Last month, I launched a survey for employees in the tech sector designed to illuminate trends in travel patterns during this era of widespread remote work. It asked participants to share details about their work and non-work trips, as well as potential interest in relocation. Here I will share some early findings, though keep in mind that the analysis phase is ongoing and there is a long road from here to a final publication. Though the contents below are unpolished, I believe it’s all still worth sharing and welcome any thoughts or suggestions for next steps.

Overview

There were over 900 responses to the survey through December 2021, but the sample sizes for each question generally varied between 300-500 since all questions were optional and many participants just didn’t complete the full set. (Respondents who said they did not work full-time in the software/technology industry or did not have an office in the San Francisco Bay Area as of February 2020 were also filtered out.) Despite the substantial drop-off rate, a staggering amount of data was collected and the majority of the findings can be considered statistically significant at the 95% confidence level, assuming a true population size of ~1.79 million remote-eligible workers in the San Francisco Bay Area.

Over 120 unique technology firms were represented, from the well-known Big Tech giants all the way down to small (sometimes unnamed) startups. Larger companies with greater numbers of employees were appropriately more represented.

And their offices were distributed all across the Bay Area, with a slight concentration in downtown San Francisco.

Employer Remote Work Policy

Almost half (46.5%) of all employers reportedly will just allow individual employees or teams to decide on remote work policies for themselves. For companies that are implementing a mandate, at least 3 days per week in the office seems to be the most common choice.

Interestingly, larger employers (5000+ employees) were far more likely to have an onsite work requirement. Small companies, especially those less than 100 employees, seemed to gravitate more towards being fully remote.

The majority of employers have already put their policies in place, with the rest planning on implementation shortly. However, the continuation of the COVID-19 pandemic with the Omicron variant likely shuffled these plans after this survey was distributed.

Vehicles Owned

The survey asked participants to report the number of vehicles they owned before February 2020 versus currently. A slight positive trend was seen for all vehicle types, including cars, but the strongest increase was seen across active mobility options (bicycles, e-bikes, and scooters) where the 95% confidence interval of mean difference (current – before COVID-19) was +0.11 to +0.23 additional vehicles per person.

Commute

The commute patterns of tech workers obviously shifted drastically during the pandemic, but it is quite clear that things will not go back exactly how they used to be even once COVID-19 is no longer perceived as a threat. A hybrid model of 2-3 days per week in the office seems to be the pattern that most tech workers are converging on.

Mode shifts are also predicted, as the rate of driving alone to work will remain high after the pandemic while usage of public transit and walking decline.

Those respondents who stated they would drive alone to work in the future also tended to have more cars in the household (1.78 on average) versus those who wouldn’t drive (1.10 on average). Also, those with commutes of less than 5 miles are much more likely to not drive alone to work.

Non-work Activities

The stickiness of new pandemic-era habits such as online food/grocery/shopping orders is one of the more surprising findings from the survey. For example, 67.4% of those who started ordering groceries for delivery apparently will continue doing so after COVID-19 is no longer a threat. 70.6% of those who started ordering food for delivery from a restaurant will also continue. This is reflected in a major loss in trips to physical stores/restaurants after COVID-19.

For those who will still make trips to physical locations (either to shop directly or just to pick up online orders), mode shifts are observed towards driving and away from public transit and walking.

Relocation

The question of whether remote work will lead to a big “exodus” away from traditional tech hubs has been debated endlessly among planners and economists throughout the pandemic. The survey results suggest that moves are happening, but they’re not going to be very far.

More than half of respondents did say they were considering a move or have already moved – not all due to remote work options, however.

Relocation interest was correlated with age, with younger folks (25-34) more likely to have already moved while older folks (55+) would most likely not consider moving.

Those who rented were also more likely to have already moved.

When asked about the reasons prompting a potential move, classic values such as “quality of life” and “cost of living” ranked highly. However, “more space” and “remote work” also had strong representation.

Respondents generally did not want to change up the type of environment they lived in, with most sticking to places similar to where they were before the pandemic. But of those who wanted change, the lure of suburbs was the strongest.

Related to this, a shift in preference towards stand-alone homes versus condos/apartments was also evident:

There was also a notable shift towards homeownership, with renters flipping to owners throughout the pandemic and beyond.

When looking at where people wanted to move to, the narrative of a mass exodus from California does seem to be a bit exaggerated. There are definitely some folks who are looking to the Pacific Northwest or the East Coast, but the vast majority of folks are staying in-state.

Public Transit Usage

In order to predict the impact of travel pattern changes on overall public transit usage, a numerical estimate of trips taken on transit needed to be established based on respondents’ answers to questions from multiple points in the survey. For example, answers to a question about how often respondents engaged in certain non-work activities (i.e., shopping, social gatherings, entertainment, and medical appointments) were converted into a number of potential trips depending on whether “public transit” was selected as a mode plus a recoded frequency variable (e.g., “few times per week” became 2*3 public transit trips, assuming a roundtrip journey three times during a week). These values were then combined with estimates for work trips based on similar recoding on commute frequency. This methodology is not exact, but the relative changes between trips before/after COVID-19 can be illustrative of broader trends.

The results paint a dire picture for the potential return of public transit patronage. An average of 14.2 trips per week before COVID-19 became 1.9 trips per week “currently” (December 2021), which was understandable as commuting trips have been completely eliminated. However, only 3.4 trips per week are expected after COVID-19 is no longer a threat, which is barely 24% of pre-pandemic levels and suggests that transit agencies cannot depend on this cohort to ride like they used to for the foreseeable future.

More work is needed to verify this finding as it is almost too drastic. An analysis of this data matched against actual ridership data from BART, SFMTA, etc. for the appropriate time periods will also provide more confidence.

Next Steps

At this stage, basic descriptive statistics and a few correlations between variables have been established, but there is potential for greater insights through predictive modeling. For example, a multivariate regression could explore factors that contribute to the change in cars owned such as age or presence of children. A more advanced random parameters binary logit model can also be constructed to predict the probability of switching to a fully remote setup given various demographic or other explanatory variables.

I’m open to suggestions and/or collaborations with anyone interested in this one-of-a-kind dataset, so feel free to reach out!