Vous êtes sur la page 1sur 30

RIDE-SHARING, FATAL CRASHES, AND CRIME *

Angela K. Dills Western Carolina University

Sean Mulholland Western Carolina University

Abstract:

Spring 2017

The advent of smart-phone based, ride-sharing applications has revolutionized the vehicle for hire market. Advocates point to the ease of use and lower wait times compared to hailing a taxi or pre-arranging limousine service. Others argue that proper government oversight is necessary to protect ride-share passengers from driver error or vehicle part failure and violence from unlicensed strangers. Using U.S. county-level data from 2007 through 2014, we investigate whether the introduction of the ride-sharing service, Uber, is associated with changes in fatal vehicle crashes and crime. We find that Uber’s entry lowers the rate of DUIs and fatal accidents. For most specifications, we also find declines in arrests for assault and disorderly conduct. Conversely, we observe an increase in vehicle thefts.

Keywords: Ride-Sharing; Uber; Crashes; Crime; Traffic fatalities

JEL Codes: R41, K42, D45

* We would like to thank Reynaldo Hernandez-Julian, Christopher Schilling, Luke Sherry, E. Frank Stephenson, and session participants at the 2016 Public Choice Society Meetings and the 2016 Association for Private Enterprise Meetings for their valuable comments. Contact information: Sean Mulholland:

Professor of Economics, Department of Economics, Management, and Product Management, College of Business, Forsyth 224D, Western Carolina University, Cullowhee, NC 28723, 828.227.3169, seanemulholland@gmail.com. All errors are most assuredly ours alone.

1

I.

Introduction

The advent of smart-phone based, ride-sharing applications has revolutionized the vehicle for hire market. An alternative to traditional taxi and limousine services, ride-sharing applications, such as Uber and Lyft, enable potential passengers to ‘hail’ nearby private drivers via geolocation. Potential passengers and drivers both broadcast their locations, quickly map the distance to one another, agree on a price, and estimate the likely wait time. Although matching drivers to potential passengers in real-time provides a greater ease of service, this innovation has encountered much scrutiny (Rogers, 2015). Much of the scrutiny stems from the lack of state and municipal safety regulations that are required of ride-sharing’s competitors: traditional taxis and limousines.

We investigate whether the introduction of the ride-sharing service, Uber, is associated with net changes in vehicular fatalities and arrest rates. Ride-sharing passengers, drivers, and others may respond to Uber in a variety of ways. With little to no regulation, ride-sharing passengers, as well as pedestrians and occupants in nearby vehicles, may be subject to a greater risk of injury from driver error or part failure. The use of smartphone applications by drivers and increases in the number of passengers per vehicle potentially increase driver distraction. The increased interaction of non- government certified drivers and passengers may result in greater violence. On the other hand, these applications reduce passenger wait times and may encourage some drowsy or intoxicated potential drivers to ride instead (Rayle, Shaheen, Chan, Dai, and Cervero 2014). Yet, this ease of use might also increase alcohol consumption and other risky behavior. All of these behavioral responses affect the risk of vehicular crashes and crime. 1 What is unknown is the direction and magnitude of these effects. We empirically estimate these effects.

To do so, we first use monthly data from the National Highway Traffic Safety Administration’s Fatality Analysis Reporting System (FARS) to study whether Uber’s entry is associated with changes in the overall rate of fatal automobile accidents. We also examine three related measures: alcohol-related fatal crashes, night-time fatal

1 We use the terms ‘crashes’ and ‘accidents’ interchangeably, though we note that the terms have unique connotations. For more see, Stromberg (2015).

2

crashes, and the number of vehicular fatalities per 100,000. Using a differences-in- differences specification, we find that fatal accident rates generally decline after the introduction of Uber. Specifically, in the unweighted regressions, we find that entry is associated with a 7 percent decline in the fatal accident rate. In both the weighted and unweighted estimations, we also discover a continued decline in the overall fatal crash rate and the rate of night-time fatal crashes for the months following the introduction of Uber.

Next we use the Federal Bureau of Investigation’s Uniform Crime Reporting (UCR) program to explore whether the introduction of Uber is associated with changes in arrests for particular types of crime: aggravated assaults, other assaults, motor vehicle thefts, driving under the influence (DUI), drunkenness, and disorderly conduct. Again employing a differences-in-differences specification, typically with county- specific trends, we find a large and robust decline in the arrest rate for DUIs. Depending upon the specification, DUI arrests are 6 to 27 percent lower after the entry of Uber. Recognizing that it takes time for potential users to become aware of the service and for current users to become more familiar with the process, we separately estimate a 2.8 to 3.4 percent decline in DUIs for each additional month Uber is available. For most specifications, we also observe declines in the arrest rates for non-aggravated assaults. However, the arrest rates for motor vehicle theft increase.

We expand the literature on transportation options, crime, and traffic rates by addressing a specific, new industry attracting significant public attention and talk of regulation. In the paper most similar to ours, Greenwood and Wattal (2015), use a differences-in-differences approach to show that the entry of Uber into California markets between 2009 and 2014 was associated with a significant drop in the rate of motor vehicle homicides. We expand this analysis geographically to encompass all entry across the United States. Further, we buttress these results with an analysis of arrest rates, including arrests for DUIs, providing completely new evidence of reduced drunk-driving following the introduction of ride-sharing services.

In the following section, we discuss the possible relationships between ride- sharing, fatal traffic accidents, and a variety of crimes. In section III, we present our data on entry, accidents, and crime. In section IV, we present our differences-in-differences

3

empirical strategy. We reveal the estimation results in section V and then conclude in section VI.

II. The interplay between transportation, traffic accidents, and crime

Much heat has been generated and ink spilled over the effects of ride-sharing. Although ride-sharing existed long before the advent of smart-phones, ride-sharing exploded when innovators such as Uber’s Travis Kalanick and Garrett Camp and Lyft’s Logan Green and John Zimmer began using the geolocation function of smart-phones to match private drivers with potential riders. These applications also enabled ride- sharing services to instantaneously alter prices in response to changes in supply and demand. By using geolocation, introducing more flexible pricing, and encouraging automated payments, ride-sharing services offer greater convenience and shorter wait times than other point-to-point transportation options. Many hail these innovations. Others, including taxi and limousine owners and drivers, note the potential safety risks to such unregulated ride-sharing services.

To determine whether the lack of oversight has exposed citizens to greater harm, we investigate whether the introduction of Uber’s ride-sharing service is associated with changes in vehicular fatality rates and arrest rates. The advent and expansion of ride-sharing services may affect traffic accidents and crime rates through a variety of mechanisms.

We first consider the potential effects of this new transportation option on traffic accidents. Ride-sharing may affect traffic accidents through a variety of mechanisms:

changes in the number of vehicles on the road; changes in rates of distracted, drowsy, or drunk driving; changes in the quality of drivers; and changes in the quality of vehicles.

First, ride-sharing services may change the total number of vehicles on the road.

If ride-sharing only substitutes for driving oneself, we would expect no change in

vehicles miles traveled. If ride-sharing increases travel, it leads to more congestion and

a greater probability of collisions. Those calling for increased regulation note that

county or municipal restrictions typically limit the number of taxis and limousines.

