Are We Adapting to Climate Change?

Marshall Burke, Mustafa Zahid, Mariana Martins, Christopher M. Callahan, Richard Lee, Tumenkhusel Avirmed, Sam Heft‐Neal, Mathew V. Kiang, Solomon Hsiang, David B. Lobell

TL;DR

Despite the common assumption that human ingenuity will naturally lead to climate adaptation, a comprehensive analysis of 50 years of global data shows that systematic adaptation is the exception, not the rule. Out of 21 societal outcomes studied—ranging from crop yields to mortality and conflict—only six showed significant signs of reduced sensitivity to climate extremes, while five actually became more vulnerable over time.

What Problem Does This Paper Address?

A central debate in climate economics is the "clever human" hypothesis: the idea that as the planet warms, societies will innovate and adjust, rendering historical climate-damage relationships obsolete. If we adapt quickly, future damages will be lower than feared. If we don't, current projections likely underestimate the catastrophic costs of inaction.

The authors point out that while there are many anecdotes of adaptation (e.g., drought-tolerant seeds or air conditioning), there is a lack of systematic, empirical evidence showing that these efforts are moving the needle at a macro level. This paper seeks to quantify the speed of adaptation across diverse sectors and geographies to determine if we are actually becoming less sensitive to a changing climate.

Methods

The researchers employed a rigorous panel econometric approach, analyzing longitudinal data to estimate "response functions"—the mathematical relationship between a climate variable (like temperature) and a societal outcome (like mortality).

Key methodological innovations include:

  • Decadal Sensitivity: Instead of assuming a fixed relationship, they allowed the response functions to vary by decade, tracking whether the "slope" of the impact flattened over time.
  • Total Sensitivity vs. Point Sensitivity: Most studies look at "point sensitivity" (e.g., how much does one 35°C day increase mortality?). This paper introduces "Total Sensitivity," which is the exposure-weighted derivative of the response function. This accounts for the entire temperature distribution, recognizing that a society might get better at handling extreme heat but worse at handling moderate warming, leading to a net increase in vulnerability.
  • Adaptation Decomposition: They separated changes in sensitivity into two drivers: "changing responses" (technological or behavioral shifts, like better medicine) and "changing exposures" (spatial shifts, like populations moving into hotter or more flood-prone regions).

Datasets

The study draws on a massive "meta-analysis" of high-quality subnational data:

  • Agriculture: Yield data for maize, wheat, and soy from the US, EU, Brazil, and India, plus global Agricultural Total Factor Productivity (TFP).
  • Health: County-level mortality in the US (since 1968) and district-level mortality across 37 European countries.
  • Economy: Global GDP growth and US personal income data.
  • Conflict & Crime: Gridded data on African civil conflict and county-month FBI data on violent crime in the US.
  • Disasters: Long-term data on US flood damages and global tropical cyclone wind fields.

Key Insights and Ideas

  • Adaptation is surprisingly rare: In roughly half of the cases studied, there was no statistically significant change in sensitivity. In the other half, the "wins" were nearly offset by the "losses."
  • We are becoming more sensitive in some areas: Brazilian agriculture and African civil conflict have become significantly more sensitive to temperature spikes over the last few decades. In agriculture, this may be a "productivity trap" where intensifying farming (e.g., higher planting density) increases yields in good years but makes the crop more fragile during heatwaves.
  • The "Sun Belt" Effect: In the US, any gains from better technology (like AC) are being partially offset by "changing exposures." As people move to warmer states, the population-weighted exposure to extreme heat increases, raising the total societal sensitivity to warming.
  • Progress has stalled: US heat-related mortality showed a remarkable decline through the 20th century, but that progress has largely stalled or even slightly reversed since the 2000s.

Key Results

The authors summarized their findings across 21 outcomes:

  • 6 Outcomes Improved (Adapted): Including US maize, EU wheat, US violent crime, and US personal income.
  • 5 Outcomes Worsened (Increased Sensitivity): Including Brazilian soy and maize, African civil conflict, and US suicide rates.
  • 10 Outcomes Showed No Change: Including global GDP growth, US flood damages, and Indian wheat.

Quantitatively, the "winners" saw sensitivity declines of roughly 1% to 7% per year, but these were often concentrated in earlier decades. For example, EU mortality sensitivity to temperature showed a strong decline (-5.3% to -7.4% per year), but African conflict sensitivity increased by 3.4% per year.

Limitations and Future Work

The authors acknowledge several caveats:

  • Measurement Error: Historical weather data is less precise than modern satellite data, which could lead to an underestimation of sensitivity in earlier decades (making it look like we are "worsening" when we are just measuring better).
  • Aggregation Bias: While a country might not show adaptation, specific cities or innovative farms within that country certainly are. The study focuses on the net effect.
  • The "Why" remains elusive: The paper proves that we aren't adapting systematically, but it doesn't fully explain why. Future research is needed to determine if the barrier is the high cost of technology, lack of information, or policy frictions like subsidized insurance that discourages risk-reduction.

Ultimately, the paper serves as a wake-up call: we cannot assume that "cleverness" will automatically save us from climate damages. Without a massive scaling of effective adaptation strategies, the costs of a warming world will continue to mount.

Summary updated

Abstract

We study whether the sensitivity of economic, health, and livelihood outcomes to climate extremes has declined over the last half century, consistent with adaptation.Understanding whether such adaptation is already occurring is central to anticipating future climate damages, to calibrating the level of ambition needed for emissions mitigation efforts, and to understanding additional investments in adaptation that could be required to avoid additional damages.Using comprehensive panel data across diverse geographies and outcomes, including data on mortality, agricultural productivity, crime, conflict, economic output, and damages from flooding and tropical cyclones, we find limited systematic evidence of adaptation to date.Across 21 outcomes we study, six show a statistically significant declining sensitivity to a changing climate, five show an increasing sensitivity, and the remainder show no statistically significant change.Our results do not imply that specific documented adaptation efforts are ineffective or certain locations have not adapted, but instead that the net effects of existing actions have largely not been successful in meaningfully reducing climate impacts in aggregate.To avoid ongoing and future damages from warming, our results suggest a need to identify promising adaptation strategies and understand how they can be scaled.
BibTeX
@misc{burke2024are,
  author = {Marshall Burke and Mustafa Zahid and Mariana Martins and Christopher M. Callahan and Richard Lee and Tumenkhusel Avirmed and Sam Heft‐Neal and Mathew V. Kiang and Solomon Hsiang and David B. Lobell},
  title = {{Are We Adapting to Climate Change?}},
  year = {2024},
  doi = {10.2139/ssrn.4964403},
  howpublished = {SSRN},
}

Loading related content...

Figure