The Impact of the COVID-19 Pandemic on Self-employment Transitions in Canada

Acknowledgments

The author would like to thank Amélie Lafrance-Cooke, Huju Liu, Wulong Gu, Marc Frenette, Ryan Macdonald, Lyming Huang and Mamour Fall for their helpful comments. The author would also like to thank Evelyne Bougie, Martha Patterson and participants of the Data Interpretation Workshop at Statistics Canada. Many thanks also to Winnie Chan for their assistance in preparing the data.

Abstract

The job transitions of individuals into and out of unincorporated self-employment (USE) and incorporated business ownership (IBO) during the COVID-19 pandemic are less documented in the context of the Canadian labour market. This study uses the Canadian Employer–Employee Dynamics Database, covering the period from 2017 to 2021, to quantify the dynamics of entry into and exit from self-employment  around the COVID-19 pandemic . It also documents how such dynamics vary by workers’ characteristics.

The study presents information on entry into and exit from self-employment as well as modelled estimates of the probability of transition. Analysis and models take into consideration province, industry and year, but not job support programs.

The findings suggest a decrease in self-employment, as evidenced by both lower entry and higher exit rates with the onset of the pandemic. This decline was greater for USE than for IBO. In addition, the results reveal some significant differences in the magnitude of change by worker characteristics. For example, with the onset of the pandemic, the largest changes in self-employment flows were observed among women, low-income earners, immigrants, workers aged 24 or younger, those with more than four children and those with a self-employed spouse.

This study underscores how workers’ characteristics relate to transitions into and out of self-employment around the COVID-19 pandemic. In doing so, it provides insights relevant for policymakers on employment transitions during economic downturns.

1 Introduction

Entrepreneurial activity, including self-employment, is an important driver of economic growth, productivity and innovation. Entrepreneurs bring new ideas, create jobs and increase competition (Schumpeter, 1942). While many studies have linked entrepreneurship to self-employment, the latter concept refers to a more diverse group, ranging from large, incorporated businesses to small, unincorporated businesses such as part-time food delivery (Schoar, 2010; Grekou & Liu, 2018). In Canada, the share of self-employment, including unincorporated self-employment (USE) and incorporated business ownership (IBO), in total employment declined from 2000 to 2023 (see Figure A1 in the appendix). Looking at the determinants of individuals’ self-employment flows, previous studies suggest that this is attributable to insufficient and inadequate opportunities into (or out of) the paid sector, low unemployment, high wages, an aging population, and a more complex business environment (Leung & Robinson, 2011). This study contributes to this literature by quantifying workers’ transitions from paid employment (PE) to self-employment and vice versa during the COVID-19 pandemic.

The pandemic presented opportunities and challenges for self-employment. On the one hand, job loss and industry disruptions because of COVID-19 (Statistics Canada, 2020) may have pushed some individuals into self-employment to maintain their livelihoods. Furthermore, COVID-19 also created pull factors such as remote work opportunities and rapid technological developments that may stimulate entrepreneurial initiatives. On the other hand, despite various programs^Note  offered to support workers and businesses, restrictive health measures and physical distancing policies may have limited opportunities for entrants and small businesses to grow, leading to closures (Leung & Liu, 2022).^Note  For example, in 2020, there were 5.4% fewer new businesses than the previous year (Lafrance-Cooke & Leung, 2024).

Despite government efforts to stimulate the labour market during the pandemic, several studies have documented lower labour market outcomes for various classes of worker, including paid employees, self-employed workers and business owners. Analyzing self-employment in Canada, Beland, Fakorede and Mikola (2020) found a large decrease in the number of business owners, employees and hours worked from February to July 2020. In the same vein, Lemieux, Milligan, Schirle and Skuterud (2020) documented that the pandemic induced a decline in aggregate weekly hours worked and employment from February to April 2020. Additionally, the pandemic had a disproportionate negative impact on the labour outcomes of certain groups, such as women (Grekou & Lu, 2021), less educated individuals (Beland, Fakorede & Mikola, 2020), visible minorities (Tam, Sood, & Johnston, 2020a), and women and immigrant majority-owned businesses (Gueye, 2024). Similar results were obtained by Kurmann, Lalé and Ta (2020) and Fairlie, Couch and Xu (2020) for minorities and immigrants in the United States.

The literature lacks evidence on the job transitions into and out of USE and IBO around COVID-19. In other words, movement between these types of work, in terms of entry into and exit out of USE and IBO, remains understudied in the Canadian context around the pandemic. Furthermore, understanding how the profile of workers intersects with such transitions informs on entrepreneurship in the context of the pandemic.

This paper seeks to determine how the dynamics of the class of worker (PE, USE and IBO) changed from 2017 to 2021 at the individual worker level. Furthermore, it investigates the relationship between a worker’s characteristics in the years before the pandemic and their job transition choice, i.e., (1) from PE to either USE or IBO, or (2) from either USE or IBO to PE. Throughout the paper, the term “self-employment” is used to refer to both USE and IBO.

This article contributes to the literature in several ways. It provides short-term labour market effects of COVID-19 at the more disaggregated level of the individual rather than the sector and business levels, as in previous studies. In addition, this paper assesses the factors governing workers’ transition choices. While the pandemic may have played a triggering role in the transition choice from one job category to another, the fact remains that this transition varies according to various factors including industry, region and worker’s characteristics. This study contributes to the literature by examining the transition of paid employees into and out of self-employment around the pandemic according to the characteristics of workers.

While previous studies have looked at impacts of deteriorating economic conditions, including the financial crisis, on self-employment (LaRochelle-Côté & Gilmore, 2009; Lin, Picot & Compton, 2000), few studies have examined changes in self employment entry and exit around COVID-19. This paper aims to quantify the transitions around the pandemic. By doing so, it also contributes to the literature seeking to understand drivers and constraints for USE and IBO.

