Road to a Master’s Thesis: Final Research Proposal

After extensive discussions with my faculty advisor and colleagues, this reflects my commitment to a research project as part of the requirements of a Masters in Urban Planning degree. Your feedback, thoughts, and potential assistance (if you work in tech!) would be greatly appreciated.

Post-Pandemic Travel Patterns of Remote-Eligible Workers


  • Transportation planners looking to model travel demand in regions with high concentrations of remote-eligible workers (e.g., large numbers of employees in IT or professional services sectors).
  • Academic researchers with an interest in the impacts of remote work / telecommuting / telework on individual and aggregate travel patterns, including changes in trip frequencies or shifts in mode share.
  • Professional staff and executives at public transit agencies seeking guidance on long-term ridership potential from “choice” riders as remote work becomes increasingly common.


The COVID-19 pandemic upended typical mobility patterns around the globe from 2020 through 2021, especially for people accustomed to commuting for work. Among so-called “knowledge workers” or the “creative class” (Florida 2019), the combination of asynchronous collaboration software and employer flexibility enabled rapid acceleration of what was already a growing trend of telework or remote work (Molla 2019). While past research suggests that only about one-third of all jobs can be done fully remotely (University of Chicago 2020) and less than half surveyed would want to stay remote on all weekdays (PwC 2021), any increase in long-term teleworking may have impacts on travel demand and subsequently traffic congestion and transit ridership.

These impacts may vary between regions depending on the prevalence of “remote-eligible” jobs and workers (Dingel and Neiman 2020). In the San Francisco Bay Area specifically, around 50% of jobs (1.79 million) are “remote-eligible” (Bay Area Council Economic Institute 2020) – significantly more than other counties in California and more than any other metropolitan area in the United States. This is partially due to the high concentration of employees in the technology sector, representing the fastest growing industry segment prior to the pandemic (Metropolitan Transportation Commission 2019) and one of the first groups to fully adopt remote work policies (Levy 2020).

Chart, scatter chart

Description automatically generated
Figure 1: National distribution of remote-eligible jobs (Dingel and Neiman 2020)

If even a small percentage of tech workers stayed remote a few days each week in the long-term, there may be profound impacts on aggregate travel patterns in the region. For example, fewer trips to offices may alleviate commute-time traffic congestion but also devastate the revenues of local transit agencies dependent on regular ridership. Greenhouse gas emissions may be reduced in the short-term but increase as people take more non-work-related trips with their additional leisure time. This research seeks to predict the magnitude of these impacts and more, using the habits of employees in the tech sector as a proxy for remote-eligible workers in general.

Relevant Terminology

Remote work / teleworkWork done from home or other location not at a traditional office
Remote eligibleCharacteristic of jobs that can be done fully remotely
Tech workersEmployees of companies in the technology, software, or information industries
Knowledge workers / creative classBroader group of professional workers that either fully engage in the creative process (e.g., innovation) or draw on bodies of knowledge to solve problems
PandemicThe COVID-19 pandemic that spread in the United States from February 2020 until the present
Pre-pandemicTime period prior to February 2020 (when the COVID-19 pandemic triggered shelter-in-place orders in the San Francisco Bay Area)
Post-pandemicTime period estimated to begin in early 2022 when the majority of remote-eligible employees will receive the option to return to re-opened office environments

Research Question

What are the unique long-term impacts of post-pandemic remote work on travel behaviors among tech workers (as representative of all remote-eligible workers) within the San Francisco Bay Area?

  • Sub-question 1: Which factors/variables are most predictive of sustained telework with this cohort?
  • Sub-question 2: How do stated/revealed preferences of this cohort compare to national averages?


Even after COVID-19 vaccinations and offices reopening, tech workers in the San Francisco Bay Area will continue teleworking 2-3 days per week on average. While the number of commute (work) trips per week will decrease, the number of non-work trips (for errands, groceries, etc.) per week will increase. These increased trips will be completed with modest shifts in travel mode – i.e., some people will take more biking and walking trips, while others will take more trips with single-occupancy automobiles. The average trip length/distance will decrease as people take fewer long commutes and stay within their immediate residential communities.

