POLICY RESEARCH WORKING PAPER 2785 Im proving Air Q uality in The annual health -.erefir . of a i 0 percent rct, Ai-rl: :nc in Metropolitan Mexico City ozone and PM 10 in Mexico City, conservatively estimated, An Economic Valuation are approximately S760 million (in 1999 U.S. dollars) annually. Reducing PM 1 0 has The Mexico Air Quality Management Team * larger estimated health benefits than reducing ozone, irh each microgram per cubic centimeter reduction in PM I O worth about $100 million per year. The World Bank Latin America and the Caribbean Region Environmentally and Socially Sustainable Development Sector Unit February 2002 I POLICY RESEARCH WORKING PAPER 2785 Summary findings Mexico City has for years experienced high levels of functions from the peer-reviewed literature. They value ozone and particulate air pollution. In 1995-99 the cases of morbidity and premature mortality avoided entire population of the Mexico City metropolitan area using three approaches: was exposed to annual average concentrations of fine * Cost of illness and forgone earnings only (low particulate pollution (particulates with a diameter of less estimate). than 10 micrometers, or PM10) exceeding 50 * Cost of illness, forgone earnings, and willingness to micrograms per cubic meter, the annual average standard pay for avoided morbidity (central case estimate). in both Mexico and the United States. Two million * Cost of illness, forgone earnings, willingness to pay people were exposed to annual average PM10 levels of for avoided morbidity, and willingness to pay for more than 75 micrograms per cubic meter. The daily avoided mortality (high estimate). maximum one-hour ozone standard was exceeded at The results suggest that the benefits of a 10 percent least 300 days a year. reduction in ozone and PM10 in 2010 are about $760 The Mexico Air Quality Management Team million (in 1999 U.S. dollars) annually in the central documents population-weighted exposures to ozone and case. The benefits of a 20 percent reduction in ozone and PM10 betwccn 1995 and 1999, project exposures in PM10 are about $1.49 billion annually. In each case the 2010, and computes the value of four scenarios for benefits of reducing ozone amount to about 15 percent 2010: of the total benefits. * A 10 percent reduction in PM10 and ozone. By estimatilng the magnitude of the beniefits from air * A 20 percent reduction in PM1O and ozone. pollution control, the authors provide motivation for * Achievement of ambient air quality standards across examining specific policies that could achieve the air the metropolitan area. pollution reductions that they value. They also provide * A 68 percent reduction in ozone and a 47 percent unit values for the benefits from reductions in ambient reduction in PM10 across the metropolitan area. air pollution (for example, per microgram of PM10) that The authors calculate the health benefits of reducing could be used as inputs into a full cost-benefit aiialysis of ozone and PM10 for each scenario using dose-response air pollution control strategies. This paper-a product of the Environmentally and Socially Sustainable Development Sector Unit, Latin America and the Caribbean Region-is part of a larger effort in the region to assist the Mexico City Metropolitan Area authorities in formulating the Third Air Quality Management Plan. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Glaura Lage, room 16-130, telephone 202-473-1099, fax 202-676- 9373, email address glage@worldbank.org. Policy Research Working Papers are also posted on the Wcb at http:// econ.worldbank.org. The authors may be contacted at mcropper@Caworldbank.org or wvergaraaworldbank.org. February 2002. (51 pages) The Policy Reseatch Workinig Paper Series dissemniinates the findinigs of work in progress to encourage the exchange of ideas about development issuies. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The -. - . interpretations, and conclusionzs expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center Improving Air Quality in Metropolitan Mexico City An Economic Valuation * The Mexico Air Quality Management Team The Mexico Air Quality Management Team was led by Walter Vergara and includes the following authors: Hennan Cesar' Xander Olsthoomn Victor H. Borja-Aburto2 Alberto Rosales-Castillo2 Pablo Cicero-Fernandez3 Gloria Soto Montes de Oca4 Kees Dorland' Victor Torres-Meza2 Roberto Mufioz Cruz4 Ricardo Uribe Ceron4 Luke Brander1 Pieter Van Beukeringt Maureen Cropper8 Eduardo Vega L6pez4 Ana Citlalic Gonzalez Martinez5 Max Magin Nifio Zarazua5 Gustavo Olaiz-Fernandez7 Miguel Angel Nifo Zarazua5 Ana Patricia Martinez Bolivar6 Walter Vergara9 ' Institute for Environmental Studies (IVM), Vrije Universiteit, Amsterdan 2 Secretaria de Salud, Mexico DF Centro Nacional de Salud Ambiental (CENSA) 3Environmental Health Sciences Department, School of Public Health, University of California Los Angeles (UCLA) 4 Secretaria del Medio Ambiente del Gobiemo del Distrito Federal, Mexico DF 5 Unidad Asesora para Comunidades Empresariales, A.C. (UfNACE) 6 Pan American Health Organization (PAHO) 7 Direcci6n General de Salud Ambiental, Mexico DF 8 Development Research Group, World Bank, Washington, 9Latin America and Caribbean Environment Department, World Bank This report is part of a larger study on "Health Risks of Atmospheric Pollution," supervised by the Technical Secre- tariat of the Environmental Metropolitan Commission. The Institute for Environmental Studies of the Free University (Amsterdam) and the Centro Nacional de Salud Ambiental are in charge of the execution of the report. Copyrights are with the Comision Ambiental Metropolitana (CAM) and the World Bank. This report is part of the assistance provided by the World Bank to the CAM for the formulation of the proposed Third Air Quality Management Plan in the Mexico City Metropolitan Area (MCMA). The assistance also includes technical assistance for the preparation of an updated emissions inventory and for the modelling of the airshed of the MCMA and assessment of measures for air quality im- provements. The Netherlands Consultant Trust Fund of the World Bank financed the preparation of the report and the International Development Research Center provided financial support for salaries. The results reported here were managed and supervised by Walter Vergara and Carl-Heinz Mumme. John Dixon and Robert Bacon provided useful comments on drafts of this report. Executive Summary The Mexico City Metropolitan Area (Zona Metropolitana del Valle de Mexico (ZMVM)) has witnessed high levels of air pollution in the past few decades. Recent efforts to curb emis- sions have been reasonably successful, and 1999 had the lowest overall level of air pollution during the last decade. With the exception of lead, carbon monoxide and sulfur dioxide (S02), however, pollution levels are still far above current air quality standards (See Table E.1). Table E.1: Number of Days Per Year that Ozone and PM10 Concentrations in Mexico City Satisfy Daily Air Quality Standards 1995 1996 1997 1998 1999 Ozone 41 39 43 45 65 PM10 273 186 212 176 345 Source: GDF (2000). Further efforts to reduce polluting emissions are being developed by the Comision Am- biental Metropolitana (CAM) under the Third Air Quality Program 2001-2010. This study presents an economic valuation of benefits from reducing pollution in the ZMVM, as the main economic rationale for controlling emissions is the welfare gain from improvements in air quality. The current study focuses on the two most important economic impacts of air pol- lution, namely health impacts and restrictions imposed on economic activities through envi- ronmental contingencies (contingencias ambientales). The health hazards associated with ozone and PM1O are studied because these sub- stances are the most important in terms of violating pollution standards. Ozone pollution stems mainly from emissions of NO, and VOCs. Their concentration levels depend on the amount and location of emitted pollutants, geographical characteristics, meteorological con- ditions, and atmospheric chemistry and transport. The chemistry of ozone formation is com- plicated and nonlinear: under certain conditions, an increase in NO, emissions could reduce ozone concentrations. PM10 pollution stems mainly from direct emissions of particles, and from reactions of NO,, and S02 with other substances in the atmosphere. Likely emission sources are building and construction (road construction), diesel trucks and buses, forest fires, open-air refuse burning, some manufacturing industries, and resuspension of road dust. The daily 1-hour maximum air quality standard for ozone is 0.110 ppm. During 1995- 99, the highest concentration observed for ozone-0.349 ppm-was measured at the Pede- gral station, in the southwest zone of the ZMVM. The Chapingo station in the northeastern zone was the least polluted, with a daily 1-hour maximum concentration of 0.210 ppm. The daily average air quality standard for PMIO is 150 ,ug/m3 and the annual average standard is 50 igIm3. All stations violated both standards with the exception of the annual average stan- dard at the Pedregal and Coacalco stations. The highest concentrations were in the east of the ZMVM with a daily maximum of 335 jig/m3 at the Netzahualcoyotl station. The highest an- nual average of 94 ug/m3 was observed at the Xalostoc station. In 1995, over 1.2 million people were exposed to concentrations above the environmental contingency Stage I level of 300 ,g/m3 at least once during the year. The baseline scenario for 2010 assumes emissions of NO, and VOCs, precursors of ozone and PM1o, to be the same as at the end of the 1990s. Likewise, we assume air quality in 2010 with respect to ozone and PM1O to be the same as the levels observed at the end of the 1990s. This assumption, however crude, seemed to be the most appropriate one in the ab- sence of an integrated model of emission projections for 2010 for fixed and mobile sources in Mexico City. Four alternative air pollution reduction scenarios for 2010 are evaluated. We do not ap- praise the policies needed to achieve the concentration reductions. The four scenarios are (population weighted exposure reductions are presented in table E.2): * a 10-percent reduction in PM1O and ozone; * a 20-percent reduction in PM10 and ozone; * improved air quality compliance at an air quality standard of 50 Ftg/m3 for PM1O and 0.11 ppm 1 -hour maximum for ozone in all ZMVM locations (AQS 1); * an air quality standard superimposing the required decrease in concentrations in the most polluted areas (Xalostoc for PM10 and Pedregal for ozone) across the ZMVM (68 and 47 percent reduction in ozone and PMIO concentrations, respectively) (AQS2). Table E.2: Reduction in Population-Weighted Exposure for the Analyzed Scenarios Population weighted Population weighted exposure to PM10 Exposure to ozone Scenario (ug/m3/person) (ppm/person) 10 percent exposure reduction 6.41 0.0114 20 percent exposure reduction 12.81 0.0227 AQS compliance in each area - AQSI 14.06 0.0702 AQS compliance in worst area - AQS2 29.99 0.0778 The health risks due to air pollution (specifically ozone and PM1O) are quantified by es- timating the relationship between the incidence of adverse health effects and air quality. To this end, a number of quantitative estimates of exposure-response relations of known health effects from various cities have been pooled together (meta-analysis). Health impacts include eye irritation, respiratory diseases, cardiovascular effects, and premature death. This paper, unlike studies such as Hernandez-Avila and others (1995), who focused only on hospital costs, assesses a wide range of health benefits of reducing air pollu- tion: (i) reduced cost of illness (COI); (ii) reduced losses in productivity; (iii) willingness to pay (WTP) for reduced acute and chronic morbidity effects; and (iv) willingness to pay for mortality effects associated with acute and chronic exposure. ii In each case the WTP concept captures aspects of the value of avoiding death and illness (for example, the pain and suffering avoided) above and beyond foregone earnings and COI (used here to refer to avoided medical costs). The largest single contributor to the benefit es- timate is WTP for premature death. Because of the debate over using WTP for valuing health benefits, in particular when WTP is estimated using the Contingent Valuation Method (CVM), we compute the health benefits both including and excluding this benefit category. Specifically, we present three sets of benefit estimates. The 'high estimate', the most comprehensive one, includes WTP to avoid illness, as well as avoided illness costs (COI) and reduced losses in productivity, to value reduced morbidity. Avoided premature mortality is valued using WTP. The 'central estimate' includes the same comprehensive measure of the value of reduced morbidity, but values avoided premature mortality using foregone earnings, a lower bound to WTP. The 'low estimate', the most conservative, values morbidity using COI and productivity measures alone and premature mortality using foregone earnings. The high and central estimates vary depending on the income elasticity used to transfer WTP es- timates for morbidity and mortality from other countries to Mexico. Income elasticities of 1.0 and 0.4 are presented; however, we view the 1.0 elasticity as our central estimate. Table E.3 summarizes the benefits of each control scenario, where results for ozone and PM,( are added together. Adding the benefits of these two pollutants is appropriate because the estimates for each pollutant controls for the level of the other pollutant. The central esti- mate of the annual benefits of a 10 percent reduction in ozone and PM1O is $759 million. High and low estimates of the value a 10 percent reduction are $1,607 million and $154 mil- lion, respectively. Obtaining air quality compliance (AQS1) offers benefits of approximately $2 billion per year, with high and low estimates of benefits of some $4 billion and $400 mil- lion, respectively. Table E.3: Summary of Benefits From Each Scenario for Ozone and PM10 Combined (in million US$ per year, 2010 values in 1999 prices, income elasticity 1.0) Estimates 10% 20% AQS1 AQS2 High 1607 3184 3952 7636 Central 759 1489 1928 3580 Low 154 275 368 618 The estimates presented in table E.3 clearly show that the calculated benefits associated with air pollution reduction provide an economic basis for expenditures to further reduce pol- luting emissions. Exactly hoW much is open to debate. Ideally, a study like this on economic benefits should be combined with estimates of emission abatement costs to determine an economically justifiable level of abatement. Hence, developing a cost-benefit model is the next logical step. Table E-4 presents alternate estimates of health benefits, as well benefits from avoiding environmental contingencies, for ozone and PMl0 separately. This is particularly useful as it iii shows that the health benefits of PM10 reductions are roughly an order of magnitude higher than those of ozone. Table E.4: Benefits from Reducing Air Pollution: Four Scenarios for Ozone and PM,o (in million US$ per year, 2010 value in 1999 prices, 3 percent discount rate) Scenario 10% 20% AQSI AQS2 Income elasticity 1.0 0.4 1.0 0.4 1.0 0.4 1.0 0.4 Ozone Health benefit estimate 1, including morbidity (Prod. Loss + COI +WTP) and WTPformortality 116 183 232 365 717 1129 794 1250 Health benefit estimate 2, including morbidity (Prod. Loss + COI +WTP) and human capital losses for mortality 75 114 151 228 465 706 515 782 Health benefit estimate 3, including morbidity (Prod. Loss + COI) and human capital losses for mortality 18 18 35 35 109 109 121 121 Environmental contingencies benefits 36 36 45 45 45 45 45 45 PM,0 Health estimate 1, including' morbidity (Prod. Loss + CO +WTP) and WTP for mortality 1451 2549 2903 5098 3186 5595 6793 11931 Health benefit estimate 2, including: morbidity (Prod. Loss + COI +WTP) and human capital losses for mortality 644 1184 1289 2367 1414 2598 3016 5540 Health benefit estimate 3, including: morbidity (Prod. loss + COI) and human capital losses for mortality 96 96 191 191 210 210 448 448 Environmental contingencies benefits 4 4 4 4 4 4 4 4 Prod. loss = Productivity losses; COI = Direct Cost of Illness; WTP = Willingness to Pay. iv 1. Introduction The Zona Metropolitana del Valle de Mexico (ZMVM) (Mexico City Metropolitan Area) is one of the world's largest urban areas and one of the most notorious for its poor air quality. In the 1990s, however, efforts to control air pollution seem to have diverted the trend. Table 1.1 shows a decline in overall air pollution during the last decade (GDF 2000). However, with the exception of lead, carbon monoxide and sulfur dioxide, pollution levels are still far above air quality standards. Table 1.1 Number of Days Per Year that Ozone and PM10 Concentrations in Mexico City Satisfied Air Quality Standards 1995 1996 1997 1998 1999 Ozone 41 39 43 45 65 PM,0 273 186 212 176 345 Source: GDF (2000). The Third Air Quality Program 2001-2010 ("the Program") being developed by the Comision Ambiental Metropolitana (CAM) includes further initiatives to improve air quality. These air quality efforts are expected to