Report No. 36946-PK Pakistan Strategic Country Environmental Assessment (In Two Volumes) Volume II: Technical Annex The Cost of Environmental Degradation in Pakistan -- An Analysis of Physical and Monetary Losses in Environmental Health and Natural Resources August 21, 2006 South Asia Environment and Social Development Unit South Asia Region Document of the World Bank Table of Contents PAGE TECHNICAL ANNEX I. URBAN POLLUTION AIR ........................................................................................... 1 11. WATER,SANITATION AND HYGIENE ...................................................................... 14 111. INDOOR AIR POLLUTION ........................................................................................ 20 IT. NATURALRESOURCEDAMAGESINPAKISTAN ...................................................... 24 V. COSTTOAGRICULTUREOFSOIL,SALINITY ........................................................... 24 VI. COST OFAGRICULTURAL SOIL EROSION .............................................................. 27 VII. kOST OFRANGELANDDEGRADATION ................................................................... 30 VIII. REFORESTATION .................................................................................................... 33 The Cost of Environmental Dewadation in Pakistan: An Analysis of Phvsicaland Monetary Lossesin Environmental Health and Natural Resources 1. This annex provides a comprehensive overview o f the data and methods used to estimate the costs o f environmental degradation in three environmental damage categories and three natural resource damage categories: (i)urban air pollution, including particulate matter and lead, (ii)water supply, sanitation and hygene, (iii) indoor air pollution, (iv) agricultural damage from soil salinity and erosion, (v) rangeland degradation, and (vi) deforestation. Data limitations have prevented estimation o f degradation costs at the national level for coastal zones, municipal waste disposal and inadequate industrial and hospital waste management. It i s hoped that the detailed analysis outlined in this annex will stimulate greater research on the costs o f degradation and will be used to update and refine damage estimates. 2. At the outset a number o f caveats are inorder. Unfortunately, much o fthe information neededto estimate social costs is lacking. To overcome this problem greater reliance has been placed on observable measures, such as financial costs. Such costs are generally thought to be lower bounds to social costs, but this is not always the case. Presumably, ifthose being damaged allocate their resources to mitigate that damage (e.g., by a doctor visit), we ,expect that this outlay i s at least equal to the monetary value o f the damage they are feeling, assuming they expect the visit to be very effective. Ifthey think it will not be effective, this outlay will underestimate social cost to them. On the other hand, if insurance picks up the tab, this link between preferences and outlays i s broken and medical (financial) costs could exceed social cost. Though this report relies ,on financial costs for a significant part o f the analysis, it i s likely that this substantially underestimatesthe true economic costs. 3. The other option i s to perform a benefit transfer. The simplest such transfer involves taking a social cost-based monetary measure o f damage from another country, say the U.S., which has many such estimates, and applying it to the target country, e.g., Pakistan. An adjustment is usually made for income differentials across the two countries, on the theory that income should constrain preferences for non- market goods (such as health), just as it does for market goods. This adjustment factor, called the income elasticity o f willingness to pay, has been estimated to be in the range o f about 0.3 to 1.0, the latter meaning that the transferred value would be proportional to the income differential; the former meaning that the value would be adjusted far less. This report uses benefits transfer, with a conservative income elasticity o f 1.O. 4. Finally, whenever there are choices to be made - between models, or parameters, we take a conservative approach and make assumptions that provide lower bounds to guard against exaggeration. I. URBANAIRPOLLUTION 5. The focus o f this report i s the health effects o f fine particulates (PM10 and PM2.5). There are three main steps to quantifying the health impacts from air pollution. First, the pollutant needs to be identified and its ambient concentration measured. Second, the number o fpeople exposed to that pollutant and its concentration needs to be calculated. Third, the health impacts from this exposure should be estimated based on epidemiological assessments. Once the health impacts are quantified, the value o f this damage can be estimated. BaselineData 6. About 35 % of the Palustan population of 149 million lives in urban areas (WDI, 2004). Only major cities have TSP and PMlO monitoring data. The monitoring data on PMlO concentrations are presented in Table 1. Population in each city was estimated from 1998 Census data (http://www.statpak.aov.pk) and applying a 1.034% annual population growth rate. Population in these major cities totals 21 million. Table 1:Average AnnualConcentrationsOf PMlO(pg/m') And PopulationInMajorPakistan 1st 2nd 3rd 4th cycle, Average Populationin Cycle, cycle, cycle, mean annual 2004, thousand mean mean mean Source: SUPARCO (2004). 7. Table 2 presents population figures for "other cities" with population above 100thousand that do not have PM monitoring data. These cities have a total population of almost 15.2 million. Excluding them from estimating the health impacts of urban air pollution would therefore represent a serious omission. Annual average P M 10 levels were therefore assignedto these cities based on scaling up of the World Bank modeling P M 10 concentrations (worldbank.org/nipr/Atrium/mapping.html.url),using an average factor for the major cities from the Table 1. Modeled concentrations were compared with actual monitoring data for each of these cities. Modeled concentrations and actual monitoring data were fond to differ by a factor of 1.8. To incorporate this uncertainty, we present two scenarios of annual average PMlO concentrations for cities without monitors. Inthe highercase we appliedthe average scaling factor equal to 1.8 to all cities without monitors. In the lower case we did not apply the scaling factor to the PMlO concentrations inthe cities with population 0.1-1 million. PM 10Average Annual Population Concentration,udm3 2004 Low case I Highcase Faisalabad(Lyallpur) 205 205 2337 Multan 193 193 1398 Hyderabad 187 187 1360 Guiranwala 199 199 1330 Sargodha 101 183 538 Sialkote 105 190 494 Bahawalpur 120 217 476 sukkur 122 221 389 Jhang 95 171 345 2 Sheikhu Pura 93 169 321 Larkana 122 221 319 L . ' Bahawalnager 116 209 130 Muridke 97 176 129 I Pahattan I 112 I 201 I 128 I Abdottabad 100 181 125 Jaranwala 100 181 122 Tando Adam 106 191 122 Chishtian 114 206 121 Daska 94 170 121 Khaitlur 117 211 121 8. The age distribution o f the urban population was estimated using urban population parameters from a 2003 Pakistan Demographic Survey. PMlOwere transformed into PM2.5 usingthe ratio 0.5 based on evidence from India (TEN, 2001). Concentration-Response Coefficients 9. The risk ratios, or concentration-response coefficients from Pope et a1 (2002) are likely to be the best available evidence of the mortality effects o f ambient particulate pollution (PM 2.5). These coefficients were applied by the WHO in the World Health Report 2002, which provided a global estimate of the health effects of environmental risk factors. For acute child mortality concentration- response coefficients from Ostro (2004) were applied. The mortality and morbidity coefficients are 3 presented in Table 3. We applied coefficients from Ostro (1994) per 100 000 of population where baseline data for Pakistanwere not available. Table3: UrbanAir PollutionConcentration-ResponseCoefficients AnnualHealthEffect Concentration- Per 1ug/m3annual response average ambient Coefficient concentrationof: Longterm mortality (% change incardiopulmonary and 0.8% * P M2.5 lungcancer mortality) Acute mortality children under five (% change in ARI 0.166% PMlO deaths) Chronic bronchitis (% change inannual incidence) 0.9% PMlO Respiratory hospital admissions (per 100,000 1.2 PMlO population) Emergencyroomvisits (per 100,000 population) 24 PMlO Restricted activity days (% change inannual incidence) 0.475% PMlO Lower respiratory illness in children (per 100,000 169 PMlO children) Respiratory symptoms (per 100,000 adults) 18,300 PMlO I I I I *Mid-range coefficient from Pope et a1(2002) reflecting a linear function o frelative risk. Source: Pope et a1(2002), Ostro (2004) for the mortality coefficients. Ostro (1994, 1998) andAbbey et a1(1995) for the morbidity coefficients. 10. Inorder to apply the mortality coefficients in Table 3 to estimate mortality from urban air pollution, baseline data on total annual cardiopulmonary and lung cancer deaths are required. The PDS 2003 data for this purpose. Urban crude mortality rate of 6.3 per 1,000 was used, along with an average cardiopulmonary and lung cancer mortality rate of 39 percent of total deaths.' A backgroundlevel of 7.5 ug/m3of P M 2.5 has been assumed. This is the same procedure used inthe World Health Report 2002 (WHO). No background level has been used for morbidity. An estimate of annual incidence of chronic bronchitis (CB) i s requiredin order to apply the CB coefficient in Table 3.1.4. In the absence o f CB incidence data for Pakistan, the rate i s fiom WHO (2001) and Shibuya et a1 (2001).Restricted activity days from ARI prevalence for the adult population is estimated to last 5 days (out o f a total of 7). This i s basedon global studies. 11. Other morbidity health endpoints considered are hospital admissions of patients with respiratory problems, emergency room visits (or hospital out-patient visits), lower respiratory infections in children, and respiratory symptoms. These are the most common health endpoints considered in most of the worldwide studies on air pollution. The coefficients are expressed as cases per 10,000 inthe absence of incidence data for Pahstan. 