Environmental Policy Challenges of the Future: A Longitudinal Analysis of Atlanta’s Air Quality Jennifer Chirico Douglas Noonan Georgia Institute of Technology ______________________________________________________________________________ Abstract Determining the future of Atlanta’s air quality is paramount to public policy makers as they make decisions about the appropriate resource allocations for air quality improvements. The EPA recently strengthened standards for ozone and lead in 2008, creating greater challenges for the City of Atlanta, which was out of attainment for ozone most years in the 1990s. We conduct a longitudinal study examining the air quality of Atlanta over a 50 year time frame and the extent to which the recent increase in air quality standards might affect the City of Atlanta. Our study is two fold: i) we compare and contrast the history of major air pollutants in Atlanta to key air quality determinants such as population, vehicle miles traveled (VMT), industrial activity, and technology developments; ii) we utilize a multivariate fractional polynomial regression that forecasts a relationship between the economic indicators and ambient concentrations based on our historical data. Our findings reveal that Atlanta has experienced great advances in air quality, yet rising populations, VMT, and industrial activity are not positively related to ambient concentration levels, making forecasting air quality a difficult task. While our estimations reveal that ambient concentrations will tend to consistently fall and then asymptote near a low level, our results point to technology and policy as playing a dominant role in the air quality system. We conclude that forecasting air quality appears to pose an enormous practical challenge to any agency tasked with monitoring it and achieving greater standards. This research is based in part on research funded by EPA STAR grant, “Air Quality, Emissions, Growth, and Change.” The U.S. Environmental Protection Agency (EPA) made recent modifications to national air quality standards for several criteria air pollutants. These modifications may create greater challenges for cities in the U.S. that were previously out of attainment in one or more of the criteria air pollutants. The City of Atlanta was out of attainment for ozone for most of the years in the 1990s, and now it must reevaluate its goals for air quality and determine how it will work to meet the new standards. Determining the key determinants of poor air quality in Atlanta is paramount for policy makers as they make decisions about the appropriate resource allocations for meeting these new federal standards. This study examines the air quality in Atlanta over the last 50 years and uses this information to estimate the future of Atlanta’s air quality through 2050. There are two primary components of this study: i) to compare and contrast the history of major air pollutants in Atlanta to key air quality determinants such as population, vehicle miles traveled (VMT), industrial activity, and technology developments; ii) to utilize a multivariate fractional polynomial regression that forecasts a relationship between the economic indicators and ambient concentrations based on historical data. Section I of this paper provides background information on air quality and how it has changed over time in Atlanta. Sections II discusses relevant literature on air quality standards and determinants. Section III provides the data and methodology utilized. Section IV analyzes provides the results of the historical analysis and the forecasted data analysis of the six criteria air pollutants and key determinants of air quality. Section V provides a discussion of the results, and Section VI concludes with the policy implications and recommendations for Atlanta. 2 I. Air Quality Background Up until the 20th century, the term “air pollution” did not exist. Air pollution was referred to as ‘smoke,’ ‘kiln,’ ‘nuisance,’ and ‘coal’ (Bowler and Brimblecombe, 1992). During the first hundred years of the United State’s existence, “smoke” complaints were considered a common law nuisance and were resolved by litigation among involved parties instead of through legislation. Around 1881, the first legislation was enacted that declared smoke emissions a public nuisance, and by the early 1900s, smoke was prohibited if it was considered “dense,” “black,” or “grey.” Even as late as the 1950s when air control programs were becoming more common at the city level, “air pollution” was not yet an official term and was still referred to as “dirty air arising from coal combustion” (Sabatier, 1988). Federal legislation on air pollution in the U.S. was not enacted until 1970. After the passage of the Clean Air Act (CAA) in 1970, air pollution control became the responsibility of the U.S. Environmental Protection Agency (EPA). The new, strict regulations imposed by the CAA coincided with a change in the public perception of air quality. The public began to perceive air pollution as not only dirty, but also unhealthy. In addition, the perception of the sources of air pollution shifted from coal-burning and factory emissions to automobiles (Sabatier, 1988). The passage of the CAA also accompanied the first Earth Day celebration, a rise in environmental concerns that coincided with rising affluence and scientific understanding of air pollution problems. Under the CAA, the EPA established National Ambient Air Quality Standards (NAAQS) that included six criteria air pollutants: lead (pb), particulate matter (pm), sulfur dioxide (SO2), and nitrogen dioxide (NOx), ozone (O3), and carbon monoxide (CO) (EPA, 2006a, 1999). The National Ambient Air Quality Standards (NAAQS) provides national air quality standards for each pollutant. 