Finally, the paper provides insights for formulating targeted policies and programs to help workers according to their profiles and their transitions between different job categories.

The rest of the paper proceeds as follows. Section 2 describes the data and methodology used in this study. Section 3 outlines the main findings. Section 4 discusses the results and concludes the paper with some limitations and future research avenues.

2 Data and methodology

2.1 Data source

This study uses the Canadian Employer–Employee Dynamics Database (CEEDD), covering the 2017-to-2021 period, which makes it possible to observe and track individual sociodemographic and economic characteristics, including employment, over time. The CEEDD is a matched database between Canadian firms and workers created by linking several administrative tax files, including individual tax files (T1 Income Tax and Benefit Return), individual employment remuneration files (T4 Statement of Remuneration Paid), the T1 Family File, the Immigrant Landing File, and corporate (T2 Corporation Income Tax Return) and unincorporated business (T1 Business Declaration) tax files. Therefore, the data provide detailed information about paid and self-employed workers, such as age, gender, marital status, immigrant status, family composition, earnings from paid jobs, USE income, income from owned corporations and related industry of work.

The study population is limited to active workers who were either paid employees, unincorporated self-employed workers or business owners in each of the three years (2017 to 2019) preceding the pandemic. Unemployed people and individuals out of the labour force are excluded from this analysis.^Note  Thus, the paper aims to explore transitions around COVID-19 for a given class of worker. In line with the related literature (Grekou & Liu, 2018; Liu & Zhang, forthcoming), this analysis considers an individual as a paid employee, unincorporated self-employed worker or business owner in any year based on the primary source of income, i.e., the highest one. These categories of employment type are used to define job transition states from 2017 to 2021.

2.2 Methodology

The methodology of this paper is twofold and consists of descriptive analysis and a regression model.

First, the descriptive analysis includes statistics on job transitions by worker characteristics. These descriptive statistics are produced using three dimensions: demographic (age, gender, marital status and immigrant status), economic (economic sector and income quintile) and family composition (number of children and the class of worker of the spouse).

Second, the econometric analysis assesses the relationship between the pandemic and job transition, as well as how this transition correlated with worker characteristics. To do so, two equations are evaluated. Equation (1) below quantifies how much more or less likely transitions were to happen during the pandemic relative to before the pandemic. The following probit model is estimated:

$$\text{Pr}(JobTrans{i}_{it}^{k,s}=1)=\alpha COVI{D}_t+X_{ijt-1}\beta +Industr{y}_j+Provinc{e}_p+{\epsilon }_{it}\text{ (1)}$$

Where $JobTrans{i}_{it}^{k,s}$
is a latent variable equal to 1 if an individual i transitions from class of worker k (in year t-1) to s (in year t). Both k and s are either PE, USE or IBO. The vector $X_{ijt-1}$
accounts for individual and business characteristics (measured in the previous year, t-1), including dummies for age category, immigrant status, number of children, an indicator of whether the spouse is in self-employment, urban area, income and business size. The model also includes industry and province fixed effects, denoted by $Industr{y}_j$
and $Provinc{e}_p$
, respectively. The complete list of definitions and variable measurements is provided in the next subsection.

Using a dummy variable to compare the period before the pandemic and the pandemic period does not account for the pre-COVID-19 trend. In Chart A.1, a clear downward trend in self-employment was observed before COVID-19. Equation (1) could potentially overestimate the impact of COVID‑19 and mask part of the variation in the transition to self-employment attributable to economic cycles.

A robustness exercise is therefore performed and presented in the appendix using an event study methodology with dummies for each year, except for 2019 (the reference year) to estimate the impact of each year on job transitions. The resulting coefficients provide a comparison of the COVID-19 effect in 2020. Formally, the following model is evaluated:^Note 

$$\text{Pr}(JobTrans{i}_{it}^{k,s}=1)={\sum }_{k=2018}^{2020}Year\_Dummy\left(k\right)+X_{ijt-1}\beta +Industr{y}_j+Provinc{e}_p+{\epsilon }_{it}\text{ (2)}$$

Where $Year\_Dummy\left(k\right)$
is a binary variable equal to 1 for year k and 0 otherwise. The year 2019 is used as the reference category and therefore is dropped from equation (2).

To further examine how the short-term effect of COVID-19 varies with worker characteristics, equation (3) is estimated with additional interaction terms as in the following:

$$\text{Pr}(JobTrans{i}_{it}^{k,s}=1)=\alpha COVI{D}_t+X_{ijt-1}\beta +COVI{D}_t * X_{ijt-1}\delta +Industr{y}_j+Provinc{e}_p\text{ (3)}$$

For ease of interpretation, coefficients estimated in all equations are presented in terms of marginal effects instead of change in the z-score (standard normal distribution) associated with a one-unit change in the predictor variable.

2.3 Definitions and measurement of key variables

2.3.1 Dependent variable: Job transition status

Let $JobTrans{i}_{it}$
denote a binary variable of the job transition of an individual i in year t. The variable equals 1 if the class of worker in year t-1 differs from that in year t and 0 otherwise. Six job transition states are considered:

1. from PE to USE
2. from PE to IBO
3. from USE to PE
4. from USE to IBO
5. from IBO to PE
6. from IBO to USE.

It is important to distinguish between these transition states given that previous studies underline substantial differences in the entry and exit^Note  dynamics of both types of self-employment (Grekou & Liu, 2018; Liu & Zhang, forthcoming). Entrants into USE and IBO are captured by states (1) and (2), respectively, while states (3) and (5) are related to exits from USE and IBO, respectively. Transition states between types of self-employment (i.e., states [4] and [6]) are not examined in this paper. Instead, this study examines entrants from PE into both types of self-employment and exits out of both types of self-employment to PE. Entrants into USE or IBO are defined as those who are unincorporated self-employed workers or business owners in the current year but who were not in the previous year. Exits, conversely, are those who were unincorporated self-employed workers or business owners in the previous year but who are not in the current year. Therefore, the entry rate in year t is defined as the number of entrants in year t divided by the population of workers in year t-1. Similarly, the exit rate in year t is defined as the number of exits from USE or IBO divided by the total population of workers in year t-1.

