Why are the work-at-home statistics so different from one source to the next?

We offer the latest data about how many people telecommute and how often they do it on our Work-at-Home Statistics page, but many have asked why the numbers that are stated in the press vary so widely. The problem is that everyone who conducts a survey comes at it with their own needs and biases (including us). Below are details about the primary sources of data on who telecommutes and offers it.

Conflicting Counts of Work-at-Home Statistics

Work-at-home numbers vary by source primarly because of who is counted. Some data includes the self-employed which taints the numbers because, at least right now, they are flat or declining. Some researchers include part-time employees and even unpaid family. Some count any work-at-home even if it’s just people taking work home to work on outside of normal working hours or working in second jobs (i.e. eBay seller, gig workers, etc.). We tend to use American Community Survey (ACS) data. Here’s why.

The Census Bureau and American Community Survey asks “What was your primary means of transportation to work during the survey week?” One of the optional answers is “worked at home.” The do not drill down into how often the respondents worked from home. Therefore, all we really know is if someone said “worked at home” they must have done so half-time or more. Most of the data you’ll find on our telecommuting statistics page, is based on ACS data. It’s not perfect, but its a huge sample size and, importantly, it allows us to separate self-employed from employee work-at-home and allows us to parse the data by public and private sector workers, geography, age, gender, occupation, industry, and many other demographics. ACS data is updated annually in the fall of each year. Therefore, 2019 data, based on surveys fielded in 2018 and 2019, will be available in September of 2020. Some of the demographic data is not updated until later in the year.

The Bureau of Labor Statistics (BLS) captures information on where and when people work. It does not distinquish between employees who work at home in lieu of going to an office from those who may be working overtime or in second jobs.

Gallup has be collecting work-at-home data for many years. Their data sometimes include the self-employed, but they also provide data on just employees who work remotely.

Conflicting Number of Employers Who Offer Work-at-Home

There is a great deal of discrepancy in the reported numbers for who offers work-from-home to their employees. BLS only counts organizations that offer it to all or most of their people (about 7% in 2019). SHRM and others tend to count companies that offer it all, regardless of how many or few they offer it to.

The IRS and the SBA gather information about home-based businesses. And the Office of Personnel Management (OPM) tracks telework practices in the federal workforce.

“Trying to find the real work-at-home numbers as a “statistical Vietnam – the data goes in, but you can’t get it out,” said Bruce Phillips, then researcher for the National Federation of Independent Business (NFIB).

As a result, studying the work-at-home population is a little like trying to study meteoroids. We know there are a lot of them and we know they’re important, but we don’t know where they all are and not everyone agrees on which ones to count.

For U.S. numbers, we lean toward Census data. They ask employed workers where their principal place of work was during the survey week. Granted, some may have had an unusual week, but it’s the most reliable and consistently updated data available.

Of course, we have our biases too. We’re interested in the people who regularly work from home because they’re the ones who have the greatest impact on global warming, fuel supplies, and traffic congestion. They’re also the ones who derive the most work-life benefits from the arrangement.

While companies such as IBM and Cisco call a high percentage of their staff teleworkers, few of them actually work both from home and at home. These road warriors, plumbers, electricians, etc. should be labeled differently than the from home/at home population if we’re to really understand the environmental, economic, and work-life benefits of telecommuting.

Here’s a round-up of the primary sources of telework data, along with a discussion of their limitations.

Census / American Community Survey (ACS)

ACS is a nationwide survey conducted annually by the U.S. Census Bureau. It produces 1-year estimates for geographic areas with a population of 65,000 or more. This includes the nation, all states and the District of Columbia, all congressional districts, approximately 800 counties, and 500 metropolitan and micropolitan statistical areas. About three million housing unit addresses and 307 million people are represented in the weighted sample.

ACS asks survey respondents: What was your primary means of transportation to work during the survey week?

  • Car, truck or van – driving alone
  • Car, truck or van – Carpooled
  • Public transportation
  • Walked
  • Taxi, motorcycle or bike
  • Worked at Home

ACS Limitations

While the question offers some insight into the WAH workforce, it falls short of providing useful answers in a number of ways:

1) While the respondent is also asked whether they work for a private or public sector organization, if they’re self-employed, or if they’re an unpaid family worker, that class of worker data, is only tied to the ‘means of transportation to work’ category in a handful of Census Bureau reports.

American Fact Finder, the primary search tool for Census data, does not allow users to determine, say, the number of non-self-employed people in the construction industry who work from home in Millville, New Jersey. It could be used to determine how many were self-employed, or how many were unpaid family workers, or how many were state government workers, and it could help you determine how many people in Millville worked from home, but it doesn’t separate out those who were not self-employed.

2) ACS only captures information about people who primarily work at home, not those who do so on an occasional basis – a group far larger than those who do so most of the time.

3) ACS does not capture information about people who work remotely from client offices, shared office centers, coffee shops, their cars, or other ‘third places’ as they’re called.

4) The statistical validity of changes in the WAH population obviously diminishes with size. In general, for total workforce populations smaller than a million – while data regarding the total regional WAH population is statistically valid – the granular changes within them may not be. For this reason, we primarily focus our reporting on larger metro regions.

Bureau of Labor Statistics (BLS)

The Bureau of Labor Statistics (BLS) surveys produces the American Time Use Survey (ATUS) and the National Compensation Survey (NCS):

American Time Use Survey (ATUS)

ATUS is conducted annually. It includes, among other things, information about where and when people work – at their workplace, at home, or at another location. The data are collected through telephone interviews.

ATUS defines the term ‘working’ as any time employed people spent doing tasks required for a job. A person who reads work-related e-mail messages for 10 minutes on a Saturday is counted as working on that day, as is someone who worked a 12-hour shift.

ATUS Limitations

ATUS does not distinguish between people who are paid to work from home and those who simply take work home.

BLS does not allow standard searches or produce standard reports that distinguish the self-employed from the non-self employed by industry, occupation, or anything but the most general data. The most recent survey that does separate the self-employed from the rest of the WAH population covers the periods of 2003 to 2007.

Respondents are asked to answer questions based on where they worked on a particular survey day, which may or may not be indicative of their regular workplace.

BLS National Compensation Survey (NCS)

The National Compensation Survey is conducted annually. It collects information from companies about the compensation and benefits they offer.

One benefits choice is ‘flexible workplace’. BLS defines this as: ‘Permits workers to work an agreed-upon portion of their work schedule at home or at some other approved location, such as a regional work center.’ They note, ‘such arrangements are especially compatible with work requiring the use of computers linking the home or work center to the central office.’

National Compensation Survey Limitations

NCS data only indicates who offers a benefit, not who’s using it, how, or how often.

Their count does not include companies that offer workplace flexibility on an ad-hoc or occasional basis.

Data on Federal Telework

Data about participation in telework programs within the federal workforce comes from the annual Status of Telework in the Federal Government – Report to Congress produced by the Office of Personnel Management.

Who is counted as a teleworker varies from agency to agency. Some, unfortunately, count people who have a signed telework agreement, but who are prohibited from teleworking by their boss.

OPM is hard at work trying to come up with a consistent measure, but as it changes the reported numbers will too. For example, the number of federal workers considered eligible for telework dropped from 66% in 2010 to 34% in 2012 simply because the definition changed. We estimate that at least half of all government workers could telework 2 or 3 days a week.

For more information on the who, when, where, and how remote workers are working, download our free white papers.

For the latest U.S. telework numbers, visit our Telework Statistics page.