3 Prior to the passage of the CAA, direct historical evidence of Atlanta’s air quality is quite limited. This poses a serious challenge for understanding Atlanta’s air pollution problems in a longer context. With the exception of lead, no systematic, direct measures of concentrations of other criteria pollutants are publicly available. Furthermore, it poses an enormous problem identifying the determinants of air quality changes and, in particular, the role of the 1970 extensions to the Clean Air Act. The conventional wisdom and mental models of discussants (as well as many formal air quality models) posit very close relationships between pollution and these measures of economic activity.1 The relationship is often thought to be quite direct, even if a bit noisy. Holding all else equal, it is commonly expected that ambient concetrations of pollution will rise when population, industrial activity, transportation, and temperatures rise. Population in Metro Atlanta has been on the rise since the 1940s, and industrial activity is a primary component. Metro Atlanta is home to more than 137,000 businesses and serves as the primary hub for transportation and manufacturing in the Southeastern U.S. (Georgia Department of Labor, 2005). Trade, transportation, and utilities (TTU) occupy the largest percentage (22.9%) of the industry sector, and currently employ about 23% of the workforce (City of Atlanta Chamber of Commerce, 2006). Atlanta has the largest rail center hub in the South, is home to four major interstates for truck and car transport, and has the busiest airport in the world, with more than 1,200 flights departing per day and more than 85 million passengers traveling through the Atlanta airport per year (Metro Atlanta, Chamber of Commerce, 2006). Manufacturing is the fifth largest industry sector in Metro Atlanta (Metro Atlanta Chamber of Commerce, 2006). In addition, technological advances in the automobile may also play a significant role in the air 1 Even the simple heuristic common to environmental studies of I=PAT (where I=impact, P=population, A=affluence, and T=technology) supports the intuition that as Atlanta’s population, affluence, and use of emitting technologies increases, its air will become more polluted. Obviously, a key variable in this relationship is technology, which has changed dramatically over time. 4 quality of Atlanta. With the extensive suburban growth in Atlanta, this rate has continued to grow, and today Metro Atlanta residents utilize the private automobile more than ever. Against this trend in rising automobile use are the changes in automobile technologies, most notably the introduction of the catalytic converters and new fuel mixes (e.g., unleaded and reformulated gasolines). In addition, the EPA claims that vehicle miles traveled (VMT) will continue to rise, and that the VMT growth rate for Atlanta is expected to be 103 percent between 1996 and 2030 (EPA, 2006c). According the Census, average commute times in Metro Atlanta grew by 20% from 1990 to 2000, compared to a 14% growth nationwide. Atlanta now has the highest traffic congestion growth rate (76% between 1993 and 2003) in the country (Metro Atlanta Chamber of Commerce, 2006). It has surpassed large urban cities, such as Denver, Houston, and Dallas. The roads in Atlanta are more congested, and the average number of hours commuters are delayed due to congestion increased by 43% between 1993 and 2003 (Metro Atlanta Chamber of Commerce, 2006). Atlanta also experiences very hot summers. The climate in Atlanta can directly affect ozone (which requires favorable weather conditions for its formation) and indirectly weather affects emissions (e.g., more air conditioning and driving in the hot summer). II. Literature Air quality standards for ozone, particulate matter, and lead were recently strengthened by the EPA. Ozone standards decreased from .08 to .075 ppm and lead decreased from 1.5 ug/m3 to .15 ug/m3 in 2008. Particulate matter (pm) standards also strengthened in 2006. Several reports have analyzed the effects on this change and found that there are many cities and counties who are projected to not be able to meet the new standards (EPA, 2008). The PAG Southwest Air Quality Review (2008) issued a report claiming that the number of counties out of attainment for ozone would increase from 104 counties to 398 counties with the new legislation. When the 5 particulate matter standard changed, the EPA speculated that the number of counties out of attainment for pm would increase from 39 to 141 (EPA, 2006f). Other state and municipality reports have found similar results with other pollutants, such as carbon monoxide (Municipality of Anchorage, 2008; Utah Division of Air Quality, 2008) and nitrogen oxides (Utah Division of Air Quality, 2008). Recent literature has also been conducted on the determinants of air quality and how they might affect future air quality. Warm temperatures are a necessary condition for some air pollutants and, therefore, warming trends could lead to greater amounts of air pollution (Georgia State Climatology Office, 2006). Climate change leading to warming temperatures in the future can potentially affect air quality (Tagaris, 2007), transportation, and emissions (Tagaris, 2006). Tagaris et al. (2007) conducted a study that simulated future pollutant concentrations based on past concentrations to examine potential changes in temperatures. They found that most pollutants, including ozone, will continue to decrease into the future in Atlanta. Transportation is also deemed important to air quality, as a recent report in Bridgeport, Connecticut claimed that transportation is the major contributor its air pollution (Bridgeport Regional Planning, 2007). This study observes air quality in concert with key determinants of air quality to determine what affects Atlanta’s air quality and what policy makers can expect air quality to be in the future. It is unique in that it is timely to the recent changes to air quality standards and examines Atlanta’s air quality over a 100 year time period. Air quality regulators are projecting that many places will be out of attainment and violate NAAQS into the distant future under these new standards. Being out of attainment is costly for cites and requires great effort to meet standards. Therefore, this study analyzes the history of air quality followed by a statistical analysis that forecasts the future of Atlanta’s air