Job transition states are used to compute the entry and exit rates at the yearly^Note  level for a given group of interest. These groups include all workers and subgroups of workers by (1) economic sector, (2) gender and marital status, (3) number of children, (4) spouse self-employment status, (5) age group, (6) immigrant status, and (7) income quintile. Let $p_{gt}^{k,s}$
be the transition rate from the k to s class of worker of a group g in year t. $p_{gt}^{k,s}$
is obtained as follows:

$$p_{gt}^{k,s}=\frac{{{\displaystyle \sum }}_i^{N_{g,t-1}}(JobTrans{i}_{it}^{k,s}=1)}{N_{g,t-1}}\text{ (4)}$$

Where $N_{g,t-1}$
is the total number of workers in group g during the previous year t-1. k and s are either PE, USE or IBO.

2.3.2 Independent variables: Individual and business characteristics

The main regressor of interest is the binary variable $COVID$,
equal to 1 in the 2020-to-2021 period and 0 otherwise. The coefficient associated with this variable captures the short-term effect of the pandemic on the probability that an individual switches from one class of worker to another.

Following the related literature on self-employment transitions (Lin, Picot & Compton, 2000; Cowan, 2020), independent variables account for individual, family and business characteristics. Individual and family information includes age, gender, marital status, an indicator of whether the spouse is self-employed, the number of children,^Note  immigrant status, income and place of residence. Gender is dichotomized as either men or women. Four age groups are considered: 15 to 24 years, 25 to 44 years, 45 to 64 years, and 65 years and older. The place of residence is classified as urban or rural. Province of residence is also included in the controls. Concerning marital status, the analysis distinguishes between individuals who are married^Note  and those who are not married. Marital status is an important factor to consider when analyzing individuals’ decision to enter or exit self-employment. Empirical studies provide evidence that family emotional help and support (including income stability) received by married people relative to unmarried people are likely to favour their entry into self-employment (Borjas, 1986; Blumberg & Pfann, 2001). Finally, immigrant status consists of recent immigrants (who arrived five years ago or less), established immigrants (in Canada for more than five years) and Canadian-born individuals. Regarding business characteristics, the preferred model controls for business size and two-digit North American Industry Classification System code.

The set of independent variables is measured at an earlier period to better estimate the causality of worker characteristics on job transitions.

3 Results

3.1 Descriptive analysis

3.1.1 The reduction in unincorporated self-employment and incorporated business ownership is proportionally larger than in paid employment

Over the 2020-to-2021 period of the pandemic, the number of workers amounted to nearly 10 million, a decrease of roughly 3 million compared with the average of the 2017-to-2019 period, as shown in Table 1.^Note  This decline was observed in all three classes of worker. Moreover, Table 1 indicates that workers were more prevalent in the PE class by about 82% before the pandemic. This proportion increased by nearly 5 percentage points on average in the 2020-to-2021 period, with perhaps a twofold explanation. First, people leaving the labour market were mostly from the self-employment class of worker rather than from PE. Second, a certain fraction of workers in self-employment in the pre-pandemic period had joined PE in the 2020-to-2021 period. For the latter argument, COVID-19 may have affected USE and IBO more than PE. Because businesses with paid employees received government support (e.g., wage subsidy), these employees were more likely to keep their jobs. For USE and IBO, income was affected negatively so that self-employment income was not the main income source anymore. Regarding the shares of unincorporated self-employed workers and business owners, they decreased by 2 percentage points over the period, suggesting individuals’ transition movements across each class of worker.

3.1.2 Less than 3% of workers changed status during COVID-19

Table 2 presents the rate of workers who transitioned between different job types over the years before and during the pandemic. A small proportion (less than 3%) of the study population transitioned over the sample period. Additionally, the raw data show a reduction in the number of switchers, in line with the overall decline in the total number of workers observed from 2020 to 2021.

Compared with the pre-pandemic period, the transition rates into USE and IBO were lower from 2020 to 2021. For example, among all workers, on average, 0.74% transitioned from PE to USE from 2017 to 2019. This rate decreased by approximately 23%, reaching 0.57% from 2020 to 2021, indicating a lower entry rate of paid employees into USE.

By contrast, the data show an increase in exit rates from self-employment to PE during the onset of the pandemic. Together, lower entry rates and higher exit rates may explain the drop in USE attributable to the pandemic. This result is consistent with previous findings that document evidence of a decrease in the self-employment rate in Canada (Beland, Fakorede & Mikola, 2020). However, the data described in Table 2 do not account for the transition from unemployment and non-employment into self-employment or out of self-employment into unemployment and non-employment. Therefore, the paper’s estimates of self-employment transition rates could be lower compared with the published data (Statistics Canada, 2025).

3.1.3 The majority of switching workers moved into a different industry

Chart 1 plots the transition rates into and out of self-employment that occur within a given industry, that is, the rate for workers who change class of worker without changing industry. The self-employment flows are related to the workers who enter self-employment from PE or exit self-employment to PE. The general pattern is still observed for those who moved within their industry, that is, a decrease in the self-employment entry rate and an increase in the exit rate during the pandemic. However, Chart 1 suggests that self-employment transition rates within the industry of work represent a lower share compared with transition outside the industry. Moreover, the difference is nearly double when considering the transition into or out of USE. For instance, in the 2020-to-2021 period, the entry rate into USE within the same industry was 0.23%, about half of the total entry rate of 0.57%.

Data table for Chart 1

Data table for Chart 1

Table summary

This table displays the results of Data table for Chart 1 USE, IBO, Total, Within prior industry, Total and Within prior industry, calculated using percent units of measure (appearing as column headers).

