Analytical Studies Branch Research Paper Series
Working Most Hours from Home: New Estimates for January to April 2022

Release date: July 17, 2023

DOI: https://www.doi.org/10.25318/11f0019m2023006-eng

Skip to text

Text begins

Abstract

Using Statistics Canada’s COVID-19 Restriction Index and estimates of telework feasibility, this study models, for the period from January 2020 to July 2022, the percentage of Canadian workers who worked most of their hours from home in a given province during a given month. Along with descriptive evidence from public transit data and Statistics Canada’s COVID-19 Restriction Index, predicted values from the model indicate that the percentage of Canadians working most of their hours from home was likely overestimated from January 2022 to April 2022. New estimates are offered at the provincial and national levels for these four months. A simple adjustment procedure is also proposed for revising estimates of work from home by other dimensions for these four months.

1 Introduction

The percentage of Canadians working most of their hours from home has more than tripled since the mid-2010s, increasing from 7.4% in May 2016 to 24.3% in May 2021 (Statistics Canada, 2022). This massive change in work arrangements has led to a demand for timely information on the number and percentage of Canadians working most of their hours from home.

To satisfy this demand for new information, Statistics Canada has introduced in the Labour Force Survey (LFS) a series of questions aimed at estimating the number and percentage of Canadians working most of their hours from home during the LFS reference week, as well as their usual work location. These questions were fielded from April 2020 to December 2021. Starting in January 2022, a new set of questions was introduced. These questions were aimed at better measuring the prevalence of hybrid work—a work arrangement where the worker usually works some hours at home and some hours at locations other than home—to track more precisely the growing diversity of work location arrangements within the Canadian labour market.

As shown below, data on public transit, data on COVID-19 restrictions and regression analyses all suggest that estimates of the percentage of individuals working most of their hours from home are biased upwards for the months from January to April 2022. To remove this bias, the study offers new estimates for these four months at the provincial and national levels. A simple adjustment procedure is also proposed for revising estimates of work from home by other dimensions for these four months.

The study is structured as follows. Section II compares the different questions used in the LFS from April 2020 to December 2022 to produce statistics on work from home. It also offers potential explanations for the likely overestimation of the percentage of Canadians working most of their hours from home during the first four months of 2022. Section III provides descriptive evidence that the estimates from January to April 2022 likely overestimate the true percentage of Canadians working most of their hours from home during this period. Section IV models the percentage of Canadians working most of their hours from home as a function of telework feasibility and Statistics Canada’s COVID-19 Restriction Index. Along with the descriptive evidence offered in Section III, the numbers shown in Section IV suggest some overestimation for the period from January to April 2022. Section V provides new estimates for this period at the provincial and national levels and a simple adjustment procedure for revising estimates of work from home by other dimensions for these four months. Concluding remarks follow.

2 Questions used to measure work from home: April 2020 to December 2022

While Statistics Canada previously collected information on work location through the Census of Population and other surveys,Note fast-changing labour market conditions during the COVID-19 pandemic required more timely and regular data collection on this topic. In response to this need, Statistics Canada introduced two new questions on work location in a supplement to the LFS in April 2020. Based on the existing data collection parameters for LFS supplements, the universe was restricted to the non-institutionalized population aged 15 to 69 living in the provinces.

The questions had two main objectives. The first was to provide estimates of how many Canadians were working mainly from home during the LFS reference week.Note The second was to identify, among those who were working from home, how many had transitioned to remote work in the context of the COVID-19 pandemic.

To establish how many workers had started to work from home because of the pandemic, the supplement first asked all employed workers (including those who were absent from their job or business) about their usual work location, excluding recent changes related to the COVID-19 pandemic (Question U1). Then, employed respondents who had worked during the LFS reference week were asked to indicate in which location they had worked most of their hours during that week (Question C1). For both questions (U1 and C1), the response categories were based on concepts from the Census of Population and covered three types of work location: at home, at a fixed location outside the home and no fixed location.Note The main indicator used to track changes in work location during the COVID-19 pandemic was the share of workers who worked most of their hours at home during the LFS reference week (based on C1).

The two questions continued to be collected until December 2021, but, starting in January 2022, they were replaced by questions designed to better capture medium- and long-term trends in the evolution of work location arrangements in the Canadian labour market, including the growing presence of hybrid work. The first change involved replacing the question on usual work location before the pandemic (U1) with a question on usual work location at the present time (U2). This new question used a mark-all format, allowing respondents to indicate that they usually worked both at home and at locations other than home. At the same time, the question on the location where the respondent worked most of their hours during the LFS reference week (C1) was changed to a question on the proportion of hours they had worked at home during that week (C2).

While the second version of the question on current work location (C2) was designed to produce an estimate of the main work location of respondent during the LFS reference week that would be comparable to the question from April 2020 to December 2021 (C1), a challenge encountered during the implementation of the supplement affected the comparability of estimates for the period from January to April 2022.

All new Statistics Canada survey questions must undergo a process of qualitative questionnaire testing prior to collection. This testing ensures that the questions are clear and accurately measure the intended concepts. The new question on current work location (C2) did not raise any concerns during qualitative testing and was well understood by all participants. However, one cannot be certain that survey respondents in the field read and interpreted the question in the same way as test participants.Note The analysis below suggests that the new question (C2) yields an overestimation of the number of Canadians who work most of their hours at home. Two interrelated issues may have contributed to this overestimation.

