Phd Dissertation

Environmental and health benefits of airport congestion pricing - The case of Los Angeles International Airport

Abstract

Airports are a source of greenhouse gases (GHG) and air pollutants such as fine particulate matter with an aerodynamic diameter under 2.5 μm (PM2.5), which adversely affect the climate and human health. This pollution is worsening with increasing aircraft congestion. Even though aviation is the second largest source of GHG emissions in the transportation sector, it was excluded from the recent COP21 Paris Agreement. Little is known about the climate change and adverse health impacts from increasing airports congestion. The purpose of this study is to start filling this gap.

In this dissertation, I estimate congestion, health, and climate benefits from airport congestion pricing for Los Angeles International Airport (LAX), the fourth busiest airport in the world by passenger numbers in 2018. I first derive the optimal congestion fee for airports like LAX that primarily serve local and regional markets. To quantify the impacts of airport congestion pricing, I analyze one year of airport operations (2014), which corresponds to 593,547 flights (both inbound and outbound). My simulation results suggest that hourly congestion pricing would on average reduce waiting time by 2.9 minutes per flight and annual PM2.5 emissions by 11.4 percent, thus decreasing the environmental impacts from aircraft landing and takeoff operations (LTO), which extend as far as 19 km downwind from the airport.

An analysis of the health gains from implementing a congestion fee that accounts for air pollution cost shows that it would annually reduce premature mortality from PM2.5 exposure by 4.6 cases, avoided hospital admissions for cardiovascular diseases by 167 cases, and avoid 8,539 lost work days. The corresponding monetary value of these health gains are $45.8 million, $21.9 million, and $1.4 million respectively (all in 2014 dollars).

For my climate change analysis, I consider both the country-level social cost of carbon (CSCC; $36 per tonne) and the global social cost of carbon (GSCC; $417 per tonne). While pricing GHG emissions with the CSCC only has a minor impact, using the GSCC helps further reduce aircraft congestion and its associated health impacts. Indeed, an aircraft congestion fee with GHG based on the GSCC would reduce premature mortality by 6 cases each year, avoided hospital admissions by 221 cases, and avoid 11,528 lost work days (95 % CI: 4,995, 18,060). The corresponding monetary value of these health gains are $60.7 million, $27.7 million, and $1.9 million respectively.

The methodology presented in this study is widely applicable. It provides engineers, planners, and policymakers a tool for reducing airport congestion and for quantifying the resulting health and climate benefits.

Phd Dissertation

Time-varying networks measurement, modeling, and computation

Abstract

Time-varying networks and techniques developed to study them have been used to analyze dynamic systems in social, computational, biological, and other contexts. Significant progress has been made in this area in recent years, resulting from a combination of statistical advances and improved computational resources, giving rise to a range of new research questions. This thesis addresses problems related to three lines of inquiry involving dynamic networks: data collection designs; the conditions needed for structural stability of an evolving network; and the computational scalability of statistical models for network dynamics. The first contribution involves a commonly neglected problem concerning data collection protocols for dynamic network data: the impact of in-design missingness. A systematic formalization is offered for the widely used class of retrospective life history designs, and it is shown that design parameters have nontrivial effects on both the quantity of missingness and the impact of such missingness on network modeling and reconstruction. Using a simulation study, we also show how the consequences of design parameters for inference vary as a function of look-back time relative to the time of measurement. The second contribution of this thesis is related to a fundamental question of network dynamics: when or where are changes in a network most likely to occur? A novel approach is taken to this question, by exploring its complement — what factors stabilize a network (or subgraphs thereof) and make it resistant to change? For networks whose behavior can be parameterized in exponential family form, a formal characterization of the graph-stabilizing region of the parameter space is shown to correspond to a convex polytope in the parameter space. A related construction can be used to find subgraphs that are or are not stable with respect to a given parameter vector, and to identify edge variables that are most vulnerable to perturbation. Finally, the third contribution of this thesis is to scalable parameter estimation for a class of temporal exponential family random graph models (TERGM) from sampled data. An algorithm is proposed that allows accurate approximation of maximum likelihood estimates for certain classes of TERGMs from egocentrically sampled retrospective life history data, without requiring simulation of the underlying network (a major bottleneck when the network size is large). Estimation time for this algorithm scales with the data size, and not with the size of the network, allowing it to be employed on very large populations. 