improve the health of the population and also reduce the number of environment-related alerts in the ZMVM. One element of the program, and the purpose of this study, is the economic evaluation of the benefits gained from improving air quality. Air pollution has a range of negative effects on human health. It may also affect eco- nomic activity when excessive levels of pollution require Contingencias Ambientales (restric- tions on environmentally polluting activities). Health-related impacts include eye irritation, respiratory diseases, cardiovascular effects, and premature death. When a Contingencia Am- bientale is declared it limits activities of a range of manufacturing industries that generate emissions of air pollutants, and also restricts traffic. In sum, current air pollution exceeds permitted standards, and prospects for the future, with some exceptions (such as lead), will not improve without more active air quality man- agement and policies designed to improve air quality. Policy measures will be most useful if they can (1) be defined in terms of specific measures and the costs involved; (2) assess changes in air pollution using some form of air quality modelling; (3) assess and evaluate changes in air pollution impacts; and (4) rank the measures in terms of cost-effectiveness. The rationale for making air quality policies is the welfare gain from improvements in air quality. This report attempts to assess in economic terms the reduced impacts on human health and economic activity associated with four prespecified air quality scenarios with a time horizon of 2010. It is limited to the impacts of ozone and PM1o because these substances are the most important in terms of exceeding their standards and because relevant health in- formation is not fully available for other pollutants (such as NO2). Earlier efforts to assess the benefits of improvements in air quality for Mexico City by Hernandez-Avila and others (1995) estimated the direct medical costs and foregone income avoided if air quality standards were met. We use a different methodology for the economic valuation of reduced health risks (see section 6), and we use recent insights into the func- tional relationships between air quality and health impacts. We also deal with the economic benefits of avoiding Contingencias Ambientales (the use of environmental contingencies). Air pollution is the outcome of a range of physical processes. To understand its impacts one needs to know (1) the spatial and temporal patterns of pollutant emissions; (2) the chemi- cal, physical and meteorological processes in the airshed; and (3) the effects of pollutants on people's health, how many people are exposed to them, what economic activities suffer from environmental contingencies, and, if the scope of interest extends beyond the urban area, how natural systems (for example, ecosystems and climate) are affected. The structure of the report is as follows. Section 2 describes current emissions and air quality management in the ZMVM. Section 3 specifies the four air quality scenarios consid- ered in a model of current and future air quality. Section 4 models the population exposure to air pollution and estimates the number contingency measures invoked. Section 5 discusses the functional relationship between exposure and health, and derives exposure-response func- tions specific to the ZMVM. Section 6 covers the economic valuation of the air quality sce- narios set out in section 3 in terms of both the reduced health impacts and the reduced num- ber of Contingencias Ambientales. Section 7 discusses the results. 2. Emissions and Current Air Quality in the ZMVM For a quantitative understanding of air quality it is necessary to have an insight into the spatial and temporal pattern of emissions. The present study does not perform atmospheric transport modelling as this is outside the scope of the study. We will instead assume scenar- ios for current and future air quality and exposure (see sections 3 and 4). To provide a context for this study, however, we briefly characterize the pollution emissions that are the root of the air quality problems in the ZMVM. This information also indicates the available options for improving air quality. We shall also give a brief overview of current air quality in the ZMVM. Finally, we discuss the environmental contingency program that is currently applied to deal with high air pollution levels in the ZMVM. Emissions In recent years a number of different emission inventories have been taken. Table 2.1 summarizes the emission inventory by sector for 1996. Table 2.2 summarizes the emission inventory for 1998, but excludes emissions from heavy industry and open-air refuse burning. 2 Table 2.1 ZMVM Emissions Inventory, 1996 (tons/year) NO, VOC PM,0 Industry 28,666 16,279 5,700 Services 7,832 234,991 337 Transport 84,961 193,100 7,745 "Natural sources" a 134,673 18,072 Total 121,459 579,043 31,854 a. Includes biogenic emissions, forest fires, and open-air refuse burning. Source: INE (1997). Table 2.2 ZMVM Emissions Inventory, 1998 (tons/year) NO, VOC PM10 Industrya 22,094 17,595 3,173 Area sourcesb 8,489 270,190 1,058 Transportc 142,603 198,253 8,545 Natural sourcesd 11,802 72,670 5,800 Total 184,988 558,708 18,576 a. Excludes heavy industry. b. Includes lubricant industry, solvent emissions, forest fires, and services sector, and others. c. Includes private vehicles, public transport, taxis, and trucks. d. Includes biogenic emissions and soil erosion. Source: Comision Ambiental Metropolitana (http://sma.df.cob.mx/inventario/emisiones 1998.htm on 25 July 2000). Ozone air pollution is formed from the emissions of NO,, and VOCs. The amount pro- duced depends on the amount and location of emitted pollutants; background pollution levels; atmospheric chemistry; geographical, climatological and meteorological characteristics; and atmospheric transport characteristics. Moreover, the chemistry of ozone formation is quite complicated and nonlinear: under certain conditions an increase in NO, emissions can reduce ozone concentrations. The origins of particulate pollution (PM1o) are less clear. PM1O may be emitted directly or formed from SO2 and NO, reacting with other substances in the atmosphere (secondary particle formation). Likely sources of directly emitted particles include building and construction (road construction), diesel trucks and buses, forest fires, open-air refuse burning, some manufacturing industries, and resuspension of road dust. The relationship between emissions and concentrations, however, is not straightforward due to secondary particle for- mation. The ambient concentration of air pollutants depends on the amount and location of emissions; the source dependent physical characteristics of the emitted PM1O and PM1O pre- 3 cursors such as SO2 and NO,'; background pollution levels (especially of ammonia); atmos- pheric chemistry; geographical, climatological and meteorological characteristics; and at- mospheric transport characteristics. Air Quality and the Programa de ContingenciasAmbientales(PCA) Most air quality information comes from measurement stations across the area-the Red Automatica de Monitereo Atmosferica (RAMA)-that compile time-averaged concentrations (see figure 2.1). The annual reports usually present this information in frequency tables giv- ing the percentage of a year that a certain concentration occurred, or as a single annual aver- age. Day-to-day air quality data available on the Internet include daily maximum concentra- tions for five pollutants-PM1O, ozone, SO2, NO,, CO-and an ultraviolet (UV) index) for five zones2: downtown, Northwest, Northeast, Southwest and Southeast. These concentra- tions are expressed in IMECA points (Indice Metropolitana del Calidad del Aire). Table 2.3 shows how concentrations relate to the indicator points (100 = standard).3 Figure 2.1 Measuring PM10 and Ozone Concentrations across the ZMVM v Pla Stl.on. Z .1nII1p.IkI.. and D.I.g.lon- If one would consider PM2 5-a smaller mass than PM1O-emissions of SO2 and NO. become more important since these substances can be converted into particulate matter (PM1.0) in the atmosphere. Furthermore, NO. and VOCs can be attached to existing particulate matter in the atmosphere. The indications are that these small parti- cles have disproportionately large health effects. However, given the lack of air quality information on PM2.5 (and appropriate epidemiological data) it impossible to take account of this. 2 http://sima.cor.mx/sima/df (April 2000). 3 For ozone, the IMECA indicator is proportional to ozone concentrations. For PM1O an IMECA number fol- lows from linear interpolation between the values indicated in the table. 4 Table 2.3 The IMECA Indicator System for PM10 and Ozone IMECA points 100 200 300 400 500 PMl0 [ig/m3 (daily average) 150 350 420 500 600 Ozone ppm (daily 1-hr. maximum) 0.110 0.232 0.355 0.477 0.600 Source: INE (April 2000) at http:///www.ine.gob.mx/dggia/indicatores. Figure 2.2 shows the trend in the IMECA points between 1990 and 1999 for ozone and between 1995 and 1999 for PM1O. Table 2.4 shows the number of days per year satisfying air quality standards. Figure 2.2 Daily Average Trends in the ZMVM for Ozone and PM10 Ozone, daily average 1-hour maximum 400 350 300 - - _ _ _ _ _ _ _ _ _ _ - _ _ 100 164. 1180 1652 173.2 169. -4 - 15.1 9 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 -Maximum * Average -Minimum AQS <= 100 IMECA units PM10, daily average 250 200 4_ __-_ E100Xf --L 100.3 *98.6 - 102.8 1 50 l_l____ _. _ 0 1995 1996 1997 1998 1999 -Maximum * Average -Minimum AQS <= 100 IMECA units Source: SMA (1999). 5 Table 2.4 Number of Days Ozone and PM,0 Levels Satisfied Air Quality Standards 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Ozone 37 12 34 41 21 41 39 43 45 65 PM10 -- = -- 273 186 212 176 345 Source: SMA (1999). For the period 1995-99, the highest value for ozone-0.349 ppm-was measured at the Pedegral station, which is located in the Southwest zone. Ozone air pollution at this station exceeded the standard for 276 days. The least polluted station was Chapingo in the Northeast, with a daily 1-hour maximum concentration of 0.210 ppm . Air quality standards for PM1O include a daily average (150 ,ug/m3) and an annual aver- age (50 ,ug/m3). All stations violated both standards apart from the annual average standard at the Pedregal and Coacalco stations (formerly Villa de las Flores). The highest concentra- tion-a daily maximum of 335 11g/m3 (190 IMECA)-was observed in the eastern zone at the Nezahualcoyotl station. The maximum annual average of 94 tg/m3 was observed at Xalostoc station. This station has traditionally recorded the highest particulate matter concentrations in the metropolitan area, exceeding particulate standards 58 days per year (16 percent of the year). As a result, in 1995 about 1.2 million people were exposed at least once a year to PM1O concentrations above 300 Ag/m3, the trigger for a Phase I contingency (see table 2.6). If air pollution goes above certain levels in one of the five zones a PCA is invoked (a contingency program). Table 2.5 describes the three levels of action: Precontingency, Phase I contingency, and Phase II contingency. Table 2.5 The Environmental Contingencies Program Target Precontingency Phase I contingency Phase llcontingency Public health Suspend outdoor Epidemiological surveillance Suspend activities in sport activities in and communication of rec- public offices, recrea- schools and ommendations to address tional activities and parks. health risks. public services. Transport sector Restrict traffic (no circulation Suspend use of all of hologram 11 vehicles hologram 11 vehicles every other day). and 80% of public ser- Suspend use of publicly vice vehicles. owned vehicles by 50%. Improve traffic circulation. Industry and services Reduce certain industrial ac- Reduce industrial activ- tivities 30-40%. Suspend ity by 50%. fuel distribution activities, red brick fabrication, and the thermoelectric power plant Jorge Luque. Public services / in- Suspend maintenance of urban infrastructure. frastructure mainte- nance Additional actions Surveillance and control of fires in natural and agricultural areas I landfills. Source: SMA (http://sma.df.gob.mx). 6 Precontingencias (precontingencies) apply to the zones where the corresponding thresh- old is exceeded. For ozone and the ozone-PM1O combination, Phase I applies to the entire ZMVM, while for PM1o alone Phase I applies only to the zone where the threshold is ex- ceeded. If the situation persists, however, the contingencia is extended to the entire ZMVM. Phase II applies to the entire ZMVM irrespective of which zone exceeds the threshold.4 Table 2.6 gives the air quality threshold levels applied since May 1998. Table 2.6 PCA Threshold Air Quality Levelsa Levels Ozone (IMECA) PMo0 (IMECA) Precontingencia 200-240 (0.233-0.281 ppm) 160-175 (270-300yg/rm3) Phase 1 240-300 (0.281-0.355 ppm) 175-300 (300-420 pg/M3) Phase II > 300 (>0.355 ppm) > 300 (> 420 Pg/M3) a. Since May 1998. 3. Air Quality Modelling Three strands of science are combined to address the research question of this report: air quality and exposure modelling, epidemiology, and economics. This section focuses on air quality modelling of current and future scenarios. The next section deals with exposure mod- elling. Figure 3.1 shows the basic elements of the two sections. Figure 3.1 Overall Approach for Exposure Modelling Air quality at Population Measurement Stations Economic activities Model of air quality across the ZMVM Exposure model 4For more information go to http://sma.df.gob.mx/contingencias2OOO. 7 Modelling of Current Air Quality The starting points for modelling air quality are the air quality measurements at specific locations in the ZMVM. Figure 2.1 shows the RAMA air quality measurement network.5 The empirical data from the measurement stations is used to derive an air quality data field for the entire ZMVM. Since the use of an emission database and an atmospheric transport model are beyond the scope of this study6, we apply a simple approach to generate the air quality fields. We use measurements at the stations between 1995 and 1999, spatial interpolation in a geo- graphical information system (GIS), and take the average over the institutional units (16 delegations in FD) and 28 municipalities in the State of Mexico. Because the measurement stations tend to be located in areas with high levels of pollu- tion, information from relatively good air quality areas is patchy. Therefore, interpolating only on the basis of measured data could give unrepresentative results. We avoid such inter- polation results by assuming pseudo air quality data at locations where low pollution is ex- pected.7 Another difficulty was selecting a reference air quality year for comparison with future air quality. Given the variability of meteorological conditions the reference air pollution year-represented as an air quality frequency distribution-was derived from the distribu- tions for 1995-99. This reference distribution preserves the following three baseline statis- tics: the average over the five-year interval; the standard deviation over the five-year interval; and the maximum concentration in the five-year interval (see Cesar and others 2000 for more details). We developed the following distribution metrics for air quality: * for ozone, a daily 1-hour maximum, a daily maximum 8-hour running average, and a daily average; * for PMIo, a daily average. Figure 3.2 presents estimates based on the regionally differentiated air quality models for annual average ozone (daily 1-hour maximum) and PM1o (daily average) over the ZMVM during 1995-99. 5 The network has 19 ozone stations and 10 PM1o stations. 6The Germnan Fraunhofer Institute developed such a model for CAM during the period of the present study. Unfortunately, results of this development were not (yet) usable for the present project. 7We consulted experts to make the best estimate of air quality across the ZMVM. 8 Figure 3.2 Regionally Differentiated Model for Air Quality by Region, 1995-99 (for 16 delegations in FD and 28 municipalities in the State of Mexico) Ozone PM,0 Annual average daily 1-hour maximum Annual average briS- ~ ~ , t Mun. & Deleg. 03 AA I h max in PPM Mun. & Deleg. 