This hypothesis is based on historical research about telework, press coverage of the proposed long-term remote work policies of prominent employers in the tech sector, and anecdotal stated preferences from peers in the industry.


Description automatically generated
Figure 2: Snapshot of Tech Company Remote Work Policies (Kim 2021)

Based on recent findings from a similar study about post-pandemic travel preferences conducted on a national scale (Barbour, Menon and Mannering 2021), hypotheses about the sub-questions can also be made. Socio-demographic factors such as age, education level, and presence of children will be the strongest predictors of a transition to sustained remote work, while the characteristic of having a job in the information technology sector will also skew preferences towards more remote work.


For decades, urban planners in the San Francisco Bay Area (and California more broadly) have been keen to address the worsening issues of traffic congestion and GHG emissions from transportation – the top source of CO2 in California (California Air Resources Board 2021). Telecommuting has been touted as a potentially effective transportation demand management (TDM) strategy for the region, to the point where it was pitched as a potential employer mandate from MTC though withdrawn due to concerns about equity and risks to downtown businesses (Waxmann 2020). However, the impact of broad and extensive adoption of telework had never been seen before the COVID-19 pandemic. The resulting chaos has led many to wonder what would really happen if widespread telework continues long-term, particularly with regards to:

  • Public transit revenues
  • Travel mode choices/shifts
  • Greenhouse gas emissions
  • Demand for commercial office space in downtown districts
  • Housing preferences (shifts towards suburbs)
  • Service sector jobs dependent on office worker populations

This research can provide some guidance and model projections, utilizing the Bay Area’s famed concentration of tech workers as a litmus test for all 1.79 million remote-eligible workers in the region. The learned habits of fully remote-eligible workers will only become more relevant as additional remote jobs are offered in the region and across the country.

Figure 3: Companies are increasingly posting new remote jobs (Oldham 2021)

The potential impacts of telework on travel patterns (as measured in aggregate VMT and GHG emissions) has been studied for the past 30 years with mostly mixed results. A few papers have suggested that telework will result in significantly reduced travel demand (Curtis 2020, Kitou and Horvath 2003, Shabanpour, et al. 2018), though these had some limitations – for example, only looking at rush hour or effects on only one company in one city. A few others concluded that telework could counterintuitively lead to increased VMT (Zhu and Mason 2014, Riggs 2020, de Abreu e Silva and Melo 2018) due to additional non-work trips being made (and using more polluting modes such as cars to boot).

However, the vast majority of the work presented mixed conclusions. Some papers found non-existent or negated impact (Mokhtarian 1998, Moeckel 2017, Gareis and Kordey 1999, Lyons 1998). Others were more uncertain or were just starting a longer-term study (Matson, et al. 2021, Hook, et al. 2020). Still others found that impacts were highly dependent on variables such as specific telecommuting arrangements or household/personal characteristics (Lachapelle, Tanguay and Neumark-Gaudet 2018, Kim, Choo and Mokhtarian 2015). Given the many potential confounding factors and flavors of telework that can be studied, perhaps it is not too surprising that there may never be a simple or definitive answer to this research question.

While these previous works have ranged quite broadly in their geographic area of study and the types of telecommuting workers they surveyed, there has not been any widely cited work on the San Francisco Bay Area or its high concentration of workers in technology/software companies specifically. Hence, this research can add to this body of knowledge around the effects of telework on travel patterns more generally, extrapolating findings from a population with a high proportion of remote-eligible workers. Any impact on VMT may become more pronounced or clearer here than in other regions.

This work can also complement parallel studies on this topic (Barbour, Menon and Mannering 2021, Menon, Keita and Bertini 2020, Matson, et al. 2021), confirming the potential differences in behavior/preferences between those working in the tech sector versus others.


Description automatically generated with low confidence
Figure 4: The geospatial distribution of a prior nationwide survey, with limited coverage in the San Francisco Bay Area (Menon, Keita and Bertini 2020)


Online Survey of Tech Sector Employees


I will conduct an online survey of employees working at major technology companies headquartered (or formerly headquartered in 2020) in the San Francisco Bay Area to get a snapshot of their current travel behaviors and stated preferences regarding long-term remote work. (Revealed preferences will be collected if employers have established long-term remote working policies that employees have firmly adapted to at the time of the survey.)

