Lessons from COVID-19 for managing transboundary climate risks and building resilience

COVID-19 has revealed how challenging it is to manage global, systemic and compounding crises. Like COVID-19, climate change impacts, and maladaptive responses to them, have potential to disrupt societies at multiple scales via networks of trade, finance, mobility and communication, and to impact hardest on the most vulnerable. However, these complex systems can also facilitate resilience if managed effectively. This review aims to distil lessons related to the transboundary management of systemic risks from the COVID-19 experience, to inform climate change policy and resilience building. Evidence from diverse fields is synthesised to illustrate the nature of systemic risks and our evolving understanding of resilience. We describe research methods that aim to capture systemic complexity to inform better management practices and increase resilience to crises. Finally, we recommend specific, practical actions for improving transboundary climate risk management and resilience building. These include mapping the direct, cross-border and cross-sectoral impacts of potential climate extremes, adopting adaptive risk management strategies that embrace heterogenous decision-making and uncertainty, and taking a broader approach to resilience which elevates human wellbeing, including societal and ecological resilience.


Transboundary impacts of COVID-19
In the two years since its first detection in Wuhan, China, in December 2019, COVID-19 has spread to all but a few countries (Dong et al., 2021;Hubbard, 2021), with over 272 million infections and over 5.3 million deaths reported worldwide (Dong et al., 2021). For many patients, the health impacts have continued after the acute phase as 'long COVID', with more than half reporting one or more persistent symptoms within the first six months after diagnosis and ~37% at least one symptom between three and six months after (Taquet et al., 2021). Beyond the direct effects of infection, COVID-19 has disrupted public health by demanding healthcare resources that have become unavailable for other health issues (Chudasama et al., 2020;Masroor, 2020), impacting healthcare workers' physical and mental wellbeing (Shaukat et al., 2020), raising public avoidance of healthcare services (Baugh et al., 2021;Hartnett et al., 2020), increasing risks to immunosuppressed (Al-Quteimat and Amer, 2020) and elderly (Banerjee, 2020) patients, and inflating other health risks due to the physical inactivity (Pinto et al., 2020;Crisafulli and Pagliaro, 2020;Sallis et al., 2021) and mental stressors (Pancani et al., 2021;Pieh et al., 2021) associated with measures such as lockdowns and social distancing. Despite the rapid development of effective vaccines (USCDC, 2021) and administration of over 8.5 billion doses globally (Dong et al., 2021), infection rates remain high (Dong et al., 2021). New viral variants continue to emerge, some with the potential to circumvent previously effective control measures and present unpredictable new challenges at all scales from individual to global. Perceptions of uncertainty at various stages of the pandemic, and responses to them, such as panic buying, have amplified impacts throughout the system (Szczygielski et al., 2021).
The pandemic has shown that most communities are able and willing to respond drastically to crises during the period when they perceive a high level of threat and are supported by institutions in collective action to reduce the threat (Botzen et al., 2021). Together, governments have announced stimulus spending of an unprecedented $17.2 trillion (Vivid Economics & F4B, 2021). Foreign aid rose to its highest level ever in 2020, with donors providing more emergency funds for developing countries suffering from the crisis. However, this barely compensated for losses; in the same year, total external private finance to developing countries fell by 13% and trade volumes declined by 8.5% (OECD, 2021). Debt levels relative to GDP, particularly in those countries facing the double hit of lockdown and declining export revenues, have risen (Adam et al., 2020). Labour markets have been disrupted on an unprecedented scale -with losses of $3.7 trillion in income (before any benefits) globally in 2020, the equivalent of around 255 million full-time jobs (International Labour Organization, 2021). Forecasts of prolonged economic downturns and the prospect of rising public debts have the potential to trigger new sovereign debt crises, in both developing and industrialised countries (Dunz et al., 2021).
The COVID-19 experience has also added (Sachs, 2020) to the evidence that socioeconomic inequalities tend to erode resilience against multiple shocks (Hart et al., 2016;Pickett and Wilkinson, 2010). Three countries -the United States, Brazil, and Mexicoaccount for one third (32%) of the world's reported COVID-19 deaths (Dong et al., 2021) although they contain only 8.6% of the global population (2021). These countries also have very high levels of income and wealth inequality (Sachs, 2020;Roser and Ortiz-Ospina, 2013). Although a systematic analysis of this correlation is needed, it is consistent with prior evidence that economic inequalities can strongly influence societies' resilience to shocks and crises (Hart et al., 2016;Pickett and Wilkinson, 2010). At the global level, unequal vaccine access due to hoarding by rich countries leaves the poorest, most vulnerable nations to bear the brunt of the pandemic while also promoting continued viral evolution and emergence of dangerous mutant strains, thereby prolonging risks to all nations (Asundi et al., 2021;Amaro, 2021;. While the pandemic has hit the most vulnerable groups hardest (International Labour Organization, 2021;Kharas, 2020;Lenhardt, 2020;Chi, 2021), the rich have not been immune to its effects. The wealth of the world's top 1000 billionaires decreased by almost 30% in the early months of the pandemic. However, their fortunes have proven more bouyant, taking just nine months to return to pre-pandemic heights (Berkhout et al., 2021). While it is too early to gather conclusive evidence, international experts warn that global inequalities are expected to further increase as a result of the pandemic, including capital, income, gender, political and racial inequalities (Ferreira, 2021). Most national governments so far have no plans in place to resolve these inequalities. Instead, business continuation has been the main priority in societal recovery plans (Berkhout et al., 2021).
