In early July, the UN released its long-awaited, inaugural Global Risk Report. The report is based on a risk perception survey of 1,000+ stakeholders from across the multilateral system, including UN Member States, UN employees, civil society, private sector, risk experts, and academics. The survey asked respondents to rate on a scale of 1 to 7 how likely 28 risks were to negatively impact a large portion of humanity by 2050 and how severe the impacts of each would be if the risk were to occur by 2050. Risk importance seems to have been defined as the product of the two.
“Climate Change Inaction” was perceived as the #1 most important global risk by respondents.
Data from the UN Global Risk Report. (2025). Figure 1.
Additionally, participants were asked to rate the ability of multilateral institutions to identify, reduce the likelihood of, and mitigate the negative impact for five randomly selected risks. Each on a scale of 1 to 7. The average score per risk was called ‘risk preparedness’.
In general, respondents thought that the more important a risk was, the better prepared for it the multilateral system was. A key outlier is “Mis-and disinformation”, which was perceived as an important risk (#3) for which the level of preparedness of multilateral institutions was perceived as low. Below is our attempt to replicate Fig. 8 of the report. The values for some risks like multilateral institutional collapse & climate change inaction slightly differ; the line we used is a simple linear trendline.
Risk preparedness & risk importance matrix. Data from the UN Global Risks Report. (2025). Data for Importance from Fig. 1, for Preparedness from Fig. 7.
Overall, we think this report is a good first step towards strengthening global risk awareness and analysis, emphasizing the need for collaboration and foresight. At the same time, we also highlight that risk perception, while valuable, has methodological limitations. Finally, we make several suggestions for how future iterations of the UN Global Risk Report could fill remaining gaps.
Three reasons why the inaugural UN Global Risk Report is a good first step
1. The multilateral system has a role to play in global risks
The global scope of risks necessitates governance at both national and international levels. Acting alone, national governments can be incentivized to underinvest in reducing risks from activities within their borders that can have global spillovers, as they don’t pay for most of the consequences. Vice versa, countries that deal with global risks originating from abroad will often focus on local mitigation measures, even though the same investment in prevention at the source might lead to much more effective risk reduction. The multilateral system, because of its global scope, needs to play a role in coordinating disaster risk management for such challenges between governments.
In the report itself, “action by multiple governments” was perceived as the most effective response to global risks, and the UN Global Risk Report is a step in that direction.
2. The report is forward looking
The world is not static. It is important to not just consider the hazards that have occurred in the past but also future and emerging risks. As highlighted in the Report on “Existential Risk and Rapid Technological Change,” many of the absolute worst-case scenarios are linked to future and emerging technologies. As such, we support the involvement of the UN Futures Lab in the Global Risk Report, the use of multiple time horizons, and the use of a range of future scenarios.
3. The report considers cascading risks
We support the report’s recognition that risks are interconnected. As argued in “Hazards with Escalation Potential,” a deeper understanding of hazard escalation can reveal potential circuit-breaker actions to slow or halt the cascading impacts of disasters.
Overall, the existence of a UN Global Risk report is important as a public good. The UN has a unique role, network, and legitimacy, and as such the report adds an important perspective from the multilateral system that complements similar reports from other institutions.
Three limitations of risk perception
The first UN Global Risk Report is a risk perception report. Risk perception reports are informative of how decision-makers or the population at large think about and prioritize global challenges. However, risk perception surveys primarily provide evidence on perceptions, not on risks. In other words, there is a difference between “What are the most important Global Risks?” and “What are perceived as the most important Global Risks?”. The following are three challenges that risk perception surveys, including the UN Global Risk Report, face.
1. Survey participants should be selected based on the report’s goals
Should the report be a representative “worry barometer” of multilateral stakeholders or should it be an expert survey of what the multilateral system ought to worry about?
Risk perception reports often mix the aims of a representative survey with those of an expert survey aiming to capture “diversity and expertise in risk perceptions”. While we understand the rationale for this, choosing either or both options without mixing them would be methodologically cleaner.
- A representative survey aims to reflect the perceptions and opinions of a reference population. This is best achieved with a large sample size and by adjusting for demographic differences between the survey participants and the reference population. In a representative survey it can also be interesting to break down results by various demographic criteria as for example provided in the “risk profiles” by the World Economic Forum.
- An expert survey aims to reflect the consensus among individuals specifically invited for their relevant expertise. It may want to consider covering different aspects of relevant expertise, but will not aim to be representative of the population at large and it is fine to have a lower absolute number of respondents.
2. Rating risks on a qualitative scale has drawbacks
The Global Risk Report evaluates risks on a qualitative scale from 1 to 7. This is easy for survey respondents and captures some variation for likelihood and severity. However, it also comes with inherent limitations. There is no guarantee that survey respondents assign the same likelihood/severity ranges to these categories, and different risks on a survey response list may span multiple orders of magnitude in likelihood and impact.
First, if we zoom out and project the risk perception on x- and y-axes that reflect the full possible answering range, all 28 risks form a cluster. This hints at range compression and centering bias, which are known cognitive biases for evaluations on a qualitative scale from 1-7.
