The Local Benefits of Global Air Pollution Control in Mexico City - Final Report of the Second Phase of the Integrated Environmental Strategies Program in Mexico ppt

During this project, the following goals have been achieved: • Estimate cost savings due to health improvements related to air pollution reductions occurring simultaneously with greenhou

Instituto Nacional de Ecología, México Instituto Nacional de Salud Publica, México

August 2003

Final Report of the Second Phase of the

Integrated Environmental Strategies

Program in MexicoThe Local Benefits of Global Air Pollution Control in Mexico City

Table of Contents

III Emission Reductions and Costs

III.2 Renovation of the taxi fleet Hojer III.3 Extension of the Metro Osnaya

III.5 Measures to reduce leaks of Liquefied Petroleum Gas McKinley III.6 Co-generation Laguna

IV Air quality modeling McKinley and Iniestra

V Health impacts analysis Zuk with Avalos, Martínez, Hernández,

González, Reynales and Valdés

VI Valuation Zuk with Avalos, Martínez, Hernández,

González, Reynales and Valdés VII Integration: The Co-Benefits model McKinley and Zuk

Appendix A Air Quality Modeling McKinley and Iniestra Appendix B Capacity Building Zuk and McKinley

Appendix C Basic User’s Guide for the Co-Benefits Model McKinley and Zuk

Contact Information:

Consultants to Instituto Nacional de Ecología:

Galen McKinley galen@alum.mit.eduMiriam Zuk mzuk@alum.mit.eduMorten Hojer mhoejer@hotmail.com

Instituto Nacional de Ecología:

Julia Martínez jmartine@ine.gob.mx Montserrat Avalos moavalos@ine.gob.mxIsabel González igmerino@ine.gob.mx Rodolfo Iniestra riniestr@ine.gob.mx Miguel Ángel Martínez mmartine@ine.gob.mxIsrael Laguna ilaguna@ine.gob.mx Patricia Osnaya posnaya@ine.gob.mx

Instituto Nacional de Salud Publica:

Mauricio Hernández mhernan@correo.insp.mx Luz Miriam Reynales lreyns@yahoo.com Raydel Valdes rayvs@insp.mx

Chapter I Executive Summary

From September 2002 to August 2003, the Second Phase of the Integrated Environmental Strategies Program in Mexico was undertaken at the Instituto Nacional de Ecología (INE; National Institute of Ecology) of Mexico In this report, activities and findings are summarized During this project, the following goals have been achieved:

• Estimate cost savings due to health improvements related to air pollution reductions occurring simultaneously with greenhouse gas (GHG) emissions reductions,

• Compare costs and benefits for the specific policy measures,

• Build capacity in the Mexican government for integrated, quantitative environmental and economic assessment, and

• Provide results and tools with relevance to emission control decision- making process in

Mexico City

We produce estimates of annualized reductions of emissions of local and global air pollutants and program costs for three transportation measures (taxi fleet renovation, metro expansion, and hybrid buses), one residential measure to reduce leaks of liquefied petroleum gas (LPG) from stoves, and one industrial measure for cogeneration for the periods 2003-2010 and 2003-2020 at several discount rates Using reduced-form air quality modeling techniques, the impacts of changed emissions on exposure are calculated Then using dose-response methodology, public health improvements due to reduced exposure are estimated Finally, various valuation metrics are applied to determine the monetized health benefits to society of the control measure

We find that the 5 measures considered in this study could reduce annualized exposure to particulate air pollution by 1% and to maximum daily ozone by 3%, and also reduce greenho use gas emissions by 2% (more than 300,000 tons C equivalent per year) for both the time periods We estimate that for both time horizons, over 4400 quality-adjusted life-years (QALYs) per year could be saved, with monetized public health benefits on the order

of $200 million USD per year In contrast, total costs are under $70 million USD per year The mean cost per QALY is estimated to be under $40,000 for the 5 measures Of the measures considered, transportation measures are most promising for simultaneous reductions of both local and global pollution in Mexico City

This analysis has been integrated in to an user-friendly modeling tool using Analytica software The Co-Benefits Model has been made available to decision- makers and their staffs in Mexico City There is interest from these groups in applying the model to their work and in modifying it for use in other regions of Mexico, particularly the City of Toluca

in the State of Mexico

Capacity building has been a major part of this project A large group of INE staff have actively contributed to the research effort Regular meetings and training sessions have been held with members of the Metropolitan Environmental Commission (CAM) and other environmental agencies in the Mexico City These meetings have encouraged active

participation in this project and aided the integration of this work with other air pollution control efforts in the region

Chapter II Project Summary

II.1 Introduction

Due to complex socio-political, economic and geographical realities, Mexico City suffersfrom one of the worst air pollution problems in the world Greenhouse gas emissions from the City are also substantial In this study, we compare the costs and benefits of a set of politically- relevant air pollution control measures for the City and simultaneously consider the greenhouse gas emission impacts of these measures We find that with 5 control measures, it would be possible to reduce annualized exposure to particulate air pollution by 1% and to peak ozone by 3%, and also to reduce greenhouse gas emissions by 2% (more than 300,000 tons C equivalent per year) for the time periods 2003-2010 and 2003-2020

We estimate that for both time horizons, over 4400 quality-adjusted life- years (QALYs) per year could be saved, with monetized public health benefits on the order of $200 million USD per year In contrast, total costs are under $70 million USD per year The mean cost per QALY is estimated to be under $40,000 for the 5 measures We find that transportation measures are likely to be the most promising for simultaneous reductions of both local and global pollution in Mexico City

II.2 Motivation

With nearly 20 million inhabitants, 3.5 million vehicles, and 35,000 industries, Mexico City consumes more than 40 million liters of fuel each day It is also located in a closed basin with a mean altitude of 2240m The combination of these and other factors has led to a serious air quality problem In 2002, Mexico City air quality exceeded local standards for ozone (110 ppb for 1 hour) on 80% of the days of the year Particulate 24- hour standards were exceeded on 5% of the days (SMA, 2002)

Greenhouse gas (GHG) emissions from Mexico City are also significant In 1998, Mexico ranked as the 13th largest GHG producing nation Mexico City emits approximately 13% of the national total (Sheinbaum et al., 2000) Using a 3.3% annual growth rate (West et al., 2003) and a 1996 base year estimate of 45,585,000 tons of CO2 (Sheinbaum et al., 2000),

we estimate that the annualized GHG emission of Mexico City for the period 2003-2010 and 2003-2020 will be 17 million tons of C equivalent per year and 20 million tons C equivalent per year, respectively

As emissions of GHG and local air pollutants are often generated from the same sources, there may exist opportunities for their joint control In this study, we have developed a cost-benefit analysis framework to analyze the trade-offs between costs, public health benefits, and GHG emission reductions for a select set of control measures In an effort to disseminate the knowledge collected in this work, we have also created a reduced- form analysis tool for use by policy makers

This study fits into an ongoing process of analysis and action regarding Mexico City air

its third air quality management plan The first plan, PICCA (Programa Integral para el Control de la Contaminación Atmosférica) was initiated in 1990 and had several major accomplishments, including the introduction of two way catalytic converters, the phase out

of leaded gasoline, and establishment of vehicle emissions standards The second program, PROAIRE (Programa para Mejorar la Calidad del Aire en el Valle de México 1995-2000) achieved the introduction of MTBE, restrictions on the aromatic content of fuels and reduction of sulfur content in industrial fuel While significant improvements in ambient air quality have improved, levels remain dangerously high, therefore the government has recently initiated the third plan, PROAIRE 2002-2010, as an extension of previous plans PROAIRE 2002-2010 includes 89 control measures targeting emissions reductions from mobile, point and area sources, as well as proposing education and institutional strengthening measures to combat the air pollution that afflicts the city While some of these measures are slowly being implemented, little quantitative analysis has been done prior to designing this plan Decision makers are now faced with the difficulty in setting priorities when presented with a such a large range of control options Several studies are currently quantitatively analyzing these issue (Molina et al., 2002) A recent study by West

et al (2003) aimed to analyze a large number of PROAIRE and climate change control measures to determine the least cost set of options for joint control This study builds on these works, by simplifying and integrating the analysis to provide real time answers to policy makers

II.3 Methodology

Emissions Reductions and Costs for Specific Control Measures

We estimate the time profiles of local pollutant (PM10, SO2, CO, NOx, and HC) and global pollutant (CO2, CH4, and N2O) emission reductions, and costs for 5 control measures that address transportation, residential and industrial emission sources We estimate emissions reductions and costs for each year from 2003 to 2020 such that the different time-profiles of the programs’ costs and impacts can be studied These two time horizons were chosen to allow us to analyze the short term on the time frame of the plan itself, and a longer term analysis on the scale of the project implementation For incorporation into the cost – benefit analysis, results are annualized using several discount rates In this Project Summary, we present results using a 5% discount rate only

