Environments and urban mortality—An econometric approach

ENVIRONMENTS A N D URBAN MORTALITY--AN ECONOMETRIC APPROACH RAJINDAR K. KOSHAL • MANJULIKA KOSHAL

Ohio University,Athens, Ohio, USA

ABSTRACT

This paper establishes a quantitative relationship between total mortality rates from all diseases and the level of air pollution and other socio-economic variables, using a macro-model. With the help of regression analysis it is observed that a l0 % increase in the air pollution would imply an increase in the mortality rate by 1.2-1.9 %. ~In overall reduction of about 50 % in the air pollution would imply a social savings of the order of $3400-5400 million per year in terms of all diseases. This analysis suggests that the value of the benefits of air pollution abatement is far greater than its cost.

INTRODUCTION

pollution may be said to be the emission of harmful or obnoxious contaminants into the atmosphere. Most of these are man-made contaminants which consist of particulate matter, gases and vapours. Particulate matter includes metallic oxides, sulphur-dioxides, siliceous material and other dusts, fumes, mists and fogs. Smoke from the incomplete combustion of burning coal carries with it ash particles, carbon and tar. Aerosols--fine, solid or liquefied particles suspended in air or gas for an extended period of time, such as smoke, fog or mist--reduce visibility and block sunlight (Gordon, 1970; Council on Environmental Quality, 1970). Air pollution also damages some crops, grains, citrus groves and fruits, deteriorates the pigment in paint, causing it to discolour and peel, soils textiles, discolours dyes and leaves grime in clothes, cars and homes. Steel deteriorates two to four times faster in air-polluted cities than in rural areas. Dirty air deteriorates rubber, causing side walls of tyres to crack, and causes nylon, leather and paper products to deteriorate. The products of incombustion from factories and power plants not only pollute the air, but also waste $300 million worth of sulphur annually (Gordon, 247 Air

Ent~iron. Pollut. (4) (1973) pp. 247-259--O Applied Science Publishers Ltd, England, 1973 Printed in Great Britain

248

RAJINDAR K. KOSHAL, MANJULIKA KOSHAL

1970). In Chicago, it is estimated that air pollution reduces sunlight by 4 0 ~ , resulting in higher lighting bills for its residents (Gordon, 1970). All these facts result from man's growing concern about pollution which has led to research in many directions. Another result of this research is that the popular belief that the earth floats in a vast sea of untainted air and that we possess a super-abundance of life-giving oxygen has been discarded as false. The present theory is that the amount of air supply can be compared to a single thin coat of varnish. Only a fraction of the total air supply is available at any one place on the globe and there is a gradual, but growing, danger that this limited air supply is becoming toxic (Gordon, 1970). Such grave findings seem to have shaken, for a moment, the basic faiths of human life and health. As such, the need for further research in all possible directions seems almost a necessity. This paper is an attempt to measure scientifically the human losses in terms of total mortality and morbidity due to air pollution. At this point, it may be important to point out that, in the United States, more than 70 ~ of the total deaths is due to causes such as heart disease, cancer, tuberculosis, influenza and pneumonia and diseases of the circulatory system (US Department of Commerce, 1969). Many studies have already shown that the mortality rate due to such diseases is high in those areas where air pollution is also high (Council on Environmental Quality, 1970, 1971; Glasser et al., 1967; Lester & Seskin, 1970; McCarroll, 1967; Sprague & Hagstorm, 1969; Winkelstein et al., 1967, 1968). The present study establishes a quantitalive relationship between the total mortality rate and the level of air pollution and other socio-economic variables. This quantitative relationship is then used to estimate the benefits of air pollution abatements. Such quantitative measures seem useful for policy-makers--both in the administrative and legislative capacities--for comparing the cost of abatement of air pollution with estimates of the benefit of such efforts. Earlier, a few studies were conducted on these lines, but all of them were limited in their approach. In other words, they either related to only a single city, like New York, or to only a few diseases. One such study is that of Winkelstein et al. (1967), where total mortality is studied in relation to air pollution levels and economic status. The paper is divided into two parts--the first relating to the role of the suspended particulate in the air and the other to the oxides of sulphur (Winkelstein et al., 1968). The study is quantitative, covers 1959-61 in Buffalo and Erie County, New York, and takes into account deaths due to all causes. The results are quite interesting, in that a positive association between air pollution level or suspended particulate matter and total mortality was found when economic factors were controlled, whereas no significant relationship was evident with regard to levels of oxides of sulphur. On the other hand, there was some indication that sulphur is less important as a cause for mortality than suspended particulates (Winkelstein et al., 1968). There are three other studies with regard to total mortality. They are by Glasser