First, some respondents may have read the question too quickly and focused on the response categories instead of the body of the question. While the body of the question asked respondents about the share of actual hours they worked from home during the LFS reference week, the response categories did not include a reference to hours worked from home. Therefore, some respondents may have interpreted Question C2 as asking them,

  1. “What share of your usual hours did you actually work last week?”

instead of,

  1. “What share of your actual hours did you work from home last week?”

A factor that may have primed respondents to interpret Question C2 in this way was the presence of a question on the number of hours lost because of the COVID-19 pandemic from the LFS “Disaster/Catastrophe Effects” module. This module was collected from April 2020 to December 2022 and immediately preceded the collection of data on work location. The reference to hours worked may have led respondents to interpret Question C2 in an erroneous way (i.e., as asking them about the share of usual hours they actually worked during the LFS reference week).

An adjustment was applied to the current work location question in May 2022, adding a reference to the home in the response categories (C3). As illustrated in the remainder of the paper, this adjustment brought the series back in line with the rest of the data series.

With the easing of public health restrictions and the waning impact of COVID-19 on the labour market, Statistics Canada will reduce the frequency of data collection on work location. Starting in 2023, supplementary LFS questions on work location will be collected less frequently, while ensuring that a comparable data series on work location is available going forward.

3 Descriptive evidence

Chart 1 plots the percentage of Canadian workers—employees and self-employed—working most of their hours from home, along with the number of Canadians (in tens of millions) using public transit. These monthly data cover the period from January 2020 to October 2022.Note

Chart 1 Percentage of workers working most of their hours from home and number of passengers using public transit, Canada, January 2020 to October 2022

Data table for Chart 1 
Data table for Chart 1
Table summary
This table displays the results of Data table for Chart 1. The information is grouped by Months and years (appearing as row headers), Passenger trips (10 millions) and Work from home (percent), calculated using percent units of measure (appearing as column headers).
Months and years Passenger trips (10 millions) Work from home (percent)
percent
2020
January 16.4 7.2
February 16.1 7.2
March 8.9 24.2
April 2.6 41.1
May 2.8 37.8
June 3.9 30.6
July 5.4 28.4
August 5.6 26.2
September 6.3 25.4
October 5.8 25.9
November 5.8 27.2
December 5.4 28.3
2021
January 4.6 33.1
February 4.9 30.9
March 5.5 29.1
April 5.2 30.7
May 5.2 30.6
June 5.5 27.9
July 6.2 25.9
August 6.5 24.1
September 8.2 23.8
October 8.7 23.7
November 9.1 23.5
December 8.2 23.8
2022
January 6.5 43.0
February 7.6 37.2
March 8.9 34.7
April 9.2 31.7
May 9.6 24.0
June 9.7 23.8
July 9.4 24.2
August 9.7 23.4
September 11.7 22.1
October 11.7 21.8

The first thing to note is that estimates of the percentage of Canadians working most of their hours from home are at least as high in January 2022 as they are in April 2020. This finding is inconsistent with the fact that the COVID-19 restrictions implemented at the end of 2021 and the beginning of 2022 in response to the emergence of the Omicron variant were not as stringent as in previous COVID-19 waves (Dekker and Macdonald 2022).

As expected, the large increase in work from home that took place from February to April 2020 was associated with a substantial reduction in the number of passengers using public transit. During that period, the number of passengers using public transit fell by 13.5 units (where a unit equals 10 million passenger trips) while the percentage of Canadians working from home increased by 33.9 percentage points. Thus, each percentage point increase in work from home was associated with a reduction of 0.40 units (-13.5 divided by 33.9) in demand for public transit. Subsequent periods during which the wording of the current work location question remained unchanged also display a sizable reaction (in absolute value) of the demand for public transit to movements in work from home. For example, this responsiveness equals -0.24 from April to September 2020 and -0.39 from January to September 2021.

In contrast, a much smaller responsiveness of the demand for public transit is observed from December 2021 to January 2022, a period during which the LFS transitioned to the second version of the current work location question (C2). During this period, the number of passengers using public transit fell marginally (by 1.7 units) even though work from home appears to have increased by 19.2 percentage points. Therefore, each percentage point increase in work from home was associated with a reduction of only 0.09 units (-1.7 divided by 19.2) in demand for public transit. This responsiveness of the demand for public transit to work from home (-0.09) is more than four times lower—in absolute value—than that recorded from February to April 2020 (-0.40) and that recorded from January to September 2021 (-0.39). Along with the counter-intuitive finding that the incidence of work from home was as high in January 2022 as it was in April 2020, this pattern suggests that the January 2022 estimate of the percentage of Canadians working most of their hours from home is biased upwards.

The responsiveness of the demand for public transit to work from home is even lower in absolute value (-0.05) from April to May 2022, a period that corresponds to the transition to the third version of the question on current work location (C3) and during which the incidence of work from home appears to have fallen by almost 8 percentage points. The large reduction in the incidence of work from home observed between these two months is inconsistent with the fact that the demand for public transit barely changed during this period.

Taken together, the patterns observed in Chart 1 are consistent with the hypothesis that estimates of the incidence of work from home are biased upwards from January to April 2022.