Reason for collecting the data

Understanding the travel patterns that have been established during the pandemic (February 2020 to present) gives a sense for how this cohort has adapted to an environment where they were forced to work from home. Their level of comfort with the situation, the new/altered trips they take, the new mode choices they have made, and the factors that contribute to heterogeneity in these variables would all be indicative of the characteristics of remote-eligible workers who continue to work this way in the long-term.

The stated/revealed preferences in a post-pandemic environment are even more valuable as they can be used to generate predictive models based on reported socio-demographic factors. They can also be directly compared to patterns seen in other regions or nationally, where they may be less of a concentration of remote-eligible workers.

Data collection procedures

First, a set of technology companies headquartered (or formerly headquartered in 2020) in the San Francisco Bay Area will be identified – starting by referencing a list of top employers (by workforce size). Smaller companies (i.e., startups) could also be included by coordinating with venture capital firms or through personal contacts. However, larger companies would be more ideal as they tend to lead in workforce policies and would enable access to many more potential survey participants.

Each company typically has someone in the human resources (HR) department who is responsible for mobility/commuter benefit programs. Efforts will be made to connect with these individuals to discuss the possibility of augmenting or taking over a survey typically given to employees once a year to determine their commute preferences. Alternatively, permission to send a separate survey to employees (ideally originating from the HR department itself or another internal champion) can be negotiated. The exact timing should be planned with company business cycles in mind for the best chance at high levels of participation, but sometime in late 2021 when longer-term remote work policies are in place and COVID-19 vaccinations have completed would be preferred. The goal would be to have 385 or more respondents across 5 or more companies for broad coverage (based on best practices for sample sizes).

Survey questions will be prepared starting from a set of well-tested questions used in a national survey conducted by (Menon, Keita and Bertini 2020), for ease of comparing results during the analysis phase. Some questions may need to be slightly modified or limited if requested by the HR representatives, though it will be critical to maintain as much consistency as possible between companies for data integrity. The survey can be implemented in Qualtrics or a similar tool and be distributed through internal channels as much as possible.

If participation directly through employers is not sufficient to obtain a sizable pool of employees willing to be surveyed, alternative methods may be attempted such as:

  • Paying directly for access to Prime Panels (a.k.a. Turk Prime), Qualtrics Panels, or Prolific Panels
  • Running targeted ads on social media to recruit participants
  • Collaborating with Employee Engagement Companies / Communications Firms such as ROI

In addition, individual interviews with HR representatives can be conducted to document official company policies and understand general employee sentiment if a formal survey cannot be conducted due to privacy concerns from the employer.

The survey itself will ask about individual behavior during various timeframes: pre-pandemic, during the pandemic shelter-in-place orders (March 2020), the current timeframe, and the hypothetical post-pandemic timeframe.

Topics covered will include but are not limited to:

  • Company attributes (i.e., size) and history/culture around remote work
    • Currently established long-term remote work policy (can be confirmed with HR)
  • Personal work travel/commute patterns
    • Recent modal shifts or change in vehicle ownership
    • Days in the office versus fully remote
  • Personal travel patterns for non-work purposes
  • Opinions and preferences around mode choices
  • Sociodemographic and household information
    • Age
    • Education level
    • Home location (ZIP code)
    • Household size / presence of children

Questions about the future will be limited to the short-term (i.e., speculation on something 5 years out would not be valuable).


Description automatically generated
Figure 5: Example survey questions from national study (Menon, Keita and Bertini 2020)

The survey will remain open for at least one month from initiation, or shorter/longer if requested by individual HR representatives. The data will be collected by the survey tool and should remain accessible to internal company stakeholders.

