Some ordinances also provide procedures for handling formal complaints by neighbors. Most California counties and a number of cities now have right-to-farm ordinances, a popularity seemingly driven by the belief on the part of local officials and others that this is an easy way to provide farmland protection that avoids hard political choices. Because they are not regulatory tools and rely primarily on the dissemination of information, however, the ordinances lack teeth and legal effect. It is uncertain to what extent they have reduced conflicts in edge areas. But the ordinances do serve a useful purpose, according to many agricultural leaders and county officials, in educating residents and asserting as a policy matter the value of agriculture in particular communities . More generally, conflicts between farmers and urban neighbors over farm activities can be addressed by a variety of techniques for dealing with community-level disputes. Practitioners in this field make a distinction between conflict resolution and conflict prevention. Resolution processes often involve a form of third party mediation, in which facilitators get both sides together, factual information on the source and elements of the dispute is developed, alternatives are deliberated,planting in pots ideas and an effort is made to reach an agreement among the parties as to actions to be taken such as changes in farm management . The state of New York has formalized such processes, with a Community Dispute Resolution Center in each county with resources for dealing with edge and other local conflicts .
Preventing edge conflicts typically involves less formal methods, with the emphasis on encouraging farm operators to maintain open lines of communication with their urban neighbors. The assumption is that friendly relations can head off serious disputes in the future over specific matters. One piece of advice to farmers in a New York state guidebook on reducing edge conflicts is to notify neighbors in advance of the timing and need for particular practices that may generate negative impacts. The guidebook goes further to suggest 15 strategies that farmers can use to foster good neighbor relations, including farm tours, providing gifts of farm produce, and setting aside an acre or two for wildlife .Given the substantial returns to higher education in this setting , this is a very high stakes exam. Every year, approximately 9 million students in China take the exam to compete for admission to approximately 2,300 colleges and universities. The NCEE has two primary tracks: the arts track and the science track.All students are tested on three compulsory subjects regardless of track: Chinese, mathematics, and English, with each worth 150 points. Students in the arts track take an additional combined test that includes history, politics, and geography worth 300 points, while students in the science track take an additional combined test that includes physics, chemistry, and biology worth 300 points. Thus, regardless of track, the maximum achievable score for each student is 750 points. In our focal provinces, the Chinese and math exams are scheduled for 9– 11:30am and 3–5pm on June 7th, and the English and track test are scheduled for 9– 11:30am and 3–5pm on June 8th.Since provinces have some discretion in the design of their tests, exam difficulty can vary by track, province, and year. Our core analysis deploys province-by-year-by-track fixed effects to account for this possibility. The NCEE tests are graded one to two weeks after the exams are completed by professionals in hotels in each of the respective provincial capitals. Since this grading occurs in locations that differ from test takers in terms of both space and time, we are confident that the effect we estimate on NCEE scores is not the result of any potential impacts on graders. The NCEE data were obtained from the China Institute for Educational Finance Research at Peking University. This dataset provides a unique identifier and the total test score for the universe of students enrolled in a Chinese institution of higher education during our study period.
The dataset also reports the subject specialization for each student, allowing us to explore heterogeneity across the science and art tracks.Importantly, the student ID contains a six-digit code for county of residence, which allows us to match students to the county administrative centers. Testing facilities are located in local schools which are universally very close to county administrative center. 7 Therefore, we use the county administrative center to approximate the testing facilities. The information on which testing facility a student is assigned is unavailable. Our core analytic sample includes observations from approximately 1.3 million students. We supplement this dataset with data on the cutoff scores that determine admission eligibility to the elite universities in order to separately examine the impacts at the upper-end of the performance distribution.Data on daily agricultural fires are collected from two satellites named TERRA and AQUA, which rely upon Moderate Resolution Imaging Spectroradiometer sensors to infer ground-level fire activity. The satellites overpass China four times a day , and report all fire points detected with 1-km resolution . The fires are detected based on thermal anomalies, surface reflectance, and land use . Since the size of a fire cannot reliably be inferred from satellite data , we treat fires in adjacent pixels as distinct fires. We exploit data on fire radiative power, a measure of fire intensity, to at least partially probe the importance of this assumption. A fire is linked to NCEE performance within a county if it occurs within a 50- km of the county administrative center during the two-day exam period in each year. Alternative distances are explored as part of our robustness analyses. Since proximity to a fire is likely correlated with the economic benefits as well as the environmental harms from fires, we eschew distance-weighting strategies on fires in our core analysis. These are, nonetheless, explored in our robustness checks. Meteorological data is important for two reasons. First, as detailed in the next section, we exploit detailed data on wind direction to contrast impacts of those upwind and downwind of a given fire. Second, weather may also confound the interpretation of our results since the incidence of agricultural fires may be correlated with meteorological conditions. Our weather data are obtained from the National Oceanic and Atmospheric Administration of the United States.