quality. Results indicate that, while air quality is 6 improving generally, measuring air quality is not a simple matter and that policy and technology may play the most prominent role in achieving desired air quality goals. III. Data and Methodology Data for this study was collected from several resources. The dependent variables in the models are the air pollutants monitored by the EPA and the independent variables are the indicators of air quality. Dependent Variables The historical criteria air pollutant concentrations were collected from archive records at the EPA. The concentrations date back to 1958 for some variables (e.g., tsp, lead) and are coded based on the EPA’s corresponding measuring unit (e.g, ppm). Lead: Lead is one of the earliest pollutants measured. Historically, lead emissions came from leaded gasoline from cars and trucks, and today most lead emissions come from industrial operations (EPA, 2006b). As illustrated in Figure 1, annual lead concentrations in Atlanta substantially decreased in the late 1970s and continued to steadily decline through 2005 (EPA, 2003, 2007 ). However, despite the substantial decline, the new NAAQS standard is only slightly above current lead levels. Particulate Matter: Particulate matter (PM) measurements are usually split into two categories: 1) PM10 which is made up of inhalable coarse particles that are smaller than 10 and larger than 2.5 micrometers; and 2) PM2.5 which is made up of fine particles that are 2.5 micrometers in diameter or smaller. PM10 was not systematically measured by the U.S. EPA until the late 1980s, and PM2.5 was not measured until the late 1990s. Prior to formal measurements of PM10 and PM2.5, the EPA measured particulate matter in the form of total suspended particles (TSP). TSP, while not a criteria pollutant as such, is a common measure of air pollution in 7 international studies. TSP is made up of particles that are suspended in the air and are larger than 10 micrometers in diameter. Public measurement records of TSP date back into the 1950s, which provides a good indicator for examining older trends in PM10 and PM2.5. Emissions of PM2.5 are usually in the form of smoke coming from power plants, industries, and automobiles reacting with the air (EPA, 2006b). Figure 2 depicts the TSP and particulate matter concentrations for Atlanta. Overall, TSP varied dramatically between 1957 and 1975 and then steadily declined from 1975 into the 1990s. PM10 decreased by about 55% from 1986 to 2005 and remained well below NAAQS (the EPA standard). PM2.5 only slightly decreased (11.6%) from 1999 to 2005 and remained consistently above NAAQS (EPA, 2007). Ozone: Ozone is one of the most discussed pollutants due to its negative health consequences, such as asthma, wheezing, and reduced lung capacity (EPA, 2006b). It is created when volatile organic compounds and nitrogen oxide react with sunlight. The focus here is on ground-level ozone rather than stratospheric ozone.2 Figure 3 illustrates Atlanta’s ozone trends. From 1987 to 2005, the ozone concentration fluctuated up and down and had an overall decrease of about 17 percent. For most of these years, Atlanta’s peak ozone concentrations were above the NAAQS and well above the new NAAQS standard. Sulfur dioxide and Nitrogen Dioxide: Sulfur dioxide (SO2) is in all raw materials that contain common metals. It is an air pollutant that is formed when fuel is burned. It can cause serious respiratory problems to humans, and it facilitates the production of acid rain, which can cause considerable damage to plants and water animals (EPA, 2006b). Nitrogen dioxide (NO2) is a part of the nitrogen oxide (NOx) family of reactive gases that contain various amounts of nitrogen and oxygen. NO2 often creates a reddish-brown layer of air over urban areas and is a component of automobile exhaust fumes. Figure 4 depicts the trends in Atlanta’s SO2 and NO2 2 “Good” ozone rests 10-30 miles above ground in the stratosphere, and “bad” ozone is at ground level (EPA, 2006). 8 levels over the past 30 years. The average annual SO2 concentration dropped dramatically in 1973 and then steadily decreased through 2005, while NO2 steadily decreased over the thirty year time period (EPA, 2007). Carbon monoxide: Carbon monoxide (CO) is the final of the six criteria pollutants regulated by the EPA. Nationwide, the EPA estimates that 56% of CO emissions come from onroad vehicles, while 22% derive from non-road vehicles (e.g., construction equipment, boats). In Atlanta, CO concentrations gradually decreased since 1972, falling below the NAAQS level in the early 1980s (EPA, 2007). Figure 6 shows some of the population trends from Census data. Atlanta’s population peaked in 1970 (496,973), at which point the population began to drop until 1990 (394,017). --Insert Figures 1-5 Here— Independent Variables Population, industrial and manufacturing activity, VMT and transportation, climate, and technology are contributing factors which routinely receive the bulk of regulatory and planning attention. Understanding the determinants of air quality can help one predict or forecast air quality trends. These determinants are the independent variables in our model. By observing the relationship between these indicators and measured air quality in Atlanta, trends in Atlanta’s air quality prior to formal measuring (“backcasting” before 1970) and in air quality after today (forecasting after 2005) can be estimated. Population: Population data was collected from the U.S. Census (1998 2006). There was a resurgence of Atlanta’s population in the late 1990s and early twenty-first century, reaching almost 471,000 in 2005 (U.S. Census Bureau, 1998, 2006). The Metro Atlanta region experienced similar growth, without the decline after the 1970s. Today, Metro Atlanta includes 9 8,376 square miles, has a population of nearly 5 million, and is expected to grow to nearly 6 million by 2015 (City of Atlanta Chamber of Commerce, 2006). Industrial Activity: Manufacturing data was found on the Bureau of Economic Analysis on Regional Economic Information System (BEA 2006). Metro Atlanta’s industrial