Final Report Summary - ACC_EXT (The Accident Externality from Driving: Evidence from Shabbat Exit and Entry)
In this project we document and measure an externality of driving, whereby a driver's decision to take to the road affects fellow drivers' risk of accident. Because religious Jews refrain from driving during the Sabbath, traffic on Israeli roads decreases sharply when the Sabbath begins each Friday, at a precisely defined time before sundown, and increases after the Sabbath ends on Saturday evening, at a precisely defined time after sunset. This pattern is particularly pronounced in more religious areas. Using plausibly exogenous variation in traffic flows associated with the Sabbath, we estimate the effect of traffic flow on the risk of fatal or injurious accident. The project addresses the issue of the identification of the causal relationship between traffic volume and accidents, an issue that did not receive much attention in the existing literature. In this project we propose a clear and highly innovative identification strategy. Additionally, the richness of the data allows us to test the validity or our identifying assumptions. The project provides direct evidence on the matter by addressing the relationship between traffic density and injurious and fatalities in automobile accidents directly, rather than using insurance costs.
In our final analysis, at Sabbath entry, a 10% increase in traffic volume is associated with a 7% increase in the probability of severe accidents, indicating that there no positive accident externality upon Sabbath entry. At the Sabbath's exit, on the other hand, the corresponding effect of a 10% increase in traffic volume is a dramatic increase in the probability of severe accidents of 20%. Thus, the positive accident externality emerges entirely from the Sabbath's exit. This result could indicate that traffic volume affects the risk of accident non-linearly, generating an accident externality only when traffic volume is heavy.
As the travel choices of non-Jewish drivers are not bound by the precise time of Sabbath entry and exit, changes in the probability of severe traffic accidents involving non-Jewish drivers, provide indication of the presence of an accident externality. We, therefore, use evidence regarding non-Jewish drivers to corroborate our interpretation of the previous results as evidence of an externality, and to corroborate our estimate of its magnitude. Consistent with our earlier results, we find that this subset of accidents, too, is similarly affected by variation in traffic flow upon Sabbath entry and exit, showing an increase of 13% in the probability of being involved in a severe accident at Sabbath exit, but with no change at Sabbath entry. The estimation results correspond very well with the earlier results, thereby providing support to our interpretation of them.
When we distinguish between more and less perilous roads according to a commonly used external measure, we find that the effect of traffic volume on the risk of accident is more than 3 times greater on more perilous roads. This result is consistent with a theory of limited driver attention or skill, whereby the challenges posed by dangerous roads and heavier traffic volume impact the risk of accident additively. This finding is consistent with recent experimental evidence on the issue
With respect to advancement beyond the state of the art in the field: Our analysis provides some new and interesting insights about the relation between traffic volume and traffic accidents. With respect to the empirical approach, in addition to providing plausibly exogenous identifying variation from a known source, the natural experiment we use allows to directly estimate the magnitude of the accident externality by examining a well-defined subset of drivers that is likely to be affected by traffic changes only through per-vehicle risk. To our knowledge, this study is the first to do so. Second, this study sheds new light on the micro-foundations of this relationship. It shows that the positive relationship between traffic and accidents tends to be concentrated on roads that are more perilous, suggesting that the underlying mechanism involves an important interaction with road characteristics. Our results suggest that on roads that are not perilous and at times at which traffic is relatively light, drivers are able to offset the increased risk generated by additional vehicles.
Project objectives since the beginning of the project
Project objectives for the reporting period:
• Hiring & training assistants: Hadas Fuchs and Elisheva Schwarz, MA students were hired trained and has worked on the project – participating in all the stages of the data analysis and research.
• Carrying out of a small-scale pilot study; Data entry; Data checks; Data analysis: The pilot analysis was completed successfully and the data was prepared for the full scale analysis.
• Present early findings at workshops, and professional seminars: The project was presented at several invited seminars including the applied economics workshops at Tel Aviv University, Hebrew University, Bar Ilan University and Université Libre de Bruxelles. The presentations were very useful, providing insightful feedback.
• Modification of research design based on results of pilot study and feedback from presentations: The project received some useful feedback and changes were made accordingly.
• Carrying out of a full-scale study; Data entry; Data checks; Data analysis: The full scale data analysis was done successfully.
• Write up of results; Preparation of manuscripts for publication: Taking into account all the useful feedback, the results were reported in a manuscript. The manuscript was published in the Journal of Urban Economics. Probably the best field journal in the area of research (Romem I, Shurtz I. The accident externality of driving: Evidence from observance of the Jewish Sabbath in Israel. Journal of Urban Economics. 2016 Nov 30;96:36-54.).