Phd Dissertation

Essays on Transportation Externalities

Publication Date

June 29, 2019

Author(s)

Abstract

This dissertation concerns the measurement and regulation of externalities with a focus on the numerous and interrelated external costs in the transportation sector. The research touches on pollution, congestion, and collisions as well as various modes of transportation. The results have important implications for public policy and the regulation of externalities both within and outside of transportation. Chapter 1 investigates the effectiveness of traffic laws which require drivers to provide a minimum amount of distance between their vehicle and cyclists when passing them on roadways in improving cyclist safety. Many believe these laws are ineffective in reducing the number of bicyclist fatalities because they are difficult for police to enforce, contain loopholes, and the minimum distance required is inadequate. This chapter tests this claim empirically using data on 18,534 bicyclist fatalities from the Fatality Analysis Reporting System and a differences-in-differences approach, in a negative binomial model, to identify the effect of minimum distance passing laws on bicyclist fatalities. The analysis fails to find a significant effect of enacting a minimum distance passing law on the number of cyclist fatalities after controlling for differences in weather, demographics, bicycling commuter rates, state level traffic, and time variation. Chapter 2 examines the effects of freight truck weight and miles traveled on both the quantity and severity of truck-involved collisions using a unique panel data set covering the universe of truck-involved collisions and 3.5 billion truck-weight observations. Estimates reveal that, though both measures of trucking activity can increase collisions, increases in truck weight skew the distribution of collisions towards more severe outcomes involving either injury or death. The results are applied to a welfare analysis of diesel fuel taxes, which have been shown to both reduce truck miles traveled and increase truck cargo weight. Though diesel taxes are shown to slightly reduce the total number of collisions, the remaining collisions become more severe. Societal gains from pollution, congestion, and collision reductions are entirely offset by the increased fatal collision costs, reducing total welfare by $39.9 billion annually. The final chapter examines the regulation of heterogeneous externalities. When demand for or damages from an externality producing good vary, uniform policy instruments are an inefficient tool for correcting market failures. This chapter examines how a policy that reflects this heterogeneity can improve efficiency. County travel demand elasticities and congestion damages are estimated to compare the efficiency of a uniform fuel tax to county-specific fuel taxes. Because elasticities, congestion damages, and pollution damages exhibit heterogeneity across regions, county-specific fuel taxes, levied in a subset of major metropolitan areas, provide welfare gains between $7-$26 per capita annually in addition to equity gains relative to a revenue neutral uniform fuel tax.

Phd Dissertation

Control Theoretic Approaches to Congestion Pricing for High-occupancy Toll Lanes

Abstract

The purpose of this study is to propose control theoretic approaches for high-occupancy toll (HOT) lanes operation. This dissertation considers different operation objectives, and provides pricing schemes for HOT lanes accordingly.

To improve the system performance, the study first proposes a simultaneous estimation and control method for the same system as that in (Yin and Lou, 2009). An integral controller is applied to estimate the average value of time (VOT) of SOVs, and the dynamic prices are calculated based on the logit model. The closed-loop system is proved to be stable and guaranteed to converge to the optimal state both analytically and numerically. Two convergence patterns, Gaussian or exponential, are revealed. The effect of the scale parameter in the logit model is also examined.

Then, a new lane choice model, i.e., the vehicle-based user equilibrium principle, is proposed to capture the lane choice of SOVs. A general lane choice model is derived based on the characteristics of the logit and the vehicle-based UE model. An insight regarding the dynamic price is obtained by analytically solving the optimal dynamic prices with constant demands of HOVs and SOVs, and then a feedback controller is designed to determine the dynamic prices without knowing SOVs’ lane choice models, but to satisfy the two control objectives: maximizing the flow-rate but not forming a queue on the HOT lanes. If the type of the lane choice model is given, the distribution of VOTs of the SOVs can be estimated.

Next, an optimal control problem is proposed to examine the statement that revenue maximization should generally coincide with the optimization of freeway performances, such as maximizing overall travel-time savings or throughput. Results show that operators need to make different strategies based on the traffic demand. In order to maximize the revenue, operators should set a higher price to make the HOT lanes underutilized if the demand of HOVs is low. However, if the demand of HOVs is high, operators need to set a lower price to attract more SOVs to create congestion on the HOT lanes.