Method of data analysis

The survey data can be analyzed in a series of progressively complex ways:

  1. Simple descriptive statistics related to current travel patterns only, segmented by demographics (particularly with regards to household/family size and presence of children) and location if participants end up being in significantly disparate geographic areas. Participants can also be grouped by the nature of their employer’s remote work policies, as some companies would be going fully remote while others may be more resistant or moving to a hybrid model.
    1. Example: 95% confidence intervals around mean of total non-work trip distances per week, for employees with children and those without
  2. Comparison between statistics during the pandemic, current timeframe, and hypothetical future.
    1. Example: 2-sample t-test (difference of means) of days in office per week, during vs. current or current vs. future
  3. Finding correlation between reported characteristics.
    1. Example: Chi-square test of company remote work policy as a categorical independent variable and mode choice for work-related trips as a categorical dependent variable
  4. Predictive modeling of future behavior based on multiple factors/variables.
    1. Example: Multivariate Regression with presence of children, age, and education level as input variables and hypothetical future number of days in the office per week as an outcome variable
    1. Example: Random parameters binary logit model (with heterogeneity in means and variances) to predict probability of switching to permanent work-from-home given the full set of explanatory variables (Barbour, Menon and Mannering 2021)

Greater availability/richness of the data collected will enable more confidence in pursuing more advanced techniques.

Optionally and if time permits, the survey results could also be paired with some external data sources to verify hypothetical patterns. Data sources may include:

  • Car and bicycle sale data
    • Second-hand vehicle market / Kelly Blue Book data
  • National Household Travel Survey
  • Traffic data from Inrix or local road sensors
  • Car crash data (potentially to add insight to a side question of whether traffic injuries/fatalities per mile have increased or decreased with car usage)

Public Transit Data Analysis


I will pull basic ridership data from one or more public transit agencies in the San Francisco Bay Area region (SFMTA, BART, and/or Caltrain) which served the pre-pandemic commuting needs of tech workers, to get measures of patronage at different points in time: pre-pandemic (January 2020), immediately after shelter-in-place orders (March 2020), and the timeframe when the survey of tech workers is conducted (October or November 2021).

Reason for collecting the data

By matching ridership levels at each timeframe to the stated levels/characteristics of remote work practiced by tech workers in the survey, it will be possible to create a model that describes expected public transit ridership given potential future extended adoption of remote work. This would better illustrate the direct impacts of post-pandemic remote work for public transit agency stakeholders.

Data collection procedures

The various public transit agencies make ridership data available as part of federal regulation. For example:

The raw data can all be downloaded into Excel (.XLS) formatted files. I will attempt to standardize on monthly ridership, extrapolating from annual counts if necessary. Ideally, a baseline of “pre-pandemic” ridership levels can be established and then the rest of the data points can be expressed as a percentage of this baseline. For example, March 2020 ridership might be 10% of the baseline while October 2021 ridership might be 60% or similar.

Chart, line chart

Description automatically generated
Figure 6: Example of month-by-month ridership data (Arvin and Siegal 2021)

Method of data analysis

A linear regression method could be used to map one of the reported factors in the survey as an independent variable and percentage of ridership recovery on a given transit system as a dependent variable. For example:

  • Independent variable: Median number of reported days in the office per week
  • Dependent variable: BART passenger trips (% of pre-pandemic level)
  • Test statistic: R2


(*) peer-reviewed journal article

Arvin, Chris, and Kat Siegal. Transit Recovery in U.S. Cities. July 2021. (accessed September 15, 2021).

* Barbour, Natalia, Nikhil Menon, and Fred Mannering. “A statistical assessment of work-from-home participation during different stages of the COVID-19 pandemic.” Transportation Research Interdisciplinary Perspectives 11 (2021): 100441.

Bay Area Council Economic Institute. “Remote Work in the Bay Area.” December 2020. (accessed April 27, 2021).

California Air Resources Board. GHGs Descriptions & Sources in California. 2021. (accessed April 17, 2021).

Curtis, Matt. (re)Awakening to the benefits and climate impacts of telework during COVID-19. Capstone Project, San Diego: UC San Diego: Climate Science and Policy, 2020.

* de Abreu e Silva, João, and Patricia C. Melo. “Home telework, travel behavior, and land-use patterns: A path analysis of British single-worker households.” Journal of Transport and Land Use 11, no. 1 (2018): 419-441.