4

Expanding the number of vehicles providing point-to-point transportation may increase vehicle miles traveled and the per capita incidence of vehicular accidents.

Second, ride-sharing may change the likelihood of driving while distracted, drowsy, or drunk. Passengers are often cited as the greatest source of distraction (NHTSA 2014). By increasing the number of private drivers with passengers, ride- sharing may increase the fraction of distracted drivers. In addition, using any type of phone based application, whether hands free or not, likely increases the level of distracted driving. 2 When a potential rider electronically hails a driver, the Uber application alerts a driver by sound, but the driver must then determine the time and distance to the potential rider. Interacting with potential riders via the smart-phone application is clearly driving while distracted. 3 This distraction increases the probability of a crash and lowers the safety of passengers as well as nearby drivers and pedestrians. Drowsy and drunk-driving, however, may decrease. In many locations, ride-sharing applications have larger geographic coverage than traditional taxis and require less notice than traditional limousine services. These customer-friendly characteristics may reduce the number of drivers who might otherwise drive under the influence or while drowsy. The net effect of ride-sharing services on impaired driving is ambiguous.

Third, ride-sharing may change the composition of point-to-point drivers through differing standards for drivers-for-hire. Ride-sharing drivers hold traditional private driver’s licenses, whereas most taxi and limousine drivers must hold commercial driver’s licenses. Commercially licensed taxi and limousine drivers undergo background checks, vehicle specific tests, a thorough driver history examination, and medical certification. This more intensive licensing process may reduce the likelihood of driver error and accidents. Medically certified commercial drivers may provide better medical care to passengers in the event of an accident.

2 The “NHTSA (2014) reports that in 2012, 16% of all police-reported crashes involved any driver distraction, and that 7% of crashes that involved some form of distraction (and 1.1% of all crashes) involved distraction due to cell phone use” (Carney et al., 2015, p.45). This is especially true for younger drivers (Doherty et al., 1998; Chen et al., 2000; Mayhew et al., 2003; Williams, 2003; Williams, Ferguson et al., 2007). According to Carney et al. (2015), distraction was a factor in 58% of automobile accidents by 16- 19 year olds. 3 For a discussion of ride-sharing and distracted driving see: Richtel, Matt. (2014). “Distracted Driving and the Risks of Ride-Hailing Services Like Uber” New York Times, December 21, 2014. 7:00 AM.

5

Fourth, the quality of vehicles on the road may change. The number and type of inspections differ for commercial and private passenger vehicles. For instance, New York City taxis are inspected 3 times per year, while private vehicles are inspected at most once a year and, in some locations, not at all (NYC Taxi & Limousine Commission 2014). Fewer vehicle inspections increase the likelihood of accidents and injuries from parts failures, especially wear items such as tires and brakes. Those seeking more regulation also note that if a crash does occur, safety controls, such as airbags and seat belts, are more likely to operate properly in a well-inspected commercial vehicle. On the other hand, ride-sharing drivers are more likely to own their vehicle. Because the owner-driver is the residual claimant of the vehicle’s value, they are likely to be more concerned with the vehicle’s appearance and safety equipment than a non-owner driver. Being an owner-driver reduces many of the principal-agent problems present in the vehicle-for-hire market, leading to safer drivers and vehicles.

The net effect of ride-sharing on accidents is ambiguous, motivating empirical testing.

Ride -sharing may also affect crime. Ride-sharing options may affect crime rates in a variety of ways: they may change the availability of victims, the cost of fleeing the scene, or increase alcohol consumption.

First, ride-sharing services may change the availability of victims both by reducing potential riders’ wait times and by increasing interaction of riders and drivers. Because potential riders are able to electronically hail a ride through the application, wait times are likely to be much shorter. Wait times are also likely to be lower because ride-sharing applications can quickly adjust prices in response to changes in the number of riders and drivers. Potential ride-share passengers do not need to physically search for a vehicle as they do for a taxi. This reduces opportunities for them to become the victim of a street crime. Potential passengers can also leave on short notice. This may reduce assaults. On the other hand, ride-sharing drivers are not subject to as thorough

6

of a driver history and criminal background check as are taxicab and limousine drivers. This may increase the risk to passengers of theft, assault, and even death. 4

Second, criminals also experience a greater ability to leave on short notice (“Uber driver unwittingly becomes getaway driver,” 2015). Because this reduces the likelihood of apprehension, it may increase crime rates. We analyze arrest rates. Greater ability to avoid arrest would appear as a decline in arrest rates holding crime rates constant.

Third, increased transportation options may change drinkers’ behavior. Ride- sharing services encourage drinkers to find a ride, reducing arrests for DUI and public intoxication. However, to the extent that this lowers the cost of drinking and therefore increases the level of drinking, violations for disorderly conduct and public drunkenness may increase.

The net effect of ride-sharing on crime is ambiguous, motivating empirical testing.

Previous literature suggests that transportation options affect crime rates. For example, Philips and Sandler (2015) use closures of subway stations to estimate how the availability of public transportation affects crime rates. They find that closing a station reduces nearby crime, primarily in locations to which perpetrators are likely to travel to find victims. Jackson and Owens (2011) consider the effects of changing public transportation schedules on alcohol-related behavior. They find that the availability of late-night public transportation likely increases alcohol consumption, leading to more arrests for minor crimes near bars but fewer DUI arrests in those areas. In the paper most similar to ours, Greenwood and Wattal (2015), use a differences-in-differences approach to show that the entry of Uber into California markets between 2009 and 2014 was associated with a significant drop in the rate of motor vehicle homicides. Given the theoretical ambiguity of many of these responses, we empirically investigate for greater clarity.

4 The high profile case of Uber driver Jason Dalton, who shot and killed six people while driving in Kalamazoo Michigan, has increased the level of scrutiny over passenger and pedestrian safety. For more on this case see: Kiplinger (2016). The website ‘Who’s Driving You?’ keeps a comprehensive list of incidents at http://www.whosdrivingyou.org/rideshare-incidents.

7

III. Data

When investigating the effects of ride-sharing on fatal vehicle crashes and crime, we chose to focus on Uber. We chose Uber for two reasons. First, Uber is the oldest and largest of the ride-sharing applications. 5 In 2014, Uber performed over 10 million rides per month, while Lyft garnered second place in market share with an average of only 2.2 million per month (Miller 2015). According to MIT Technology Review, Lyft revealed in 2015 that the company provided only “7 percent of rides summoned over the Internet in [New York City], compared with Uber’s 90 percent” (Bradley 2015). Uber has eight times the funding of Lyft and several times Lyft’s roughly 100,000 active drivers (Bradley 2015). Second, Uber’s website provides the day and month Uber began service in a city, county, or region (Uber 2015). This enables us to easily merge monthly accident and crime data with entry dates from Uber.