Chart 2 Work from home and COVID-19 Restriction Index, Ontario, January 2020 to July 2022

Data table for Chart 2 
Data table for Chart 2
Table summary
This table displays the results of Data table for Chart 2. The information is grouped by Months and years (appearing as row headers), COVID-19 Restriction index and Work from home, calculated using index value and (percent) units of measure (appearing as column headers).
Months and years COVID-19 Restriction index Work from home
index value (percent)
2020
January 1.5 7.1
February 5.8 7.1
March 34.2 26.8
April 82.6 46.5
May 73.9 44.8
June 64.1 36.2
July 54.7 34.0
August 42.6 31.4
September 33.3 30.5
October 43.4 30.5
November 51.1 31.5
December 68.2 32.4
2021
January 74.8 39.8
February 70.6 36.9
March 61.9 33.9
April 74.7 37.5
May 82.1 37.7
June 62.3 34.8
July 40.9 30.5
August 33.9 29.0
September 27.7 28.6
October 28.6 28.0
November 36.7 27.2
December 37.7 28.1
2022
January 51.1 49.5
February 30.8 41.4
March 15.1 39.5
April 13.2 35.6
May 13.2 28.7
June 13.2 28.6
July 13.2 28.8

Chart 2 provides additional evidence supporting this hypothesis. From February to April 2020, the COVID-19 Restriction Index in Ontario increased by 76.7 points. Meanwhile, the percentage of workers working most of their hours from home in this province increased by 39.4 percentage points, thereby implying a responsiveness of work from home to COVID-19 restrictions of 0.51 (39.4 divided by 76.7).

In contrast, the responsiveness observed from December 2021 to January 2022 (1.60) and from March to May 2022 (5.77) is much higher. Furthermore, the incidence of work from home in Ontario fell substantially from April to May 2022 (-6.8 percentage points) even though the COVID‑19 Restriction Index remained unchanged during these two months. Similar patterns are observed in Quebec.

In sum, the changes in the incidence of work from home observed from December 2021 to January 2022 and from April to May 2022—two pairs of months associated with transitions to new questions on work location—are not consistent with either data on public transit ridership or data on COVID-19 restrictions. These changes suggest that the percentages of Canadians working most of their hours from home are biased upwards from January to April 2022. This hypothesis is investigated more formally in the next section.

4 Regression results

The percentage of Canadian workers working most of their hours from home in province p MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGWbaaaa@370B@ during month m MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGTbaaaa@3708@ of year t MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWG0baaaa@370F@ , WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@ , is modelled as

WF H pmt =  θ r +  β 1 FEASIBILITY_pmt+ β 2 COVI D mt + β 3 RESTRICTION S pmt + MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaak8qacqGH9aqpcaGGGcGaeqiUde3damaaBaaaleaape GaamOCaaWdaeqaaOWdbiabgUcaRiaacckacqaHYoGypaWaaSbaaSqa a8qacaaIXaaapaqabaGcpeGaamOraiaadweacaWGbbGaam4uaiaadM eacaWGcbGaamysaiaadYeacaWGjbGaamivaiaadMfacaGGFbGaamiC aiaad2gacaWG0bGaey4kaSIaeqOSdi2damaaBaaaleaapeGaaGOmaa WdaeqaaOWdbiaadoeacaWGpbGaamOvaiaadMeacaWGebWdamaaBaaa leaapeGaamyBaiaadshaa8aabeaak8qacqGHRaWkcqaHYoGypaWaaS baaSqaa8qacaaIZaaapaqabaGcpeGaamOuaiaadweacaWGtbGaamiv aiaadkfacaWGjbGaam4qaiaadsfacaWGjbGaam4taiaad6eacaWGtb WdamaaBaaaleaapeGaamiCaiaad2gacaWG0baapaqabaGcpeGaey4k aScaaa@6E9E@

β 4 RESTRICTION S pmt * D pmt + u pmt    MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqaHYoGypaWaaSbaaSqaa8qacaaI0aaapaqabaGcpeGaamOuaiaa dweacaWGtbGaamivaiaadkfacaWGjbGaam4qaiaadsfacaWGjbGaam 4taiaad6eacaWGtbWdamaaBaaaleaapeGaamiCaiaad2gacaWG0baa paqabaGcpeGaaiOkaiaadseapaWaaSbaaSqaa8qacaWGWbGaamyBai aadshaa8aabeaak8qacqGHRaWkcaWG1bWdamaaBaaaleaapeGaamiC aiaad2gacaWG0baapaqabaGcpeGaaiiOaiaacckaaaa@5266@ (1)

Where FEASIBILITY_pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGgbGaamyraiaadgeacaWGtbGaamysaiaadkeacaWGjbGaamit aiaadMeacaWGubGaamywaiaac+facaWGWbGaamyBaiaadshaaaa@42C6@   denotes the percentage of workers who hold jobs that can be done from home (Deng et al. 2020) and COVI D mt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGdbGaam4taiaadAfacaWGjbGaamira8aadaWgaaWcbaWdbiaa d2gacaWG0baapaqabaaaaa@3C6A@   is a binary indicator that equals 1 from March 2020 onwards or 0 otherwise. The term RESTRICTION S pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGsbGaamyraiaadofacaWGubGaamOuaiaadMeacaWGdbGaamiv aiaadMeacaWGpbGaamOtaiaadofapaWaaSbaaSqaa8qacaWGWbGaam yBaiaadshaa8aabeaaaaa@4336@   is Statistics Canada’s COVID-19 Restriction Index for the whole population of a given province during month m MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGTbaaaa@3708@   or year t MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWG0baaaa@370F@ . Clarke et al. (2022) argue that,

“At a value of 41, restrictions tend to go from being an inconvenience (e.g., wear a mask and gather only in smaller groups) to being a burden (e.g., in-person schooling is cancelled, nonessential retail and personal services are closed, or stay-at-home orders are issued). The value of 41, therefore, represents a point at which restrictions tend to become more binding for personal and business activities, and thus represents a level above which increases in restrictions can lead to more noticeable changes in activity.”