ENVIRONMENTS AND URBAN MORTALITY--AN ECONOMETRIC APPROACH

249

et al. (1967), McCarrol (1967), and Lave & Seskin (1970). The Glasser study examines the patterns of mortality and morbidity during high levels of pollution, particularly during the Thanksgiving Day weekend of 23-26 November 1966. It was observed that there were 24 more deaths per day during this period than in a normal period and, in all, there was a total of 168 excess deaths. Significant increases were found with regard to heart diseases, vascular lesions of the central nervous system, cancer of the respiratory system and other diseases of the circulatory system. However, the location of the study was New York City. McCarrol's study was similar to Glasser's and was also limited to New York City. It referred both to mortality and morbidity and was pertinent for three years as to the morbidity effect. The result, with regard to mortality, was the same as observed by Glasser et al. The paper by Lave & Seskin is a quantitative study but the data pertain to 1960. A review of these studies indicates that most are limited in their scope because of either narrow geographical coverage or the inclusion of too few diseases. One of the studies in the Nashville Air Pollution series also relates to total mortality (Sprague & Hagstorm, 1969) but, when studied closely, one finds that it is more a study of foetal and infant mortality than the total human mortality. This paper tries to be significant through its broad coverage. It takes into account 40 cities of the United States and the mortality rate of all the diseases. The period of study is eight years, from 1960-1967.

MODEL

The total mortality rate in a city--the mortality rate due to all causes--is assumed to be a function of (i) the level of suspended particulate matter, (ii) the level of benzene soluble organic matter, (iii) population density, (iv) the amount of sunshine, (v) percentage of non-white population, and (vi) percentage of population aged 65 and over. Specifically the functional relation is: M r = aA lbA2cPaSeNYG°lO v

(1)

where M, is the total mortality rate per 1000 population; A 1 is annual arithmetic average of suspended particulate matter measured in/~g/m 3 ; A 2 is annual arithmetic average of benzene soluble organic matter measured in pg/m 3 ; P is the population/ km 2 of area; S is the annual average percentage of days with sunshine; N is the percentage of non-white population; G is the percentage of the population aged 65 and over; U is the random error term; a is a constant and b, c, d, e , f a n d g are measures of partial elasticities of mortality rates with respect to A1, A2, P, S, N and G, respectively. A priori, one would expect b, c, d, f and g to have positive signs and e to have a negative sign. These expectations are based on the fact that most of the studies mentioned earlier indicate a positive relationship between total

250

RAJINDAR K. KOSHAL, MANJULIKA KOSHAL

mortality, both suspended particulate and benzene soluble matter, and population density. Population density measures, among other things, urbanisation, overcrowding and noise pollution. Such factors would have a positive relationship with the total mortality rate. Regarding sunshine, it has been known for a long time that exposure to sunshine usually reduces illness. This might be because sunshine gives a more pleasant and comfortable environment. Non-white people usually live in sub-standard conditions and are liable to higher death rates. On the other hand, given other factors, places where the percentage of old people is high would also have a higher death rate. This functional form of the model (1) is chosen on the basis of statistical fit. Other forms, such as linear, were also considered, but the statistical fit was not as good as for model (1). It may be added that on this criteria we feel that model (1) is a reasonable approximation to reality and that deductions from it are suggestive rather than conclusive.

DATA

Data for this study pertain to 40 cities in the United States. The selection of the cities was limited to those for which the statistics on the level of air pollution were available for the years 1960-67. Data on air pollution, population density and sunshine were collected from the Statistical Abstract of the United States, 1969. Population data were collected from the Editor and Publisher Market Guide, 1967. Numbers of deaths were compiled from the Vital Statistics of the United States, 1967. At this juncture, one needs a word of caution. Due to the limitations of the data and of the model (1), the results of this study should be interpreted with due care. For example, air pollution monitoring systems are spotty in coverage. Sampling stations are generally in downtown areas and the deterioration of air quickly away from these regions, where the greatest amount of industrialisation and urbanisation has been taking place, is often not even monitored (Editor & Publisher Co., 1966).