0.08 - 0.094 PM10 AA in ugIm3 0.094 - 0.109 40 - 50 0.109 - 0.123 50 - 60 0.123 - 0.138 60 - 70 0.138 - 0.152 70-80 o80 -95 Modelling Future Air Quality The previous section developed a baseline for air quality for the ZMVM thought to be representative of the end of the 1990s. Since a program to improve air quality would typically take some years to show results, we chose the year 2010 as the reference year for the future. So two research questions arise: * What reference or business-as-usual scenario for the air quality in 2010 do we use, assum- ing no air quality policy beyond current measures? * What future air quality scenario do we want to evaluate, assuming some air quality policy beyond current measures? The Future Reference Scenario For an assessment of future air quality one needs insight into factors that determine air quality. The main determinants are emissions and climate, although the latter is not likely to change significantly over the next 10 years. So, the key question is "How will emissions de- velop over the next decade?" In our speculations about future emissions we deal separately with ozone and PM1o, since the origins of these two problems are for the most part unrelated. Ozone pollution is generated in the presence of NO, (nitrogen oxides) and VOCs (vola- tile organic compounds or hydrocarbons) in the atmosphere, and depends on geographical, climatological and meteorological conditions. Most emissions of NO, and VOCs-50 to 75 9 percent-come from the use of gasoline vehicles and associated gasoline distribution sys- tems. As table 3.1 shows, emissions standards for new gasoline vehicles have helped to re- duce emissions. The share of modem cars with three-way catalysts (electronic systems) has risen to over 10 percent of the total Mexican fleet of gasoline cars (about 70 percent of all cars). In the ZMVM all new gasoline cars are equipped with catalytic systems. According to SMA (1999), however, 32 percent of the cars in the ZMVM are pre-1980 models. And we know that emissions from pre-1986 cars are ten times greater than 1999 cars. Hence, total fu- ture emissions depend on the rate at which old cars are replaced with newer models. Table 3.1 Environmental Characteristics of Cars Sold in Mexico around 1998 Year/model Features Percent of vehicle fleet Pre-1986 With carburator 37.2 1986-1992 Fuel injection 23.8 1992-1993 Catalytic converter 28.6 1994 and after Full electronic systems 10.4 Total 100.0 Source: Mexican Association of Vehicle Distributors (April 2000) as cited at www.tradeport. orQ/ts/ countries/mexico/isa/isar0013.html. As the composition of the future vehicle fleet in Mexico City is unknown, it is assumed here-in the absence of an integrated model on emission projections for 2010 for mobile sources-that emissions in 2010 will equal those of 1998. This could be the case, for in- stance, if all improvements in vehicle emissions were exactly offset by the growing impor- tance of road transport in the ZMVM. This would also hold for VOC emissions. Hence, in the absence of more detailed information on emission patterns, we assume that NOx and VOC emissions will be the same as at the end of the 1990s. Once again, we note the enor- mous uncertainty of baseline predictions.8 Table 3.2 Main Assumptions on Reference Scenario Pollutant Main observations Result NO, and There are two opposing trends: (i) increase in Due to lack of information it is as- VOCs cars, buses and other pollution sources; (ii) the sumed that the baseline situation in emissions per unit is decreasing over time. The 2010 is equal to current conditions. resultant of these two trends is inconclusive. PM,0 Origins of PM10 are uncertain. Trends in meas- Due to lack of information it is as- ured air quality (since 1995) are inconclusive. sumed that the baseline situation in 2010 is equal to current conditions. 8As mentioned in the previous section, meteorology is an important explanatory variable for actual air quality, but is unlikely to change during the time horizon of this analysis. We, therefore, assume meteorological condi- tions in 2010 to be similar to the meteorological conditions of the 1995-99 reference air pollution year. Economic variables are equally uncertain and different growth patterns will greatly influence the actual future emissions. I0 Changes in PM1O emission sources and their contribution to air quality is even more un- clear. Relevant sources of directly emitted particles include building and construction (for example road surface works), diesel engine vehicles (about 30 percent of all vehicles (SMA 1999)), forest fires, industry, and open-air refuse burning. Measurements of PM1o since 1995 (the year in which continuous PM1o air quality measuring started) do not indicate a trend, al- though in 1999 the number of days that standards were violated were the lowest in the five- year period. In the absence of any concrete trend data or integrated model on emission pro- jections for 2010 for fixed and mobile sources, we assume the reference case in 2010 to be equal to the 1998 baseline air quality. We note the arbitrariness of 2010 baseline. The Future Scenarios To examine the implications of different levels of pollution control we developed four alternative 2010 scenarios in addition to the 2010 reference scenario: * a 10 percent air pollution reduction scenario; * a 20 percent air pollution reduction scenario; * an air quality standard compliance scenario assuming air quality would improve to the standard (50 JIg/M3 for PMI0 and 0.11 ppm 1-hour maximum for ozone) in all locations in the ZMVM-the AQS 1 scenario; * an air quality standard compliance scenario superimposing the needed percentage decrease in concentrations in the most polluted areas (Xalostoc for PM1O and Pedregal for ozone) across the ZMVM (68 percent and 47 percent reduction in ozone and PM1o concentra- tions, respectively)-the AQS2 scenario. To enhance the potential plausibility of the scenarios, we compared them with air quality trends for the South Coast Air Basin of California (see box 3.1). The trends for this area show a decline in maximum ozone concentration of 24 percent in 10 years due to strict controls. Similarly, the PM1O annual average decreased by 40 percent, but maximum PMlo concentra- tions remained the same. Hence, the first two scenarios proposed seem plausible but the third and especially fourth scenario would be very difficult, but not impossible, to achieve. We do not consider the policies needed to achieve the concentration reductions in this study. II Box 3.1 Air Quality Trends for the South Coast Air Basin of California Year Ref Ozone _ Ref PM1o Year 1 hour max Stndrzd 8 hour max Stndrzd Year 24 hour max Stndrzd AA Stndrzd oDrn to WO DOM to YO ua/m3 to YO ua/m3 to YO 1980 0 0.49 1.00 0.34 1.00 1981 1 0.39 0.80 0.28 0.84 1982 2 0.40 0.82 0.27 0.79 1983 3 0.39 0.80 0.26 0.77 1984 4 0.34 0.69 0.25 0.74 1985 5 0.39 0.80 0.29 0.86 1986 6 0.35 0.71 0.25 0.75 1987 7 0.33 0.67 0.21 0.62 0 219 1.00 73 1.00 1988 8 0.35 0.71 0.26 0.77 1 289 1.32 82 1.11 1989 9 0.34 0.69 0.25 0.75 2 271 1.24 81 1.11 1990 10 0.33 0.67 0.19 0.58 3 475 2.17 67 0.91 1991 11 0.32 0.65 0.20 0.60 4 179 0.82 65 0.89 1992 12 0.30 0.61 0.22 0.65 5 649 2.96 62 0.85 1993 13 0.28 0.57 0.20 0.58 6 231 1.05 58 0.79 1994 14 0.30 0.61 0.21 0.62 7 161 0.74 56 0.76 1995 15 0.26 0.52 0.20 0.60 8 219 1.00 52 0.71 1996 16 0.24 0.49 0.17 0.52 9 162 0.74 52 0.71 1997 17 0.21 0.42 0.17 0.51 10 227 1.04 56 0.77 Parameter Ozone PM10 1 hour max 8 hour max 24 hour max AA R Square 0.87 0.78 0.06 0.81 Obs 18 18 11 11 Intercept 0.88 0.88 1.53 1.07 P-value >0.001 >0.001 0.004 >0.001 m -0.0238 -0.0216 -0.0498 -0.0400 P-value >0.001 >0.001 0.475 >0.001 Annual red. -2.4% -2.2% -5.0% -4.0% Decade red. -24% -22% -50% -40%/o Source: California Air Resources Board (p. 90, 1999). Although the definition of baseline air quality in 2010 is highly uncertain, the benefit analyses that follow are valid for the reductions in ambient air pollution associated with each scenario. This is because the dose-response functions used to quantify health benefits and the economic values applied to these benefits are independent of baseline levels of air pollu- tion-they depend only on changes in ambient pollutant concentrations. 4. Exposure Modelling and Contingency Estimates This section describes both the modelling of exposure and the estimation of environ- mental contingencies (that is, alerts). First, exposure of the population to pollution is mod- elled by combining the air quality maps (section 3) with information on population distribu- tion. This model will be used to estimate the health impacts of air pollution in the next sec- tions. Second, we estimate the number of environmental contingencies declared to value the economic cost of these alerts (in section 6). 12 Population The Mexico City Metropolitan Area (with a population of 17 million in 1995 (Instituto Nacional de Estadistica, Geografia e Informatica, INEGI, 1997) is composed of the Federal District (containing Mexico City and its 8.5 million inhabitants) and part of the State of Mex- ico. Demographic information used for the population distribution was obtained from Mexi- can National Institute of Statistics, Geography and Information (INEGI). Figure 4.1 presents a population distribution map of the ZMVM. The GIS working group at the Federal District Government provided political boundaries and geographic definitions. Each locality (repre- sented by a point) is assigned to a municipality (in the State of Mexico) or to a delegation (in the Federal District). Figure 4.1 Population Map for ZMVM by Municipality, 1995 (in the State of Mexico) or Delegation (in the Federal District) J - Population 1995 _7 14975 - 90543 90544 - 255838 266839 - 552183 552184 - 839692 839693- 1696609 Exposure of the Population By combining the map of population distribution and the maps of air quality it was pos- sible to assess exposure of the population to air pollutants. This assessment of pollution ex- posure should correspond with the format of exposure defined and used in the epidemiologi- cal studies that our exposure-response models are based on (section 5). Exposure-response relations are constructed from epidemiological information and air quality measurements made at monitoring stations. The statistical exposure-response relationships reflect several factors in the cause-effect chain between air quality and health effects. 13 One of these factors is actual exposure and inhalation. This element is important since indoor air quality differs greatly from outdoor air quality, and an individual's behavior de- termines what he/she actually inhales. In epidemiological studies (see meta-analysis in sec- tion 5), however, it is assumed that exposure is proportional to measured air quality at a spe- cific outdoor air quality measurement station. Actual exposure will, however, differ from measured concentrations, depending on human activity patterns. Exposure-response functions incorporate the behavior of people in the particular study area and we need to assume that the behavior of people in Mexico City is similar to that of the people in the study areas where the epidemiological studies are performed. Also other factors, such as state of health, age, diet, and so forth may lead to differences in the exposure-response functions in different locations. Due to the lack of information and knowledge as to how these characteristics influence the estimated functions we have to assume there are no differences between the characteristics of people represented in the epidemiological studies and those living in Mexico City. For exposure to PMIo, the metric we use is the annual average of the 24-hour average concentration in an area. For exposure to ozone, the metric we use is the annual average of the daily 1-hour maximum concentration in an area (see Cesar and others 2000 for more de- tails.) Figure 4.2 summarizes population exposure in the 1995-99 reference air quality scenario we developed in the previous section. The baseline scenario for exposure can be summarized by computing population-weighted exposure for each pollutant. For PM1o and ozone this is 64.06 pLg/m3 /person and 0.114 ppm/person, respectively. The 10 percent reduction scenario would lead to a reduction of 6.41 ,ug/m3 PM,W/person and 0.0114 ppm ozone/person, respec- tively. A 20 percent reduction would double this figure. The AQSl scenario would result in reduced exposures of 14.06 p.g/m3/person and 0.0702 ppm/person for PM1o and ozone, re- spectively. The AQS2 scenario would result in the reduced exposures of 29.99 jIg/m3 /person and 0.0778 ppm/person for PM1o and ozone, respectively. These results are summarized in table 4.1. This table indicates that for PM1o, a compliance strategy aimed at achieving the air qual- ity standard in each area and no further air quality improvement would have only a slightly higher benefit than a uniform 20 percent reduction in the annual average. Hence, an emission abatement strategy should target sources in highly polluted and populated areas of the metro- politan area. 14 Figure 4.2 Reference Scenario for Population Exposure to Ozone and PM10 for the ZMVM, 1995-99 Ozone, maximum and annual average daily 1-hour maximum Population exposed to the stated concentration 20,000,000 1 03-max 18,000,000 901% _ B 03_AA 16,000,000 14,000,000 12,000,000 10,000,000 _ ___ 8,000,000 6,000,000 - _ _ 4,000,000 - _____ 2,000,000 0.000 0.050 0.100 0.150 0.200 0.250 0.300 0.350 Ozone concentration (ppm) PM10, maximum daily and annual average population exposure Population exposed to the stated concentration 18000000 ----....... 16000000 -O-|PM10_max -O -PM1O AA 14000000 - _ . AQS 12000000 10000000 8000000 6000000 4000000 _____- __ 2000000 = _ _ _ , _ 0 0 50 100 150 200 250 300 350 PM15 concentration (:g/m3) 15 Table 4.1 Reduction in Population-Weighted Exposure for the Analyzed Scenarios Population weighted Population weighted exposure to PM1O Exposure to ozone Scenario (pg/rM3 /person) (ppm /person) 10 percent exposure reduction 6.41 0.0114 20 percent exposure reduction 12.81 0.0227 AQS compliance in each area-AQS1 14.06 0.0702 AQS compliance in worst area-AQS2 29.99 0.0778 Environmental Contingencies This section estimates the number of times a contingency is declared; that is, the number of days that ozone or PM1O concentration measured at each station exceeds the relevant stan- dard for the reference scenario and for each of the future scenarios. Implicitly it is assumed that a contingency is declared if the measured concentration is above the concentration levels stated for the contingency (see section 2 for details on the Contingencias Ambientales pro- gram). In practice this is not always the case, for instance when air quality on a specific day is expected to improve considerably because of changing meteorological conditions. There- fore, our predictions give an upper estimate of the number of contingencies. The highest ozone concentrations in the period 1995-99 have been observed at the Pedegral station. The five-year composite frequency distribution of the highest daily 1-hour maximum ozone concentrations for the whole ZMVM is shown in figure 4.3 together with the future scenarios (see section 3). Contingencia I is triggered at IMECA=240, which corresponds to 0.281 ppm. According to the frequency distribution a contingencia I would be invoked for 10 days, and a contingencia II would be invoked for 0 days. (In the 10 percent reduction scenario: 2 days for contingencia I. In the 20 percent reduction scenario: 0 days). Noted that the threshold levels from May 1998 are used to make this calculation. Table 4.2 gives the results based on pre-May 1998 threshold levels. Figure 4.3 Cumulative Frequency Distribution of Pedegral Station and Ozone Air Quality Scenarios, 1995-99 Station Pedregal 0 .4 0 0 - .. .. .. ...... . . ..... .... ......... . ... --------- -------- ---------------------------~~--- 0.350 - IF + Baseline , t ~~~~~~~~~~~~o 10% Red E0.300- o 20% Redi b \ - ~~~~~~~~~~~~~~at AQS 0.250- 0.200 0 10 0.000 0 50 100 150 200 250 300 350 400 Days 16 Table 4.2 Days above Ozone Daily 1-hour Maximum Standards and Contingency Stages Ozone (ppm) Baseline 10 % Red 20% Red At standard >0.355 0 0% 0 0% 0 0% 0 0% >0.281 8 3% 2 1 % 0 0% 0 0% >0.233 60 18% 25 8% 5 1 % 0 0% >0.110 AQS 319 87% 306 84% 285 79% 0 0% The highest frequency distribution for the PM1O daily maximum concentration in the ZMVM shows that air quality at that specific place would trigger one contingencia (see table 4.3 and figure 4.4). The highest values are measured at the stations Nezahualcoyotl or Xalostoc. Table 4.3 Days above the PM,0 Daily Maximum Standards and Contingency Stages PM10 (pg/m3) Baseline 10% Red 20% Red At standard >420 0 0% 0 0% 0 0% 0 0% >300_Neza 1 1 % 1 0% 0 0% 0 0% >270_Xal 2 1 % 1 0% 0 0% 0 0% >15OAQS 87 16% 59 12% 34 7% 0 0% Figure 4.4 Cumulative Frequency Distribution of Xalostoc Station PM,0 Air Quality Scenarios, 1995-99 Station Xalostoc 300 - - . ._ - F ~~~~~~~~~~~~~~~~~~~+ Baseline 4F ° 10% Red 250 a I 0% Red e t - ~~~~~~~~~~~~~~~~~at AQS ab 200- I.00 50 0 0 50 100 150 200 250 300 350 400 Days 17 5. The Physical Effect of Air Pollution This section deals with assessing the effect of improvements in air quality in the ZMVM on human health and environmental contingencies. We begin with an introduction to the health effects of ozone and PM10. Then we discuss the relationships between public health and air quality (exposure-response relations) and quantify the health benefits of the air qual- ity improvement scenarios in physical terms. The economic benefits of these health benefits and of contingencies will be quantified in section 6. Effects of Ozone and PM1o on Health Not all air pollutants have the same capacity to damage human health. The differences in toxicity are due to the physical and chemical properties of the components of pollution. First, we briefly discuss the types of health effects caused by exposure to air pollution. Then we describe the properties of PM1o and ozone as they relate to toxicity. More details can be found in Cesar and others (2000). In the second section we discuss the development of the exposure response functions that are used in this