The Largest Low Emission Zone of Europe: Traffic and Air Quality in London

Size: px

Start display at page:

Download "The Largest Low Emission Zone of Europe: Traffic and Air Quality in London"

Grace Norman
5 years ago
Views:

1 The Largest Low Emission Zone of Europe: Traffic and Air Quality in London Hendrik Wolff and Muxin Zhai University of Washington November 23, 2014 H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

2 Table of contents 1 Motivation and Background 2 Data 3 Model 4 Results 5 Robustness Checks and Supportive Evidence H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

3 Table of Contents Motivation and Background 1 Motivation and Background 2 Data 3 Model 4 Results 5 Robustness Checks and Supportive Evidence H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

4 Why do we care? Motivation and Background H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

5 Why do we care? Motivation and Background Exposure to air pollution accounted for 7 million deaths worldwide in 2012, including almost 600,000 in the WHO European Region. (WHO report, March 24, 2014) In Europe alone, the direct costs to society from air pollution amount to about e23 billion per year, and the external costs from health impacts are estimated at e billion (3-9% of EU GDP). (The European Commission, EUROPA, 2014) H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

6 Motivation and Background Why do we care? Exposure to air pollution accounted for 7 million deaths worldwide in 2012, including almost 600,000 in the WHO European Region. (WHO report, March 24, 2014) In Europe alone, the direct costs to society from air pollution amount to about e23 billion per year, and the external costs from health impacts are estimated at e billion (3-9% of EU GDP). (The European Commission, EUROPA, 2014) H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

7 Motivation and Background Traffic Regulation Direct regulation: Command-and-control License plate program Davis, L. (2008 JPE): The Effect of Driving Restrictions on Air Quality in Mexico City Chen, Y., Ginger, Z., Kumar, N., Shi, G. (2013 JEEM): The Promise of Beijing: Evaluating the Impact of the 2008 Olympic Games on Air Quality Low emission zone Wolff, H. (forthcoming): Keep Your Clunker in the Suburb: Low Emission Zones and Adoption of Green Vehicles Indirect regulation: Economic-incentive Congestion charge zone Leape, J. (2006 JEP): The London Congestion Charge Low emission zone: London H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

8 This paper... Motivation and Background uses a unique dataset on air quality and weather conditions throughout England is the first empirical analysis on the world largest low emission zone - the London LEZ is the first empirical research to exploit the spatial heterogeneity on both: the type of vehicles (i.e. Heavy vs. Light truck) WHERE these vehicles drive (road stations vs. industrial stations) H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

9 Motivation and Background London Low Emission Zone: Map H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

10 Motivation and Background London Low Emission Zone: Timeline Announcement May 3, 2007 H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

11 Motivation and Background London Low Emission Zone: Timeline 200 HGV, truck Announcement May 3, 2007 Feb 8, 2008 Phase I H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

12 Motivation and Background London Low Emission Zone: Timeline 200 HGV, truck 100 LGV Announcement Phase I Phase II May 3, 2007 Feb 8, 2008 Jul 7, 2008 H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

13 Motivation and Background London Low Emission Zone: Timeline 200 HGV, truck 100 LGV 100 Larger van, minibus Announcement Phase I Phase II Phase III May 3, 2007 Feb 8, 2008 Jul 7, 2008 Oct 4, 2010 H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

14 Motivation and Background London Low Emission Zone: Timeline 200 HGV, truck 100 LGV 100 Larger van, minibus 200 Bus, coach, HGV Announcement Phase I Phase II Phase III Phase IV May 3, 2007 Feb 8, 2008 Jul 7, 2008 Oct 4, 2010 Jan 3, 2012 H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

Motivation and Background London Low Emission Zone: Timeline 200 HGV, truck 100 LGV 100 Larger van, minibus 200 Bus, coach, HGV Announcement Phase I

15 Motivation and Background London Low Emission Zone: Timeline 200 HGV, truck 100 LGV 100 Larger van, minibus 200 Bus, coach, HGV Announcement Phase I Phase II Phase III Phase IV May 3, 2007 Feb 8, 2008 Jul 7, 2008 Oct 4, 2010 Jan 3, 2012 EU III EU IV H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

16 Motivation and Background London Low Emission Zone: Subject vehicles Vehicle composition in Great Britain 2007 Bus & coach 9% 4% 2% 1% Car 2% HGV LGV Motorcycle 83% Other H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