12. The health effects o f air pollution can be converted to disability adjusted life years (DALYs) to facilitate a comparison to health effects from other environmental risk factors. DALYs per 10 thousand cases of various health end-points arepresentedinTable 4. 'PDS 2003 presents that 23 percent o f all deaths are from heart attacks, pneumonia and asthma. Using GBD 2002 for EmroD and Sear DWHO regions, another 10 percent couldbe added for hypertensive mortality and strokes and 5 percent for other ARI. A correction factor o f 0.79 specific for the combination o fthe Amro Dand Sear D was applied reflecting deaths o fthose over 30. Annual ARI deaths for children under 5 were estimated applying GBD 2002 mortality estimates for the WHO subregions. ARI deaths are about 11-18 percent o f total urban children under 5 mortality, which is 79 per 1000 (estimated from PDS 2003 and GBD 2002). 4 Table 4: DALYsfor HealthEffects Health Effect DALYslostper 10,000 cases Mortality adults 75,000 Mortality children under 5 340,000 Chronic Bronchitis (adults) 22.000 Respiratory hospital admissions 160 Emergency Room visits 45 Restricted activity days (adults) 3 Lower respiratory illness inchildren 65 Respiratory symptoms (adults) 0.75 13. Table 5 presents the disability weights and average duration o f illness that have been used inthis report to calculate the DALYs as presented in Table 7. The weights for lower respiratory illness (LRI)' and chronic bronchitis (CB) are disability weights presented by the National Institute o f Health, United States.' Disability weights for the other morbidity end-points are not readily available, and are estimates by Larsen(2004a) based on weights for other comparable illne~ses.~Average durationo f CB is estimated based on age distribution inPakistan and age-specific CB incidence inShibuya et a1 (2001). Years lost to premature mortality from air pollution i s estimated from age-specific mortality data for cardiopulmonary and lung cancer deaths, and have been discounted at 3 percent per year. Average duration o f illness for the other health end-points i s from Larsen (2004a). Table 5: Calculation of DALYsPer Case ofHealthEffects Disability Average Duration o f Weight Illness Mortality 1.o (7.5 years lost) or 70 years lost for children under 5 Lower respiratory Illness - Children 0.28 10days ResPiratorv Svmntoms -Adults 0.05 0.5 davs IRestrictedActivityDays -Adults Activitv Davs I 0.1 I 1day dav I Emergency RoomVisits 0.30 5 days Hospital Admissions 0.40 14days* Chronic Bronchitis 0.2 20 years EstimatedHealth Impacts 14. Usingthe information inTables 2-5, the annual health effects o f ambient particulate air pollution inPakistan are presented inTable 6. Urbanair particulate pollution is estimated to cause around 22,000 premature deaths among adults and 700 deaths among children under 5 annually. Estimatednew cases of chronic bronchitis are about 8,000 per year. Annual hospitalizations due to pollution are estimated at close to 80 thousand, and emergency room visits/outpatient hospitalizations at 1600thousand per year. In See: htta:iiwww.tic.nih.eovldcvuiwe~~hts.xls The disability weight for mortality is 1.0. 5 terms o f annual DALYs lost, mortality accounts for an estimated 60 percent, chronic bronchitis around 5 percent, restricted activity days (RADS)for 7 percent, andrespiratory symptoms for 25 percent. Table 6: EstimatedHealthImpactofUrbanAir Pollution Healthend-points Total Total Cases 1 DALYs EstimatedCostof HealthImpacts 15. The estimated annual cost o f urban air pollution health effects i s presented in Table 7. Cost o f mortality i s based on the human capital approach (HCA) for children and the value o f statistical life (VSL) for adults. The range in cost i s due to the uncertainty o f monitoring data inPalustan. For VSL we used benefit transfer from the United States and Europe with a conservative approach using market exchange rate and income elasticity o f WTP equal to one. 16. A measure o f the welfare cost o f morbidity is often based on the willingness-to-pay (WTP) to avoid or reduce the risk o f illness. This measure i s often found to be several times higher than the cost o f medical treatment and the value o f time losses (Cropper and Oates 1992), and reflects the value that individuals place on avoiding pain and discomfort. There are however not a sufficient number o f WTP studies from Pakistan. For this reason, the cost-of-illness (COI) approach (mainly medical cost and value oftime losses) was applied as the only measure o fmorbidity cost (see cost o fmorbidity inTable 7). Table 7: EstimatedAnnual Cost ofHealthImpacts(BillionRs.) I Healthcategories I TotalAnnual Cost* I Percent of TotalCost* I (Mean) Mnrtalitv I Adults I 58-61 I 92.5% I Children under 5 0.83 1.3% Morbidity: Chronic bronchitis 0.06 0.1% Hostital admissions 0.28 0.4% Emergency room visitdoutpatient hospital visits 0.80 1.2% Restrictedactivity days (adults) 2.06 3.2% Lower respiratory illness inchildren 0.84 1.3% Respiratory symptoms (adults) 0.00 0.0% 6 Total cost of Morbidity 4.05 6.3% TOTAL COST (Mortality andMorbidity) 62 - 65 100 % * Percentagesare rounded to nearest percent. 17. Since for morbidity we estimated only cost of illness, no values were assigned to respiratory symptoms, therefore they have zero cost inthe Table 7. 18. Table 8 presents the estimated cost per case of mortality and illness (health end-point) based on the data inTable 9. Some of these require explanation. The value of time for adults is based on urban wages. Economists commonly apply a range of 50-100 percent of wage rates to reflect the value of time. The rate of Rs135 per day is an average urban wage inPakistan. Furthermore 75 percent of this rate has been applied for both income earning and non-income earning individuals. There are two reasons for applying the rate to non-income earning individuals. First, most non-income earning adult individuals provide a household function that has a value. Second, there i s an opportunity cost to the time of non- income earning individuals, becausethey could choosetojoin the paidlabor force.4 Table8: EstimatedUnit Cost by HealthEnd-Point(000' Rs.) Cost Per Case Cost-of-Illness Per Case Mortality adults 2740 Mortality children under 5 1260 Chronic bronchitis 7.92 Hospital admissions 3.41 Emergency room visitdoutpatient hospital visits 0.50 Restricted activity days (adults) 0.03 Lower respiratory illness inchildren 0.17 Respiratory symptoms (adults) 0.00 19. There i s very little information about the frequency o f doctor visits, emergency visits and hospitalization for CB patients in any country in the world. Schulman et a1 (2001) and Niederman et a1 (1999) provide some information on this from the UnitedStates and Europe.' Figuresderivedfrom these studies have been applied to Palustan. Estimated lost work-days per year i s based on frequency of estimated medical treatment plus an additional 7 days for each hospitalization and one extra day for each doctor and emergency visit. These days are addedto reflect time neededfor recovery from illness. 20. To estimate the cost o f a new case of CB, the medical cost and value o f time losses have been discounted over a 20-year duration of illness. An annual real increase o f 2 percent in medical cost and value o f time has been applied to reflect an average expected increase in annual labor productivity and real wages. The costs are discounted at 3 percent per year, a rate commonly applied by WHO for health effects. 4Some may argue that the value oftime based on wage rates should be adjusted by the unemployment rate to reflect the probabilityo f obtaining paid work. CB is a major component o f COPD which is the focus of the referenced studies. 7 Table 9: Baseline Data for Cost Estimation and Niedennan et a1(1999) Year RestrictedActivity Days: Average number o f days o f illness (per 10 cases) 2.5 Lower Respiratory Illness in Children: Number o f doctor visits 1 Total time o f care givingby adult (days) 1 Estimated at 1-2 hours per day LeadExposure 21. The annual cost o f lead (Pb) exposure i s estimated at 38-52 billion Rs per year, with a mean estimate o f 45 billion Rs, or 0.7 percent o f GDP in2004. This estimate i s basedon lead exposure from all sources (leaded gasoline, industry and possible other sources such as water, soil, paint and food) for the populationliving in cities with more than 100 thousand inhabitants, totaling nearly 36.3 million people or about 26 percent of the Palustanpopulation.6IQ losses (reduced intelligence) represent 78 percent o f total cost, and mild mental retardation (MMR) 15 percent. Cardiovascular mortality and elevated blood pressure morbidity inadults constitute only 7 percent o f total cost. Inaddition, lead exposure i s estimated to cause 660,000 annual new cases o f gastrointestinal effects in children, and 580,000 new cases o f anemia inchildren. This correspondsto the populationfor whichthe cost ofPMpollution was estimated. 8 22. The estimated cost o f lead exposure i s based on blood lead level (BLL) measurements in children from 1994-2003 and rough estimates for adults. A s little is known about current blood lead levels in the urbanpopulation, the cost estimates are highly uncertain. BaselineData 23. Lead exposure can come through breathing, drinking and eating lead particles. The original sources o f lead can include leaded gasoline, industrial lead emissions to air, water, and land (e.g., from smelters), leached lead from lead pipes carrying drinking water, contaminated food, lead paint, and pottery. And once in the environment, lead accumulates in soil and water. Significant amounts o f lead were found ingasoline in the 1990s (0.42 g/l inregular gasoline and 0.84 g/1 in high octane gasoline). In 2001-2002 all four major refineries announced that they would move to production o f lead free gasoline (Paul et al, 2003). However, it will be quite some time until the lead phase-out policy brings significant results. 