The pandemic has also raised awareness of the extent to which economic health, as measured by conventional metrics, depends on increasing product consumption, mobility and pollutant emissions. Simultaneously, it has shown the potential for demand-side actions to mitigate a broad range of environmental impacts and reduce global CO 2 emissions at a rate approaching the 7.6% per year now required to prevent average global warming of more than 1.5 • C, in line with the Paris Agreement (Tollefson, 2021). Total global emissions in 2020 are estimated to have fallen by 5.8% relative to the 2019 level (IEA, 2021), with a global reduction of 8.8% during the first half of the year . This correlated with a 4% reduction in energy consumption, 50% of which came from the transport sector (IEA, 2021). The average peak of daily reductions across individual countries was as high as 26% (Le Quéré et al., 2020). These reductions were larger than during previous economic downturns and World War 2 Ritchie and Roser, 2020), and were accompanied by other environmental benefits such as large reductions in other air pollutants, water pollution, noise pollution and reduced human encroachment into wildlife habitats (Kumar et al., 2020;Arora et al., 2020). However, emissions are now projected to increase by 4.9% in 2021 (Friedlingstein et al., 2021), returning almost to pre-pandemic levels, driven by economic recoveries in a few advanced economies while the developing world continues to struggle under the pandemic (World Bank, 2021). Adverse effects of the resulting overheating of rich economies, such as increased fuel prices, are further adding to the burden on the poor (McHugh et al., 2021). We discuss implications of these findings for building climate resilience in Section 4.
The abatement of environmental damage due to the pandemic was in part caused by pervasive behavioural changes. These included a large reduction in air travel (accompanied by a widespread switch to online conferencing) (Jack and Glover, 2021;Puttaiah et al., 2020), reduced overall personal mobility, increased use of active transportation modes like walking and cycling (Borkowski et al., 2021;Kim, 2021), and reduced discretionary spending (Puttaiah et al., 2020). Despite the pandemic's many negative impacts on citizens, societal responses have also brought benefits to some, such as fewer road crashes and health benefits from cleaner air and increased exercise by those using more active transportation (Kim, 2021). These widely appreciated upsides of the difficult COVID-19 experience have motivated public investments in, for example, cycling infrastructure upgrades (Frost, 2021;Puttaiah et al., 2020), and inspired calls for a persistent shift to online conferencing (Jack and Glover, 2021). Notably, the conditions that enabled these silver linings were created by strong regulations that affected everyone more or less equally. They did not allow exceptions based on, for example, a market in tradeable permits to violate lockdown rules, which may have helped to preserve the status quo and reinforce inequalities (e.g. online conference attendance may be disadvantageous if others, who can afford to, meet in person (Jack and Glover, 2021).

Complexity of systemic risks in a connected world
A complex system comprises many components whose interactions, with each other and their environment, give rise to 'emergent' system properties that cannot be deduced from the components' properties alone (Thurner et al., 2018). The climate, societies and economies are quintessential complex systems with networked interactions from local to global scales. The interconnectedness of these systems provides some measures and mechanisms for resilience to local crises: rapid spread of information, trade, regional and global security arrangements, and financial transfers can provide resilience to disruptions by enabling fast responses and fallback options. Examples include international assistance in cases of natural disaster, the ability to import food from alternative suppliers if one source cannot deliver, and international insurance markets (Altenberg, 2020). Trade (e.g. in medical equipment) helped to meet shortages during the COVID-19 crisis (Altenberg, 2020).
The same interconnectedness can, however, enable propagation of negative impacts from a crisis like COVID-19 between organisations, regions or sectors (Valdez et al., 2020;. The virus' rapid spread was facilitated by international trade and travel. The subsequent response, in the form of lockdowns and mobility restrictions, affected consumer demand, supply networks, company revenues, poverty levels and national economies, with impacts spreading through networks (Fig. 1), many of which were designed to maximise short-term profits and consumer choice rather than societal resilience (Kharrazi et al., 2020). For instance, automated trading on international markets, just-in-time long-distance supply chains and globalised mass-tourism promote overexploitation of resources (Myklebust, 2020;Gössling and Peeters, 2015) and concentrate market power in the hands of few, increasing vulnerability of the overall system (Klimek et al., 2015).
