Monday, June 27, 2016

Conflict in a changing climate: Adaptation, Projection, and Adaptive Projections (Guest post by Tamma Carleton)

Thanks in large part to the authors of G-FEED, our knowledge of the link between climatic variables and rates of crime and conflict is extensive (e.g. see here and here). The wide-ranging effects of high temperatures on both interpersonal crimes and large-scale intergroup conflict have been carefully documented, and we’ve seen that precipitation shortfalls or surpluses can upend social stability in locations heavily dependent on agriculture. Variation in the El Niño cycle accounts for a significant share of global civil conflict, and typhoons lead to higher risks of property crime.

Much of the research in this area is motivated by the threat of anthropogenic climate change and the desire to incorporate empirically derived social damages of climate into policy tools, like integrated assessment models. Papers either explicitly make projections of conflict under various warming scenarios (e.g. Burke et al., 2009), or simply refer to future climate change as an impetus for study. While it’s valuable from a risk management perspective to understand and predict the effects that short-run climate variation may have on crime and conflict outcomes today, magnitudes of predicted climate changes and the lack of clear progress in climate policy generally place future projections at the heart of this body of work.

However, projections are also where much of the criticism of climate-conflict research lies. For example, see Frequently Heard Criticisms #1 and #6 in this post by Marshall. These criticisms are often reasonable, and important (unlike some other critiques you can read about in that same post). The degree to which impacts identified off of short-run climate variation can effectively predict future effects of long-run gradual climate change involves a lot of inherent (statistical and climatological) uncertainty and depends critically on likely rates of adaptation. Because the literature on adaptation is minimal and relies mostly on cross-sectional comparisons, we are limited in our ability to integrate findings into climate policy, which is often the motive for conducting research in the first place. 

Recently, Sol, Marshall and I published (yet another) review of the climate and conflict literature, this time with an emphasis on adaptation and projection, to try to address the concerns discussed above. I’m going to skip the content you all know from their detailed and impressive previous reviews, and talk about the new points we chose to focus on.


1.     Only some of the existing tests for evidence of adaptation inform us about mechanisms, and therefore only some are helpful for making better projections of future impacts.

Once Sol, Marshall and Ted convinced us that there are strong causal relationships between climate and conflict outcomes, everyone wanted to know about mechanisms. While many papers are silent on the particular pathway through which impacts occur, there are two central theories: shifting economic incentives (climate shocks lower economic productivity and thus lower the opportunity cost of crime, reduce state capacity, or reallocate resources across actors), and direct psychological impacts (temperature may affect serotonin and other neurotransmitters, changing proclivity for aggressive behavior). Pinning down these mechanisms is important for policy today, but also for informing possible channels of adaptation – if climate-induced economic shocks are responsible, then income-based adaptation may make future, richer, societies less vulnerable than those we study today.  

Many papers look for evidence of adaptation, but the three main categories of adaptation tests in existing work don’t equally inform us about the mechanisms behind the climate impact identified. Without clear information on mechanisms, it’s hard to use evidence of adaptation to improve predictions of the future (discussed in point 3 below).

The three categories of tests are:

a. Long-run evolution of short-run sensitivity

In these tests, authors take their causal estimates of climate’s impact on a particular conflict outcome, usually identified off of short-run climate variation, and see how that response changes over time by splitting their sample into sub-periods. A declining response over time suggests some form of adaptation. These tests are well identified, but they don’t tell us much about mechanisms – a falling response (or lack of one) over time could be due to income growth, changing institutions, or even to upward trends in average temperature. Without knowing why a response changed (or didn’t) over time, the only way to incorporate these results into a projection is to assume that the same rate of change in sensitivity will occur in the future as had in the past.

b. Long-run sensitivity to gradual change

This strategy estimates a long-run response using gradual trends in both climate and conflict (see Marshall and Kyle Emerick’s paper on U.S. agriculture). If conflict outcomes respond similarly to these gradual trends as they do to short-run variation, this is evidence that our high-frequency empirical estimates can be applied to future projections without worrying about accounting for adaptation (see Sol’s recent paper for more details on using this approach to test for adaptation). If, however, the two approaches lead to different sensitivities, we have evidence of adaptation, but, like the tests in (a), little to point to in terms of what caused that adaptation to occur. 

c. Identifying specific adaptation pathways

In the climate and conflict literature, we have two recurring hypotheses about adaptation – acclimatization and economic incentives – and in this approach authors test these directly to look for possible adaptation pathways. Heterogeneous responses across space and/or time can be correlated with levels of average temperature or income to test whether these covariates affect sensitivity.