Below, key aspects of the control measures analyzed in this study are outlined In Tables II.1 and II.2, the estimated emissions reductions and costs of these measures are presented

Taxi fleet renovation

• 80% of old taxis are replaced by 2010

• Fuel efficiency increases from 6.7 km/L to 9 km/L

• Tier I technology is assumed in 1999 and newer models

• Changes in emissions of primary particulate matter are not estimated

Metro expansion

• 76 km of new construction by 2020 (5 km between 2003 and 2010, 71 km from 2011 to 2020)

• Riders assumed to come from microbuses and combis

• Recuperation value of capital is included, using a 30 year useful life

• Stove maintenance is performed in 1 million households to eliminate leaks

• This is a combination of 4 measures that each address a specific part of LPG stove systems (TUV, 2000)

Cogeneration

• Installation of 160 MW of capacity by 2010

• Recuperation value of capital is included, using a 20 year useful life

Table II.1 Annualized emissions reductions (tons / year)

Table II.2 Annualized abatement costs (2003 million US$ / year)

Control Measure Public Investment Private Investment Fuel, Operations,

Maintenance Total Cost Time Horizon 2003-2010

changes occurring with changes in hydrocarbon and NOx emissions

In order to account for the spatial relationship of population and pollution concentrations,

as well as to account for annual exposures, we use reduced form models to provide a reduction fraction (RF) of pollutant concentration (Cesar et al., 2002; USEPA, 1999) This reduction fraction is then multiplied by projected population-weighted concentrations for the appropriate time horizon These projected concentrations use as a baseline the mean 1995-1999 observed, population-weighted (1995 census) 24- hour mean PM10 (64.06 ug/m3)

or O3 maximum concentration (0.114 ppm), from Cesar et al (2000) The projection to future population-weighted concentrations is achieved by a linear interpolation of mean concentration results from the Multiscale Climate Chemistry Model (MCCM) model for

1998 and 2010 based on the emissions inventory for 1998 and emissions inventory projection for 2010 of the CAM (PROAIRE, 2002; Salcido et al 2001)

To estimate changes in PM10 concentrations, the chemical species in the observed particulate matter are attributed to primary pollutants based on chemical analyses of the composition of particulate matter in the MCMA (Chow et al 2002) Fractional changes in the emission inventories of primary pollutants can then be related to fractional reductions in particulate concentrations Results of chemical analyses of the composition of particulate matter from 6 sampling sites during the IMADA campaign of March 1997 (Chow et al 2002) are averaged, with weighting based on the total mass of each sample In order to attribute organic carbon to its primary (combustion) and secondary (hydrocarbon) sources, observed organic carbon is disaggregated into its primary and secondary contributions Following Turpin et al (1991), we estimate the primary organic contribution to total organic carbon based on a fixed ratio to elemental carbon mass of 1.9, a mean value for the

Los Angeles basin The mass of secondary organic carbon is then the difference of the total organic carbon mass and the mass of primary organic carbon Total primary particulate mass from combustion sources (25%) is the sum of primary organic and elemental carbon

Secondary organic carbon mass (2%) is attributed to hydrocarbon emissions Additionally, the mass of particles associated with geological sources (45%) is attributed to primary PM10emissions from geologic sources; the mass of particles associated with total particulate ammonium nitrate (7%) is attributed to NOx emissions; and the mass of particles associated ammonium sulfate (11%) is attributed to SO2 emissions

The peak mean O3 reduction fraction (RO3max) is estimated from the fractional reductions

in hydrocarbon (RHC) and NOx (RNOx) by:

RO3max = 0.5353*RNOx - 0.2082*(RNOx)2 + 0.1112*RHC This equation is derived from a series of runs of the MCCM for Mexico City (Salcido et al., 2001) where HC and NOx emissions are varied in equal proportion from all sources and O3

concentration changes were recorded The above equation results from a polynomial regression fit to the results of Salcido et al (2001)

These reduced- form air quality modeling approaches are limited by the still large uncertainty about fundamental processes responsible for ozone and particulate formation in the Mexico City Valley Further, the approaches have uncertainty due to the lack of spatial and temporal resolution and imperfections in the modeling and measurement techniques on which the approaches are based An exact quantification of the uncertainty is beyond the scope of this analysis Based on the work of Cohen et al (2003) and comparisons made during this study, we make a conservative estimate of 30% uncertainty on primary particulate results, and 50% uncertainty on the secondary particulate and maximum ozone results

In Table II.3, concentration change estimates based on Source Apportionment and the Ozone Isopleth methods are shown for each of the control measures

Table II.3 Annual particulate and maximum ozone exposure changes (ìg/m 3 )

Particulates (PM 10 ) Maximum Daily O 3

Health Impacts Analysis

Results from epidemiological studies are used to estimate avoided cases of mortality and morbidity (Hij) due to reductions in ambient concentrations of ozone and PM10 A standard dose response equations with the following form is used:

N C R

Where âij is the dose-response coefficient for the i th effect from the j th pollutant (% increase

in cases/year/person/ ì g/m3), R i is the background rate of the effect of interest

(cases/year/person), C j is the change ambient concentration of pollutant j (µg/m3) averaged

across the entire population as determined by the air quality module, and N is the

population at risk (persons)

A set of 19 health impacts, including premature mortality, chronic bronchitis, medical attention for cardiovascular and respiratory disease, and work loss days are analyzed in this study Dose response coefficients for each outcome are gathered from three main meta-analyses (USEPA, 1999; Cesar et al., 2002; Evans et al., 2002), with supplementary studies for information on select outcomes Greater weight is placed on evidence originating from Mexico Uncertainty in epidemiological evidence is included in our modeling, by including

a distribution of possible dose response values A detailed description of the sources for each coefficient and a summary table are included in Chapter V

Information on rates of hospitalizations and emergency room visits for respiratory and cardiovascular diseases were gathered in a co-study conducted by the National Institute of Public Health (INSP) using the database from the IMSS social security system This system covers approximately 80% of the population of the Federal District and nearly 30%

of the state of Mexico This database was chosen due to its data quality and availability While it does not represent the entire Mexico city population, it accurately captures the trends in the city Furthermore, the data gathered from this database account for less than 10% of the total monetary impacts Tables II.4a and b summarize results of the health impacts for the two time horizons

Table II.4a Annual mean health impacts (cases/year)

Time horizon 2003-2010

Taxi Renovation

Metro Expansion

Hybrid Buses LPG Leaks Cogeneration

Table II.4b Annual mean health impacts (cases/year)

Time horizon 2003-2020

Taxi Renovation

Metro Expansion

Hybrid Buses LPG Leaks Cogeneration

us to calculate cost per QALY ratios

For the economic valuation we use three methodologies to determine the total social benefit due to reductions in health impacts: 1 Direct health costs 2 Productivity loss and 3 Willingness to pay (WTP) These three methods are combined to give the total social benefits from reductions in health impacts, removing some impacts to avoid overlap Direct health costs were derived from an analysis by the Mexican National Institute of Public Health (INSP) of costs of hospitalizations and emergency room visits Productivity loss is calculated by the salary loss over the duration of an illness or years lost due to premature mortality Finally, for WTP, we use results from a study conducted in Mexico (Ibarrarán

et al., 2002) as well as those from the international body of literature adjusted to Mexican income, placing more weight on the Mexican study

Table II.5 Monetary benefits (2003 million US$ / year)

an intervention are simply the sum of quality-adjusted life years gained by avoiding premature mortality and disease QALYs are calculated by the following equation:

i

i T H u

Where u(H i ) is a utility weight assigned to a given health outcome (zero to one), and T i is

the duration of that health outcome The utility weights we use here are from several international studies (Fryback et al., 1993; Liu et al., 2000; Stouthard et al., 2000), as none have yet been done in Mexico The duration of illnesses are obtained from the IMSS databases, whereas the life years lost per premature mortality are calculated from a separate INSP study