It has long been known that drivers’ departure time choice behavior is one fundamental cause of congestion. In the last part of this dissertation, pricing schemes are proposed to consider both lane choice and departure time choice. In the study period, the demands for the HOT and GP lanes are higher than their capacities, which means the whole freeway is congested. However, the congestion period on the HOT lanes is short than that on the GP lanes. So, the HOT lanes are “underutilized”. It turns out that flat (instead of dynamic) pricing schemes are able to meet the following two constraints: (1) the total travel time and scheduling cost is minimized; and (2) the costs for each non-switching and switching SOV are the same. We show that different revenue and tolling constrains for certain type of vehicles lead to different pricing schemes.

Phd Dissertation

Commute Mode Choice, Parking Policies, and Social Influence

Publication Date

May 31, 2019

Author(s)

Abstract

This dissertation examines the impact of parking policies and social influence on commute mode choice using discrete choice analysis. A key feature of the dissertation is overcoming the problem of insufficient data by using unique datasets, building unique datasets, or exploring appropriate estimation strategies and assumptions.

Chapter 1 studies the impact of parking prices on the decision to drive to work using the California Household Travel Survey. The chapter tackles estimation challenges posed by insufficient parking information. The first challenge is the estimation of parking prices for those who do not drive, which is addressed by using a sample selection model. The second challenge is to understand the effect of the extent of the prevalence of Employer-Paid parking coupled with incentive programs offered in-lieu of parking. To address this challenge, two extreme scenarios are examined, and a range for the marginal effects of parking prices is estimated; one scenario assumes everyone receives Employer-Paid parking coupled with in-lieu of parking incentives, and the second assumes that no one is offered such incentives. The results suggest that higher parking prices reduce driving, regardless of the followed approach. It is estimated that a 10% increase in parking prices leads to a 1 – 2 percentage point decline in the probability of driving to work. Moreover, there seems to be no evidence of sample selection bias. The evidence suggests that parking pricing can indeed be an effective transportation demand management tool.

Chapter 2 extends the analysis of Chapter 1 to simultaneously estimate the impact of parking pricing, parking availability, and urban form on commute mode choice. The joint role of these three factors is examined using a dataset that is constructed by merging three major different data sources. The California Household Travel Survey data are matched to two unique datasets on parking for Los Angeles County; one for prices and the other availability. Chapter 2 first examines how these three factors affect the binary decision of whether to drive, while controlling for a rich set of covariates. The analysis then becomes more specific and examines how these factors affect particular commute modes in a multinomial context. The results indicate that parking prices have a significant negative impact on the decision to drive to work, where a 10% increase in parking prices is associated with a 1.1% drop in the probability of driving to work. Both on-street and off-street parking availability at home, as well as urban form measures of the workplace tract, are found to significantly affect commute mode choices. These findings have important policy implications in terms of minimum parking requirements, maximum parking standards, employer-paid parking, and parking pricing policies.

Chapter 3, on the other hand, examines the impact of a number of fundamental determinants of commute mode choice on transit use, and introduces the role of social influence. The determinants explored cover socioeconomic characteristics, built environment and neighborhood characteristics, transit accessibility, and trip characteristics. Social interactions have been found to affect many of the decisions of economic agents, and are likely to play a role in the decision to use transit. A unique dataset is built to conduct this analysis across a number of major US cities and examine the effects in both the residence and workplace neighborhoods, where a neighborhood is defined as a census tract. Social influence is explored along three different dimensions: space (neighborhood), income, and race. A novel instrumental variable is constructed in order to identify spatial social influence, and an alternative identification strategy is devised to identify income-group and racial social influence. The evidence suggests that spatial social influence exists among both coworkers and residential neighbors, and that peer effects among coworkers are larger than those among residential neighbors. Moreover, income-group social influence, among both coworkers and residential neighbors, plays a significant role in the rich commuter’s decision to use transit. However, racial social influence does not affect a commuter’s decision to use transit, regardless of race.

policy brief

Compact, Accessible, and Walkable Communities Help Support Gender Equality

Abstract

In California, Senate Bill 375 mandates regional planning organizations align their transportation plans with sustainable land use and development strategies to achieve reductions in greenhouse gas emissions. In response, the Southern California Association of Governments’ 2016 Regional Transportation Plan/Sustainable Community Strategy directs nearly 50% of housing and employment growth between 2010 and 2040 into walkable and compact neighborhoods within a one-half mile walking distance from well-serviced transit stops. This approach to land use development can encourage shorter driving trips, greater transit usage, and increased walking and cycling as a result of daily activity destinations being clustered near residential and work locations.1Another bi-product and benefit of compact and accessible communities may be improving gender equality related to travel and activity patterns. Prior research shows segregated and dispersed land uses (i.e., suburban sprawl) can exacerbate gender disparities in daily household travel by separating the public and private realms, and can also constrain women to their immediate neighborhoods.2,3 In contrast, neighborhoods with pedestrian accessible mixes-use centers have been shown to help counter social isolation of women in suburbia.4In addition, compact communities with denser land use and better transit service has been shown to reduce the disproportionate amount of chauffeuring women conduct on behalf of the household.

