不确定气候下的脱碳投资策略

1,2, Florent McIsaac2, and Stéphane Hallegatte22World Bank Group, Washington,The Paris Agreement established that global warming should be limited to “well below” 2°C andAchieving international climate goals will require transitioning the global 1 of 19 For example, adjustment costs (Lucas, 1967; Mussa, 1977) raise the price tag of a swift transition to a greenhousegas emissions‐free economy. This is because an induced transition from dirty to clean capital creates an op-portunity cost to direct scarce resources, such as skilled labor or productive capacity, from polluting production togreenhouse gas‐abating economic activity and tends to slow the diffusion of new technologies (Morriset al., 2019). As an example, if one was to transition a region's electrical grid from relying on natural gas toprimarily relying on renewables with battery storage, completing this transition in 1 year would be far moreexpensive than if it was done over the course of a decade. In this example, the “adjustment cost” might be the costof training additional electricians to install the required solar panels, or the cost of buying more expensive bat-teries to accommodate the speed of the transition. Additional adjustment costs could arise from supply chain andtrade constraints that place limits on the rate certain technologies can be bought, imported, and deployed. Modelsincluding adjustment costs tend to favor more near‐term investments in clean technologies (Campiglioet al., 2022; Ha‐Duong et al., 1997) conflicting with a number of models in the literature that do not consider theseeffects (e.g., Nordhaus (2017)) and recommend an initially low, rising investment pathway.Another complicating factor is that there are a number of options one can choose to abate fossil fuel emissions;how does one choose between abating emissions in the energy sector, say, versus heavy industry? And howshould these efforts be allocated over time? Conventional marginal abatement cost‐based approaches wouldsuggest that one should start with cheap mitigation options and progressively move toward more expensivetechnologies; however, this policy advice can be challenged if one includes the impact of adjustment costs, whichhas been shown to lead to more investment happening earlier in expensive‐to‐abate sectors (Vogt‐Schilb &Hallegatte, 2014). Throughout, we will refer to sectors with both high marginal abatement costs and highemissions intensities as “hard‐to‐abate” (e.g., heavy industry and agriculture).The challenge to financing decarbonization posed by each of these economic factors is compounded by thepresence of uncertainty in the physical climate system. While the targets in the Paris Agreement are deterministic,in the sense that the targets themselves are not uncertain, the geophysical timing of their actualization is unclearowing to uncertain climate feedbacks (Sherwood et al., 2020). For example, the true value of the remaining carbonbudget (or simply “the carbon budget”), a geophysical quantity that corresponds to the amount of emissions onehas left to emit before a given long‐term global temperature target is nearly certain to be reached, is ambiguousowing to uncertainty in the zero‐emissions commitment, future aerosol emissions, and the transient climateresponse to emissions (Jenkins et al., 2022; Matthews et al., 2009, 2018, 2021). The presence of climate un-certainty has been generally shown to increase the stringency of climate policy (see Bauer, Proistosescu, andWagner (2024), Cai (2021), Lemoine (2021) and Lemoine and Rudik (2017), and for a few examples) and implymore climate damages (Calel et al., 2020); this is especially the case if so‐called “climate tipping points” areconsidered (Cai & Lontzek, 2019; Dietz et al., 2021; Lemoine & Traeger, 2016; Lenton et al., 2008).In this paper, we amend an economic model of abatement investment that includes convex adjustment costs andheterogenous sectors (Vogt‐Schilb et al., 2018) with a representation of climate uncertainty to jointly explore theinfluence of adjustment costs and climate uncertainty on optimal decarbonization investment strategies. Inparticular, we task the social planner with decarbonizing the economy prior to breaching some temperature targetfor the least cost when the true value of the carbon budget is hidden until some point in time; once the carbonbudget is known, the planner's policy can be adjusted (either to be more stringent or lax) in accordance with thetrue value. The result is that the social planner has to formulate a policy which is robust under a number ofpotential future risk states that will be revealed later on (similar to the approach of, e.g., Ackerman et al. (2013),Crost and Traeger (2014), Morris et al. (2018), and Okullo (2020)). Our approach allows us to vary the time whenthe carbon budget is learned, and analyze how learning the true value of the carbon budget in 2030, for example,impacts the resulting policy in comparison to learning about the carbon bu