Climate Change will Increase Ozone Levels and unravel Air Quality Regulations

: Using daily data for the United States over the period 1980-2019, we estimate the impacts of temperature on ambient ozone concentrations, accounting for adaptation to climatic change. We find that even with adaptation, rises in temperature will steeply increase ozone levels by over 9 ppb on days above 25 ◦ C. By mid-century, we calculate that 189 additional counties will be violating the air quality standards, with 33 million more residents exposed to unhealthy 15 levels of ozone. Climate change will thus likely increase the costs of compliance with existing ambient ozone standards. In light of a recent EPA ruling that would effectively remove co-benefits of ozone precursor reductions from the cost-benefit analysis of those standards, they will be in peril, further threatening public health.


Main Text:
It is well established that ambient ozone causes detrimental health effects, leading to mortality and morbidity (1)(2). Estimates of the global burden of disease indicate that 9-23 million annual asthma emergency room visits globally in 2015 could be attributable to ozone, representing 8-20% of the annual number of visits (3). Ozone affects both the young and the old, even at levels 5 below the current U.S. Environmental Protection Agency's (EPA) National Ambient Air Quality Standards (NAAQS) (4). Yet, progress on reducing the levels of ozone relative to other criteria pollutants has been limited over the last forty years (5), and a deeply flawed EPA ruling at the end of the Trump Administration has threatened the grounds for the current level of regulatory oversight on ozone (6). The rule, which remains in effect, disregards indirect co-benefits of the 10 ozone NAAQS in the form of improved public health associated with the concomitant reduction of other air pollutants. Because ozone standards rely on indirect benefits from reductions in particulate matter concentrations to comfortably pass a cost-benefit analysis, they are in jeopardy for the next revision in a few years (7).
Climate change may further put the ozone standards at risk if rising temperatures increase ozone 15 concentrations, and ultimately the costs of compliance with these standards. There is a growing recognition that climate change will affect ozone levels. Indeed, there is a literature on the socalled climate penalty on ozone -that is, the impact of climatic changes on ozone -indicating that this penalty can be substantial (8,9). But important knowledge gaps remain. First, it is still not clear exactly how much climate change will increase ozone concentrations because there can 20 be adaptation to climate change, and the ozone NAAQS themselves may act as catalysts for such adaptation (10). If a county violates the standard because average temperature has risen -not because ozone precursor emissions of NOx and VOCs have risen -individuals and firms may still be forced to curb emissions of ozone precursors in order to comply with air quality regulations. These adjustments may preclude excessive increases in ozone levels even under climatic changes, but may come with associated increases in compliance costs. Second, ozone may respond differently to higher temperatures across regions. Places that have experienced extreme temperatures relatively often may have already made key adjustments. Therefore, 5 adaptation might manifest differently in regions that experience varying baseline climates.
In this study, we build on an econometric method we recently developed (11) to estimate the impact of long-run (climatic) temperature changes on ozone in the United States. Our approach has two main elements: The first is the decomposition of meteorological variables into long-run climate normals and short-run weather shocks. The second is the ability to identify responses to 10 weather shocks and climatic changes in the same estimating equation. Individuals and firms respond to information on climatic variation they have observed and processed over the years. In contrast, they may be constrained in their response to short-term, unanticipated weather shocks.
Our measure of adaptation is the difference between those two responses by the same economic agents. We combine daily ozone data from the EPA with daily weather data from the National 15 Oceanic and Atmospheric Administration (NOAA) for the period 1980-2019 (Figs. S2 -S5 & SI section S2). Splitting temperature into 5 • C bins (Figs. S8 -S7), we estimate the differential relationship between climate normal temperature and ozone across the temperature distribution nationally and for each of the nine NOAA climate regions (Tables S1, S2 & SI section S4). We then predict ozone levels by mid-century, relying on climate change projections (12). Ultimately, 20 we are able to project how many more counties would violate the NAAQS for ozone by midcentury, in the absence of further action to control ozone concentrations. Finally, we assess the consequences of climate change to the cost-benefit analysis of ozone standards, and suggest that 4 -in the absence of co-benefits -climate change will unravel ozone standards and, as a result, may further threaten public health. We highlight three main findings.