Although Uber’s pilot program included a small number of trial runs in both New York and San Francisco in early 2010, Uber officially began service on May 31, 2010 in San Francisco (Uber 2015). Figure 1 shows the number of areas in the United States served by Uber by month and year. By the end of 2011, Uber was serving 11 cities or counties; by the end of 2012, the number was up to 20. By the end of 2014, Uber was serving 155 counties throughout the United States. 6

To investigate whether Uber’s entrance is associated with changes in the rate of fatal crashes, we use the National Highway Traffic Safety Administration’s (NHTSA) Fatality Analysis Reporting System (FARS). The NHTSA’s FARS data reports fatal automobile accident rates, whether a fatal crash is classified as alcohol-related, the time of the crash, and the total number of vehicular fatalities. 7 As reported in Table 1, there are 4.15 fatal vehicular crashes per 100,000 people. There are 0.50 fatal alcohol-related crashes per 100,000 and 5.12 vehicular fatalities per 100,000. Given our investigation of

5 Lyft and Sidecar were second and third in market share of ride-sharing services nationwide during the period we examine. In December 2015, Sidecar ceased operations.

6 Table A1 in the appendix lists, by FIPS code, all areas served between 2010 and 2014.

7 Night time accidents are frequently used as a proxy for alcohol-related accidents as most alcohol-related accidents occur during this time; time of accident may be more accurately coded than alcohol involvement.

8

whether Uber’s entry is associated with crash rates, we also report rates for areas without Uber and areas where Uber is available for at least one month. Areas where Uber is never present have higher fatal accident rates overall and for each type. Therefore, it does not appear that Uber enters areas with higher accident rates. If anything, this may suggest that Uber enters areas where accident rates are lower and thus any endogeneity may cause an upward bias of the estimates.

For our measures of crime rates, we use the FBI’s Uniform Crime Reports Statistics from 2007 through 2014 (United States Department of Justice, various years). We report the arrest rate summary statistics in the top portion of Table 2. 8 There are 29.0 arrests for aggravated assaults and 114.6 arrests for other types of assaults per 100,000. There are 5.4 arrests for vehicle thefts and 121.0 for DUIs per 100,000. In the bottom portion of Table 2 we investigate whether the arrest rates in areas that never have access to Uber are different than those where Uber is or becomes available. Looking at assaults, we find that areas where Uber is never present witness half as many assault arrests per person as places that have access to Uber. Moreover, places that are ultimately home to Uber have a higher assault rate (314.6 per 100,000) before the introduction of Uber than after (282.4). Areas without Uber have much lower DUI rates. For areas that witness the entry of Uber between 2010 and the end of 2014, the DUI rate is higher (301) before the ride-sharing service is available than after (266). This is true for most other types of crime as well and thus suggests that Uber’s entry may be correlated with arrest rates. To help mitigate concerns about endogeneity, the empirical method described in the next section includes county fixed effects and county-specific linear trends. To further address endogeneity concerns, we supplement the estimates using the full sample with estimates using only the sample of counties that will witness Uber entry.

IV. Empirical Methods

To determine whether the changes in fatal accidents and crime are associated with Uber’s entry, we estimate for county i in month m and year t the following:

8 We drop Carroll County, IN (FIPS = 18015) from the crime data analysis because the values reported do not appear to be reliable.

9

= + + +

The equation is a standard differences-in-differences specification. The monthly outcomes include per capita arrests for a variety of potentially related crimes as well as fatal vehicle accidents per capita. The variable of interest, uber, indicates whether the ride sharing service UberX was available to individuals located in the county in that month and year.

We control for other factors including demographic variables, a variety of alcohol and driving-related laws, real per capita income, the maximum welfare benefit for a family of three, and the legal status of marijuana. The specification includes month-by-year fixed effects and county-specific linear time trends. Standard errors are clustered by county. Some specifications are weighted by county population.

A typical concern with differences-in-differences estimation is the possibility of endogenous entry. If, for example, Uber enters areas with differing trends in the propensities to go out, to drive drunk, or to commit crimes, the estimated effect of entry also captures these differing propensities. The county-specific linear time trends capture much of this possibility by controlling for trends in the outcome variables specific to the geographic area. In addition, we separately estimate the results on the sample of counties who will witness Uber entry. This sample identifies the effect of Uber on differential timing of its entry into the county.

One concern with including county-specific trends is that the time trend in the outcome may be confounded with the effect of the policy change of interest. For example, if the effect of Uber increases over time as more residents, visitors, and drivers become aware of the possibility of using or driving for the service then this effect will be picked up in the county-specific trend. Wolfers (2006) points out that fitted county- specific linear trends capture both the pre-existing trend and the policy response. To account for this possibility, we estimate specifications that allow the effect of Uber to grow (or shrink) the longer Uber exists in a county. We estimate specifications that allow entry to affect both the mean and the post-entry trend:

= 1 + 2 ∗ + + +

10

We explore a variety of fatal crash rates, the number of vehicular fatalities per 100,000, and arrest rates for a variety of crimes. The predicted effect of Uber on fatal accident rates is ambiguous; increased transportation options have ambiguous effects on the number of vehicles on the road, driver conditions, vehicle ownership, and vehicle quality. The predicted effect of Uber on arrest rates is crime-specific. Increased alcohol-consumption may increase crimes such as assaults, DUIs, drunk, and disorderly conduct. Ability to leave a tense situation or to not linger outside waiting for a cab likely reduces crimes such as assaults. Greater ability to find alternate transportation likely reduces DUIs, drunkenness, and disorderly conduct violations. Increased interaction with strangers via ride-sharing potentially increases assaults or thefts.

V. Results

a. Traffic fatalities

Table 3 presents results examining whether the entry of Uber into a county is associated with any change in fatal crashes. We consider four measures of fatal traffic accidents per capita: total, alcohol-related, night time, and the number of fatalities. Panel A presents results using the differences-in-differences specification with county- specific linear trends. In the unweighted regressions, entry of the ride-sharing service reduces accidents. Specifically, entry corresponds to a 0.28 reduction in fatal crashes per capita, a 7 percent decline at the mean. Night time accidents experience a decline of 0.09, a 9 percent decrease, and vehicular fatalities fall by 4 percent at the mean, although these estimates are not statistically different from zero. Our findings are similar to those estimated by Greenwood and Wattal (2015) who find that vehicle homicides fall by about 3.6% in locations treated by Uber X in the state of California. The estimated effects on alcohol-involved crashes is negative, but not statistically significant. Weighting the results by county population, leads to estimates that are mostly negative, smaller, and statistically insignificant.

At first glance these magnitudes may seem large. Eisenberg (2003) considers a range of drunk-driving related policies and estimates effects from 4 percent to 9.4 percent on drunk-driving fatalities. Dills (2010) finds a 9 percent decline in drunk- driving fatalities due to social host laws. Thus, our estimates are not far off from the effects of changes in drunk-driving policies and social host laws. Yet, Uber is, in many

11

ways, different from these legal restrictions. Unlike most policy changes, the adoption and use of Uber is voluntary. That is, potential drivers voluntarily choose to ride-share because they view it as the best alternative and not because there is a threat of punishment. Moreover, those most likely to adopt and use Uber are exactly the groups mostly likely to be involved in collisions: traffic fatality rates are higher among younger drivers and fall by age until 65 (Chang 2008). Smartphone owners are more likely to be aged 18 to 35 (Smith 2012). And most importantly, forty percent of Uber’s users are aged 25 to34 and another 28 percent are 35 to44. 9 Providing what customers believe to be a superior point-to-point transportation alternative for those most likely to be in a collision is likely to have larger effects than those found in broad, punishment based policies.