Work progress and achievements since the beginning of the project
Research progress made: Hadas Fuchs, an economics MA student was hired trained and has worked on the project – participating in all the stages of the data analysis and research. Later she was replaced by Elisheva Schwarz, a statistics MA student. With the help of Hadas and Elisheva, we prepared the data for the project. We compile administrative police report data from Israel spanning the universe of all accidents in which at least one person was injured in 2000-2010 to create a panel of accidents. These data contain information on:
• Accident characteristics: exact time of occurrence in fifteen minutes intervals, exact location, type of road, road conditions, weather condition.
• Persons involved characteristics (including driver): age, gender, years of driving experience (driver), ethnic group, severity of injuries, seatbelt (on or off), type of locality of residence.
• Vehicles characteristics: year of manufacturing, size, ownership, type of car (e.g. stolen)
In order to examine the relationship between traffic and accident we combine these data with data on traffic. We use a representative sample of non-urban roads in Israel in 2000-2010. These data are created by the Israeli central bureau of statistics by sampling a set of about six hundred non-urban road segments for (at least) an entire week (168 hours), using specialized pneumatic counting devices. Traffic in each segment is counted every year in a different time of the year in order to create coverage of all four seasons for each road segment in a four-year sampling cycle (which are counted form 1982). We match these data with data on accidents in non-urban roads in the same time period to analyze the relation in a country wide perspective.
The project was presented in various seminars including the applied economics workshops at Tel Aviv University, Hebrew University, Bar Ilan University and Université Libre de Bruxelles. The presentations were extremely useful and we made changes in the analysis accordingly. One interesting idea that came up in the seminars is to exploit information about non-Jewish drivers to corroborate the results further. This approach turned out to be very fruitful.
In our final analysis, we find that at Sabbath entry, a 10% increase in traffic volume is associated with an 7% increase in the probability of severe accidents, indicating that there no positive accident externality upon Sabbath entry. At the Sabbath's exit, on the other hand, the corresponding effect of a 10% increase in traffic volume is a dramatic increase in the probability of severe accidents of 20%. Thus, the positive accident externality emerges entirely from the Sabbath's exit. This result could indicate that traffic volume affects the risk of accident non-linearly, generating an accident externality only when traffic volume is heavy.
As the travel choices of non-Jewish drivers are not bound by the precise time of Sabbath entry and exit, changes in the probability of severe traffic accidents involving non-Jewish drivers, provide indication of the presence of an accident externality. We, therefore, use evidence regarding non-Jewish drivers to corroborate our interpretation of the previous results as evidence of an externality, and to corroborate our estimate of its magnitude. Consistent with our earlier results, we find that this subset of accidents, too, is similarly affected by variation in traffic flow upon Sabbath entry and exit, showing an increase of 13% in the probability of being involved in a severe accident at Sabbath exit, but with no change at Sabbath entry. The estimation results correspond very well with the earlier results, thereby providing support to our interpretation of them.
When we distinguish between more and less perilous roads according to a commonly used external measure, we find that the effect of traffic volume on the risk of accident is more than 3 times greater on more perilous roads. This result is consistent with a theory of limited driver attention or skill, whereby the challenges posed by dangerous roads and heavier traffic volume impact the risk of accident additively. This finding is consistent with recent experimental evidence on the issue
Advancement beyond the state of the art in the field: Our analysis thus far provides some new and interesting insights about the relation between traffic volume and traffic accidents. With respect to the empirical approach, in addition to providing plausibly exogenous identifying variation from a known source, the natural experiment we use allows to directly estimate the magnitude of the accident externality by examining a well-defined subset of drivers that is likely to be affected by traffic changes only through per-vehicle risk. To our knowledge, this study is the first to do so. Second, this study sheds new light on the micro-foundations of this relationship. It shows that the positive relationship between traffic and accidents tends to be concentrated on roads that are more perilous, suggesting that the underlying mechanism involves an important interaction with road characteristics. Our results suggest that on roads that are not perilous and at times at which traffic is relatively light, drivers are able to offset the increased risk generated by additional vehicles.
Impact and transfer of knowledge: Impact and transfer of knowledge: The main effect is through the publication of the study (Romem I, Shurtz I. The accident externality of driving: Evidence from observance of the Jewish Sabbath in Israel. Journal of Urban Economics. 2016 Nov 30;96:36-54.). The study is also taught in the PI’s courses at the Hebrew University. It will be reported in a major newspaper in Israel in the coming months. The published results will also be embedded in the public economics course which I will offer to undergraduate and graduate students at the Hebrew University.
