Risks, uncertainties and complexity: Perspectives from a macroeconomist

Written by Dr Angelia Grant, First Assistant Secretary and G20 Sherpa, Multilateral Economic Engagement Division, Australian Department of the Prime Minister and Cabinet, with support from Karen Goodwin BA, LLB, FGIA.

In May 2024, Dr Grant delivered a keynote presentation at the Governance and Risk Management Forum, held by the Victorian chapter of the Governance Institute of Australia. This article, based on the presentation, summarises her perspectives as a macroeconomist on risks, uncertainties and complexity in the context of the times in which we find ourselves, with significant challenges that are intertwined across the international, domestic, economic, and security policy spaces. Dr Grant believes that the sharing of frameworks across sectors and across disciplines is critical to successfully navigating the current environment, and shares her framework in this spirit. Dr Grant draws on her background as a macroeconomist for the Australian Government, including her experiences in economic forecasting, macroeconomic and climate change modelling, and her work in international settings as an Alternate Executive Director on the Board of the International Monetary Fund and as Australia’s G20 Sherpa.

Risks

Economists have a particular way of assessing risk preferences – the appetite that an individual has for risk – and it all comes down to their expected utility of a risk, and whether an individual gets higher utility from the same expected outcome if they take a risk or not.

This is best explained with a simple example – imagine you have a choice between a guaranteed payment of $10,000 or taking a risk that gives you an expected payment of $10, which is based on the probability of two outcomes – there is a 50 per cent chance that you get nothing and a 50 per cent chance that you get $20,000 (0.5 x $0 + 0.5 x $20,000). Would you like the guaranteed payment, or would you like to take the risk in the hope that you get $20,000? Those who would take the guaranteed payment are considered risk averse because their expected utility is higher when they have certainty about an outcome. Those who would take the risk are considered risk loving because their expected utility is higher from taking a risk. And those who are indifferent between the guaranteed payment and taking the risk are risk neutral.

To assess risk preferences using the expected utility of a risk, it is necessary to know the probabilities of particular outcomes. So, simply put, I have been taught that risks have a probability, a mathematical likelihood of occurring. That means we can work out the expected outcomes and consequences of taking a particular risk, undertake sensitivity analysis, and make fairly well considered decisions.

Sensitivity analysis allows for the assessment of the impact of key assumptions. It allows you to ask ‘what ifs’ that are focused on what happens if one of the parameters is different – you can figure out its probability and the approximate outcome if the risk is realised. For example, Statement 8 in Budget Paper No. 1 of the 2024-25 Budget sets out sensitivity analysis if iron ore prices are higher or lower than expected – it simply allows for one key variable to change and reports on how sensitive the estimates of the underlying cash balance are to that change in assumption.

The requirement that risks have a probability means that risks sit in a fairly comfortable zone for many decision makers. We collect and use all the information available on the decision, we work out the options, assess the probabilities of each particular outcome occurring, undertake sensitivity analysis and choose an option that best meets our purposes, achieves our specified outcome, or addresses our specified problem.

One exception to risks being able to be assessed in this manner is when we need to consider tail risks. Tail risks are risks that sit out in the tail of the probability distribution, so have low probabilities but significant, and often dire, consequences if realised. While central forecasts are based on the balancing of risks, the balancing of tail risks is not really possible so they are best considered as uncertainties.

Uncertainties

Unlike risks, within my economic framework, uncertainties do not have a probability that can be accurately estimated. This may be due to a lack of reliable data, possibly because the future is not expected to be representative of the past. Uncertainties may have a range of probabilities and they may be contingent on other uncertain factors that make the probability so uncertain that it is not useful to use it in the same way as it can be used for risks.

While sensitivity analysis is where we can go for most risks, scenario analysis is where we can go for tail risks and uncertainties. Scenario analysis is used when we are being asked to choose, or deal with, different futures. Rather than asking ‘what ifs’ about a specific parameter, we are asking ‘what ifs’ about a possible future state. We are making more significant changes to our assumptions, parameters and variables. And we are doing so in a way that requires significant judgement, as we are determining what matters most to us when it comes to possible futures, and how we think those values, priorities and underlying variables may change.

Climate change is a good example of an uncertainty that can be better understood with the use of scenario analysis. The United Nations Intergovernmental Panel on Climate Change (IPCC) has set out ‘shared socioeconomic pathways’, which are scenarios that depend on the interaction of a number of choices and outcomes. They examine different ways in which the world may evolve and how that will impact the climate.

