THE SOLVABLE CHALLENGE OF AIR POLLUTION IN INDIA Anant Sudarshan, - - PowerPoint PPT Presentation

Anant Sudarshan, Michael Greenstone (University of Chicago), Santosh Harish (EPIC-India), Rohini Pande (Harvard University) Anant Sudarshan

THE SOLVABLE CHALLENGE OF AIR POLLUTION IN INDIA

India Policy Forum 2017, July 11 2017

OUTLINE

Air Quality Regulation: The Health Rationale Reviewing Command and Control

• Technology mandates
• Bans and rationing

Implications

• Collect data reliably and transparently
• Incentive-compatible and transparent regulation
• Rigorously evaluate pilots

Air Pollution and Health

STRONG EVIDENCE THAT PM REDUCES HUMAN LIFESPAN

Source: Chen et al (2013)

660 MILLION PEOPLE MAY LOSE 3.2 YEARS OF LIFE

• We use data from 456

monitors across 190 cities, combined with ground calibrated satellite estimates from Dey et al (2012) to produce the map on left.

• Drawing upon research

from across the world and China, we estimate life- expectancy drops by 1.1 to 3.2 life years because

• f non-compliance with

Satellite and Ground Monitoring Data Source: Greenstone et al (2015)

THE EVIDENCE SUGGESTS POLLUTION CAN BE REDUCED

Reviewing Command and Control

REVIEWING COMMAND-AND-CONTROL

Efficient environmental regulation requires i. High Quality Data ii. Low Costs and Incentive Compatible Design iii. Rigorous Evaluation

AIR POLLUTION LEVELS VARY SIGNIFICANTLY OVER TIME

100 200 300 400 500 PM2.5 conc (mg/ m3)

Jan round April round

LIMITED SPATIAL MONITORING…

…BUT LOTS OF SPATIAL VARIATION

100 200 300 400 PM2.5 conc. (mg/m3)

Jan 2016 April 2016

TECHNOLOGY DOES NOT ENSURE DATA AVAILABILITY

Eric Dodge and Rohini Pande, January 19 2016. http://www.indiaspend.com/cover-story/to-cut-delhis-air-pollution-pinpoint-the-source- 40763

MONITORING EMITTERS CAN BE EVEN HARDER

Duflo, E., Greenstone, M., Pande, R., and Ryan, N. (2013). Truth-telling by third-party auditors and the response of polluting firms: Experimental evidence from India. The Quarterly Journal of Economics, 128(4):1499–1545.

COMMON PROBLEM IN TRANSPORT SECTOR ALSO

• 79 percent emission testing centers in Mexico City accept

bribes and substitute emissions readings of failing cars (Oliva, 2015)

• Private centers in California fail vehicles at half the rate at

which government run centers (Wenzel, 1998)

• CPCB 2013 audit of pollution checking centers around

Delhi:

• Manpower poorly trained and unaware of testing

protocols

• Equipment not maintained, and rarely properly calibrated
• Software used to generate dummy measurements

CONTINUOUS EMISSIONS MONITORING SYSTEMS MIGHT HELP, BUT AGENT INCENTIVES STILL MATTER

CASE STUDY: DELHI’S ODD-EVEN PROGRAM

• Delhi government ran two rounds of the car-rationing

program

• January 1-15, 2016
• April 15- 30, 2016
• Ban on driving a car with an odd numbered license plate on
• dd dates, and an even numbered plate on even dates
• Ban in effect between 8am and 8pm with important

exceptions (two-wheelers, female drivers, taxis)

• Fines for non-compliance: Were these enforced?

DELHI TIME SERIES YIELDED MIXED REVIEWS…

CHANGES IN TRENDS WITHIN AND OUTSIDE DELHI

Jan January 19, 19, 2016 2016 May 13, 13, 2016 2016

DID THE ODD-EVEN PROGRAM REDUCE EMISSIONS?

• We compare changes in time trends for monitors within

and outside Delhi before, during and after the pilot

• Difference-in-Differences method uses outside Delhi

monitor trends as a control for all exogenous factors – meteorology, crop burning, wind direction – that influence the pollution time-series

• Especially in January, monitors within and outside Delhi

had similar trends before the pilot, diverged during the pilot, and reverted to being similar after the pilot

DID THE ODD-EVEN PROGRAM REDUCE EMISSIONS?

50 100 150 200 250 300

PM2.5 conc. mg/ m3

Non-Delhi Jan Delhi Jan Non Delhi- April Delhi- April

IMPACTS CLEAR IN JANUARY, NOT IN APRIL

• 140
• 120
• 100
• 80
• 60
• 40
• 20

20 40 60 1 2 3 4 5 6 7 8 9 101112131415161718192021222324 HOUR OF DAY January round April round

Change in concentrations during the dates when odd-even was implemented (microgram/m3)

ONE EXPLANATION: WEATHER WAS THE DIFFERENCE

S.K. Guttikunda, R. Goel, 2013. Health impacts of particulate pollution in a megacity— Delhi, India. Environmental Development 6 (2013) 8–20

• Dispersion is faster when

temperatures are higher

• Reduction in emissions may not

translate into reduction in concentrations that are identifiable

BUT COMPLIANCE MAY HAVE BEEN LOWER…

• In January, traffic surveys by the School of Planning and

Architecture, travel time queries from Google Maps and self- reported behavior all suggested high compliance (Kreindler 2016 survey of 960 commuters).