24. A number o f studies were identified that analyzed the BLL inPalustan. Table 10 below presents major results o f the studies: Table 10: BloodLeadLevelsin ChildreninUrbanAreas Bloodlead, mean Year of Sample size Study (ug/dl) study 15.6 2000 400 White F.et al., 2001 21.6 2003 53 Hozhabri S.. et al. 2004 21.2 1994 374 Khwaija M.,2003 16.8 1994 126 Khwaija M.,2003 22.8 1995 230 Khwaija M.,2003 16.1 2002 138 White F. et al, 2001 25. No studies were identified with the BLL in adults. We applied BLL equal to mean BLL of 16 ug/dlinchildren, as recommended by WHO ifBLLinadults i s unknown (Pruss-Ustin et al, 2004). EstimatedHealthImpactsfromLeadExposure 26. BLLfrom Table 10was appliedto the Fewrell et a1model with two major adjustments. Some o f the studies o fBLL inPakistan date back to 1994 and the average BLL does not reflect the recent phase- out program o f lead ingasoline. While there is great uncertainty about how much BLL will decline from a lead phase-out program, international experience indicates that a program over a five-year period could lead to a 40 percent reduction in BLL. Applying this adjustment factor for a two year period gives an average BLL o f 16 ug/dl in children, which i s well above any threshold for health effects. Of course, a part o f the population has a BLL below 16 ug/dl as reflected by the standard deviation reported in the studies. 9 Table 11: Estimated HealthEffects per 1000People Age Rate o f event/ illness per 1000people O t o 4 I 5to 14 I 15+ IO (1) loss o f 1.95 Doints - 98 Notes: IQ reference to intelligence; BP =blood pressure; MMR= mild mental retardation. is in 27. The adjusted BLL and the range in standard deviation are applied in the model to estimate population BLL. The result suggests that an estimated 62 percent o f the children and 61 percent o f the adults have BLL>10 ug/dland an estimated 44-45 percent o f the children and adults have B L D 2 0 ug/dl. These are extraordinarily large estimates. Estimated health effects per 1000 children and per 1000adults are presented inTable 11. It i s assumed that IQ losses take place duringthe first 5 years o f a child's life, while gastrointestinal effects and anemia can occur inchildren under 15 years of age. Inadults, the health effects are increasedbloodpressure (BP) and anemia. 28. Loss of Intelligence: Studies have found an average loss o f 1.3 IQ per 5 ug/dl BLL in points children. Fewtrell et a1 (2003) apply a lower threshold o f 5 ug/dl BLL below which no IQ occurs, loss and an upper threshold o f 20 ug/dl BLL above which no further IQ losses are expected (i.e., a loss o f about 3.5 IQ points for BLL > 20 ~g/dl).~ we noted above, we adjusted the model, applying a As threshold o f 10ug/dl. For some children an IQ loss will cause mildmental retardation (MMR), occurring at an IQ o f 50-70 points. Thus children with an IQ70.5-73.5 points are at risk o f MMR from lead o f exposure. Following the assumption o f a normal distribution o f IQ in the population, the number of children with MMR from lead exposure i s estimated by applying the results in Table 3.2.3 to the estimated chldren with IQ o f 70-73.5 points. Estimated annual loss o f intelligence from lead exposure are presentedinTable 12, totaling about 2,200 thousand IQ points and 18,000 cases o f MMR. IOPoint Losses (thousands) IQ-lossof1.95pointsperchild IQ-lossof3.25pointsperchild (1) 199 (2) 239 IQ-lossof3.50pointsperchild (3) 1,649 Total Losses (thousands) 2,188 MMR Number o f children with MMR 17,000 29. Other Health Effects: Other health effects o f lead exposure are gastrointestinal effects in children, anemia inchildren and adults, and elevated blood pressure inadults resulting ina higher risk o f Fewtrell et a1 (2003) apply a linear relationship through the mid-point o f each 5 ug/dl BLL interval with a maximumloss o f 3.5 IQ points. 10 cardiovascular disease and mortality. Gastrointestinal effects and anemia are found to develop at BLL exceeding 60-80 ug/dl. Estimated number o f cases ispresented inTable 13. 0 to 4 5 to 14 15+ Gastrointestinal effects 220,000 440,000 Anaemia - children 195.000 385.000 30. Elevated Blood Pressure. The level o f cardiovascular disease resulting from lead exposure is estimated using the attributable fraction. The proportions of the adult population with different BLL are equated with the relative risks for cardiovascular diseases to calculate the attributable fraction (AF), which was associated with the increased bloodpressure. AF = C 5 years) 0.32-0.46 Estimatedfrom a combination o fpriority disease regstered from (pakistan.gov.pk) and PIHs2001/2002 Hospitalizationrate (% o f all diarrheal cases) -children 0.75 % Internationalexperience under 5 years Hospitalizationrate (% o f all diarrheal cases) -children 0.5 % under 5 years Percent o f diarrheal cases attributable to inadequate water 90 % WHO (2002b) supply, sanitationandhygiene DALYs per 100 thousand cases o f diarrhea in children 70 under5 Estimatedfrom WHO tables DALYsper 100 thousand cases o f diarrhea inpersons >5 100-130 years DALYs per 100 thousand cases of typhoid in persons 190-820 under 5 and over 5 DALYs per case o f diarrheal and typhoid mortality in 32-34 children over 5 andunder 5 EstimatedHealth Impacts from InadequateWater, Sanitation and Hygiene. 42. Table 21 presents the estimated health impacts from inadequate water, sanitation and hygiene. The estimates are based on the data inTable 19, taking into account the WHO estimate that 90 percent o f diarrheal illness i s attributable to water, sanitation and hygiene. loIt shouldbenotedthat someresearchers elect notto use age weighting, or reportsDALYswith and without age weighting. 15 "Low" "Hi&`'" Diarrheal child mortality 35,505 56,470 Diarrheal child morbidity 24,477,3 00 24,477,300 Diarrheal adults morbidity 34,42 1,400 50,067,900 Typhoidparatyphoid mortality 27,000 27,000 Typhoidparatyphoid morbidity 1,350,000 1,350,000 Total Disability Adjusted Life Years (DALYs)-mortality and morbidity (mean) 2,522,755 IIEstimatedP nualDALYs "Low" "High" %o fTotal DALYs Children (under the age o f 5 years) - increaseddiarrheal mortality 1,207,179 1,919,989 56-66 % . Children (under the age o f 5 years) - increaseddiarrheal morbidity 17,134 17,134 1% Population over 5 years o f age - increased diarrheal morbidity 34,077 66,090 2-3 % Children (under the age o f 5 years) - increasedtyphoid morbidity 965 965 0 Population over 5 years o f age - increased typhoid morbidity 6,989 6,989 0 Children (under the age o f 5 years) - increasedtyphoid mortality 340,000 340,000 12-16% Population over 5 years o f age - increased typhoid mortality 544,000 544,000 19-25% TOTAL I 2,150,344 2.895.167 1 I EstimatedCost ofDiarrhealHealthImpacts. 43. Total annual cost o f diarrheal illness associated with inadequate water, sanitation and hygiene i s estimated at 55-84 billion Rs. (Table 4.4). The cost o f mortality i s based on the human capital approach (HCA) since both diarrheal and typhoid mortality predominantly affects children. The cost o f morbidity includes the cost o f illness (medical treatment, medicines, and value o f lost time). Cost-of-illness i s presented in Table 24 for diarrheal morbidity. About 50 percent o f these costs are associated with the value o f time lost to illness (including care giving), and another 50 percent are from cost o f treatment and medicines. Table 23: EstimatedAnnualCost of DiarrhealIllness (Rs. Billion) EstimatedAnnualCost "Low" UHi h99 Mortality Children under age 5 diarrheal mortality 45 72 Morbiditv., Diarrheal morbidity 10 12 TOTAL ANNUAL COST 55 84 16 EstimatedAnnual Cost (BillionS/.) "Low" "High" Cost o f medical treatments (doctors, hospitals, clinics) 2 3 Cost o f medicines 3 3 Cost o f time lost to illness 5 6 TOTALANNUAL COST 10 12 Baseline data for the cost estimates are presented inTable 25. Table 25: BaselineDatafor CostEstimation I Baseline Source: Percent o f diarrheal cases treated at medical facilities PMS2001/2002 (children <5 years) and with medicines 82% Percent o f diarrheal cases treated with ORS (children <5 PIHS2001/2002 years) 54% Percent o f diarrheal cases treated at medical facilities 56-82% Estimated from a (population > 5 years) and with medicines combination o f PakistanDHS 1990/91 andpriority disease statistics at www.pakistan.gov.pk Average Cost o f doctor visits (urban and rural) -Rs. 50 Per consultations with Average Cost o fmedicines for treatment o f diarrhea -Rs. 50 pharmacies, medical Average cost of ORS per diarrheal case inchildren (Rs.) 30 service providers, and health authorities Average duration of diarrheal illness in days (adults and - . 3-7 PMS2001/2002 children) Hours per day of care giving per case of diarrhea inchildren 2 Assumption Hours per day lost to illness per case o f diarrhea in adults 2 Assumption Value o f time for adults (care giving and illadults) -Rs./hour 7.71 Based on urban and rural wages inPalustan (see Outdoor air pollution section) Hospitalizationrate (% o f all diarrheal cases) -children under 0.75 % Adjusted basedon 5 years evidence from Egypt Hospitalization rate (% o f all diarrheal cases) -population 0.5 % (Larsen 2004) over 5 Average length o f hospitalization(days) 2 Adjusted from (Larsen 2004) Time spent on visitation (hours per day) 4 Assumption Average cost o f hospitalization(Rs. per day) 500 Per consultations with hospitals Percent o f diarrheal cases attributable to water, sanitation and 90 % (WHO 2002b) hygiene 17 Typhoid and Paratyphoid 44. Recorded annual deaths o f typhoidparatyphoid inPakistan by age are available from PDS 2003. Using the India typhoid study by Sinha et al, 1999, annual cases o f typhoid fever were estimated. The resulting mortality rate i s about 2 percent, which i s consistent with the evidence from the literature (1 percent mortality in the US) (See Louisiana Department of Public Health, 2004). Results o f the estimation are presentedinTable 4.7 for the year 2000-2003. 