EMPIRICAL RESULTS

In order to estimate the values of a, b, o, d, e, f and g through multiple regression analysis, model (1) is transformed into a linear model by taking logarithms of both sides. The model (1) reduces to: log Mr = log a + b log A 1 "3t- c log A 2 + d log P + e log S +flogN+glogG

+ U

(2)

ENVIRONMENTS AND URBAN MORTALITY--AN ECONOMETRIC APPROACH

251

where all variables are as defined above. Using the data for 1967 and applying multiple regression analysis, the following results were obtained: log Mr = 0.2426 + 0.1781 log A 1 - 0-0039 log A 2 + 0.0536 log P (2.61) (0.07) (1.39) - 0.2678 log S + 0-0556 log N + 0.6582 log G (1.47) (2.09) (6.44) R 2 = 0.8476

(3)

F-Ratio = 30.59**

The values in parentheses below the coefficients are their t-values. R 2 denotes the coefficients of determination. The F-ratio tests the overall fit. A 1 ~ level of significance is denoted by ** Statistically, the results of eqn. (3) are impressive and the signs of the coefficients, except the coefficient of A 2, are as expected. The coefficient for air pollution levels, A2, is also not significantly different from zero. This is perhaps due to multicollinearity between log A x and log A 2, which increases the standard errors of these coefficients, thus reducing the t-values. Such a multicollinearity is not merely a possibility, but a fact. The coefficient of correlation between log A ~ and log A2 is 0-57. Furthermore, when in eqn. (2) either log A ~ or log A 2 is taken as the pollution index, the t-values for the coefficient of log A ~ and log A 2 and the other coefficients are increased. Equations (4) and (5), calculated in this manner, give the results: log M r = 0.2458 + 0.1758 l o g A a + 0.0537 l o g P - 0.2701 log S (3.00) (1.42) (1.53) + 0.0551 log N + 0.6559 log G (2.19) (6.88) R 2 = 0.8476

(4)

F-Ratio = 37.82**

log M r = 0"3487 + 0"0674 log A z + 0.0785 log P - 0"2399 log S (1.27) (1"95) (1-21) + 0-0536 log N + 0-6493 log G (1-87) (5-87) R z = 0.8161

(5)

F-Ratio = 30.17"*

Thus it is clear that the presence of multicollinearity between log A 1 and log A2 reduces the significance level of A ~ and A 2 in eqn. (3). We should, however, concentrate on eqn. (3) for quantitative results because both types of air pollutions are important from the point of view of both public policy and general health. Furthermore, future data are likely to have such multicollinearity between log A and log A 2.

252

RAJINDAR K. KOSHAL, MANJULIKA KOSHAL

Equation (3) explains about 85 ~ of the variations in the total mortality rates. According to this relationship, a 10 ~ increase in total air pollution, while keeping other factors at a constant level, would bring an increase of about 1.7 (10 • b + 10 • c) ~ in the total mortality rate, which is quite substantial. It is interesting to note that eqn. (3) suggests that, keeping other factors at a constant level, a I0 ~ increase in population density would imply an increase of 0.6 ~ in the total mortality rate in the United States. One policy solution would be somehow to control population density. On the other hand, a 1 0 ~ increase in exposure to sunshine would reduce the mortality rate by 2.7 ~. Man has almost no control over this factor; therefore, for policy purposes, sunshine is not as important a variable as air pollution levels. However, one could also argue that, to a certain degree, air pollution obstructs sunshine and therefore a reduction in the level o f pollution might increase the sunshine. It has been established by Gordon (1970) in his study that pollution reduces sunlight by 40 ~ in Chicago. A ten point increase in the percentage of non-white population would bring an increase of about 0.6 ~o in the death rate. Similarly, keeping all other factors at a constant level, a ten point increase in the percentage of people over 65 years old would increase the death rate by 6.6~. To test the effect of age on mortality rates, two additional models were tried. The results of these models are: log Mr = -0.1950 + 0.1634 log A 1 - 0-0023 log A2 (2.21) (0.04) + 0.0688 log P - 0.2486 log S + 0.0597 log N (1.53) (1-33) (2.17) + 0-2100 log G~8 + 0.7153 log G65 (0.67) (5.34) R 2 = 0.8497