• In April, travel speeds (an indirect proxy related to traffic)

and self-reported behavior suggested compliance.

• BUT, surveys found traffic volumes were higher during the

second round of the program than the first round, and that there was a large shift to two-wheelers.

SIMILAR EXPERIENCES IN OTHER PARTS OF THE WORLD

Mexico City (Davis, 2008)

• Restrictions led to an increased adoption of used cars.
• Substitution to relatively older vehicles on restricted days

may have led to a net increase in pollution. Beijing (Wang et a 2014)

• Non-compliance may have been as high as 48 percent
• Car owners who traveled during peak hours and/or for

work trips, were more likely to break the driving restriction rules.

THE REGULATOR IS LEFT CHASING A MOVING TARGET

• Ban on brick kilns in Delhi,

led to movement just

• utside Delhi border
• Enforceable regulation, but

no change in polluting behavior and limited impacts

• High costs imposed on

industrial activity

• Challenge: Regulate

emissions not technology and minimize costs of compliance

S.K. Guttikunda, R. Goel (2013) “Health impacts of particulate pollution in a megacity—Delhi, India Environmental Development 6 8–20

TECHNOLOGY MANDATES- EASY TO ENFORCE, BUT DO NOT TRANSLATE TO REDUCTIONS IN EMISSIONS

• 1. Two emission sources attached

each with a cyclone and bag filter;

• 2. Two emission sources attached,

each with a cyclone and scrubber;

• 3. Two emission sources attached,

each with a cyclone, scrubber and bag filter;

• 4. A single boiler attached with a

cyclone and bag filter;

• 5. A single boiler attached to a

cyclone and scrubber. Source: CPCB survey of 1000 industries in Gujarat, Tamil Nadu, Maharashtra

500 1,000 1,500 1 2 3 4 5

Source: Baseline Survey

(Gujarat)

Box plot of PM Concentration by Stack Attachments

NOTWITHSTANDING DESIGN, REGULATION HAMSTRUNG BY INFLEXIBLE PENALTY OPTIONS AVAILABLE TO SPCBS

• Do not have the

ability to levy penalties commensurate with offence

• Very strict

penalties can be difficult to enforce, leading to effectively lax regulation

“The Value of Discretion in the Enforcement of Regulation: Experimental Evidence and Structural Estimates from Environmental Inspections in India” (with Esther Duflo, Michael Greenstone and Rohini Pande). NBER Working Paper #20590.

Reforming Environmental Regulation

THREE STEPS SEEM ESSENTIAL

1. Collect reliable and transparent data 2. Move towards incentive-compatible and efficient regulation 3. Pilot and evaluate impact of new policies

Two Important Examples

WHY EXPERIMENT WITH MARKET-BASED REGULATION?

• Lowers costs of

compliance

• Encourages monitoring

transparency and quality

• Incentivizes performance

beyond an absolute standard

• Introduces flexible,

financial penalties

Progra m Impact Estimate (cost savings / emissions reductions) Study US SO2 Trading Progra m U.S. $30m Carlson et al. (2000) U.S. $358m per year Ellerman et al. (2000) U.S. $153 - 183m per year Keohane (2006) EU ETS Progra m - EU power sector 88 Mton CO2 Delarue, Voorspools, D’haeseleer (2008) 34 Mton CO2 Delarue, Ellerman, D’haeseleer (2008)

ADAPTED FROM ANTHOFF AND HAHN (2010)

EMPIRICAL EVIDENCE MOTIVATING ETS PILOT IN SURAT

• For an emissions cap at the

100 mg/Nm3 level, we estimate a 55 percent reduction in costs for industries trading relative to fixed standards

• Gupta (2002) estimated that

emissions trading would be 2.69 times cheaper than using fixed standards to reduce TSP levels by 50% across 15 polluting industries in Maharashtra

Without Trading With Trading

MAHARASHTRA STAR RATING SCHEME

• Maharashtra is the most industrialized state of India with
• ver 1,00,000 industries:
• 12,500 high pollution potential
• 15,500 medium pollution potential
• 47,000 low pollution potential
• MPCB conducts extensive industry stack and ambient air
• monitoring. More than 20,000 stack samples collected

between September 2012 to January 2017

• The Star Rating Program aims to decrease air pollution by

increasing transparency and removing information asymmetries

7/13/2017 33

Visit: http://www.mpcb.gov.in/star_rating/browseRating

Thank You

Evaluating using a Difference-in-Differences Method

• We used regulatory monitors in the National Capital Region, inside and
• utside Delhi and ambient PM2.5 data from November 2015 to April 2016
• Ytm = α + β. 1(m є Delhi) + γ.1(t є oddeven)
• + δ.1(m є Delhi) X 1(t є oddeven) + λm + ηt + εtm
• Ytm is the particulate (PM2.5) concentration at time t (on hour h and day d) for

monitor m.

• Explanatory variables include an indicator variable for the treatment area (Delhi), an

indicator variable for the days that the odd-even program was in place (termed

• ddeven), and their interaction term.
• β and γ are the coefficients for the treatment area and period indicator variables.
• The interaction coefficient δ estimates the program impact on particulate

concentration.

• λm are ηt capture fixed effects at the monitor level and for each hour.

REGRESSION RESULTS

13% 13% red reduction in Ja January No

• st

statisti tically ly si significant red reductio ion in n Apr pril il