For this reason, equation (1) allows the effect of COVID-19 restrictions on work from home to increase when RESTRICTION S pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGsbGaamyraiaadofacaWGubGaamOuaiaadMeacaWGdbGaamiv aiaadMeacaWGpbGaamOtaiaadofapaWaaSbaaSqaa8qacaWGWbGaam yBaiaadshaa8aabeaaaaa@4336@   is equal to or greater than 41, in which case the binary indicator D pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGebWdamaaBaaaleaapeGaamiCaiaad2gacaWG0baapaqabaaa aa@3A1A@   is equal to 1. Last, θ r MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqaH4oqCpaWaaSbaaSqaa8qacaWGYbaapaqabaaaaa@391E@   is a vector of region fixed effects that allows for the possibility that workers in the Atlantic provinces—the reference group in this vector—might have weaker preferences for work from home than their counterparts in large provinces (Mehdi and Morissette 2021), possibly because of shorter commuting distances.

The first column of Table 1 shows the results obtained when estimating equation (1) over the period from January 2020 to July 2022 (N=310=10 provinces times 31 months). In this model, the parameter estimate for RESTRICTION S pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGsbGaamyraiaadofacaWGubGaamOuaiaadMeacaWGdbGaamiv aiaadMeacaWGpbGaamOtaiaadofapaWaaSbaaSqaa8qacaWGWbGaam yBaiaadshaa8aabeaaaaa@4336@   is not statistically significant. Neither Quebec nor British Columbia appears to differ significantly from the Atlantic provinces regarding the incidence of work from home. The adjusted R-squared of the model equals 0.70.


Table 1
Regression results
Table summary
This table displays the results of Regression results. The information is grouped by Variables (appearing as row headers), Work from home and period, Model 1 , Model 2, Model 3, Model 4, Initial estimates January 2020 to July 2022, Initial estimates January 2020 to July 2022, New estimates
January 2020 to July 2022, Initial estimates January 2020 to July 2022 except January to April 2022, Parameter estimates and Significance level, calculated using number units of measure (appearing as column headers).
Variables Work from home and period
Model 1 Model 2 Model 3 Model 4
Initial estimates January 2020 to July 2022 Initial estimates January 2020 to July 2022 New estimates
January 2020 to July 2022
Initial estimates January 2020 to July 2022 except January to April 2022
Parameter estimates Significance level Parameter estimates Significance level Parameter estimates Significance level Parameter estimates Significance level
number
Intercept -29.00 <0.0001 -12.63 <0.0001 -11.17 <0.0001 -10.73 0.0003
FEASIBILITY_pmt 0.96 <0.0001 0.47 <0.0001 0.43 <0.0001 0.42 <0.0001
COVID_mt 13.48 <0.0001 8.80 <0.0001 8.85 <0.0001 9.20 <0.0001
RESTRICTIONS_pmt 0.03 0.3309 0.15 <0.0001 0.15 <0.0001 0.13 <0.0001
RESTRICTIONS_pmt*D_pmt 0.07 0.0010 0.03 0.0227 0.04 0.0033 0.05 0.0005
Provinces
Atlantic provinces (reference group) Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable
Quebec 0.82 0.4823 3.58 <0.0001 3.43 <0.0001 3.48 <0.0001
Ontario 2.87 0.0386 6.85 <0.0001 6.90 <0.0001 6.98 <0.0001
Manitoba -1.60 0.0868 -0.98 0.10 -1.33 0.0168 -1.36 0.0335
Saskatchewan 0.27 0.7742 0.18 0.77 0.35 0.5322 0.24 0.7131
Alberta 2.09 0.0348 4.07 <0.0001 4.26 <0.0001 4.22 <0.0001
British Columbia 0.95 0.4010 4.12 <0.0001 4.51 <0.0001 4.52 <0.0001
Month and year
December 2021 Note ...: not applicable Note ...: not applicable -2.14 0.0238 -1.98 0.0279 Note ...: not applicable Note ...: not applicable
January 2022 Note ...: not applicable Note ...: not applicable 13.34 <0.0001 -0.02 0.9846 Note ...: not applicable Note ...: not applicable
February 2022 Note ...: not applicable Note ...: not applicable 11.41 <0.0001 -0.07 0.9399 Note ...: not applicable Note ...: not applicable
March 2022 Note ...: not applicable Note ...: not applicable 11.88 <0.0001 0.10 0.9127 Note ...: not applicable Note ...: not applicable
April 2022 Note ...: not applicable Note ...: not applicable 8.36 <0.0001 0.17 0.8585 Note ...: not applicable Note ...: not applicable
May 2022 Note ...: not applicable Note ...: not applicable 1.15 0.2545 1.18 0.2185 Note ...: not applicable Note ...: not applicable
Number of observations 310 Note ...: not applicable 310 Note ...: not applicable 310 Note ...: not applicable 270 Note ...: not applicable
Adjusted R-squared 0.70 Note ...: not applicable 0.88 Note ...: not applicable 0.87 Note ...: not applicable 0.86 Note ...: not applicable

The second column of Table 1 adds to equation (1) a vector of monthly indicators for the six months covering the period from December 2021 to May 2022. Adding this vector substantially improves the fit of the model: the adjusted R-squared increases from 0.70 to 0.88. The parameter estimate for RESTRICTION S pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGsbGaamyraiaadofacaWGubGaamOuaiaadMeacaWGdbGaamiv aiaadMeacaWGpbGaamOtaiaadofapaWaaSbaaSqaa8qacaWGWbGaam yBaiaadshaa8aabeaaaaa@4336@   increases from 0.03 to 0.15 and now becomes statistically significant. Both Quebec and British Columbia now display, all else equal, a greater incidence of work from home than the Atlantic provinces.Note The binary indicators for the months of January to April 2022 are all statistically significant and suggest that—even after controlling for telework feasibility and COVID-19 restrictions—the incidence of work from home during these months was between 8 and 13 percentage points higher than during the months outside the period from December 2021 to May 2022. This finding indicates that the transition to the second version of the LFS questions on work from home (from U1 and C1 to U2 and C2) led to overestimating the percentage of Canadians working most of their hours from home. In contrast, the binary indicator for May 2022 is not statistically significant, thereby suggesting that the transition to the third version (U3 and C3) of the LFS questions on work from home appears to have solved this problem.

