Contrasting influences of biogeophysical and biogeochemical impacts of historical land use on global economic inequality

Human-driven land-use and land-cover change (LULCC) has extensively transformed natural landscapes into cropland, grazing land, and urban areas, and intensified land management (irrigation, fertilization, wood harvest). LULCC affects climate through biogeophysical (BGP) processes (altering albedo, evapotranspiration, energy and water fluxes) and biogeochemical (BGC) processes (carbon emissions). While the macroeconomic impacts of greenhouse gases and aerosols via annual mean temperature (AMT) changes have been examined, the economic consequences of LULCC remain unclear due to its dual BGP/BGC effects and spatial heterogeneity (e.g., deforestation cools extratropics via albedo but warms tropics via evapotranspiration reduction). Moreover, beyond AMT, day-to-day temperature variability strongly influences economic growth, with heightened vulnerability in low-latitude, low-income countries. This study asks how historical LULCC since 1850, via BGP and BGC impacts on AMT and day-to-day temperature variability, has affected country-level economic growth and global economic inequality over 1993–2012.

Prior work documents that LULCC contributes substantially to anthropogenic climate change, with global BGP forcing around −0.15 W/m² and notable BGC contributions to historical CO₂ increases. Deforestation warms the tropics and cools extratropics through contrasting BGP mechanisms. Empirical studies link temperature to economic growth via a nonlinear temperature-growth response function with an optimal annual mean temperature; warming tends to harm warm, developing countries and benefit cooler, developed countries, thereby increasing inequality. Aerosol-induced cooling has been shown to offset some GHG warming and may reduce inequality. Recent evidence indicates day-to-day temperature variability can reduce GDP per capita growth more than changes in AMT, with stronger effects in low-latitude and low-income regions. LULCC has also been implicated in changes in temperature variability and extremes. However, the integrated economic implications of LULCC’s BGP and BGC impacts on both AMT and day-to-day variability had not been quantified.

Study period and units: Country-level cumulative economic impacts over 1993–2012 are assessed via changes in annual mean surface air temperature (SAT) and day-to-day SAT variability attributable to historical (1850–2014) LULCC. Climate data and experiments: Multi-model CMIP6 simulations are used. Two concentration-driven experiments—“historical” (with evolving natural and anthropogenic forcings, including land-use) and “hist-noLu” (identical but land use fixed at 1850)—isolate BGP impacts as the SAT difference due to land-surface changes under prescribed identical CO₂. The BGC impact is estimated by (1) computing annual LULCC-induced CO₂ emissions from the difference in net land–atmosphere CO₂ flux between “historical” and “hist-noLu”, (2) converting emissions to accumulated atmospheric CO₂ concentration using a pulse response function (airborne fraction), yielding ~25.3 ppm added by 2014 (about +9% vs. 1850), and (3) mapping SAT responses using the “1pctCO2” experiment by selecting the year corresponding to the relative CO₂ increase (e.g., 9th year for +9%). BGP and BGC SAT impacts are summed for combined effects. SAT fields use 5-year running means to reduce interannual variability. Ensembles and models: For annual mean SAT, BGP impact uses 16 members from 8 GCMs; BGC and combined impacts use 9 members from 4 GCMs (carbon cycle availability). For day-to-day SAT variability, 5 members from 3 GCMs are used (daily data availability). All outputs are interpolated to 2.5° × 1.9°. Observed/reanalysis and population weighting: Factual SAT uses ERA5 (central) and MERRA-2 (sensitivity). Grid SAT is aggregated to country level using GPWv4 population weights. A delta method constructs counterfactual SAT without LULCC by subtracting modeled LULCC-induced SAT differences from reanalysis. Economic data: World Bank statistics provide GDP per capita, growth rates, and population for 147 countries (1993–2012). Economic impact via AMT: The temperature-growth response function f(T) = β₁T + β₂T² (nonlinear, bootstrapped by country 1000 times) yields 1000 parameter sets and temperature optima (median ~13.12 °C; interquartile 11.80–14.55 °C). Applying factual and counterfactual SAT to f(T) produces ΔGrowth due to LULCC. Counterfactual GDP per capita time series are then iteratively computed starting from 1993 using Growth_NLULCC = Growth + ΔGrowth (delta method), and cumulative effects to 2012 are expressed as percent differences between factual and counterfactual. Economic impact via day-to-day variability: Annual mean, seasonally adjusted day-to-day SAT variability is the intra-monthly standard deviation of daily SAT averaged over 12 months. LULCC-induced changes (ΔTVAR) are mapped from CMIP6. A damage function ΔGrowth = α × ΔTVAR, with α negative and dependent on countries’ seasonal temperature range (larger magnitude α in low-latitude, small-seasonality countries), is applied. Seasonal decomposition (MAM, JJA, SON, DJF) uses season-specific α to quantify seasonal contributions. Inequality metrics: Global economic inequality is assessed via 80:20 and 90:10 ratios of the population-weighted GDP per capita percentiles. Probabilistic statements follow IPCC guidance, with significance if ≥ two-thirds of ensemble members agree on sign. Sensitivity and robustness: Results are tested with ERA5 vs. MERRA-2 SAT, alternative temperature-growth response functions and lags, and show consistency. The study focuses on 1993–2012 due to socioeconomic data availability. Data sources and access: LUH2 for land use; CMIP6 outputs; ERA5 and MERRA-2 SAT; GPWv4 population; World Bank socioeconomic data; published response functions for temperature and variability impacts.