0-4 years 5+ years All groups TyphoidRaratyphoid ~. ~. mortality 10,000 17,000 27,000 TyphoidRaratyphoidmorbidity 500,000 850,000 1,350,000 45. The estimated annual cost o f these illnesses i s presented in Table 27. Mortality i s 95 percent o f total cost. About 13 percent o f estimated morbidity cost i s from hospitalization and doctor visits, 52 percent i s from time losses for the illindividuals and their caregivers during illness. More than 70 percent o f the cost o f time losses i s associated with illindividuals and almost 30 percent with care giving. Table 27: EstimatedAnnual Cost of Typhoidmaratyphoid I EstimatedTotal Annual Cost (BillionRs.) I Mortality Childrenunder age 5 typhoid mortality 12.7 People over 5 typhoid mortality 23.5 Morbidity Cost o f Hospitalizationand doctor visits 0.2 I Cost o f Medication I 0.7 I I Total Annual Cost I 38.1 I Averting Expenditures 46. Inthe presence ofperceivedhealthrisks, individuals often take avertingmeasures to avoid these risks. Economists usually consider these measures a cost o f healthrisks. Ifconsumers perceive there i s a risk of illness from the municipal water supply, or from other sources of water supply they rely, some consumers are likely to purchase bottled water for drinking purposes, or boil their water, or install water purification filters. Bottled Water. Rosemann (2003) estimates that about 70 Million liters o f bottled water are sold in Pakistan annually and some market growth is predicted. We use 100 Million liters as a highbound bottled water consumption estimate. Total annual cost o f bottled water consumption i s estimated at 1-1.5 billion Rs. The lower bound represents a 75 percent mark-up o f average factory price. The upper bound represents an arithmetic average o f retail prices for the most commonly sold quantities o f bottles and containers. Average retail price was about 15 Rs./liter. 18 BoiZing of Water. According to the Luby et al, (2001) study for Karachi, 40 percent of households boil their drinking water, either all the time or sometimes. Table 4.9 presents the estimated annual cost ofboiling water for those households, totaling 2-5 billion Rs. per year. 47. Tables 28 and 29 present the data usedto estimate the annual cost of boiling drinlungwater. It i s assumed that the average daily consumption o f dnnking water per person i s 0.5-1.0 liters among households boiling water. Residential cost of energy i s estimated based on data from Pakistan 2004 Statistical Yearbook. The average stove efficiency is for electric, natural gas and kerosene. Lower efficiency was applied for wood stove. Table28: EstimatedAnnual Cost ofBoilineDrinkineWater EstimatedAnnualCost "Low" UHi hV Annual cost--using fuel wood for water boiling 1.37 3.43 Annual cost--using kerosene for water boiling 0.05 0.14 Annual cost--using natural gas for water boiling 0.40 0.99 Annual cost--using other types of energy for water boiling 0.14 0.35 TotalAnnual Cost 1.96 4.90 Data: Percentageofhouseholds that boil their drinkingwater 40 % Lubyet al, (2001) Average daily consumptiono f drinlungwater 0.5-1.0 Liters perperson per day Percent of households usingfuel wood for coolung 69% Census 1998 Percent of households using:kerosene for cooking; 4% Cost of kerosene 7 Rs/liter Average cost of fuel wood 70 Rs per 40 Pakistan 2004 Statistical kg Yearbook Table 30: EstimatedTotalAnnual HouseholdCost ofAvertingExpenditures TotalAnnualCost (BillionRs.) "Low" 6 b H ihV Cost of bottled water consumption 1.o 1.5 Cost ofhousehold boilingdrinhngwater 2.0 5.1 Totalannualcost 3 6.6 19 111. INDOORAIR POLLUTION 48. There are two main steps in quantifying the health effects. First, the number of people or households exposed to pollution from solid fuels needs to be calculated, and the extent o f pollution, or concentration, should ideally be measured. Second, the health impacts from this exposure should be estimated based on epidemiological assessments. Once the health impacts are quantified, the value o f this damage can be estimated. TraditionalFuelUse 49. The Pakistan Census conducted in 1998 reports that 86 percent o f rural and 32 percent o f urban households use solid fuels for coolung inPakistan. The national weighted average i s about 67 percent. HealthRiskAssessment 50. Desai et a1 (2004) provides are review o f research studies around the world that have assessedthe magnitude o f health effects from indoor air pollution from solid fuels. The odds ratios for acute respiratory illness (ARI)and chronic obstructive pulmonary disease (COPD) are presented in Table 31. The odds ratios represent the risk o f illness for those who are exposed to indoor air pollution compared to the risk for those who are not exposed. The exact odds ratio depends on several factors such as concentration level o fpollution inthe indoor environment and the amount o ftime individuals are exposed to the pollution. A range o f ratios reflects the bounds o f uncertainty. The odds ratios have been applied to young children under the age o f five years (for ART) and adult females (for ARI and COPD) to estimate the increase in mortality and morbidity associated with indoor air pollution." It is these population groups who suffer the most from indoor air pollution. This i s because they spend much more o f their time at home, and/or more time cooking than older children and adult males. Table31: HealthRisksofIndoorAir Pollution OddsRatios(OR) "Low" "High" Acute Respiratory Illness (ARI) 1.9 2.7 Chronic obstructive pulmonary disease (COPD) 2.3 4.8 BaselineHealthData. 51. To estimate the health effects o f indoor air pollution from the odds ratios baseline data for ARI and COPD need to be established. These data are presented in Table 32, along with unit figures for disability adjusted life years (DALYs) lost to illness and mortality. Data on COPD mortality and especially morbidity incidence, according to international disease classifications, are not readily available for Pakistan. The national average two-week prevalence rate o f ART inchildren under 5 years as inMICS database (1996) i s used to estimate total annual cases o f ARI inchildren under 5. The average duration o f ARI is assumed to be about 7 days. This implies that the two-week prevalence captures halfo f the ARI prevalence inthe week prior to and the week after the two-week prevalence period. Although Desai et a1 (2004) present odds ratios for lung cancer, this effect o f pollution i s not estimated in tlus report. This is because the incidence o f lung cancer among rural women is generally very low. The number o f cases inruralPakistanassociated with indoor air pollution is therefore likely tobeminimal. 20 52. There is no information on ARI prevalence in adults. However, the information i s available for quarterly ARI-reported morbidity among people under and over 5 years of age.'* This database provides an indication o f the annual incidence o f ARI per child relative to annual incidence for the rest o f the population. An analysis o f the database suggests that ARI incidence in the population above 5 years o f age i s 0.36 o f the incidence in children under 5 years. It should be noted however that the database contains information on cases o f ARI treated at health facilities. In general, the percentage o f cases o f ARI that are treated at health facilities is higher among young children than older children and adults. The annual cases o f ARI per person among the population above 5 years o f age, presentedinTable 5.3, i s therefore estimated inthe range o f 0.36 to 0.42 [(1/(0.85))*0.36] o f the annual casesper child under-5. 53. ARI mortality in children under 5 years is presented in Table 5.3. The l o w bound o f ARI mortality o f 11 percent i s estimated from PDS 2003, the high bound o f 18 percent i s estimated from a combination o f GBD 2002 mortality tables for Searo D and Emro D regions o f WHO, reflecting uncertainty over all-cause and cause-specific child mortality statistics. 54. Table 32 also presents DALY per cases o f ARI and COPD, which are used to estimate the number o f D A L Y s lost because o f indoor air pollution. The disability weight for ARI morbidity i s the same for children and adults (i.e., 0.28), and the duration o f illness is assumed to be the same (i.e., 7 days). The DALYs per 100 thousand cases o f ARI i s however much higher for adults. This is because DALY calculations involve age weighting that attaches a low weight to young children, and a higher weight to adults, that corresponds to physical and mental development stages.13 For ARI child mortality the number o f D A L Y s lost i s 34. This reflects an annual discount rate o f 3 percent o f life years lost. Baseline Source: Urban I Rural Female COPD mortality rate (% o f total 3.1 % WHO (2002) and Shibuya et a1(2001) female deaths) Female COPD incidence rate (per 100,000) 63 ARI 2-week Prevalence inChildren under 5 24% 24% MICS 1996 years Estimated annual cases o f ARI per child 4.1 4.1 Estimated from MICS 1996 under 5 years Estimated annual cases o f ARI per adult I 1.5 I 1.75 I Estimated from a combination of female (> 30 years) MICS 1996 and pakistan.gov.pk ARI mortality in children under 5 years (% 11-18% PDS 2003, GBD 2002 (WHO) for o f child mortality) Searo Dand Emro D DALYs per 100 thousand cases o f ARI in 165 165 children under 5 DALYs per 100 thousand cases o f ARI in 700 700 Estimatedfrom WHO tables female adults (>30) DALYs per case o f ARI mortality in 34 34 children under 5 DALYs per case o f COPD morbidity in 2.25 2.25 adult females DALYsper case o fCOPDmortality inadult 6 6 females 12Informationabout Priority Diseases at pakistan.gov.pk. 13It shouldbe notedthat some researchers elect not to use age weighting, or reports DALYs with andwithout age weighting. 