Recognizing that it takes time for potential users to become aware of the service and for current users to become more familiar with the process, Panel B of Table 3 allows the effect of entry to differ as time passes. Our unweighted estimates are consistent with Uber leading to larger declines in fatal accidents the longer the service is available. Fatal crashes decline by 0.5 percent for each additional month or 1.5 percent for each additional quarter Uber is available. Night-time fatal crashes decline by 0.9 percent for each additional month or 2.7 percent per quarter. The number of fatalities decline by 0.37 percent for each additional month or 1.1 percent for each additional quarter Uber is available. Our estimates are a third of the size as those in Greenwood and Wattal (2015) who find a “3.6% – 5.6% decrease in the rate of motor vehicle homicides per quarter [or 0.9% - 1.4% per month] in the state of California.” In the weighted regressions, the estimated effect over time tends to be smaller and statistically significant. We observe statistically significant and economically meaningful declines in fatal accidents, fatal night time accidents, and the number of fatalities the longer Uber is available.

Our results in Table 3 may be biased if Uber enters areas that differ from the rest of the country. To alleviate this concern, we restrict the sample to only those areas that witness the entry of Uber. In this restricted sample, any variation between the control and treatment groups for the differences-in-differences specifications must come from differences in the timing of Uber’s entry into an area. The much smaller sample yields

12

less precise estimates. Estimates in panel A of Table 4 are mostly positive and statistically insignificant. Allowing for the change in slope suggests that declines in accidents may occur several months to a year post-entry. Coefficients in the unweighted sample are statistically insignificant. In the population-weighted regressions, we observe statistically significant declines in the rates of fatal crashes, night-time fatal crashes, and traffic fatalities that occur within four to eight months post-entry. The magnitudes of these results are, at the ever-Uber means, 0.6 percent per month for fatal crashes, 0.5 percent per month for fatal night-time crashes, and 0.8 percent per month for traffic fatalities; these are roughly similar to the monthly declines in the full sample.

Overall, our findings suggest that Uber does not increase overall fatal crash rates and, for some specifications, is associated with a decline in fatal crash rates.

b.

Arrests

Many articles have been written about assaults and homicides at the hands of ride-share drivers (Kiplinger, 2016). Citing these incidents, some people seeking government regulation note that commercial drivers are subject to more thorough driver history audits and criminal background checks than ride-sharing drivers. These requirements may reduce the risk of driver-on-passenger violence. Therefore, ride- sharing passengers may be more at risk of driver-on-passenger violence. Commercial drivers also receive training on how to handle unruly passengers. Without proper training, the risk to ride-sharing drivers may be higher as well.

To investigate whether ride-sharing is associated with changes in crime, we consider the effect of Uber’s entrance on arrests per 100,000 for a variety of crimes. Many of these crimes are likely to be affected by additional transportation options: DUI, drunkenness, and disorderly conduct. Another group of crimes may be affected by the increased interaction of strangers: aggravated assault and other assaults. Because ride- sharing offers a new point-to-point transportation option, we also include motor vehicle thefts.

Table 5 presents estimates for all counties. Panels A and B contain the differences-in-differences estimates without and with population weights, respectively. The results are similar with and without weights: counties with Uber experience statistically significant declines in arrests for other assaults and DUIs. The magnitudes are economically important and typically larger for the weighted estimates. For other

13

assaults, the entrance of Uber is associated with a 11 to 18 percent decline. The availability of Uber is associated with a 6 to 27 percent decline in DUIs. Counties experience a 55 to 157 percent increases in arrests for motor vehicle thefts after the introduction of Uber. This may come from an increased propensity for Uber passengers to leave personal vehicles parked in public locations.

The specifications in Panels C and D allow Uber’s entry to also affect the post- entry trend. For DUIs, we witness a 2.8 to 3.4 percent decline for each additional month of Uber service. We continue to observe declines in arrests for assault; each additional month of Uber availability is associated with a 2.4 percent decline in assaults in the unweighted estimate. The results for motor vehicle thefts are also consistent across specifications with some evidence of increasing thefts over time.

Because we are concerned that Uber may enter areas with characteristics that are correlated with crime rates, we restrict the sample to only those areas where Uber services have been offered. We then repeat the same analysis as in Table 5. As shown in Panel B of Table 6, arrests for DUI decline by 17 percent with the entry of Uber. Including both the entry and trend effects, the Panel C and D estimates reveal a 2.7 to 3.9 percent decline in DUIs for each additional month Uber service is available. Motor vehicle thefts increase following the entry of the ride-sharing service. The results for assaults, however, become statistically insignificant.

To investigate whether any findings may be spurious, we analyze a number of crimes that are unlikely to be associated with ride-sharing: liquor law violations, fraud, and embezzlement. 10 Table 7 presents these estimates for the specifications and samples in Tables 5 and 6. As expected, we find no relationship between Uber on the arrest rates for liquor law violations, fraud, or embezzlement.

10 Liquor law violations are “The violation of state or local laws or ordinances prohibiting the manufacture, sale, purchase, transportation, possession, or use of alcoholic beverages, not including driving under the influence and drunkenness. Federal violations are excluded.” (US DoJ September 2010 “Offense Definitions” accessed Feb. 1, 2016 https://www2.fbi.gov/ucr/cius2009/about/offense_definitions.html). In results not reported here, we examine effects on arrests for family violations and curfew/loitering violations. These two crimes may be less clearly counterfactuals as family violations may be correlated with alcohol consumption and curfew/loitering may be affected by availability of transportation. Nonetheless, we estimate no effects of Uber on these arrest rates.

14

Our estimates reveal that the introduction of Uber lowers arrests due to DUIs and may lower assaults. Overall, this suggests that the introduction of Uber increases the safety of citizens. We also witness little to no change in liquor law violations, fraud, or embezzlement. This suggests that our findings are not due to overall declines in crime rates. We do, however, witness an increase in the theft of vehicles.

VI. Conclusion

Claiming consumer protection, some community’s leaders have sought greater government oversight and limits on the entry of ride-sharing services (for example Moore 2016 and O’Sullivan 2016). Articles often cite concerns about the safety of riders and drivers in this comparatively unregulated service (whosdrivingyou.org, 2016).

We investigate and find that many of these concerns are, at least on net, unwarranted. Using a differences-in-differences specification and controlling for county-specific linear trends, we find that the entry of ride-sharing tends to decrease fatal vehicular crashes. Our (unweighted) estimated 1.1 percent decline in vehicle fatalities for each additional quarter are smaller than those found by Greenwood and Wattal

(2015).

We also observe declines in arrests for assault and DUI. Specifically, we find that Uber’s entry lowers DUIs rates by 6 to 27 percent. The magnitude of our findings are smaller than those found by Jackson and Owens (2011) who show that DUIs decreased by 40% when the Washington DC Transit Authority expanded late night Metro transportation services. In many cases, these declines become larger the longer the service is available in an area. These beneficial declines are somewhat offset by increases in arrests for motor vehicle thefts.