F-ratio = 25.85**

and log Mr = -0"7237 + 0"1550 log A 1 + 0.0408 log A2 (1"98) (0"65) + 0.0473 log P - 0"4565 log S - 0"0151 log N + 1"3557 log G,, (1.02) (2"27) (0-60) (4.83) R z = 0"7988

F-Ratio = 21.84"*

where G1 s is the percentage of population of up to 18 years of age and Gm is the median age in years. These results are quite similar to the results presented in eqn. (3) above.

TABLE 1

1966-67

1965-67

1964-67

1963-67

1962-67

1961-67

1960-67

1

2

3

4

5

6

7

0-2428

0"2217

0'2142

0'2223

0"1742

0"1289

0'1836

log a

0"1842 (2"39) 0"1773 (2"08) 0"1499 (1'72) 0"1534 (1"72) 0"1615 (1 "69) 0'1553 (1"57) 0"1482 (1 "47)

A1

--0'0131 (0"19) 0"0143 (0"18) 0"0418 (0"48) 0'0291 (0"31) 0"0128 (0"12) 0"(222 (0"21) 0"0372 (0"34)

A2

0"0652 (1"73) 0'0672 (1'78) 0"0652 (1"69) 0"0621 (1"60) 0"0645 (1"64) 0"0639 (1"63) 0'0627 (1 "61)

P

S

--0"2553 (1"40) --0"2426 (1"36) --0"2608 (1-44) --0"2773 (1"50) --0"2713 (1 "46) --0'2778 (1 "50) --0"2928 (1 '57)

Coefficient o f log

Notes: (1) Figures in parentheses below the coefficients are their t-values. (2) ** Significant at 1 ~ level.

Average pollution for years

Equation

0'0607 (2"30) 0'0587 (2'25) 0"0581 (2'17) 0"0591 (2'17) 0'0616 (2"25) 0-0614 (2 26) 0'0601 (2"22)

N

0"6537 (6'40) 0"6415 (6"31) 0"6391 (6"14) 0"6447 (6'08) 0"6444 (5"92) 0"6459 (6-00) 0"6419 (6"01)

G

29'97**

29'56**

29"55**

30"02**

30"62**

31'30"*

30"41"*

F-ratio

REGRESSION RESULTS BASED ON AIR POLLUTION AVERAGES FOR DIFFERENT Y E A R S - - L O G M r AS THE DEPENDENT VARIABLE

0"8449

0"8431

0"8431

0"8451

0"8477

0'8505

0'8469

R2

t~h

to

>o

>

©

Z

o -t >

> Z

>

.<

TABLE 2

1966

1965

1964

1963

1962

1961

1960

1

2

3

4

5

6

7

0.4194

0.2635

0.2740

0.3494

0.2963

0-0524

0-1575

log a

0.1442 (1.91) 0.1239 (1.29) 0.o641 (0-79) 0.1546 (1.90) 0.1858 (1-70) 0.0941 (0.88) 0.1028 (1.07)

A1

-0.0182 (0.26) 0.0583 (0.65) 0.0846 (1-02) -0-0347 (0-36) -0.0729 (0.71) 0-0439 (0.44) 0.0803 (0.74)

A2

0.0791 (2.05) 0-0789 (1.99) 0.0665 (1-62) 0.0596 (1.50) 0.0704 (1-73) 0.0650 (1-60) 0.0529 (1.38)

P

-0-2259 (1.21) -0.2063 (1.16) -0.2791 (1.47) -0.2993 (1.51) -0.2654 (1.39) 0.2738 (i .45) -0.3750 (1.9o)

S

Coefficient o f log

Notes: (1) Figures in parentheses below the coefficients are their t-values. (2) ** Significant at 1 ~ level.