These challenges resulting from economic connectivity are aggravated by an unprecedented level of social connectivity, particularly through online social media. This has driven human collective behaviour into a regime that is significantly altered from the conditions of our evolutionary past, for which we may be maladapted. The functional consequences are emerging in real time as we struggle to understand and manage them without a developed scientific framework (Bak-Coleman et al., 2021). This emerging regime of social connectivity includes many benefits, such as rapid spread of evidence-based information, and transnational and transdisciplinary collaborations. However, these are accompanied by harmful phenomena such as ideological echo chambers and polarisation, eroded trust in government, and rapid spread of fake news and misinformation that can undermine collective intelligence (Bak-Coleman et al., 2021). In the case of the COVID-19 pandemic, this has hindered public acceptance of evidence-based safety measures such as mask wearing, widespread testing and vaccination (Zarocostas, 2020;Roozenbeek et al., 2020). In the case of climate change, it has contributed to overrepresentation of climate contrarians, public confusion, political inaction, and stalled support for or rejection of mitigation policies (Bak-Coleman et al., 2021;Lamb et al., 2020;Treen et al., 2020).
The same combination of network connectivity, emergent collective behaviour, political responses, and social vulnerability has potential to spread and amplify climate change impacts. These will likely take diverse, more diffuse forms than COVID-19, although health issues and disease spread across borders may be among them. Indeed, although the current pandemic is not clearly attributable to climate change, its likelihood was elevated by large-scale human impingement on natural habitats, biodiversity loss and ecosystem degradation (Johnson et al., 2020;Shield, 2020), and possibly also by a climate change-driven increase in bat species 1 richness in southern China (Beyer et al., 2021). Some climate-driven disruptions of global socio-economic networks may arise from a single climatic event (Lenton et al., 2019). Others may emerge from multiple scattered, smaller events and their interactions with each other and other social-ecological dynamics . For example, simultaneous crop losses due to droughts in key production regions can drive up global food prices, potentially triggering mass human displacement (UNFCCC, 2017) and unrest (Gaupp et al., 2020). 2 These in turn can influence politics in consumer and migrant-receiving regions. In such a case, pre-existing inequalities and lack of social safety nets in food importing countries would weaken the ability to cope and intensify the crisis.
Linked collections of climate change impacts should be understood as systemic risks (Aglietta and Espagne, 2016), which are characterised by the potential for cross-sectoral impacts to reverberate across geographical and political boundaries (Folke et al., 2011). They are likely to materialise as a result of cascading failures (Pescaroli and Alexander, 2015;Valdez et al., 2020) whereby impacts accumulate, impairing or triggering breakdown of entire systems (McGlade et al., 2019). Risks are greatest where there is strong interdependence between conditions that amplify risk (Willner et al., 2021), trigger secondary disasters (Jongman et al., 2014) and affect critical nodes such as infrastructure or major hubs of trade or finance (Mandel et al., 2021). A growing number of public and private financial institutions, including central banks, banks (Carlin et al., 2021), and insurance and asset managers (BlackRock, 2021), recognise that climate risks could affect their portfolios' performances and have destabilising effects on the international financial system. Accordingly, interest in and pressure to disclose, assess and respond to climate-related financial risks is growing (Smith, 2021). This could lead to positive or negative revaluation of entire asset classes due to 'physical risks' such as exposure to climatic hazards, or 'transition risks' such as decarbonization. Climate-related financial risks have been recognized by the European Commission (European Commission, 2018) and financial supervisors, including over 90 central banks and financial regulators that have joined the Network for Greening the Financial System (Elderson and Heemskerk, 2020;Basel Committee on Banking Supervision, 2021;NGFS, 2019). In terms of more general theoretical understanding, the dependence of systemic resilience on network properties including connectivity remains only partially characterised and is a subject of ongoing research Kharrazi et al., 2020).

Systemic resilience and its dependence on network properties
Different fields of study define resilience in different ways and its meaning in relation to social-ecological systems has evolved in recent years. While it was initially conceptualised as the capacity to absorb disturbances and continue to function as before, it is now understood as the ability to cope with disturbances not just by withstanding and recovering from them but also through adaptive changes and, if necessary, transformation into a qualitatively different state (Walker, 2020). In this sense, resilience is a process rather than simply a property. For example, resilient coastal towns may at first withstand and recover from a single flooding event, then adapt to more frequent flooding due to sea level rise by raising the heights of sea walls, but eventually find it necessary to relocate to higher ground Walker, 2020).
Studies of a wide range of systems have identified redundancy, diversity and modularity as key properties that contribute to systemic resilience (Kharrazi et al., 2020). Redundancy is the inclusion of replicated pathways, functions or components, which increases fault tolerance. Diversity may refer to the number of categorical types in a system (e.g. economic sectors), the distribution of 1 Bats are the likely zoonotic origin of several coronaviruses, including that which causes COVID-19 (SARS-CoV-2). However, significant evolutionary gap between SARS-CoV-2 and the closest known animal viruses suggests decades of evolutionary divergence, and a more immediate animal reservoir for SARS-CoV-2 has not been identified (Beyer et al., 2021;Holmes et al., 2021).