published journal article

Integrating Autonomous Vehicles in Multimodal Peer-to-peer Shared Mobility Systems and its Network Impacts

Abstract

As public perception of sharing economy in transportation has changed, mobilephone-hailed ridesharing is gaining prominence. The key aspect of capitalizing and promoting better shared-mobility systems depends on the matching rate between the supply and demand for rides. Peer-to-peer (P2P) ridesharing systems devise higher matching rate than pure ridesharing systems by attracting more drivers. Even relaxing the spatiotemporal constraints for participants could increase the chances to be matched. However, we notice that sole P2P ridesharing systems still do not guarantee matching when the number of drivers is limited. We propose the utilization of a fleet service to cover the unmatched riders in P2P ridesharing. While it can be any type of fleet services such as taxis, Uber/Lyft, or paratransit, we explore the idea of utilizing shared autonomous vehicles as a fleet, as they can be dispatched without labor. We model an integrated system for P2P ridesharing and shared autonomous fleet vehicles (SAFVs). The proposed algorithm is designed to maximize matching ratio while optimizing the number of required SAFVs. Based on a simulated study on the northern Los Angeles, the integrated shared-mobility system is shown to have high potential to serve a high fraction of riders.

policy brief

Transit Investments are Having an Impact on Land Use Beyond the Half-Mile Mark

Abstract

Recent years have witnessed a growing interest in transit-
oriented development (TOD) and other transit-centered
initiatives. It has been widely presumed that transit investment
can significantly contribute to curbing sprawl and creating
a more compact (and thus more sustainable) pattern of
urban land use, while providing a broader range of travel
options. However, little is known about how investments in
the public transit system modify urban land use patterns and
the geographical extent of impacts. Prior research tends to
assume transit lines and stations are homogeneous and have
similar impacts without careful consideration of development
history, service quality, or other variations. In addition, prior
research and current practice often assume transit impacts
are concentrated within a half-mile, which has limited the
understanding of how transit investments impact the broader
vicinity.

Phd Dissertation

Real Options Models for Better Investment Decisions in Road Infrastructure under Demand Uncertainty

Abstract

An efficient transportation system requires adequate and well-maintained infrastructure to relieve congestion, reduce accidents, and promote economic competitiveness. However, there is a growing gap between public financial commitments and the cost of maintaining, let alone expanding the U.S. road transportation infrastructure. Moreover, the tools used to evaluate transportation infrastructure investments are typically deterministic and rely on present value calculations, even though it is well-known that this approach is likely to result in sub-optimal decisions in the presence of uncertainty, which is pervasive in transportation infrastructure decisions. In this context, the purpose of this dissertation is to propose a framework based on real options and advanced numerical methods to make better road infrastructure decisions in the presence of demand uncertainty. I first develop a real options framework to find the optimal investment timing, endogenous toll rate, and road capacity of a private inter-city highway under demand uncertainty. Traffic congestion is represented by a BPR function, competition with an existing road is captured by user equilibrium, and travel demand between the two cities follows a geometric Brownian motion with a reflecting upper barrier. I derive semi-analytical solutions for the investment threshold, the dynamic toll rates and the optimum capacity. The result shows the importance of modeling congestion and an upper demand barrier — features that are missing from previous studies. I then extend this real options framework to study two additional ways of funding an inter-city highway project: with public funds or via a Public-Private Partnership (PPP). Using Monte Carlo simulation, I investigate the value of a non-compete clause for both a local government and for private firms involved in the PPP. Since road infrastructure investments are rarely made in isolation, I also extend my real options framework to the multi-period Continuous Network Design Problem (CNDP), to analyze the investment timing and capacity of multiple links under demand uncertainty. No algorithm is currently available to solve the multi-period CNDP under uncertainty in a reasonable time. I propose and test a new algorithm called “Approximate Least Square Monte Carlo simulation” that dramatically reduces the computing time to solve the CNDP while generating accurate solutions