1. Pollution and Climate Change ALLISON LARR, COLUMBIA MATTHEW NEIDELL, COLUMBIA AND NBER

2. Climate change  air quality  child well-being Climate change and ozone ◦Ozone = f(VOC, NOx, temperature, sunlight) ◦If unabated: higher temperatures  higher ozone Climate change mitigation and pollution (co-benefits) ◦Fossil fuels ◦CO2 emissions ◦Criteria pollutants (PM 2.5, SO 2, NOx) ◦If abated: less fossil fuels  less PM 2.5 and ozone ◦E.g. Clean power plan: reduce SO 2 and NOx Air pollution and well being ◦Early years: birth weight, gestational length, infant mortality ◦Later years ◦Respiratory diseases. E.g., asthma ◦Performance in school, labor market

3. Outline 1. Review studies on climate and air quality ◦Overview of AQ projections ◦Studies that project both: ◦PM 2.5 and ozone ◦BAU and mitigation 2. Review QE studies on PM 2.5 /ozone and well-being ◦Pollution not randomly assigned. E.g., sorting ◦QE studies limit OVB 3. Combine 1 & 2 above to project future changes in well-being ◦Fraught with heroic assumptions ◦Need evidence in light of uncertainty 4. Comment on other (developing) nations

4. AQ modeling: 4 steps 1. Carbon emissions: IPCC scenarios (A1, A2, B1, B2) ◦Fossil fuel reliance ◦Economic and population growth ◦Technological progress 2. Climate change projections ◦General circulation model (GCM) of earth’s climate: temperature, precipitation, etc. ◦Results on coarse spatial resolution, ignore local conditions. E.g., 100 km 2, topography 3. Downscaling ◦Modify spatial resolution: country, state, city, etc. ◦Provide local weather 4. AQ models: Community Multi-scale Air Quality (CMAQ) model ◦Simulates chemical and physical processes involved in atmospheric chemical transport ◦Combine emissions & weather  local pollution

5. Table 1. Ozone and PM2.5 projection model scenarios

6. How does AQ affect children? Health effects ◦Ozone: lung irritant  shortness of breath, lung inflammation, asthma exacerbation ◦PM 2.5 : travels though lungs into bloodstream ◦Respiratory effects ◦Cardiovascular effects (heart attacks, blood pressure, etc.) Human capital effects ◦Indirectly via health effects ◦Direct effects: PM 2.5 travels up olfactory nerves to brain ◦Latent effects via epigenetic changes (FOH)

7. Quasi-experimental approach Major concern: sorting ◦AQ higher in more productive areas (e.g., cities)  wealthier families live in dirtier areas ◦AQ capitalized into housing prices  wealthier families live in cleaner areas Two approaches ◦Policy event ◦1980-82 recession ◦CAAA as an instrument ◦Military relocation ◦Area fixed effects ◦People sort on average pollution ◦SR variation within area exogenous Focus on QE ozone and PM (TSP, PM 10, PM 2.5 ) studies

9. Projections Focus on Tagaris et al. (2007) ◦Baseline (2001), BAU (2050), and mitigation (2050) ◦Same model for PM 2.5 and ozone ◦Regional projections ◦Smaller unabated ozone projections than others: 1 vs. 5 ppb Projections for: ◦PM 2.5 and infant mortality ◦PM 2.5 and adult earnings ◦Ozone and hospitalizations

10. Table 2. Ozone and PM2.5 Projections by Region

11. Infant mortality and PM 2.5 projections 1. Number of births by region: NVSS (2012), assume constant 2. δIM/δPM from: ◦Chay and Greenstone: 5.5 per 100k ◦Knittel et al.: 18 per 100k 3. Conversion to PM 2.5 ◦TSP to PM 2.5 : 0.16  CG 34 per 100k ◦PM 10 to PM 2.5 : 0.54  KMS 33 per 100k 4. δPM 2.5 /δCC from Tagaris et al. 5. Impact = 1 X 2 X 3 X 4

12. Table 3A. Infant mortality and contemporaneous PM 2.5

13. Adult earnings and PM 2.5 projections 1. Per capita income by region (BEA REIS) 2. δearnings/δPM: 1.1% for 10 unit TSP (Isen et al.) 3. TSP to PM 2.5 : 0.68% for 1 unit PM 2.5 4. δPM 2.5 /δCC from Tagaris et al. 5. Impact = 1 X 2 X 3 X 4

14. Table 3B. Adult earnings and early childhood exposure to PM2.5

15. Hospitalizations and ozone projections 1. δhosp/δozone from: ◦Lleras-Muney ◦1 SD increase  8-23% decrease; 15.8% preferred ◦1 SD =.008 ppm ◦1 ppb ozone  1.97% change in hosp. ◦Beatty & Shimshack ◦10% ozone increase  2.5-3.3% increase in hospitalizations; choose lowest ◦10% = 2.43 ppb ◦1 ppb ozone  1.03% change in hosp. 2. δozone/δCC from Tagaris et al. 3. Impact = 1 X 2 (No good baseline of hospitalizations by region)

16. Table 3C. Respiratory hospitalizations and contemporaneous ozone

17. Other countries Comparable wealth to US: comparable effects LMIC could be different ◦More rapid development, higher levels of pollution ◦Past US pollution like current pollution

18. Figure 1. Trends in Air Pollution for US, China, and Mexico China Pittsburgh Mexico

19. Other countries Comparable wealth to US: comparable effects LMIC could be different ◦More rapid development, higher levels of pollution ◦Past US pollution like current pollution ◦Arceo-Gomez et al: dose-response comparable (Mexico vs. US) ◦Non-linear relationship (log(y)): larger effects ◦Often equatorial  larger temperature increases ◦If unabated, ozone increases likely bigger ◦Less likely to mitigate. E.g., Kyoto ◦No co-benefits Suggest pollution via CC will have larger effects in LMIC countries

20. Conclusion Unabated climate change  minimal effects on child well-being (relative to baseline) Mitigating climate change  large “co-benefits” (relative to BAU and baseline) ◦Reduced infant mortality ◦Reduced respiratory illnesses ◦Improved test scores ◦Increased earnings Many unknowns involved in projections ◦Suggest immediate benefits from mitigation policies ◦Not for geoengineering (CCS, solar radiation management) ◦Useful starting point