Scenario analysis, by its very nature, requires more vision and creativity than sensitivity analysis. It requires that we imagine different futures and consider the choices that will influence those different futures; and it gives us a deeper sense of the consequences of making particular decisions. For scenario analysis to be most useful, we need to be conscious of the inherent subjectivity of this approach, and how it is affected by our underlying values, ideologies and unconscious biases. As such, it benefits from diverse and inclusive teams, and from collaboration across disciplines, industries and sectors.

The helpfulness of scenario analysis lies not in any final specific number, but in a deeper understanding of the assumptions, parameters and variables that matter most when it comes to the future states we are considering.

Complexity

Complexity occurs when risks and uncertainties are combined. It requires us to use both sensitivity analysis and scenario analysis to assess options and make choices, and it makes our assessments much more difficult but also more enriching.

In our current environment, risks and uncertainties are not only combined across one issue, but across multiple issues with different potential directions and trade-offs. How do we undertake analysis on the scenarios of climate change, artificial intelligence, global development, and shifting geopolitical relationships when all these scenarios affect each other? In this environment, scenario analysis can seem like an overwhelming prospect on the one hand, but an empowering prospect on the other hand.

Scenario analysis allows us to consider the reality of different envisioned futures, both positive and negative, which result from significant, interconnected, structural issues. It allows us to be clear about risks versus uncertainties, and lets us more deeply analyse them, and their interconnectedness, within different possible futures.

This underlines the importance of multi-disciplinary teams in dealing with complexity, and the importance of sharing frameworks across sectors and across disciplines. Further, it makes it understandable in the current environment, with interconnected risks and uncertainties, that we are seeing an increase in expertise in futures analysis.

The combination of risk and uncertainty across multiple issues with different directions and trade-offs makes it difficult to define and breakdown problems because there are often multiple objectives and issues at play. When considering the difference between complicated and complex problems, I like the . https://thecynefin.co/ And I think that it sits neatly with a framework that draws a distinction between risk, uncertainty and complexity.

The Cynefin framework says that for complicated problems, it is necessary to assess the situation, collect the relevant data and expertise to analyse it, weigh up the different options using a strong framework and good practice, and then decide on the best response. The assessment of the best response depends on a clear assessment of the problem that is being solved and the framework that determines success. For example, a macroeconomist may choose an option based on the criteria of improving productivity, resilience or welfare.

This sounds a lot like what we do when considering risks. However, complex problems are more unpredictable, as there is incomplete data and information, no amount of extra searching is going to help, and we cannot control the situation. There is an insufficient amount of reliable data to calculate the probability of specific risks, and there are intertwined and dependent relationships at play.

This sounds like uncertainty, and it requires that we set up our problem solving and decision making in a different way. It requires that we take a collaborative approach that focuses on the next step of action, sets out a trajectory, and looks at the points where available data and information will emerge, and experiments can be conducted. Communication about the overarching aim is essential, even though we may not be exactly sure how to get there yet.

Complex problems require us to use scenario analysis in a way that helps us break down our assumptions, parameters and variables at each step of the process rather than in a state-to-state comparison. We can locate and understand decision points and the interconnections between decisions and consequences for our future state. Scenario analysis also helps us identify the conditions for trigger points. These conditions may occur suddenly or they may build slowly, with scenario analysis enabling us to identify cumulative and interconnected risks and uncertainties.

Conclusion

We find ourselves in times of elevated risks, uncertainties and complexity.It is vital that we collaborate and share frameworks across disciplines, industries and sectors. It is vital that we use both sensitivity analysis and scenario analysis to understand and navigate risks, uncertainties and complexity. And it is vital that we adapt and shift how we solve problems and make decisions.

Our current world presents an opportunity for all of us to become even better problem solvers, decision makers, and leaders. Leaders who constantly look for new sources of information, feedback loops and learnings; leaders who engage honestly and constructively; leaders who build the capability of their teams; and leaders who acknowledge their own need to adapt, and who adapt with purpose.

We must know who we are, what we are working towards, and how we are going to work with others to achieve it. We must embrace uncertainty and complexity, and weigh up options with clarity about our values and purpose, and the direction in which we want to be headed, because the choices we make today are going to shape our tomorrow.