Years for which poilution level is used

Equation

0.0663 (2-46) 0-0569 (2-18) 0-0560 (1.97) 0.0653 (2.31) 0.0700 (2.53) 0.0626 (2.32) 0.0559 (2.12)

N

0.6641 (6.40) 0-6374 (6.31) o.6411 (5.89) 0-6766 (6.21) 0.6641 (6.01) 0.6738 (6.45) 0.6506 (6-52)

G

30.87**

27.64**

27.10,,

27.96**

27.89**

31.20**

28.07**

F-ratio

REGRESSION RESULTS BASED ON AIR POLLUTION FOR DIFFERENT Y E A R S - - L O G M r AS THE DEPENDENT VARIABLE

0-8488

0-8341

0.8313

0.8356

0.8353

0-8501

0-8361

R2

e,

z

©

z ~7

t,a

TABLE 3

1966-67

1965-67

1964--67

1963-67

1962-67

1961-67

1960-67

1

2

3

4

5

6

7

0"2262

0"2140

0'2094

0"2554

0"1509

0'1211

0"1932

log a

0-1751 (2'96) 0"1877 (3"13) 0-1798 (2-95) 0"1734 (2"86) 0"1705 (2"78) 0'1715 (2-78) 0"1750 (2.85)

A1

0"0657 (1 "77) 0"0660 (1"80) 0"0615 (1 '65) 0.0602 (1 '60) 0-0634 (1 "68) 0"0622 (1 '64) 0"0603 (1.60)

P

--0"2609 (1-47) --0"2371 (1"36) --0-2~45 (1-39) --0"2654 (1 "49) --0"2673 (1 '49) --0'2714 (1 "51) --0'2798 (1.56)

S

Coefficient o f log

Notes: (1) Figures in parentheses below the coefficients are their t-values. (2) ** Significant at 1 ~ level.

Average air pollution level (A l)

Equation

0-0593 (2'37) 0.0690 (2"42) 0.0622 (2"49) 0'0621 (2"47) 0"0628 (2"48) 0-0632 (2"50) 0-0631 (2.51)

N

0"6480 (6"74) 0'6472 (6-82) 0"6576 (6"88) 0'6574 (6"83) 0-6506 (6"69) 0"6552 (6"75) 0-6565 (6.81)

G

36"90**

36"49**

36'51"*

36'99**

37"54**

38-65**

37"56**

F-ratio

REGRESSION RESULTS BASED ON THE AVERAGE SUSPENDED PARTICULATE MATTER IN MICROGRAMS PER CUBIC METRE IN THE AIR F O R D I F F E R E N T Y E A R S - - L O G M r AS T H E D E P E N D E N T V A R I A B L E

0-8444

0"8429

0-8431

0.8447

0'8466

0'8504

0"8467

R2

to

o >

>

0 Z o

Z m

>,4 ,..]

> Z

0 Z m Z ,-1 >

rtl Z <

1965-67

1964-67

1963-67

1962-67

1961--67

1960-67

2

3

4

5

6

7

0"3174

0-2910

0"2929

0.3232

0.2873

0.2711

0'3171

log a

0"0910 (1.57) 0"1332 (2.18) 0-1473 (2.34) 0.1473 (2-20) 0-1447 (2"11) 0'1496 (2"20) 0.1636 (2.37)

A2

0"0812 (2.05) 0'0856 (2"23) 0.0837 (2.20) 0.0797 (2-07) 0-0835 (2.16) 0"0816 (2-13) 0.0783 (2"06)

P

--0.2473 (1"27) --0.2603 (1-39) -0.2745 (1.47) --0-2894 (1-52) -0-2767 (1 "46) --0.2812 (1.49) --0.3040 (1 "61)

S

Coefficient o f log

Notes: (1) Figures in parentheses below the coefficients are their t-values. (2) ** Significant at 1% level.

1966-67

Average air pollution level (A 2) for years

1

Equation

TABLE 4

0.0535 (1"91) 0.0504 (1.86) 0.0485 (1-80) 0-0482 (1.77) 0"0502 (1 '84) 0.0511 (1.90) 0"0501 (1 "88)

N

0.6405 (5"88) 0"6187 (5.84) 0.6094 (5"78) 0-6104 (5-70) 0.6052 (5"54) 0.6086 (5.67) 0.6058 (5'73)

G

34"34**

33.52**

33"09**

33-51"*

34.20**

33.44**

31.06"*

F-ratio

REGRESSION RESULTS BASED ON THE AVERAGE BENZENE SOLUBLE SOLUTION ( A 2) IN MICROGRAMS PER CUBIC METRE IN THE AIR FOR DIFFERENT Y E A R S - - L O G M r AS THE DEPENDENT VARIABLE