We collect daily average weather data on temperature, precipitation, dew point, wind speed, wind direction and atmospheric pressure from 44 local weather stations during our sample period. Daily average wind direction is reported based on the hourly wind direction and wind speed through vector decomposition .8 Given the sensitivity of wind direction to topography and other quite localized factors, we assign wind to test locations based on monitor data from the source closest to the county administrative center, and drop counties with no wind stations within 50 km.9 We extract other weather data during the exam time and then convert from station to county using the inverse-distance weighting method . The basic algorithm calculates weather for a given site based on a weighted average of all station observations within a 50-km radius of the county center, where the weights are the inverse distance between the weather station and the county administrative center. While the detrimental impacts of agricultural fires on air quality have been documented in the environmental science literature,growing blueberries in pots data availability does not allow us to make this link explicitly in our setting. Ground monitoring pollution data at the station-day level in China is not available prior to 2011, and there are infamous stories of data manipulation of the Air Pollution Index and PM10 in China apply to the period prior to 2013 .10 In addition, satellite data is not well suited for ground-level measurement at fine temporal and spatial scales required for our analyses, especially during burning seasons with smoke plumes . Nonetheless, we provide a first-stage estimation, of sorts, by estimating the relationship between air pollution and agricultural fires using data from a more recent period: 2013–2016. Since NCEE data is not available for this period, we view this analysis as one designed to shed light on the mechanisms through which agricultural fires might impact cognitive performance. Daily pollution data are obtained from the China National Environmental Monitoring Center , which is affiliated with the Ministry of Environmental Protection of China. Monitoring stations report data for the six major air pollutants – particulate matter less than 10 microns in diameter , particulate matter less than 2.5 microns in diameter , sulfur dioxide, nitrogen dioxide, ozone, and carbon monoxide – that are used to construct the daily Air Quality Index in China. For each pollutant, we construct a two-day average concentration level, corresponding to the length of the exam period. Fires that took place more than 50 km from a county center are excluded from this analysis. We select all pollution monitoring stations within 50 km from a county administrative center and calculate the pollution level at the center using the IDW method. Our analysis relies on data from 212 distinct pollution monitors, with an average distance of 24.5 km. In this section, we explore the heterogeneity of our core results along two dimensions, as shown in Table 3. The first column simply reproduces the results from our preferred specification for our primary results .
Columns and of Table 3 explore heterogeneity along another dimension: the subject track. It appears that the impacts are negative and highly statistically significant for those in the science track while only marginally significant for those in the arts track. This may reflect the differential sensitivity of the prefrontal cortex – the part of the brain responsible for more mathematical style reasoning, and is consistent with other evidence on the impacts of environmental stressors on cognitive performance . This pattern of results might also, at least partly, be driven by the gender composition of students across tracks. While we do not have individual level gender data, the male ratio is typically much higher in science track than arts track and other work has found the cognitive performance of males to be more sensitive to PM pollution than females . The next four columns of Table 3 examine how the impacts of agricultural fires vary across the student ability distribution by estimating Equation using a quantile regression approach. This regression is especially important for two reasons. First, since we only observe NCEE scores for students that were eventually admitted to an institution of higher learning, we might be worried about sample selection resulting from negative effects at the lower end of the ability distribution. Second, differences in impacts across the ability distribution could have profound long-run impacts on income inequality given the highly nonlinear returns to scores. Our results find no impacts among low ability students, thus minimizing concerns about selection bias. Moreover, the impacts appear to be concentrated near the very top of the performance distribution – above the 75th percentile. This can be seen most clearly in Figure 5, which further breaks down estimates by decile. Column offers another perspective on the higher end of the ability distribution by focusing on the impacts of agricultural fires on the likelihood of admission into an elite university in China based on the cutoff scores that govern that process. The cutoff score in each province is the lowest score of students admitted to the first-tier universities in China. It is determined by the admission quota of each university and the ranking of student scores in each province. Upwind fires continue to have a significant negative impact on test performance. A one percentage point increase in the difference between upwind and downwind fires, decreases the probability of admission to an elite university by 0.027 percent . Given the sizable impacts of an elite education in China on lifetime earnings , these impacts should be viewed as economically meaningful, even if they may be largely re-distributional by privileging the admission of students from less exposed counties over those from more exposed ones. In this section, we provide a number of robustness checks. We begin by exploring alternative ways to assign the exposure of test takers to agricultural fires. The first column of Table 4 reproduces our main results, which limit our focus to fires within 50 km of a testing center.