Supplementary data to : "Macroeconomic amplification of climate change damages: modeling the role of socioeconomic drivers"

Appendix A - Climate impact module The productivity loss is calibrated in order to equalize direct impacts and the value returned by the damage function in each region. The labor productivity (measured as units of output per working hour), production capacity, and the efficiency of input conversion into output from region r at time t are all multiplied by a productivity decline factor 1 − 𝜔𝑟,𝑡, with 0 < 𝜔𝑟,𝑡 < 1 if productivity decreases. A decline in productivity increases unit production costs, thereby reducing output. The direct GDP loss is thus the difference between two GDPs computed from two static equilibriums sharing the same parameters, except that they have different productivity decline factors: 1 − 𝜔𝑟,𝑡 for the economy affected by climate impacts at time t, and 1 for the one which is not affected at time t. However, the relative drop in GDP is not equal to the productivity decline factor 1 − 𝜔𝑟,𝑡. This is partly because lower input conversion efficiency compounds the effect of higher input costs, and most sectors are either their own customers or the customers of their customers. As a result, the higher the share of intermediate inputs in production costs, the greater the GDP loss relative to 𝜔𝑟,𝑡. The gap between 𝜔𝑟,𝑡 and GDP loss therefore depends on each region’s economic structure and evolves over time. To capture this, regional productivity losses are re-calibrated at each time step using a Newtonlike iterative scheme, adjusting productivity until the resulting GDP change matches the value predicted by the damage function. This approach accounts for sectoral interdependencies and ensures a consistent mapping from productivity shocks to economic impacts. B - Fossil Fuels The oil supply module is comprehensively described in (Waisman et al 2012). OPEC producers can reduce or increase oil extraction to stabilize world oil prices around their target. In SSP1 and SSP2, they target a high price of $100/bbl, in SSP3 they target a lower price of $40/bbl. The oil extraction dynamics are endogenous and follow hubbert curves, with a progressive increase of production followed by a decline. We contrast three cases about future oil resources. Low oil resource (SSP1): Production starts to decline when 50% of reserves remain, the deployment of production capacities is slow. Medium oil resources (SSP2): Production starts to decline when 25% of reserves remain, the deployment of production capacities is slow. High oil resources (SSP3): Production starts to decline when 25% of reserves remain, the deployment of production capacities is faster. C - Energy Efficiency The energy efficiency module is described in (Bibas et al 2015). The maximum annual rate of price-driven energy efficiency improvement is 50% higher in SSP1 than in SSP2, and the convergence is complete in SSP1, meaning that countries converge towards the energy efficiency level of the most efficient country in each sector (the “leader”), while in SSP2 they only converge towards 90% of this level. The convergence speed towards this targeted level is also much faster in SSP1 than in SSP2. More precisely, the price index of energy defines a characteristic timescale of convergence (the higher the price index, the faster the technology spillover, and thus the faster the convergence and the lower this characteristic timescale). After this timescale, the energy efficiency gap between leader and followers is divided by 2 in SSP2, and by 100 in SSP1, ensuring complete convergence. Furthermore, the energy efficiency of the leader improves 4 times faster in SSP1 than in SSP2. In SSP3, there is no energy efficiency improvement. DPopulation E - labor productivity F - Supplementary figures Supplementary figure F1 | Sensitivity of the total and direct GDP losses in 2100 to scenario factors. Variations in total and direct losses are expressed as the difference, in percentage points, between the total and direct impacts (expressed in GDP percentages) across the SSP. Supplementary figure F2 | Net capital inflows in each region, expressed as a percentage of regional investment, for each SSP. Supplementary figure F3 | Large capital availability case: investments are multiplied by 10 at each timestep. Global economic impacts, expressed in GDP loss, as a function of global mean temperature increase. Bibas R, Méjean A and Hamdi-Cherif M 2015 Energy efficiency policies and the timing of action: An assessment of climate mitigation costs Technological Forecasting and Social Change 90 137– 52 Waisman H, Rozenberg J, Sassi O and Hourcade J-C 2012 Peak Oil profiles through the lens of a general equilibrium assessment Energy Policy 48 744–53