Abstract

This dissertation introduces a novel model intended for integration within an Agent-Based Model (ABM) framework to dynamically estimate and predict workers’ commuting behaviors under various policy scenarios. The model is designed to aid policy-making by providing insight into commuting patterns and their potential responsiveness to policy interventions. In particular, the focus is on changes in Working from Home behavior due to the COVID-19 pandemic. The methodology encompasses a three-step process, starting with the identification of worker commuting preference classes. Employing an unconditional latent class analysis model, the study categorizes workers into distinct groups based on their telecommuting preferences and behaviors. This classification is foundational for understanding diverse work-related travel patterns. The second step is predicting class membership. Post-classification, the study considers demographic features to determine their impact on class membership. This analysis is critical for predicting shifts in commuting behavior in relation to demographic changes. Third, estimating commuter type within each commuter type class: This concluding step uses logistic regression to estimate the likelihood of an individual being a commuter, a hybrid commuter, or a telecommuter, with adaptability to policy changes for exploring varied outcomes. The study produced several key findings. First, diverse worker classes were identified: The analysis of the ASU Covid Future Panel Survey data revealed several distinct worker classes based on telecommuting experiences and preferences. These include a telecommuter class, a regular commuter class, pre-Covid home remote worker class, and a class exhibiting significant demographic changes during the pandemic. Particularly noteworthy is a class that shows a strong propensity to shift to high-frequency telecommuting under supportive policies, despite an initial preference for hybrid or regular commuting. Distinct class characteristics and predictors were identified within each class, serving as predictors for class membership. This finding is essential for understanding and predicting changes in commuting behaviors. The study also included an intra-class commuter type estimation and factor analysis to identify the factors influencing these classifications. This provides deeper insights into the motivations and constraints affecting commuting choices. 

Abstract

An activity-travel chain is a series of consecutive trips to multiple destinations. By influencing activity decisions (e.g., activity location, duration, and start time) and travel decisions (e.g., trip mode, route, and departure time), activity-travel chaining can directly impact roadway congestion, vehicle miles traveled by mode, transit ridership, energy consumption, and emissions of harmful pollutants. In this context, my dissertation uses the 2017 National Household Travel Survey (NHTS) and 2018-2019 Household Travel Survey from four Metropolitan Planning Organizations (MPOs) to (i) identify distinct activity-travel chain types, (ii) quantify the effect of activity-travel chaining propensity on peak and off-peak person-miles traveled (PMT), and (iii) explore how activity-travel chain makers use emerging transportation modes (i.e., ride-hail). To perform these three analyses, I employ several statistical modeling techniques, including Latent Class Analysis (LCA), multi-level Poisson regression, structural equation modeling, and logistic regression. In Chapter 3, I identify four distinct types of activity-travel chains. The most representative type involves simple car-based activity-travel chains with short-duration stops, typically for maintenance activities. The classification also reveals one group that exclusively represents non-motorized transport (NMT)- and transit-based activity-travel chains. In addition to identifying distinct activity-travel chains, I also model the propensity of travelers to conduct each type of activity-travel chain. I find that travelers in households with children and older travelers more frequently make car-based activity-travel chains for maintenance activities. Moreover, travelers in single-member households, and travelers who are younger and male more frequently make NMT- and transit-based activity-travel chains for maintenance activities. I expect the identification of these distinct activity-travel chain types, and the models of propensity of travelers to perform each activity-travel chain type, to be useful in agent- and activity-based travel forecasting modeling frameworks. In Chapter 4, I investigate the structural relationship between activity-travel chaining propensity and motorized person-miles traveled (PMT) during the peak and off-peak periods of the day. Moreover, I differentiate between workers and non-workers. Using structural equation modeling techniques, and mediating factors I find that chaining of maintenance and discretionary activities increases peak motorized PMT for workers and non-workers, providing the strongest evidence in the literature that activity-travel chaining can exacerbate traffic congestion during peak travel periods. Moreover, I find possible substitution of maintenance activities (e.g., shopping, dining, etc.) in peak-hour with same/similar chained activities in off-peak hour. Finally, in Chapter 5, I analyze activity-travel chain mode choice and show that young persons, frequent transit users, and those having long-duration stops prefer ride-hailing over car. Also, activity-travel chain makers headed to healthcare and social/recreational activities have a particularly high tendency to use ride-hail. Understanding the use of ride-hailing in activity-travel chains should help in formulating policies to better align ride-hailing services with compatible activity-travel patterns and consequently improve accessibility and mobility. 