21 EstimatedHealthImpacts 55. Annual new cases o f ARI and COPD morbidity and mortality (Di) from fuel wood smoke was estimated from the following equation: where OB i s baseline casesDi illness or mortality, i(estimated from the baseline data inTable 5.2), and o f= PAR *DB (1) PAR i s givenby: PAR =PP*(OR-I)/(PP*(OR-I)+l) (2) where PP i s the percentage o fpopulation exposed to fuel wood smoke (32 percent o f the urban and 86 percent o frural population according to Pakistan Census 1998), and OR i s the odds ratios. The results are presented in Table 33. Estimated cases o f ARI child mortality and ART morbidity (children and female adults) from indoor air pollution represent about 38-53 percent o f total ARI in Pakistan. Similarly, the estimated cases o f COPD mortality and morbidity represent about 46-72 percent o f total estimated female COPD from all causes. EstimatedAnnual DALYs "Low" "High" % o fTotal DALYs Acute Respiratory Illness (AM): Children (under the age o f 5 years) - increased mortality 745,7 18 1,056,029 77% Children (under the age o f 5 years) - increased morbidity 48,690 68,95 1 5% Females (30 vears and older) -increasedmorbiditv 75.282 106.609 8% , d Chronicobstructivepulmonary disease (COPD): Adult females -increasedmortality 44,450 68,600 5% Adult females -increasedmorbiditv I 49.163 I 75.873 I 5% 56. Table 34 presents the estimated health impacts interms o f disability adjusted life years (DALYs). An estimated 963-1,376 thousand DALYs are lost each year due to indoor air pollution. About 77 percent i s from mortality, and about 30 percent from morbidity. 22 EstimatedCost ofHealthImpacts 57. Total annual cost o f indoor air pollution i s estimated at 55-70 billion Rs, with a mean estimate o f 62 billion (Table 35). The cost of mortality for adults i s based on the value of statistical life (VSL) and for children on HCA. The cost o f morbidity includes the cost o f illness (medical treatment, value o f lost time, etc). "Low" "Hi~h" Acute RespiratoryIllness (AH): Children (under the age o f 5 years) -increasedmortality 27.83 39.40 Children (under the age o f 5 years) -increasedmorbidity 4.26 6.03 Adult females -increasedmorbidity 2.04 2.89 Chronic obstructivepulmonary disease (COPD): Adult females -increasedmortality 25.84 25.84 Adult females -increasedmorbidity 0.12 0.18 TOTAL 60.08 74.34 58. Baseline data for the cost estimates o f morbidity in Table 5.5 are presented in Table 36. The percentage o f ARI cases inthe age group older than 5 years treated at medical facilities is estimated from percent o f treated cases among children (MICS 1995) and the ratio o f treated cases among children under 5 to treated cases among the population above 5 years o f age. There i s very little information about the frequency o f doctor visits, emergency visits and hospitalization for COPD patients in any country in the world. Schulman et a1 (2001) and Niedennan et a1 (1999) provide some information on this from the United States and Europe. Figures derived from these studies are applied to Pakistan in this report. Estimated lost work-days per year i s based on frequency o f estimated medical treatment plus an additional 7 days for each hospitalization and one extra day for each doctor and emergency visit. These days are added to reflect time needed for recovery from illness. To estimate the cost o f a new case of COPD, the medical cost and value o f time losses have been discounted over an assumed 20-year duration o f illness. An annual real increase o f 2 percent in medical cost and value o f time has been applied to reflect an average expected increase in annual labor productivity and real wages. The costs are discounted at 3 percent per year, a rate commonly appliedby WHO for health effects Table 36: BaselineDatafor CostEstimation Source: Percent o f ARI cases treated at medical facilities (children < 5 years) 53 MICS 1995 Cost o f medicines for treatment o f acute respiratory Per consultations with illness (population < 5 years) 50 pharmacies Percent o f ARI cases treated at medical facilities (females > 30 years) 49 Internationalexperience Cost o f medicines for treatment o f acute respiratory Per consultations with illness (females > 30 years) (Rs) 50 pharmacies Percent o f COPD patients beinghospitalizedper year 1.5 Assumption based on Percent o f COPD patients with an emergency Schulman et a1(2001) and doctorhospital outpatient visits per year Niedennan et a1(1999) 23 age number o f doctor visits per COPD patient per frequency o f doctor visits, amacies, me IAverage lengthofhospitalizationfor COPD (days) I 10 I Larsen(2004b) IV. NATURALRESOURCEDAMAGESINPAKISTAN 59. Natural resources damages are estimated for croplands, rangelands and forests. Cropland losses include losses from soil salinity due to improper irrigation practice and human-induced soil erosion. Inthe absence of data on the annual increase insalinity and eroded croplands andrangelands, the annual loss of agncultural production (crop and rangeland fodder) i s estimated based on accumulated degradation. This estimate may be more or less than the net present value (NPV) of annual production losses depending on the rate of annual increase in degradation. Annual deforestation data are available and annual losses resultingfrom deforestation are therefore estimated basedonNPV oflost forested area. V. COSTTO AGRICULTURE OF SOIL SALINITY 60. Soil salinity reduces the productivity of agncultural lands and, if salinity levels are highenough, can eliminate cultivation. From conventional welfare economics and assuming agricultural markets are competitive, the economic costs o f salinity are the losses in consumer surplus (consumer willingness to pay above market price) and producer surplus (profit) associated with this loss in productivity. Such losses could be direct, inthe sense of crops that cannot be plantedor, ifplanted, yield lower output than if planted in less saline soil. They could also be related to losses from crop substitution to more saline- tolerant but lessprofitable crops. Because of a lack o f data, these losses are approximated by the value of "lost" output relatedto the salinity, with some simpleadjustment for changes incropping patterns. Total irrigated land inPakistan is about 18.2 million hectares (Table 37). Nearly 25 percent of this land suffers from various levels o f ~alinity'~.Table 38 presents salinity levels o f irrigated lands in Palustan. About 1.4 million hectares of lands with salinity 15-20 dS/m are no longer cultivated. Table 39 presents soil salinity thresholds and yield effects of salinity from the internationalempirical 1iterat~e.l~ l4In addition, as much as 2.0 millionhectares are reported to have soil salinity exceeding 20 dS/m (Agricultural Statistics o fPakistan 2003-2004. Economic Wing, Ministryo fFood, Agriculture and Livestock, Government o f Pakistan, Islamabad). Eventhe most salt tolerant crops, such as cotton, wheat, andbarley, would have severe difficulties insuch saline conditions. We assume that these 2 millionha o f landwere never cultivated. l5The dS/m values are roundedto nearest integer and percent, andrelatively conservative values have beenused. 24 Table 37: Irrigated land in Pakistan (2002-2003) I I Million hectares 1 Punjab 13.94 Sindh N.W.F.P. Balochistan Total Source: Agricultural Statistics ofPakistan. (2003-2004). Table 38: Salinity Levels of Irrigated Lands Salinity Level h g a t e d dS/m (millionha) % Source: Agricultural Statistics of Pakistan2003-2004. Table 39: Crop Salinity Tolerance and Yield Effects Salinity Yield decline Threshold per 1dS/m (dS/m) over threshold Pulses 1.5 15% Sugar cane 1.7 7% I Fodder I 2 I 7% I Vegetables 2 10% Maize 2 12% I Rice I 3 I 12% I Soybean 5 20% Wheat 6 5% Barlev 8 5yo Source: Salinity threshold and Kotuby-Amacher, J. et a1 (1997), and Cullu M.A. (2002). 61. There are no comprehensive data available on cropping patterns in relation to specific levels o f soil salinity in Palustan. To estimate the cost o f salinity, it i s therefore necessary to make a simplifying assumption that more salt sensitive crops are cultivated on the lands with lower salinity. Optimal adaptation, if salinity was the only soil characteristic affecting crop choices, would imply that the salt sensitive crops are cultivated on the land that has salinity lower than 4 dS/m while crops on more saline land are mainly wheat, barley and cotton. 25 62. We consider two scenarios o f cropping patterns on saline land. The first scenario assumes that only cotton i s cultivated on the most saline lands (15-20 dS/m), and that more cotton than wheat is cultivated on land with salinity 8-15 dS/m. The second scenario assumes that only wheat i s cultivated on the most saline lands, and that more wheat than cotton i s cultivated on land with salinity 8-15 dS/m. These cropping patterns are in practice unlikely. The two scenarios therefore represent an upper and lower bound o f the cost of salinity because o f the different market value and salinity tolerance o f cotton and wheat. Table 40: AssumedCroppingPatternson Irrigated Lands Scenario (1) Scenario (2) Level Minimal Pulses, sugar cane, vegetables, Pulses, sugar cane, vegetables, li-s salinity maize, fodder, rice, and soybean; maize, fodder, rice, and soybean; plus plus Wheat (6.3 millha), cotton (0.6 Wheat (4.7 mill ha) and cotton (2.2 mill.ha) mill ha) & Slight Wheat (0.4 millha), cotton (0.4 mill Wheat (0.4 mill ha), cotton (0.4 salinity ha) mill ha) Wheat (0.4 millha), cotton (1.O mill Wheat (1.O mill ha), cotton (0.4 ha), barley (0.06 mill ha) mill ha), barley (0.06 millha) Cotton (1millha) Wheat (1mill ha) 63. The assumed cropping patterns on saline land correspond to a cropping intensity o f 1.4 on the land with minimal salinity, 1.3 on the land with slight salinity, 1.2 on land with moderate salinity, and 1.O on land with severe salinity. These cropping intensities are estimated based on provincial cropping intensities reported by the provincial agriculture departments in Palustan and on distribution o f saline land. 64. To estimate crop losses from salinity it i s necessary to estimate crop yields that would prevail in the absence o f salinity. The following equations are solved for cotton, wheat andbarley for this purpose: where Xis observed average yield; Xi is yield on land with salinity="i"; a, b and c are share o f land with salinity level="i"; and di is yield reduction on land with salinity level="?'