	USE		IBO
	Total	Within prior industry	Total	Within prior industry
	percent
Notes: USE stands for unincorporated self-employment. IBO stands for incorporated business ownership. This chart depicts the rates of entry into and exit from self-employment within the prior industry relative to the entire sample of workers. Source: Author calculations using the Canadian Employer–Employee Dynamics Database, 2017 to 2021.
Entry rates
Average (2017 to 2019)	0.74	0.31	0.41	0.18
2020 to 2021	0.57	0.23	0.36	0.17
Exit rates
Average (2017 to 2019)	0.79	0.33	0.48	0.27
2020 to 2021	0.82	0.35	0.51	0.31

For ease of presentation, the rest of the descriptive statistics use the term self-employment to refer to USE and IBO. Furthermore, the transition into (or out of) self-employment is related only to workers transitioning from (or to) PE.

3.1.4 Arts, entertainment and recreation, and accommodation and food services were among the main sources of workers’ self-employment flows during the pandemic

Previous studies have documented the major role of economic sector of work in affecting individuals’ entrepreneurial activities (Wan, 2017). Chart 2 depicts the rates of entry into (left part) and exit from (right part) self-employment before and during the pandemic by economic sector of origin, that is, the individual’s economic sector of work a year prior to their transition choice. This chart aims to examine the economic sectors in which the self-employment transition rates were important. From 2020 to 2021, entry rates into self-employment were highest for workers coming from arts, entertainment and recreation (3.60%), followed by other services (except public administration) (3.50%) and professional, scientific and technical services (3.43%). By contrast, paid workers from public administration, management of companies and enterprises, and utilities were least likely to become self-employed. Regarding the exit rates out of self-employment, the right part of Chart 2 indicates that during the 2020-to-2021 period workers leaving self-employment came mainly from accommodation and food services (2.45%), followed by information and cultural industries (2.32%) and administrative and support, waste management and remediation services (2.32%).

Data table for Chart 2

Data table for Chart 2

Table summary
This table displays the results of Data table for Chart 2 Average (2017 to 2019) and 2020 to 2021, calculated using percent units of measure (appearing as column headers).

	Average (2017 to 2019)	2020 to 2021
	percent
Source: Author calculations using the Canadian Employer–Employee Dynamics Database, 2017 to 2021.
Entry rates by economic sector of origin
Public administration	0.00	0.00
Management of companies and enterprises	0.16	0.12
Utilities	0.54	0.40
Manufacturing	0.57	0.44
Mining, quarrying, and oil and gas extraction	0.90	0.77
Accommodation and food services	0.95	0.75
Retail and wholesale trade	1.15	0.88
Finance and insurance	1.61	1.25
Information and cultural industries	2.36	1.77
Construction	3.17	2.44
Administrative and support, waste management and remediation services	3.51	2.61
Real estate and rental and leasing	3.52	2.70
Transportation and warehousing	3.60	3.14
Educational services	3.96	3.27
Agriculture, forestry, fishing and hunting	3.99	3.17
Health care and social assistance	4.10	3.45
Other services (except public administration)	4.56	3.50
Professional, scientific and technical services	4.70	3.43
Arts, entertainment and recreation	5.11	3.60
Exit rates by economic sector of origin
Agriculture, forestry, fishing and hunting	1.34	1.37
Health care and social assistance	1.53	1.58
Finance and insurance	1.89	1.85
Utilities	1.91	1.85
Public administration	1.84	1.86
Retail and wholesale trade	2.19	1.97
Manufacturing	1.97	1.98
Management of companies and enterprises	2.07	1.99
Professional, scientific and technical services	2.07	2.05
Other services (except public administration)	2.12	2.05
Arts, entertainment and recreation	1.94	2.06
Real estate and rental and leasing	1.77	2.08
Educational services	2.01	2.09
Transportation and warehousing	2.11	2.16
Mining, quarrying, and oil and gas extraction	2.36	2.26
Construction	2.40	2.30
Administrative and support, waste management and remediation services	2.31	2.32
Information and cultural industries	2.33	2.32
Accommodation and food services	2.31	2.45

In Canada, the effects of the pandemic have differed across industries, with significant negative impacts in service sectors. Sectors that rely on social interactions (such as face-to-face personal services), non-essential spending and foreign sales have been especially hard hit by COVID-19, with significant losses in revenues and jobs (Bernard, Fell, & Li, 2021). Employees and businesses in such sectors had few or no options for working remotely and may have transitioned into or out of self-employment to balance their financial resources. Reflecting this, Chart 2 shows higher entry and exit rates of workers from service sectors, including arts, entertainment and recreation, and accommodation and food services.

3.1.5 Married workers in paid employment were more likely to enter self-employment during COVID-19 compared with unmarried ones

Self-employment transition rates vary by individuals’ gender and marital status (Lin, Picot & Compton, 2000). From 2020 to 2021, married workers exhibited higher entry rates into self-employment than their unmarried counterparts, as shown in Table 3. These rates are greater for men than women. Regarding exits out of self-employment, the data reveal an increase in the rates with the onset of the pandemic, regardless of gender and marital status. For instance, while the entry rate rose by 0.05 percentage points for married men, reaching 1.78%, it increased by 0.08 percentage points for married women during the 2020-to-2021 period. Similar results are obtained for unmarried workers.