activity generally increased along with its population. The manufacturing industry grew from 3,513 establishments in 1990 to 4,997 in 2006 (Georgia Department of Labor, 2005). Interestingly, employment in the manufacturing industry had the opposite effect—overall, it slightly decreased over the last several decades from about 190,000 employees in 1969 to approximately 185,000 in 2004 (see Figure 7). More, smaller establishments are producing increasingly valuable manufacturing output. Although the manufacturing sector in Metro Atlanta is hardly shrinking in raw size (measured in terms of real earnings), its share of total earnings has declined from 22 percent in 1969 to under 8 percent today (see Figure 7). VMT: Vehicle miles traveled data was collected from the Georgia Department of Transportation (GDOT 2006). Atlanta’s VMT increased by 55% between 1990 and 2004 (see Figure 8). The 3.2% annual growth rate for VMT for the City of Atlanta lags slightly behind Metro Atlanta’s 3.4% population growth rate over the same time span (Georgia Department of Transportation, 2006). Climate: Atlanta climate data was found at the Georgia State Climatology Office (2006) out of the University of Georgia The temperatures in Atlanta fluctuate from year to year, but have experienced a slight overall increase since 1940 (see Figure 9). Atlanta experienced its highest recorded average annual temperatures in the 1990s. --Insert Figure 6-9 Here— 10 At a national level, the EPA regularly produces reports to tout the efficacy of their efforts. Figure 10 reprints a graph from one of the EPA’s recent reports (U.S. EPA, 2006a). While reductions in emissions are not the same as reductions in ambient concentrations, the steady declines over the past 25 years suggest some success. This accomplishment occurs alongside fairly steady increases in population, VMT, energy use, and economic activity. Comparable measures for the Atlanta area suggest a similar story by overlaying Figures 1-5 on top of Figures 6-9. --Insert Figure 10 Here— IV. Methodology The methodology for this study uses a straightforward and flexible statistical modeling approach to forecast Atlanta’s air quality. A multivariate fractional polynomial regression is estimated separately for each of the major pollutants. This approach fits a regression of the annual ambient concentration measure on several economic construction earnings, real transportation earnings, and average temperatures). In fitting the regression, the best-fitting fractional polynomials of the predictors are found.3 This allows us to estimate the ambient concentration measure for each year as a flexible nonlinear function of the economic indicators. While this approach offers a richer description of the data and enables the estimation of confidence intervals, it maintains its naïveté regarding atmospheric chemistry, regulatory changes, technological innovation, and many other important details. Rather, this parsimonious model incorporates only a handful of variables that are commonly linked to air pollution policy and debates. It does so without restrictions on the way in which they might enter the model or explain data. If, controlling for other factors, increases in VMT are associated with cleaner air, 3 The polynomials are chosen from powers of -2, -1, -0.5, 0.5, 1, 2, 3, and ln. Up to four power terms can be chosen for each predictor. 11 then the model estimates will reflect this. Interpreting the model, therefore, requires some caution, because important intervening factors may not be controlled for and might account for the observed relationships. Before proceeding with the multivariate forecasting exercise, an additional remark is necessary about forecasts of the economic indicators that serve as “independent variables” or predictors. The fractional polynomial regressions are based on observed values during roughly thirty years (1972-2005), with a few missing values imputed. The model estimates a relationship between the economic indicators and ambient concentrations. To predict the concentrations in the future requires some knowledge about the future values of the predictors.4 We accomplished this by producing forecasts of the economic indicators in a straightforward manner. A basic (univariate) VAR model is estimated for the Metro Atlanta population with a three period lag, and then forecast out 50 years. The same is done for the temperature and VMT variables. Then, a two-equation VAR model is estimated for the Metro Atlanta population variable and one other economic indicator (e.g., real per-capita earnings). This allows for the economic indicator to be determined by its own time path and the time path of the metropolitan population. This twoequation VAR model is repeated for the population variable and for each of the other economic indicators (i.e., real earnings in manufacturing, construction, and transportation). Annual forecasts are then made out 50 years. These forecasted values are used to generate the forecasts of air quality measures after estimating the fractional polynomial regression models. A sample of the forecasts is depicted in Figure 11. In Figure 11A, the gray and blue dots reflect actual populations for the city and the MSA, respectively, while the orange and maroon dots are their forecasted values. In Figure 11B, the gray and maroon dots are historical income and VMT 4 This approach does assume that these predictors are exogenous, or that the economy is not itself influence by air quality. There is some evidence to suggest that this assumption may not hold. Addressing the endogeneity of economic indicators is beyond the scope of this paper. 