Warning from tropical deforestation reduces worker productivity in rural communities

The study investigates whether and how the loss of trees—and the associated reduction in local cooling from shade and evapotranspiration—affects the productivity of rural outdoor workers in low-income, tropical settings. While the cooling benefits of trees are well documented in urban areas, little causal evidence exists for rural contexts where baseline temperatures are already high and adaptive infrastructure (e.g., air conditioning) is limited. The authors posit that deforestation can produce rapid, substantial local warming, exacerbating heat stress and reducing labor productivity among outdoor workers engaged in physically demanding tasks. They further highlight gaps in understanding real-time behavioral adaptations (e.g., rest-taking, work pacing) and whether financial incentives can offset heat-related productivity losses. The primary hypotheses are that heat exposure in deforested areas reduces productivity relative to forested areas and that higher financial incentives might mitigate productivity declines by altering behavior (e.g., fewer breaks).

Prior literature establishes that trees and forests provide local cooling through shade and evapotranspiration, a benefit well recognized in urban heat island mitigation. However, evidence on rural, low-income settings is sparse, despite these areas often exceeding safe heat thresholds. Studies have reported that outdoor workers adapt their behavior to heat (e.g., changing timing and intensity of work), but evidence from high-frequency, real-time data is limited; most research emphasizes longer-term adaptations (e.g., cooling technologies). There is also an open question whether heat imposes a binding physical constraint on productivity versus one that can be offset by increased incentives. The authors position their work to address these gaps with a randomized field experiment, high-frequency physiological and behavioral measurements, and an explicit test of whether increased piece-rate incentives can overcome heat-induced constraints.

Design: A randomized field experiment was conducted with 361 healthy, working adults from ten rural villages in Berau Regency, East Kalimantan, Indonesia. Participants were randomly assigned to work in either forested or deforested settings and cross-randomized to either a standard or a high piece-rate incentive in a 2×2 factorial design. The work session lasted 90 minutes and mimicked common harvesting activities: packing twelve 500 g items into a backpack, carrying it 25 m, unpacking, and stacking, repeated for the session duration. Water, snacks, and shade were provided, and participants could rest ad libitum. Sampling and setting: Villages were selected via multistage sampling to represent areas above and below the median intact forest cover within a 5 km buffer, with additional inclusion criteria for accessibility and land/water cover characteristics. Individuals were eligible if aged ≥21, able to hike ≤10 km, and without recent chronic respiratory or cardiac issues. Informed consent was obtained. Ethics approvals were secured (AAEA RCT Registry AEARCTR-0002789; University of Washington IRB). Measurements: High-frequency data were collected on output, break-taking behavior, and core body temperature at one-minute intervals, and physical effort via accelerometry at one-second intervals (used to derive minutes in moderate-to-vigorous physical activity, MVPA). Thermal environment was characterized using wet bulb globe temperature (WBGT). Core body temperatures were estimated from oral temperatures and heart rate using a validated algorithm. Subjective assessments captured perceived heat impacts on work speed and quality. Statistical analysis: Average treatment effects were estimated via ordinary least squares with indicator variables for experimental factors, interactions, robust standard errors at the individual level, and village fixed effects. Transformations (e.g., inverse hyperbolic sine) were applied as appropriate. Time-to-break analyses used cumulative time to breaks during the 90-minute session. Subjective outcomes were analyzed using multinomial logit to estimate marginal effects. Analyses were conducted in Stata 14. Participants with missing key variables or sensor data were excluded from corresponding analyses; randomization checks indicated excluded samples were similar to included ones.

- Productivity: Participants working in forested settings were, on average, 8.22% more productive than otherwise equivalent participants in deforested settings (p = 0.009), implying productivity was 8.22% lower in deforested areas. - Thermal environment: Forested sites were, on average, 2.8–2.84 °C cooler in WBGT than deforested sites, consistent with forests providing substantial local cooling. - Physiological strain: Workers in forested settings had significantly lower core body temperatures. Median core body temperature was 0.14 °C lower than in deforested settings (p ≈ 0.0098). Time spent in moderate hyperthermia (core temperature > 38.5 °C) was 39.3% lower in forested settings (p = 0.0002). - Mechanism via behavioral adaptation: Productivity losses in deforested areas were driven by increased rest-taking rather than reduced work effort; accelerometer-derived MVPA did not differ significantly between settings. Workers in forested settings took their first break, on average, 12 minutes later than those in deforested settings, with cumulative differences in time to breaks growing over the session. - Incentives: Doubling the piece-rate incentive did not affect strain, effort, breaks, or productivity, suggesting heat imposed a binding physical constraint not offset by financial incentives. - Awareness: Participants in deforested settings were 15% more likely to report lower work speed and 12% more likely to report lower work quality because of heat, indicating awareness consistent with observed behavioral adaptations.

The experiment provides causal evidence that loss of forest cooling services increases heat exposure and reduces the productivity of rural outdoor workers over even short (90-minute) periods. Differences in WBGT and core body temperature confirm that thermal environment is the key pathway. Behavioral adaptation—specifically, increased and earlier breaks—appears to be the primary mechanism, not reduced physical effort. The lack of response to doubled piece-rate incentives indicates that at the observed temperatures, physiological limits constrain output, supporting the view that heat exposure imposes a binding physical constraint on productivity. These findings underscore the importance of conserving forests and integrating trees into working landscapes (e.g., agroforestry, silvopasture) to provide local cooling that can bolster resilience of subsistence workers to climate warming, with implications for both local well-being and national climate strategies. The study situates productivity impacts within broader rural vulnerability, where declines in output can undermine household resilience in settings with limited access to credit, insurance, and cooling infrastructure.

This study demonstrates that deforestation-driven warming reduces rural worker productivity through increased heat exposure, with a quantified 8.22% productivity decline in deforested versus forested settings and significant differences in WBGT and core body temperatures. Short-run behavioral adaptations (more frequent and earlier breaks) drive observed output losses, and increased financial incentives do not overcome heat-related constraints. The results provide policy-relevant evidence that maintaining and restoring tree cover can deliver local cooling services that enhance the adaptive capacity and resilience of rural communities while aligning with broader climate and conservation goals. Future research should examine longer work durations and real-world conditions, different task types and labor contexts, seasonal and longitudinal adaptations, and how complementary interventions (e.g., shade structures, hydration strategies, scheduling) interact with landscape-level tree cover to protect health and productivity.

- External validity and generalizability: The experiment was conducted in one regency (Berau, East Kalimantan) and focused on a specific harvesting-mimicking task; results may differ for other activities, regions, and populations. - Conservative setting: The 90-minute protocol with ready access to water, snacks, and shade likely underestimates heat impacts relative to typical working conditions (longer days, variable access to shade/hydration). - Short-run focus: Findings capture immediate behavioral adaptations, not longer-term physiological or behavioral acclimatization/adaptation. - Missing sensor data and exclusions: Some participants lacked complete sensor data; while randomization checks suggest missingness is random, exclusions reduce sample size for certain outcomes. - Incentive structure and ceiling effects: High baseline incentives and near-thermal limits may have limited scope for additional output, potentially masking marginal incentive effects in other contexts. - Measurement constraints: Core temperature estimated via a validated algorithm based on oral temperature and heart rate rather than direct continuous core measurements; MVPA is a proxy for effort.