- Historical LULCC’s BGP and BGC impacts have contrasting climate effects: BGP generally cools extratropics (via increased albedo from deforestation) and warms the tropics (via reduced evapotranspiration), while BGC produces near-global warming due to added CO₂. The combined effect yields net warming in most countries, dominated by BGC. - Day-to-day SAT variability changes are dominated by BGC: variability decreases at high northern latitudes (north of 60°N) and increases across the tropics and subtropics (30°S–30°N). - Economic impacts via annual mean SAT (1993–2012): Warming from combined LULCC harms warm-climate, typically lower-income countries by moving them further above the temperature optimum (e.g., India cumulative damage by 2012; UK benefits by moving closer to the optimum). At the global level, combined LULCC warming reduces GDP per capita by about −1.30% in 2012 (25th–75th percentile: −3.02% to −0.11%). BGP cooling alone yields global gains of about +0.88% (25th–75th: +0.10% to +2.19%), whereas BGC warming drives the net global damage. - Distributional impacts via annual mean SAT: Many low-latitude countries benefit from BGP cooling but are harmed by BGC warming. High-latitude cool-climate countries (e.g., Russia, Canada, Norway) are harmed by BGP cooling but benefit from BGC warming. Where BGP cooling dominates (e.g., Canada), combined effects can reduce growth. - Inequality via annual mean SAT: By 2012, combined LULCC increases global inequality with 80:20 ratio up by +5.10% (interquartile +1.18% to +12.75%) and 90:10 by +2.64% (−0.80% to +5.23%), driven by BGC and partly offset by BGP. - Economic impacts via day-to-day variability: Combined LULCC increases variability (and economic damages) in many low-latitude, low-income countries, while reducing variability (and yielding benefits) in many extratropical high-income countries (e.g., Canada, US, Western Europe). Inequality increases further: 80:20 ratio +9.36% and 90:10 +2.49% by 2012 due to variability changes. - Seasonal contributions: Spring (MAM) variability changes contribute most to annual economic impacts in many countries due to higher economic sensitivity in spring and prominent LULCC impacts on spring variability. - Robustness: Findings hold across reanalysis datasets, alternative temperature-growth response specifications, and ensemble members.

The study disentangles the biogeophysical cooling and biogeochemical warming effects of historical LULCC and links these climate responses to country-level economic outcomes. Because BGC-induced warming dominates, combined LULCC effects raise annual mean temperatures in most places, which, through the nonlinear temperature-growth response, tends to damage warm, developing economies and benefit cooler, developed ones, widening between-country inequality. Additionally, LULCC-driven increases in day-to-day temperature variability in the tropics and decreases in high latitudes impose further asymmetric economic impacts that exacerbate inequality. These results directly address the research question by quantifying how distinct LULCC climate pathways translate into heterogeneous economic growth effects and demonstrate that LULCC has contributed to increasing global economic inequality over recent decades. The findings underscore the importance of accounting for both mean and variability changes, and for both BGP and BGC processes, when assessing socioeconomic impacts of land-use policies.

Historical LULCC has produced opposing BGP and BGC climate effects, with BGC-dominated warming leading to net economic damages in warm, developing countries and benefits in cooler, developed countries, thereby increasing global economic inequality. Changes in day-to-day temperature variability further intensify this inequality. The study integrates multi-model CMIP6 climate responses with empirical economic response functions to provide country-level estimates and inequality metrics. Policy implications include the need to evaluate land-use strategies (e.g., deforestation/afforestation trajectories and land management) for their long-term climate and economic distributional effects. Future work should quantify scenario-dependent LULCC economic impacts, include non-CO₂ greenhouse gas forcings explicitly, extend analyses over longer periods, and improve subnational assessments where climate impacts are highly heterogeneous.

- For BGC impacts, CO₂ effects are inferred via the 1pctCO2 experiment, which is transient and not equilibrated; no dedicated CMIP6 experiment isolates historical LULCC BGC impacts directly. Non-CO₂ GHGs and ozone precursors from LULCC are not fully included in the main estimates; their warming would likely increase the magnitude of inequality impacts. - Historical land-use forcing uncertainties (LUH2 high/low reconstructions) affect magnitude though spatial patterns are robust. Ensemble sizes are limited for some components due to model and data availability (especially daily outputs for variability; fewer models with interactive carbon cycles). - Bias correction uses a delta method applied to reanalysis; residual model/reanalysis biases may remain. The analysis period (1993–2012) is constrained by socioeconomic data availability; longer periods would yield larger cumulative effects. - The country-level focus may mask important within-country heterogeneity; precipitation and other climate variables can affect regional economies even if country-level precipitation effects on GDP per capita are weak. Adaptation heterogeneity and risk management differences are represented indirectly via empirical response functions and may vary over time.