Evidence from Canadian studies suggests that self-employment rates were substantially higher for men than women, with a continuously closing gap over time (Leonard, Emery & McDonald, 2017). With the onset of the pandemic, women entrepreneurs were hit especially hard, as they tend to be in service-related fields such as retail, accommodation, tourism and food services (Grekou & Lu, 2021). Previous studies have also documented that having a spouse in business (self-employed) substantially increases the likelihood of people becoming self-employed, because a self-employed spouse often attracts the other to either join the family business or start their own (Lin, Picot & Compton, 2000). This could partially explain the generally higher turnover rate in self-employment of married workers compared with unmarried workers.^Note 

3.1.6 Workers with more than four children and with a self-employed spouse were less likely to enter self-employment during the 2020-to-2021 period relative to the pre-pandemic period

Charts 3 and 4 below describe the self-employment transition rates by number of children and whether workers had a self-employed spouse, respectively. Regardless of the selected factors, results indicate a decline in entry rates into self-employment and an increase in exit rates out of self-employment with the onset of the pandemic. However, the magnitude of the variation differs across subgroups. Workers with more than four children and with a self-employed spouse were less likely to enter self-employment during the 2020-to-2021 period relative to the pre-pandemic period.

Data table for Chart 3

Data table for Chart 3

Table summary
This table displays the results of Data table for Chart 3 Average (2017 to 2019) and 2020 to 2021, calculated using percent units of measure (appearing as column headers).

	Average (2017 to 2019)	2020 to 2021
	percent
Source: Author calculations using the Canadian Employer–Employee Dynamics Database, 2017 to 2021.
Entry rates by number of children
No children	1.12	0.92
1 or 2	2.03	1.50
3 or 4	2.02	1.42
More than 4	3.12	0.39
Exit rates by number of children
No children	1.24	1.31
1 or 2	2.11	2.26
3 or 4	2.82	2.86
More than 4	3.25	2.62

The results by number of children (Chart 3) seem intuitive and consistent with the related literature (Lin, Picot & Compton, 2000; Cowan, 2020). In fact, one would expect that individuals have less opportunity to change jobs as their number of children increases. This is justified by having less time available to start a business and the low propensity to bear the risk and income loss induced by entering self-employment. However, the counter-argument is that self-employment provides a balance between work and family time. So, as the number of children increases, people find self-employment more attractive, as shown by the entry rate from 2017 to 2019 in Chart 3. Regarding the impact of having a self-employed spouse (Chart 4), the literature discusses positive and negative evidence on the effect on a worker’s self-employment transition. As a result, it is unclear what to expect in terms of the incidence of having a self-employed spouse on the individual transition into or out of self-employment.

Data table for Chart 4

Data table for Chart 4

Table summary
This table displays the results of Data table for Chart 4 Average (2017 to 2019) and 2020 to 2021, calculated using percent units of measure (appearing as column headers).

	Average (2017 to 2019)	2020 to 2021
	percent
Source: Author calculations using the Canadian Employer–Employee Dynamics Database, 2017 to 2021.
Entry rates by spouse self-employment status
Yes	1.65	1.32
No	1.13	0.96
Exit rates by spouse self-employment status
Yes	2.81	2.96
No	1.27	1.33

3.1.7 Low-income earners and workers aged 24 or younger were most likely to be pushed out of self-employment during the pandemic

It is important to consider the different transition rates into self-employment by age group of workers given the disproportionate impacts of the pandemic on younger age groups. Chart 5 suggests that the pandemic has lowered entry rates into self-employment and has increased the exit rates for all age groups. More precisely, workers aged 15 to 24 years had the largest decrease in entry rates, from an average of 1.24% during the two years preceding the pandemic to 0.91% in the 2020-to-2021 period (i.e., a decrease of more than 0.30 percentage points). Regarding exit rates, estimates from Chart 5 show that from 2020 to 2021, workers aged 24 or younger were the most pushed out of self-employment, with an exit rate of 1.07% (i.e., an increase of 0.17 percentage points relative to the average rate from 2017 to 2019).

Data table for Chart 5

Data table for Chart 5

Table summary
This table displays the results of Data table for Chart 5 Average (2017 to 2019) and 2020 to 2021, calculated using percent units of measure (appearing as column headers).

	Average (2017 to 2019)	2020 to 2021
	percent
Source: Author calculations using the Canadian Employer–Employee Dynamics Database, 2017 to 2021.
Entry rates by age group
15 to 24	1.24	0.91
25 to 44	1.62	1.35
45 to 64	1.35	1.16
65 or older	0.13	0.11
Exit rates by age group
15 to 24	0.90	1.07
25 to 44	1.84	1.95
45 to 64	1.47	1.56
65 or older	0.33	0.34

Focusing on changes in employment and hours worked induced by the pandemic, Lemieux, Milligan, Schirle and Skuterud (2020) documented that younger workers were among the most affected groups. Moreover, it is well established in the literature that self-employment turnover rates are significant among young adults for need purposes or because young adults are prevalent in temporary jobs with lower and unstable incomes (Beland, Fakorede & Mikola, 2020; Brochu, Créchet & Deng, 2020; Nichols, 2023). Consistent with the latter argument, the decline in self-employment flows was greater for workers in the lowest income quintile, as shown in Chart 6. Compared with the average of the 2017-to-2019 period, the entry rates into self-employment of workers in the lowest income quintile were around 0.79% from 2020 to 2021, a decrease of roughly 0.32 percentage points. The higher flows into self-employment among younger workers and low earners, as described in charts 3 and 4, are consistent with the relevant literature on the greater vulnerability of these workers in times of hardship.

Data table for Chart 6

Data table for Chart 6

Table summary
This table displays the results of Data table for Chart 6 Average (2017 to 2019) and 2020 to 2021, calculated using percent units of measure (appearing as column headers).