study in more detail. Health Effects Due to Short- and Long- Term Exposure to Air Pollutants Susceptibility to air pollution exposure varies greatly among individuals. Individual risk is determined by genetics, age, nutritional state, presence and severity of respiratory and car- diac conditions, and the use of medications. The variability in the estimates found in epide- miological studies may reflect these differences in the populations studied. A good example of variation in individual risk occurs in the evaluation of maximum expiratory flow in healthy children, children with minor respiratory disease and those with asthma, with and without pharmacological treatment, and all exposed to various environmental pollutants. The results show an association between exposure and disease only in children with asthma under phar- macological treatment, in other words, those children who are most seriously ill (Roemer and others 1999). Genetic susceptibility is another factor that could be associated with respiratory diseases (Moller, Schuetzle, and Autrup 1994). Age is an important factor as well, with pre- adolescents (<13 years) and the elderly (>65) at greatest risk (Wilson and Spengler 1996, Ghio and others 1999). Toxic effects attributable to short-term exposure to high levels of air pollution (hereafter "acute effects and acute exposure") vary widely. Episodes of high pollution and the associ- ated increases in diverse respiratory and heart diseases and death have been reported since the beginning of the industrial revolution. The most serious acute effect is mortality. Many re- ports describe an increase in total mortality (not including accidental death) associated mainly with exposure to particulate matter (PM), ozone, and sulfates. (Schwartz 1994a, Wil- son and Spengler 1996). Many studies report increases in mortality due to respiratory complications, and this re- lationship can obviously be related to exposure to air pollution. Many reports also claim an increase in death due to cardiovascular disease, which would also imply an indirect effect 18 from air pollution. Both causes of death are associated with exposure to PM, ozone, and sul- fates. Mortality attributable to exposure to air pollution occurs mainly in individuals who al- ready suffer from cardiac and/or respiratory diseases. Increased mortality in these groups oc- curs within one to five days following the hazardous exposure (Schwartz 1994a, Wilson and Spengler 1996). Short-term exposure to high levels of air pollutants is also associated with diseases of the respiratory tract, both upper and lower: bronchitis, pneumonia, chronic obstructive pulmo- nary disease, and cough with phlegm. Symptoms aggravated by exposure to certain pollutants such as ozone and PM include asthmatic attacks, cough without phlegm, and wheezing (Wil- son and Spengler 1996, Ghio and others 1999). Episodes of extremely high pollution documented in cities around the world have dem- onstrated the consequence of human exposure to high concentrations of air pollution. These episodes, however, occur sporadically, whereas exposure to low concentrations of pollutants over long periods of time is a daily phenomenon. Recent studies have focussed on establish- ing the effects of long-term exposure to low levels of air pollutants. Health effects due to long-term, low-level exposure to air pollution (hereafter "chronic effects and chronic exposure") are similar to those reported for short-term exposure to high levels of air pollution. A synthesis of the available information concerning chronic exposure is an extremely complex task because many different factors can cause the same symptoms. There are several reports of increased mortality related to chronic exposure, however, most cases involve mainly elderly individuals for whom respiratory and cardiovascular problems are already the principal cause of death (Pope and Dockery 1999). Increased respiratory dis- eases (such as bronchitis) have also been reported as associated with chronic exposure. In both acute and chronic exposure to air pollutants, populations are exposed to a com- plex mixture of compounds whose combined toxic effects could differ from that of each compound alone. A study performed on volunteers exposed to ozone with and without preex- posure to H2SO4 showed that the preexposed group suffered more severe toxic effects than the group that was not preexposed (Wilson and Spengler 1996). Particulate matter and ozone are often correlated spatially and over time, making it diffi- cult to separate the effects of the individual pollutants. The mixture of PM1O and ozone, how- ever, has proven more toxic than the individual compounds alone (Katsouyanni 1995). Un- fortunately models and protocols to analyze the different interactions among environmental pollutants are not yet available (Samet and Speizer 1993). Thus, it is not clear how much each pollutant individually influences elevated mortality and morbidity rates. As a result some cost-benefit studies have chosen to use one index air pollutant rather than estimating ef- fects for multiple air pollutants individually and then adding their effects to get a total air pol- lution effect. Focusing on a single pollutant provides a conservative approach to estimating air pollution effects. In fact, recent analyses (for example Thurston and Ito 1999) suggest that ozone and PM air pollution effects are relatively independent, since controlling for one pol- lutant has only modest effects on the concentration-response of the other. Thus, use of a sin- gle index pollutant underestimates the overall public health effects and monetary valuations 19 of air pollution changes. Given that the effect of ozone on mortality independent of particu- lates is still being debated, we re-evaluated the effect of ozone restricting the analysis to those studies that controlled for particles in the statistical analysis. Properties of PM,0 and Ozone Aerosol air pollutants (molecular aggregations) have been shown to be more toxic than gases. This is because gaseous compounds are eliminated by the respiratory system much more easily than aerosols, which are rapidly deposited or absorbed. (Wilson and Spengler 1996). PM1O In the field of air pollution epidemiology, there is now much more interest in the study of PMlo and PM2.5 particles, and the organic and inorganic compounds in them (Wilson and Spengler 1996, Pooley and Mille 1999). The particles produce toxic effects according to their chemical and physical properties. Their effects on susceptible individuals are much more severe than those produced in normal individuals (Schlesinger 1995, Wilson and Spengler 1996). The extent of particle penetration into the respiratory system is determined by particle size. Only particles less than 10 p.m in diameter enter the respiratory system. This is the rea- son for focusing on PM1o (particles less than about 10 [im).9 Once particles have entered the respiratory tract, depending on their size, they can accumulate in different sites. Evidence suggests that many of the health effects associated with PM1o can be attributed to even smaller particles (Pope and Dockery 1999, Ghio and Samet 1999). Since, however, most epi- demiological information refers to PM10, and for the ZMVM there is little air quality information on smaller particles, we restricted our analysis to PM1O. The chemistry of suspended particles complicates empirical epidemiology enormously and has not been analyzed in detail yet, so little epidemiological evidence is available on the influence of the chemical composition of particles. Since little is known about the chemistry of PM1o found in Mexico City, this aspect is not accounted for. This contributes to the uncer- tainties associated with exposure-response modelling. Ozone. Ozone is a poorly soluble but highly reactive gas. "Bad" ozone (as opposed to "good" ozone in the stratosphere) is mainly produced in the troposphere (ground level) by a series of sunlight-driven reactions involving oxides of nitrogen and volatile organic com- pounds. Inhaled ozone is partially depleted in the upper airways but a major fraction reaches the lower airways. In the body ozone can react with uric acid, which is secreted by human submucosal airway glands and is present in near millimolar/liter (mmolI1) concentrations of nasal surface liquid. Pryor and his colleagues have proposed that some of the toxic products of the latter reaction (hydroxyhydroperoxides, hydroxyaldehides) are important mediators of ozone effects on underlying epithelium. Bromberg (1999) has calculated that ozone per se does not even reach the epithelial cell apical membrane in conducting airways. 9 Actually, the metric is not the size, but the aeolic behavior of particles, as measured in equipment that mimics the human respiratory system. This metric comes close to size. 20 The proportion of ozone uptake attributed to surface liquid dynamics decreases progres- sively as a surface liquid thins and its reactivity with ozone diminishes, so that the highest epithelial tissue dose is predicted for the terminal bronchiole-respiratory bronchiole region, which is, indeed, a site of damage in ozone-exposed animals. Bronchoscopic sampling along airways also indicates that a substantial fraction (35 percent) of orally inspired ozone is taken up in the upper airway and trachea and that ozone in exhaled air is limited to the initially ex- pired volume representing airways dead space (Bromberg 1999). The toxicity of ozone inhalation in large airways is supported by evidence of ciliated cell loss and increased epithelial mitotic index in small animals, netrophilic inflammation in hu- mans, increased bronchial artery blood flow in sheep, and by the symptoms of cough and substernal pain exacerbated by deep inspiration in humans (Bromberg 1999). Development of Exposure-Response Models for Mexico City Meta-analysis Although the number of published studies on the health effects of air pollution has grown during the past decade, specific studies of the ZMVM are still limited. We, therefore, decided to summarize relevant international and national published reports through a meta- analysis, which combines the results from various studies to identify consistent patterns. Due to the rapid growth of the field of epidemiology since the 1960s, the number of publications is overwhelming and a classical narrative review is no longer appropriate for summarizing findings. Despite limitations, statistical analysis of compiled published results has become more common when dealing with an extensive offering of differing and inconclusive results. Identirication, Selection and Classfication of Bibliographical Information The meta-analysis involved an exhaustive search of published studies on human health effects due to exposure to ozone and PMIo using Medline, Pubmed, Biomed-net and Aries da- tabases. Manual library searches examined mainly Mexican publications. Not all the biblio- graphic material collected was appropriate for the statistical analysis. Criteria for inclusion were * peer-reviewed published papers evaluating the association between exposure to ozone or particles and clinically identifiable human health effects (biochemical and molecu- lar effects were not included); and * papers that quantified any type of particles: Total Suspended Particles (TSP), black smoke (BS), coefficient of haze (CoH), or any PM.' 0 0 We used the approach of Dockery and others (1993) to convert air quality expressed in these metrics into PM1O concentrations. 21 Criteria for exclusion were * papers that did not present information for the variance, standard error or confidence intervals of the association estimate; * reports based on small populations or excessively large confidence intervals or stan- dard errors; * papers that did not control for temperature and seasonal variation over the study time period; and * papers that did not correct for ozone effects when addressing PM1O and vice versa. According to these criteria, 126 publications were selected for the statistical analysis of ozone and PM1o health effects (the list appears in Cesar and others 2000). Exposure-Response Functions Most studies express the health effect (y) as a function of the degree of change in health and the measured change in air pollutant levels (AC). The calculation of the corresponding change in health impact (Ay) depends on the exposure-response (ER) functions from epide- miological studies. The ER-function estimations may differ from each other in several ways, for example, in the use of standard definitions of health endpoints, baseline populations and the functional form of the estimated relationship. Some studies assume linear relationships, while others use log-linear functions. The linear relationship is of the form y=a+ lBC (5.1) The log linear relationship is of the form: y = a * e Fc or, equivalently ln(y) = (x + ,-C (5.2) Despite some statistical limitations, results from different studies were transformed to percent changes in the health effect for each 10 units of variation in the pollutant concentra- tion. Pooled Estimates We obtained a single pooled estimate of the health effects reported from the selected studies by using a weighted average. Deciding on a method to obtain an average estimate is not an easy task. Estimates from different studies could be different because of random varia- tion and also because of a true difference coming from differences in exposure and suscepti- bility factors. To take into account heterogeneity of effects of reported studies we applied a random-effects model to pool the studies (DerSimionian and Laird 1986.) Random-effect models assume that the true effect is decomposed into the mean population effect and be- tween-study variability. With a random-effects model the estimate of the average value is the 22 weighted average of the study estimates taking into consideration the sampling error and the between-study variability. Note that the within-study variability is not taken into account and only the average estimate is used in the quantification of the health benefits. Since the analysis applies to Mexico City, articles based on Mexico City population were given double the weight of international cases because they are thought to better reflect the Mexican reality in terms of susceptibility and sociodemographic characteristics. The po- tential influence of long-term exposures on health, and especially in the reduction of life ex- pectancy, could be one of the most influential end points. This is discussed in more detail in box 5. 1. An example of mortality due to acute exposure to PM1o is presented in box 5.2. Table 5.1 summarizes the ER functions and the background rates for health effects as they are used in the present study. The exposure response coefficients in the second and third columns of the table come from the meta-analyses described above. (For some health end- points, a meta-analysis was not possible and the source of the estimate is a single study.) The studies used to derive each coefficient reported in table 5.1 may be found in Cesar and others (2000). In the next two sections we will include only nonoverlapping health endpoints to pre- vent double counting of the benefits from air pollution reduction. Box 5.1 Premature Mortality Due to Long-Term Exposure to PM10 Cross-sectional studies and cohort studies have been conducted to study the effect of long-term exposure to particles and premature mortality. Cohort studies are preferable to cross-sectional studies because cohort studies can control for other factors related to mor- tality such as smoking status or occupation. To date three cohort studies in the United States have followed a significant number of individuals for at least 8 and up to 17 years (Dockery and others 1993, Pope and others 1995, Abbey and others 1993). During the study period, air pollution data were gathered from local monitoring stations to estimate average pollution exposures for individuals within the study. The 1999 U.S. EPA Report to Congress on the Benefits and Costs of the Clean Air Act 1990 to 2010 and other authors prefer to use the Pope and others (1995) study, which is based on extensive evaluation of confounders, as well as a larger sample size and greater geographic coverage. This co- hort study found a concentration-response coefficient of 17 percent for a 24.5 ug/m3 in- crease of PM2 5, or a 6.6 percent increase for a 10 pg/iM3 increase of PM2 5. For compari- son with PM10 studies, this is equivalent to an increase in mortality rates of 3.84 percent for a 10 ,g/m3 increase of PM10. 23 Table 5.1 Best Estimates of Exposure-Response Functions for the ZMVM for the General Population (unless stated otherwise) Percent Percent change Background change per per 10,ug/m3 rates 10 ppb daily daily average (per 100,000 Endpoints 1-h max ozone PM1o persons) Notes Hospital admission Respiratory 3.76 1.39 411 Cardiocerebrovascular 0.98 0.60 403 Congestive heart failure - 1.22 5.1 Emergency room visits (ERV) Respiratory 3.17 3.11 3,168 Restricted activity days (RAD) Total (adults) - 7.74 646,050 Work loss days (adults)9 - 7.74 236,520 Assumed same as total RAD background rate Total (children)' 7.74 646,050 in adults Work loss days women due to RAD in childrenh - 7.74 332,000 Minor restricted activity days (MRAD) Total (adults) 2.20 4.92 780,000 Effects in Asthmatic' Asthma attacksa 2.45 7.74 12,740 Cough without phlegm (chil- dren) - 4.54 21,200 0. 