Should we be climate adaptation optimists?
by: Marshall B, with helpful feedback & additions from Andrew W and Chris C
Yes, argues a recent working paper by Matt Burgess, Patrick Brown, Matt Kahn, and Roger Pielke Jr. The paper nicely crystallizes various arguments that members of the author team have been making for years, and is a great read for anyone interested in climate adaptation or in how we can reduce climate damages. The paper also touches on a bunch of work we’re doing in the lab, some of which is out in working paper form (see here and here) and some of which is underway, so it seemed worthy of a post. So, should we be climate adaptation optimists?
The paper makes a range of arguments, some explicitly and some more implicitly. I take the main arguments to be:
Most of us probably don’t care about climate or climate change per se, we care mainly about how it shapes outcomes that we care about (livelihoods, health, crop yields, life enjoyment, etc). [This is implicit, but I think underlies their argument]
These outcomes can evolve over time for reasons related to climate and reasons unrelated to climate. Mostly, these outcomes have improved dramatically over time, for reasons unrelated to climate.
For many outcomes, estimated climate damages are very small relative to these longer-run changes.
Improving these outcomes by investing in climate mitigation – i.e. reducing damages by reducing climate change – can be a very expensive way to improve these outcomes, and for most individual actors in the world (including large emitters), the costs of unilateral mitigation likely exceed the benefits they would directly experience in terms of reduced damages, perhaps substantially.
Investments in “adaptation” or “development” (and here these terms get conflated in a way that’s a bit confusing) are likely cheaper ways to improve outcomes
Given all this, research and policy focuses too much on mitigation and not enough on adaptation. Climate-focused orgs, and governments more broadly, should re-balance their efforts more in favor of adaptation.
As much as I enjoy arguing with Matt Kahn in particular, I broadly agree with all of these points. I think the authors are also correct that, in climate circles, saying some of these points out loud is somehow a bit taboo, since they might seem to undermine support for climate action. This was basically the reaction to the now infamous Gates Memo, which makes notionally similar points to this paper.
However, I would frame some key points a bit differently, and I am not sure the evidence on these points is always as clear cut as they suggest. And I think they omit some key considerations. Here’s how I think about it:
We have outcomes that we care about: health, incomes, livelihoods, happiness, biodiversity, etc. For simplicity, we’ll follow Burgess et al and call these O. (For the pedants, O is a vector but I am not using bold face and I will further abuse notation below) .
We have climate C which is changing, and again for simplicity, assume these changes on average negatively affect O, so dO/dC is negative (pedantic derivative notation, denoting the effect of small changes in climate dC on changes in outcomes dO).
Imagine that we have three basic tools to improve O: (1) “development” policies, which improve O separate from what happens with the climate; (2) “mitigation” policies, which improve O by explicitly trying to reduce climate change (i.e. making dC smaller), and (3) “adaptation” policies, which improve O by explicitly trying to reduce the damages of a given change in climate (i.e. making dO/dC smaller).
These policies can interact in important ways. Some key interactions:
Development policies, by making people better off, can also reduce marginal damages (dO/dC), if you think that wealthier individuals or societies are in general less harmed by a change in climate. They can potentially also make climate change worse, if fossil energy is used as an input to development. This in turn can worsen O, but the effects are small the more local the development policy.
Mitigation policies, if they divert local resources from development, could reduce O locally or even globally (or cause it to increase more slowly). Alternatively, they could boost O by delivering win-win economic benefits, such as solar energy savings and energy efficiency (a point that Burgess et al. discuss).
Adaptation policies, if they come at the expense of mitigation policies, could increase C.
I see the Burgess et al paper as being mainly about which of our three basic policies should we choose, although I think they confuse (or don’t usefully distinguish) “development” policies, which target O directly or possibly indirectly through changes in income, from “adaptation” policies, which we typically think of as directly targeting the damages from a changing climate dO/dC. Development policies can be adaptive, but need not be.