Table II.6 Total QALYs saved per year

II.4 Results

We find that the combination of these 5 measures will substantially reduce emissions of local air pollutants, as well as GHG These measures will reduce PM10 exposure by approximately 1% (0.6 ìg/m3) for both time horizons; and will reduce maximum ozone concentrations by approximately 3% (6.2 ìg/m3 and 4.8 ìg/m3, respectively for 2003-2010 and 2003-2020), while eliminating emissions of more than 300,000 tons C equivalent per year and 400,000 tons C equivalent per year, respectively Together, these reductions will save more than 4,600 and 4,400 QALYs per year, respectively Monetized benefits are estimated to be $225 million USD per year and $210 million USD per year, respectively, for the combined 5 controls Total annualized costs are less than 30% of the estimated benefits: we estimate costs to be $66 million per year for 2003-2010 and $50 million USD per year for 2003-2020

Each measure contributes uniquely to these results The impact of each individual measure

is discussed below

For the 2003-2010 time horizon, the benefits of the Taxi Fleet Renovation are far greater than the costs (Table II.2 and II.5) Costs are small for this measure because of significant fuel efficiency gains realized with newer vehicles Benefits are high because of large ozone reductions, and also because of significant reductions in secondary particulate concentrations reductions (Table II.3) We estimate that approximately 3,000 QALYs per year could be saved with the measure (Table II.6), at mean cost of approximately $3,000 per QALY On the longer time horizon, net costs turn into net savings as the fuel cost savings continue to accumulate without additional investment costs Annualized benefits are still large, though less so, for the long time horizon because there is deterioration in emissions among aging vehicles that gradually increases local emissions, and thus decreases local benefits with time For 2003-2020, we estimate that approximately 2,000 QALYs per year could be saved (Table II.6) at the same time as cost savings are realized Consistent with existing government proposals, this analysis assumes that only 5 km of Metro would be built from 2003-2010, and an additional 71 km from 2011-2020 For this reason, it appears as to be a relatively small, inexpensive measure on the short time horizon, but much larger undertaking on the long horizon (Table II.2) Because Metro Expansion involves significant capital investment, the inclusion of the recuperation value for the Metro (30 year useful life) offsets a significant portion of these initial costs We find that the local emission reduction benefits (Table II.5, II.6) can also be large and compensate for

a majority, if not all, of the net costs for both time horizons For example, for 2003-2020,

we estimate that approximately 950 QALYs per year could be saved (Table II.6) at a cost

of approximately $50,000 per QALY by the expansion of the Metro This analysis assumes that the extension of the Metro causes a significant reduction in the use of on-road public bus transportation, which means local emissions are significantly reduced However, increase in Metro length requires more electricity and increases emissions from power plants that are primarily located outside the valley Thus, the Metro Expansion causes a net transfer of local emissions from inside to outside the valley We assume that population density is substantially lower where the electricity is generated than in Mexico City, and for this reason, public health impacts will be negligible from increased power generation This

transfer of local emission helps to make local benefits large enough to offset much, if not all, of the costs for this measure

The Hybrid Buses measure has large upfront investment costs due to the expensive nature

of the technology, but also generates significant cost savings on the long term due to greatly enhanced fuel efficiency (Table II.2) Benefits are large for both time horizons primarily because of large reductions in primary particulate emissions For both time horizons, we find that approximately 1,000 QALYs per year could be saved (Table II.6) This measure is implemented between 2003 and 2006 Annualized costs are, therefore, lower and benefits higher for the long time horizon than for the short time horizon; thus the cost per QALY reduces from approximately $60,000 for 2003-2010 to $20,000 for 2003-2010

The LPG leaks reduction measure, on the other hand, has low costs because of the low unit costs for each stove repair Benefits are much larger than the costs because of the significant reduction in hydrocarbon emissions that reduces both ozone and secondary organic particulate exposure For both time horizons, approximately 500 QALYs per year could be saved (Table II.6) at a cost of approximately $50,000 per QALY

For Cogeneration, net costs are low due to the significant gains in fuel efficiency and the inclusion of the recuperation value of the equipment at the end of each time horizon (20 year useful life) Local benefits are not very large for this measure because the gains in efficiency derive from simultaneous on-site production of thermal and electrical energy that replaces off-site electricity generation and on-site thermal energy production As explained above, only a small portion (3.1%) of the electricity consumed in Mexico City is generated

in the valley Though Cogeneration significantly reduces the total emissions by substantially increasing efficiency, the measure moves emissions of local pollutants into the valley, and thus local benefits are small QALYs saved are on the order of 30 per year for both time horizons (Table II.6) at a cost of approximately $25,000 per QALY

In Figure II.1, we compare local and global net benefits The local net benefits are defined

as the Monetized Health Benefits (Table II.5) minus Costs (Table II.2), while the global net benefit is the reduction in GHG emission Figure II.1 illustrates that the Taxi Fleet Renovation measure is clearly the best measure from the joint local – global perspective The Hybrid Bus measure for 2003-2020 and the LPG Leak measure on both time horizons are the next- most promising for joint local / global control The Metro Expansion, in large part because of its very high costs, is less promising from the joint perspective Cogeneration also does not have sufficient local benefits to make it interesting for joint local – global control

Figure II.1: Net Health Benefits vs C equivalent Reduction

II.5 Discussion and Conclusions

Taxi fleet renovation offers the most promising opportunity for the joint control of local and global pollution of the measures studied here Furthe r, benefits might be found to be significantly larger than estimated here if changes in primary particulate matter emissions could be estimated The large potential benefits of this measure have already been recognized by decision- makers in Mexico City, and the implementation of this measure has begun as of 2002-2003 with public funding for the replacement of 3,000 taxis

The LPG leak measure also provides benefits than are much larger than the total costs Emissions reductions and local benefits from this measure are small compared to the taxi fleet renovation, but investment costs are quite small, making implementation of the LPG leak measure relatively feasible from a decision- making standpoint

Cogeneration provides more than 50% of the GHG benefits from this set of measures, but essentially no local benefit because it moves emissions of local pollutants into the valley, and health benefits from the reduced emissions at power plants located outside the valley are assumed to be negligibly small Were a similar study conducted at the national level, Cogeneration may turn out to be a promising joint local / global option because health benefits derived in populations living near to power plants could be considered This will depend, of course, on populatio n exposure to emissions generated by electricity production across the country

Metro Expansion has large local benefits, particularly for the long time horizon when the measure has been fully implemented However, the extremely high initial investment costs

required for the measure make its implementation unlikely

Finally, the Hybrid Bus measure may have positive net benefits if the long time horizon is considered However, the analysis of this measure has large uncertainty because the emission factors used were derived for the altitude, driving conditions, and fuel mix of New York City, not for Mexico City Altitude has been shown (Yanowitz et al 2000) to significantly impact emissions behavior from heavy-duty vehicle technology, but these impacts ha ve not been specifically calculated for the technologies under consideration here

We recommend that a better understanding of emissions factors be obtained and also that the cost-effectiveness of other types of advanced technologies (e.g Cohen et al., 2003) also

be considered in order to determine what would be the best advanced bus technology to introduce in Mexico City

This work indicates that measures to improve the efficiency of transportation are key to joint local / global air pollution control in Mexico City The three measures in this category that are analyzed here all have monetized public health benefits that are larger than their costs when the appropriate time horizon is considered Global benefits, due to improved fuel efficiency, are also large In contrast, we find that traditional “no-regrets” electricity efficiency do provide large GHG emission reductions, but do not provide local benefits to Mexico City because the majority of electricity is produced outside of the valley in which Mexico City is located

Further work is needed to analyze more measures that cover a wider range of opportunities for joint local / global air pollution control Also very important is to quantify the air pollution improvements and cost savings that could be acquired from reduced congestion

in the MCMA Such an analysis may indicate that the benefits from transportation efficiency improvement are, in fact, much larger than estimates here Improved understanding of emission factors from new and old vehicles under Mexico City driving conditions is also greatly needed, and could significantly impact results

II.6 References

CAM, Comisión Ambiental Metropolitana (2002) “Programa para Mejorar la Calidad del Aire de la Zona Metropolitana del Valle de México, 2002-2010” (PROAIRE), Comisión Ambiental Metropolitana, México City

Cesar, H., et al (2000) “Economic valuation of Improvement of Air Quality in the Metropolitan Area of Mexico City,” Institute for Environmental Studies (IVM)

Cesar, H., et al (2002) “Air pollution abatement in Mexico City: an economic valuation,” World Bank Report

Chow, J.C., J.G Watson, S.A Edgerton, and E Vega (2002) “Chemical composition of

PM2.5 and PM10 in Mexico City during winter 1997,” The Science of the Total Environment

287, p.177-201

Cohen, J.T., J.K Hammitt, and J.I Levy (2003) Fuels for urban transit buses: A

cost-effectiveness analysis Environ Sci Technol 37 1477-1484

Evans et al (2002) “Health benefits of air pollution control,” in Air Quality in the Mexico Megacity: An Integrated Assessment, Kluwer Academic Publishers, Boston, 384 pp