1 * chronic cough without Cough with phlegm (children) - 3.32 2,120 phlegm Cough with phlegm & bron- chodilator usaged - 10.22 56,174 Some respiratory symptoms Same as cough (children) 0.66 - 21,200 without phlegm Lower respiratory symptoms 0.23 - 8,810 Respiratory symptoms Upper respiratory symptomsd 1.50 4.39 22,400 1 Lower respiratory symptomsd 2.20 6.85 9,000 1 Wheezed 1.32 - 10,600 Acute bronchitisd - 11.0 4,400 Morbidity-Chronic Exposure Chronic bronchitis, additional cases - 3.60 707 Chronic cough, prevalence (children) - 0.30 5,770 Mortality-Chronic Exposure Totale - 3.84 b 1 Mortality-Acute Exposure Total' 0.59 1.01 577.9 lnfant' - 3.52 3,133 Source: Cesar and others (2000), Summary Tables, 11.4. a. Included in MRAD (U.S. EPA 1999). b. Estimated with life expectancy and survival probability tables by 1-year age interval, see section 111.2.2. c. ER-functions to be applied to asthmatics in population only (5 percent of population). d. Included in RAD for PM,o (U.S. EPA 1999). e. Originally identified for people age 30+ but applied to all population. f. Not included in aggregated benefit estimates because of methodological problems of separating morality asso- ciated with acute exposure from mortality associated with chronic exposure. g. Assumed the same ER-function as RAD total adults. h. WLD in adult women due to RAD of their children. 1. Boletin de Informaci6n Estadistica. Dafos a la Salud. Secretarla de Salud, Septiembre, Mexico (1996). 24 Box 5.2 Percent Change in Mortality Due to Acute Exposure to PM10 Of all the toxic effects attributed to PM10, death has been the most thoroughly documented. Death due to the acute effects of air pollution occurs generally between one and five days after the hazardous exposure. Since the 1950s studies have recorded increased mortality associated with high levels of pollution. This analysis includes the major studies carried out in the Americas, Europe, Australia and Asia since 1970. The figure below shows the percent change in general mortality associated with an in- crease in air pollution. The percent change, considering all the cases, establishes an in- crease in mortality of between 0.06 and 2.82 percent. These data are for total, nonaccidental deaths. 6 5- 4 3 ~~~~~~~T T~~~~~~~~~~~~ 2 T E + T -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 Author Note: Percent change in general, nonaccidental mortality for each 10 pg/iM3 increase in PM1o. The numbers represent the following studies: 1. Anderson and others 1996 (London), 2. Ballester and oth- ers 1996 (Valencia), 3. Borja-Aburto and others 1997 (Mexico), 4. Bremner and others 1999 (London), 5. Dockery and others 1992 (St. Louis), 6. Dockery and others 1992 (Tennessee), 7. Gamble and Lewis 1996 (Chicago), 8. Gamble and Lewis 1996 (Utah), 9. Ito and Thurston 1996 (Chicago), 10. Kel- sall and others 1997 (Philadelphia), 11. Kinney, Ito, and Thurston 1995 (Los Angeles), 12. Lee and Schwartz 1999 (Seoul), 13. Mazumdar and Sussman 1983 (Pittsburgh), 14. Moolgavkar and Luebeck 1996 (Ohio), 15. Moolgavkar and others 1996 (Philadelphia), 16. Neas, Schwartz, and Dockery 1999 (Philadelphia), 17. Ostro 1995 (California), 18. Ostro and others 1996 (Santiago), 19. Pope and Kalk- stein 1996 (Utah), 20. Pope III 1999 (Ogdem), 21. Pope, Hill, and Villegas 1999 (Provo), 22. Pope, Hill, and Villegas 1999 (Utah), 23. Samet and others 1998 (Philadelphia), 24. Schwartz 1994c (Cincinnati), 25. Schwartz and Dockery 1992a (Philadelphia), 26. Schwartz and Dockery 1992b (Steubenvile), 27. Schwartz 1993 (Birmingham), 28. Schwartz 1994b (Detroit), 29. Schwartz 1994c (Ohio), 30. Simpson and others 1997 (Brisbane), 31. Spix and others 1993 (Erfurt), 32. Sunyer and others 1996 (Barce- lona), 33. Touloumi and others 1996 (Athens), 34. Touloumi, Samoli, and Katsouyanni 1996 (Athens), 35 Verhoeff and others 1996 (Amsterdam), 36. Wordley, Walters, and Ayres 1997 (Birmingham), 37. Zmirou and others 1996 (Lyon), 38. Castillejos and others 2000 (M6xico), 39. Cropper and others 1997 (Delhi), 40. Pooled estimate. Pooled Estimate of the Effect of PM10 in Total Mortality Mortality Mean C195% Total 1.01 0.83,1.19 25 Health Effects The morbidity avoided due to a reduction in exposure to PM,o and ozone are now calcu- lated as follows: I = AY * Yb ACpopw * Pop (5.3) With: I = Impact AY = ER-function coefficient (percent change in impact per unit of pollutant) Yb = Background health impact rate (impacts/100,000 persons) ACp,pw = Population weighted change in exposure (concentration/person) Pop = Population exposed (persons) Box 5.3 presents estimates of avoided hospital admissions for respiratory problems with a 10-percent reduction in annual average daily 1-hour maximum ozone concentrations. The avoided morbidity impacts are expressed in number of cases. When quantifying the avoided mortality impacts it is important to take into account that the exposure-response functions provide estimates of premature mortality rather than addi- tional deaths. The economic valuation of an additional death is quite different from the valua- tion of only a limited number of years of life lost (YOLL). Following ExternE (1999) we have assumed that acute and chronic premature mortality leads, on average, to 0.75 and 5 years of life lost per case respectively. The quantification of the number of YOLL related to mortality associated with acute exposure is thus equal to the number of premature deaths times the average YOLL (0.75 years). The quantification of the avoided YOLL related to mortality associated with chronic exposure is more complex as death occurs later. Therefore, the age-specific life expectancy and death rates are taken into account. A more detailed dis- cussion of the method followed is presented in Cesar and others (2000). Tables 5.2 and 5.3 present the morbidity health benefits for the air pollution reduction scenarios discussed in sections 3 and 4. Table 5.4 gives the mortality health benefits for the air pollution reduction scenarios. Box 5.3 Estimation of Avoided Hospital Admissions for Respiratory Problems Due to Ozone Pollution Improvements in 2010 * A 10 percent improvement of air quality results in a reduction of population weighted exposure of 0.011357 ppm/person (ppm relates to annual average daily 1-hour max ozone concentration), see section 4. . The background rate for this type of hospital admissions is 411 per 100,000 per- sons per year. * The exposure-response relation is 3.76 percent per 10 ppb ozone concentration change. So the number of avoided admissions is: 0.0376/10 change/ppb x 1,000 ppb/ppm x 0.00411 admissions/person x 0.011357 ppm/person x 18,787,934 persons = 3,300 admissions. 26 Table 5.2 Reduction in Morbidity Health Impacts Due to Ozone Pollution Reduction Scenarios for the ZMVM in 2010 Scenario Endpoints 10% 20% AQSI AQS2 Hospital admission Respiratory 3,300 6,600 20,404 22,597 Cardiocerebrovascular 842 1,684 5,207 5,767 Emergency room visits Respiratory 21,429 42,857 132,501 146,746 Minor restricted activity days Total (adults) 2,495,805 4,991,610 15,432,494 17,091,616 Effects in asthmatics Asthma attacksb 3,330 6,660 20,591 22,805 Some respiratory symptoms 404 809 2,501 2,770 (children) Table 5.3 Reduction in Morbidity Health Effects Due to PM10 Pollution Reduction Scenarios for the ZMVM In 2010 Scenario Endpoints 10% 20% AQS1 AQS2 Hospital admission Respiratory 688 1,376 1,510 3,221 Cardiocerebrovascular 291 582 638 1,361 Congestive heart failure (elderly) 0.36 0.71 0.78 1.66 Emergency room visits Respiratory 11,858 23,717 26,029 55,507 Restricted activity days Total (adults) 4,102,282 8,204,565 9,004,464 19,202,173 Work-loss days (adults) 998,116 1,996,233 2,190,854 4,672,035 Total (children) 1,630,710 3,261,421 3,579,391 7,633,112 Work-loss days for women due 428,269 856,537 940,045 2,004,662 to RAD in children Minor restricted activity days Total (adults) 3,148,315 6,296,630 6,910,516 14,736,794 Effects in asthmatics Cough without phlegm (children) 1,569 3,139 3,445 7,346 Cough with phlegm (children) 115 230 252 537 Chronic morbidity Chronic bronchitis, new cases 3,063 6,126 6,723 14,337 Chronic cough, prevalence 574 1,148 1,260 2,686 (children) 27 Table 5.4 Reduced Deaths or YOLL Related to Ozone and PMl0 Pollution Reduction Scenarios for the ZMVM in 2010' Scenario Endpoints 10% 20% AQSI AQS2 Mortality-acute exposure Total population-YOLL Ozone 546 1,091 3,374 3,737 Modality-chronic exposure Total population-YOLL PM10 14,131 28,261 31,016 66,143 a. 3% discount rate, average YOLL per death are 0.75 and five years for morality associated with acute and chronic exposures, respectively. 6. Economic Valuation of Scenarios Earlier studies suggested that improving air quality in Mexico City would bring limited benefits (Hernandez-Avila and others 1995). These studies, however, used a narrow defini- tion of health benefits. Estimates of the effects of air pollution on human health were quanti- fied for fewer endpoints than in section 5. In addition, health benefits were valued using a very narrow definition of benefits. Reductions in premature mortality were valued by the as- sociated increase in eamings (the human capital approach). Reductions in illness were valued using the savings in medical costs and reductions in lost work time that result from reducing illness (the Cost of Illness approach). In the present study we use a broader definition of the value of health benefits: In addition to valuing avoided illness costs and productivity losses, we estimate the amount that people are willing to pay to avoid the discomfort associated with illness and the disutility associated with premature death. This section first discusses the methods used to value health benefits in this study. Next we present the main results of the health benefit analysis and the economic benefits of a reduction in the number of contingen- cies experienced." Economic Valuation of Premature Mortality and Morbidity Economists value avoided premature mortality by the amount that people are willing to pay to reduce their risk of dying (Hernandez-Avila and others 1995). Ideally, "willingness to pay" (WTP) should capture the loss in satisfaction-from consumption, leisure time, interac- tion with friends and family-that occurs when life is shortened.'2 It should, in particular, ex- ceed the monetary value of the consumption (or income) lost when a person dies prema- turely. In studies conducted in the United States (Viscusi 1993) estimates of WTP to reduce risk of death suggest that WTP is between 8 and 20 times as large as the corresponding gain in earnings from living longer. (The methods used to estimate WTP for reduced risk of death are discussed below.) Ideally, changes in premature mortality should be valued using WTP. The value of earnings lost when a person dies prematurely (the Human Capital measure of " A more detailed discussion of the methodology, the assumptions and the results is presented in Cesar and others (2000). 12 Typically, WTP to reduce risk of death is expressed in terms of the Value of Statistical Life (VSL). If each of 10,000 people are willing to pay $100 to reduce their risk of dying by I in 10,000, they are together WTP $1,000,000 for risk reductions that sum to one statistical life. The $1,000,000 is termed the Value of a Statistical Life. 28 the value of reduced risk of death) will, in general, understate the economic value of reduced risk of death (Freeman 1993). Avoided morbidity is also valued by the amount a person will pay to avoid a particular illness. For minor illnesses (such as respiratory infections) the correct valuation concept is what an individual would pay to avoid the illness with certainty.'3 This should capture the value of the pain and suffering avoided, as well as the value of time lost due to illness (both leisure and work time) and the costs of medical treatment. In cases where some of these costs are not bome by the individual, and are therefore not reflected in his WTP, the value of the avoided costs must be added to WTP to measure the social benefits of reduced morbidity. It is often the case that the costs of medical treatment (hereafter referred to as COI) and time lost from work (Productivity Loss) are not borne by the sick person. We therefore measure the value of avoided mortality by WTP to avoid lost leisure time and the discomfort associ- ated with illness, but add to this the value of lost productivity and the costs of medical treat- ment. As in the case of mortality, it can be argued that the avoided value of lost productivity and medical costs alone will understate the economic value of reduced morbidity (Freeman 1993). Estimates of WTP to reduce risk of death and estimates of WTP to avoid illness unfortu- nately do not exist for Mexico. It is therefore necessary to transfer to Mexico estimates from countries where WTP studies have been conducted. When extrapolating estimates of WTP from one country to another, adjustments must be made for the effect of income on WTP. In general, WTP (both for mortality and for morbidity) should increase with income. In transfer- ring estimates from country A to Mexico the formula used is WTPMEXICO = WTPA [IncomeMExlco/IncomeA] where c represents the income elasticity of WTP-the percentage change in WTP corre- sponding to a one percent change in income. It should be acknowledged that there is considerable uncertainty regarding estimates of the income elasticity of WTP, especially for mortality, as well as uncertainty regarding the estimates of WTP themselves. We handle this uncertainty in two ways. First, we use two es- timates of the income elasticity of WTP-1.0, and 0.4. Holding WTPA constant, the 0.4 elas- ticity results in a larger WTP estimate for Mexico than the 1.0 elasticity. Indeed, when WTP estimates from the United States are transferred to Mexico using Purchasing-Power-Parity- adjusted income, an income elasticity of 0.4 implies a WTP for Mexico that is about the size of WTP in the US. We therefore view WTP estimates based on an income elasticity of 0.4 as upper bound estimates, and estimates based on an income elasticity of 1.0 as central case es- timates. Second, to handle uncertainty about the size of WTP, especially WTP for reduced mor- tality, we also present conservative, lower bound estimates of the value of mortality and mor- bidity. Specifically, we measure the value of avoided premature mortality using the Human Capital/foregone eamings approach, as well as by transferring estimates of WTP to reduce 13 In the case of rarer events, such as heart attack or stroke, the correct valuation concept is what a person would pay to reduce his risk of the illness occurring. 29 risk of death from other OECD countries to Mexico. In the case of morbidity we present es- timates of avoided illness costs and productivity losses alone (i.e., without WTP) as conservative, lower bound estimates of the benefits of reduced morbidity. To summarize, 3 sets of benefit estimates are provided for each of the four air quality scenarios analyzed. (See Table 6.1.) Health Benefit Estimate 1, the most comprehensive, in- cludes WTP to avoid illness, as well as avoided illness costs (COI) and reduced losses in productivity, in valuing reduced morbidity. Avoided premature mortality is valued using WTP. Health Benefit Estimate 2 includes the same comprehensive measure of the value of reduced morbidity, but values avoided premature mortality using foregone earnings, a lower bound to WTP. Health Benefit Estimate 3, the most conservative, values morbidity using COI and productivity measures alone and premature mortality using foregone earnings. Health Benefit Estimates 1 and 2 vary depending on the income elasticity used to transfer WTP estimates for morbidity and mortality from other countries to Mexico. For reasons described more fully below, we view Health Benefit Estimate 1, with an income elasticity of 1.0 used for benefits transfer, as a Central Estimate of the value of health benefits. Health Benefit Estimate 1, using an income elasticity of 1.0, is viewed as a High Estimate and Health Benefit Estimate 3 as a Low Estimate. We interpret Health Benefit Estimate 1, using an income elasticity of 1.0 for benefits transfer, as a "Central Esti- mate" of the health benefits of pollution reduction. This is motivated by the belief that the es- timates of WTP for reduced morbidity used in the analysis are more reliable (and certainly less controversial) than the estimates of WTP for reduced risk of death. It is also the case that WTP for reduction in risk of death is based on small risk changes. Applying a marginal WTP estimate to the large risk changes in AQS 1 and AQS2 may yield implausibly large estimates of WTP. Health Benefit Estimate 3, which uses a lower bound estimate for morbidity (= Pro- ductivity Loss + COI) and mortality (Human Capital approach), is a conservative, lower bound estimate to benefits. Table 6.1 Overview of Health Benefit Estimates Presented in the Study Income elasticity of WTP Components of Health Benefits 1.0 1. Health benefit estimate 1 including morbidity High estimate (Prod. Loss + COI +WTP) and WTP for mortality 2. Health benefit estimate 2 including morbidity Central estimate (Prod. Loss + COI +WTP) and human capital losses for mortality 3. Health benefit estimate 3 including morbidity Low estimate (Prod. Loss + COI) and human capital losses for mortality The following sections explain in more detail how productivity losses, COI, human capi- tal losses and WTP are measured. For a more detailed discussion of these methods see Cesar and others (2000). 