Obviously there are identification challenges in this approach, but some studies have isolated exogenous variation in a particular mechanism to test its effectiveness at reducing sensitivity (e.g. here and here). Even without careful identification, correlations between direct adaptation pathways and estimated climate-conflict sensitivities can be incredibly useful for projections, since predictions for covariates (e.g. country-specific income growth) are often publicly available.

2.     We provide the first large-scale evidence that income can mitigate the relationship between temperature and both intergroup conflict and interpersonal crime.


Only a few papers have followed approach (c) above to directly test whether rising incomes can mitigate climate-conflict relationships. While studies taking approaches (a) and (b) often conclude that adaptation is minimal or undetectable, some of the best identified work employing (c) suggests that income may play a substantial role in facilitating adaptation (again, see here). 

We apply this approach to the 56 empirical studies reanalyzed and reviewed in our paper to look for large-scale cross-sectional evidence that income plays a role in determining heterogeneous sensitivities across space and time. We take the standardized effects (percent in conflict risk per standard deviation increase in a given climate variable) and pair them with real income data to estimate simple cross-sectional correlations for each type of conflict and for temperature and rainfall separately. Our results are in the following figure:



We find a clear relationship for temperature: rising incomes mitigate the temperature sensitivity of both intergroup and interpersonal conflict, with slopes of -6.1 and -2.6, respectively. These slopes imply that for a 10% increase in income, the marginal effect of temperature falls by 0.6%/σ for intergroup and 0.3%/σ for interpersonal conflict. Given the lower average marginal effect for the interpersonal category, this implies a relatively larger effect of income on crimes like rape, murder and assault, compared to violence outcomes like riots or war. 

As is clear from the figure, we find no evidence of income-based adaptation for precipitation effects.

3.     Incorporating evidence of adaptation into future projections is important.

Nearly all projections of future impacts of climate change on conflict apply a historical relationship to climate model output without accounting for likely adaptation (or maladaptation). A notable exception is the American Climate Prospectus, where differential sensitivities were measured across different long-run average climates within the U.S. to empirically ground adaptation into the future. However, even this application of adaptation does not address the income channel. Incorporating income-based adaptation could be critical for improving the accuracy of future projections, particularly in developing countries where predicted rates of economic growth are high.

Acknowledging that our income adaptation exercise is correlational only, we use our estimates of the income gradient shown above to provide suggestive evidence of the adaptive potential of economic growth for future conflict outcomes. We apply the meta-analysis mean of the direct climate effect of temperature in each conflict category to a standardized projection of warming based on model output for 21 different climate models. The influence of incorporating adaptation can be seen with two example countries:
  • In Tanzania, expected warming by 2050 under business-as-usual is approximately 3σ, where σ is the current standard deviation of annual temperature. Ignoring the adaptive effect of income growth, a simple projection estimates that the relative risk of intergroup conflict will rise 44% and of interpersonal conflict 8% under this scenario. But when we account for the income gradient, assuming incomes grow in Tanzania as they have since 1989, these relative risks fall enormously – to 11% and nearly zero for intergroup and interpersonal conflict, respectively.
  • In The Netherlands, expected warming by 2050 is approximately 2σ. Again, ignoring adaptation, a back-of-the-envelope projection would suggest that intergroup conflict risk should rise 30% and interpersonal 6% by 2050. Adding in the adaptive effect of income lowers these predictions to 20% and 2%, respectively.
Our estimates are cursory and use a simple projection and basic cross-sectional correlation to capture the income gradient. Nonetheless, we think this thought experiment is useful for illustrating the potentially important role of generating “adaptive projections”, particularly within the conflict literature where economic mechanisms have played a central role both theoretically and empirically.

We conclude by emphasizing that future adaptation research should focus on causally identifying mechanisms through which adaptation occurs, and incorporating these estimates of adaptation into projections under climate change. 

This is a guest post by Tamma Carleton, a doctoral student at UC Berkeley.

1 comment:

  1. I couldn't agree more on the importance of factoring potential adaptation explicitly into forecasts. This is a solid step in the right direction. Great post!

    ReplyDelete