Dingel, Jonathan, and Brent Neiman. “How Many Jobs Can be Done at Home?” June 19, 2020. (accessed April 27, 2021).

Florida, Richard. The Changing Geography of America’s Creative Class. August 27, 2019. (accessed April 27, 2021).

Gareis, Karsten, and Norbert Kordey. “Telework – An Overview of Likely Impacts on Traffic and Settlement patterns.” NETCOM 13, no. 3-4 (1999): 265-286.

* Hook, Andrew, Victor Court, Benjamin K Sovacool, and Steve Sorrell. “A systematic review of the energy and climate impacts of teleworking.” Environmental Research Letters 15, no. 9 (2020): 3003.

Kim, Kyum. Work From Home Policies Will Make Tech Employees Leave or Stay. June 21, 2021. (accessed September 14, 2021).

* Kim, Seung-Nam, Sangho Choo, and Patricia L. Mokhtarian. “Home-based telecommuting and intra-household interactions in work and non-work travel: A seemingly unrelated censored regression approach.” Transportation Research Part A: Policy and Practice 80 (October 2015): 197-214.

* Kitou, Erasmia, and Arpad Horvath. “Energy-Related Emissions from Telework.” Environ. Sci. Technol. 37, no. 16 (2003): 3467-3475.

* Lachapelle, Ugo, Georges A Tanguay, and Léa Neumark-Gaudet. “Telecommuting and sustainable travel: Reduction of overall travel time, increases in non-motorised travel and congestion relief?” Urban Studies 55, no. 10 (2018): 2226-2244.

Levy, Ari. The tech industry’s early work-from-home mandates helped California and Washington flatten the curve. April 16, 2020. (accessed April 27, 2021).

* Lyons, Glenn. “An assessment of teleworking as a practice for travel demand management.” Proceedings of the Institution of Civil Engineers: Transport (Thomas Telford) 129, no. 4 (January 1998): 195-200.

Matson, Grant, Sean McElroy, Yongsung Lee, and Giovanni Circella. “Longitudinal Analysis of COVID-19 Impacts on Mobility: An Early Snapshot of the Emerging Changes in Travel Behavior.” Research Report, 3 Revolutions Future Mobility Program, UC Davis, Davis, 2021.

Menon, Nikhil, Yaye Keita, and Robert L. Bertini. Impact of COVID-19 on Travel Behavior and Shared Mobility Systems. Tech Report, Tampa: University of South Florida. Center for Urban Transportation Research, 2020.

Metropolitan Transportation Commission. “Jobs by Industry.” Vital Signs. August 2019. (accessed April 27, 2021).

* Moeckel, Rolf. “Working from Home: Modeling the Impact of Telework on Transportation and Land Use.” Transportation Research Procedia 26 (2017): 207-214.

* Mokhtarian, Patricia L. “A Synthetic Approach to Estimating the Impacts of Telecommuting on Travel.” Urban Studies 35, no. 2 (1998): 215-241.

Molla, Rani. How remote work is quietly remaking our lives. October 9, 2019. (accessed April 27, 2021).

Oldham, Ethan. Future of Remote Work in a Post-COVID World. July 27, 2021. (accessed September 14, 2021).

PwC. It’s time to reimagine where and how work will get done. January 12, 2021. (accessed April 27, 2021).

Riggs, William. Telework and Sustainable Travel During the COVID-19 Era. SSRN, 2020.

* Shabanpour, Ramin, Nima Golshani, Mohammad Tayarani, Joshua Auld, and Abolfazl Kouros Mohammadian. “Analysis of telecommuting behavior and impacts on travel demand and the environment.” Transportation Research Part D: Transport and Environment 62 (July 2018): 563-576.

University of Chicago. With much of U.S. staying at home, how many jobs can be done remotely? March 30, 2020. (accessed April 27, 2021).

Waxmann, Laura. MTC ditches Bay Area telecommuting mandate for more flexible plan. November 23, 2020. (accessed April 17, 2021).

* Zhu, P., and S.G. Mason. “The impact of telecommuting on personal vehicle usage and environmental sustainability.” Int. J. Environ. Sci. Technol. 11 (2014): 2185-2200.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.