30 Loss of Productivity Loss of productivity (also referred to as the "change in productivity" method or "effect on production") is a valuation method that computes the loss in output due to illness or some other event. The loss of productivity method is applied in two situations. First, environment-induced health effects reduce production. Foregone income as a re- sult of illness, which is assumed to be evenly distributed over time, is valued by using the av- erage wages in the formal and informal sectors (see Cesar and others 2000 for a further dis- cussion). Assuming an annual increase of 2.45 percent the formal and informal daily wage level in 2010 are US$ 24.8 and US$ 10.3, respectively (2010 values in 1999 prices).14 For those air pollution-related health effects where we are not able to identify the age of the peo- ple affected we use the population-weighted wages for the whole ZMVM population. This leads to an average daily wage of US$ 6.49 (2010 values in 1999 prices). Using total popula- tion-weighted wages to estimate morbidity damage for specific age groups in the ZMVM would lead to an underestimation of the damages if only adults or children are affected and an overestimation of the damages if only the elderly are affected. 15 For effects in the elderly we assume no economic losses occur. For effects in adults and children we use an adult population-weighted wage in of US$ 9.52 (2010 values in 1999 prices). The assumed number of "days lost" due to air pollution is presented in table 6.2 14 In the absence of data on the expected wage growth in Mexico we have used the growth in GNP per capita as a proxy by deducting the population growth rate (1 percent) from the expected growth in GNP (3.7 percent). 15 Effects resulting in loss of time for children leads to productivity losses in adults resulting from care for the children. 3] Table 6.2 Days Lost Per Case in Mexico City for the General Population (unless stated otherwise) Endpoints Days lost' Source Hospital admission Respiratory 8 ExternE (1999, 2000) Cardiocerebrovascular 45 ExternE (1999, 2000) Congestive heart failure (elderly) 7 ExternE (1999, 2000) Emergency room visits Respiratory 5 ExternE (1999, 2000) Restricted activity days Total (adults and children) ob Work-loss day I Minor restricted activity days Total (adults) 0 Assumed Effects in asthmatics Asthma attacks 1 ExtemE (1999, 2000) Cough without phlegm (children) 1 Assumed Cough with phlegm (children) 1 Assumed Some respiratory symptoms (children) 1 Assumed Chronic morbidity Chronic bronchitis, additional cases 7 Extrapolated from ERISCAC Chronic cough, prevalence (children) 7 Extrapolated from ERISCAC a. Including recovery days at home. b. The loss-of-productivity part is accounted for by the work-loss-day part of RAD. c. See Cesar and others (2000). Second, loss of productivity occurs during environmental contingencies (ECs) or alerts. As discussed in section 2.2 these alerts lead to temporary closures in production infrastruc- ture to avoid further air pollution. Here the loss of productivity is measured by estimating the difference in gross national product (GNP) with and without an environmental alerts. A dis- tinction has been made between production losses in the industry and the transport sectors. Cost of Illness The cost of illness for the different morbidity endpoints is quantified in terms of direct costs for treatment of an illness. These costs are dependent on the social security system. In Mexico the most common health systems are the public health insurance system for unin- sured people (Poblacion Abierta), the public health system for low-income employed people (IMSS), and the private health insurance system (Privado). Hemandez-Avila and others (1995) conducted a COI study for Mexico by including the costs of consultations, laboratory tests, and medication. The inflation-corrected numbers they obtained are presented in table 6.3. 32 Table 6.3 Cost of Illness Per Case in Mexico for the General Population (unless stated oth- erwise) (costs in US$, 2010 values in 1999 prices) Cost of Illnessa Public Endpoints services IMSS Private Others Averageb Hospital admission Respiratory 939 1,252 3,131 1,565 1,870 CardiocerebrovascularC 2,818 3,757 9,392 4,696 5,611 Congestive heart failure (elderly)d 939 1,252 3,131 1,565 1,870 Emergency room visits Respiratory 211 50 83 50 91 Restricted activity dayse 10 10 10 10 10 Minor restricted activity days Total (adults) ng ng ng ng ng Effects in asthmatics Asthma attacks 271 199 572 199 337 Cough without phlegm (children) ng ng ng ng ng Cough with phlegm (children) ng ng ng ng ng Some respiratory symptoms (children) ng ng ng ng ng Respiratory symptoms 10 10 10 10 10 Chronic morbidity Chronic bronchitis 153 168 326 168 218 Chronic cough (children) 169 136 279 136 190 ng = Assumed negligible. a. From Hernandez-Avila and others (1995). b. Based on National Health Survey, ENSA 111994, 18.6% public insurance, 31.9% IMSS, 33.3% private insurance, and 16.2% other. c. Assumed three times respiratory hospital admissions d. Assumed same as respiratory hospital admissions. e. Assumed same as respiratory symptoms. Only 46% of the work-loss-day portion of restricted activity days are valued with a COI component (Krupnick 2000). Human Capital Loss The human capital approach is used for valuing the lost productivity associated with mortality. This approach assumes that the value of a person is equal to what he or she would have produced, that is, the discounted present value of a person's expected future earnings. The value of lost productivity may also include nonmarket productivity, for example, the value of household production. Other dimensions of illness and death, such as pain, suffering, and loss of leisure are excluded. The difference between the productivity loss and human capital is that the former accounts for the short-term production losses caused by morbidity, while the latter focuses on the production losses in the long term caused by increased mortal- 33 ity. Therefore discounting is applied only to human capital loss. Following Pearce and Ulph (1995) a social discount rate of 3 percent has been used.'6 Willingness to Pay As noted above, economists consider the appropriate value of avoided premature mortal- ity to be what an individual would pay to reduce his risk of death. This should reflect the value of foregone consumption and leisure time and the loss of contact with loved ones. WTP can be estimated using the contingent valuation method (CVM) and hedonic pricing. CVM estimates the WTP or willingness to accept (WTA) a change in the quantity and/or quality of a good by using survey techniques (Mitchell and Carson 1989 and Hoevenagel 1994). In the questionnaire a hypothetical change is described and the respondent is asked di- rectly for his WTP or WTA this change. The main values derived through the CVM in this study are for health impacts such as asthma attacks and premature death. Hedonic pricing estimates the WTP/WTA through (i) the difference in the value of the same property in different areas with different environmental risks (property value differen- tial); or (ii) the wage differential people are willing to pay (or accept) for a decrease (or in- crease) in risk of death related to a job. In this study we focus on the WTP estimated through CVM and wage differential stud- ies. Because CVM is a costly and complex method, studies have been conducted in only a limited number of countries for a limited number of environmental goods and services. In the United States and Europe numerous CVM studies have been conducted on the WTP to re- duce the risk of mortality and morbidity impacts. Wage differential studies are also numerous in these countries. WTP/WTA estimates, based on both CVM and wage differential studies, are not available for Mexico. Therefore, we estimate the WTP for risk reduction through "benefit transfer" of WTP studies perfortned outside Mexico. Benefit transfer is an application of monetary values from a particular valuation study in one area to a policy decision setting in another geographic area (Navrud 1999). When trans- ferring values it is important to know when data from other studies can be used and under what conditions. The value that people attach to avoided health risks depends on the type and magnitude of risk (low probability, high impact), the extent to which the risk is experienced voluntarily, on cultural settings, income, and the futurity of the risk. The most important fac- tors for applying benefit transfer in this study are the level of real per capita income, repre- sented by purchasing power parity (PPP) per capita income, and the income elasticity of WTP (6).17 For reasons explained above we assume a best estimate for the income elasticity of 1.0.18 16 A more extensive discussion on the discounting can be found in Cesar and others (2000). " It seems plausible that risk preferences might also change with the status of development. However, we have not included this difference in risk aversion between countries in our benefit transfer due to lack of data. js A more elaborate discussion on WTP and benefit transfer is presented in Cesar and others (2000). 34 The original values of WTP for the morbidity endpoints are based on report from the Centre for Social and Economic Research on the Global Environment (Pearce and others 1999), U.S. EPA (1999), and ExternE (1999).i9 Table 6.4 shows the values derived for Mex- ico using different income elasticities. Table 6.4 WTP Estimates for Morbidity Impacts Obtained with CVM (in US$, 2010 values in 1999 prices) Income elasticity Health endpoint 0 0.4 1 Hospital admission Respiratory 550 330 153 Cardiocerebrovascular 550 330 153 Congestive heart failure (elderly) 550 330 153 Emergency room visits Respiratory 284 170 79 Restricted activity days Totala,b 49 35 21 Minor restricted activity daysc Total (adults) 49 35 21 Effects in asthmatics Asthma attacks 52 31 15 Cough without phlegm (children)c 49 35 21 Cough with phlegm (children)c 49 35 21 Some respiratory symptoms (children)c 49 35 21 Chronic morbidity Chronic bronchitis, new cases 422,991 253,899 118,074 Chronic cough, prevalence (children) 287 199 116 a. All RADs are valued using WTP. The work-loss days, a subset of RADs, are not valued sepa- rately to prevent double counting. b. We value restricted activity days (RADs) as a cough episode and thus equal to a minor re- stricted activity day. This underestimates the WTP for RADs. However, Pearce and others (1999) found the value of a bed day, which might be seen as an overestimate of the RAD as not all RADs are bed days, to be only 30% higher than WTP to avoid cough. c. Following ExternE (1999) we value most cases of effects in asthmatics, cases of respiratory symptoms, and minor restricted activity days (MRADs) as a cough case (episode). For MRADs this give the same value as used by U.S. EPA (1999). In estimating the WTP for premature mortality it is important to realize that the number of life-years lost due to acute and chronic exposure to air pollution is limited. Because we in- tend to value only the reduction in the number of life-years lost, the "years of life lost" (YOLL) approach has been applied.20 The YOLL approach is particularly recommended for deaths arising from exposure to air pollution. The value will depend on a number of factors, such as how long it takes for the exposure to result in an illness and eventually death. In this 19 See Cesar and others (2000) for a more detailed discussion 20 An alternative for the YOLL approach is the "value of a statistical life" (VSL) approach. A comparison of the two approaches is provided in the Cesar and others (2000). 35 study, the YOLL approach is used both in cases where the hazard has a significant latency period before impact (mortality associated with chronic exposure), and cases where the im- pact takes place within a short period of time (mortality associated with acute exposure). In estimating the values of mortality arising from chronic exposure to particulate matter we as- sume that latency and mortality risks are spread out evenly over a period of 15 years and the life time reduction is 5 years on average (ExternE 1999). For mortality associated with acute exposure in the general population we assume no latency and the average life time reduction to be 0.75 years (ExternE 1999). The resulting "value of life-year" lost (VOLY) based on benefit transfer using the PPP approach is reproduced in table 6.5. Table 6.5 Value of Life Year (VOLY) (in US$, (2010 values in 1999 prices, 3 percent discount rate) a VOLY mortality-acute exposure VOLY mortality-chronic exposure Male Female Male Female Income elasticity = 0 184,750 179,776 140,611 138,308 Income elasticity = 0.4 131,961 128,409 100,434 98,789 Income elasticity = 1.0 79,660 77,515 60,628 59,635 a. Using a VSL of 4.28, 3.06, and 1.85 million US$ (2010 values in 1999 prices) after benefit transfer of the European estimate of VSL of 3.36 million US$ (1999 values in 1999 prices) with income elasticity 0, 0.4 and 1, respectively. b. Differences in values for males and females arise from unequal distributions of survival prob- abilities and life expectancy. Results This section presents the main results of the economic valuation of the benefits of im- proving air quality in Mexico City. A distinction is made between economic health benefits for the air pollution reduction scenarios presented in section 4 and the benefits arising from the reduction in environmental contingencies. Both categories of effects are then aggregated and summarized. Economic Health Benefits As explained in earlier sections health-related benefits consist of effects resulting from reducing acute morbidity, mortality associated with acute exposure, chronic morbidity and mortality associated with chronic exposure. For the first three categories a straightforward procedure is followed by multiplying the physical health impacts (see section 5) by the mone- tary values for each unit of health impact (see previous section). The procedure to assess the damages from mortality associated with chronic exposure-combining information on life expectancy, age dependent mortality rates, and VOLYs-is more complicated. 36 Figure 6.1 presents the configuration of the health-related benefits of a reduction in air pollution. The productivity losses, the cost of illness, and the willing to pay are included in the estimated morbidity benefits. In the mortality benefit estimates either the human capital benefits or the WTP are included.21 Tables 6.6 to 6.9 present the results for WTP estimates derived by benefit-transfer with income elasticities of 0.4 and 1.0. The results show that the main health damages are caused by WTP for a reduction of health impacts. For PMIo, the economic value of preventing pre- mature death dominates the overall outcome. A summary of the damages including and ex- cluding WTP benefits are presented in table 6.10. Figure 6.1 Health-Related Benefits of Reduction of Air Pollution Loss of Productivity Acute & chronic + morbidity Cost of Illness + J Health-related Willingness to Pay Benefits Human Capital Loss Acute & chronic or mortality Willingness to Pay motly 21 In the literature used here, only studies that specifically differentiated between the costs categories (COI, productivity loss, and WTP) were considered. 37 Table 6.6 Health Benefits of Ozone Air Pollution Reduction in the ZMVM (in million US$ per year a for income elasticity 0.4, 2010 values in 1999 prices) Scenario Endpoints 10% 20% AQSI AQS2 Morbidity impacts (Prod. Loss + COI + WTP) Hospital admission Respiratory 7.43 14.86 45.95 50.90 Cardiocerebrovascular 5.25 10.50 32.45 35.94 Emergency room visits Respiratory 6.30 12.60 38.95 43.14 Minor Restricted activity days Total (adults) 86.49 172.98 534.79 592.28 Effects in asthmatics Some respiratory symptoms (children) 0.02 0.04 0.11 0.12 Lower respiratory symptoms 0.01 0.02 0.06 0.06 Respiratory symptoms Upper respiratory symptoms 3.67 7.34 22.69 25.13 Lower respiratory symptoms 2.14 4.28 13.22 14.64 Wheeze 1.23 2.46 7.60 8.41 Morbidity impacts (Prod. Loss + COI) Hospital admission Respiratory 6.34 12.69 39.22 43.44 Cardiocerebrovascular 4.97 9.94 30.74 34.04 Emergency room visits Respiratory 2.65 5.29 16.37 18.13 Effects in asthmatics Some respiratory symptoms (children) 0.00 0.01 0.02 0.03 Lower respiratory symptoms 0.00 0.00 0.01 0.01 Respiratory symptoms Upper respiratory symptoms 1.19 2.37 7.33 8.12 Lower respiratory symptoms 0.69 1.38 4.27 4.73 Wheeze 0.19 0.39 1.20 1.33 Mortality impacts-WTP Mortality-acute exposure (total) 70.07 140.13 433.25 479.83 Mortality impacts-Human capital losses Mortality-acute exposure (total) 1.67 3.34 10.33 11.44 Total - Morbidity (Prod. Loss+ COI +WTP) and 183 365 1129 1250 WTP for mortality Total - Morbidity (Prod. Loss+ COI +WTP) and 114 228 706 782 human capital losses mortality Total - Morbidity (Prod. Loss + COI) and 18 35 109 121 human capital losses mortality a. Discount rate 3%. 