Fryback, D., E Dasbach, R Klein, B Klein, N Dorn, K Peterson, and P Martin (1993)

"The beaver dam health outcomes study: initial catalog of health-state quality factors,"

Medical Decision Making, 13: 89-102

Ibarrarán, M., E Guillomen, Y Zepeda, and J Hammit (2002) “Estimate the economic value of reducing health risks by improving air quality in Mexico City,” preliminary results

Liu, J., J Hammitt, J Wang, and J Liu (2000) “Mother’s willingness to pay for her own

and her child’s health: a contingent valuation study in Taiwan,” Health Economics, 9:

319-326

M.J Bradley & Associates, Inc (2000) “Hybrid-electric drive heavy-duty vehicle testing

project: Final emissions report.” http://www.navc.org/Navc9837.pdf

Salcido et al (2001) “MCCM Parametric Studies: Estimation of the NOx, HC and PM10

emission reductions required to produce a 10% reduction in the Ozone and PM10 surface concentrations and compliance with the MCMA air quality standards, with reference to the

2010 MCMA Emission Inventory,” Grupo de Modelación de la Comisión Ambiental Metropolitan (CAM), 42 pp

Sheinbaum P., C., L Ozawa, O Vázquez, and G Robles (2000) “Inventario de emisiones

de gases de efecto invernadero asociados a la producción y uso de la energía en la Zona Metropolitana del Valle de México: Informe final.” Grupo de Energía y Ambiente, Instituto

de Ingeniería, UNAM, report to the CAM and the World Bank

SMA, Secretaria del Medio Ambiente del Distrito Federal (2002) Red Automática de Monitoreo Atmosférico (RAMA)

Stouthard, M., M Essink-Bot and G Bonsel (2000) “Disability weights for disease: a

modified protocol and results for a western European region,” European Journal of Public

Health, 10: 24-30

Turpin, B.J., J.J Huntzicker, S.M Larson and G.R Cass (1991) “Los Angeles summer

midday particulate carbon: Primary and secondary aerosol,” Envi Sci Technol., 25(10)

1788-1793

TUV Rheinland de Mexico, S A de C V (2000) “Programa para la reducción y eliminación de fugas de Gas LP, en las instalaciones domésticas de la Zona Metropolitana del Valle de México.”

U.S Environmental Protection Agency (1999) "The Benefits and Costs of the Clean Air

Act 1990-2010," Washington, D.C., Office of Air and Radiation, EPA report no

410/R-99/001

West, J.J., P Osnaya, I Laguna, J Martínez, A Fernández (2003) “Co-control of urban air pollutants and greenhouse gases in México City.” Final report to US National Renewable Energy Laboratory, subcontract ADC-2-32409-01

Yanowitz, J., R.L McCormick and M.S Graboski (2000) “In-use emissions from

Heavy-Duty diesel vehicles.” Environ Sci Technol 3, p 729-740

III.1 General Methodology for Estimating Emissions Reductions and Costs

III.1.1 Introduction

We estimate the time profiles of local pollutant (PM10, SO2, CO, NOX, and HC) and global pollutant (CO2, CH4, and N2O) emission reductions, and direct costs for 5 control measures that address transportation, residential and industrial sources of local and global air pollution emissions Detailed descriptions of each measure is outlined in sections III.2 through III.6 We also report emission reductions of PM2.5, calculated as a fraction of PM10

emissions (US EPA, 2000) for illustrative purposes, but do not use these estimates of emission reduction in the rest of the analysis

As described below, for each measure an emissions baseline is defined given currently measured or otherwise determined emissions factors and activity levels, combined with reasonable future predictions regarding their behavior without intervention Control measures cause a change from this baseline by altering future activity levels and / or emissions factors While emissions factors used in the study are meant to capture current driving conditions, the cost savings and changes in emissions due to reduced congestion could not be calculated because this was far beyond the scope of this study We encourage the pursuit of improved understanding of congestion impacts in future work since these impact may, in fact, be large

Our objective is to estimate emissions reductions and costs for each year from 2003 to

2020 In this way, the different time-profiles of the programs costs and impacts can be studied For incorporation into the cost – benefit and ancillary benefits analyses that are the goal of this study, we annualize the results obtained over these time horizons using several different discount rates Annualized costs and emissions reduction can be considered as a constant annual flux of costs or emission reductions over the time-period that gives an equivalent net present value to the net present value estimated from the actual time-profile

of the program In this way, annualized results allow direct comparisons between measures with different time-profiles

Further, annualized results allow cost-benefit and ancillary benefit calculations to be much simplified since it is only necessary to calculate air quality changes and health impacts based on a single set of emissions reductions that appropriately represent the entire time horizon, as opposed to having to do such calculations for each year The fact that our reduced-form air quality models (see Chapter IV) are essentially linearfacilitates the use of annualized emissions reductions

III.1.2 Choice of Time Horizon

We study both a short time horizon (2003 through 2010) that is consistent with Mexico City’s Program for Improved Air Quality in the Valley of Mexico (Programa para Mejorar

la Calidad de Aire en el Valle de Mexico, PROAIRE) 2002-2010 We also study a long time horizon (2003 through 2020) that allows consideration of the lasting effects of control

measures implemented up to 2010, and also allows consideration of realistic long-term implementation plans for the Metro Expansion control measure

III.1.3 Choice of Discount Rate

We calculate costs and emissions reductions using 3 discount rates, 3%, 5% and 7% We also present results when discounting is ignored, or 0% Our benchmark scenario, for which results are considered in Chapters IV to IX, uses a discount rate of 5%

III.1.4 Equations used for Discounting and Annualization

Discounting to estimate the Net Present Value (NPV) in 2003 (where j is the year from

2003, “value” is the emission reduction or cost in that year, and dr is the discount rate) uses Equation III.1.1

dr

value NPV

III.1.5 References

U.S Environmental Protection Agency (2000) "National Air Pollutant Emission Trends:

1900 - 1998," Washington, D.C., EPA report no 454/R-00-002

III.2 Renovation of the Taxi Fleet

III.2.1 Introduction

In 1998 approximately 109,400 taxis were circulating in the Mexico City Metropolitan Area (MCMA); 103,298 in the Federal District and the rest in the State of Mexico According to official figures, the total number of taxis accounted for 3.4 percent of the entire vehicle fleet in the metropolitan area that year (CAM, 2002a, Table 5.2.2.2) In the Federal District alone, taxis accounted for about 5 percent of the vehicle fleet and about 20 percent of the total vehicle kilometers traveled (CAM, 2002a, Table A.2.6) The emissions from these activities are estimated at 188 tons per year of PM10; 535 tons of SO2; 115,200 tons of CO; 10,366 tons of NOX; and 13,733 tons of HC, respectively (CAM, 2002a, Table 5.2.2.8)

By their nature taxis are high- use vehicles Over time their emission control systems would

be expected to deteriorate more rapidly than those of other vehicles used less intensively (however, see Kojima and Bacon, 2001) This is one reason why taxis are sometimes subject to more frequent tests in vehicle inspection and maintenance (I/M) programs High-use vehicles also consume more fuel, which contributes particularly to emissions of greenhouse gases (GHG), and which makes up an important part of the vehicle operating costs The problems associated with emissions from taxis are thus similar to the ones of the private car fleet, but they tend to be exacerbated by a more intense use of taxi vehicles

The weighted average age of taxis in the Federal District was 5.7 years in 1998 Four years later, this number had grown considerably and, according to some estimates, 49% of the fleet was more than 10 years old and should have been taken off the road in order to comply with existing regulations (Gonzalez, 2002) However, there are large uncertainties associated with these estimates A reliable vehicle registration database does not exist, and

it is difficult to obtain time-series data While new vehicle sales are added to the existing population every year, vehicle retirement is often not captured As a result, large differences have been measured when the official figures are compared with data from extensive field surveys (Kojima and Bacon, 2001)

The inconsistencies observed in the official records of the overall fleet size and composition are recognized by the Metropolitan Commission for Transport and Roadways (COMETRAVI, 1999a), and are similar to problems encountered in other parts of Latin America (for a discussion in the context of the MCMA, see Gakkenheimer et al 2002) Modeling the evolution of the taxi fleet is also complicated by the fact that most taxis are traded on the market for used vehicles, and that an unknown number of vehicles have been turned into taxis illegally