In light of this evidence, a sensible approach is to re-estimate equation (1) for the period from January 2020 to July 2022 with the exception of the four months (January, February, March and April 2022) for which the percentages of Canadians working most of their hours from home are assumed to be overestimated. This yields a total of 270 observations (i.e., 10 provinces times 27 months).

The results are shown in the fourth column of Table 1. As expected, greater telework feasibility and more stringent COVID-19 restrictions are associated with a greater incidence of work from home, especially when these restrictions equal 41 or more. The COVID-19 binary indicator also suggests, all else equal, a greater incidence of work from home starting in March 2020. Lastly, the four largest provinces (Quebec, Ontario, Alberta and British Columbia) all display—as expected—larger percentages of individuals working most of their hours from home than the Atlantic provinces, even after controlling for telework feasibility. Along with the adjusted R‑squared, all parameter estimates are similar to those of the second column, in which equation (1) augmented with monthly indicators for the six months of December 2021 to May 2022 was estimated for the entire period from January 2020 to July 2022.

To assess the predictive performance of the model, the parameter estimates shown in the fourth column of Table 1 are multiplied by the regressors of equation (1). This exercise is done for all 310 observations associated with the period from January 2020 to April 2022, including the months of January to April 2022. The predicted values of WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@   are then compared with its actual values for each province.

Charts 3 and 4 show the results of this exercise for Quebec and Ontario. For both provinces and most months of the period from January 2020 to July 2022, predicted values of WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@   track reasonably well its actual values. This is not the case, however, for the months of January to April 2022. For these four months, predicted values of WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@   substantially underestimate its actual values. For example, the predicted value of WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@   for Ontario in January 2022 equals 34.4%, which is 15.1 percentage points lower than the actual value of 49.5%. In contrast, predicted values of WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@   are fairly similar to its actual values from May to July 2022, a period during which the third version of the current work location measure (C3) was administered.

Chart 3 Percentage of workers working most of their hours from home, predicted values and actual values, Quebec

Data table for Chart 3 
Data table for Chart 3
Table summary
This table displays the results of Data table for Chart 3. The information is grouped by Months and years (appearing as row headers), Actual values and Predicted values, calculated using percent units of measure (appearing as column headers).
Months and years Actual values Predicted values
percent
2020
January 6.7 9.7
February 6.7 9.9
March 23.4 23.8
April 40.1 37.3
May 34.5 34.6
June 29.1 28.4
July 27.3 22.6
August 25.4 21.9
September 24.3 23.2
October 26.7 29.6
November 27.1 29.6
December 27.7 30.7
2021
January 34.9 33.7
February 30.8 31.0
March 28.5 29.7
April 29.3 33.5
May 28.4 33.0
June 26.6 26.9
July 26.5 22.3
August 23.6 22.0
September 22.2 23.8
October 23.1 23.8
November 23.0 23.6
December 23.4 24.2
2022
January 44.8 29.8
February 39.7 23.7
March 34.0 22.1
April 33.5 21.5
May 23.9 21.4
June 23.3 21.3
July 24.8 19.9

Chart 4 Percentage of workers working most of their hours from home, predicted values and actual values, Ontario

Data table for Chart 4 
Data table for Chart 4
Table summary
This table displays the results of Data table for Chart 4. The information is grouped by Months and years (appearing as row headers), Actual values and Predicted values, calculated using percent units of measure (appearing as column headers).
Months and years Actual values Predicted values
percent
2020
January 7.1 14.8
February 7.1 15.0
March 26.8 28.4
April 46.5 41.1
May 44.8 38.5
June 36.2 35.4
July 34.0 32.9
August 31.4 30.2
September 30.5 28.1
October 30.5 31.6
November 31.5 33.2
December 32.4 36.5
2021
January 39.8 38.3
February 36.9 37.4
March 33.9 35.6
April 37.5 37.7
May 37.7 39.4
June 34.8 35.6
July 30.5 28.4
August 29.0 27.3
September 28.6 28.3
October 28.0 28.1
November 27.2 29.2
December 28.1 29.1
2022
January 49.5 34.4
February 41.4 28.4
March 39.5 25.6
April 35.6 26.0
May 28.7 25.9
June 28.6 26.1
July 28.8 24.7

The patterns shown in charts 3 and 4 are observed for all provinces. They confirm the hypothesis that the percentages of workers working most of their hours from home are overestimated for the months of January to April 2022.Note

Last, the third column of Table 1 replicates the second column but uses a new series of estimates of the incidence of work from home, instead of the initial estimates. The new series corresponds to the predicted value of WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@   for the four months of January to April 2022 and to the initial estimates for other months. Parameter estimates from the third column are similar to those of the second and fourth columns with an important exception: all binary indicators for the four months of January to April 2022 are now close to zero and statistically insignificant. This finding shows that after controlling for telework feasibility and COVID-19 restrictions, the new estimates do not generate—contrary to initial estimates—unexpected differences in the incidence of work from home for these four months.