38 Table 6.7 Health Benefits of Ozone Air Pollution Reduction in the ZMVM (in million US$ per year a for income elasticity 1.0, 2010 values in 1999 prices) Scenario Endpoints 10% 20% AQS1 AQS2 Morbidity impacts (Prod. Loss + COI + WTP) Hospital admission Respiratory 6.85 13.70 42.35 46.91 Cardiocerebrovascular 5.10 10.20 31.53 34.93 Emergency room visits Respiratory 4.35 8.69 26.87 29.76 Minor Restricted activity days Total (adults) 55.21 104.42 322.83 357.54 Effects in asthmatics Some respiratory symptoms (children) 0.01 0.02 0.08 0.08 Lower respiratory symptoms 0.01 0.01 0.04 0.04 Respiratory symptoms Upper respiratory symptoms 2.68 5.37 16.60 18.39 Lower respiratory symptoms 1.56 3.13 9.67 10.71 Wheeze 0.82 1.64 5.06 5.60 Morbidity impacts (Prod. Loss + COI) Hospital admission Respiratory 6.34 12.69 39.22 43.44 Cardiocerebrovascular 4.97 9.94 30.74 34.04 Emergency room visits Respiratory 2.65 5.29 16.37 18.13 Effects in asthmatics Some respiratory symptoms (children) 0.00 0.01 0.02 0.03 Lower respiratory symptoms 0.00 0.00 0.01 0.01 Respiratory symptoms Upper respiratory symptoms 1.19 2.37 7.33 8.12 Lower respiratory symptoms 0.69 1.38 4.27 4.73 Wheeze 0.19 0.39 1.20 1.33 Mortality impacts-WTP Mortality-acute exposure (total) 42.30 84.59 261.53 289.65 Mortality impacts-Human capital losses Mortality-acute exposure (total) 1.67 3.34 10.33 11.44 Total - Morbidity (Prod. Loss+ COI +WTP) and 116 232 717 794 WTP for mortality Total - Morbidity (Prod. Loss+ COI +WTP) and 75 151 465 515 human capital losses mortality Total - Morbidity (Prod. Loss + CO) and 18 35 109 121 human capital losses mortality a. Discount rate 3%. 39 Table 6.8 Health Benefits of PM,0 Air Pollution Reduction in the ZMVM (in million US$ per yeara for income elasticity 0.4, 2010 values in 1999 prices) Scenarios Endpoints 10% 20% AQS1 AQS2 Morbidity Impacts (Prod. Loss + COI + WTP) Hospital admission Respiratory 1.55 3.10 3.40 7.25 Cardiocerebrovascular 1.81 3.63 3.98 8.48 Congestive heart failure (elderly) 0.00 0.00 0.00 0.00 Emergency room visits Respiratory 3.49 6.97 7.65 16.32 Restricted activity days Total (adults) 161.10 322.20 353.62 754.09 Work-loss days (adults) 14.32 28.63 31.42 67.01 Total (children) 64.04 128.08 140.57 299.76 Work-loss days (working women due to RAD in chil- 6.14 12.28 13.48 28.75 dren) Minor restricted activity days Total (adults) 109.10 218.20 239.47 510.68 Effects in asthmatics Cough without phlegm (children) 0.07 0.14 0.15 0.32 Cough with phlegm (children) 0.01 0.01 0.01 0.02 Chronic morbidity Chronic bronchitis, new cases 778.48 1,556.96 1,708.75 3,643.94 Chronic cough, prevalence (children) 0.26 0.52 0.57 1.22 Morbidity impacts (Prod. Loss + COI) Hospital admission Respiratory 1.32 2.65 2.90 6.19 Cardiocerebrovascular 1.72 3.43 3.77 8.04 Congestive heart failure (elderly) 0.00 0.00 0.00 0.00 Emergency room visits Respiratory 1.46 2.93 3.22 6.86 Restricted activity days Total (adults) 18.94 37.89 41.58 88.67 Work-loss days (adults) 14.32 28.63 31.42 67.01 Total (children) 7.53 15.06 16.53 35.25 Work-loss days (working women due to RAD in chil- 6.14 12.28 13.48 28.75 dren) Minor restricted activity days Total (adults) 0.00 0.00 0.00 0.00 Effects in asthmatics Cough without phlegm (children) 0.01 0.03 0.03 0.07 Cough with phlegm (children) 0.00 0.00 0.00 0.01 Chronic morbidity Chronic bronchitis, new cases 0.81 1.61 1.77 3.77 Chronic cough, prevalence (children) 0.15 0.29 0.32 0.69 Mortality impacts-WTP Mortality (Acute exposure) - Infant - - - - Mortality (Chronic exposure) - Total 1,408.53 2,817.07 3,091.71 6,593.13 Mortality impacts - Human capital losses Mortality (Acute exposure) - Infant - - - - Mortality (Chronic exposure) - Total 43.28 86.55 94.99 202.57 Total - Morbidity (Prod. Loss+ COI +WTP) and WTP for 2,549 5,098 5,595 11,931 mortality Total - Morbidity (Prod. Loss+ COI +WTP) and human 1,184 2,367 2,598 5,540 capital losses mortality Total - Morbidity (Prod. Loss + COI) and human capital 96 191 210 448 losses mortality a. Discount rate 3%. 40 Table 6.9 Health Benefits of PM,0 Air pollution Reduction in the ZMVM (in million US$ per yeara for income elasticity 1.0, 2010 values in 1999 prices) Scenarios Endpoints 10% 20% AQSI AQS2 Morbidity impacts(Prod. Loss + COI + WTP) Hospital admission Respiratory 1.43 2.86 3.13 6.69 Cardiocerebrovascular 1.76 3.52 3.87 8.24 Congestive heart failure (elderly) 0.00 0.00 0.00 0.00 Emergency room visits Respiratory 2.40 4.81 5.28 11.26 Restricted activity days Total (adults) 104.76 209.52 229.94 490.36 Work-loss days (adults) 14.32 28.63 31.42 67.01 Total (children) 41.64 83.29 91.41 194.92 Work-loss days (working women due to RAD in 6.14 12.28 13.48 28.75 children) Minor restricted activity days Total (adults) 65.86 131.72 144.56 308.28 Effects in asthmatics Cough without phlegm (children) 0.05 0.10 0.10 0.22 Cough with phlegm (children) 0.00 0.01 0.01 0.02 Chronic morbidity Chronic bronchitis, new cases 362.46 724.92 795.59 1,696.61 Chronic cough, prevalence (children) 0.21 0.43 0.47 1.00 Morbidity impacts (Prod. Loss + COI) Hospital admission Respiratory 1.32 2.65 2.90 6.19 Cardiocerebrovascular 1.72 3.43 3.77 8.04 Congestive heart failure (elderly) 0.00 0.00 0.00 0.00 Emergency room visits Respiratory 1.46 2.93 3.22 6.86 Restricted activity days Total (adults) 18.94 37.89 41.58 88.67 Work-loss days (adults) 14.32 28.63 31.42 67.01 Total (children) 7.53 15.06 16.53 35.25 Work-loss days (working women due to RAD in 6.14 12.28 13.48 28.75 children) Minor restricted activity days Total (adults) 0.00 0.00 0.00 0.00 Effects in asthmatics Cough without phlegm (children) 0.01 0.03 0.03 0.07 Cough with phlegm (children) 0.00 0.00 0.00 0.01 Chronic morbidity Chronic bronchitis, new cases 0.81 1.61 1.77 3.77 Chronic cough, prevalence (children) 0.15 0.29 0.32 0.69 Mortality impacts-WTP Mortality (Acute exposure) - Infant - - - - Mortality (Chronic exposure) - Total 850.28 1,700.55 1,866.35 3,980.02 Mortality impacts - Human capital losses Mortality (Acute exposure) - Infant - - - - Mortality (Chronic exposure) - Total 43.28 86.55 94.99 202.57 Total - Morbidity (Prod. Loss+ COI +WTP) and 1,451 2,903 3,186 6,793 WTP for mortality Total - Morbidity (Prod. Loss+ COI +WTP) and hu- 644 1,289 1,414 3,016 man capital losses mortality Total - Morbidity (Prod. Loss + COI) and human 96 191 210 448 capital losses mortality a. Discount rate 3%. 41 Table 6.10 Summary Health Benefits Due to Ozone and PM10 Air Pollution Reduction (in million US$ per yeara 2010 values in 1999 prices, 3 percent discount rate) Scenario 10% 20% AQS1 AQS2 Income elasticity 1.0 0.4 1.0 0.4 1.0 0.4 1.0 0.4 Ozone benefits Total - Morbidity (Prod. Loss+ COI +WTP) 116 183 232 365 717 1129 794 1,250 and WTP for mortality Total - Morbidity (Prod. Loss+ COI +WTP) 75 114 151 228 465 706 515 782 and human capital losses mortality Total - Morbidity (Prod. Loss + COI) 18 18 35 35 109 109 121 121 and human capital losses mortality PM10 benefits Total - Morbidity (Prod. Loss+ COI +WTP) 1,451 2,549 2,903 5,098 3,186 5,595 6,793 11,931 and WTP for mortality Total - Morbidity (Prod. Loss+ COI +WTP) 644 1,184 1,289 2,367 1,414 2,598 3,016 5,540 and human capital losses mortality Total - Morbidity (Prod. Loss + COI) 96 96 191 191 210 210 448 448 and human capital losses mortality Environmental Contingencies The economic effects of ECs in the ZMVM industry sector have been explored through an analysis of value added losses in affected industries. Using 1994 data, value added has been estimated here to decrease 39 percent during one day of PM1o contingency and 42 per- cent in an ozone contingency.22 The costs of a one-day contingency for PMjo are lower than for ozone as fewer industries are involved in PMjo (the service sector is left out in PM1O, but included in ozone). As mentioned in the earlier sections environmental contingencies (ECs) have two main cost components: production losses in industry and transportation. Productivity losses in the transport sector are much less straightforward to estimate, and the lack of data proved to be more severe in the transport sector. Given these constraints we focussed solely on production losses in industry. Knowing the value of production, the value added, and labor costs per day (percent participation in the total costs) allowed us to calculate the costs of production in a normal situation (without environmental contingency). From the normal situation scenario, the costs of ECs can be derived considering a decline of 33 percent of production per day (see table 6.1 1).23 22 This is larger than the average decrease of production of 33 percent for all industries combined, as explained in detail in Cesar and others (2000). The discrepancy stems from the fact that most workers go to work and get paid even on days when the production stops partially. The result is a higher cost of production per unit of prod- uct. For a more detailed analysis, see Cesar and others (2000). 42 Table 6.11 Value Added Losses to ZMVM Industry during PM10 and Ozone Contingencies (value added per day in thousands US$ 1995 values, 1 US$=9.28 Mex. Peso) Percent growth Percent growth in PM1O contin- in ozone VA loss in VA loss in gency contingency PM10 Ozone Industry subsectors (percent) (percent) contingency contingency Nonmetallic minerals extraction -42 -42 17.45 17.92 Food, beverages, and tobacco -41 -41 259.01 392.08 Textiles and leather industry -43 -43 68.61 99.89 Wood and wood products -45 -45 7.86 5.10 Paper industry and printing -46 -47 40.24 123.16 Chemical industries -43 -43 215.15 247.84 Nonmetallic minerals industries -38 -39 282.05 173.06 (no oil) Basic Metallic industry -57 -58 70.89 45.09 Metallic products -48 -47 236.40 276.49 Other manufacturing -41 -43 16.20 0.99 Services - -39 - 2.36 Electricity generation -22 -22 -7.26 -7.00 TOTAL -39 -42 1,306.57 1,376.94 Source: d= datgen. Emissions Inventory 2000; i= INEGI . Instituto Nacional de Estadistica, Geo- grafia e Informatica, Censo Industrial. Mexico (1997). To calculate the total losses resulting from environmental alerts, the costs per day has been multiplied by the number of days that the contingency is expected to be in place for each of the scenarios described in section 4. Here, the number of days for the first phase con- tingencies are counted. The precontingencies do not have explicit economic costs, while the second phase is never attained. The costs per year for phase I contingencies are given in table 6.12 in millions of US$ per year.24 Table 6.12 Industry Losses in Four Scenarios for PM10 and Ozone Phase I Contingencies (value added per year in millions of US$, 2010 values in 1999 prices) Scenario Scenario Base Scenario Scenario IlI IV case 1(10%) II (20%) (AQSI) (AQS2) Days with PM1O contingency 1.0 0.0 0.0 0.0 0.0 Production losses due to PM10 Phase I contingency 4.8 0.0 0.0 0.0 0.0 Days with ozone contingency 10.0 2.0 0.0 0.0 0.0 Production losses due to ozone Phase I contingency 45.4 9.1 0.0 0.0 0.0 24 An exchange rate of 4 Pesos to the US$ was taken for 1995. These numbers were converted to 1999 US$. 43 7. Conclusions and Recommendations The health benefits included in this study are:25 (i) reduced cost of illness, (ii) reduced productivity losses, (iii) willingness to pay (WTP) for reduced acute and chronic morbidity effects, measured using the contingent valuation method (CVM); and (iv) WTP for reduced mortality effects associated with acute and chronic exposure. The WTP concept in each case captures aspects of the value of avoiding death and illness (for example the pain and suffer- ing avoided) above and beyond foregone earnings and COI. The largest single contributor to the benefit estimate is WTP for premature death due to chronic exposure to air pollution. Given the continuing debate over the use of WTP for valuing health benefits, particularly CVM, we estimate the health benefits both including and excluding this benefit category. The human capital and COI can then be interpreted as lower bounds to WTP for reduced mortal- ity and for reduced morbidity, respectively. Table 7.1 presents the overall benefit estimates from this study at different income elas- ticities used in the benefit transfer of WTP estimates from Europe and the United States to Mexico. The central estimate of the annual benefits of a 10-percent reduction in ozone and PMIO is $759 million (1999 US$). High and low estimates of the value a 10-percent reduc- tion are $1,607 million and $154 million, respectively. Because estimates of the health benefits of reducing each pollutant control for the levels of other pollutants, it is appropriate to add the benefits of ozone and PM reduction together for each scenario. This is done in table 7.2, which summarizes the benefits of each control scenario, assuming an income elasticity of one in benefits transfer. The 'high' estimate given in Table 7.2 uses benefits of reduced morbidity in terms of Productivity loss, cost of illness and willingness to pay and of reduced mortality in terms of willingness to pay. The 'central' estimate is the same as the 'high' estimate except that mortality is measured in human capital losses rather than WTP. The 'low' estimate deviates from the 'central' case in that it excludes WTP estimates for morbidity. 25 In the literature on economic valuation of morbidity effects used here, only studies that specifically differentiated between the costs categories (COI, productivity loss, and WTP) were considered. 44 Table 7.1 Summary of Total Benefits of a Reduction in Air Pollution in Four Scenarios for Ozone and PM10 (in million US$ per year, 2010 value in 1999 prices, 3 percent discount rate) Scenario 10% 20% AQSI AQS2 Income elasticity 1.0 0.4 1.0 0.4 1.0 0.4 1.0 0.4 Ozone Health benefit estimate 1 including: Morbidity (Prod. loss+ COI +WTP) and WTP for mortality 116 183 232 365 717 1129 794 1250 Health benefit estimate 2 including: Morbidity (Prod. loss+ COI +WTP) and human capital losses for mortality 75 114 151 228 465 706 515 782 Health benefit estimate 3 including: Morbidity (Prod. loss + COI) and human capital losses for mortality 18 18 35 35 109 109 121 121 Environmental contingencies benefits 36 36 45 45 45 45 45 45 PM10 Health estimate 1 including: Morbidity (Prod. loss+ COI +WTP) and WTP for mortality 1451 2549 2903 5098 3186 5595 6793 11931 Health benefit estimate 2 including: Morbidity (Prod. loss+ COI +WTP) and human capital losses for mortality 644 1184 1289 2367 1414 2598 3016 5540 Health benefit estimate 3 including: Morbidity (Prod. loss + COI) and human capital losses for mortality 96 96 191 191 210 210 448 448 Environmental contingencies 4 4 4 4 4 4 4 4 Prod. loss = Productivity losses; COI = cost of illness; WTP = willingness to Pay. Table 7.2 Summary of Benefits From Each Scenario Using Income Elasticity of 1.0 (in million US$ per year, 2010 values in 1999 prices) Estimates 10% 20% AQSI AQS2 High 1607 3184 3952 7636 Central 759 1489 1928 3580 Low 154 275 368 618 The estimates in tables 7.1 and 7.2 clearly show that the calculated benefits associated with air pollution reduction give an economic basis for spending to further reduce polluting emissions. Exactly how much is open to debate. Ideally, this study on economic benefits should be combined with estimates of emission abatement costs to determine an economi- cally justifiable level of abatement. Hence, conducting a cost-benefit analysis is the logical next step. This would also necessitate further advances in atmospheric chemistry modeling for Mexico City, which is needed to compare costs from emissions reductions with benefits of lower concentrations of pollutants. 45 The current valuation study uses meta-analyses and benefit transfers. Additional epide- mological and health-economic studies in Mexico City would allow estimates of health bene- fits solidly based on local data. Also, the uncertainties regarding all the estimates presented above are considerable. Further research that allows a reduction of these uncertainties is highly recommended. It should also be noted that the monetary estimates of health benefits give a lower boundary to actual benefits. For instance, the human misery associated with a person suffer- ing from chronic pollution-related morbidity may be much larger than monetary estimates in- dicate. This is especially the case if this person is the main wage earner of a poor family who could slide further into poverty due to a lack of safety net. As other studies have found (U.S. EPA 1997, 1999), the health benefits from reducing ozone and PMIo are dominated by the benefits of reducing particulate matter. In the case of Mexico City, the benefits associated with reductions of PM1o are roughly an order of magni- tude higher than those of ozone. The results must, however, be interpreted with caution. Pol- lution control strategies that reduce the precursors of ozone, especially NOR, may also reduce particles. References Abbey, D., B. Wang, R. Burchette, T. Vancuren, P. Mills. (1995). Estimated Long-Term Ambient Concentrations of PM1o and Development of Respiratory Symptoms in a Non Smoking Popula- tions. Archives of Environmental Health 50(2):139-145. Abbey, D. E., F. Petersen, P. K. Mills, and W. L. Beeson. (1993). Long-term ambient concentrations of total suspend particulates, ozone, and sulfur dioxide and respiratory symptoms in a nonsmoking population. Archives of Environmental Health 48(l):33-46. Anderson, H. R., A. Ponce de Le6n, J. M. Bland, J. S. Bower, and D. P. Strachan. (1996). Air pollu- tion and daily mortality in London: 1987-92. BMJ. March 16; 312:665-669. Ballester, F., D. Corella, S. P&ez-Hoyos, and A. Hervas. (1996). Air pollution and mortality in Valen- cia, Spain: a study using the APHEA methodology. Journal of Epidemiology and Community Health 50:527-33. Borja-Aburto, V. H., D. P. Loomis, S. I. Bangdiwala, C. M. Shy, and R. A. Rascon-Pacheco (1997). Ozone, Suspended Particulates, and Daily Mortality in Mexico City. Am. J. Epidemiol. 