) 1998. Human Choice and Climate Change: An International Assessment, Volume 1, The Societal Framework, Volume 2, Climate Change Resources and Technology, Volume 3, The Tools of Policy Analysis, Volume 4, What Have We Learned? Battelle Press, Columbus, Ohio.
Lempert R., Nakicenovic N., Sarewitz D. and Schlesinger M., 2004. Characterizing Climate-Change Uncertainties for Decision-Makers. An Editorial Essay. Climatic Change, 65(1), pp. 1-9.
Lempert R.J., Groves D.G., Popper S.W. and Bankes S.C., 2006. A General, Analytic Method for Generating Robust Strategies and Narrative Scenarios. Management Science, 52(4), pp. 514-528.
Wilby, R.L., & Dessai, S. (2010). Robust adaptation to climate change. Weather, 65(7), 180-185.