12 values, respectively, while red and green dots represent their forecasts. In Figure 11C, the blue, green, and gray dots represent historical real earnings in manufacturing, construction, and transportation, respectively. Their forecasted counterparts are in maroon, orange, and red. --Insert Figures 11A-11C-V. Results The resulting forecasts of air quality measures offer some insight beyond what is already suspected. Perhaps, however, the biggest lesson is not in the central tendency of the forecasts but rather the scope of the uncertainty (i.e., confidence intervals) around those forecasts. These forecasts treat the projected trends in population and economic indicators as though they are certain or inevitable. Even with this unrealistic assumption of certainty, the fractional polynomial regressions produce forecasts with enormous uncertainty. Forecasts for lead, ozone, SO2, NOx, and CO fall into three general categories. Forecasts for lead and ozone both show rapidly rising levels with overwhelming noise in the prediction. For (8-hour) ozone, the polynomial fits the data fairly well (adjusted-R2 = 0.69), although notably worse than for the other criteria pollutants that follow. Most terms enter linearly, although only VMT and construction earnings do so significantly at the 0.10 level.5 The forecasted ozone levels are presented in Figure 12A. Here, the forecasted levels (in maroon) quickly fall below zero before rising rapidly after 2030. Although the VAR forecasts predicts leveling off of ozone levels, the multivariate analysis with continued growth of the economic indicators (e.g., VMT) overwhelms the ozone trend in the coming decades. For lead, the polynomial regression also fits the data well (adjusted R2=0.79). Here, temperatures (negatively) and transportation (positively) predict lead levels. The forecasted values in Figure 12B shown, 5 Curiously, there is a positive time trend and temperature is negatively related to ozone, although neither is significant beyond the 0.30 level. 13 similar to ozone in 12A, rapidly rising lead levels in coming decades. Even though recent history suggests lead levels are asymptoting near zero, the growth in the transportation sector augurs higher lead levels in this approach. These inauspicious forecasts must be interpreted in light of the models’ limitations as well as the other dominant feature of Figures 12A and 12B: the enormous confidence intervals. The models yield very little precision in their forecasts of ozone and lead, even when their prominent economic determinants are assumed with certainty. The second set of models, for SO2 (see Figure 12C) and NOx (see Figure 12D), continue the bleak forecasts. These forecasts resemble those for ozone and lead, except that their confidence intervals are somewhat smaller and tend to not include negative values of ambient concentrations. Accordingly, the forecasts for SO2 and NOx are both fit with more precision (adjusted R2=0.96 for both). Income and VMT, nonlinearly, predict increases in SO2, while transportation earnings predict decreases. NOx levels decline over time and, nonlinearly, in VMT, but rise in income and also depend significantly on the manufacturing sector. SO2 levels are forecast to rise, surpassing the NAAQS threshold before 2020. NOx levels are also forecast to rise, surpassing the threshold after 2030. Both of these trends diverge from the VAR forecasts that predicted concentrations asymptoting to much lower levels. Furthermore, considerable uncertainty in the estimates remain, although less so than for ozone and lead. The forecasts for CO (Figure 12E) resemble the others in its uncertainty, but differ in the expectation of rapidly falling CO levels. The model fit for CO is good (adjusted R2=0.91). There is a positive time trend in CO levels, which are expected to fall at an increasing rate as population rises and to rise at a decreasing rate as VMT grows. Thus, over the projected growth scenario for Atlanta, the CO concentrations are forecast to continue falling. While the forecast of negative CO levels makes little sense, the prevailing negative relationship between Metro 14 Atlanta population size and CO concentrations does suggest that further growth may not exacerbate CO pollution. --Insert Figures 12A-12E-VI. Discussion Prior to 1970, very little was known about the distribution of the criteria pollutants in many areas. However, since the 1970s, Atlanta has experienced a downward trend in most air pollutants. Lead decreased significantly between 1971 and 1987 and then remained relatively stable through 2005. Recent concentrations for lead, however, are only slightly below the new policy standards, which could potentially pose threats to unattainment for Atlanta. PM10 decreased overall since 1986, and PM2.5 only decreased slightly since 1999, thereby making it borderline for meeting NAAQS standards. Ozone fell before 1987 and then fluctuated between 1987 and 2005. While there was an overall decrease (17%) in ozone, the concentrations were above the NAAQS for most years and well above the recent NAAQS modifications. Sulfur dioxide dropped significantly in 1973 and then steadily decreased through 2005. Nitrogen dioxide and carbon monoxide steadily decreased between 1972 and 2005. For the key determinants, results suggest that rising populations, VMT, and industrial activity are not positively related to ambient concentration levels. If anything, they are negatively related to them.6 As Atlanta has experienced a population and transportation boom since 1970, 6 Glancing at the pairwise correlations between ambient concentrations of the six criteria pollutants and economic measures (population, real per capita earnings, real manufacturing earnings, real transportation earnings, and average temperatures) confirms this. Each of the pollution levels has a negative and statistically significant correlation (at a 5% level) with each of the economic variables; the only exception being PM 10, which lacks sufficient observations. Many of these correlations are quite strong, and all are negative. The same set of pairwise correlations, when examined in first-differences, exhibits no significant correlations. Annual changes in ambient concentrations, up or down, appear uncorrelated with annual changes in the economic indicators. This suggests that the air quality and economic variables may be structurally unrelated, but both enjoy their own independent (and opposite) trends. Interestingly, this pattern also holds for the pairwise correlations between the changes in concentrations