Yet, despite these challenges there seems to be a consens us within the local governments of the MCMA that something needs to be done about the emissions from the existing taxi fleet High- use vehicles (i.e., taxis and microbuses) are currently required to be renewed after a certain number of years, but the restrictions are not effectively enforced and the age

of an increasing number of these vehicles is higher than their age limit

Apart from their impact on air quality and human health, there are also other problems related to the taxis In particular, 60-70 percent of the taxi owners have only one vehicle as their main source of household income (Gonzalez, 2002) As a consequence, these owners work between 8 and 12 hours a day and typically they do not have any kind of social security Public policies to reduce emissions from taxis ought to be sensitive to this fact In the present analysis, however, we shall focus on the total emission reductions and the direct costs of such policies, while ignoring their implications for the distribution across individuals and households

III.2.2 Description of the Measure

In response to growing concerns about the emissions from taxis, an ambitious program has been designed to scrap 80,099 old taxis in the Federal District, and to replace them by vehicles that comply with more stringent emissions standards The program is being implemented over a four year period, provided sufficient public funds are available There are four overall goals of the program (Gonzalez, 2002)

First, in order to reduce emissions of local air pollutants, such as CO, NOX and HC, old taxi vehicles will be replaced by newer vehicles that comply with at least Tier 1 emission standards The replacement is facilitated by an incentive for present taxi owners to scrap their old vehicle in exchange for a premium of 1,500 U.S dollars In addition, subsidies are given to owners of new taxis in terms of reduced purchase prices from the automobile industry, a special tax relief from the government of the Federal District, interest rate subsidies from credit institutions, and subsidies on spare parts and services

Second, a requirement is included in the program that new vehicle engines must comply with a minimum fuel economy of 12.6 km per liter Compared with the existing taxi fleet, the requirement would imply not only considerable savings in fuel cost, but also a reduction

in GHG emissions Note, however, that this is based on the assumption of no “rebound effect” from an improvement in the fuel economy of new vehicles (NRC, 2002; Portney, 2002)

Third, as emphasized above a number of other problems surround the organization of the taxi fleet About 90 percent of the vehicles in service are so-called “free” taxis that circulate the streets empty looking for passengers In contrast with fixed-site taxis, which typically operate from a coordinated taxi stand, free taxis are not formally organized They produce more emissions per passenger kilometer traveled and are generally considered to

be less safe In the taxi renewal program, provisions are therefore included to increase the share of fixed-site taxis as a means to reduce the emissions and improve the safety of the passengers simultaneously However, it remains an open question to what extent the operators of free taxis will have sufficient incentives to join a taxi stand, or another form of coordinated operation Consequently, we shall not consider this element of the program in the analysis

The fourth goal of the program is to improve the income of the taxi owners through public and private subsidies and through increased social security Financial support is thus

recurring expenditures on vehicle operation and maintenance (i.e., interest rate subsidies and subsidies on spare parts and services) In addition, since taxi credits are generally considered by the commercial banks to be a risky asset leading to a prohibitively large risk premium on private commercial loans, a mechanism has been designed between the private financial sector, the government of the Federal District, and the National Development Bank (Nacional Financiera) to provide guaranteed loans at fixed interest rates An insurance scheme for taxi owners is also being considered jointly with the loan for the purchase of a new vehicle (Gonzalez, 2002; SETRAVI, 2002a)

According to the announced plan, the taxi renewal program is being implemented from

2002 to 2006 as part of an overall effort to integrate transport and environmental policies in the Federal Dis trict (CAM, 2002b; SETRAVI, 2002b) However, the financial viability of the program remains insecure Not only are the financial resources of the Federal District scarce, but there are also large imbalances in the public finances of the transport sector These imbalances stem in part from a massive underpricing of public transport and infrastructure, such as the metro system and the road network, and in part from the inability

of the local Secretariat of Transport and Roadways to raise public revenues For the fiscal year of 2002, it is estimated that only 37% of the total expenditures in the transport sector are covered by the revenues raised (Gakkenheimer et al., 2002; SETRAVI, 2002b)

From the documents available it is difficult to get a clear picture of the current state of the taxi substitution program In the preliminary Integrated Transport and Roadways Program (Programa Integral de Transporte y Vialidad, PITV) for 2002-2006, a total amount of 10 million US dollars has been designated to a fund for the substitution of 10,000 free taxis (SETRAVI, 2002b) In the Program for Improved Air Quality in the Valley of Mexico (Programa para Mejorar la Calidad de Aire en el Valle de Mexico, PROAIRE) 2002-2010, about 80,000 of the oldest taxis are expected to be gradually replaced at a total cost of 800 million U.S dollars, of which 80 million dollars would be financed by the public sector and

720 million dollars by the private sector (CAM, 2002b) Finally, in a brief summary of the progress of PROAIRE, Paramo (2003) comments on the availability of funds for the substitution of only 3,000 taxis for the fiscal year 2002

These discrepancies are probably a reflection of the financial insecurity of the program It is also a fact that the fiscal budget covers expenditures only one year ahead, while the scrappage and replacement of taxis is a multi- year effort that cuts across institutional boundaries within and outside the government of the Federal District In this respect, the program should be contrasted with the only other known scrappage program of a comparable magnitude, which was considered for almost a decade in California to improve air quality in the greater Los Angeles area, but which was subsequently abandoned by policy makers (Dixon and Garber, 2001a, 2001b; Dixon, Garber, and Porche, 2002)

Some taxis in the Federal District have already been scrapped and replaced Information about these experiences would be useful for the evaluation of the program Yet, data on the costs and emissions characteristics of both the old taxis that are scrapped and the new vehicles introduced have not been available for the purpose of the analysis We therefore conduct a prospective analysis of the program, based on our expectations about its likely impacts, and assume a period of implementation from 2003 to 2007

In the analysis, we focus on the real social costs, as opposed to the financial costs, associated with the scrappage and replacement of taxis, the implications for the emissions reductions, and the human health impacts in the Federal District In particular, we are interested in the question of whether the taxi renewal program is desirable from an overall societal perspective, taking into account only the allocative efficiency of the measure This means that we include the real resource costs associated with the scrappage and replacement of taxis (i.e., scrappage subsidy, vehicle replacement cost, and fuel cost), while

we omit the financial costs associated with a loss to some and a gain to other agents of the economy (i.e., public and private transfers)

III.2.3 Data Requirements

For the purpose of the evaluation, a wide range of data is needed to estimate the baseline emissions trajectory for the taxi fleet without the control measure These data include the base year (1998) emissions inventory for the MCMA, distributed between the Federal District and the State of Mexico (CAM, 2002a)

Data is also needed to extend the inventory with estimates of PM2.5 and GHG emissions For this extension, we introduce a number of simplifying assumptions In particular, we are interested in an explicit calculation of the average annual fuel consumption of taxis, which

at the same time can be used to estimate fuel consumption in the future Emissions of GHG are then straightforward to calculate on the basis of emissions factors (in grams per kilogram of fuel consumed) reported by the Intergovernmental Panel on Climate Change (IPCC, 1997)

Finally, the baseline emissions for the period 1998 to 2020 are estimated on the basis of expectations about the annual rates of change in both the size and composition of the taxi fleet and its emissions characteristics To the extent possible, these data are obtained from publicly available documents Where such data are unavailable, alternative assumptions are discussed and justified explicitly

Once the baseline scenario has been specified, it is an easy matter to impose the control measure according to the number of taxis to be replaced and the time period of implementation described above The annual emissions are then re-calculated in the control scenario for the time period of analysis, and the emissions reductions derived as the difference between the two scenarios Great care needs to be taken in order to ensure that relevant parameters in the control scenario are correctly adjusted If, for example, emissions standards are introduced in the control scenario that do not already exist in the baseline scenario, such as more stringent tailpipe emissions or fuel economy standards, this change needs to be reflected in the parameter values (i.e., the emissions factors and the fuel economy of the new vehicles)

On the basis of the changes introduced in the control scenario, data are needed on the incremental capital costs and operation and maintena nce (O&M) costs of each new taxi The capital cost include the initial scrappage subsidy (1,500 USD) and the incremental cost

changes in fuel consumption, valued at constant real prices over time We assume that government administration costs of the program are negligible, since no emissions testing is associated with the scrapping of old vehicles Also, monitoring and enforcement costs are not included During the first years of the program, when the oldest vehicles are replaced,

we believe that these costs can be ignored, because incentives are provided for taxi owners

to join the program, in part, through the scrappage subsidies and, in part, through the subsidies for new vehicles However, during later years of the program, when younger vintages of vehicles are retired, participation in the program will eventually become unattractive as the used car prices of younger vehicles raise above the scrappage subsidy This is clearly in opposition with the objectives of the program, and requires more careful consideration of the enforcement mechanisms needed to replace almost 80% of the taxi fleet in the Federal District