145(3):258-68. Bremner, S. A., H. R. Anderson, R. W. Atkinson, A. J. McMichael, D. P. Strachan, J. M. Blandi, and J. S. Bower. (1999). Short term associations between outdoor air pollution and mortality in London 1992-4. Occup. Environ. Med. 56:237-244. Brombreg, P. A. (1999). Air Pollution and Health. S. Holgate, J. Samet, H. Koren, and R. Maynad, eds. Academic Press, U.S.A. Califomia Air Resources Board. 1999. "The 1999 Califomia Almanac of Emissions and Air Quality." Castillejos, M., V. H. Borja-Aburto, D. W. Dockery, D. R. Gold, and D. Loomis. (2000). Airbome coarse particles and mortality. Inhalation Toxicology 12(Suppl 1):61-72. 46 Cesar, H., K. Dorland, X. Olsthoorn, L. Brander, P. van Beukenrng, V. H. Borja-Aburto, V. Torres Meza, A. Rosales-Castillo, G. Oliaz Fernendez, R. Mufioz Cruz, G. Soto Montes de Oca, R. Uribe Ceron, E. Vega L6pez, P. Cicero-Fernandez, A. Citlalic Gonzalez Martinez, M. M. Nifio Zarazua, and M.A. Nifo Zarazua. (2000). Economic valuation of Improvement ofAir Quality in the Metro- politan Area of Mexico City, Institute for Environmental Studies (IVM) WOO/28 + WOO/28 Appen- dices (http://www.vu.nl/ivm) , Vrije Universiteit, Amsterdam. Cropper, L., N. B. Simon, A. Alberinni, and P. K. Sharma (1997). The health benefits of air pollution control in Delhi. American Journal of Agricultural Economics 79(5):1625-29. DerSimionian, R. and N. Laird (1986). Meta-analysis in clinical trials. Controlled Clinical Trials 7:177-188. Dockery, D. W., C. A. Pope III, X. Xu, J. D. Spengler, J. H. Ware, M. E. Fay, B. G. Ferris, and F. E. Speizer. (1993). An association between air pollution and mortality in six U.S. cities. The New England Journal of Medicine 329(24):1753-59. Dockery, D. W., J. Schwartz and J. D. Spengler. (1992). Air Pollution and Daily Mortality: Associa- tions with Particulates and Acid Aerosols. Environmental Research 59:362-73. Dusseldorp, A., Kruize, H. Brunekreef, H. Hofschreuder, P. Meer, G. Van Oudvorst A. (1995). Asso- ciations of PM,o and airborne iron with respiratory health of adults living near a steel factory. Am J Respir. Crit. Care Med. 152:1939-39. ExternE. (1999). ExternE - Externalities of Energy, Vol. Methodology 1998 update, (Ed. Holland M., J. Berry and D. Foster), European Commission DGXII, Science Research and Development, EUR 19083, ISBN 92-828-7782-5, Luxembourg, 518 p. Gamble, J., and J. Lewis. (1996). Health and respirable particulate (PMIo) air pollution: A causal or statistical association? Environ. Health Perspect. 104:838-50. Gobiemo del Distrito Federal.. (1998). Emisi6n de contaminantes atmosfericos de origen vehIcular. Technical Report. Secretaria de Medio Ambiente, Direccion General de Prevenci6n y Control de la Contaminaci6n. Subdirecci6n de Inventarios de Emisiones. Gobierno del Distrito Federal, Mexico City. Gobierno del Distrito Federal. (2000), 1999. El Mejor ano ambiental de la Decada, Secretario del Medio Ambiente ZMVM. Gobierno del Distrito Federal, Mexico City. Ghio, A. J. and J. M. Samet. (1999). Air pollution and health: metals and air pollution particles. Aca- demic Press. I-{ernandez-Avila, M., and others. (1995). Valuacion economica de los beneficios de reducir la conta- minacion del aire en la Ciudad de Mexico, INSP, Mexico. Processed. Hoevenagel, R. (1994). The Contingent Valuation Method: Scope and Validity. Thesis, Institute for Environmental Studies, Vrije Universiteit, Amsterdam. INEGI (Instituto Nacional de Estadistica, Geografia e Informantica). (1997). Anuario estadistico de los Estados Unidos Mexicanos. Mexico D.F. INE (Instituto Nacional de Ecologia) and CENICA. (1997). Primer Informe Sobre la Calidad del Aire en Ciudades Mexicanas 1996. Ministry of the Environment, Natural Resource and Fisheries (SEMARNAP), Mexico City. 47 Ito, Kasuhiko, and G. D. Thurston. (1996). Daily PMlo/mortality associations: An investigation of at- risk subpopulations. Journal of Exposure Analysis and Environmental Epidemiology 6(1):79-95. Katsouyanni, K. (1995). Health effects of air pollution in southem Europe: Are there interacting fac- tors? Environ Health Perspect 103(Suppl 2):23-27. Kelsall, J. E., J. M. Samet, S. L. Zeger, and J. Xu. (1997). Air Pollution and Mortality in Philadelphia, 1974-1988. Am JEpidemiol. 146(9):750-62. Kinney, P., and H. Ozkaynak. (1991). Associations of Daily Mortality and Air Pollution in Los Ange- les County. Environmental Research 54:99-120. Kinney, P. L., Kasuhiko Ito, and G. D. Thurston. (1995). A sensitivity analysis of mortality/PMo0 as- sociations in Los Angeles. Inhal. Toxicol. 7:59-69. Krupnick, A. (2000), Personal communication, Washington, D.C., Oct 17 2000. Lee, J. T., and J. Schwartz. (1999). Reanalysis of the effects of Air Pollution on Daily Mortality in Seoul, Korea: A Case-Crossover Design. Environ Health Perspect. 107(8, August):633-36. Mazumdar, S., and N. Sussman. (1983). Relationships of Air Pollution to Health: Results from the Pittsburgh Study. Arch Environ Health 38(1, January/February): 17-24. Mitchell, R. C., and R. T. Carson. (1989). Using Surveys to Value Public Goods: The Contingent Valuation Method, Resources for the Future, Washington, D.C. Moller, L., D. Schuetzle, and H. Autrup. (1994). Future research needs associated with the assessment of potential human health risk from exposure to toxic ambient air pollutants. Environ. Health Per- spect. 102 (Suppl 4):193-210. Moolgavkar, S. H., and E. G. Luebeck. (1996). A Critical Review of the Evidence on Particulate Air Pollution and Mortality. Epidemiology 7(4, July):420 -28. Moolgavkar, S. H., E. G. Luebeck, T. A. Hall, and E. L. Anderson. (1996). Air pollution and daily mortality in Philadelphia. Epidemiology 6(5, September):476-84. Navrud, S. (1999). Linking weak and strong sustainability indicators: Critical Loads and Economic Values. John Wiley & Sons Ltd. NCP (1998). La situacion demografica de Mexico, National Population Council (Consejo Nacional de Poblacion), Mexico DF, p. 9-18. Neas, L. M., J. Schwartz, and D. Dockery. (1999). A Case-Crossover Analysis of Air Pollution and Mortality in Philadelphia. Environ Health Perspect 107:629-631. Ostro, B.D. (1987), Air Pollution and Morbidity Revised: A Specification Test. Journal of Environ- mental Economics and Management 14:87-98. Ostro, B.D. (1989). Air Pollution and Acute Respiratory Morbidity-An Observational Study of Mul- tiple Pollutants. Environ Res. 50(2):238-47. Ostro, B. (1995b) Fine Particulate Air Pollution and Mortality in Two Southern California Counties. Environmental Research 70:98-104. Ostro, B., M. Lipsett, and J. Mann (1995'). Air pollution and asthma exacerbations among African- American children in Los Angeles. Inh Tox. 7:711-722. 48 Ostro, B. and S. Rothschild (1989). Air pollution and acute respiratory morbidity: an observational study of multiple pollutants. Environmental Research 50:238-247. Ostro, B., J. M. Sanchez, C. Aranda, and G. S. Eskeland. (1996). Air Pollution and Mortality: Results from a Study of Santiago, Chile. Journal of Exposure Analysis and Environmental Epidemiology 6(l):97-l 14. Pearce, D., R. Dubourg, B. Day, G. Atkinson, S. Navrud, R. Ready, 0. Kuik, F. Spaninks, X. La- bandeira-Villot, M. Vasquez Rodriguez, F.S. Machado, and S. Mourato. (1999). Benefit Transfer and the Economic Valuation of Environmental Damage in the European Union: With Special Ref- erence to Health. Report to the European Commission under the European Union's Environmental and Climate Change Research Programme (1994-1998), CSERGE, University College London and University of East Anglia, United Kingdom. Pearce, D. W., and D. Ulph. (1995). A social discount ratefor the United Kingdom. Centre for Social and Economic Research on the Global Environment, University College London, London. Processed. Pooley, F. D., and M. Mille. (1999). Air pollution and health: Composition of air pollution particles. Academic Press. Pope, C. A , and D. Dockery. (1999). Air Pollution and Health: Epidemiology of Particle Effects. S. Holgate, J. Samet. H. Koren, and R. Maynard, eds. Academic Press. Pope, C. A., and L. Kalkstein. (1996). Synoptic Weather Modelling and Estimates of the Exposure- Response Relationship between Daily Mortality and Particulate Air Pollution. Environmental Health Perspectives 104(4, April):414-20. Pope, C. A., A., M. Thun, M. Namboodiri, D. Dockery, J. Evans, F. Speizer, and C. Heath. (1995). Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am. J. Crit. Care Med. 151:669-74. Pope III, C.A., R. Hill, and G. Villegas. (1999). Particulate Air Pollution and Daily Mortality on Utah's Wasatch Front. Environ Health Perspect 107:567-73. Roemer, W., G. Hoek, B. Brunekreef, J. Haluszka, A. Kalandidi, and J. Pekkanen. (1998). Daily varia- tions in air pollution and respiratory health in a multicentre study: the PEACE project. Pollution ef- fects on asthmatics children in Europe. Eur Respir. J. 12(6):1354-61. Roemer, W., G. Hoek, and B. Brunekreef (1993). Effect of ambient winter air pollution on respiratory health of children with chronic respiratory symptoms. Am. Rev. Respir. Dis. 147:118-124. Roemer, W., J. Clench-Aas, N. Englert, G. Hoek, K. Katsoyanni, J. Pekkanen, and B. Brunekreef. (1999). Inhomogeneity in response to air pollution in European children (PAECE project). Occup. Environ. Med. 56:86-92. Samet, J., and F. Speizer. (1993). Introduction and recommendations: Working group on indoor air and other complex mixtures. Environ. Health. Perspect. 101(Suppl. 4):143-47. Samet, J., S. Zeger, J. Kelsall, J. Xu, and L. Kalkstein. (1998). Does Weather Confound or Modify the Association of Particulate Air Pollution with Mortality? Environmental Research Section 77:9-19. Schlesinger R. B. (1995). Toxicological evidence for health effects from inhaled particulate pollution: does it support the human experience? Inhalation Toxicology 7:99-109. 49 Schwartz, J. (1994a). Air Pollution and Daily Mortality: A Review and Meta-analysis. Environmental Research 64:36-52. Schwartz, J. (1994b). Air Pollution and Hospital Admissions for the Elderly in Detroit, Michigan. Am. J. Respir. Crit. Care Med. 150:648-55. Schwartz, J. (1994c). Total Suspended Particulate Matter and Daily Mortality in Cincinnati, Ohio. Environmental Health Perspectives 102(2, February): 186-89. Schwartz, J. (1994d). Ozone, PM1o, and hospital admissions for the elderly in Minneapolis-St Paul, Minnesota. Arch. Env. Health 49(5):366-74. Schwartz, J. (1993). Air pollution and daily mortality in Birmingham, Alabama. American Journal of Epidemiology 137(10):1136-1147. Schwartz, J., and D.W. Dockery (1992a). Increased mortality in Philadelphia associated with daily air pollution concentrations. Am. Rev. Respir. Dis. 145:600-604. Schwartz, J., and D.W. Dockery. (1992b). Particulate Air Pollution and Daily Mortality in Steuben- ville, Ohio. American Journal of Epidemiology 135:12-19. Schwartz, J., and R. Morris. (1995). Air Pollution and Hospital Admissions for Cardiovascular Dis- ease in Detroit, Michigan. American Journal of Epidemiology 142:23-35. Schwartz, J., D. Slater, T. Larson, and W. Pierson. (1993). Particulate Air Pollution and Hospital Emergency Room Visits for Asthma in Seattle. Am. Rev. Respir. Dis. 147:826-31. Simpson, R. W., G. Williams, A. Petroeschevsky, G. Morgan, and S. Rutherford. (1997). Associations between Outdoor Air Pollution and Daily Mortality in Brisbane, Australia. Archives of Environ- mental Health 52(6, November/December):442-54. SMA. (1999). Towards an Air Quality Programme 2000-2010, Clean Air Initiative. Secretaria del Medio Ambiente del Gobiemo del Distrito Federal, Comision Ambiental Metropolitana, Mexico DF. Spix, C., J. Heinrich, D. Dockery, J. Schwartz, G. Volksch, K. Schwinkowski, C. Collen. and H. E. Wilchmann. (1993). Air pollution and daily mortality in Erfurt, East Germany, 1980-1989. Environmental Health Perspectives 101(6, November):518-26. Sunyer, J., J. Castellsague, M. Saez, A. Tobias, and J. M. Ant6. (1996). Air pollution and mortality in Barcelona. Journal of Epidemiology and Community Health 50(1):S76-S80. Thurston, G. D., and Kasuhiko Ito. (1999). Epidemiological Studies of Ozone Exposure Effects. In S. Holgate, J. M. Samet, H. S. Koren, R. L. Maynard, eds., Air Pollution and Health. Cambridge UK: Academic Press. Toulourni, G., E. Samoli, and K. Katsouyanni. (1996). Daily mortality and "winter type" air pollution in Athens, Greece. A time series analysis within the APHEA project. J. Epidem and Commun. Health 50(Suppl. 1):s47-s51. Touloumi, G., S. J. Pocock, K. Katsouyanni, and D. Trchopolous. (1994). Short-term effects of air pollution on daily mortality in Athens: A time series analysis. Int. J Epidemiol. 23:957-67. U.S. EPA. (1999). The Benefits and Costs of the Clean Air Act, 1990 to 2010. Paper prepared for Congress, U.S. Environmental Protection Agency, Cincinnati, Ohio, USA. 50 Verhoeff, A. P., G. Hoek, J. Schwartz, and J. H. Van Wijnen. (1996). Air Pollution and Daily Mortal- ity in Amsterdam. Epidemiology 7(3, May):225-30. Wilson, R., and J. Spengler. (1996) Particles in Our Air. Harvard University Press, USA. Wordley, J., S. Walters, and J. G. Ayres. (1997). Short-tenn variations in hospital admissions and mortality and particulate air pollution. Occup. Environ. Med. 54:108-16. Zmirou, D., T. Barumandzadeh, F. Balducci, P. Ritter, G. Laham, and J. P. Chilardi. (1996). Short- term effects of air pollution on mortality in the city of Lyon, France, 1985-90. J. Epidemiol. Com- munity Health 50 (April): Suppl I :S30-S35. 51 Policy Research Working Paper Series Contact Title Author Date for paper WPS2765 Inequality Aversion. Heaith Adam '.' ! i January 2002 H. Sladovich Inequalities. and Health Achievement 37698 WPS2766 Autonomy. Participation. and Learning Gunnar S. Eskeland January 2002 H. Sladovich in Argentine Schools: Findings and Deon Filmer 37698 Their Implications for Decentralization WPS2767 Child Labor: The Role ot Income Rajeev H. Dehejia January 2002 A. Bonfield Variability and Access to Credit in a Roberta Gatti 31248 Cross-Section of Countries WPS2768 Trade. Foreign Exchange. and Energy Jesper Jensen January 2002 P. Flewitt Policies fn the Islamic Republic of David Tarr 32724 Iran: Reform Agenda. Economic Implications, and Impact on the Poor WPS2769 Immunization i,- Developing Countries Varun Gauri January 2002 H. Sladovich Its Poltica and C-, i-. I - l Peyvand Khaleghian 37698 reterminants WPS2770 Downsiz ng and Productivity Gains Martin Raina January 2002 H. Sladovich In the Public arnd Prvate Sectors Constance Newman 37698 of Colombia WPS2771 Exchange Rate ,'ii::: l ., Labor Norbert M. Fiess February 2002 R. lzquierdo Market Rigidities. and iN I,l Marco Fugazza 84161 William Maloney WPS2772 Governance Matters II Updated Daniel Kaufmann February 2002 E. Farnand Indicators for 2000-01 Aart Kraay 39291 Pablo Zoido-Lobatbn WPS2773 Household Enterprises in Vietnam: Wim P. M. Vijverberg February 2002 E. Khine Survival, Growth, and Living Jonathan Haughton 37471 Standards WPS2774 Child Labor in Transition in Vietnam Erc Edmonds February 2002 R. Bonfield Carrie Turk 31248 WPS2775 Patterns of Health Care Utilization in Pravin K. Trivedi February 2002 R. Bonfield Vietnam: Analysis of 1997-98 31248 Vietnam Living Standards Survey Data WPS2776 Child Nutrition. Economic Growth. Paul Glewwe February 2002 E. Khine and the Provision of Health Care Stefanie Koch 37471 Services in Vietnam in the 1990s Bui Linh Nguyen WPS2777 Teachers Incentives and Professional Gladys Lopez-Acevedo February 2002 M. Geller Development in Schools in Mexico 85155 Policy Research Working Paper Series Contact Title Author Date for paper WPS2778 Technology and Firm Performance Gladys L6pez-Acevedo February 2002 M. Geller in Mexico 85155 WPS2779 Technology and Skill Demand Gladys L6pez-Acevedo February 2002 M. Geller in Mexico 85155 WPS2780 Determinants of Technology Adoption Gladys L6pez-Acevedo February 2002 M. Geller in Mexico 85155 WPS2781 Maritime Transport Costs and Port Ximena Clark February 2002 E. Khine Efficiency David Dollar 37471 Alejandro Micco WPS2782 Global Capital Flows and Financing Ann E. Harrison February 2002 K. Labrie Constraints Inessa Love 31001 Margaret S. McMillan WPS2783 Ownership. Competition, and Geroge R. G. Clarke February 2002 P. Sintim-Aboagye Corruption: Bribe Takers versus Lixin Colin Xu 37644 Bribe Payers WPS2784 Financial and Legal Constraints to Thorsten Beck February 2002 A. Yaptenco Firm Growth: Does Size Matter? Asli Demirguc-Kunt 38526 Vojislav Maksimovic