2 When discussing the effects of direct climate impacts such as drought, it is important to recognize the role of historical political governance and resource management dynamics, including failure of resilience and adaptation measures, in the shape and severity of ensuing human impacts and responses.
something across available categories (more even distribution means greater diversity), or the disparity between the categories themselves (larger differences means more diversity). Diversity can enable more flexible responses to system shocks. Modularity is the extent to which a network's structure is partitioned into modules (or communities) in which node connections are much stronger than connections between nodes in different modules. More modular systems are better able to contain stresses or contagions within modules, thereby protecting the rest of the system. While redundancy, diversity and modularity are known to individually contribute to systemic resilience in some cases, less is known about how tradeoffs between them, and scaling effects, may affect resilience; these are important subjects for future work (Ringsmuth et al., 2019;Kharrazi et al., 2020).

Building resilience to systemic crises
COVID-19 has revealed how challenging it is to manage systemic and compounding crises. Institutional mechanisms were not aligned, national and global disaster risk reduction networks were not prepared, and access to reliable and verifiable data were, at least initially, poorly facilitated (Mechler et al., 2020). Although civil society organisations have been able to mobilise their constituencies to monitor systemic and compound risk (Shaw et al., 2020), this has only partially addressed the gaps left by the state. Similarly, responsibilities within many current governmental structures to identify, track and build resilience to climate risks, which are still often treated as external risks (Benzie and Persson, 2019), are unclear. Effective risk management requires a clear and endorsed 'ownership' of the risk (Young and Jones, 2018;Young and Jones, 2017): an organisational unit that takes responsibility to implement risk mitigation measures. For complex and systemic risks with multiple drivers and stakeholderssuch as COVID-19 or climate change this is challenging. It requires the development of resilience measures that include incentives to enable networks of government, market actors and civil society to co-generate timely, flexible, and actionable solutions (e.g. cost/benefit standards that factor in future generations' costs and benefits). In practical economic terms, resilience can be framed as a "portfolio of dividends": a collection of risk management options and investments which work across different sectors, regions, generations, etc. to minimise overall risk (Mechler and Hochrainer-Stigler, 2019). In this portfolio, we can increase the overall resilience of the global system by spreading investments and targeting the most vulnerable subsystems. These procurements should reduce nations' risks related to the economic, ecological and/or political fragility in countries that supply essentials such as energy, raw materials, labour or food.
The pandemic has shown the importance of impact monitoring and openly sharing data across boundaries for building resilience. This enables faster detection, earlier notification to others, and more informed management of cascading impacts. It also allows experts examples of climate change impacts that illustrate parallels to, and interactions with, the pandemic. Impacts spread and interact within and between networks of citizens, production systems and financial systems, which are also subject to multiple influences and impact responses.
around the world to research challenges that arise, analyse associated risks, and share their findings for collective benefit. For example, surveillance systems detected clusters of pneumonia cases as early as December 2019 and the novel coronavirus genome sequence was released by mid January 2020, enabling a global research effort into tests, treatments and vaccines (European Commission, 2021). Researchers used online preprint servers to share research results more quickly than is usually possible through academic journals (Fraser et al., 2021;Taraborelli, 2020) (although the absence of peer review means that sharing results in this way requires safeguards against overinterpretation of preliminary findings, including careful communication and management of uncertainties). Despite these efforts, the International Panel for Pandemic Preparedness and Response concluded that the international alert system did not work fast enough (European Commission, 2021). The European Commission has since recommended that a new global surveillance system be put into place, with capabilities to share data at the earliest possible stage (European Commission, 2021). Investing in a similarly global system for monitoring climate impacts and openly sharing data across boundaries would enable better risk management and resilience across system levels. The practice of frequent 'event attribution studies' (Otto et al., 2018;van Oldenborgh et al., 2021), identifying the contribution of anthropogenic climate change to the intensity or probability of adverse weather events, is a potential contribution to such a climate impact monitoring system.
However, as also demonstrated by the pandemic, plans to respond to threats detected through monitoring do not amount to preparedness without dedicated resources. Many governments were forced to respond to COVID-19 with ad hoc, temporary measures because pandemic response preparedness was under-resourced (European Commission, 2021). This speaks to the importance of putting adequate resources in place for climate impact response preparedness. There have been a variety of initiatives, including from governments, civil society and multilateral institutions, to allocate resources in this direction, and to sustainability more broadly, by using the pandemic as an opportunity to 'reset' the global economy and spur 'green recovery'. These initiatives are based on the understanding that investments in resilience and growth areas such as transport infrastructure, circular economy, digital transformation (European Commission, 2020), and renewable energy will create jobs and prepare for future competitiveness by investing in and up-skilling the global population. Meaningful government and corporate action on this is currently impeded by the lack of sustainability impact criteria applied in the allocation of recovery funds. Only 10.6% of this public money is estimated to be earmarked for projects that help to reduce emissions or restore the natural environment (Vivid Economics & F4B, 2021). Much more is tabled for projects which would appear to stimulate consumption and emissions. For example, the EU funds have been criticised for at least partly preserving business as usual in some countries, rather than being deployed for transformational change to enable low-carbon economies (Climate Action Network Europe, 2021). To be sustainable, such investment would also have to be accompanied by far-reaching reforms in areas such as taxation, subsidy and environmental regulation (OECD, 2020).
