Heat, Households, and Human Development

Understanding the Behavioral and Developmental Consequences of Rising Temperatures

Andrés Felipe Camacho
A data story based on Cuartas & Camacho (2025) and Cuartas et al. (2025)

Global temperatures are rising rapidly: 2024 was the warmest year on record and the first to exceed ~1.5 °C above the pre-industrial average, capping a decade of unprecedented heat according to WMO and Copernicus analyses.

This research project links household microdata from UNICEF's Multiple Indicator Cluster Surveys (MICS) with gridded climate reanalysis from ERA5-Land (monthly aggregates, 1950–present) to examine how heat translates into family life and child outcomes. We focus on two dimensions with direct human-development implications:

• Parenting behaviors — e.g., violent punishment and psychological aggression
• Early childhood development — measured via MICS' Early Childhood Development Index (ECDI)

1. Temperature Trends and Anomalies

Raw temperature series reveal strong seasonality. To make warming visible and comparable across places, we use anomalies relative to a 30-year climatological baseline (1991–2020).

2. Data and Linkage: ERA5-Land × MICS

We merge ERA5-Land monthly maximum temperature with MICS geocoded clusters to align children and households to the climate they experience. Each MICS cluster contains 25-30 households at a specific geographic location, allowing us to link household-level outcomes with precise climate data.

The analytic sample includes children from Georgia, The Gambia, Madagascar, Malawi, Sierra Leone, and the State of Palestine (MICS 2017–2020).

Geographic Distribution of Survey Clusters

The maps below show the spatial distribution of MICS survey clusters across all six countries in our sample. This geographic variation allows us to capture diverse climate conditions and household characteristics.

The Spatial Join Process

To link climate data with household outcomes, we perform a spatial join at high resolution. For each child in the sample, we:

• Extract monthly temperature from ERA5-Land at the cluster's coordinates (~0.1° or ~10km resolution)
• Aggregate across time periods from the child's birth date to interview date, as well as specific windows (e.g., 6 months before interview)
• Calculate anomalies relative to the 1991-2020 baseline to capture unusual heat exposure
• Match to household clusters containing 25-30 households each, preserving geographic precision

This process creates a powerful dataset linking precise temperature exposures to child and household outcomes at specific points in time.

3. Methods and Identification

Both papers apply fixed-effects designs on geocoded, time-stamped data to isolate the causal effect of temperature deviations on child and family outcomes.

Outcome Measures

Discipline & Aggression (MICS discipline module):

• Physical punishment (e.g., shook, spanked)
• Severe physical punishment (e.g., hit with object, beat up)
• Psychological aggression (e.g., shouted, called names)
• Non-violent discipline (e.g., explained, took away privileges)

Early Childhood Development (ECDI):

A 10-item parent-reported scale measuring developmental milestones (ages 36-59 months) across four domains: social-emotional, literacy-numeracy, approaches to learning, and physical. Children are classified as "on track" if they exhibit at least 50% of skills in each domain.

Econometric Model

Where:

• Y_ict: outcome for child i in cluster/region c at time t
• Temp_ict: recent or cumulative temperature exposure
• X_i: child and household covariates
• τ_c: subnational region fixed effects
• φ_t: month-year fixed effects

This design holds constant the "typical climate" of places and seasons; remaining variation is the unexpected local heat at interview timing.

4. Heat and Parenting Behavior: Exploratory Patterns

Beyond physical discomfort, heat can affect how people think, feel, and act. Does temperature influence parenting behaviors and discipline practices?

In the pooled sample, prevalence rates of harsh discipline are high: physical punishment ~64%, severe physical punishment ~56%, psychological aggression ~80%, non-violent discipline ~88%. But do these behaviors vary systematically with temperature exposure?

The exploratory visualization reveals an intriguing gradient: children who experienced more severe forms of discipline were exposed to somewhat higher temperatures in the months before the survey. However, this simple correlation could reflect many confounding factors—baseline climate differences across countries, seasonal patterns, wealth, urbanicity, and more.

Why Use Binned Temperature Approaches?

To rigorously test whether heat causes increases in harsh parenting, we cannot assume the relationship is linear. Heat effects may exhibit thresholds—little impact at moderate temperatures, but sharp increases beyond certain points (e.g., 30-32°C). Binned approaches allow us to:

• Avoid functional form assumptions: We don't impose linearity if the true relationship has thresholds or plateaus
• Identify non-linearities: See where temperature effects accelerate or level off
• Increase transparency: Visual binscatter plots show the data directly, not just regression coefficients

The first paper uses both continuous (linear) and binned specifications with comprehensive fixed effects and controls to isolate the causal impact of unusual heat exposure.

Interpretation: Heat loads increase physiological stress and impair emotion regulation, raising the propensity for harsh responses during caregiving. The magnitude of these effects suggests that climate change may be meaningfully increasing the risk of violent discipline in vulnerable populations.