5 New estimates

5.1 New estimates by province

To help users assess the evolution of work from home from 2020 to 2022, the upper panel of Table 2 provides, for each province, estimates of the percentage of workers working most of their hours from home for the period from January 2020 to December 2022. The values for January and February 2020 are drawn from the 2016 Census of Population, while the values from April 2020 onwards come from the LFS. The values for March 2020 are a simple average of the February and April 2020 values. In all cases, the sample consists of individuals aged 15 to 69 who were working (i.e., who were not absent) during the census or LFS reference week and who were not full-time members of the Armed Forces. Both employees and self-employed individuals are included in the estimates.


Table 2
Percentage of workers working most of their hours from home, by province, January 2020 to December 2022
Table summary
This table displays the results of Percentage of workers working most of their hours from home. The information is grouped by Month and year (appearing as row headers), Province, All provinces, Newfoundland and Labrador, Prince Edward Island, Nova
Scotia, New Burnswick, Quebec, Ontario, Manitoba, Saskatchewan, Alberta and British Columbia, calculated using percent units of measure (appearing as column headers).
Month and year Province All provinces
Newfoundland and Labrador Prince Edward Island Nova
Scotia
New Burnswick Quebec Ontario Manitoba Saskatchewan Alberta British Columbia
percent
2020
January 4.6 6.7 6.5 5.6 6.7 7.1 6.6 10.2 7.5 8.3 7.2
February 4.6 6.7 6.5 5.6 6.7 7.1 6.6 10.2 7.5 8.3 7.2
March 16.0 21.3 20.6 21.2 23.4 26.8 19.2 22.4 23.1 22.4 24.2
April 27.3 36.0 34.7 36.8 40.1 46.5 31.8 34.7 38.6 36.6 41.1
May 28.3 30.7 31.1 32.2 34.5 44.8 28.4 30.0 33.0 35.4 37.8
June 18.6 23.6 26.6 19.6 29.1 36.2 20.6 22.1 27.8 29.0 30.6
July 15.7 21.1 22.5 17.6 27.3 34.0 18.2 16.8 26.1 25.6 28.4
August 13.1 18.8 22.8 16.8 25.4 31.4 16.8 15.5 21.9 24.3 26.2
September 12.6 15.9 18.9 17.3 24.3 30.5 16.1 13.7 22.6 23.9 25.4
October 12.2 18.4 19.5 17.5 26.7 30.5 16.5 15.0 21.4 24.1 25.9
November 11.8 16.5 18.7 18.7 27.1 31.5 20.9 15.5 23.6 26.4 27.2
December 12.4 19.8 20.0 18.8 27.7 32.4 23.7 18.4 27.1 26.5 28.3
2021
January 14.0 16.5 19.6 20.3 34.9 39.8 23.8 19.2 28.2 27.5 33.1
February 26.8 16.6 20.2 20.5 30.8 36.9 22.5 19.9 25.6 27.0 30.9
March 23.0 15.8 20.4 17.8 28.5 33.9 21.9 19.2 25.2 28.0 29.1
April 15.5 17.5 20.7 17.5 29.3 37.5 21.2 20.7 25.8 29.4 30.7
May 18.3 15.4 30.3 16.8 28.4 37.7 21.0 17.7 28.2 25.4 30.6
June 15.0 15.1 23.6 14.6 26.6 34.8 20.7 16.1 22.4 24.0 27.9
July 17.0 14.4 20.0 16.3 26.5 30.5 16.9 14.1 22.4 23.1 25.9
August 16.6 13.6 20.4 14.8 23.6 29.0 16.9 13.3 19.5 21.7 24.1
September 12.9 16.8 18.5 16.1 22.2 28.6 15.7 12.6 21.7 21.8 23.8
October 11.6 17.1 16.9 18.5 23.1 28.0 15.0 11.6 22.3 21.8 23.7
November 10.8 15.0 17.6 21.1 23.0 27.2 16.8 12.9 21.3 22.2 23.5
December 11.1 17.0 18.7 18.0 23.4 28.1 17.7 12.4 21.1 21.4 23.8
2022
January 37.1 35.8 36.3 34.6 44.8 49.5 34.1 27.2 35.6 37.5 43.0
February 26.1 31.3 30.3 33.1 39.7 41.4 32.2 27.0 31.5 32.4 37.2
March 23.3 32.8 34.1 27.5 34.0 39.5 28.5 22.2 32.1 31.5 34.7
April 19.3 25.9 25.0 23.0 33.5 35.6 25.4 20.5 27.5 28.8 31.7
May 13.9 16.4 16.3 16.0 23.9 28.7 16.1 13.1 20.4 21.4 24.0
June 14.7 15.4 19.2 16.3 23.3 28.6 15.1 12.9 19.3 22.1 23.8
July 14.0 17.4 17.6 18.1 24.8 28.8 16.1 13.5 18.2 22.5 24.2
August 12.9 17.6 18.6 17.2 23.0 28.0 16.9 12.6 18.0 22.3 23.4
September 12.0 19.7 16.7 16.2 22.6 26.7 13.7 11.1 16.4 20.6 22.1
October 13.1 19.7 15.6 15.3 21.4 26.3 13.8 11.5 16.5 21.5 21.8
November 12.9 17.2 18.1 18.5 21.1 26.3 14.8 13.3 17.1 20.5 21.9
December 13.2 19.6 19.0 17.9 22.5 26.4 15.0 11.6 17.2 19.7 22.1
New estimates for January to April 2022
January 23.4 22.8 22.6 23.7 29.8 34.4 18.6 15.7 22.8 24.8 28.8
February 21.2 19.0 19.2 19.3 23.7 28.4 16.9 15.1 22.9 23.0 24.6
March 15.3 17.2 16.8 17.3 22.1 25.6 15.3 15.2 20.2 21.6 22.4
April 13.8 15.9 16.4 16.2 21.5 26.0 14.7 15.5 20.0 21.8 22.4