15

Works Cited

Bradley, Ryan. (2015). Lyft’s Search for a New Mode of Transport. MIT Technology Review. October 13, 2015. Available on February 22, 2016 at

Chang, Dow (2008) Comparison of Crash Fatalities by Sex and Age Group. Traffic Safety Facts Research Note, July, DOT HS 810 853 available at http://www-

nrd.nhtsa.dot.gov/Pubs/810853.pdf

Chen, L. H., Baker, S. P., Braver, E. R., & Li, G. (2000). Carrying passengers as a risk factor for crashes fatal to 16-and 17-year-old drivers. JAMA, 283(12), 1578-1582.

Cher Carney, Dan McGehee, Karisa Harland, Madonna Weiss, Mireille Raby (2015). Using Naturalistic Driving Data to Assess the Prevalence of Environmental Factors and Driver Behaviors in Teen Driver Crashes. AAA Foundation for Traffic Safety. (March 2015)

Dills, Angela K. (2010). Social host liability for minors and underage drunk-driving accidents. Journal of Health Economics 29: 241-249.

Doherty, S. T., Andrey, J. C., & MacGregor, C. (1998). The situational risks of young drivers: The influence of passengers, time of day and day of week on accident rates. Accident Analysis & Prevention, 30(1), 45-52.

Eisenberg, Daniel. (2003). “Evaluating the Effectiveness of Policies Related to Drunk Driving.” Journal of Policy Analysis and Management 22(2): 249-274.

Greenwood, Brad N. and Wattal, Sunil, Show Me the Way to Go Home: An Empirical Investigation of Ride Sharing and Alcohol Related Motor Vehicle Homicide (January 29, 2015). Fox School of Business Research Paper No. 15-054. Available at: http://ssrn.com/abstract=2557612 orhttp://dx.doi.org/10.2139/ssrn.2557612

Jackson and Owens. (2011). One for the road: public transportation, alcohol consumption, and intoxicated driving. Journal of Public Economics 95:106-121.

16

Kiplinger, Lisa. (2016). Kalamazoo shooting: A look at Uber background checks. USA TODAY 7:55 a.m. EST February 22, 2016. http://www.whosdrivingyou.org/rideshare-incidents#deaths (Viewed 2/29/16).

Mayhew, D. R., Simpson, H. M., & Pak, A. (2003). Changes in collision rates among novice drivers during the first months of driving. Accident Analysis & Prevention, 35(5), 683691

Miller, Dennis. (2015). Lyft vs. Uber: Just How Dominant Is Uber in the Ridesharing Business? (http://www.fool.com/investing/general/2015/05/24/lyft-vs-uber-just- how-dominant-is-uber-ridesharing.aspx). (Viewed 2/29/16).

Moore, Daniel. (2016). Lobbying efforts intensify as ride-sharing bill enters final stretch Pittsburgh Post-Gazette. (http://www.post-gazette.com/business/tech-

National Highway Transportation Safety Administration (NHTSA). (2014). Traffic Safety Facts, Distracted Driving 2012, DOT HS 812 012. April.

National Highway Traffic Safety Administration’s (NHTSA) Fatality Analysis Reporting System (FARS). (2016). http://www.nhtsa.gov/FARS (Viewed 2/2/16).

NYC Taxi & Limousine Commission (2014). 2014 Taxicab Fact Book. Office of Policy and External Affairs.

O’Sullivan, Jim. (2016). Uber, Lyft would face new rules under state bill. Boston Globe. MARCH 04, 2016. https://www.bostonglobe.com/metro/2016/03/04/uber-lyft-

(Viewed 3/4/2016).

Phillips, David C. and Danielle Sandler. (2015). Does public transit spread crime? Evidence from temporary rail closures. Regional Science and Urban Economics 52:

13-26.

17

Rayle, Lisa, Shaheen, Susan, Chan, Nelson, Dai, Danielle, and Cervero, Robert. (2014). App-Based, On-Demand Ride Services: Comparing Taxi and Ridesourcing Trips and User Characteristics in San Francisco. University of California Transportation Center (UCTC) Working Paper.

http://tsrc.berkeley.edu/sites/default/files/RidesourcingWhitePaper_Nov2014Up

date.pdf

Richtel, Matt. (2014). “Distracted Driving and the Risks of Ride-Hailing Services Like Uber” New York Times, December 21, 2014. 7:00 AM.

Rogers, Brishen. (2015). The Social Costs of Uber. 82 U Chi L Rev Dialogue 85

Smith, Aaron. (2012). “46% of American adults are smartphone owners.” PewInternet. Available April 25, 2016 at

http://www.niu.edu/facdev/m/twti/2012/Smartphone_ownership_2012.pdf

Stromberg, Joseph (2015). We don’t say “plane accident.” We shouldn’t say “car accident” either. Vox. http://www.vox.com/2015/7/20/8995151/crash-not-accident. (Viewed 5/24/2016).

2/29/16).

Williams, A. F. (2003). Teenage drivers: patterns of risk. Journal of Safety Research, 34(1),

5-15.

Williams, A. F., Ferguson, S. A., & McCartt, A. T. (2007). Passenger effects on teenage driving and opportunities for reducing the risks of such travel. Journal of Safety Research, 38(4), 381-390.

Wolfers, Justin. (2006). “Did Unilateral Divorce Laws Raise Divorce Rates? A Reconciliation and New Results.” American Economic Review, 96(5): 1802-20.

18

United States Department of Justice, Federal Bureau of Investigation. Crime in the United States, 2007-2014. https://www.fbi.gov/about-us/cjis/ucr. (Viewed 1/8/16).

Uber (2015). Uber.com (Viewed 2/2/15).

“Uber driver unwittingly becomes getaway driver in Parkville armed robbery.” (2015, October 9). WBAL. http://www.wbaltv.com/news/uber-driver-unwittingly- becomes-getaway-driver-in-parkville-armed-robbery/35746542. (Viewed

5/24/2016).

19

Figure 1: Number of Areas Served by Uber, By Month and Year

Jul 2010 Jul 2011 Jul 2012 Jul 2013 Jul 2014 Date 100 150 200 50
Jul 2010
Jul 2011
Jul 2012
Jul 2013
Jul 2014
Date
100
150
200
50
0
of Areas Served by Uber, By Month and Year Jul 2010 Jul 2011 Jul 2012 Jul
of Areas Served by Uber, By Month and Year Jul 2010 Jul 2011 Jul 2012 Jul
of Areas Served by Uber, By Month and Year Jul 2010 Jul 2011 Jul 2012 Jul
of Areas Served by Uber, By Month and Year Jul 2010 Jul 2011 Jul 2012 Jul
of Areas Served by Uber, By Month and Year Jul 2010 Jul 2011 Jul 2012 Jul

20

Table 1: Summary Statistics: FARS data

 

Standard

Minimu

Maximu

 

Mean

Deviation

m

m

Fatal Crashes per 100,000 (N=285,780) Total Alcohol-involved Night-time Motor Vehicle Fatalities per 100,000

