The carbon costs of global wood harvests

The study addresses how to accurately account for the greenhouse gas (GHG) effects of wood harvests, a major driver of carbon loss from vegetation and soils. Prevailing accounting methods—such as lifecycle assessments treating wood as carbon neutral when harvests are ‘sustainable’, and national GHG reporting that nets harvest emissions with forest regrowth on managed lands—often attribute carbon gains from forest growth that would occur anyway to new harvests. This can create an impression that harvests, particularly in temperate regions with recovering forests, have little or even beneficial climate impacts, while tropical harvests appear costly. Conversely, studies that report only gross emissions omit potential regrowth after harvests. Given policy emphasis on near-term emission reductions, the authors propose valuing emissions and removals over time using discounting to better reflect the climate costs of new wood harvests relative to an unharvested counterfactual.

The paper reviews accounting practices in LCA for wood products and bioenergy that often assume carbon neutrality if harvests do not exceed forest growth, sometimes crediting storage in long-lived products or attributing average stand carbon stocks to harvested wood. National inventories under IPCC guidelines can net all stock changes on managed forests, including regrowth from past land-use changes and climate-driven growth (CO₂ fertilization, warming, nitrogen deposition), obscuring the effects of new harvests. Scientific estimates of land-use change emissions often separate the residual land sink but still may report net effects combining new harvests with regrowth from previous harvests. Regional dynamics differ: temperate countries largely show forest recovery from past clearing and harvesting, whereas tropical countries face expanding agriculture and harvesting. The authors cite extensive literature arguing that forest growth that would occur anyway should not be credited against new harvest emissions and note other studies reporting gross emissions from harvests, especially in the tropics. Prior work on time dynamics includes radiative forcing-based approaches and discussions of substitution effects of wood versus steel/concrete, with acknowledged variability and uncertainty.

The authors develop the Carbon HARvest Model (CHARM), a global forest carbon model that tracks carbon flows among storage pools (live vegetation, roots, slash, wood products of varying lifetimes, and landfills). The atmospheric impact of a harvest in any year is quantified as the difference between carbon stored across all pools under harvest versus a counterfactual of no harvest with forests evolving naturally (growth, mortality, decay). Annual emissions/removals are the year-to-year changes in this differential stock. To value time, they apply a 4% real discount rate to emissions and removals for 40 years after each year’s harvest, producing ‘harvest-year equivalent emissions’ consistent with approaches valuing near-term mitigation and a middle-ground estimate of a constant social cost of carbon. Sensitivity analyses also consider discount rates from 0–6% and 100-year payback horizons. Future wood consumption (by country) is projected using a fixed-effects model relating consumption of four product groups—long-lived products (LLP: sawn wood, panels, other industrial roundwood), short-lived products (SLP: paper and paperboard), very-short-lived products harvested for energy (VSLP-WFL: wood fuel), and very-short-lived industrial wastes burned for energy (VSLP-IND)—to population, GDP, and time (technology proxy), with separate fits for developed and developing countries. These consumption projections are translated to roundwood harvest requirements using FAO-based material flow relationships (including use of manufacturing residues and recovered paper). Seven wood supply scenarios are analyzed for 2010–2050: (1) secondary forest harvest with regrowth; (2) secondary forest harvest with conversion to productive plantations; (3) mixed harvest where half of harvested secondary forests are mature; (4) establishment of new tropical plantations on 2 Mha yr⁻¹ of agricultural land (counterfactual: secondary forest regrowth); (5) 25% higher productivity in existing plantations; (6) higher harvest efficiency in tropical secondary forests (reduced unharvested felled wood); (7) reduced wood fuel demand (linear reduction to 50% of projected 2050 baseline). Substitution benefits are estimated separately as reductions in fossil and process emissions when wood replaces concrete/steel in construction or propane for energy, using mid-range literature values, acknowledging that substitution does not change biogenic emissions from harvest. Harvested area is estimated in ‘clear-cut equivalents’ to harmonize across thinning/selective logging where area data are sparse. Robustness is evaluated via sensitivity analyses of key parameters, soil carbon is excluded due to uncertainty, and indirect effects (e.g., roads) and biophysical non-GHG effects are noted but not modeled.