	Average (2017 to 2019)	2020 to 2021
	percent
Source: Author calculations using the Canadian Employer–Employee Dynamics Database, 2017 to 2021.
Entry rates by income quintiles
Lowest (first and second)	1.11	0.79
Third	1.93	1.56
Fourth	0.84	0.74
Highest	1.16	1.03
Exit rates by income quintiles
Lowest (first and second)	0.53	0.54
Third	2.27	2.38
Fourth	1.94	1.93
Highest	1.81	1.79

3.1.8 After the onset of the pandemic, immigrant workers experienced higher turnover rates in self-employment compared with Canadian-born workers

Compared with Canadian-born workers, recent and established immigrant workers experienced the largest declines in entry rates into self-employment and the largest rises in exit rates from self-employment in the 2020-to-2021 period relative to the pre-pandemic period. Chart 7 reveals that the entry rate into self-employment for established and recent immigrants averaged 1.23% and 1.66%, respectively, from 2017 to 2019 and dropped to 1.06% and 1.46%, respectively, for 2020 to 2021. This change of nearly 0.2 percentage points was also observed in their exit rates out of self-employment, with an increase from 1.67% and 2.03%, respectively, before the pandemic to 1.89% and 2.24%, respectively, for 2020 to 2021. In addition, Chart 7 indicates that turnover rates in self-employment were the lowest for Canadian-born workers.

Data table for Chart 7

Data table for Chart 7

Table summary
This table displays the results of Data table for Chart 7 Average (2017 to 2019) and 2020 to 2021, calculated using percent units of measure (appearing as column headers).

	Average (2017 to 2019)	2020 to 2021
	percent
Source: Author calculations using the Canadian Employer–Employee Dynamics Database, 2017 to 2021.
Entry rates by immigrant status
Recent immigrants	1.66	1.46
Established immigrants	1.23	1.06
Canadian-born individuals	1.12	0.92
Exit rates by immigrant status
Recent immigrants	2.03	2.24
Established immigrants	1.67	1.89
Canadian-born individuals	1.22	1.28

There are various reasons why the pandemic may have disproportionately affected immigrants relative to the Canadian-born population and why that effect may be larger for recent immigrants. Immigrants tend to work in sectors that were hard hit by the pandemic, such as tourism (hotels, restaurants and leisure), and they have short tenure and hence are the first to be laid off in economic downturns (Slade, 2022; Hou, Picot & Zhang, 2020). Moreover, entrepreneurs who are recent immigrants face barriers to accessing credit because of their limited credit history in Canada (Cukier et al., 2017). To cope with these weak labour market outcome perspectives, immigrant workers often use self-employment to supplement employment income, resulting in higher turnover rates in self-employment relative to Canadian-born workers, as shown in Chart 7. 

In sum, the descriptive analysis provides some insights on the effect of the pandemic on individuals’ transition into or out of self-employment. There are notable differences in characteristics between workers leaving self-employment for paid jobs and those entering from paid jobs. A small proportion of workers changed class of worker between PE and self-employment before and after the pandemic. Moreover, findings reveal that the pandemic was correlated with a decline in self-employment through both a decrease in the entry rate and a rise in the exit rate. However, the analysis also shows that there are substantial differences in self-employment flows when considering some worker characteristics. For instance, married individuals, women, young workers, low-income earners, immigrants and workers with more than four children were among those with the highest turnover rate in self-employment. In addition, the paper documents that arts, entertainment and recreation, and accommodation and food services were among the largest sources of self-employment flows.

This unconditional descriptive analysis is not sufficient to derive a clear relationship of the pandemic on workers’ job transitions into or out of self-employment. A more convenient approach consists of evaluating the pandemic impacts after controlling for worker characteristics prior to transition. For that purpose, a probit model to estimate equations (1) to (3) is used, and results are described in the next subsection.

3.2 Multivariate results

3.2.1 The drop in self-employment rates induced by the pandemic is evidenced by both a decline in entrants and an increase in exits

This subsection explores the initial impacts of the pandemic associated with job transition in USE and IBO for workers either into or out of PE. The main results are presented in tables 4 and 5 for probabilities of entry into and exit out of self-employment, respectively. For ease of interpretation, the coefficient estimates in all tables correspond to the marginal effects instead of the change in the z-score (standard normal distribution) associated with a one-unit variation in the predictor variable. The marginal effects measure, in percentage points, how a given predictor affects the probability of entry into self-employment from PE or vice versa, relative to the reference group.

Table 4 reports estimates of equation (1) on the probability of entry into self-employment from PE. In column (1), the dependent variable is a binary variable equal to 1 if a worker transitions from PE to USE and 0 otherwise. Similarly, the dependent variable in column (2) is a binary variable equal to 1 if a worker transitions from PE to IBO and 0 otherwise. All specifications control for selected worker characteristics measured in the year before the transition, t-1. The models also include province and industry fixed effects.

Results from Table 4 indicate that the pandemic is negatively associated with entry of workers both into USE and into IBO from PE. In terms of magnitude, column (1) suggests that the pandemic is associated with a decline in the probability of entering USE by 0.67 percentage points. This impact is higher than the decline of the probability of entering IBO, as shown in column (2), i.e., 0.41 percentage points.

Additionally, findings from Table 4 reveal that entrants into self-employment vary by previous worker characteristics. For instance, while women were less likely to enter into self-employment than men, recent immigrants and workers with a self-employed spouse were more likely to do so relative to Canadian-born workers and those without a self-employed spouse, respectively. These estimates are consistent with the descriptive analysis presented in the previous subsection.

Regarding the probability of exiting self-employment, estimates from Table 5 suggest that the pandemic is associated with a higher probability of workers transitioning out of self-employment. The pandemic is associated with an increase in exit out of USE by 0.93 percentage points and out of IBO by 0.57 percentage points.

Table 5 also highlights how individual characteristics are associated with self-employment exit. Overall, these characteristics (i.e., gender, age group and spouse’s self-employment status) are the same as the ones identified in the case of entry into self-employment. However, for the probability of exiting self-employment, women and established immigrants have opposite signs compared with the probability of entering self-employment.

Together, the results of tables 4 and 5 show that the drop in self-employment rates associated with the pandemic as evidenced by both a decline in entry rates into self-employment and a rise in exit rates out of self-employment. These estimates are statistically significant and consistent with the results of the descriptive analysis. However, one may argue that these estimates are higher than the true effect associated with COVID-19 given that they include the decreasing trend in self-employment attributable to economic cycles. To test this argument, equation (2) is evaluated, and results are reported in Table A.1 in the appendix. Findings are similar to those obtained in equation (1), suggesting that the model properly isolates the association with COVID‑19.