The relationship between temperature and ozone resembles a reclining beach chair, with
steep increases in ozone of over 9 ppb on days above 25 • C (Fig. 1). The relationship between temperature and ozone is relatively flat until the 20-25 • C temperature bin, but the slope becomes 5 steep at 25-30 • C and beyond. Indeed, even though adaptation reduces the effect of temperature on ozone by approximately 40% (Table S3), ozone concentrations still increase substantially by over 9 and 15 ppb when temperatures are 25-30 • C or above 30 • C, respectively, relative to 15-20 • C. Furthermore, under the business-as-usual greenhouse gas emissions scenario embedded in the Representative Concentration Pathway (RCP) 8.5, average U.S. temperatures will increase by 10 1.6 • C by mid-century (12), implying a significant right-ward shift in the distribution of temperature days with more than double the number of days above 30 • C during the ozone season by mid-century.
The relationship between temperature and ozone exhibits substantial heterogeneity among the nine U.S. climate regions (Fig. 2). The relationship between temperature and ozone is 15 steepest for regions which have higher levels of baseline ozone, such as the Northeast and Ohio Valley (Table S4). In contrast, regions with lower average ozone concentrations (Northwest, Rockies) or those that have consistently faced hot temperatures (South, Southwest) have a flatter relationship, or a significantly reclined beach chair. However, climate projections indicate these warmer regions will have substantial increases in the number of days above 30 • C by mid- 20 century. All regions are projected to see an increase in the number of days between 25-30 • C or above 30 • C by mid-century.
Under current ozone abatement levels, 189 additional counties will be violating ozone standards by mid-century, with 33 million more residents exposed to unhealthy levels of ozone (Fig 3A). Since (Fig. 3B), with a few exceptions associated with the more muted response by some climate regions as described above (Fig. 3C). Projected changes in ozone standard violations under the RCP 4.5 climate scenario are qualitatively similar (Fig. S10). 15 Our results imply that increases in ozone concentrations due to increases in average temperature would lead many additional counties to become out of attainment -unless they step up pollution abatement efforts to reduce ozone levels. Climate projections under business-as-usual emissions scenarios imply that ozone concentrations in July -typically the worst month of the year for ozone -would increase by 4 ppb by mid-century (Table S5). To put this climate-induced 20 increase in ozone levels into perspective, EPA's last change of the ozone standards in 2015 decreased the threshold from 75 to 70 ppb. Thus, 80% of that 5-ppb decrease would be undone by mid-century, on average. In essence, for counties to remain in attainment with the current 70 ppb standards, it would be as if they had to meet a standard of at least 66 ppb. Such reductions will be costly. By EPA's calculations in their 2015 regulatory impact analysis (13), reducing the ozone standards to 65 ppb would incur an additional $14.6 billion (2011 dollars) per year due to increased compliance costs in 64 counties. In order to keep the 189 additional counties projected to go out of attainment by mid-century below the 70 ppb standard, annual compliance costs could increase by up to $43.6 billion based on current technology (Table S6 & SI section S5). 5 A similar back-of-the-envelope calculation of the benefits from achieving compliance with a 70 ppb standard for these counties using the same EPA analysis indicates an average annual benefit of $59.1 billion (Table S7). Comparing these costs and benefits would suggest that the ozone standards be retained. However, about two-thirds of the total benefits under the current ozone standards are derived from co-benefits due to concomitant reductions in particular matter. The 10 recent 2020 EPA ruling based on deeply flawed analysis (6) would exclude such co-benefits from future benefit-cost analyses of ozone and other pollution standards -despite analysis showing that as long as co-benefits are carefully estimated, they should indeed count as much as direct benefits (14).
Excluding co-benefits, the back-of-the envelope benefits would fall below the compliance costs 15 at just under $20 billion. Thus, the combination of rising compliance costs due to climate change, and the artificial reduction in calculated benefits due to this ruling, threatens to unravel ozone regulations and compromise public health in the United States.

Methods
A body of literature in environmental economics and related fields has attempted to quantify the economic impacts of climate change (15). Earlier methods to estimate the impact of climate change on economic outcomes have either used cross-sectional approaches that rely on permanent meteorological conditions (16, 17) or panel fixed-effects approaches that exploit 5 unanticipated weather shocks (18, 19). We developed a unifying approach that has two key elements: First, a decomposition of meteorological variables. Second, the ability to recover estimates of both the short-and long-run impacts of climate change in the same equation. Taking the results of this model, we then apply climate projections from the RCP 4.5 and 8.5 scenarios to predict Ozone levels and county violations of the standards by mid-century (SI section 3). 10 Decomposition of meteorological variables: norms vs. shocks -daily temperature is decomposed into its long-run component, and its short-run deviation from this value: where Temp C represents climate normal temperature, and Temp W (≡ Temp−Temp C ) deviations from the norm. We focus on temperature around the location of ozone monitor i in day t of month m 15 and year y, and construct 5°C temperature bins before decomposing each bin into Temp C and Temp W by defining Temp C as the 30-year monthly moving average (MA) of past temperatures.
We then average each bin by month -in essence, these monthly averages now reflect the "share" of the month in which the daily temperature fell within the respective bin. Next, we create the 30-year monthly MA -"climate norms" -by taking the 30-year average of these monitor-level 20 monthly averages for each bin. 8 Econometric Model -The econometric specification is:  For these reasons, our results should be seen as a lower bound of the impacts of climate change 20 on ozone. Therefore, retaining the ozone standards in this changing climate is even more crucial.  Table S1. of currently monitored counties will be in violation of the standards by mid-century -higher than