Risk preparedness & risk importance matrix with full axes. Data from the UN Global Risks Report. (2025). Data for Importance from Fig. 1, for Preparedness from Fig. 7.
At the same time, averages can hide richer information: Whether respondents converged on values near these averages or whether they disagree. Crossed with demographic data (e.g., field, age, sector), such disaggregated analysis could help understand various risk perception profiles (akin to the WEF risk profiles).
Second, it is inherently difficult to compare risks with a large range of likelihoods and impacts on a scale. One challenge that we have highlighted before is that extreme outlier events matter a lot for overall disaster deaths. Based on the International Disaster Database (EM-DAT) 0.044% of disasters caused more than 500,000 deaths and account for 52% of total deaths from disasters. Yet extreme impact risks are systematically underassessed on an impact scale from 1 to 7. In national disaster risk frameworks, where a similar number of impact categories are tied to quantitative ranges, deaths from the COVID-19 pandemic have already exceeded the maximum category threshold by more than 100x.
Quantitative point estimates of annualized expected deaths and damages require more time from survey respondents but they would avoid these distortions.
3. Risk perceptions are subject to availability bias
Availability or recency bias means that survey respondents assess a risk as higher if they can easily recall recent examples of that risk category. Risk perception surveys are more news-cycle driven and have a shorter half-life than risk assessments by domain experts. This is particularly relevant for:
- low probability, high impact risks (e.g. new pandemic, weapons of mass destruction) where risk perception is better understood as a lagging indicator of recent disasters rather than a reliable predictor of future risk levels;
- risks with persistent but highly dispersed impacts (e.g. chronic health issues), which generally receive less media attention than disasters with multiple deaths per event.
As such, risk perception also has greater year-to-year variance than domain expert risk assessments. The data for the Global Risk Report was collected between February and May 2024 and we hope the UN will reduce this lag in future iterations. For example, we might expect that “economic fragmentation” would rank higher on the list of concerns after “Liberation Day” in April 2025.
Three ideas for future reports
We are aware that the UN operates under time, personnel, and financial resource constraints. Still, we do want to highlight a few future directions for the next report, scheduled for December 2026, that we would find promising.
1. Time the release of the report with key UN events
The success of the World Economic Forum’s Global Risks Report cannot be copied through its methodology alone. A key source of the WEF report’s persistent relevance is arguably that it is linked to the Forum’s Annual Meeting and performs an (informal) agenda-setting function for that meeting. In that sense, the original target of releasing the UN Global Risk Report in the run-up to the Summit of the Future would potentially have been more impactful. Options for events to which future reports could be tied to include UNGA High-level Week, the High-level Political Forum on Sustainable Development, the SDG Summit, or the Global Platform for Disaster Risk Reduction.
2. Consult domain experts for in-depth assessments
There is a challenge that expertise on global risks is fragmented and that many experts who are very good at assessing the specific risk they have expertise in, are not necessarily best positioned to accurately assess and compare a wide range of 28 risks without significant additional research. Rather than asking everyone to assess all risks, national risk assessments often seek out experts on specific risks and iterate with them to better understand the likelihood and impacts of these risks. Comparability across risks is achieved by using the same impact metrics and/or ensuring different impact metrics (e.g. loss of life, monetary damages, cultural loss) can be monetized for comparison. Such an approach does not have to replace risk perception, but it would provide an important second pillar. A third, innovative approach, could be to also use individuals with a proven track record in forecasting and compare their forecasts with both risk perception and expert assessments.
3. Consider specific gaps that national risk assessments leave at the global level
Aggregate and compare national risk assessments of global risks: Cross-border risks are included in national disaster risk assessments to the degree that they affect the specific country in question. However, comparing national risk assessments shows that governments assign different likelihoods to the same global risks. The UN Global Risk Report could be a good place to compare national estimates of global risks.
Map global critical systems and their interlinkages: Many national disaster risk management agencies have lists of critical infrastructures within their territory. Several have also mapped interdependencies between critical infrastructures. The mapping of critical global systems—from the global food system, to the global energy system, to global digital infrastructure—is still underdeveloped (see here for an attempt). Including a more infrastructure-oriented perspective in the Global Risk Report could help to identify potential points of intervention, with the caveat that a detailed list may not be suitable for publication due to information hazards.
Discuss measures to rapidly scale the global supply of critical global goods: A severe global catastrophe with systemic impact requires a structurally different response than a smaller one. In a local disaster, actors can redistribute national system-capacity to help. In a national disaster, governments mobilize international help to bring critical goods and services to the affected areas. However, in the worst case, there may be insufficient global capacity to provide critical goods and services. In a global system shock, we cannot rely on help from Mars. Rather, we need to research and build the capacity to rapidly scale the global supply of critical goods and services, such as food, in a crisis.
Global risks require continued engagement
Overall, we welcome the UN Global Risk Report as a new addition to the global risks community, and we hope this still young ecosystem continues to evolve and grow to foster effective and evidence-based global risks management.