17 Motivation and Background London Low Emission Zone: Signs Sourses: Traffic for London H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

18 Table of Contents Data 1 Motivation and Background 2 Data 3 Model 4 Results 5 Robustness Checks and Supportive Evidence H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

19 Data Data Sources Airborne particulate matter (PM10) data Data Source: London Air Quality Network (LAQN) Air Quality England (AQE) Number of stations: 58 Data time range: , daily level Weather data Data source: WeatherSpark We match each LAQN and AQE station with the nearest WeatherSpark station H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

20 Station Types Data Each LAQN or AQE station has a classification according to its relative distance to major sources of pollution. H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

21 Station Types Data Each LAQN or AQE station has a classification according to its relative distance to major sources of pollution. Roadside station H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

22 Station Types Data Each LAQN or AQE station has a classification according to its relative distance to major sources of pollution. Roadside station Industrial station H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

Roadside station Industrial station Urban

23 Station Types Data Each LAQN or AQE station has a classification according to its relative distance to major sources of pollution. Roadside station Industrial station Urban background station H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

24 Data Station Locations: LAQN and AQE!.!. XY #0 #0 XY XY XY #0!. XY XY!. X X!XX X XX!!! ## #!!!! # # kg! G kg Station locations in southern England! Treated - Roadside!. Control - Roadside k Outside - Roadside X XY G Treated - Urban Background Control - Urban Background Outside - Urban Background # Treated - Industrial #0 Control - Industrial H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

25 Data Station Locations: LAQN and AQE!!!!!!!!!!!!!!!!!!!!!! X X X X X X X X X X X # # # # # Station locations within Greater London # Industrial X Urban Background! Roadside H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

26 Data Summary Statistics Table 3: Summary statistics Variable Mean Std. Dev. Min. Max. N Air pollutant PM10 (µg/m 3 ) Station groups Treated (1=treated) Outside (1=outside) Control (1=control) Station Types Roadside (1=roadside) Industrial (1=industrial) Urban background (1=urban) Weather conditions Humidity (%) Hours of rain (hours) Light rain (1=light rain) Moderate rain (1=mod. rain) Heavy rain (1=heavy rain) Air pressure (mbar) Temperature ( o C) Average wind speed (m/s) Maximum wind speed (m/s) Other covariates H.Wolff, Weekend M.Zhai (UW) (1=weekend) 0.29London LEZ November , / 34

27 Table of Contents Model 1 Motivation and Background 2 Data 3 Model 4 Results 5 Robustness Checks and Supportive Evidence H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

28 Model Difference-in-Differences Approach Main specification: τ=3 log(pm10 it ) = τ=0 τ=3 β τ P τ + τ=3 γ τ P τ T r i + δ τ P τ Out i (1) τ=0 τ=0 + θ i + f (λ t ) + X it Γ + ε it P τ is the indicator for LEZ phases (0=announcement, 1=phase I, 2=phase II, 3=phase III) Tr i and Out i are time-invariant group indicators for treated and outside stations θ i captures unobserved station specific fixed effects (FE) f (λ t ) captures year fixed effects (FE) and month fixed effects (FE) X it is a covariate matrix ε it is robust standard errors clustered at station level H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

29 Table of Contents Results 1 Motivation and Background 2 Data 3 Model 4 Results 5 Robustness Checks and Supportive Evidence H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

30 Results Graphical Representation: PM10 Roadside stations Deseasonalized PM Pre-announcement Announcement Phase II Phase I Phase III 24 01jan jul jan jul jan jul jan jul jan jul jan jul jan2011 Treated stations Outside stations Control stations H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

31 Results Graphical Representation: PM10 Roadside stations Deseasonalized PM Pre-announcement Announcement Phase II Phase I Phase III 24 01jan jul jan jul jan jul jan jul jan jul jan jul jan2011 Treated stations Outside stations Control stations H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

32 Results Graphical Representation: PM10 Roadside stations Deseasonalized PM Pre-announcement Announcement Phase II Phase I Phase III 24 01jan jul jan jul jan jul jan jul jan jul jan jul jan2011 Treated stations Outside stations Control stations H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

33 Results Graphical Representation: PM10 Industrial stations 60 Pre-announcement Announcement Phase I Phase II Phase III Deseasonalized PM jan jul jan jul jan jul jan jul jan jul jan jul jan2011 Treated stations Control stations H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