. Observed average yields, and estimated yields inthe absence o f salinity and on severely saline land are presented inTable 41. Observed average yield Estimatedyield in Estimatedyield on on irrigatedland absence o f salinity severely saline land (toisha) (tonsiha) (tinsiha) * Seed Cotton 1.8 1.9-2.3 1.2 Wheat 2.5 2.6-2.9 1.2 Barley 1.o 1.2 NA 26 65. Tables 42-43 present the estimated annual cost to agnculture of soil salinity (scenario 1, 2). World prices were applied for wheat and barley, and producer prices in 2005 for seed cotton, i.e., 9400 Rs, 5300 Rs and 22500 Rs, respectively. The total annual cost of crop loss from salinity is estimated at 15-55 billion Rs, not including lost opportunities from cropping on the 1.4 million hectares o f land with salinity that has reached 15-20 dS/m (unproductive land). The cost o f salinity on this land i s estimated at net farm income on land with minimal levels of salinity, i.e., 10-18 thousand Rs.per hectare (Dost 2002). This bringsthe total estimated cost of salinity to 30-80 billion Rs., with a mean cost of 55 billion Rs, or 0.9 percent o f GDP in2004. Unproductiveland 15-20 14 20 25 0.32% Total Loss 33 57 80 0.92% Wheat 4-8 0 0.5 1 0.01% Wheat 8-15 3 7.5 12 0.12% Wheat 15-20 12 15.7 19 0.26% Barley 8-15 0 0.1 0 0.00% Cotton 8-15 0 2.9 6 0.05% Total Loss (includingunproductive land) 29 47 I 64 I 0.76% VI. COST OFAGRICULTURAL SOIL EROSION 66. Land degradation caused by wind and water erosion increased by almost 3.5 million hectares from 1993 to 2003 (Ahmed and Rashid 2003; Brandon 1995) and comprised about 18 million hectares in total in2003. Figure A1 presents the increase in eroded lands by province. The provinces most affected by soil erosionduringthis period are Sindh(about 1.5 million hectare increaseineroded land of which an estimated 360 thousand hectares i s an increase in eroded crop lands) and Balochistan (about 2 million hectares increase in eroded land of which an estimated 500 thousand hectares i s an increase in eroded crop lands). 27 Punjab Sindh NWFP& FATA Baloohlrtan 1 I1993 El2003 Source: Ahmed and Rashid (2003); Brandon (1995). 67. We applied the same assumptions as in Brandon (1995) about cropping patterns and yield loss due to soil erosion for five major crops. The analysis i s based on three categories o f soil erosion (light, moderate and severe) disaggregated by province. Each type o f degraded land was allocated in Brandon (1995) across the cropping pattern for each province, and a corresponding yield reduction factor was applied. Table 44 presents yield reduction factors for different crops (Brandon, 1995). Crop Level o f erosion Light Moderate Severe Paddy 2% 50% 60% Wheat 2% 5% 10% Maize 2% 5% 10% Cotton 2% 5yo 10% Sugarcane 2% 5% 10% 68. After the eroded cropping area inSindh and Balochistanwas adjusted assuming the increase o f eroded croplands presented above, the eroded area under five major crops was estimated, utilizing initial estimates from Brandon (1995). Table 45: The Eroded Area under Five Major Crops in 2003 SLIGHT DEGRADATION Paddy Wheat Maize Cotton Sugarcane Total 000 ha NWFP 1,927 29,461 16,375 12 3,099 51 Punjab 177,994 837,638 47,117 355,3 18 78,106 1,496 Sindh 84,071 157,592 1,727 56,742 35,409 336 Balochistan 1,999 6,390 70 4 12 8 0 Sum000 ha 266 1,031 65 412 117 1,891 28 Paddy Wheat Maize Cotton Sugarcane Total 000 ha NWFP 20,187 308,690 171,574 130 32,474 533 Punjab 88,135 414,762 23,33 1 175,938 38,675 741 Sindh 33,001 61,861 678 22,272 13,901 132 Balochistan 48,898 156,204 1,700 82 241 207 Sum000ha 672 2,873 370 780 261 4,957 69. Using data from 2003 on eroded crop land, and applying world prices for wheat, maize, paddy and cotton lint and producer prices for sugar cane, we estimate economic losses from soil erosion in Pakistan at around 15 billion Rsper year, or 0.25 percent of GDP Paddv 220 I Wheat I 160 I Maize 100 Cotton lint 1100 Sugarcane 22 Source: http:llwww.favri.iastate.eduloutlook2005ltablesl6 ComPrices.xls; http:l/www.orvza.codvriceslasia,shtml; ht~://64.233.179.104/search?~=cache:3rParkkb7-EJ:www.~aktribune.comlnewslindex.php%3Fid%3D 127995+ price+sunarcane+Pakistan&hl=en&al=us&ct=clnk&cd=3 29 Table 47: Annual Erosion-RelatedLosses by Crop and Provincein Pakistan(tons) NWFP Punjub Sindh Balochistan Palustan Paddy 66,054 151,917 88,341 476,285 782,596 Wheat 52,666 158,077 25,680 74,521 310,944 Maize 34,326 5,547 64 385 40,322 Cotton 5 19.588 1.473 8 21.074 Sugarcane 197,013 280,474 144,024 1,590 623,101 Palustan 350,064 615,603 259,582 552,788 1,778,038 Table 48: Annual Erosion-RelatedLosses by Province Billion Rs Percent GDP NWFP 1.8 0.03% Punjab 5.1 0.08% Sindh 1.7 0.03% Balochistan 6.9 0.11% VII. COST OFRANGELANDDEGRADATION 70. The National Forest and Rangeland Resource Assessment Study (NFRRAS) (2004) presents estimates of rangelands inPakistan over a period of ten years, reporting that rangeland area declined from 28.5 million hectares in 1992 to about 23.5 million hectares in 2001. The net decrease i s 5 million hectares, which i s a rate of 2 percent per year. Figure A2 presents rangeland area in Palustan in 1992- 2001. FigureA2: RangelandArea in 1992-2001 25 20 !!-615 g B Es2001 10 5 0 NWFP Punjab Slndh Baloohlrtan N.AS. AJK Total Source: NFRRAS, 2004. 71. The most substantial reduction occurred in 1992-1997. Figure A3 presents annual reduction of rangelands as percent o f total province land area. The most substantial relative reduction o f rangelands occurred in Northern Areas @.As) and Balochistan in 1992-1997. In absolute terms rangelands in Balochistan declined the most. The province was losing annually about 527,000 hectares of grazing lands. 30 Intotal about 3.6 millionhectares were lost inBalochistanin 1992-2001as result ofdesertification and denudation of vegetation from drought and continuous overgrazing. FigureA3: Percentof RangelandChangeinDifferentProvinces 12.0Y. 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% NWFP Punjab Slndh Baloohlstm N.M. -2.0% ource; NFRRAS, 2004 Figure A4: AccumulatedRangelandChangein 1992-2001 I I I I Source: NFRRAS, 2004 72. In1995 FA0presented estimates oflandusebasedon satellite imagery. Bothdegraded andnon- degraded rangelands were estimated as well as alpine rangelands, which are mostly non-degraded (Table 49). Nearly 25 millionhectares were classified as degraded, or 85 percent of total rangelands. 31 Table 49: Degraded and Non-degraded Rangelands in 1995 I IDegraded, 000' ha I Non-degraded, 000' ha IAlpine, 000' ha ~ ~~~ , AzadKashmir 731 79 Balochistan 11,674 892 Northern Areas 896 705 NWFP 4,106 519 269 Puniab 4,466 1.293 Sinih 2,809 68 Total Area 24,682 2,772 1,053 73. Two methods are usedto estimate the cost of rangeland degradation and losses. The first method i s an estimate of losses o f rangeland fodder yield valued at the price o f fodder. The second method i s an estimate of foregone livestock income from loss of fodder basedon livestock feed requirement. 74. Very limiteddata are available on rangeland fodder yields. Mackintosh (1993) reports an average yield of nearly 0.4 tons o f dry matter (DM) per hectare in 1974. According to interviews with rangeland experts in Palustan in the process of preparing this report, average yield i s now estimated at 0.2 tons of DMperhectareon degradedrangelands. This impliesthat the yield decline from cumulativedegradation i s at least 0.2 tons per hectare. Based on a rangeland area of 23.5 million hectares in 2001, and assuming that the area of non-degraded rangeland was constant from 1995 to 2001, degraded rangeland area in 2001 was nearly 20 million hectares. Applying the yield decline to this area suggests a total loss o f 4 million tons of DMper year from cumulative degradation. At a fodder price of 1000-1500 Rs per ton of DM(Dost 2000), thislossrepresents acost of2.4-3.6 billionRsperyear basedona sustainablerangeland fodder utilization rate of 60 percent, as applied in Brandon (1995). In addition, the loss of 5 million hectares of rangeland from 1992 to 2001 suggests a loss of nearly 2 million tons of DM per year, assuming the original yield on this rangeland was 0.4 tons per hectare. At current fodder prices, and a sustainable utilization rate of 60 percent, this represents a cost o f 1.2-1.8 billion Rs per year. The total cost of rangeland degradation over the last 30 years and rangeland losses over the last decade i s therefore estimated at 3.6-5.4 billion Rsper year. 75. Data neededto estimate foregone livestock income (i.e., the secondmethod o f estimating the cost of rangeland degradation and losses) are presented in Table 50. Total feed requirement in DM for the animal stock in Pakistan i s estimated at 145 million tons per year. Inthe absence o f the rangeland yield decline of 4 million tons of DM per year, of which 60 percent could be sustainably utilized, rangelands could have supported an additional 1.7 percent of current animal stock. Total household net income from livestock i s around 150 billion Rs per year (PIES 2001-02). The loss in rangeland yield i s therefore equivalent to 2.5 billion Rs in foregone livestock income. Also, the loss of 5 million hectares of rangeland from 1992-2001, with a fodder loss of 2 million tons of DM per year, o f which 60 percent could be utilized, could have supported an additional 0.85 percent of animal stock. This i s equivalent to 1.3 billion Rs in foregone livestock income per year. Total cost of rangeland degradation and losses i s therefore estimated at 3.8 billion Rsper year. 32 Feedrequirement, Weight, kg TDMtlyear Total animal, mill 2000 Total TDM, millt Buffalo 550 4 18.64 75 Cattle 450 3.3 17.30 57 Sheen 75 0.5 13.13 7 Goat 20 0.1 34.76 5 Total 83.84 145 Billion Rs Percent o f GDP Market value o f fodder losses 3.6-5.4 0.06-0.09% Foregone livestock income from fodder losses 3.8 0.06% Mean cost 4.2 0.07% VIII. DEFORESTATION 77. The cost o f forest degradation i s the aggregate social loss associated with degraded or deforested lands. These costs include, in theory, a wide range o f local, regional, national, and even global costs. Examples include timber, fuel wood and non-timberproduct losses (see below), recreation and tourism losses, indirect use losses (such as those associated with damages to ecosystem services, such as water supply and carbon sequestration), and non-use value loss associated with loss o f forests. This section examines each o f these categories as data permits, but refrains from presenting the carbon sequestration values. 