0.8347

0'8314

0"8295

0.8313

0-8341

0.8310

0.8204

R2

C)

z

Z

ENVIRONMENTS AND URBAN MORTALITY--AN ECONOMETRIC APPROACH

257

At this point one could suggest that total disease mortality in 1967 is not due to the air pollution levels of 1967 but perhaps to the levels of pollution for earlier years. To test this proposition, two approaches are taken. Firstly, eqn. (2) is fitted, using the averages of air pollution levels for periods 2, 3, 4, etc., up to eight years prior to 1967. The results of the statistical analysis are summarised in Table 1. Secondly, eqn. (2) is fitted to the air pollution levels of each of the years, starting with 1960. The results of this analysis are summarised in Table 2. A glance at these tables should convince the reader that these results are similar to the results presented in eqn. (3). According to the results shown in Table 1, keeping all other factors at a constant level, a 1 0 ~ increase in air pollution would be accompanied by an increase of about 1.7-1.9 ~o in the total mortality rates. Similarly, Table 2 suggests that a 1 0 ~ increase in the air pollution would be accompanied by an increase of about 1.2-1.8 ~ in the total mortality rates. As shown in eqns. (4) and (5), because of multicollinearity between the two pollution indices, the results are statistically much more significant when the pollution indices are considered separately. We made the same analysis as before, taking the averages of air pollution levels for two, three . . . . . and eight years prior to 1967. These results are summarised in Tables 3 and 4, which suggest conclusions similar to those reached on the basis of Tables 1 and 2, but with a higher level of confidence. However, with the present state of our knowledge, it is not possible to pinpoint the lag between contact with the air pollution and the time of death due to various diseases. Nevertheless, our analysis suggests that a 10 ~ increase in the air pollution would be accompanied by an increase of about 1-2-1-9~ in the total mortality rates due to various diseases in the United States.

ECONOMIC COSTS

For the cities under study, the range for suspended particulate matter in 1967 is 38 to 169/~g/m 3 and the range for benzene soluble organic matter is 1-8 to 12.7 pg/m 3. Thus, air pollution in many of the cities is more than six times higher than that of the lowest level of air pollution existing in the sample. Now, the question which policy-makers might like to ask is: What would be the social savings in terms of reduction of all disease when, on the average, air pollution is reduced, by, say, 5 0 ~ ? In order to answer this question we need data not only on mortality but also on morbidity from all diseases. Unfortunately, the data on morbidity for these cities are not available. However, overall estimates of costs of morbidity and mortality due to all diseases for the United States are available in a detailed study by Rice (1966). These costs for 1963 are listed in Table 5. This table shows that total economic costs in 1963 amounted to about $93,500 million It may be pointed out that mortality costs are the sum total of discounted (4 ~ ) future earnings lost

258

R A J I N D A R K. K O S H A L , M A N J U L I K A K O S H A L

for all persons who died in 1963. These calculations take into account life expectancy for different age and sex groups, the changing pattern of earnings at successive ages, varying labour force participation rates and imputed value of housewives' services. According to our statistical analysis, an overall reduction of 50 ~ in the air pollution would reduce the total mortality rate by 6 to 9-6 ~ . Assuming a similar relation for mortality and morbidity with air pollution, a 50 ~o improvement in the air would imply a social savings of $3400-5400 million/year.t In terms of 1967 prices and number of deaths, these savings would even be of a higher order. TABLE

5

TOTAL ECONOMIC COSTS OF ALL DISEASES

Categories

Millionsof dollars

Direct Expenditures Morbidity Total Mortality

22530'0 21042.2 49928.1

Total

93500.3

Source: Rice (1966). At this stage one may ask the question: Are costs of air pollution abatement more than its benefits? To answer this question, we need to add to the above estimates other losses due to air pollution. According to the Council on Environmental Quality (I 97 I) the direct costs of air pollution on both material and vegetation are estimated at $4900 million annually. In addition, $5200 million per annum are lost in terms of lower property values due to air pollution. Thus, the annual toll of air pollution on human health, vegetation, material and property values would be in the range of $13,500 to 15,500 million per annum. On the other hand, to meet the present air quality standards, which would reduce the bulk of these damages, the nation would have to incur an annual cost of $4700 million by 1975 (Council on Environmental Quality, 1971). Thus, these estimates suggest that the value of benefits is far greater than the costs of air pollution abatements.