Abstract

Apart from the COVID-19 pandemic, two chronic problems affecting Californians are high housing costs and road congestion. Although high housing costs and the determinants of commuting have separately received a lot of attention from academic researchers, to my knowledge very few papers have analyzed the linkage between them. In this dissertation, I present three essays that will enhance our understanding on the relationship between commuting, land use, housing costs, and the impact of COVID-19 on telecommuting. In all three essays, I use Structural Equation Model (SEM). In my first essay, I propose a framework for understanding the impact of housing costs on commuting time and commuting distance in one worker-households in Los Angeles County, which is the most populous county in the US. After analyzing data from the 2012 California Household Travel Survey (CHTS), I find that households who can afford more expensive neighborhoods have on average a commute 3.1% shorter per additional $100k to their residence median home values. In my second essay, I analyze the commutes of two-worker households to understand some of the trade-offs they need to make, since two-worker households have dual work constraints. My data for this essay come from 2017 National Household Travel Survey (NHTS) respondents who reside in five U.S. MSAs (San Francisco, Los Angeles, Dallas, Houston, and Atlanta). Results show that women do not commute as far as men on average, although their commuting time is not necessarily shorter than men’s, and that the commuting times of men and women are weakly positively correlated. Moreover, households have faster commutes by 14.5% for men and 22.7% for women per additional $1000 to their residence median monthly housing cost. My third essay investigates the impact of the COVID-19 pandemic on telecommuting by analyzing a unique dataset collected at the end of May 2021 by IPSOS via a random survey of California members of KnowledgePanel®. I find that an additional 4.2% of California workers would engage in some level of telecommuting and more educated workers are expecting to telecommute more (0.383* for bachelor’s degree) post-pandemic. Teasing out the impact of high housing costs on commuting is important at a time when concerns about the environmental impacts of transportation have turned reducing vehicle-miles traveled (VMT) into a policy priority. More generally, a better understanding of the determinants of commuting is critical to inform housing and transportation policy, improve the health of commuters, reduce air pollution, and achieve climate goals.

Abstract

For several decades now, matching mechanisms have been deployed, sometimes implicitly, to solve economic and social problems with unprecedented efficiency. Most notably, the kidney transplant matching algorithm has been key in saving many lives, 39,000 donation in 2019 alone. Likewise, the National Residency Matching Program successfully employed in the United States today uses a matching mechanism that places medical students into hospital residencies. A similar mechanism is used in college and public school admissions around the United States, most famously, in the Boston and New York City. This dissertation will continue this long history of applying matching mechanisms towards efficiently solving social problems.

We begin with a summary of relevant definitions and other terminology in chapter1, all of which will be applied in the chapters that follow. Chapter 2 will examine two concurrent social problems, over-production of resources that contributes to global waste, and lack of access to wasted resources by people living on the economic margin. We will contrast two possible solutions to both problems, that is, a decentralized vs a centralized matching solution. The feasibility of both solutions will be tested through theoretical investigation and two qualitative case studies in food surplus redistribution in the United Kingdom and allocation of housing to unhoused household in Los Angeles County during the pandemic.

Chapter 3 will give a more detail examination of allocation of housing to the unhoused by examining the efficiency and robustness to manipulation of the algorithm that was employed by LA county vs a centralized matching mechanism. Whereas Chapter 4 will explore an online matching mechanism solution to the problem of traffic congestion pricing. The proposed solution combines a matching algorithm, which assigns drivers to routes at the time of travel, with an anticipatory pricing mechanism that determines how much each traveling driver pays if they choose to use a congested route.

The conclusion will present open problems implied in the preceding three chapters.