78. The cost o f deforestation i s very difficult to estimate. Deforestationmay contribute to increased frequency and severity o f flooding and landslides, and is likely contributing to agricultural land erosion problems. It i s also associated with impacts on water resources quality. However, it is practically very difficult to identify and isolate these costs o f deforestation at the national level, and they are not included in the estimated cost in this section. Further, studies in Pakistan are insufficient to estimate the full economic value o f the country's forests, and thus the cost o f deforestation. Estimates of forest values for the forests in other countries are therefore applied in this report, using a range o f values to reflect the uncertainties o f applying these values to Pakistan. 79. Because o f the high degree o f uncertainty about these costs, we present average as well as high and low estimates, drawing on background studies by Pearce et a1 (1999) and Lampietti and Dixon (1994). High-end estimates are based on the assumption that it i s possible to internalize all forest benefits based on a forest "inventory" by the local community in the short term, which is obviously an overestimation. Low-end estimates are based on the possibility that almost no forest benefits can be internalized because o f an absence o f market infrastructure, roads, favorable public policy and very high discount rates. 33 80. Social forest values should be considered from a long-term perspective. Therefore, financial flows from concessions and profit o f predatory logging are not estimated. Only flows out o f sustainable forest management are analyzed. Although it i s generally accepted that in the short-term profits from predatory logging are higher than from sustainable forest management, in the long term with a real discount rate less than 20 percent, sustainable management theoretically has a higher net present value (Pearce et al, 1999). Pearce et a1 (1999) reviews an abundant literature that debunks common perceptions about higher profitability o fpredatory logging. 81. We start by considering diverse estimates o f forested and deforested lands inPakistan. Forests in Pakistan in 1990 were estimated to cover about 3 percent of total land area (FAO, 2005). Today, forest cover i s less than 2 percent (FAO, 2005). Even with irrigatedplantations and other wooded areas, forest land was estimated to be no more than 4.3 percent o f the Palustan territory. However, there i s a great controversy about this estimate in Palustan. Government sources suggest that forested areas are about 5 percent o f Pakistan territory and growing (Table 52). Table 52: ForestArea inPakistanin1990-2005 I Year 1990-1991 IIForestarea (millionha) YOincreaseor decrease 3.46 1991-1992 I 3.47 0.3% I 1992-1993 I 3.48 I 0.3% I 1993-1994 3.45 -0.9% 1994-1995 3:6 4.3% I 1995-1996 I 3.61 I 0.3% I 1996-1997 3.58 -0.8% 1997-1998 3.6 0.6% 1998-1999 3.6 0.0% 1999-2000 3.78 5.O% 2000-2001 3.77 -0.3% 2001-2002 3.81 1.1% 2002-2003 4.04 6.0% 2003-2004 4.04 I 0.0% Average annual change 1.2% 82. The National Forest Institute o f Palustan facilitated the FA0 effort on the Global Forest Resource Assessment for Pakistan in 2005. Inour judgment, the FA0 effort i s the more reliable. The figures A5 and A6 present details o f these estimates, covering forest land area composition and the relative share of natural forest inthe total forest area in 1990-2005,as presented inFAO, 2005. 34 Figure AS: Forest Land Area Composition 2000 i 1 1500 IForests(exsl. plantations) hB lrrioated plantations (Forests) 0OtherWooded Land 1000 500 0 1990 2000 2005 ource: FAO, 2005 FigureA6: Forest Share inthe Total Forest Area ! mother wooded Land Ff)Irrigatedplantations(Forests) MForosta (exsl. plantations) 1990 2000 2005 Source: FAO, 2005 83. Figure A5 demonstrates that the forest (FA0 definition, NFRAS, 2004) is declining inPalustan. The estimated deforestation rate over the 1990-2005 period is 2.1 percent or 47 thousand hectares annually. Forest types included in the FA0 definition of forests are coniferous forest, riverain and mangrove forest. The most valuable coniferous forest i s declining at the rate 40,000 hectares annually. Northern Areas and NWFT have the highest annual rates o f deforestation (about 34,000 hectares in NorthernAreas and 8000 hectares inNWFT16). l6 Coniferous forests area increased for about 2,000 hectares annually inAJK 35 Figure A7: Composition of Coniferous forest by Province I ........................... ................ ,800 1 1600 1400 1200 1000 0Baloshl.tan OSlndh 800 600 400 200 I 0 1992 1997 2001 I Source: NFRAS, 2004 84. Riverain and mangrove forests are also decreasing in area at the rate o f 2,300 and 4,900 hectares annually, respectively (Figure A8). This i s an alarming rate given the quite highecological value o f these types o f forest (NFRAS, 2004). Figure AS: Riverain and mangrove forest area 1992 1997 2001 Source: NFRAS, 2004 85. InFA0 2005 the share ofproductive forest was estimated as 32 percent in2005. We apply this rate to the total area o f deforestation o f coniferous forest (40,000 hectares) and get about 13 thousand hectares. Brandon (1995) reported an annual average sustainable yield o f 1-2 cubic meters o f timber per hectare o f productive forest. This i s quite a reasonable estimate given that the commercial growing stock inconiferous forests is estimatedinNFRAS (2004) at about40 cubic metersoftimber perhectare. 86. The annual timber loss from deforestation i s estimated at about 114 million Rs by applying a net stumpage value o f 100 USDper cubic meter of sustainable timber harvest (Brandon, 1995) on one hectare o f productive forest (upper bound estimate)." Ifwe apply the domestic timber price in2005 (2925 Rs per l7Ths assumes an average sustainable yield of 1.5 cubic meters o ftimber per hectare ofproductive forest. 36 cubic meter from FAO, 2005) and assume 50 percent production cost, then the corresponding timber value loss i s 28 million Rs. (lower bound estimate). 87. Following RWEDPA (1997) estimates that one hectare o f coniferous forests in Palustan supplies 1.1-1.25 ton o f fuel wood (1.17 on average), it i s possible to estimate fuel wood losses from deforestation. We apply 50 percent o f the current price o f fuel wood to account for fuel wood production cost.18 The estimated value o f one hectare o f forest related to fuel wood production i s then about 1025 Rs. The annual cost o f fuel wood losses from 40 thousand hectares o f coniferous forest losses per year i s then about 41 million Rs. 88. FAO, 2004 estimates non-timber values for the Pakistan forest. Excluding fodder value that we already accounted for in the rangeland degradation section, non-timber value per hectare o f forest i s estimated at 9 USD. This value is close to estimates inLampietti and Dixon (1994) for non-timber values inCentral and SouthAmerica equal to US $9-10 per hectare. We apply the value of9 USDper hectare of deforested land (including riverain and mangrove forest losses). The annual cost o f non-timber losses i s then 25 million Rs. 89. Another direct use value i s ecotourism. Pearce et a1 (1999) estimates these values in the range o f U S $5-10 per hectare o f forest and stresses their local-specific character. We use 10 USDper hectare as a lower estimate, implying a total cost o f 28 million Rs. from annual loss o f recreational forest value. In Khan 2004 recreational value o f a Natural Park near Islamabad was estimated at about 25 U S D per hectare. Usingthis as an upper estimate gives 70 million Rs. inannual loss o frecreational forest value. 90. Indirect use values o f forest include watershed protection, nutritional and erosiodflood prevention, and waterhtrient recycling. Although there is n o definite agreement in the literature about the magnitude o f this forest value, Pearce et a1 (1999) presents a higher end estimate o f U S D 30 per hectare o f forest generalized from the literature review. Applyingthis value to the annual forest losses in Pakistan gives an annual cost o f 84 millionRs. 91. Pierce et a1 also give a wide range for the option value o f forest (bioprospecting, i.e. prospects o f new drugs to be developed in the future usingrich forest biodiversity) in the range o f USD 0.01-21 per hectare. Applying the highest value inthis range gives a high-end cost o f deforestation o f 58 million Rs per year. We assign zero as the low-end estimate. Existence value o f forest associated with forest preservation i s estimated in Pearce et a1 (1999) at U S D 13-27 per hectare, derived from his literature review. This implies an annual deforestation cost o f 36-75 millionRs. 92. As mentionedbefore, this section refrains from including lost carbon storage values o f forest as a cost o f deforestation due to the uncertain magnitude o f the carbon price at this point in time. Carbon markets are only just emerging and currently deforestation reduction i s not eligible for any compensation. However, the situation could change inthe near future. Then forest values should be updated usingcarbon market prices and the eligible share o fthe forest that can be counted for carbon sequestration. 93. The estimated costs o f deforestation inPakistan are summarized in Table 53. NPV i s the present value o f the stream o f costs from one year o f deforestation. The direct use values, reflecting local private forest losses, include the losses from sustainable logging, non-timber products and tourism and recreation. "Low" and "high" non-use values differ by a factor o f three. This reflects an essential disadvantage o f non-use value estimation. The studies use contingent valuation approaches that are based on the solicitation o f the values from respondents in surveys. Different survey questionnaires will result in a different non-use value, and will often vary substantially across countries and for different locations 1870 Rs.per40 kg of helwood (Pakistan Statistical yearbook, 2004). 