Adoption of such policies at the scale and rate required for adequate climate action has been deliberately delayed (Lamb et al., 2020) and also hindered by misinformation about climate change, particularly through nontraditional, digital media (Treen et al., 2020). Recent research has found that lower susceptibility to misinformation about both COVID-19 and climate change is associated with higher risk perception, higher capacity to think deliberately and critically, a tendency to update prior beliefs based on new evidence, and higher trust in scientists and mass media (Roozenbeek et al., 2020;Gruener, 2021). There is also evidence that susceptibility to misinformation about one subject is correlated with susceptibility to misinformation about another subject (Gruener, 2021), so measures taken to counteract misinformation on one subject may also help to counteract it elsewhere. Such measures include educational approaches, 'inoculation', technological solutions, corrective and collaborative approaches, and regulation, although all are subject to criticisms and caveats (Treen et al., 2020).
The impacts of the pandemic on different countries and social groups have shown that a well-functioning social security and public health system can act as insurance against various types of future risks. Reducing economic and other social inequalities, and investing in systems that care for the most vulnerable in society, improves resilience to a wide spectrum of possible risks. For example, higher numbers of beds and staff in hospitals than are needed on average (or increased flexibility to quickly scale this capacity up and down) can greatly reduce the effects of unforeseeable crises and disasters. The need to support the costs of such public services further emphasises that changes in taxation and subsidy flows are likely needed. A higher contribution from the wealthiest and measures against tax havens and tax avoidance may be crucial. At the same time, it is now widely recognized that underlying health conditions including chronic lung diseases, heart conditions, cancer and obesity increase the chances of severe or fatal effects of the virus (USCDC, 2021). Given that these also remain the top causes of death globally, it would seem that a much more integrated approach to public health, with greater attention to prevention and well-being, would both increase societal resilience and reduce long-term costs.
Many already existing internationally agreed policy processes stress the need for enhancing resilience, including the Sustainable Development Goals (SDGs) (United Nations, 2021), the Paris Agreement (United Nations, 2015), and the Sendai Framework for Disaster Risk Reduction 2015-2030(United Nations, 2015. However, progress in interpretation, uptake and implementation of their various recommendations has thus far been slow and ineffective. It has been argued that if the SDGs, for example, had been fully embraced and implemented, the world would have been in a much better position to respond to the COVID-19 pandemic, which may have been less likely to occur in the first place (Ottersen and Engebretsen, 2020). Counterfactual speculation aside, crucial resiliencebuilding may come from diagnosing and overcoming reasons behind the limited effectiveness of existing policy processes rather than developing entirely new ones.
On the other hand, a number of researchers have recently argued that the limited effectiveness of climate change mitigation efforts to date is due mainly to more fundamental sociometabolic dynamics 3 intrinsic to the global economic system, and therefore largely immune to policy that does not address these dynamics (Hagens, 2020;King, 2020;Leiva and Schramski, 2020;Garrett et al., 2020). Since the creation of the United Nations Framework Convention on Climate Change in 1992, annual global CO 2 emissions have risen every year during periods of GDP growth, except in 2015, when emissions fell by 0.08% (Ritchie and Roser, 2020), a relative reduction 73 times smaller than was achieved by economic restraint in 2020. Although the fraction of global primary energy supplied by renewable sources grew rapidly before the pandemic, fossil fuel consumption also grew over the same period (BP, 2020) such that global energy growth outpaced decarbonisation and emissions continued to climb (Jackson et al., 2018). That is, renewables have so far not displaced fossil fuels but rather only added to them (BP, 2020), to support the energetic demands of global economic growth.
Given that large and compounding emission reductions are now needed immediately to remain within an agreed safe carbon budget (IPCC, 2021), prospects for decoupling GDP from emissions at the scale and rate required are negligible (Haberl et al., 2020;Hagens, 2020;Hickel and Kallis, 2020;Ward et al., 2016). Serious commitment to climate risk reduction should therefore make us question whether returning to sustained GDP growth at the global level is a sound goal for pandemic recovery. Moreover, the possibility of an approaching decline in net energy yields from primary energy sources, whether fossil-fuelled or renewable, suggests that the world may soon face protracted energetic constraints to economic activity (Brockway et al., 2019;Capellán-Pérez et al., 2019;King and van den Bergh, 2018). This is another reason to consider alternatives to GDP growth as the prime objective and performance measure for economies.