CONCLUSIONS

As pointed out earlier, due to the limitation of the data, and the present technical knowledge about air pollution and its effects, the results of this study should be taken as suggestive rather than conclusive. Our analysis suggests that a 50~o increase in the air pollution would increase the mortality rate by about 6-9.5 ~ . t These calculations are based on the fact that in 1967 about 60 ~ of the total deaths occurred in the urban areas. Thus $3"4 = $93.5 x 0-06 x 0-60 and $5.4 = $93'5 x 0.095 x 0'60.

ENVIRONMENTS AND URBAN MORTALITY--AN ECONOMETRIC APPROACH

259

A r e d u c t i o n o f a b o u t 50 % in the air p o l l u t i o n w o u l d i m p l y a social savings o f the o r d e r o f $3400 to 5400 m i l l i o n / y e a r in terms o f all diseases. This study also suggests t h a t the air p o l l u t i o n a b a t e m e n t benefits are far greater t h a n its cost.

ACKNOWLEDGEMENTS This p a p e r was w r i t t e n d u r i n g the first a u t h o r ' s s a b b a t i c a l at the I n d i a n Statistical i n s t i t u t e (Planning Unit), N e w Delhi. T h e a u t h o r s are t h a n k f u l to Professor N. B h a t t a c h a r y a a n d Professor P. M a l l e l l a for their c o m m e n t s a n d suggestions, also to D r J. W. Doering, M. D. L o g a n , Ohio, f o r clarifying certain medical terms. T h a n k s are also e x t e n d e d to W a l t e r Stewart for his help with the c o m p u t a t i o n a n d collection o f the data. A l l c o m p u t a t i o n s were m a d e at O h i o University C o m p u t e r Center.

REFERENCES COUNCIL ON ENVIRONMENTALQUALITY (1970). Environmental quality--the first annual report. Washington D.C., US Government Printing Office. COUNCIL ON ENVIRONMENTALQUALITY(1971). Environmental quality--the second annual report. Washington D.C., US Government Printing Office. EDITOR & PUBLISHERCOMPANYINC. (1966). Editor and Publisher Market Guide, 1967. New York. GLASSER,M., GREENBURG,L. & FIELD,F. (1967). Mortality and morbidity during a period of high levels of air pollution. Archs envir. Hlth, 15, 684-94. GORDON, A. H. (1970). Air pollution Part L Law and the Municipal Ecology: air, water, noise, over-population, 1-3. National Institute of Municipal Law Officers. LAVE,L. B. & SEESKIN,E. P. (1970). Air pollution and human health. Science, N.Y., 169, 723-33. McCARROLL, J. (1967). Measurements of morbidity and mortality related to air pollution. J. Air Pollut. Control Ass., 17, 203-9. RICE, D. P. (1966). Estimating the cost of illness, 1963. US Department of Health, Education and Welfare, Public Health Service, Series No. 6, 109. SPRAGUE, A. H. & HAGSTORM,R. (1969). The Nashville air pollution study: mortality multiple regression. Archs envir. Hlth, 18, 503-7. US DEPARTMENTor COMMERCE(1969). Statistical abstract of the United States, 1969. Washington, D.C., US Government Printing Press. US DEPARTMENTOF HEALTH,EDUCATIONANDWELFARE(1969). Vital statistics of the United States, 1967. Washington, D.C., US Government Printing Press. WINKELSTEIN,W., JR., KANTOR, S., DAVIS,E. W., MANERI,C. S. & MOSHER, W. E. (1967). The relationship of air pollution and economic status to total mortality and selected respiratory system mortality in men, I. Suspended particulate. Archs envir. Hlth, 14, 162-9. WINKELSTEIN,W. JR., KANTOR, S., DAVIS, E. W., MANERI,C. S. & MOSHER, W. E. (1968). The relationship of air pollution and economic status to total mortality and selected respiratory system mortality in men, II. Oxides of sulfur. Archs envir. Hlth, 16, 401-5.