37 within a country. The non-use forest values are therefore not included in the estimate of the cost o f deforestation for Pakistan inthis report. Annual Cost Forest service L o w High NPV* estimate Mean estimate (mean) Direct use values 122 186 250 1860 Sustainable timber production 28 71 114 710 Fuelwoodproduction 41 41 41 410 Non timber Droducts 25 25 25 250 Tourism and recreation 28 49 70 490 Indirect use values 84 84 84 840 Non-use values I 36 I 84.5 I 133 I 845 I Option value (bioprospecting) 0 29 58 290 Existence value 36 55.5 75 555 Direct PlusIndirect 206 270 334 2700 Total value 242 354.5 467 3545 38 References Abbey, D.et a1(1995). Long-TermAmbient ConcentrationsofParticulatesandOxidants andDevelopment of Chronic Diseaseina Cohort o fNonsmolung CaliforniaResidents. InhalationToxicology, Vol7: 19- 34. Agha F., SadaruddinA,, KhatoonN.(2005). Effect of Environmental LeadPollutiononBloodLeadLevels inTraffic PoliceConstablesinIslamabad,Pakistan.JournalofPalustanMedicalassociation.Vol. 55 (10): 410-13. Agricultural Statistics ofPakistan. Economic Wing, MinistryofFood, Agriculture and Livestock, (2003- 2004). Govemment ofPakistan, Islamabad AhmedandRashd(2003) Fertilizers andtheir UseinPakistan;NFDC/P&DD/GoP; Brandon C. (1995). Valuing Environmental CostsinPalustan:The economy-wide impactofEnvironmental degradation Cohen A. et a1(2004). UrbanAir Pollution. In"Comparative quantification ofhealthrisks" ed. by Ezzati et al. WHO. Geneva Cropper, M.andOates, W. (1992). Environmental Economics: A Survey. Journal ofEconomic Literature. Vol. XXX, pp.675-740. CulluM.A.(2002). Estimation o fthe Effect of Soil Salinity on Crop Yield UsingRemote Sensingand GeographicInformation System, TurkJ Agric For 27 (2003) 23-28, http://journals.tubitak.gov.tr/agriculture/issues/tar-03-27-l/tar-27- 1-4-0209-13.pdf Desai, M.,Mehta, S., Smith, K.(2004). Indoor smoke from solidfuels. Assessing the environmental burden of diseaseat national and local levels. Environmental BurdenofDisease Series, No. 4. WHO. Dockery D.W., PopeC.A. III, X, et a1(1993). An associationbetween air pollution andmortality insix Xu U S cities. New EnglandJournal ofMedicine, 329: 1753-1759. DostM.(2000) Fodder Production for Pen-urbanDairies inPakistan. http:/lwww.fao.org/ag/agp/agpc/doc/pasture/dost/fodderdost.h~ DostM.(2002). Country PastureEorageResourceProfiles. Palustan http://www.fa0.org/WAICENTEAOINFO/AGRICULT/AGP/AGPC/d~/Counpro~akista~a~stan. h tm Esrey, Potash, Roberts, and Shiff(1991). Effects of improvedwater supply and sanitation on ascariasis, diarrhoea, dracunculiasis, hookworminfection, schistosomiasis, andtrachoma. Bulletinofthe World HealthOrganization, 69(5):609-621. Esrey, S.A., andHabicht, J-P. (1988). Maternal literacy modifiesthe effect oftoilets andpipedwater on infant survival inMalaysia. American JournalofEpidemiology, 127(5): 1079-1087. FA0(1998). FA0IrrigationandDrainagePaper56. PreparedbyAllen, R.et al. FA0(2005) GlobalForestAssessment Report.Palustan.Countryreport. FewtrellF., Priiss A, KaufhannR.(2003). Guide for assessmentof EBDat national and local level: Lead. WHO, Geneva. Fewtrell, L.andJ. Colford Jr. (2004). Water, Sanitation andHygiene: Interventions andDiarroea-A systematicreviewandmeta-analysis. HNP Discussion Paper. World Bank. Gratten, Zeng, Shannon, Roberts (2002). k c e i s more sensitiveto salinity thanpreviously thought. CaliforniaAgnculture, Vol. 56, Number 56. November/December. 39 Hozhabri, S., White F.,RahbarMH, Agboatwalla M., Luby S. (2004). ElevatedBloodLeadLevelsAmong ChildrenLivingInA FishingCommunity, Karachi, Palustan.Archives of Environmental Health, JanUary KhanH.(2004) Demandfor Eco-Tourism. Estimating RecreationalBenefits from the MargallaHills NationalPark inNorthernPakistan. SANDEE.Worlung Paper# 5-04. KhwajaMh.(2003). EnvironmentalHealth: LeadExposureandIts Impact on Children. SDPIResearchand News.Bulletin.Vo12 (2): 1-3,7. Kotuby-Amacher, J. et a1(1997). Salinity andPlant Tolerance. UtahSateUniversity. UnitedStates. Lampietti, J., DixonJ. (1994). To See the Forestfor the Trees: A Guideto Non-Timber ForestBenefits. WorldBank. Washington DC. Larsen, B.(2003). HygieneandHealthinDevelopingCountries: DefiningPriorities through Cost-Benefit Assessments. InternationalJournal ofEnvironmental HealthResearch. Vol 13: S37-S46. Larsen, B.(2004a). Cost ofEnvironmental Degradation: A Socio-Economic andEnvironmental Health Assessment inDamietta, Egypt. Preparedfor SEAMII, Egypt.Cairo, Larsen, B.(2004b) Columbia. Cost OfEnvironmental Damage: A Socio-Economic and Environmental HealthRiskAssessment. Final Report. Preparedfor the Ministryof Environment, HousingandLand Development ofRepublic of Colombia. LouisianaDepartment ofPublic Health. (2004). TyphoidFever. www.oph.dhh.state.la.us/infectiousdisease/docs/disease~updates~O4~~hoi~an~l.pdf Luby S., Agboatwalla M.,RazzA., SobelJ. (2001) A Low-CostIntervention for CleanerDridungWater in Karachi, Pakistan.PakistanInternationalJournal ofInfectious Diseases. 5(3): 144-150. MaQ.(1999) Asia-Pacific Forestry SectorOutlook Study: Volume I- Socio-Economic, Resources andNon- Wood Products .Working PaperNo: APFSOS/WP/43 .ForestryPolicy andPlanningDivision, Rome. Regonal Office for Asia andthe Pacific, Bangkok http://www.fa0.org/documents/show~cdr.asp?url~file=/docrep/x26 13eIx2613e00.HTM Mackintosh J.B. (1993). Shhep Production inPakistan.PakistanResearchCouncil, Islamabad Mirza S., Thomson E., Akhbar G,Sattar Alvi A.,and Rafique S. Balochstan. Searchingfor a strategy. http://www.icarda.org/Publications/Caravan/Caravan4/Car411.Html MohammadNoor. (1989). RangelandManagementinPalastan.Publishedby the InternationalCentre for IntegratedMountainDevelopment (ICIMOD) Kathmandu, Nepal. Mrozek, J. andTaylor, L.(2002). What Determinesthe Value ofLife? A MetaAnalysis. Journal ofPolicy Analysis andManagement.Vol21 (2): 253-270. NationalCommission on Agriculture. (1988). Report ofthe NationalCommissionon Agriculture, Ministryof FoodandAgnculture, Government ofPakistan, Islamabad NationalForestandRangelandResourceAssessment Study (NFRR4S) (2004) PakistanForestInstitute. Peshawar. Niedennan, M.et al. (1999). Treatment Cost ofAcute Exacerbationsof Chronic Bronchitis. ClinicalTherapy, 21(3): 576-91. Ostro, B.(1994). Estimating the Health Effects ofAir Pollution: A MethodwithanApplication to Jakarta. Policy ResearchWorking Paper, World Bank. Othman, J.(2002) HouseholdPreferencesfor SolidWaste ManagementinMalaysia, Economy and Environment Program for Southeast Asia(EEPSEA) ResearchReportNo. 2002-RR8, International Development ResearchCentre. 40 PakistanEconomic Survey (2004-2005), Economic Advisors Wing, Ministryo fFinanace, Government of Palustan PakistanStatisticalYearbook 2004. Paul R., White F.,Luby S. (2003). Trends inLeadContent ofPetrol inPahstan. Bulletin O fWorld Health Organization. 81 (6): 468. Pearce, D.,PutzF., Vanclay J. (1999). A SustainableForestFuture? CSERGE Working Paper GEC99-15. London. PIES (2001-02). PahstanIntegratedEconomic Survey 2001-02. PopeCA m,Burnett RT,ThunMJ, et a1(2002). Lungcancer, Cardiopulmonary mortality, and Long-term exposure to Fineparticulate air pollution. Journal of the American Medical Association, 287: 1132- 1141. Pope CA III,ThunMJ, NambudiriMM,et a1(1995). Particulateair pollutionas apredictor ofmortality ina prospective study ofU S adults. AmericanJournal ofRespiratory and Critical CareMedicine, 151: 669- 674. Pruss-Ustin ,A., FewtrellF., LandriganP. (2004). LeadExposure. In"Comparative quantification ofhealth risks" ed. by Ezzati et al. WHO. Geneva RahmanA., Maqbool E., Zuberi H.(2002). LeadAssociatedDeficitsinStature, MentalAbility andBehavior inChldreninKarachi.Annals ofTropicalPediatrics,22: 301-311. Rosemann, N.(2003) DrinlungWater Crisis inPakistanandthe Issue ofBottledWater: The case ofNestle's 'Pure Life". DrinkingWater Crisis inPakistan andthe Issue ofBottledWater: The case ofNestle's 'Pure Life", http://www.swisscoalition.ch/english/files/l-WrNn.pdf Rutstein, S.O. (2000). FactorsAssociated withTrendsinInfant and ChddMortality inDevelopingCountries duringthe 1990s. Bulletinofthe WorldHealthOrganization. Vol. 78(10): 1256-1270. SalkeverDS.(1995). Updatedestimatesof earningsbenefits from reducedexposureof chldrento environmental lead. Environmental Research. 70: 1-6 Schulman, RoncaandBucuvalas, Inc. (2001). ConfrontingCOPD inNorthAmerica andEurope: A Survey ofPatientsandDoctors inEightCountries. SchwartzJ. (1994). Societalbenefits o freducinglead exposure.Environmental Research66:105-12 Shi,A.(1999). The Impact ofAccessto UrbanPotableWater andSewerageConnection onChildMortality: City Level Evidence, 1993. Economists' Forum, Vol. 1,World Bank 1999. Shibuya, K.,Mathers, C., and Lopez, A. (2001). Chronic Obstructive Pulmonary Disease(COPD): ConsistentEstimatesofIncidence, Prevalence, andMortalitybyWHO Regon. Global Programme on Evidence for HealthPolicy. World HealthOrganization. November 2001. SinhaA. et al. (1999). TyphoidFever inChildrenAged LessthenFiveYears. Lancet. 354:734-37. Strukova, E.(2004). Opportunity Cost O fDeforestationInThe BrazilianAmazon: Aggregated Estimate Per Ton OfAvoided Carbon Emission. Working Paper.Washington DC SUPARCO(2004). Materials on Ambient Air Quality inMajor Cities of Palustan. TERI(2001). ReviewofPastandOn-Going Work onUrbanAr Quality inIndia. NewDelhi. Tata Energy ResearchInstitute. TERIProject ReportN2001 EE41. Viscusi, W.K. andAldi, J.E. (2002). The Value o fa Statistical Life: A criticalreviewofmarket estimates throughout the world. DiscussionPaper No. 392. HarvardLaw School. Cambridge, MA. United States. White F., et a1(2001). ElevatedBloodLeadLevelsinKarachi Children. BulletinWorldhealth Organization, 79(2): 173. 41 WHO (2001). GlobalBurdenofDisease 2001. The WorldHealthOrganization. WHO (2002a). GlobalBurdenofnsease 2002. The WorldHealthOrganization. WHO (2002b). The World HealthReport2002. The WorldHealthOrganization. WorldBank(2002a). EgyptCostAssessmentofEnvironmentalDegradation. ReportNo. 25175-EGT. .WashngtonDC. WorldBank(2004b). WorldDevelopmentIndicators2004. WashngtonDC. PDS2003 PMS2001/2002 Pakistan2004 StatisticalYearbook MICS. (1995) UNICEF.http://www.chldinfo,org/eddb/Diarrhoea/dtabase.htm PakistanCensus 1998 42