A growing literature is now exploring options for so-called degrowth, which aims to promote human wellbeing and ecological resilience without requiring GDP growth (Hickel et al., 2021;Kallis, 2017;Keyßer and Lenzen, 2021;Wiedmann et al., 2020;. The experiences of the last few years, including COVID-19, may have shifted public understanding and opinion on these issues, especially in richer countries, which are best placed to lead with new policies. In a recent survey of people in G20 countries, 74% of respondents supported the idea that their country's economic priorities should move beyond profit and increasing wealth to focus more on human wellbeing and ecological protection (Gaffney et al., 2021). Overcoming the research-implementation gap remains a challenge, however. We do not yet know how degrowth policies could be adopted at scale or how they would interact with our current economic processes and patterns of behaviour. Nonetheless, our climate predicament demands openness to new approaches. To better understand the options, further research into basic relations between societal metabolism and social-ecological wellbeing is crucial (Haberl et al., 2019;Hagens, 2020;King, 2020;Leiva and Schramski, 2020;Schramski et al., 2015;West, 2018;Fischer-Kowalski and Haberl, 2015;Garrett et al., 2020;Giampietro et al., 2013;Ringsmuth et al., 2016). Although the many challenges and risks associated with a fundamental economic transition are formidable, arguably, they may be overcome on the time scale of a human life (decades). 4 Conversely, the risks associated with ever more radiative forcing as we avoid adequately reducing emissions include transitions of the Earth system to inhospitable states that may not be reversible on the time scales of civilisations (centuries to millennia) (Lenton et al., 2019;Steffen et al., 2018). The growing literature on tipping dynamics in socioeconomic systems may offer reasons for optimism about societies' potentials for a timely rise to the challenge of systemic transformation (Otto et al., 2020;Lenton, 2020).

Research methods for assessing and managing complex systemic risks
In order to build overall systemic resilience and preparedness for cascading impacts of climate change, we need to understand how they may affect systems and how systems can adapt. Existing conceptual frameworks lay the groundwork for such assessments by, for example, making it possible to distinguish between the initial impact of a climate trigger and its downstream consequences as the impact propagates through an impact transmission system (Carter et al., 2021). Simulating and visualising the risks and impacts of interventions can support policy actions to limit systemic risks and understand the dependencies on internal and external interactions.
A major challenge in managing the rapidly shifting climate risk landscape is the development of mental and formal models that adequately capture systemic complexity. Integrated Assessment Models (IAMs), either cost-benefit optimization models such as DICEstyle IAMs (Nordhaus, 1992) or process-based simulation models (Kriegler et al., 2013;McCollum et al., 2018;Rogelj et al., 2019) have been used to inform climate policy and the Intergovernmental Panel on Climate Change (IPCC). They vary in terms of structure, assumptions and outcomes (Krey et al., 2019), such as the optimal cost of carbon, and climate mitigation trajectories compatible with 2 • C-aligned carbon budgets (Weyant, 2017). Despite their ubiquity, IAMs face several limitations for the analysis of complex and compound shocks that embed nonlinearity. IAM economic modules can range from partial to computable general equilibrium models. They embed representative agents with perfect foresight and forward-looking expectations (i.e. it is assumed that they know perfectly what will happen in the future), and solve to equilibrium (Balint et al., 2017;Mercure et al., 2016). Such models cannot capture nonlinear dynamics that emerge from interactions between heterogeneous agents who are subject to imperfect information and have adaptive expectations about the future, which amplifies the complexity of decision making and uncertainty over outcomes (Monasterolo, 2020;. IAMs have been criticised for neglecting agents' heterogeneity, which is important for modelling the complexity of societal transitions, and for over-or under-representing technological change (Keppo et al., 2021). Furthermore, IAMs do not embed finance and its complexity; there are no financial actors that decide whether to allocate capital based on risk assessment, which translates into financing costs for high or low-carbon investments . Given the importance of these dynamics in the real world, it may be useful to complement or integrate IAMs with models such as dynamical systems/system dynamics models (Lade et al., 2022;Sterman, 2000), agent-based models (Andersson et al., 2021;Haer et al., 2020;Steinbacher et al., 2021), adaptive network models (Wunderling et al., 2021;Thurner et al., 2018), or Bayesian networks (Kaikkonen et al., 2021), that can more fully accommodate the complexity of nonequilibrium, nonlinear, networked systems like societies, economies, ecosystems and Earth systems (Balint et al., 2017;Farmer et al., 2015). Moreover, incorporating alternative economic performance metrics into these models and considering degrowth scenarios can assist with understanding potential future pathways that have not usually been taken into account by integrated assessment modellers (Keyßer and Lenzen, 2021).
Policymakers and managers tend to use mental models that may have been valid when the risk landscape was more stable, leading to expectations that the world will change only incrementally and 'go back, more or less, to normal' soon after a shock such as the COVID-19 pandemic or an extreme weather event. Such models now diverge ever further from reality. Even if carbon emissions are reduced, atmospheric concentrations will continue to rise, and extreme climate events and sea level rise will stay with us for decades to come (IPCC, 2021)(). There is no longer a stable 'normal' climate on which to base planning decisions. Managers would therefore do well to adopt mental models in which climate risks cannot always be quantified, and extreme and cascading impacts are ever more likely (Shepherd et al., 2018). Such a change in mental models can be supported by interactive tools such as crisis simulators (Smith, 2004), serious games (Mochizuki et al., 2021;Solinska-Nowak et al., 2018), event storylines (Sillmann et al., 2021) or policy simulations (Duke and Geurts, 2004), in which managers are confronted with severe climate triggers leading to inter-connected, acute impacts and must respond to them. Outcomes from these processes, in the form of storylines and policy responses, can provide input for formal complex system models, which in turn produce results that help to build, in a process of iterative development, real-world adaptation pathways.