4.18

17.19

0

1674.64

0.50

3.76

0

952.381

1.01

7.80

0

1108.647

5.16

31.26

0

4701.554

 

Never

Uber

Ever Uber (N=15,204)

N=270,57

 

6

All

Uber = 0

Uber = 1

Fatal Crashes per 100,000 Total Alcohol-involved Night-time Motor Vehicle Fatalities per 100,000

4.31

1.82

1.87

1.49

0.52

0.21

0.22

0.15

1.04

0.60

0.61

0.51

5.33

2.11

2.17

1.69

21

Table 2: Summary Statistics: UCR arrest data Standard

 

Mean

Deviation

Minimum

Maximum

UCR data: Arrests per 100,000 (N = 237,852)

 

Aggravated Assault Other Assaults Motor vehicle thefts DUI Drunkenness Disorderly Liquor law violations Forgery Embezzlement

28.2

63.2

0

6452

114.3

499.4

0

158465

5.3

22.2

0

6452

121.0

1184.1

0

474715

56.1

2367.1

0

837170

64.4

987.9

0

474763

56.9

2345.4

0

1107611

6.7

31.5

0

6452

1.5

11.0

0

3226

Never

Uber

Ever Uber (N=13,316)

 

N=224,536

All

Uber = 0

Uber = 1

Aggravated Assault Other Assaults Motor vehicle thefts DUI Drunkenness Disorderly Liquor law violations Forgery Embezzlement

24.4

91.0

88.6

109.7

102.2

317.0

321.0

286.4

4.9

13.0

12.7

14.8

110.1

303.5

308.6

264.1

52.9

110.0

110.6

105.4

58.5

162.4

167.8

120.6

53.6

113.0

118.1

73.8

6.3

14.6

15.1

11.0

1.4

3.3

3.4

2.6

22

Table 3: Uber entry and fatal traffic crashes per 100,000

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

 

Alcohol-

Night-

Alcohol-

Night-

involved

time

involved

time

 

Fatal

fatal

fatal

Vehicular

Fatal

fatal

fatal

Vehicular

Crashes

crashes

crashes

Fatalities

Crashes

crashes

crashes

Fatalities

 

unweighted

 

population weighted

 

Panel A: Differences-in-differences with linear county trends

 

Uber

-0.279**

-0.0231

-0.0876

-0.224

-0.0523

-0.000651

0.00183

-0.0234

(0.119)

(0.0260)

(0.0583)

(0.191)

(0.0441)

(0.00714)

(0.0208)

(0.0892)

R-squared

0.080

0.033

0.038

0.056

0.127

0.073

0.061

0.071

Panel B: Differences-in-differences with linear county trends, post-entry trend

 

Uber

-0.163

-0.00607

-0.0392

-0.126

-0.00235

0.000765

0.0140

0.0526

(0.106)

(0.0222)

(0.0506)

(0.155)

(0.0434)

(0.00657)

(0.0193)

(0.0753)

Uber*trend

-0.0226**

-0.00332

-0.00942*

-0.0191

-0.0116***

-0.000330

-0.00283*

-0.0177***

(0.0115)

(0.00215)

(0.00497)

(0.0186)

(0.00344)

(0.000490)

(0.00150)

(0.00680)

R-squared

0.080

0.033

0.038

0.056

0.127

0.073

0.061

0.071

There are 285,780 observations. All specifications include month-by-year fixed effects, county fixed effects, and county-specific linear trends. We also control for whether marijuana is decriminalized, medicalized, or legalized; the percent of the population who are black, aged 20 to 24, aged 25to 34, aged 35 to 54, and aged 55 and over; indicators for whether the state has a graduated drivers licensing law, zero tolerance law, maximum legal blood alcohol concentration of 0.08; state real per capita personal income, and the state maximum welfare benefit for a family of three. Standard errors are clustered by county. *** p<0.01, ** p<0.05, * p<0.1

23

Table 4: Uber entry and fatal traffic crashes per 100,000, sample of counties ever hosting Uber

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

 

Night-

Night-

 

Alcohol-

time

Alcohol-

time

 

Fatal

involved

fatal

Vehicular

Fatal

involved

fatal

Vehicular

Crashes

fatal crashes

crashes

Fatalities

Crashes

fatal crashes

crashes

Fatalities

 

unweighted

 

population weighted

 

Panel A: Differences-in-differences with linear county trends

 

Uber

0.00536

0.00748

-0.0337 0.0889

0.0271

0.00763

0.0135

0.0921

(0.0751)

(0.0179)

(0.0516)

(0.122)

(0.0461)

(0.00710)

(0.0204)

(0.0844)

R-squared

0.290

0.148

0.136

0.177

0.359

0.262

0.219

0.149

Panel B: Differences-in-differences with linear county trends, post-entry trend

 

Uber

0.0219

0.00977

-0.0271

0.107

0.0340

0.00753

0.0159

0.102

(0.0772)

(0.0179)

(0.0506)

(0.119)

(0.0457)

(0.00697)

(0.0200)

(0.0821)

Uber*trend

-0.0134

-0.00185

-0.00528

-0.0147

-0.00968**

0.000148

-0.00340*

-0.0144*

(0.00994)

(0.00155)

(0.00364)

(0.0145)

(0.00409)

(0.000609)

(0.00190)

(0.00845)

R-squared

0.290

0.148

0.136

0.177

0.359

0.262

0.219

0.149

There are 15,204 observations. All specifications include month-by-year fixed effects, county fixed effects, and county-specific linear trends. We also control for whether marijuana is decriminalized, medicalized, or legalized; the percent of the population who are black, aged 20 to 24, aged 25to 34, aged 35 to 54, and aged 55 and over; indicators for whether the state has a graduated drivers licensing law, zero tolerance law, maximum legal blood alcohol concentration of 0.08; state real per capita personal income, and the state maximum welfare benefit for a family of three. Standard errors are clustered by county. *** p<0.01, ** p<0.05, * p<0.1

24

Table 5: Uber entry and Arrests per 100,000: Differences-in-differences estimates

(1)

(2)

(3)

(4)

(5)

(6)

 

Motor

 

Aggravated

Other

vehicle

Disorderly

Assault

Assaults

thefts

DUI

Drunk

conduct

Panel A: with linear county trends (unweighted)

 

Uber

-1.184

-12.53*

2.945***

-7.525

7.066

-3.612

(2.398)

(6.512)

(0.889)

(10.21)

(15.15)

(11.79)

R-squared

0.688

0.167

0.155

0.068

0.053

0.048

Panel B: with linear county trends (weighted)

 

Uber

-14.77

-20.58**

8.393**

-32.16***

-10.36

4.554

(12.96)

(8.954)

(4.009)

(11.56)

(9.830)

(6.518)

R-squared

0.974

0.710

0.894

0.501

0.150

0.489

Panel C: with linear county trends, post-entry trend (unweighted)

 

Uber

1.211

2.014

0.861

10.22

13.54

6.644

(3.250)

(7.446)

(0.863)

(10.27)

(16.60)

(12.52)

Uber*trend

-0.451

-2.740***

0.393*

-3.344***

-1.220

-1.933*

(0.373)

(0.914)

(0.207)

(1.152)

(0.789)

(1.079)

R-squared

0.688

0.167

0.155

0.068

0.053

0.048

Panel D: with linear county trends, post-entry trend (weighted)

 

Uber

-16.03

-13.38

0.493

-11.42

-1.721

12.05*

(16.68)

(10.81)

(1.663)

(9.040)

(14.16)

(6.882)

Uber*trend

0.253

-1.447

1.588

-4.171***

-1.736

-1.508

(0.846)

(1.080)

(1.000)

(1.231)

(1.580)

(1.070)

R-squared

0.974

0.710

0.897

0.501

0.150

0.489

There are 237,852 observations. All specifications include month-by-year fixed effects, county fixed effects, and county-specific linear trends. We also control for whether marijuana is decriminalized, medicalized, or legalized; the percent of the population who are black, aged 20 to 24, aged 25to 34, aged 35 to 54, and aged 55 and over; indicators for whether the state has a graduated drivers licensing law, zero tolerance law, maximum legal blood alcohol concentration of 0.08; state real per capita personal income, and the state maximum welfare benefit for a family of three. Standard errors are clustered by county. *** p<0.01, ** p<0.05, * p<0.1

25

Table 6: Uber entry and Arrests per 100,000, sample of counties ever hosting Uber:

Differences in differences estimates

(1)

(2)

(3)

(4)

(5)

(6)

 

Motor

 

Aggravated

Other

vehicle

Disorderly

Assault

Assaults

thefts

DUI

Drunk

conduct

Panel A: with linear county trends (unweighted)

 

Uber

-2.570

0.170

1.403

-4.570

0.0274

-1.299

(4.075)

(6.227)

(1.133)

(7.727)

(3.386)

(11.81)

R-squared

0.765

0.850

0.148

0.898

0.939

0.697

Panel B: with linear county trends (weighted)

 

Uber

-24.69

-10.58

5.128***

-20.71*

3.271

5.020

(21.36)

(10.26)

(1.825)

(11.17)

(9.709)

(5.168)

R-squared

0.980

0.852

0.935

0.957

0.952

0.887

Panel C: with linear county trends, post-entry trend (unweighted)

 

Uber

-2.084

2.424

0.811

1.045

0.999

1.792

(4.369)

(6.553)

(1.061)

(6.835)

(3.799)

(11.59)

Uber*trend

-0.279

-1.295

0.340

-3.226**

-0.558

-1.776

(0.374)

(1.033)

(0.217)

(1.404)

(0.766)

(1.122)

R-squared

0.765

0.850

0.148

0.898

0.939

0.697

Panel D: with linear county trends, post-entry trend (weighted)

 

Uber

-24.16

-9.860

2.714***

-14.09

5.414

7.813

(21.71)

(10.62)

(0.973)

(10.10)

(11.37)

(5.485)

Uber*trend

-0.371

-0.511

1.702

-4.666***

-1.511

-1.969*

(0.473)

(1.138)

(1.030)

(1.523)

(1.888)

(1.110)

R-squared

0.980

0.852

0.937

0.957

0.952

0.887

There are 13,316 observations. All specifications include month-by-year fixed effects, county fixed effects, and county-specific linear trends. We also control for whether marijuana is decriminalized, medicalized, or legalized; the percent of the population who are black, aged 20 to 24, aged 25to 34, aged 35 to 54, and aged 55 and over; indicators for whether the state has a graduated drivers licensing law, zero tolerance law, maximum legal blood alcohol concentration of 0.08; state real per capita personal income, and the state maximum welfare benefit for a family of three. Standard errors are clustered by county. *** p<0.01, ** p<0.05, * p<0.1

26

Table 7: Uber entry and Arrests per 100,000, counter-factual crimes: Differences-in-differences estimates

(1)

(2)

(3)

(4)

(5)

(6)

Full sample

Ever Uber sample

Liquor

Liquor

law

Embezzlemen

law

Embezzlemen

 

violations

Forgery

t

violations

Forgery

t

Panel A: with linear county trends (unweighted)

 

Uber

11.22

-4.693

-0.660

8.505

-2.487

-1.249

(14.73)

(4.423)

(1.094)

(7.505)

(3.302)

(1.104)

R-squared

0.039

0.167

0.085

0.441

0.108

0.041

Panel B: with linear county trends (weighted)

 

Uber

-2.020

0.141

0.385

6.658

-0.353

0.0985

(7.378)

(1.157)

(0.350)

(6.066)

(1.511)

(0.328)

R-squared

0.064

0.686

0.239

0.455

0.838

0.227

Panel C: with linear county trends, post-entry trend (unweighted)

 

Uber

24.59

-4.629

-0.671

11.35

-2.560

-1.124

(16.75)

(4.650)

(1.148)

(8.029)

(3.374)

(1.081)

Uber*trend

-2.518

-0.0120

0.00207

-1.635

0.0422

-0.0719**

(1.580)

(0.0905)

(0.0245)

(1.530)

(0.108)

(0.0354)

R-squared

0.039

0.167

0.085

0.441

0.108

0.041

Panel D: with linear county trends, post-entry trend (weighted)

 

Uber

3.635

-0.252

0.398

7.623

-0.425

0.153

(6.740)

(1.345)

(0.369)

(5.994)

(1.561)

(0.344)

Uber*trend

-1.137

0.0791

-0.00248

-0.680

0.0508

-0.0381

(1.731)

(0.109)

(0.0217)

(1.750)

(0.104)

(0.0290)

R-squared

0.064

0.686

0.239

0.455

0.838

0.227

There are 237,852 observations in the full sample and 12,680 in the ever-Uber sample. There are 13,316 observations in the ever-Uber sample. All specifications include month-by-year fixed effects, county fixed effects, and county-specific linear trends. We also control for whether marijuana is decriminalized, medicalized, or legalized; the percent of the population who are black, aged 20 to 24, aged 25to 34, aged 35 to 54, and aged 55 and over; indicators for whether the state has a graduated drivers licensing law, zero tolerance law, maximum legal blood alcohol concentration of 0.08; state real per capita personal income, and the state maximum welfare benefit for a family of three. Standard errors are clustered by county. *** p<0.01, ** p<0.05, * p<0.1

27

Table A1: Areas Served from 2010 through 2014

Area

Fips Code

Area

Fips Code

Area

Fips Code

Auburn, AL Tuscaloosa Anchorage Flagstaff Phoenix Tucson Little Rock Fayetteville, AR Oakland Fresno Bakersfield Los Angeles Monterrey Orange County Palm Springs Sacramento San Bernardino San Diego San Francisco San Luis Obispo Santa Barbara Palo Alto Modesto Ventura Boulder Denver Eagle Fort Collins Pitkin Summit Connecticut Connecticut Athens Atlanta Honolulu Boise Chicago Indianapolis Bloomington South Bend West Lafayette Des Moines Wichita Kansas City, KS Lexington Louisville Baton Rouge Portland, ME Annapolis Baltimore Boston Worcester