The quantitative initial impacts of the pandemic on self-employment flows at the worker level outlined in this paper echo findings of previous studies based on Canadian data at a more aggregate level (sector or business level). Beland, Fakorede and Mikola (2020) found a large decrease in the number of business owners of small enterprises from February to July 2020 at the onset of the pandemic. Similarly, Lemieux, Milligan, Schirle and Skuterud (2020) found a 32% decline in aggregate weekly hours worked from February to April 2020 and a 15% decrease in employment attributable to the pandemic. The decline in incorporated self-employment and USE induced by COVID‑19 is also documented in U.S. studies (Kalenkoski & Pabilonia, 2022).

3.2.2 Interaction effects of COVID-19 on job transition probabilities into and out of self-employment by worker characteristics

This study further examines the heterogenous marginal effects associated with the pandemic on self-employment transitions by worker characteristics measured in the years preceding the transition. These characteristics include gender, age group, family characteristics (self-employed spouse and number of children), income quintile and immigrant status. Results are reported in Table 6. Panels (a) to (f) of Table 6 represent specific groups based on worker characteristics prior to the pandemic. Based on equation (3), Table 6 reports coefficient estimates of the interaction terms for certain groups on the probability of entry into self-employment (columns [1] and [2]) and exit out of self-employment (columns [3] and [4]). All specifications control for selected worker characteristics measured in the year before the transition, t-1. The models also include province and industry fixed effects.

3.2.3 Gender

Findings from panel (a) of Table 6 indicate that with the onset of COVID-19, women were less likely to enter into USE (by 0.61 percentage points versus 0.70 percentage points for men) and more likely to exit from USE (by 0.99 percentage points versus 0.87 percentage points for men) than men. Similar relationships applied to transition into or out of IBO, with lower magnitude of point estimates.

There are several potential reasons why COVID-19 may be associated with different impacts on self-employment transitions for men and women. One reason could be differences in occupational choice. If women were more likely than men to be involved in activities with a higher exposure risk during the pandemic (e.g., working in hotels, restaurants and tourism), then one may expect COVID-19 to impact them more (Gueye, 2024; Tam, Sood & Johnston, 2020b; Choi, Harrell & Watkins, 2022).

3.2.4 Age groups

Panel (b) of Table 6 reports the estimated coefficients associated with the interaction term between age categories and the COVID-19 dummy. Regardless of the age groups, COVID-19 is associated with a decline in self-employment transitions, with lower entry rates and higher exit rates. However, findings reveal that workers aged 24 or younger were the least affected by a decline in entry into USE, compared with other worker age groups. In terms of magnitude, COVID-19 is associated with a 0.52 percentage point decline in the likelihood of entering USE of workers aged 15 to 24. Regarding exit flows, results suggest that COVID-19 associated impacts are largest on the probability of workers aged 25 to 44 leaving USE for PE. COVID-19 is also associated with an increased exit probability from IBO. This pattern in distributional effects by age group is consistent with that of the related literature, including Beland, Fakorede and Mikola (2020) and Lemieux, Milligan, Schirle and Skuterud (2020).

3.2.5 Family characteristics: Number of children and self-employed spouse

Panels (c) and (d) provide estimates of marginal effects of COVID-19 on workers’ self-employment transitions by number of children and whether they had a self-employed spouse. With the onset of the pandemic, workers with children experienced contrasting impacts regarding self-employment transitions, as evidenced by a decline in entry and exit rates. This may suggest that these workers were less likely to take risks and transition into a different class of worker. Regarding the self-employment status of workers’ spouses, the findings of panel (d) indicate that workers with a self-employed spouse had a lower decline in entry into self-employment and a higher exit rate out of self-employment compared with those without a self-employed spouse. This result is consistent with the main findings of the paper.

3.2.6 Income quintiles and immigrant status

Broken down by income quintile and immigrant status, as reported in panels (e) and (f), the results suggest that self-employment transition flows for each category were similar, i.e., they showed a decline in entry and an increase in exit during the pandemic. What differs among the results is the magnitude of the COVID-19 associated impacts by given worker characteristics. For example, findings from panel (e) suggest that, with the onset of the pandemic, low-income earners had the lowest decline in the probability of entering USE (0.02 percentage points) and the highest increase in the probability of exiting USE (1.12 percentage points).

Like young workers, those in the lowest income quintiles are likely to be found in more precarious and unstable jobs (Beland, Fakorede & Mikola, 2020). Moreover, they are often employed or operate in industries with a high employment turnover rate—such as restaurants and food services. Findings from panel (e) reflect this greater vulnerability of low-income workers, with higher transition turnover rates in self-employment.

Panel (f) of Table 6 reports the coefficient estimate of interaction terms with indicators for immigrant status. Focusing on USE, results indicate different relationships to the pandemic period on the entry of immigrant workers. Compared with the pre-pandemic period, recent immigrants were the least likely to experience a decline in entry into USE (0.05 percentage points) and an increase in exit from USE (0.14 percentage points) during the 2020-to-2021 period. By comparison, established immigrants were the most likely to have a decrease in the entry rate (0.7 percentage points) and an increase in the exit rate (1.41 percentage points).

4 Discussion and conclusion

This paper examines and quantifies the magnitude of changes in workers’ transitions into and out of self-employment associated with the pandemic. Furthermore, it investigates how this relationship is related to worker characteristics before the transition. Using the Canadian Employer–Employee Dynamics Database (CEEDD) for the period from 2017 to 2021, the study points to a decrease in self-employment, as evidenced by lower entry rates and higher exit rates with the onset of the pandemic. This decline was greater for USE than for IBO. In addition, the results reveal some significant differences in the magnitude of the relationship by worker characteristics. For example, with the onset of the pandemic, the largest changes on self-employment flows were observed among women, low-income earners, immigrants, workers aged 24 or younger, and those with more than four children and a self-employed spouse.