34 Results Graphical Representation: PM10 Industrial stations 60 Pre-announcement Announcement Phase I Phase II Phase III Deseasonalized PM jan jul jan jul jan jul jan jul jan jul jan jul jan2011 Treated stations Control stations H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

35 Results Graphical Representation: PM10 Industrial stations 60 Pre-announcement Announcement Phase I Phase II Phase III Deseasonalized PM EU daily limit EU annual average limit 10 01jan jul jan jul jan jul jan jul jan jul jan jul jan2011 Treated stations Control stations H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

36 Results Estimation Results: Pooled Data Table 4: Difference-in-Difference Estimation Results (1) (2) (3) (4) (5) (6) Treatment effects LEZ announcement treated (0.042) (0.042) (0.042) (0.041) (0.041) (0.041) LEZ phase I treated (0.038) (0.035) (0.035) (0.033) (0.033) (0.033) LEZ phase II treated (0.026) (0.027) (0.027) (0.027) (0.027) LEZ phase III treated (0.040) (0.036) (0.036) (0.036) Spillover trends LEZ announcement outside (0.041) (0.041) (0.041) (0.040) (0.040) (0.040) LEZ phase I outside (0.046) (0.035) (0.035) (0.034) (0.034) (0.034) LEZ phase II outside (0.039) (0.039) (0.039) (0.039) (0.039) LEZ phase III outside (0.043) (0.039) (0.038) (0.039) Time trends LEZ announcement (0.039) (0.039) (0.039) (0.039) (0.039) (0.039) H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

37 Results Estimation Results: Spatial Heterogeneity Table 5: Spatial Heterogeneity of LEZ s effects (1) (2) (3) Roadside Industrial Urban Background Treatment effects LEZ announcement treated (0.031) (0.136) (0.062) LEZ phase I treated (0.044) (0.076) (0.060) LEZ phase II treated (0.022) (0.063) (0.052) LEZ phase III treated (0.052) (0.077) (0.052) Spillover effects LEZ announcement outside (0.028) (0.059) LEZ phase I outside (0.043) (0.056) LEZ phase II outside (0.024) (0.050) LEZ phase III outside (0.051) (0.057) Time trends LEZ announcement H.Wolff, M.Zhai (UW) London (0.024) LEZ (0.124) (0.058) November 23, / 34

38 Results Graphical Representation: Compliance Non-compliant vehicles Phase I Phase II Deseasonalized PM10 01jun aug oct dec feb apr jun aug2008 date 01oct dec feb apr jun aug oct2009 Phase I noncompliant vehicles Deseasonalized PM10 Phase II concompliant vehicles H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

39 Results Effect of Compliance on PM10 Table 9: Effect of compliance to phase I and II on PM10 Roadside stations Industrial stations (1) (2) (3) (4) (5) Inside Nearby Far Inside Far Panel A: Non-compliance to phase I Non-compliant vehicles to phase I (0.002) (0.006) (0.006) (0.008) (0.010) Observations R Panel B: Compliance rate of phase I Compliance rate of phase I (0.105) (0.293) (0.317) (0.452) (0.513) Observations R Panel A-0: Subject vehicles to phase I Subject vehicles to phase I (0.001) (0.002) (0.001) (0.002) (0.006) Stations H.Wolff, M.Zhai (UW) London 22 LEZ 2 5 November 5 23, / 34

40 Panel A: Non-compliance to phase I Non-compliant vehicles to phase I H.Wolff, M.Zhai (UW) London LEZ November 23, / 34 Results Effect of Compliance on PM10 Table 9: Effect of compliance to phase I and II on PM10 Roadside stations Industrial stations (1) (2) (3) (4) (5) Inside Nearby Far Inside Far Panel C: Non-compliance to phase II Non-compliant vehicles to phase II (0.001) (0.004) (0.004) (0.006) (0.008) Observations R Panel D: Compliance rate to phase II Compliance rate of phase II (0.055) (0.157) (0.160) (0.283) (0.309) (0.321) (0.164) (1.845) (1.157) (4.498) Observations R

41 Robustness Checks and Supportive Evidence Table of Contents 1 Motivation and Background 2 Data 3 Model 4 Results 5 Robustness Checks and Supportive Evidence H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