The lower panel presents new provincial estimates for the months of January to April 2022. These new provincial estimates simply equal the predicted values of WF H pmt MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGxbGaamOraiaadIeapaWaaSbaaSqaa8qacaWGWbGaamyBaiaa dshaa8aabeaaaaa@3BC5@   obtained from the fourth column of equation (1) for these months. New estimates are also presented for all provinces taken together. For a given month, these new national estimates equal a weighted average of the new provincial estimates, where the weights equal the share of workers of a given province in the overall workforce during a given month. For the period from January to April 2022, users are advised to use the new estimates instead of the initial estimates.

5.2 New estimates by other dimensions

The model-based estimates obtained from equation (1) use telework feasibility estimates developed by Deng et al. (2020) and Statistics Canada’s COVID-19 Restriction Index. While telework feasibility estimates can be disaggregated by other dimensions (such as age, sex, education or industry), this is not the case for Statistics Canada’s Restriction Index, which is built only at the provincial level. As a result, the modelling approach used in Section IV cannot be replicated for these other dimensions.

Nevertheless, it is possible to produce new estimates of the incidence of work from home for the period from January to April 2022 for other dimensions (such as age, sex, education or industry), using a simple adjustment procedure outlined in Appendix 2.

Last, users should be aware that the work location questions introduced from 2020 to 2022 by Statistics Canada in the LFS were asked only to a subset of the LFS sample. As a result, the ability to disaggregate data on work from home by various dimensions is more limited than it would be with the entire LFS sample.

6 Concluding remarks

The substantial increase in work from home triggered by the COVID-19 pandemic has led to growing demand for information on the incidence of telework. Using public transit data and multivariate analyses that take advantage of estimates of telework feasibility and of Statistics Canada’s COVID-19 Restriction Index, this study shows that the percentage of Canadians working most of their hours from home was likely overestimated during the months of January, February, March and April 2022. New estimates are offered at the provincial and national levels for these four months. In addition, a simple adjustment procedure is proposed for disaggregating numbers by other dimensions such as age, sex, education or industry.

Users should keep in mind that the issues documented in this study affect only the concept of “working most hours from home.” They do not affect the concepts of “hybrid work arrangements” and “working exclusively from home,” two useful concepts that have been measured by the LFS since January 2022. With the updated estimates provided in this study and the data that will be collected by the LFS going forward, users will be able to assess the evolution of the percentage of Canadians working most of their hours from home from 2020 onwards.

7 Appendix 1: Labour Force Survey questions on work from home

First set of questions: April 2020 to December 2021

U1. Which of the following best describes #{__DT_NAMEE}’s usual place of work at #{__DT_HIS_HER} main job or business?

Exclude any recent changes related to COVID-19.

Would you say:

  1. Work at a fixed location outside the home
  2. Work outside the home with no fixed location e.g., driving, making sales calls
  3. Work at home

C1. Last week, in which of these locations did #{__DT_NAMEE} work the most hours?

Last week is from #{__DT_REFWEEK_E}.

Would you say:

  1. At a fixed location outside the home
  2. Outside the home with no fixed location e.g. driving, making sales calls
  3. At home
  4. Absent from work

Second set of questions: January to April 2022

U2. At the present time, in which of the following locations does #{__DT_NAMEE} usually work as part of #{__DT_HIS_HER} main job or business?

Select all that apply.

  1. At a fixed location outside the home
  2. Outside the home with no fixed location
    1. e.g., driving, visiting clients
  3. At home
    1. Include all work done at the same address as home, including farm work

C2. Last week, what proportion of #{__DT_HIS_HER} work hours did #{__DT_NAMEE} work at home as part of #{__DT_HIS_HER} main job or business?

Last week is from #{__DT_REFWEEK_E}

Include all work done at the same address as home, including farm work.

Would you say:

  1. All hours
  2. More than half, but not all
  3. One quarter to half
  4. Less than a quarter
  5. No hours

8 Appendix 2: Approximating the incidence of work from home for other dimensions

It is possible to produce new estimates of the incidence of work from home for the period from January 2022 to April 2022 for other dimensions (e.g., age, sex, industry and education), using a simple adjustment procedure. For example, one can approximate the percentage of individuals working most of their hours from home by industry in January (April) 2022 by using the following steps:

Step 1: Compute ΔNWF H t MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqqHuoarcaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGa amiDaaWdaeqaaaaa@3C17@ , the overall change in the number of individuals working most of their hours from home observed from December 2021 to January (April) 2022, based on the new estimates for the first four months of 2022. Table 3 shows that the overall change observed from December 2021 to January (April) 2022 amounts to 567,800 (-282,700) workers.