Dissemination
The dissemination activity of the study includes a strong scientific publication that we will hope lead to numerous citations (Romem I, Shurtz I. The accident externality of driving: Evidence from observance of the Jewish Sabbath in Israel. Journal of Urban Economics. 2016 Nov 30;96:36-54.). Presentations at invited talks at seminars at Tel Aviv University, Hebrew University, Bar Ilan University and Université Libre de Bruxelles. The results will also be embedded in the public economics course which I will offer to undergraduate and graduate students at the Hebrew University. The project has a webpage too (link via my personal website: https://sites.google.com/site/ityshurtz/). We are currently working on a newspaper piece and it will be reported in a major newspaper in Israel in the coming months.
In our final analysis, at Sabbath entry, a 10% increase in traffic volume is associated with a 7% increase in the probability of severe accidents, indicating that there no positive accident externality upon Sabbath entry. At the Sabbath's exit, on the other hand, the corresponding effect of a 10% increase in traffic volume is a dramatic increase in the probability of severe accidents of 20%. Thus, the positive accident externality emerges entirely from the Sabbath's exit. This result could indicate that traffic volume affects the risk of accident non-linearly, generating an accident externality only when traffic volume is heavy.
As the travel choices of non-Jewish drivers are not bound by the precise time of Sabbath entry and exit, changes in the probability of severe traffic accidents involving non-Jewish drivers, provide indication of the presence of an accident externality. We, therefore, use evidence regarding non-Jewish drivers to corroborate our interpretation of the previous results as evidence of an externality, and to corroborate our estimate of its magnitude. Consistent with our earlier results, we find that this subset of accidents, too, is similarly affected by variation in traffic flow upon Sabbath entry and exit, showing an increase of 13% in the probability of being involved in a severe accident at Sabbath exit, but with no change at Sabbath entry. The estimation results correspond very well with the earlier results, thereby providing support to our interpretation of them.
When we distinguish between more and less perilous roads according to a commonly used external measure, we find that the effect of traffic volume on the risk of accident is more than 3 times greater on more perilous roads. This result is consistent with a theory of limited driver attention or skill, whereby the challenges posed by dangerous roads and heavier traffic volume impact the risk of accident additively. This finding is consistent with recent experimental evidence on the issue
With respect to advancement beyond the state of the art in the field: Our analysis provides some new and interesting insights about the relation between traffic volume and traffic accidents. With respect to the empirical approach, in addition to providing plausibly exogenous identifying variation from a known source, the natural experiment we use allows to directly estimate the magnitude of the accident externality by examining a well-defined subset of drivers that is likely to be affected by traffic changes only through per-vehicle risk. To our knowledge, this study is the first to do so. Second, this study sheds new light on the micro-foundations of this relationship. It shows that the positive relationship between traffic and accidents tends to be concentrated on roads that are more perilous, suggesting that the underlying mechanism involves an important interaction with road characteristics. Our results suggest that on roads that are not perilous and at times at which traffic is relatively light, drivers are able to offset the increased risk generated by additional vehicles.
Project objectives since the beginning of the project
Project objectives for the reporting period:
• Hiring & training assistants: Hadas Fuchs and Elisheva Schwarz, MA students were hired trained and has worked on the project – participating in all the stages of the data analysis and research.
• Carrying out of a small-scale pilot study; Data entry; Data checks; Data analysis: The pilot analysis was completed successfully and the data was prepared for the full scale analysis.
• Present early findings at workshops, and professional seminars: The project was presented at several invited seminars including the applied economics workshops at Tel Aviv University, Hebrew University, Bar Ilan University and Université Libre de Bruxelles. The presentations were very useful, providing insightful feedback.
• Modification of research design based on results of pilot study and feedback from presentations: The project received some useful feedback and changes were made accordingly.
• Carrying out of a full-scale study; Data entry; Data checks; Data analysis: The full scale data analysis was done successfully.
• Write up of results; Preparation of manuscripts for publication: Taking into account all the useful feedback, the results were reported in a manuscript. The manuscript was published in the Journal of Urban Economics. Probably the best field journal in the area of research (Romem I, Shurtz I. The accident externality of driving: Evidence from observance of the Jewish Sabbath in Israel. Journal of Urban Economics. 2016 Nov 30;96:36-54.).
Work progress and achievements since the beginning of the project
Research progress made: Hadas Fuchs, an economics MA student was hired trained and has worked on the project – participating in all the stages of the data analysis and research. Later she was replaced by Elisheva Schwarz, a statistics MA student. With the help of Hadas and Elisheva, we prepared the data for the project. We compile administrative police report data from Israel spanning the universe of all accidents in which at least one person was injured in 2000-2010 to create a panel of accidents. These data contain information on:
• Accident characteristics: exact time of occurrence in fifteen minutes intervals, exact location, type of road, road conditions, weather condition.