and the changes in economic indicators in the preceding year. One exception to this arises for NOx, 15 its air quality has noticeably improved (or, perhaps in the case of ozone, stagnated). The data might better support a conclusion that people and traffic cause cleaner air, not dirtier air. While the recent trends clearly depict improving air quality and increasing economic trends, perhaps the most noteworthy finding is that the new modifications to the NAAQS standards for ozone, lead, and particulate matter may be difficult to overcome for Atlanta. Currently, Atlanta does not pass the new standard for ozone and only barely passes the standard for lead and particulate matter. These findings suggest potential challenges for the City of Atlanta. However, the limitations of this study prevent us from making forecasts with any degree of certainty. The graphs and bivariate correlations are inadequate to form predictions of what might be a much more complex system. It is possible that there is no relation between the economic indicators and the air quality measures. But it is also possible that the relationship appears in more subtle ways. For instance, a steady decline in ambient pollutant levels associated with the passage of time (perhaps because of technological innovation or regulation) might well fade over time and be overcome by the effects of population, VMT, or some other determinant experience rapid growth. In addition, measuring air quality is not a simple matter in the best circumstances. Air quality may vary substantially over space. The airshed in Atlanta is quite large, and a single “air quality index” value might not do well to describe the air quality in the area. Air quality can also vary widely over time, whether time is measured in hours or in months. The graphs above depict variation at an annual level, but they obscure important and dramatic variations at smaller time scales. Ozone levels, for instance, fluctuate dramatically over the course of a day, can vary where NOx levels tend to rise in the year following increases in income and manufacturing and decreases in temperatures. 16 widely from day to day, and certainly have important seasonal fluctuations as well. Air quality is also multidimensional with respect to the many chemicals that constitute the ground-level atmosphere. Atlanta may be relatively clean with respect to some pollutants, like lead, but remains polluted with respect to others, like PM2.5 or even airborne carcinogens.7 Measuring air quality, across the many dimensions of time, space, and chemicals, poses an enormous practical challenge to any agency tasked with monitoring it. It may be little wonder that the data that do exist suffer from coding errors and other possible outliers. (e.g., the average and maximum SO2 levels fell sixfold from 1972 to 1973). Furthermore, omitted variables, such as technological and policy events can lead to biases. VII. Conclusion Overall, the story of Altanta’s air quality is dynamic and, thus far, promising. Despite powerful growth trends in most of the major urban air pollution indicators, Atlanta has experienced little deterioration and, in most instances, great advances in quality. Recent history has seen major improvements in many air pollution levels. This has occurred against a backdrop of growing population, rising VMT, and growing manufacturing and construction sectors. Clearly, either the simple model used in policy discussions (more cars and people and industry means worsened air quality) has not applied to Atlanta or something is missing. Most likely, the story of Atlanta’s air quality history has been dominated by technological shifts. These changes have made the other factors second-order at best. Whether the technologies have been “forced” via regulation (e.g., unleaded gasoline, catalytic converters, smokestack scrubbers) or arisen with macroeconomic shifts (e.g., growth of service- and information-based industries) or 7 This analysis has focused on the criteria air pollutants. These chemicals fall under a separate regulatory category from air toxics, or cancer-causing chemicals. These air toxics might pose some of the greatest threats to public health, although they are only more recently receiving much attention. For the purposes of this paper, however, it should suffice to note that trends in air toxics in Atlanta have roughly followed similar patterns to criteria pollutants like SO2. 17 something else is unclear from this analysis. Moreover, simply examining trends cannot identify the counterfactual or “what would have happened” in the absence of one factor or another. At this point, even after 30 or more years of intensive federal air quality regulation, there are a great many unanswered questions about urban air quality. These ambiguous and mixed results point to technology and policy as playing a dominant role in the air quality system. Understanding the innovation and diffusion of new technologies that impact air quality is no small feat, and predicting it is still tougher. Regulations can favor certain technologies, with both positive and negative impacts on air quality. A holistic approach to the air quality system appreciates the complex interactions between technology, policy, economics, and the environment. Just as one affects the other, those changes are likely to feedback and influence the other aspects of the system. These feedback loops can be particularly important in making long-term forecasts about Atlanta’s air quality. And, to date, very little is understood about these mechanisms. Most air quality models take economic and policy trends as “exogenous” to the model, or completely determined from outside the system. Yet, as Atlanta struggles to manage its growth, it is clear that policies react to changes in both environmental quality and economic developments. Many economic development models likewise take air quality as fixed. Integrating the subsystems involves a major interdisciplinary undertaking. To date, far too little effort has gone into understanding how (technology and policy) innovation responds to environmental and economic dynamics, and vice versa. The shortcomings of the conventional predictions (more traffic, population, construction leads to worsening air quality) reflect this “partial” modeling of the situation. 