Yet, there is some confusion in the official perception of how the taxi substitution program

is enforced In the Secretariat of Transport and Roadways (SETRAVI, 2002a), the program

is viewed as voluntary This means that the decision to scrap an old taxi and replace it by a new one is left entirely to the owner Incentives therefore need to be put in place for the program to take effect (see, for example, Dixon and Garber, 2002a)

In the economics literature, these incentives are typically analyzed in models of so-called

“rational scrappage”, where the optimal decision of the owner to keep or scrap the vehicle

is based on the minimization of the present value of the costs from the two alternatives, with all the relevant costs included (e.g., Hahn, 1995; Alberini et al., 1995, 1998) Vehicle scrappage rates above the natural rate of retirement are then achieved by policies (e.g., scrappage subsidies, emission fees, differentiated ownership taxes, or more stringent inspection and maintenance) that change the relative costs in favor of scrappage Public policies based on this type of scrappage is also sometimes referred to as “voluntary accelerated vehicle retirement” (VAVR) programs (U.S EPA, 1998; ESMAP, 2002)

In contrast with this viewpoint of the taxi substitution program as voluntary, the Metropolitan Environmental Commission describes it in PROAIRE as mandatory (CAM, 2002b) Since taxi owners in the Federal District are allowed to operate only under a system

of public concessions, compliance with the taxi substitution program in ensured in PROAIRE by making the renewal of the concession for each taxi owner dependent on participation in the program Failure to participate in the program by not scrapping the old taxi means that the license of the owner to own and operate a taxi expires

For convenience, we adopt the latter viewpoint Estimating a model of rational scrappage is beyond the scope of the present analysis Given the available time and data, we therefore assume that the incentives provided by the program are sufficient to make the taxi owners comp ly If they are not sufficient, we assume that compliance can be enforced through the system of concessions currently in place in the Federal District This greatly simplifies the analysis However, the assumption of compliance is questionable given a past history of problems with concessions in the MCMA, particularly with respect to the operation of urban buses (Estache, 2001; Gakkenheimer et al., 2002) One should therefore be cautious

in the interpretation of the results from the present analysis

III.2.4 Determining baseline emissions

Following the emissions inventory (CAM, 2002a), a bottom- up approach is used to estimate the total emissions in 1998 by multiplying the level of activities (i.e., the number

of taxis and their annual vehicle kilometers traveled) with the level of emissions per unit of activity (i.e., the emissions factors in grams per kilometer) First, we describe the data used

in this approach We then turn to the projection of the activities and the emissions characteristics of the taxi fleet

The base year (1998) emissions

A vehicle registration database is not available for the MCMA In its place, data on the size and composition of the vehicle fleet, as well as its emissions characteristics, can be obtained from the vehicle verification program The program requires an emissions test to

be performed every six months on vehicles circulating in the Federal District and the State

of Mexico (see Gakkenheimer et al., 2002)

The activity data for the base year emissions inventory are specific to each model year vehicle in 1998 and spans a total of 25 model year vintages Figure III.2.1 shows the age distribution of taxis and private cars in the Federal District (CAM, 2002a, Table A.2.2) The age distribution of taxis is from the vehicle verification program in the second semester

of 1999 The figure illustrates that the taxi fleet is not very old, compared with the private car fleet, and that taxis appear to be retired faster than private cars This is probably because taxis travel more kilometers every year, and therefore deteriorate more rapidly due to wear and tear Other interpretations are also possible related to factors external to the vehicles themselves, such as differences in the price of maintenance and repair (Hamilton and Macauley, 1998)

Figure III.2.1 Age distribution of taxis and private cars in the Federal District (1998)

The taxis are assumed to travel 200 kilometers per day during 6 days a week This yields a total distance traveled of 62,600 kilometers per year (COMETRAVI, 1999) The estimate represents an upper bound of the annual vehicle kilometers traveled (VKT) per taxi, compared with other estimates of odometer readings taken from the verification program in the period from 1996 to 1999 (Kojima and Bacon, 2001) These estimates indicate that taxis

in the Federal District on average travel 30.000 km per year – about half the estimate we use in the present study

Notice that we do not differentiate across model year vehicles in terms of their annual VKT In agreement with most empirical observations, other studies typically assume that old vehicles travel less than new ones (e.g., Mostashari, 2003) This pattern has also been found in the MCMA, although at a very aggregate level (Kojima and Bacon, 2001) Here

we simplify the analysis and leave the quantitative significance of such a variation for further study

Given the total distance traveled for each model year, the emissions of criteria pollutants (CO, NOX, and HC) are calculated with the emissions factors from the emissions inventory shown in Table III.2.1 (CAM, 2002a, Annex A) In the inventory, emissions factors for diesel fueled vehicles and motorcycles are estimated through the MOBILE5-MCMA model The MOBILE model was originally developed by the U.S EPA, and has subsequently been adjusted for use in Mexico, including the Mexico City Metropolitan Area (Radian International, 1997; ERG and Radian International, 2000; Burnette et al., 2001) The model is part of a larger, on-going effort to improve the capacity within Mexico for the development of emissions inventories However, the MOBILE model has been subject to critical scrutiny in the U.S in recent years, particularly as a means to estimate the expected emissions reductions from mobile source control measures (Harrington et al., 1998; NRC, 2000) A new generation of the model has therefore been developed to address some of its limitations (U.S EPA, 2001, 2002)

For gasoline fueled vehicles in Mexico City, including taxis, emissions data have been obtained from tunnel studies and measurement campaigns conducted, in part, by the Mexican Petroleum Institute (IMP) during the 1990s Focusing on the emissions of hydrocarbons, two tunnel studies report results on the measurement of exhaust emissions profiles for motor vehicles in operation, as well as hot soak emissions from vehicles in a parking garage (Mugica et al., 1998; Vega et al., 2000) These results were then combined with ambient air quality measurements to develop a source apportionment model, which shows that somewhere between 55% and 64% of the ambient concentrations of non-methane hydrocarbons (NMHC) can be attributed to the emissions from motor vehicles

However, despite these and other efforts, it is a very complicated and time consuming task

to develop a comprehensive emissions inventory, which at the same time can be validated through the use of various independent methods (for an excellent discussion, see Molina et al., 2002) Estimating the emissions of taxis in the MCMA is no exception, and it is not clear from the 1998 emissions inventory what are the sources of the emissions factors for taxis (CAM, 2002a, Annex A) The data are shown in Table III.2.1, but measurement results have been obtained only for model years 1991 to 1998 For all the previous model years, the emissions factors of private cars were used instead

Table III.2.1 Emissions factors of local air pollutants from taxis in the MCMA (g/km)

of PM in the two areas These factors include differences in temperatures, the urban driving cycles used to test emissions, characteristics of the vehicle fleet and the sample of vehicles

in the test For example, under the driving cycle defined by the Federal Test Procedure (FTP), the Denver study finds considerable differences between the emissions of PM10

from new vehicles (2.82 mg per mile, model years 1991-1996) and the emissions from older vehicles (95.5 mg per mile, model years 1971-1980) during the summer During the winter, this difference is narrowed somewhat (Cadle et al., 1999, Table 4)

In the emissions inventory for the MCMA, no distinction is made between model years with respect to the emissions of PM10 This is unfortunate for the evaluation of the taxi substitution program, because the emissions reductions from this program are obtained precisely from the differences between old and new vehicles Failure to take these

differences into account in the emissions inventory therefore means that the emission reductions are at best underestimated, at worst they are completely ignored

Reductions of PM10 can still be obtained from the taxi substitution program, if the new vehicles in the control scenario comply with more strict exhaust emissions standards than new vehicles in the baseline scenario This is in fact the source of PM10 reductions in PROAIRE (CAM, 2002b), where new vehicles in the taxi substitution program are assumed to comply with the strict Tier 2 emissions standards, while new taxis in the baseline are assumed to comply with the 1998 model year emissions reported in Table III.2.1 We believe that such estimates are unfounded, because they are based on a comparison of certification standards for Tier 2 vehicles, when they leave the factory (as stated in the U.S regulations), with the measurement of in-use emissions from new vehicles in Denver Moreover, the estimates tend to obscure the fact that there might be important PM emissions reductions from a difference between model years that is not taken into account An effort should therefore be made to remedy this in the emissions inventory