Table 3
Estimated number of individuals working most of their hours from home, Canada, December 2021 to April 2022
Table summary
This table displays the results of Estimated number of individuals working most of their hours from home. The information is grouped by Month and year (appearing as row headers), Number (thousands) and Change from December 2021 to month m (thousands) (appearing as column headers).
Month and year Number (thousands) Change from December 2021 to month m (thousands)
December 2021 4,198.2 Note ...: not applicable
January 2022 4,766.0 567.8
February 2022 4,297.2 99.0
March 2022 3,836.7 -361.5
April 2022 3,915.5 -282.7

Step 2: Disaggregate ΔNWF H t MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqqHuoarcaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGa amiDaaWdaeqaaaaa@3C17@ , the overall change in the number of individuals working most of their hours from home, by industry, using the industry shares of work from home observed in December 2021. For example, to compute the change in the number of workers working from home in industry i from December 2021 to January (April) 2022, use the following equation:

ΔNWF H it = ΔNWF H t *SHAR E i_Dec 2021    MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqqHuoarcaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGa amyAaiaadshaa8aabeaak8qacqGH9aqpcaGGGcGaeuiLdqKaamOtai aadEfacaWGgbGaamisa8aadaWgaaWcbaWdbiaadshaa8aabeaak8qa caGGQaGaam4uaiaadIeacaWGbbGaamOuaiaadweapaWaaSbaaSqaa8 qacaWGPbGaai4xaiaadseacaWGLbGaam4yaiaacckacaaIYaGaaGim aiaaikdacaaIXaaapaqabaGcpeGaaiiOaiaacckaaaa@5556@ (2)

Where ΔNWF H it MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqqHuoarcaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGa amyAaiaadshaa8aabeaaaaa@3D05@   equals the change in the number of workers working from home in industry  i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@   from December 2021 to January (April) 2022 and SHAR E i_Dec 2021 MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGtbGaamisaiaadgeacaWGsbGaamyra8aadaWgaaWcbaWdbiaa dMgacaGGFbGaamiraiaadwgacaWGJbGaaiiOaiaaikdacaaIWaGaaG Omaiaaigdaa8aabeaaaaa@42FA@   equals the percentage of workers working from home employed in industry  i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@   in December 2021.

Step 3: Add ΔNWF H it MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqqHuoarcaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGa amyAaiaadshaa8aabeaaaaa@3D05@   to the number of workers working from home in industry  i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@   in December 2021 ( NWF H i_Dec2021 MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGaamyAaiaa c+facaWGebGaamyzaiaadogacaaIYaGaaGimaiaaikdacaaIXaaapa qabaaaaa@4111@ ) to obtain NWF H it MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGaamyAaiaa dshaa8aabeaaaaa@3B9F@ , the estimated number of workers working from home in industry  i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@   in January (April) 2022:

NWF H it = ΔNWF H it +NWF H i_Dec2021   MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGaamyAaiaa dshaa8aabeaak8qacqGH9aqpcaGGGcGaeuiLdqKaamOtaiaadEfaca WGgbGaamisa8aadaWgaaWcbaWdbiaadMgacaWG0baapaqabaGcpeGa ey4kaSIaamOtaiaadEfacaWGgbGaamisa8aadaWgaaWcbaWdbiaadM gacaGGFbGaamiraiaadwgacaWGJbGaaGOmaiaaicdacaaIYaGaaGym aaWdaeqaaOWdbiaacckaaaa@5205@ (3)

Step 4: Divide NWF H it MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGobGaam4vaiaadAeacaWGibWdamaaBaaaleaapeGaamyAaiaa dshaa8aabeaaaaa@3B9F@   by the number of individuals working in industry  i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@   in January (April) 2022, to obtain the estimated percentage of workers working from home in industry  i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@   in January (April) 2022.

This four-step procedure ensures that the sum of disaggregated numbers equals the overall numbers. It can be used for other dimensions such as age, sex or education. Its main limitation is that it assumes if, for example, 10% of all teleworkers worked in industry A in December 2021, then the change in the number of teleworkers observed from December 2021 to January (April) 2022 in industry A will equal 10% of the overall change in the number of teleworkers during that period. This assumption may provide a good approximation in some cases but a poor one in other cases. Users should keep this limitation in mind.

References

Clarke, S., J. Dekker, N. Habli, R. Macdonald and C. McCormack, C. 2022. Measuring the Correlation Between COVID-19 Restrictions and Economic Activity. Analytical Studies: Methods and References, no. 40. Statistics Canada Catalogue no. 11-633-X. Ottawa: Statistics Canada. https://www150.statcan.gc.ca/n1/en/pub/11-633-x/11-633-x2022003-eng.pdf?st=bOzLbU1b

Dekker, J., and R. Macdonald. 2022. “COVID-19 restrictions index update,” Economic and Social Reports, Statistics Canada Catalogue no. 36-28-0001. https://www150.statcan.gc.ca/n1/pub/36-28-0001/2022008/article/00002-eng.htm

Deng, Z., D. Messacar, and R. Morissette. 2020. “Running the Economy Remotely: Potential for Working from Home During and After COVID-19.” StatCan COVID-19: Data to Insights for a Better Canada, no. 26. Statistics Canada Catalogue no. 45280001. Ottawa: Statistics Canada.

Mehdi, T., and R. Morissette. 2021. “Working from Home: Productivity and Preferences.” StatCan COVID-19: Data to Insights for a Better Canada, no. 12. Statistics Canada Catalogue no. 45280001. Ottawa: Statistics Canada.

Statistics Canada. 2022. “Commuting in Canada during the COVID-19 pandemic: What changed from 2016 to 2021?”, https://www150.statcan.gc.ca/n1/pub/11-627-m/11-627-m2022081-eng.htm


Date modified: