Loading...
Please wait, while we are loading the content...
Similar Documents
When the Field Is Not Enough: Using Complementary Field and Lab Experiments to Investigate Congestion Pricing Responses
| Content Provider | Semantic Scholar |
|---|---|
| Author | Andersen, Steffen Dixit, Vinayak V. Martínez-Correa, Jimmy Rutström, E. Elisabet |
| Copyright Year | 2014 |
| Abstract | We use data from a field experiment collecting route choices using GPS recorders and a lab experiment eliciting risk attitude characteristics of drivers. We find that drivers behave in predicted ways in response to variations in average travel times, and in response to manipulations of route pricing. Demand functions are downward sloping in tolls and in travel times. We do not find that drivers respond to variations in the unreliability of travel times, but this is not due to their risk attitudes. Almost all drivers are risk averse, and express a revealed preference in favor of the relatively safe route that is increasing in risk aversion. Because the alternate routes are located at some distance apart, we suspect that the lack of response to travel time unreliability is to be found in biases in drivers expectations of unreliability. Authors are listed in alphabetical order. Professor, Copenhagen Business School, Department of Economics, Denmark sa.eco@cbs.dk, (Andersen), Senior Lecturer, Research Centre for Integrated Transport Innovation (rCITI), School of Civil and Environmental Engineering, University of New South Wales, Australia, v.dixit@unsw.edu.au, (Dixit); Assistant Professor in Applied Microeconomics, Copenhagen Business School, Department of Economics, Denmark, jima.eco@cbs.dk, (Martínez-Correa); Professor and Director, Dean's Behavioral Economics Laboratory, J. Mack Robinson College of Business and Andrew Young School of Policy Studies, Georgia State University, USA, erutstrom@gsu.edu (Rutström). All online materials available at http://dbel.robinson.gsu.edu/researchprojects-fhwa/. This material is based upon work supported by the Federal Highway Administration under Agreement No. DTFH61-09-H-00012. Any opinions, findings and conclusions or recommendations expressed in this publication are those of the Author(s) and do not necessarily reflect the view of the Federal Highway Administration. We are grateful to Dr. Karen White, Economist in the Office of Transportation Policy Studies at the Federal Highway Administration for guidance and feedback on this research. We are grateful to our other collaborators on the research program including Dr. Glenn W. Harrison, Georgia State University, Dr. Steffen Andersen and Dr. Morten Lau, Copenhagen Business School, and Dr. Essam Radwan, University of Central Florida, for invaluable contributions to the research program. All errors in the paper are those of the authors. Introduction Traffic congestion is a rapidly growing concern in metropolitan areas worldwide. According to the 2012 Urban Mobility Report (Schrank, Eisele, and Lomax (2012)) in the US alone the estimated cost of excess travel time and wasted fuel consumption was $121 billion in 2011. The average annual delay per auto has increased from 16 hours in 1982 to 38 in 2011. About 2.9 billion gallons of fuel were wasted due to clogged highways, and the drivers experienced 5.5 billion hours of traffic delays. The report puts these costs in perspective and equates them to enough time for five vacation days, and the equivalent amount that Americans spend online shopping every year. As the volume of traffic increases, not only do average travel times increase but there is also an exponential increase in the variability of travel times, adding a great deal of uncertainty to travel and transportation services. With additional congestion the number of incidents will increase, and the delay caused by any one incident will be longer. Important determinants to the welfare effects of congestion and congestion pricing are the reactions of the individual drivers. In this paper we provide evidence on how drivers respond to congestion and congestion pricing. We estimate responses and decision models based on data from complementary field and lab experiments. The field experiments are set in participants’ natural traffic environment: common commuter corridors in two major cities: Orlando and Atlanta. Congestion conditions are therefore natural, generated by the traffic system that participants are familiar with. We measure these congestion conditions using GPS recorders in the participants’ cars and are able to categorize travel routes by six timesofday intervals (3 during morning commutes and 3 during afternoon commutes), for each weekday, during 3 different time periods in 2011-2012. There is considerable variation in travel time across time, and we add to these natural variations experimental conditions that vary the monetary costs of choosing various routes. By basing our analysis on revealed preferences over route choices, using GPS recorders and real monetary route prices, we avoid the hypothetical biases that can be the result when using stated preference techniques. We do not find any significant influence from the unreliability of travel time on the route choices. However, this cannot be taken as evidence that our drivers are risk neutral since in the field data inferences of such preferences may be confounded by biases in the perceptions of travel time variations. Dixit, Harrison and Rutstrom (2013) demonstrate the importance of both risk perceptions and risk attitudes to choices made in a driving simulator experiment when there is an accident risk. In order to separate perceptions from preferences, we measure risk preferences in a lab task for the same participants, and find that these preferences correlate with route choice behavior in expected ways. We also find evidence that drivers reveal prefer the Expressway when it is faster on average, and when it has a lower relative price. 1 Fifer, Rose and Greaves (2014) demonstrate significant hypothetical bias in stated choice data of responses to various driving charges. Experimental Design We collect data from a field driving task and from a stylized lab task involving lottery choices from the same subjects. We include two commuter corridors in Orlando, one on the east and the other on the west side of downtown, and two in Atlanta, one on the northeast and the other on the northwest side of downtown. In the field driving part of the experiment the participants are paid to take certain trips, but they have a limited set of route and departure time choices. We only study drives during rush hours Monday – Friday 7 am – 9:30 am in the direction towards downtown and 4 – 8 pm in the opposite direction. Thus, trips that start before 7 am or 4 pm, or start after 9:30 am or 8 pm are not observed. We also include Saturdays 2 – 8 pm but those drives can be in either direction. Each route is approximately five miles long. Slight variations in the exact length of the route depend on availability of intersections and on-off ramps. These restrictions in time and location serve to limit the variability in traffic conditions and trip purposes so that drives are more comparable. The five mile segments were selected because they were highly congested, and because it was more likely that the recipients of our invitations frequently used these segments in full. This ensured that our study drivers were highly familiar with the routes and their congestion conditions. If we had used longer segments it would have increased the likelihood that many drivers would need to enter and exit the segments in their interiors, thus not making the routes comparable across drivers. To be observed, a drive must involve continuous travel on either of the two routes during the valid times. If the participant stopped or deviated in any way the drive would not be included as an observation. These segments also allowed us to present arterial and expressway route options that were reasonably close to each other and parallel so that they could serve as substitutes for the same trip purpose and be similar in access cost. If the two routes were too far apart, the distance to the start of one would be longer than the distance to the start of the other, and access cost on the former would be higher. For participants on the west side of Orlando participant drivers can choose to take either SR408, the expressway, or SR50, the arterial. We observe route choices between Good Homes Road and John Young Parkway during these times. Figure 1 shows a map of these routes. The expressway has 2 or 3 lanes in either direction with a speed limit of 55 mph, whereas the local road has 2 or 3 lanes with a speed limit of 45 mph and 12 traffic signals. The two routes are 0.3 and 0.8 miles apart at the end points. Once the route is selected the participant cannot deviate from that route, so if the drive starts on SR408 in the morning rush hour traffic at Good Homes Road going towards downtown, a valid study drive must involve staying on SR408 the entire distance to John Young Parkway. Participants are not restricted to stay on the same route each time they travel but can vary their choice freely. Drivers on the east side of Orlando travel either on SR408 or SR50 between Goldenrod and Mills Avenue where again SR408 is a tolled express way and SR50 is a local road. See Figure 1b for a map showing these routes. On the east side of Orlando, SR408 has 4 or 5 lanes in each direction with a speed limit of 55 mph and SR50 has 2 or 3 lanes in each direction, a speed limit of 45 mph and 15 traffic signals. The two routes are 1.2 and 2 miles apart at the end points. Drivers in northeast Atlanta travel either on the expressway SR400 or on the local road SR9. See Figure 1c for a map of these routes. SR400 has 4 lanes in each direction and a speed limit of 55 mph, and SR9 has 2 lanes in each direction plus a middle turning lane, a speed limit of 35 – 45 mph and 15 traffic signals. The two routes are 0.5 and 1.3 miles apart at the end points. Finally, drivers in north-west Atlanta travel either on the expressway I75 or on the local road SR41. See Figure 1d for a map of these routes. I75 has 5-7 lanes in each direction and a speed limit of 55 mph whereas SR41 has 2 lanes in each direction plus a center turning lane, a speed limit of 35 – 45 mph |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.rciti.unsw.edu.au/sites/rciti/files/u48/RiskTransport/1_Elisabet%20Rutstrom%20-%20When%20the%20Field%20is%20Not%20Enough.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