While crisis simulation can be an essential tool in building preparedness to shocks and resilience for the new normal, it is equally important to include a long-term perspective in the mental models of managers and policy makers. Policy simulation -also known as policy exercise -is an interactive, participatory method to develop strategic insight. This approach allows stakeholders to explore real policy issues, using design elements known from serious games to structure communication (Duke, 2011) as well as to include feedback that participants receive based on their decisions. Participants work with real world data and take roles that often represent their roles in the real-life system (Harvey et al., 2009). Simulation experience helps to understand how problems emerge in complex systems and where points of policy intervention may lie. They can explore possible strategies, or "pathways" for their specific internal group of actors embedded in a range of external scenarios (Zurek and Henrichs, 2007;OECD, 2006). Such scenarios can be drawn, for example, from Shared Socioeconomic Pathways (Riahi et al., 2017), event storylines rooted in historic events (Sillmann et al., 2021), or other global scenarios. Within these scenarios, stakeholders use their knowledge and available data in a deliberative manner to develop strategies that can lead them to responses that are consistent with desirable future developments, identifying challenges, seeking solutions, and negotiating trade-offs.
Policy simulations have been successfully used in multiple areas including systemic liquidity crises in banking (Gai et al., 2010), climate policy as a business opportunity for venture capital in Europe (Kasemir et al., 2000) and river floodplain management (Stefanska et al., 2011). The Cascading Climate Impacts (CCI) policy simulation (Jarzabek et al., 2020) was developed in 2020 to account for cascading effects and systemic interconnection 5 . It brings together stakeholders from different thematic areas: Trade, Supply & Value Chains; Security, Development, and Foreign Policy; and Business and Finance. The simulation is based on a storyline -a series of climate triggered events cascading across continents with final impacts on Europe. Although the objective is not to predict specific events, quite often the future that unfolds is quite close to proposed storylines -for example a Suez Canal blockade was part of the simulated series of events in April 2020, shortly before one happened in the real world.

Recommendations for improving transboundary climate risk management
Based on the lessons learned, we draw three recommendations from COVID-19 that are relevant to climate risk management. First, acknowledge the systemic nature of climate change impacts. Most climate risk assessments and adaptation strategies focus on nations and sectors, addressing clearly identified risks, actors and options to reduce risk. However, the systemic, cross-sectoral and transboundary nature of many climate change drivers and impacts requires broad, systemic transformation and better preparation for the unavoidable impacts of climate change. Complex interactions and impact cascades will shape the overall risk profile. Actions to approach climate change as a systemic rather than a localised risk include: 1) Understanding and mapping the direct, cross-border and cross-sectoral impacts of potential climate extremes (including compound events) to guide effective risk mitigation strategies. Cross-sectoral model intercomparison projects such as the Intersectoral Impact Model Intercomparison Project (ISIMIP) (Frieler et al., 2017) and recent conceptual advances defining compound climatic events  and cross-border impacts (Carter et al., 2021) provide the foundations for modelling and developing scenarios of plausible impact chains affecting key regions (e.g. reduced water availability, extreme temperatures and population increase) with local stakeholders to analyse regional risks (such as conflict dynamics and migration), and trans-regional impacts; 2) Redefining the goals of climate adaptation plans, including a wider definition of resilience and the targeted scope of these actions.
For example, at the international level, these could include developing or sharpening legal or fiscal interventions that increase the resilience of global trade, finance, development cooperation and food security networks (such as EU regulations on corporate transparency, dedicated stress-tests and policy that directs sustainable international collaboration). At both the regional and national levels, they may look to increase the modularity of energy and food systems; 3) Reorganising risk management procedures and institutions to establish responsibility for systemic risks. For example, exploring mandates for risk mitigation strategies at international and cross-sectoral levels, by designing appropriate risk contingency programs involving public and private institutions (Benzie and Persson, 2019).
Second, accept limitations to predictability of impacts, develop models that account adequately for systemic complexity, and adopt adaptive risk management strategies that embrace uncertainty. In a highly interconnected world, quantitatively analysing how systemic hazards propagate to societal impacts is a challenge. While probabilistic assessments of local and near-term climate-related impacts are generally feasible, the ability to map indirect and cascading effects of shocks proves to be limited. Nonetheless, there is potential for going beyond current models with methods that can more fully account for systemic complexity and incorporate alternative economic performance metrics. Furthermore, we need to develop adaptation strategies that accommodate heterogeneous decision makers working under uncertain conditions, to be able to adjust to new insights, developments and risk management options over time. For instance, the concept of 'dynamic adaptation policy pathways' (Haasnoot et al., 2013) is designed to incorporate deep uncertainty into decision making, by identifying short-term actions and a framework to guide future actions towards the desired goal, whilst allowing for adaptation over time to meet changing circumstances. Steps to adopt adaptive risk management strategies include: 1) Visualising dynamic and responsive pathways and storylines to anticipate unknown events and prioritise options and measures. For instance, digesting lessons learned from COVID-19 and other systemic crises into a representative set of cause-effect networks, or exploring potential evolution of historic shocks under possible future (climate) conditions (Sillmann et al., 2021); 2) Exploring adaptation strategies to unprecedented but plausible risk pathways for 'critical entities' (critical services, supplies and infrastructure (Haasnoot et al., 2020;European Commission, 2020). For example, co-developing adaptive risk strategies for selected services and institutions (such as urban water management or power supply).
