ESG Economist - How big is the impact of extreme weather on GDP?

Introduction

In November 2023, the fourth vintage Network for Greening Financial Services (NGFS) climate scenarios were published. We earlier wrote a general note on the fourth vintage scenarios here. We now go into more details. One of the main steps forward in the fourth vintage was the more extensive and more granular incorporation of acute physical risk. In this note, we look further into these acute physical risks.

Risk, transition, chronic physical and acute physical

Risks associated with climate scenarios come in three flavours. Transition risk is the risk associated with the transition towards a greener economy. The size of transition risk depends partly on the speed and extent of the transition taking place, relative to the capacity and readiness of the private sector to adjust. If the transition is orderly, gradual and expected, transition risk is lower than when the transition occurs in a disorderly, sudden, unexpected or delayed fashion. Chronic physical risk is the risk associated with the average temperature change, for instance through trend changes to labour productivity and agricultural productivity as the earth warms. Acute physical risk is the risk associated with the increased frequency and severity of extreme weather, such as heatwaves, drought or floods.

The NGFS climate scenarios illustrate that an immediate coordinated transition will be less costly than inaction or a disorderly transition in the longer term. This is due to the prominence of physical risk. Chronic physical risk becomes gradually more prominent over time, especially in the hot house world scenario Current Policies (this scenarios assumes continuation of current policies without any further action and leads to global warming of 2.9 degrees Celsius at the end of the century). But the most sizeable source of risk across all scenarios and time horizons is acute physical risk, and this source of risk increases further in relative terms towards the longer term: since physical risk is unaffected by mitigation efforts in the short run (given that the impact works with a lag), acute physical risk is similar across scenarios until 2040, with a strong surge in damages in Current Policies thereafter.

Hazard impacts

Acute physical risk modelling has been enriched to include more hazards and increasing geographical granularity. In the previous version, there was some coverage on a global scale of acute physical risk. The fourth vintage expands the coverage of acute physical climate risks and specific hazards, such as heatwaves, droughts, floods, wildfire, and storms, and their impact on the macro economy. This is done through more granular indicators, the use of global climate models, and impacts at specific locations. These impact are then put into the economic model NIGEM (the macroeconomic model used to transmit the transition and physical risks into the fundamental macroeconomic variables). GDP impacts are estimated in NIGEM using specific transmissions mechanisms.

Heatwaves can affect economic activity in several ways ranging from the impact on productivity, the disruption of supply chains due to impacts on infrastructure, and the impact of water scarcity. The hazard is measured by humid heat (“wet bulb temperature”). Humid heat is particularly harmful for human health as it affects the body’s cooling capabilities which are based on evaporation of sweat. When exposed to such conditions severe health risks could be the consequence, which can result in heatstroke. To identify heat stress for each location, a relative threshold for that location is set, combined with an absolute threshold which goes for all regions. Climate models are used to determine the frequency of these occurrences under different emissions scenarios.

The population exposed to the hazard is then determined, at a certain location (grid point). In some regions, humid heat is projected to dramatically increase. But not all the regions are densely populated. The impact is larger in more densely populated areas. The aggregation of these grid points then leads to an impact in terms of productivity and demand shocks. These shocks are then added as inputs to determine the general economic impact in the NIGEM model. The results suggest that damages from heatwaves reach about 3.1% by 2050 in a Current Policies scenario, which is 1.9 percentage points more damage than the orderly transition scenario. The loss estimated varies across countries, with highest impacts in Europe and Asia.

Drought conditions are detrimental to ecosystems and a broad range of sectors, including agriculture, energy, and other water-intensive sectors. The hazard is detected via the Standardized Precipitation-Evapotranspiration Index (SPEI), which is a drought indicator based on relevant variables such as precipitation and evaporation and considers long-lasting drought conditions (as they are particularly detrimental to food security). The existence of this hazard in a certain location is then combined with the agricultural intensity of that location and aggregation of the locations (grids) then determines national crop yields impacts. These impacts then affect the general economy via shocks to productivity, exports and prices in NIGEM. Resulting estimates show that projected GDP losses at world level in a Current Policies scenario exceed 4.2% by 2050, more than 2.4 percentage points above an orderly transition scenario. Higher drought risk is shown in countries around tropical and subtropical climate zones, in particular in South America and Asia.

Floods are estimated via capital stock damages affecting the economy. These capital stock damages are estimated based on output from global hydrological models, combined with flood protection data, gridded capital stock estimates, and depth-damage functions. These capital stock damages are transmitted through the NIGEM model via investment premia shocks, which can be used to introduce a capital shock in the model. Projected GDP losses from riverine floods at world level in a Current Policies scenario exceed 0.6% by 2050, about 0.15% above an orderly transition scenario. The losses distribution shows higher losses in tropical and subtropical areas, in particular in Asian and African regions

Tropical cyclones are modelled to have a direct impact on the capital stock due to asset damages. A climate model is used to simulate a probabilistic set of tropical cyclones under different scenarios. The hazard data is then combined with exposure and vulnerability to estimate capital stock damages, affecting the economy via investment premia shocks in NIGEM. Resulting estimates show that projected losses from tropical cyclones at world level in a Current Policies scenario amount to only about 0.2% of GDP loss by 2050. This probably reflects the fact that tropical cyclones are limited to a latitude band and primarily affect coastal regions.

Conclusions

The new estimates of acute physical risk are significantly more sizeable in the fourth NGFS scenarios. Acute physical risk associated with the four modelled hazards is estimated to result in GDP losses of 8% by 2050 in the Current Policies scenarios. For comparison, in Phase III the overall acute GDP losses were estimated to be about 1.4% relative to the baseline for the Current Policies scenario. Droughts and heatwaves are estimated to represent the largest source of risk across regions. Countries in Europe and Asia are assessed to be most exposed to heatwaves. Countries in Africa, and North America are primarily exposed to drought. The lesser impact of floods and cyclones might be driven by the localised nature of these phenomena and by the different modelling approach.

Despite the fact that damage estimates for physical risk are substantially higher in this fourth vintage of the NGFS scenarios, it is clear that this is still an underestimation of physical risk, both of the chronic and the acute kind. Not all hazards and all effects of hazards are taken into account. For instance, the impact of heat stress goes beyond the effect on human health and labour productivity. Second round effects (for instance mass migration), potential tipping points (such as in permafrost thaw) and the fact that the climate change currently appears to be happening more quickly than anticipated have not been taken into account. Also, potential compound risks, where shock events overlap and potentially interact or aggravate each other, are not modelled. Crucially, in an environment of heightened risk of extreme weather events, economic agents might become more risk averse, while credit conditions would tighten, weighing more heavily on economic activity than direct impacts.