5. Heat and Early Childhood Development: Exploratory Patterns

Does heat exposure affect early childhood development outcomes? The ECDI provides a comprehensive measure of developmental progress across cognitive, social-emotional, and physical domains.

In the analytic sample, 63% of children (ages 36-59 months) were classified as developmentally "on track" and 37% were not. The question is whether temperature exposure—measured from birth to interview—predicts these developmental outcomes.

The exploratory pattern suggests a concerning trend: as temperature bins increase, the probability of being on track declines, with sharper drops appearing beyond ~30-32°C. However, as with the parenting results, this could reflect confounding by baseline climate, seasonality, poverty, or infrastructure differences.

The Case for Binned Temperature Specifications

Developmental impacts of heat may not be linear. There may be:

• Physiological thresholds: Human thermoregulation works well up to a point, then degrades rapidly
• Behavioral tipping points: Heat stress may only become severe enough to disrupt caregiving or child health beyond certain temperatures
• Adaptation limits: Households may cope with moderate heat but struggle once temperatures exceed local norms

The second paper uses binned temperature specifications with 1°C bins (reference: <26°C) alongside the same rigorous fixed-effects design—subnational region and month-year fixed effects—to control for all time-invariant regional characteristics and common seasonal patterns.

Interpretation: Heat exposure during early childhood appears to directly burden children's physiology, sleep, and caregiving environments. The developmental losses are not evenly distributed—they concentrate among the most vulnerable populations who lack adaptive capacity, creating climate-driven inequality in life chances from the earliest years.

6. Synthesis: Connecting Mechanism and Evidence

The exploratory patterns and rigorous paper findings converge on a coherent story about how heat affects families and children.

A Clear Mechanism

Heat increases household stress and reduces the quality of caregiving environments while directly burdening children's physiology, sleep, and cognitive functioning. The effects intensify as temperatures move into the 30–35°C range—precisely where many of our sample countries are experiencing increasing exposure due to climate change.

The econometric designs in both papers focus on within-place variation to avoid confounding from baseline climatic differences and seasonality. By comparing children in the same regions interviewed in the same months but under different temperature conditions, the fixed-effects models isolate the effect of short-run temperature deviations.

The Role of Vulnerability

The developmental impacts are not evenly distributed. They concentrate among:

• Poorer households with limited resources for cooling, healthcare, and nutritious food
• Urban areas experiencing heat island effects and crowded living conditions
• Communities without improved water/sanitation facing compounded environmental stress

This heterogeneity reveals that heat exposure creates and amplifies inequality from the earliest years of life—children already disadvantaged face the largest developmental losses when temperatures rise.

7. Policy Implications and Future Directions

These findings have immediate relevance for child protection and development policy in warming regions.

Policy Recommendations

• Recognize heat as a risk factor: Extreme heat should be included in child protection frameworks as a documented risk factor for maltreatment and developmental delays in climate-vulnerable populations.
• Target interventions strategically: Parenting support programs and early childhood interventions should be prioritized during hot periods and in heat-exposed communities, with additional resources for the most vulnerable households.
• Invest in adaptive infrastructure: Cooling infrastructure (e.g., community cooling centers, improved housing), water/sanitation systems, and green spaces can buffer vulnerable families from heat stress.
• Build climate-sensitive social services: Social service systems should account for seasonal and climate-driven fluctuations in child and family stress, with surge capacity during extreme heat events.
• Address inequality: Since heat impacts concentrate among the poor, urban dwellers, and those without basic services, climate adaptation must be equity-focused to avoid exacerbating existing disparities.

Future Research Directions

Building on this merged ERA5-Land × MICS dataset, several promising research directions emerge:

• Exposure decomposition: Contrast short heat shocks (e.g., prior month) vs cumulative exposure (birth→interview)
• Adaptation mechanisms: Stratify by wealth, urbanicity, water/sanitation to identify where adaptation policies are most needed
• Mediating pathways: Add proxies for caregiver stress or illness episodes; test if they mediate heat → aggression or heat → ECDI
• Place-based moderators: Merge land cover/greenness (MODIS) or nighttime lights (VIIRS) to measure urban heat islands and infrastructure buffers
• Policy simulations: Simulate cooling access and safe water expansions to estimate potential reductions in harsh discipline and ECDI losses

8. References

Data Sources

• UNICEF MICS: Multiple Indicator Cluster Surveys (2017-2020). Geocoded household and child-level microdata from Georgia, The Gambia, Madagascar, Malawi, Sierra Leone, and State of Palestine.
• ERA5-Land: ECMWF Reanalysis v5 Land (1950-present). Monthly aggregates of 2-meter maximum temperature at ~0.1° (~10km) spatial resolution.
• Climate baseline: 1991-2020 climatological normal used for anomaly calculations, following WMO standards.

Methods References

• UNICEF (2017). Discipline practices in the home. In MICS6 Tools.
• Loizillon, A., Petrowski, N., Britto, P., & Cappa, C. (2017). Development of the Early Childhood Development Index in MICS surveys. MICS Methodological Papers, No. 6. UNICEF.