Third, build resilience at multiple, interconnected levels. Resilience against system shocks is desired but difficult to achieve. A redefinition of strategic economic objectives can be helpful: from an efficiency-focussed paradigm towards a forward looking, multiobjective approach that takes into account social and intergenerational equity. Incentives for building resilience can be developed by: 1) Redefining performance metrics to have a long-term focus and target human wellbeing, including societal and ecological resilience.
For example, understanding the multidisciplinary SDGs as performance indicators applicable to economies at different scales, including citizens, private business organisations, sectors and multi-level governments; 2) Redesigning international and national solidarity mechanisms that help to recover from shocks of various kinds. For instance, operationalising the Global Goal on Adaptation (United Nations, 2015) and ensuring international support and cooperation under the Paris Agreement to achieve just and systemic resilience to current and future climate change, including by building rapid crisis response capacity and making sufficient financial and political investments in societal resilience at all scales. 3) Establishing the study of human collective behaviour in the digital era as a 'crisis discipline' 6 tasked with developing methods for ethical stewardship (Bak-Coleman et al., 2021), particularly in the face of crises, even though a complete understanding of system dynamics is not available. Leveraging the benefits of global digital communication tools, while also effectively navigating and mitigating their pathologies, will be crucial to future risk and crisis management. 4) Jointly utilising COVID-19 recovery packages to meet the interconnected goals of fiscal and financial stability, societal resilience and environmental quality. For example, improving public health care systems and expanding investment in integrated public health with greater emphasis on disease prevention, social protection and labour laws, strengthening corporate tax policy and making fiscal measures that support carbon-intensive businesses conditional on alignment with international climate targets (Battiston et al., 2020).

Conclusions
This review has aimed to distil lessons related to the transboundary management of systemic risks from the COVID-19 experience, to inform climate change policy and resilience building. We have identified challenges that arise from highly interconnected and adaptive global systems. These include risks of transboundary cascades of both climate impacts and maladaptive responses to them Carter et al., 2021), as well as reduced response effectiveness due to pathologies in human collective behaviour, such as widespread digital misinformation and discourses of action delay (Bak-Coleman et al., 2021;Lamb et al., 2020;Treen et al., 2020). The massive stimulus spending so far announced (Vivid Economics & F4B, 2021) has the potential to contribute to a 'new normal' of increased climate resilience, but until now has strongly favoured the recovery of consumption and, accordingly, emissions (Vivid Economics & F4B, 2021). The large drop in emissions in the early months of the pandemic  demonstrated the potential for demand-side restraint to reduce emissions at the scale and rate now required for effective climate action (Tollefson, 2021). The rapid resumption of emissions along with rebounding economic growth following the lifting of lockdown restrictions calls 6 'Crisis disciplines' have been described as distinct from other fields of urgent, evidence-based research in their need to consider the degradation of an entire complex system without a complete description of the system's dynamics (Bak-Coleman et al., 2021). Examples include medicine, conservation biology and climate science. Such a discipline for the study of collective human behaviour would need to take a transdisciplinary approach, drawing on many established fields from within and also beyond the social sciences (Bak-Coleman et al., 2021). into question the soundness of a return to global GDP growth as a goal for economic recovery (Hickel et al., 2021;Keyßer and Lenzen, 2021).
Social-ecological resilience building is essential to climate impact preparedness, and it is important to think of resilience as an adaptive and/or transformative process, rather than simply a capacity to persist Walker, 2020). COVID-19 has shown the value of impact monitoring and open data sharing for building resilience, and there have since been calls to develop an integrated global monitoring system to guard more effectively against future pandemics (European Commission, 2021). A similarly global system for monitoring climate impacts and openly sharing data across boundaries would enable better climate risk management and resilience across levels. However, dedicated resources for responding to impacts detected through monitoring must also be put in place (European Commission, 2021). Developing strategies for ethical stewardship of human collective behaviour would further help to build climate impact resilience (Bak-Coleman et al., 2021), as would effectively diagnosing and overcoming reasons for the limited effectiveness of existing policy processes.
Effective risk management through clear and endorsed risk 'ownership' (Young and Jones, 2018;Young and Jones, 2017) is difficult for complex and systemic risks with multiple drivers and stakeholders, so resilience measures need to include incentives that enable diverse stakeholders to co-generate timely, flexible, and actionable solutions. Reducing economic and other social inequalities, and investing in systems that care for the most vulnerable groups, would also improve resilience to a wide spectrum of possible risks. Based on the pandemic experience, we recommend that decision makers must take into account the systemic nature of climate change impacts, adopt adaptive risk management strategies that embrace uncertainty, and implement a range of strategies for building resilience at multiple, interconnected system levels. The suite of transdisciplinary tools and methods reviewed here can support these endeavours.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.