Emissions of PM2.5 are assumed to be a constant fraction of PM10 The fraction is obtained from the national emissions trends observed in the U.S over the last decade (U.S EPA, 2000) Estimates of PM2.5 are not included in the official emissions inventory for the MCMA, and those presented here must be considered rather hypothetical Since the fine particulate fraction is used as an all purpose estimate, the qualifications above for PM10 also applies to PM2.5

Emissions of GHG (CO2, CH4, and N2O) are estimated from emissions factors reported in IPCC (1997, Table 1-27) in grams per kilogram of fuel consumed The total annual fuel consumption is in turn estimated on the basis of assumptions about the average fuel economy of taxis model year 1998 and earlier on the one hand (i.e., the existing fleet), and model year 1999 and later on the other hand (i.e., the new fleet) The fuel economy, the aggregate fuel consumption in 1998, and the emissions factors are shown in Table III.2.2 The estimate of the fuel economy of the existing fleet is calculated backwards, given data from Mexican Petroleum (PEMEX) on total gasoline consumption in the MCMA in 1998, and given the total VKT of taxis A density of 0.736 kg/liter for PEMEX Premium and 0.73 kg/liter for PEMEX Magna is used to convert fuel consumption in liters to kilograms, before applying the emissions factors

Table III.2.2 GHG emissions factors for taxis in the MCMA (g/kg)

Model

Year

Fuel economy (km/liter)

Consumption Premium 1998 (liter)

Consumption Magna 1998 (liter)

CO2(g/kg)

CH4 (g/kg)

N2O (g/kg)

automobile manufacturer with an average fuel economy of 12.6 km per liter Adjusted for the urban driving cycle and an observed bias in laboratory measurements, this reduces to 9.0 km per liter (U.S EPA, 2001)

The data applied so far in the analysis on vehicle activities and emissions characteristics for the local criteria pollutants coincide with the data used in the 1998 emissions inventory (CAM, 2002a) The total emissions in 1998 are therefore identical, except for a small difference in rounding For the projection of the baseline activities and emissions, however, there are differences in both the methodology and the data used For the purpose of comparison, the reader is referred to the calculation of projections and emissions reductions

in PROAIRE until 2010 (CAM, 2002b)

Vehicle fleet and travel demand projections

In the evaluation of measures to control emissions from mobile sources, it is customary to develop models that are able to generate forecasts of future travel demand These models are based on expectations about the growth in income per capita and other socio-economic characteristics of the population, which may serve as explanatory variables Some models generate simple estimates of changes in the number and distance of trips, distributed over different modal alternatives (i.e., private cars, taxis, microbuses, etc.) Other models involve more complex econometric estimation There are also models which include changes in land use among the driving forces behind vehicle ownership and use (Harrington and McConnell, 2003)

To some extent, all these different alternatives are relevant to the estimation of the future emissions from taxis in the MCMA, given changes in the number of vehicles, their age distribution, and the total distance traveled (Mostashari, 2003) In the present analysis, however, we side-step the issue of travel demand modeling for two reasons

First, although the taxis in the MCMA are privately owned, the ownership and use of taxis

is conditioned on public concessions issued by the government of the Federal District and the State of Mexico These concessions, if effectively monitored and enforced, act like a constraint on the expansion of the number of taxis Rather than being a variable in need of explanation, the growth in the taxi fleet thereby becomes a parameter over which the policy makers assume direct control

In the Federal District, no new concessions are currently issued as the result of an explicit political choice (SETRAVI, 2002b) This is seen as a means to reduce the share of taxis in the vehicle fleet over time, since they are generally considered to be in oversupply In the analysis, we therefore assume a zero percent growth rate of new taxis in the Federal District Obviously, this parameter can be changed in order to see the implications from the choice of different alternatives In the State of Mexico, an annual growth rate of 2 percent is expected according to SENER (2000)

Second, the demand for vehicles and their use is on occasion seen as a derived demand for transport services with certain characteristics What is demanded is not the vehicle per se,

safety But, although taxi owners have preferences over these alternatives, the essential decision with respect to travel seems to be one of supply, not demand From the viewpoint

of the taxi owner, assuming he is also the driver, the problem can therefore be stated as one

of choosing how many kilometers to supply, given alternative prices (i.e., the taxi fare), capital and labor costs, and a labor- leisure trade off In other words, whereas the private car

is most easily seen as a durable consumer good, the taxi is more like a producer good This ought to lead to differences in the modeling strategy of future travel behavior

Vehicle fleet turnover

Given the growth rates of new taxis, the total size of the taxi fleet until 2020 is determined Since we do not discriminate between model years in terms of annual distance traveled, the total VKT of taxis is also determined If old vehicles are assumed to be driven less than new vehicles, as the evidence seems to indicate, the total VKT depends not only on the total number of vehicles, but also on the age distribution

In order to determine the age distribution of the taxi fleet over time, we develop a simple model of the fleet turnover, which consists of two basic elements; a natural rate of retirement and a rate of replacement We assume that the two rates are identical each year,

so that the old taxis retired are automatically replaced by new ones This means that the turnover of the taxi fleet is independent of the overall fleet size, a fact which helps us interpret the results The natural rate of retirement (or the natural scrappage rate) is determined through the specification of age specific “death” probabilities, with the property that old vehicles are more likely to be retired than new vehicles This property is supported

by the empirical literature The retirement rates of taxis are calculated on the basis of a linear function in 1999, which produces a total retirement of taxis that year equal to 4% of the fleet This function is kept constant during all the subsequent years The retirement rates are shown in Figure III.2.2

Figure III.2.2 Age specific retirement rates for taxis in the MCMA

From the cumulated retirement rates, a survival function is derived in Figure III.2.3 The figure shows, in percentage terms, how many taxis of each model year would be expected

to survive in comparison with the number of taxis in the fleet from the beginning Given the retirement and survival rates, it is easy to confirm that 17 years is the maximum age all taxis in the MCMA This is deliberately a conservative estimate

Figure III.2.3 Age specific survival rates for taxis in the MCMA

III.2.5 Estimating emissions reductions and costs for the measure

The previous section has described in considerable detail the development of the baseline scenario and the choice of parameter values for that purpose In this section, we merge the baseline with the specification of a benchmark control scenario The benchmark refers to a choice for the analysis of what we believe are moderate parameter values, as well as a control scenario that is not too stringent The combination for this analysis is shown in Table III.2.3

Table III.2.3 is a “policy analysis matrix” which illustrates the key assumptions and policy variables in the analysis The policy variables are those over which policy makers and others exercise control in the taxi renewal program, such as the number of old taxis to be replaced, and the standards to be required from new taxis Although the control scenario is quite ambitious in the number of taxis to be scrapped and replaced, it is not very stringent

in the other policy variables This is easily observed in the table for the tailpipe emissions and the fuel economy standards of new vehicles in the control scenario, which are assumed

to be identical with the standards of new vehicles in the baseline scenario Since there are

no additional requirements associated with these standards, they have no incremental costs either

Table III.2.3 Benchmark control scenario for renovation of the taxi fleet

Benchmark control scenario Baseline

Parameters

Parameter values

Taxi ownership concessions

Accelerated turnover/

scrappage

Fuel economy standards

Tailpipe emission standards

Vehicle ownership growth rate

Concessions issued by the government

Vehicle replacement rate

Vehicle kilometers traveled (VKT)

“No Driving Day” not included

Fuel economy and fuel economy standards

Tailpipe emissions standards

Emissions deterioration and durability requirements

Model year 1998 IMP emis

factors Model year 1999 Adjusted

IMP factors

No durability requirements

The policy matrix is easy to extend in both the vertical and horizontal dimensions With respect to the baseline parameter assumptions, it may serve as a useful tool for identifying possible variations in the parameter values, with the aim of conducting a sensitivity analysis and, eventually, include the control measure in the Analytica software With respect to the benchmark control scenario, one can imagine a number of other, more stringent control measures, such as larger improvements in the fuel economy of new vehicles, Tier 2 emission standards, and standards for the durability of the emission control technology If combined with estimates of the associated incremental costs, an increasingly more strict control scenario can be subject to an incremental cost-effectiveness analysis

From Table III.2.3 it appears that an important caveat apply to the analysis Thus, it should

be noted that the No Driving Day program (Hoy No Circula) has been ignored in the estimation of the total distance traveled The program restricts vehicles from circulating in the MCMA for one or two days during a week, if the vehicle exceeds certain in- use emission standards in the vehicle verification test Assuming that No Driving Day remains unchanged over the time period of the analysis, including the program would restrict a growing number of vehicles in the baseline according to an increasing emissions

deterioration By contrast, when these vehicles are replaced with new ones, the restrictions from the No Driving Day would be suspended due to the improved emissions control The net effect of this change would therefore be to reduce the emissions reductions obtained in the present analysis