18 The rest of the system remains a mystery – one for political debate and opportunism. Lacking the ability to forecast air quality with much confidence, we might be content in knowing that the sky is neither falling nor pristine. Our air quality fate is likely to be determined by the winds of change that blow elsewhere. This may take the form of larger technological innovations (perhaps induced by policy, perhaps by continuing deindustrialization) that are difficult to foresee by just looking at air quality monitoring stations and regional planning documents. While we have made tremendous progress in improving air quality over the last several decades, important gaps remain in our understanding of the system that generates that air quality. It seems we understand least the most salient determinants of air quality. In the coming decades, Atlanta faces a formidable task in filling in those knowledge gaps and achieving additional improvements. If the past three decades are any guide, there is cause for guarded optimism. 19 References ARC (Atlanta Regional Commission). (2004). RDP Technical Report – 2004 Update. pp.3-22. http://www.atlantaregional.com/cps/rde/xbcr/arc/rdp_Ch3_Natural_Resources.pdf BEA (Bureau of Economic Analysis). (2006). Regional Economic Information System (REIS) 1969–2004. Available online at: http://bea.gov/regional/reis/ Bowler, C., & Brimblecombe, P. (1992). Archives and air pollution history. Journal of the Society of Archivists, 13(2), 136. Brimblecombe, P. (2005). The globalization of local air pollution. Globalizations, 2(3), 429-441. Brimblecombe, P. (1987). The Big Smoke: A History of Air Pollution in London since Medieval Times. Methuen: London. Bridgeport Regional Planning. (2007). Regional Air Quality Emissions Assessment. Retrieved from http://www.gbrpa.org/TransportationPlan/AQ_Conformity%20SummaryMarch%202007.DOC. Brimblecombe, P. (1976). Attitudes and responses towards air pollution in Medieval England. Journal of Air Pollution Control Association. 26, (10):941-5. Brookings Institution. (2006). Moving Beyond Sprawl: The Challenge for Metropolitan Atlanta. http://www.brookings.edu/es/urban/atlanta/lessons.htm. Census, B. o. t. No. HS-28. National Air Pollutant Emissions: 1900 to 2000. http://www.census.gov/statab/hist/HS-28.pdf. Collins, J., Hidy, G. M., Ganesan, K., Hansen, D. A., & Henry, R. C. (1984). Trends in Historical Acid Precursor Emissions and Their Airborne and Precipitation Products. Journal of the Air Pollution Control Association, 34(4), 333. Colvile, R. N., Kaur, S., Britter, R., Robins, A., Bell, M. C., Shallcross, D., et al. (2004). Sustainable development of urban transport systems and human exposure to air pollution. Science of the Total Environment, 334-335, 481-487. CVIOG (Carl Vinson Institute of Government). 2002. “Report on the Public Opinion Research for the Atlanta Regional Commission’s ‘Platforms for Progress’ Project.” December, 2002. Georgia Department of Labor. (2005). Georgia Labor Market Explorer: Historical Data. http://explorer.dol.state.ga.us/mis/historical.htm. Georgia State Climatology Office. (2006). Historical Climate Data and Summaries. http://climate.engr.uga.edu/historic.html. 20 Georgia Department of Transportation. (2006). Office of Transportation Data: Georgia’s Roadway Mileage and Characteristics Reports (400 Series Reports). http://www.dot.state.ga.us/DOT/. Goklany, I. M. (1999). Clearing the air : the real story of the war on air pollution / Indur Goklany. Washington, D.C. :: Cato Institute,. Metro Atlanta Chamber of Commerce. (2005). A Look at Atlanta: An Executive Profile. http://www.metroatlantachamber.com/. Metro Atlanta Chamber of Commerce. (2005). Transportation Briefing. http://www.metroatlantachamber.com/. Municipality of Anchorage. (2008). Analysis of the Probability of Complying with the National Ambient Air Quality Standard for CO in Anchorage between 2007 and 2023. Air Quality Program, Department of Health and Human Services. PAG Southwest Air Quality Forum. (2008). Ozone NAAQS Review. PAG Southwest Air Quality Forum. Preston, H. L. H. L., 1944- Automobile age Atlanta : the making of a southern metropolis, 1900-1935 / Howard L. Preston. Athens :: University of Georgia Press,. Sabatier, P. (1988). An Advocacy Coalition Framework of Policy Change and the Role of Policy-Oriented Learning Therein. Policy Sciences. 21, 129-68. State of Georgia, Secretary of State. (2006). History of Atlanta. http://www.sos.state.ga.us/tours/html/atlanta_history.html. Stern, A. C. (1982). History of Air Pollution Legislation in the United States. Journal of the Air Pollution Control Association, 32(1), 44. Tagaris et al. (2007). Impacts of global climate change and emissions on regional ozone and fine particulate matter concentrations over the United States. Journal of Geophysics Research Atmospheres. Tagaris et al. (2006). Uncertainties of Global Climate Change Impacts on Regional Air Quality over US. American Geophysical Union, Fall Meeting. U.S. Department of Congress. (2006). Bureau of Economic Analysis: Regional Economic Accounts. http://bea.gov/bea/regional/reis/ U.S. Department of Energy. (2006). Energy Information Administration: Official Energy Statistics from the U.S. Government. http://www.eia.doc.gov/. 