They rightly make the important point that the policy choice depends substantially on the scale at which we’re trying to affect outcomes. At the local level, unilateral mitigation has basically no effect on climate, and so Policy 2 won’t get you much, at least in the case where we ignore “co-benefits” such as reductions in local air pollution (a substantial omission). At a global scale, we seem to have lots of options where, even if we ignore co-benefits, the benefits of abating a ton of emissions substantially exceed the costs. [to wit: global benefits, as estimated by the social cost of carbon, are at least $200/ton and potentially way higher, see here and here; many abatement technologies will remove carbon for well under $100/ton.] In contrast, and in principle, investments in a development or adaptation policy are perhaps more scale neutral – local investments could yield local benefits, and national investments could yield national benefits. So a local policymaker might then sensibly prefer Policy Options 1 or 3 if their goal is to increase O.
In practice, though, the optimal policy choice is an empirical question, and the empirics are often a lot less clear than we want. For instance:
Climate change might be a first order challenge for development. If you believe that a warming climate reduces economic growth, a point on which there is now strong evidence (again see here and here), then dO/dC can be big enough to really matter for longer run trends in outcomes. We estimate that a warming climate could knock 1-2 percentage points off of GDP growth rates in hot places. Often their baseline growth rates are only 1-2 percent/year, meaning warming literally stops growth in many places. There are potentially worse outcomes if you consider conflict dynamics. If true, then it is incorrect to say that long run secular trends in outcomes are going to overwhelm climate impacts.
It is not always apparent that development (income growth) reduces the impacts of climate on a population, at least for reasonable changes in income. In data from the US, Mexico, and EU, we find income has a very limited effect on temperature-related mortality (see Fig 1 here). In a global study of the same topic, Carleton et al find that a much larger increase in incomes might reduce heat-related mortality by about 20%, but the estimate is not statistically significant. In GDP data, we find that for two countries with the same average temperature, the wealthier one is again not statistically less sensitive to warming. In a meta-analysis across a bunch of different outcomes, we find that most outcomes have not become less sensitive to climate over time, despite incomes having gone up more than 2x in many settings.
The effects of income do show up strongly in other settings. In ongoing work, we estimate that the impacts of a 99th percentile (i.e. very hot) day on under-5 mortality is 20-30x lower in the US than in Sub-Saharan Africa. But, US per capita incomes are 40x African incomes, and Africa would have to grow at roughly 2%/year per capita for nearly 200 years to reach US income levels. So saying development improves outcomes (clearly true) and development reduces climate damages (maybe true) does not tell us that development is going to help very quickly.
Estimates of the co-benefits of mitigation from local air pollution reduction are often massive, often outweighing the “direct” climate benefits. This changes the calculus on the returns to local mitigation efforts, but the magnitude really depends on the context and the type of mitigation intervention.
We actually don’t really know what “adaptation” policies work at a local level. This is an area where, as the authors point out, we need a lot more research! But existing work, including syntheses our group has tried to put together (paper here), suggest that we often really don’t know what works to reduce damages in a lot of settings. So it’s not like we know how to just “do” adaptation.
We also don’t really know what “development” policies work at a local or national level. We certainly have a lot of great micro-evidence on how to improve certain outcomes like health, but we know a lot less about how to make countries grow faster, at least in any directed way.
If you’re a person who mainly cares about improving a specific health outcome or reducing overall mortality (see: Bill Gates), then reducing mortality through mitigation is probably an incredibly expensive way to do it relative to other leading health interventions. I did some simple math on this a while back and estimated that mitigation is maybe 100x more expensive per life saved than well-tested health interventions like malaria bednets or vitamin A supplementation (we are currently working on a more comprehensive analysis). This ignores air pollution co-benefits and a range of other concerns, but here I am with the authors that we can be pretty confident that your marginal dollar is probably best spent on something other than climate mitigation if you care about specific outcomes like mortality.
Caveats on caveats: some mitigation is probably very inexpensive, and the costs of mitigation generally fall as we do more of it, but in ways that are hard to predict
So, should we be climate adaptation optimists? I guess my view is that we should at least be climate adaptation apologists: we should acknowledge what our overarching goals are (for many of us, this will be in improving outcomes like health and livelihoods), acknowledge that there are multiple paths to achieving these goals – mitigation, adaptation, or development policies, in the above framing – and likely real tradeoffs among them, and better try to estimate which approach looks most beneficial in which setting. In many settings, and especially for local actors, the best choice is probably not going to be mitigation policy. It could be adaptation, or it could be something else. We know a lot about the costs and benefits of mitigation policy, but we are in a near-evidence-free zone currently on much adaptation policy. To really be adaptation optimists, we need to find some things that we know work. Optimism is not a plan.