• Persons involved characteristics (including driver): age, gender, years of driving experience (driver), ethnic group, severity of injuries, seatbelt (on or off), type of locality of residence.
• Vehicles characteristics: year of manufacturing, size, ownership, type of car (e.g. stolen)
In order to examine the relationship between traffic and accident we combine these data with data on traffic. We use a representative sample of non-urban roads in Israel in 2000-2010. These data are created by the Israeli central bureau of statistics by sampling a set of about six hundred non-urban road segments for (at least) an entire week (168 hours), using specialized pneumatic counting devices. Traffic in each segment is counted every year in a different time of the year in order to create coverage of all four seasons for each road segment in a four-year sampling cycle (which are counted form 1982). We match these data with data on accidents in non-urban roads in the same time period to analyze the relation in a country wide perspective.
The project was presented in various seminars including the applied economics workshops at Tel Aviv University, Hebrew University, Bar Ilan University and Université Libre de Bruxelles. The presentations were extremely useful and we made changes in the analysis accordingly. One interesting idea that came up in the seminars is to exploit information about non-Jewish drivers to corroborate the results further. This approach turned out to be very fruitful.
In our final analysis, we find that at Sabbath entry, a 10% increase in traffic volume is associated with an 7% increase in the probability of severe accidents, indicating that there no positive accident externality upon Sabbath entry. At the Sabbath's exit, on the other hand, the corresponding effect of a 10% increase in traffic volume is a dramatic increase in the probability of severe accidents of 20%. Thus, the positive accident externality emerges entirely from the Sabbath's exit. This result could indicate that traffic volume affects the risk of accident non-linearly, generating an accident externality only when traffic volume is heavy.
As the travel choices of non-Jewish drivers are not bound by the precise time of Sabbath entry and exit, changes in the probability of severe traffic accidents involving non-Jewish drivers, provide indication of the presence of an accident externality. We, therefore, use evidence regarding non-Jewish drivers to corroborate our interpretation of the previous results as evidence of an externality, and to corroborate our estimate of its magnitude. Consistent with our earlier results, we find that this subset of accidents, too, is similarly affected by variation in traffic flow upon Sabbath entry and exit, showing an increase of 13% in the probability of being involved in a severe accident at Sabbath exit, but with no change at Sabbath entry. The estimation results correspond very well with the earlier results, thereby providing support to our interpretation of them.
When we distinguish between more and less perilous roads according to a commonly used external measure, we find that the effect of traffic volume on the risk of accident is more than 3 times greater on more perilous roads. This result is consistent with a theory of limited driver attention or skill, whereby the challenges posed by dangerous roads and heavier traffic volume impact the risk of accident additively. This finding is consistent with recent experimental evidence on the issue
Advancement beyond the state of the art in the field: Our analysis thus far provides some new and interesting insights about the relation between traffic volume and traffic accidents. With respect to the empirical approach, in addition to providing plausibly exogenous identifying variation from a known source, the natural experiment we use allows to directly estimate the magnitude of the accident externality by examining a well-defined subset of drivers that is likely to be affected by traffic changes only through per-vehicle risk. To our knowledge, this study is the first to do so. Second, this study sheds new light on the micro-foundations of this relationship. It shows that the positive relationship between traffic and accidents tends to be concentrated on roads that are more perilous, suggesting that the underlying mechanism involves an important interaction with road characteristics. Our results suggest that on roads that are not perilous and at times at which traffic is relatively light, drivers are able to offset the increased risk generated by additional vehicles.
Impact and transfer of knowledge: Impact and transfer of knowledge: The main effect is through the publication of the study (Romem I, Shurtz I. The accident externality of driving: Evidence from observance of the Jewish Sabbath in Israel. Journal of Urban Economics. 2016 Nov 30;96:36-54.). The study is also taught in the PI’s courses at the Hebrew University. It will be reported in a major newspaper in Israel in the coming months. The published results will also be embedded in the public economics course which I will offer to undergraduate and graduate students at the Hebrew University.
Dissemination
The dissemination activity of the study includes a strong scientific publication that we will hope lead to numerous citations (Romem I, Shurtz I. The accident externality of driving: Evidence from observance of the Jewish Sabbath in Israel. Journal of Urban Economics. 2016 Nov 30;96:36-54.). Presentations at invited talks at seminars at Tel Aviv University, Hebrew University, Bar Ilan University and Université Libre de Bruxelles. The results will also be embedded in the public economics course which I will offer to undergraduate and graduate students at the Hebrew University. The project has a webpage too (link via my personal website: https://sites.google.com/site/ityshurtz/). We are currently working on a newspaper piece and it will be reported in a major newspaper in Israel in the coming months.