21 U.S. Census Bureau. (2006). U.S Census Bureau Fact Finder. http://factfinder.census.gov/. U.S. Census Bureau. (1998). Population of the 100 Largest Cities and Other Urban Places in the United States: 1790 to 1990. http://www.census.gov/population/www/documentation/twps0027.html. U.S. EPA. (2007). Air Quality System. U.S. Environmental Protection Agency. http://www.epa.gov/ttn/airs/airsaqs/ U.S. EPA. (2006a). National Ambient Air Quality Standards. U.S. Environmental Protection Agency. http://www.epa.gov/air/criteria.html. U.S. EPA. (2006b). What are the Six Common Air Pollutant? U.S. Environmental Protection Agency. http://www.epa.gov/air/urbanair/6poll.html U.S. EPA. (2006c). VMT Growth Factors (%) by County. U.S. Environmental Protection Agency. http://www.epa.gov/ttn/naaqs/ozone/areas/sitemap/vmtgagf.htm. U.S. EPA. (2006d). The Inside Story: A Guide to Indoor Air Quality. U.S. Environmental Protection Agency. http://www.epa.gov/iaq/pubs/insidest.html U.S. EPA. (2006e). Air Emissions Trends - Continued Progress Through 2005. Retrieved. from http://www.epa.gov/airtrends/econ-emissions.html. U.S. EPA. (2006f). September 2006 Revisions to the National Ambient Air Quality Standards for Particulate Matter. Retrieved from http://www.epa.gov/oar/tribal/attachmts/Overview%20of%20PM%20NAAQS%20Final %20Rule%20&%20Maps%2009%2028%2006.ppt. U.S. EPA. (2005). Approval and Promulgation of Implementation Plans Georgia: Vehicle Miles Traveled State Implementation Plan for the Atlanta 1-Hour Ozone Nonattainment Area. http://www.epa.gov/fedrgstr/EPA-AIR/2005/April/Day-12/a7333.htm. U.S. EPA. (2005). Air Emissions Trends - Continued Progress Through 2004. Retrieved. from http://www.epa.gov/air/airtrends/2005/econ-emissions.html. U.S. EPA. (2004). The Particle Pollution Report: Current Understanding of Air Quality and Emissions through 2003. http://www.epa.gov/air/airtrends/aqtrnd04/pmreport03/pmlooktrends_2405.pdf#page=1. U.S. EPA. (2003). National Air Quality and Emissions Trends Report, 2003 Special Studies Edition. U.S. EPA. (1994). 1994 Air Quality Trends Summary Report. 22 U.S. EPA. (1971). EPA Sets National Air Quality Standards. EPA Press Release. http://www.epa.gov/hisotry. Utah Division of Air Quality. (2008). Baseline 2020: Air Quality. Retrieved from. http://governor.utah.gov/dea/qget/Baseline/6.htm. 23 Figure 1: Annual Lead Concentrations, Atlanta, GA, 1958-2005 Based on Annual Maximum Average8 Lead Concentration, ug/m3 3 2.5 2 1.5 1 0.5 19 71 19 73 19 75 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 0 Year Concentration, ug/m3 "Old" NAAQS Standard "New" NAAQS Standard Figure 2: Annual TSP, PM 2.5, and PM 10 Concentrations, Atlanta, GA, 1958-2005 Based on Annual 4th Maximum Average 250 200 150 PM TSP PM 2.5 100 PM 10 50 19 57 19 61 19 65 19 69 19 73 19 77 19 81 19 85 19 89 19 93 19 97 20 01 20 05 0 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 19 87 19 88 19 89 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 Ozone 8hr Concentration, ppm Figure 3: Annual Ozone Concentrations, Atlanta, GA, 1987-2005 Based on Annual 8 Hour 4th Maximum Average Year Annual Ozone Concentrations "Old" NAAQS Standard "New NAAQS Standard 8 The lead values between the years 1959-1962 were unavailable. The average of 1958 and 1963 were used for these years. 24 Figure 4: Annual NO2 and SO2 Concentrations, Atlanta, GA, 1972-2005 Based on Annual Averages 0.1200 0.1000 0.0800 SO2 0.0600 NO2 0.0400 0.0200 19 72 19 75 19 78 19 81 19 84 19 87 19 90 19 93 19 96 19 99 20 02 20 05 0.0000 20 15 10 5 0 19 7 19 2 7 19 3 7 19 4 7 19 5 7 19 6 7 19 7 7 19 8 7 19 9 8 19 0 8 19 1 82 19 8 19 3 8 19 4 8 19 5 8 19 6 8 19 7 8 19 8 8 19 9 9 19 0 9 19 1 9 19 2 9 19 3 94 19 9 19 5 9 19 6 9 19 7 9 19 8 9 20 9 0 20 0 0 20 1 0 20 2 0 20 3 0 20 4 05 CO Concentration, ppm Figure 5: Annual CO Concentrations, Atlanta, GA, 1972-2005 Based on 8-Hour Second Maximum Annual Averages Year Annual CO Concentrations, Atlanta NAAQS CO Atlanta population 4 3 2 1 0 0 MSA population, in millions 5 100000 200000 300000 400000 500000 Figure 6: Population Trends, Atlanta, GA, 1860-2005 1850 1900 1950 2000 year MSA population, in millions 25 Atlanta population .25 .2 .1 .15 Share of earnings 220000 200000 180000 160000 1970 1980 1990 Year 2000 2010 Employment in manufacturing Share of earnings from manufacturing 40000000 35000000 30000000 25000000 20000000 15000000 10000000 5000000 0 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04 VMT Figure 11: VMT Trends, Atlanta, GA, 1984-2004 Year Figure 12: Climate Trends, Atlanta, GA, 1940-2003 66 65 Average Temperature Employment 240000 Figure 10: Manufacturing Employment and Sector Size, Metro Atlanta, 1969-2005 64 63 62 61 60 59 58 57 56 1940 1950 1960 1970 Year 26 1980 1990 2000 Figure 13: National Trends in Emissions and Other Factors 400000 0 0 200000 ATL city population 5 10 ATL MSA population 15 20 600000 Figure 11A: Historical and Forecasted Populations 1850 1900 1950 year 2000 2050 ATL MSA population (millions) forecasted metro pop (millions) predicted censusATL forecasted city pop ATL city population 80000 0 20000 40000 60000 per capita income (1982-83$) 6.00e+07 2.00e+07 4.00e+07 3.00e+07 5.00e+07 1.00e+07 Vehicle miles traveled Figure 11B: Historical and Forecasted Income and VMT 1960 1980 2000 2020 2040 2060 year fitted VMT line VAR VMT forecast real income/capita VMT fitted income/capita VAR income/capita forecast Figure 11C: Historical and Forecasted Earnings 27 0 1980 2000 2020 2040 2060 year real manufacturing earnings real transportation earnings real construction earnings VAR forecast of manuf. earnings VAR forecast of transport. earnings VAR forecast of construct. earnings -.5 0 .5 1 Figure 12A: Historical and Forecast Ozone Levels, from Multivariate Regression 1960 1980 2000 2020 2040 2060 Year 95% confidence interval for ozone forecast VAR ozone forecast Prediction NAAQS Ozone (8hr) 5 10 Figure 12B: Historical and Forecast Lead Levels, from Multivariate Regression 0 earnings (1982-83$) 2.00e+07 4.00e+07 6.00e+07 8.00e+07 1960 1960 1980 2000 Year 95% confidence interval for lead forecast VAR lead forecast 2020 2040 Prediction NAAQS Lead Figure 12C: Historical and Forecast SO2 Levels, from Multivariate Regression 28 .6 .4 .2 0 -.2 1960 1980 2000 2020 2040 2060 Year 95% confidence interval for SO2 forecast VAR SO2 forecast Prediction NAAQS SO2 0 .1 .2 .3 .4 Figure 12D: Historical and Forecast NOx Levels, from Multivariate Regression 1980 2000 2020 Year 2040 95% confidence interval for NOx forecast VAR NOx forecast 2060 Prediction NAAQS NO2 -60 -40 -20 0 20 Figure 12E: Historical and Forecast CO Levels, from Multivariate Regression 1960 1980 2000 Year 95% confidence interval for CO forecast VAR CO forecast 29 2020 2040 Prediction NAAQS CO