1081

Flint Lansing Kalamazoo Grand Rapids Ann Arbor Detroit Minneapolis St. Paul Oxford Columbia, MO Kansas City, MO St Louis Omaha Lincoln Las Vegas Reno Manchester, NH Atlantic County, NJ Bergen County, NJ Burlington County, NJ Camden County, NJ Cape May County, BJ Essex County, NJ Gloucester County, NJ Hudson County, NJ Middlesex County, NJ Monmouth County, NJ Morris County, NJ Ocean County, NJ Union County, NJ Albuquerque Santa Fe Brooklyn Brooklyn Asheville, NC Fayetteville, NC Raleigh-Durham Winston-Salem Greensboro, NC Charlotte Wilmington, NC Raleigh-Durham Raleigh-Durham Cleveland Columbus Cincinnati Toledo Dayton Akron Norman Oklahoma City Tulsa

26049

Eugene Salem Oregon Metro Salem Oregon Metro Pittsburgh Philadelphia Providence Charleston, SC Greenville, SC Myrtle Beach Columbia, SC Columbia, SC Nashville Chattanooga Knoxville Memphis Austin San Antonio College Station Dallas Dallas Dallas El Paso Houston Dallas Lubbock Waco Corpus Christi Amarillo Dallas Austin Austin Salt Lake City Burlington Blacksburg Charlottesville, VA Chesapeake, VA Hampton, VA Newport News, VA Norfolk, VA Richmond, VA Roanoke Virginia Beach Vancouver, WA Seattle Tacoma Spokane Green Bay Madison Milwaukee

41039

1125

26065

41047

2020

26077

41051

4005

26081

41053

4013

26161

41067

4019

26163

42003

5119

27053

42101

5143

27123

44007

6001

28071

45019

6019

29019

45045

6029

29165

45051

6037

29510

45063

6053

31055

45079

6059

31109

47037

6065

32003

47065

6067

32031

47093

6071

33011

47157

6073

34001

48015

6075

34003

48029

6079

34005

48041

6083

34007

48085

6085

34009

48113

6099

34013

48121

6111

34015

48141

8013

34017

48201

8031

34023

48257

8037

34025

48303

8069

34027

48309

8097

34029

48355

8117

34039

48381

9001

35001

48397

9009

35049

48453

13059

36009

48491

13121

36025

49035

15003

37021

50007

16001

37051

51121

17031

37063

51540

18097

37067

36103

18105

37081

51650

18141

37119

51700

18157

37129

51710

19153

37135

51760

20173

37183

51770

20209

39035

51810

21067

39049

53011

21111

39061

53033

22033

39095

53053

23005

39113

53063

24003

39153

55009

24510

40027

55025

25025

40109

55131

25027

40113

 

28

Table A2: Uber entry and fatal traffic crashes per 100,000, sub-sample of Non-California counties

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

 

Alcohol-

Alcohol-

involved

Night-

involved

Night-

 

Fatal

fatal

time fatal

Vehicular

Fatal

fatal

time fatal

Vehicular

Crashes

crashes

crashes

Fatalities

Crashes

crashes

crashes

Fatalities

 

unweighted

 

population weighted

 

Panel A: Differences-in-differences with linear county trends

 

Uber

-0.305**

-0.0285

-0.100* -0.257

-0.0697

-0.00821

-0.0105

-0.0468

(0.119)

(0.0263)

(0.0598)

(0.193)

(0.0511)

(0.00796)

(0.0233)

(0.103)

R-squared

0.080

0.033

0.038

0.056

0.125

0.072

0.060

0.070

Panel B: Differences-in-differences with linear county trends, post-entry trend

 

Uber

-0.181

-0.0109

-0.0477

-0.149

-0.0108

-0.00626

0.00320

0.0371

(0.111)

(0.0234)

(0.0539)

(0.162)

(0.0488)

(0.00692)

(0.0213)

(0.0832)

Uber*trend

-0.0248**

-0.00350

-0.0104**

-0.0215

-0.0133***

-0.000442

-0.00309*

-0.0190**

(0.0119)

(0.00214)

(0.00491)

(0.0189)

(0.00380)

(0.000531)

(0.00176)

(0.00805)

R-squared

0.080

0.033

0.038

0.056

0.125

0.072

0.060

0.070

There are 284,648 observations. All specifications include month-by-year fixed effects, county fixed effects, and county-specific linear trends. We also control for whether marijuana is decriminalized, medicalized, or legalized; the percent of the population who are black, aged 20 to 24, aged 25to 34, aged 35 to 54, and aged 55 and over; indicators for whether the state has a graduated drivers licensing law, zero tolerance law, maximum legal blood alcohol concentration of 0.08; state real per capita personal income, and the state maximum welfare benefit for a family of three. Standard errors are clustered by county. *** p<0.01, ** p<0.05, * p<0.1

29

Table A3: Uber entry and Arrests per 100,000: Differences-in-differences estimates in non-California counties

(1)

(2)

(3)

(4)

(5)

(6)

 

Motor

 

Aggravated

Other

vehicle

Disorderly

Assault

Assaults

thefts

DUI

Drunk

conduct

Panel A: with linear county trends (unweighted)

 

Uber

-1.957

-14.14**

1.677***

-5.629

6.536

-4.467

(2.303)

(7.024)

(0.640)

(10.78)

(15.98)

(12.84)

R-squared

0.565

0.166

0.113

0.067

0.053

0.048

Panel B: with linear county trends (weighted)

 

Uber

-2.714

-24.52**

2.991**

-20.98**

-17.69

5.050

(4.245)

(10.46)

(1.389)

(9.729)

(11.79)

(8.248)

R-squared

0.833

0.687

0.525

0.379

0.133

0.485

Panel C: with linear county trends, post-entry trend (unweighted)

 

Uber

2.280

4.245

0.736

13.93

15.21

8.756

(3.246)

(8.202)

(0.817)

(11.15)

(17.93)

(13.93)

Uber*trend

-0.812*

-3.523***

0.180

-3.747***

-1.663*

-2.534*

(0.435)

(1.081)

(0.114)

(1.383)

(0.855)

(1.352)

R-squared

0.565

0.166

0.113

0.067

0.053

0.048

Panel D: with linear county trends, post-entry trend (weighted)

 

Uber

3.710

-6.130

1.285

3.685

-9.267

20.94**

(4.550)

(11.75)

(0.815)

(7.870)

(13.58)

(9.090)

Uber*trend

-1.230**

-3.521**

0.327**

-4.723***

-1.613

-3.043*

(0.536)

(1.536)

(0.152)

(1.567)

(1.527)

(1.664)

R-squared

0.833

0.687

0.525

0.379

0.133

0.486

There are 236,700 observations. All specifications include month-by-year fixed effects, county fixed effects, and county-specific linear trends. We also control for whether marijuana is decriminalized, medicalized, or legalized; the percent of the population who are black, aged 20 to 24, aged 25to 34, aged 35 to 54, and aged 55 and over; indicators for whether the state has a graduated drivers licensing law, zero tolerance law, maximum legal blood alcohol concentration of 0.08; state real per capita personal income, and the state maximum welfare benefit for a family of three. Standard errors are clustered by county. *** p<0.01, ** p<0.05, * p<0.1

30