III.2.6 Costs and Emissions Reductions with the Measure

The undiscounted emissions reductions from the benchmark control scenario are shown in Table III.2.4 As explained above, there are no reductions of PM emissions, since no distinction is made between the emissions factors of old and new vehicles in the emissions inventory By contrast, the emissions reductions of criteria pollutants and GHG are quite substantial The time profile shows that the emissions reductions peak in 2007, the last year when taxis are scrapped and replaced through the program The emissions reductions then start to decline This observation is a logical consequence of the scrappage program as a measure that produces emissions reductions only temporarily Since none of the fundamental parameters behind the turnover of the taxi fleet are affected (i.e., growth and retirement rates), the size and composition of the fleet in the control scenario will automatically converge to the original size and composition of the fleet in the baseline scenario And since no permanent changes are assumed in the emissions characteristics of the new taxis, over and above the new vehicles in the baseline scenario, the total emissions

in the control scenario will eventually return to the baseline level

Table III.2.4 Emission reductions, renovation of the taxi fleet, without discounting

Table III.2.5 Annualized emissions reductions from the renovation of the taxi fleet

(ton/yr)

The annualized emissions reductions are shown in Table III.2.5 for a plausible range of discount rates The table shows that the results are not very sensitive, neither to changes in the discount rate, nor to changes in the time horizon of the analysis With the exception of the NOX reductions, the changes are quite small The larger change in NOX reductions from

2010 to 2020 is due to a sharp decline in NOX emissions reductions, which even become negative from 2015 and onwards This seeming paradox is explained by the emissions factors for NOX in the emissions inventory, which show a remarkable jump in the emissions for the 1989 and 1990 model year vehicles However, while this jump had a justification in the past, as a technical artifact of the existing taxi fleet, it is unlikely to be replicated in the future A more careful modeling of the future emissions of taxis, including the (NOX) emissions deterioration rates, might show a less drastic return to the underlying baseline trend Such a modeling exercise could be conducted on the MOBILE5 or MOBILE6 models

The time profile of the undiscounted cost estimates and the annualized costs are shown in Table III.2.6 and Table III.2.7 It is clear from the undiscounted costs that although both the public and private costs of capital are substantial, there are also considerable savings in fuel costs due to an improved fuel economy of new vehicles Note that the fuel savings peak in the year 2007 for the same reasons as the emissions reductions Similarly, the annual savings will eventually become zero at some point beyond the time horizon of the analysis

Given the time profile of the costs, with large up- front capital costs and fuel savings distributed over the entire time period of analysis, it is expected that the total cost estimates are sensitive to both the choice of discount rate and the choice of time horizon A lower discount rate leads to much lower total costs, because the future savings are discounted less compared to the initial capital costs Likewise, when the time horizon of the analysis is extended from 2010 to 2020, the fuel savings are included over a longer period, while the capital costs remain unchanged This leads to negative total costs for the 2003 to 2020 period, independently of the discount rate applied in the analysis

Table III.2.6 Costs of renovation of the taxi fleet without discounting

Table III.2.7 Annualized costs of renovation of the taxi fleet (millions US$/yr)

Compliance Cost (2003 millons US$/ yr) Discount rate Public

Investment Private Investment

O&M Cost

Total Cost Time Horizon 2003-2010

Among the parameters in the present analysis, some can be identified as playing a key role for the final results They include the emissions factors of PM10, the emissions of old high-use vehicles, emissions deterioration rates for the future taxi fleet, and the age specific rate

of retirement These variables, in particular, should be subject to more careful analysis of their quantitative impact on the results In addition, we have adopted a moderate control scenario in the present analysis If cost estimates for Tier 2 vehicles can be obtained, it would be interesting to see how large further emissions reductions can be achieved Since this would imply a permanent shift in the emission rates of new vehicles, the emission reductions would also be more permanent

Considerable uncertainty also exists with respect to the implementation of the taxi substitution program and its expected level of compliance Since this has been conveniently assumed away in the present analysis, the question of the enforcement of scrappage and replacement does not arise This is a form of model uncertainty that should be taken seriously

III.2.8 Discussion and Next Steps

On the basis of the present analysis, the scrappage and replacement of taxis seems to be a worthwhile measure to adopt in order to control the large share of emissions from mobile sources in the MCMA However, this conclusion depends on a number of uncertain factors, whose influence has not been formally analyzed A careful sensitivity analysis could provide a more firm ground upon which the taxi substitution program is recommended to policy makers and communicated to the public In this respect, an effort should be made to explain, to what extent the results obtained from the scrappage and replacement of taxis can

be expected to carry over to the in- use private car fleet

A fundamental source of conflict exists in the current design of the program This conflict means that, under the worst circumstances, an over- valued price of the oldest taxis is being used to achieve emissions reductions which are highly uncertain, and which we have been unable to verify and document within the time available for this study The uncertainty derives from a poor understanding of the average emissions of old taxis, for which there are

no directly estimated emissions factors in the emissions inventory It also derives from a supposedly large distribution over this average, which has been observed in numerous measurement studies of private cars If these observations hold also for taxis, an improved cost-effectiveness of the taxi substitution program might be achieved by an improved targeting of the program As a thought experiment, a better targeting is associated with improved knowledge of high-emitting taxis and their costs in the used car market This calls for an increased testing of the taxis in the program, possibly at the expense of scrapping 80% of the fleet

III.2.9 References

Alberini, A., W Harrington, and V McConnell (1995), “Determinants of Participation in Accelerated Vehicle Retirement Programs”, The RAND Journal of Economics, 26(1), 93-

112

Alberini, A., W Harrington, and V McConnell (1998), “Fleet Turnover and Old Car Scrap Policies”, Discus sion Paper 98-23, Resources for the Future, Washington, D.C

Burnette, A.D., S Kishan, and M.E Wolf (2001), “MOBILE5-Mexico: An Emission Factor Model for On-Road Vehicles in Mexico”, Paper presented at the 10th Annual Emission Inventory Conference, May 2, 2001, Denver, Colorado

Cadle, S.H, P Mulawa, E.C Hunsanger, K Nelson, R.A Ragazzi, R Barrett, G Gallagher, D.R Lawson, K.T Knapp, and R Snow (1999), “Light-Duty Motor Vehicle

Exhaust Particulate Matter Measurements in the Denver, Colorado Area”, Journal of the

Air and Waste Management Associaction, 49, 164-74

Cadle, S.H., P Mulawa, P Groblicki, C Laroo, R.A Ragazzi, K Nelson, G Gallagher, and B Zielinska (2001), “In-Use Light-Duty Gasoline Vehicle Particulate Matter

Emissions on Three Driving Cycles”, Environmental Science and Technology, 35, 26-32

CAM (Comisión Ambiental Metropolitana) (2002a), Inventario de Emisiones de la Zona

Metropolitana del Valle de México, 1998 (Mexico, D.F.: CAM)

CAM (Comisión Ambiental Metropolitana) (2002b), Programa para Mejorar la Calidad

del Aire de la Zona Metropolitana del Valle de México, 2002-2010 (Mexico, D.F.: CAM)

COMETRAVI (Comisión Metropolitana de Transporte y Vialidad) (1999), “Diagnóstico de

las Condiciones del Transporte y sus Implicaciones sobre la Calidad del Aire”, in Estudio

Integral de Transporte y Calidad del Aire para la Zona Metropolitana del Valle de México, Vol 1 (Mexico, D.F.: COMETRAVI and CAM)

Dixon, L and S Garber (2001a), Fighting Air Pollution in Southern California by

Scrapping Old Vehicles (Santa Monica, CA.: RAND Institute for Civil Justice)

Dixon, L and S Garber (2001b), “Scrapping Old Vehicles Would Improve Southern

California Air Quality at Reasonable Cost”, Research Brief, RB-9033, RAND Institute for

Civil Justice, Santa Monica, CA

Dixon, L., S Garber, and I Porche (2002), “Driven Into a Corner”, RAND Review, 26(3),

10-15

ERG (Eastern Research Group) and Radian International (2000), MOBILE5-Mexico:

Documentation and User’s Guide (Denver, CO.: Western Governors’ Association and

Binational Advisory Committee)

ESMAP (UNDP/World Bank Energy Sector Management and Assistance Program) (2002),

“Can Vehicle Scrappage Programs Be Successful?”, South Asia Urban Air Quality Management Briefing Note No 8, World Bank, Washington, D.C

Estache, A (2001), “Privatization and Regulation of Transport Infrastructure in the 1990s”,