Teleworking trends under the new normal

The pandemic changed everything in the transportation field – every mode of transport was impacted.

Due to increased teleworking, people changed their commuting patterns, avoiding peak times and flattening out those curves. Many even changed where they live. Of course, when people move, travel demands and travel patterns in an area change.

Researchers, government departments and companies wanted to know about these patterns, and there have been many studies into whether they will continue to fluctuate in the future, or if and when transport will return to pre-COVID levels.

A very different opportunity

We first heard about the Cintra project from our advisor Dr. Patricia Mokhtarian, Professor of Civil & Environmental Engineering (Transportation) at Georgia Tech.

This was a very different opportunity for us. Previously our research projects have been collaborations with or funded by national/state Departments of Transportation (DOTs). This was the first time we have worked with a company in this way.

It was a unique and very positive experience. Cintra provided a lot of input and ideas throughout the study, which helped us shape our project in mutually beneficial ways from beginning to end. Cintra staff were very aware and respectful of our commitment to academic rigor, and conversely we understood (and learned even more than we initially knew) that the staff needed reliable descriptive statistics (how much teleworking and commuting is there right now, compared to before the pandemic? what are the expectations for the future?) even more than our fancy academic models.

We had monthly feedback meetings with the Cintra team, and were aware of the ongoing conversations within the company using our results.

We see partnerships between academia and industry as really beneficial to both sides. Companies gain the specialized knowledge and expertise of academics, while academics can shape their research to be more directly useful in real-world settings.

Two types of study combined to give the bigger picture

In this research, we wanted to know the impact of COVID-19 on travel and transport demand: how much people travel (particularly for commuting), the number of (commute) trips they take, total VMT (vehicle-miles travelled) – a proxy of travel demand – and whether that level would be recovered in the near future.

That is a really key question in many respects. What are the implications for congestion, vehicle emissions, gas tax revenues, vehicle ownership, transit operations, residential location patterns? And an infrastructure company like Cintra, which builds and operates toll roads and managed lanes, wanted to know whether the overall travel demand, and of course revenue, would return to pre-COVID levels soon.

This research combined disaggregate level and aggregate level studies.

The disaggregate part involved asking people questions about their travel behavior and work habits and how this had changed from pre- to during to (expected) post-COVID, analyzing their responses, opinions and behaviors at an individual level. Meanwhile, the data analysis in the aggregate study looked at important macro variables like GDP per capita and fuel prices, in forecasting national and regional vehicle miles travelled (VMT). To our best knowledge, there are very few aggregate analyses of the transportation impacts of teleworking, so we think that is one of the key contributions of this study.

Ensuring our data was representative of the wider employed population

For the disaggregate study we looked at two regions, Dallas and Washington DC. Using an online survey, we asked respondents key questions about their expectations for the future, such as how much they will continue to telework, and whether they will travel more or relocate to another home. We summarized those findings and used them as inputs to our aggregate model when using it to forecast the future.

In Dallas, we had two main data sources – the Cintra customer database and an online opinion panel (a group of people who agree to complete surveys in return for rewards). For Washington DC, the data source was mainly the online panel since Cintra’s concession was not yet operational there.

In recruiting the online panel respondents, we ensured a diverse representation in terms of demographic variables such as gender, income, education level and age. Then, in the analysis, we weighted all cases so that the sample matched the population distribution on those key demographic variables as well as on the different types of teleworkers in the population.

We believe that the extra care we took in recruitment, data cleaning (checking for inconsistent, nonsensical, or inattentive responses and removing such cases), and weighting sets our study apart from most others having similar aims. If we had not carefully weighted the sample, our descriptive statistics would have been quite biased (for example, our unweighted sample heavily overrepresented teleworkers, which is not surprising considering the survey was entirely online).

Increasing numbers and changing demographics of teleworkers

The key takeaway from the survey is that – based on respondents’ expectations for post-COVID – the teleworking occasions tripled from pre- to post-COVID. However, we believed it was likely that people wouldn’t telework post-COVID as much as they thought they would when the survey was taken in Spring 2021. Once we made some evidence-based adjustments to those expectations – another distinctive feature of our study – our best estimate is that the teleworking occasions would be 2 – 2.5 times the pre-COVID number.

We also found that the demographics of teleworkers changed. The newer teleworkers tend to have lower household incomes or different occupations, compared to pre-COVID teleworkers. Thus, the teleworking population is now more diverse. Whether this may also influence road use and travel behaviors post-COVID is something we don’t yet know.

Forecasting is not an exact science, but still so valuable

At the time of our survey, overall VMT was still smaller than pre-COVID and had not fully recovered. However, our forecast suggested that in 2022, it would be slightly larger than in 2019 – so there were signs of recovery in travel demand even if there were still differences in its temporal distribution.

But there are many assumptions in this and so many things can shift within a pandemic situation. If something really new happens, such as a new COVID variant, then everything could be different. As always, forecasting is subject to volatility.

Nevertheless, as VMT is a primary indicator in many transportation applications and transport policymaking, we expect that other agencies will find these near-term forecasts to be useful.

Our research has quickly been put into practice

We have been very happy to see our research results so quickly applied in practice, in helping Cintra adjust their travel and transport models. Many research projects remain firmly in the academic sphere, so it is fulfilling to see our research being used to address real-world challenges.