• Global wood harvests are projected to increase by 54% from 3.7 billion m³ (2010) to 5.7 billion m³ (2050): LLP +69%, SLP +128%, VSLP-WFL (wood fuel) +22%, VSLP-IND +91%.
• Annualized, time-discounted carbon costs of global wood harvests (2010–2050, 4% discount over 40 years) are estimated at 3.5–4.2 Gt CO₂e yr⁻¹ across scenarios, comparable to common estimates of 3–4 Gt CO₂e yr⁻¹ for agricultural land-use change.
• Existing (2010-level) wood demand accounts for about 78% of the costs (Scenario 1: 3.2 Gt CO₂e yr⁻¹), with rising demand accounting for the remainder. On average, industrial wood and wood fuel each contribute roughly half of the carbon costs.
• Substitution benefits (production emissions avoided by using wood instead of concrete/steel or propane) are estimated at 0.8–0.9 Gt CO₂e yr⁻¹ globally but do not offset biogenic emissions from harvest; overall climate benefits are not implied.
• Estimated harvested area (clear-cut equivalents) across scenarios is 756–855 Mha over the period. Increasing plantation productivity by 25% reduces harvest area by ~60 Mha; reducing wood fuel demand by 50% by 2050 reduces harvest area by ~70 Mha relative to Scenario 1.
• Sensitivity analyses indicate total carbon costs remain around 3–5 Gt CO₂e yr⁻¹ when focusing on decadal timescales; results are relatively insensitive to discount rate choices between 2–6% or extending payback to 100 years (e.g., 100-year horizon at 4% reduces costs by ~3%; 2–6% over 100 years varies costs by −12% to +1%). Only 0% discounting over 100 years lowers annual costs substantially (−40%).
• Even ambitious supply-side scenarios (e.g., full conversion to productive plantations, new tropical plantations, higher efficiency, higher plantation productivity) do not reduce annualized carbon costs below ~3.5 Gt CO₂e yr⁻¹.
• The analysis suggests significant, often overlooked carbon costs of wood harvests and identifies reduced harvesting—especially wood fuel—as a potential mitigation wedge.

By explicitly accounting for time via discounting and comparing harvest trajectories to an unharvested counterfactual, the study clarifies that new wood harvests impose substantial near- to medium-term carbon costs that are not canceled by background forest growth unrelated to the harvest. This addresses misperceptions from netting approaches in LCA and national inventories that can imply carbon neutrality or climate benefits of harvesting in regions with strong forest regrowth or climate-driven sinks. The magnitude of annualized costs (3.5–4.2 Gt CO₂e yr⁻¹) underscores that wood harvests are a climate factor comparable to agricultural land-use change. Substitution benefits modestly lower production emissions relative to alternative materials but do not negate biogenic emissions, and their magnitude may decline as steel and concrete decarbonize. Scenario analyses show that enhancing plantation productivity, improving harvest efficiency, expanding plantations, or reducing wood fuel demand can influence areas harvested and total costs but cannot eliminate substantial carbon impacts; demand reduction (notably for wood fuel) yields notable savings. Although some economic models might posit alternative land-use counterfactuals (e.g., more cropland without wood harvests), absolute emissions from harvests remain large and policy-relevant; recognizing them can inform integrated strategies that pair harvest reductions with measures to avoid leakage (e.g., preventing cropland expansion). Overall, the findings emphasize the value of prioritizing near-term emission reductions consistent with climate targets and reconsidering assumptions of carbon neutrality in wood use.

The paper introduces CHARM, a time-discounted, pool-based global carbon accounting framework for wood harvests, and shows that global harvests from 2010–2050 likely entail annualized carbon costs of 3.5–4.2 Gt CO₂e yr⁻¹, on par with emissions from agricultural land-use change. It challenges accounting practices that offset new harvest emissions with background regrowth and demonstrates that even with regrowth, harvests impose significant near-term climate costs. The results suggest a meaningful, underrecognized mitigation opportunity: reducing wood harvests—especially traditional wood fuel use—improving harvest efficiency, and enhancing plantation productivity can lower carbon costs and harvested areas, helping ‘buy time’ for broader decarbonization. Future research should refine input data and econometric relationships, incorporate soil carbon dynamics and indirect effects (e.g., infrastructure), better quantify biophysical non-GHG effects (e.g., cloud and albedo impacts), improve representation of selective logging areas, and explore integrated economic counterfactuals and policy mechanisms to avoid leakage while reducing harvests and shifting material demand toward lower-emission alternatives.

• Soil carbon impacts of harvests are omitted due to uncertain loss and recovery rates, likely making estimates conservative; meta-analyses indicate average soil carbon losses (~11% in upper layers) and potentially larger belowground losses, with substantial losses reported in logged tropical forests.
• Indirect effects of forestry (e.g., road building) are not included; tropical estimates suggest these can be several times direct effects.
• Biophysical, non-GHG effects (albedo, cloud formation) are not modeled; current uncertainties preclude reliable inclusion.
• Consumption projection model fits are reasonable but imperfect; wood fuel projections are particularly uncertain due to heterogeneous energy transitions across income levels.
• Material flow and area estimates rely on FAO data and assumptions (e.g., clear-cut equivalents), with limited data on selective harvest areas and intensities.
• Substitution benefits are uncertain and context-dependent; future decarbonization of steel and concrete could reduce or eliminate such benefits.
• Discounting choices and payback horizons influence results, though sensitivity shows modest effects except under extreme assumptions (0% over 100 years).