42 Robustness Checks and Supportive Evidence Robustness Checks over Standard time. We errors resolve this issue by clustering standard errors at stationyear level Baseline while- clustered maintaining standard the station errors atspecific station level unobserved fixed effects. This modification Alternatives -allows clustered autocorrelation standard errors within at station-year for each level, station type-region and assumes level, independence and newey-west in random standard shocks errorsacross year and station. Table 7 column Sample (2) observations presents our estimation results using full sample. We find that the significance levels of our estimates stay roughly the same with baseline Baseline - full sample specification. Alternatives - exclude likely outliers that exceed 95 percentile of station-level PM 10 distribution Table 7: Robustness Checks, All Types of Stations Full sample Short sample ( 95 percentile) (1) (2) (3) (4) (5) (6) (7) (8) Baseline S.E. clust. S.E. clust. Neweywest Baseline S.E. clust. S.E. clust. Newey- by stationyearegioyearegion by type- s.e by station- by typewest s.e Treatment effects LEZ announcement treated (0.041) (0.033) (0.034) (0.018) (0.042) (0.032) (0.036) (0.017) LEZ phase I treated (0.033) (0.035) (0.028) (0.026) (0.032) (0.034) (0.028) (0.023) LEZ phase II treated (0.027) (0.023) (0.027) (0.022) (0.025) (0.021) (0.028) (0.020) LEZ phase III treated (0.036) (0.034) (0.039) (0.027) (0.035) (0.032) (0.040) (0.025) Spillover effects LEZ announcement H.Wolff, M.Zhai outside (UW) (0.040) (0.037) (0.036) London (0.026) LEZ (0.041) (0.035) November (0.035) 23, 2014 (0.024) 29 / 34

43 Robustness Checks and Supportive Evidence e if our conclusions are valid, we use urban background stations inside LEZ s control Robustness and roadside checks and industrial stations inside LEZ as treated stations, ssuming that urban and residential areas inside LEZ are not affected by the gulation. Control We do stations not estimate the spillovers of LEZ in this construction due data limitation. Baseline - use stations located in other areas of England as controls Alternatives - use urban background stations located inside LEZ as controls Table 9: Alternative difference-in-differences estimation Overall Roadside Industrial (1) (2) (3) (4) (5) (6) Treatment effects LEZ announcement treated (0.027) (0.027) (0.027) (0.027) (0.025) (0.074) LEZ phase I treated (0.039) (0.037) (0.037) (0.036) (0.037) (0.057) LEZ phase II treated (0.030) (0.030) (0.030) (0.029) (0.028) (0.054) LEZ phase III treated (0.027) (0.027) (0.027) (0.027) (0.029) (0.048) Time trends LEZ announcement (0.022) (0.022) (0.023) (0.023) (0.020) (0.032) LEZ phase I (0.034) (0.028) (0.028) (0.028) (0.028) (0.020) LEZH.Wolff, phase M.Zhai II (UW) London LEZ November , / 34

44 Robustness Checks and Supportive Evidence Supportive Evidence: NO2 70 Pre-announcement Announcement Phase II Phase I Phase III Deseasonalized NO jul jul jul jul jul jul2010 Treated stations Outside stations Control stations H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

45 H.Wolff, M.Zhai (UW) Figure 9: Annual London trafficlez flow on major roads November 23, / 34 Robustness Checks and Supportive Evidence cars and taxis on all major roads 21 in each region. The base year is 2008, the year that both phase I and phase II of LEZ are launched. The traffic flow in each year is then divided by the traffic flow in year 2008 to get the percentage change relative to the base year for each region. Traffic Flow: London vs. Other Regions Panel A: Heavy goods vehicles 1.05 Panel B: Light goods vehicles 1.1 Traffic Index 1.95 Traffic Index Panel C: Buses1.06 and coaches Panel D: Cars and taxis Traffic Index Traffic Index Traffic Index Year London South East North West East of England Yorkshire and The Humber

46 Robustness Checks and Supportive Evidence Fuel Consumption: London By type of fuel By type of vehicles Fuel consumption in thousand tonnes in London Year Diesel Petrol Annual diesel consumption in thousand tonnes Year Heavy good vehicles (HGVs) Diesel light good vehicles (LGVs) Buses Diesel cars H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

47 Robustness Checks and Supportive Evidence The end H.Wolff, M.Zhai (UW) London LEZ November 23, / 34

The London Low Emission Zone. Nick Fairholme Head of the LEZ Transport for London

The London Low Emission Zone. Nick Fairholme Head of the LEZ Transport for London The London Low Emission Zone Nick Fairholme Head of the LEZ Transport for London An introduction to the LEZ. Air Quality in London Air pollution is a serious problem in London - 1,000 premature deaths