The results described in this paper are consistent with those from previous studies at more aggregated levels, such as the enterprise or sector level. However, some considerations may help to improve the estimates. These considerations include (1) overcoming data limitations and (2) adjusting the model specification to consider additional relevant factors – particularly employment support programs.

The data used in this study cover the 2017-to-2021 period, which includes only two years of the pandemic. If COVID-19 impacts are not immediate but occur with certain lags, the analysis would benefit from including more years of the COVID-19 period to refine the estimates. Alternatively, in the absence of additional years, one could ideally consider monthly data such as those from the Labour Force Survey (LFS) to examine the immediate effects and assess whether the paper’s main findings are supported. Furthermore, with monthly observations, it is possible to evaluate how job transition correlates with the severity of physical distancing restrictions and health measures implemented over months. However, the LFS does not allow tracking the individual dynamics of a class of worker as in the CEEDD and therefore prevents analyses examining the job transitions of a given individual over time.

The main specification of the paper models the transition into or out of self-employment as exclusive, i.e., it considers workers as entering or exiting self-employment if this class of worker is their primary source of income or not. Therefore, it is silent about the workers who remain in self-employment or PE but with a lower income than in the previous period. This is likely to affect the true value of the transition rates. A more robust specification should distinguish entry into (or exit out of) self-employment with no income or little income in PE (or in self-employment). Moreover, the transition dynamics into and out of self-employment are likely to depend on whether workers anticipate entering self-employment with or without paid employees. This is a minor issue as the model controls for business size.

Finally, an attempt to differentiate the effect of economic cycles on workers’ self-employment transitions from that of the pandemic was made using an event study methodology. However, the short length of the pre-pandemic period may prevent capturing the effect of economic cycles in explaining the decreasing trend of self-employment. Future work could include a longer pre-pandemic data period, in addition to key economic information such as employment support programs, unemployment rates and differences in labour market features at the regional level.

This study speaks to the growing literature on the economic impacts of COVID-19 and sheds light on the role the pandemic played in changes to workers’ self-employment flows. With more recent data, future research could advance the paper and evaluate the outcomes after the transition.

Appendix

Data table for Chart A.1

Data table for Chart A.1

Table summary
This table displays the results of Data table for Chart A.1 IBO, USE and Total SE, calculated using rate (percent) units of measure (appearing as column headers).

	IBO	USE	Total SE
	rate (percent)
Note: IBO stands for incorporated business ownership. USE stands for unincorporated self-employment. SE stands for self-employment. Source: Author calculations using Statistics Canada, Table 14-10-0027-01 Employment by class of worker, annual (x 1,000).
2000	6	10	16
2001	5	10	15
2002	5	10	15
2003	6	10	16
2004	6	9	15
2005	6	9	15
2006	6	9	15
2007	6	9	15
2008	6	9	15
2009	6	9	15
2010	6	9	15
2011	6	9	15
2012	6	9	15
2013	6	9	15
2014	6	8	14
2015	7	8	15
2016	7	8	15
2017	7	8	15
2018	7	8	15
2019	7	8	15
2020	7	8	15
2021	6	8	14
2022	6	7	13
2023	6	7	13

References

Beland, L.-P., Fakorede, O., & Mikola, D. (2020). Short-term effect of COVID-19 on self-employed workers in Canada. Canadian Public Policy, 46(S1), pp. S66-S81.

Blumberg, B.F., & Pfann, G.A. (2001). Social Capital and the Uncertainty Reduction of Self-Employment. Available at SSRN 270796.

Borjas, G. (1986). The Self-Employment Experience of Immigrants. Journal of Human Resources, 21(4).

Brochu, P., Créchet, J., & Deng, Z. (2020). Labour market flows and worker trajectories in Canada during COVID-19 (No. 32). Working Paper Series.

Choi, S.L., Harrell, E.R., & Watkins, K. (2022). The Impact of the COVID-19 Pandemic on Business Ownership Across Racial/Ethnic Groups and Gender. Journal of Economics, Race, and Policy, 5, 307–317.

Cowan, B.W. (2020). Short-run effects of COVID-19 on U.S. worker transitions (No. w27315). National Bureau of Economic Research.

Cukier, W., Hon, H., McGraw, S., Latif, R. and Roach, E., (2017). Immigrant entrepreneurship: Barriers and facilitators to growth. The Diversity Institute, p.7.

Fairlie, R.W., Couch, K., & Xu, H. (2020). The Impacts of COVID-19 on Minority Unemployment: First Evidence from April 2020 CPS Microdata. National Bureau of Economic Research, No. w27246.

Grekou, D., & Liu, H. (2018). The Entry into and Exit out of Self-employment and Business Ownership in Canada. Analytical Studies Branch Research Paper Series, Statistics Canada Catalogue no. 11F0019M, no. 407.

Grekou, D., & Lu, Y. (2021). Gender differences in employment one year into the COVID-19 pandemic: An analysis by industrial sector and firm size. Economic and Social Reports, Statistics Canada Catalogue no. 36-28-0001.

Gueye, B. (2024). Survival and growth of women-owned and immigrant-owned businesses during the COVID-19 pandemic. Economic and Social Reports, Statistics Canada Catalogue no. 36-28-0001.

Hou, F., Picot, G., & Zhang, J. (2020). Transitions into and out of employment by immigrants during the COVID-19 lockdown and recovery. StatCan COVID-19: Data to Insights for a Better Canada, Statistics Canada Catalogue no. 45-28-0001.

Kalenkoski, C.M., & Pabilonia, S.W. (2022). Impacts of COVID-19 on the self-employed. Small Business Economics, 58(2), pp.741-768.

Kurmann, A., Lalé, E., & Ta, L. (2020). The Impact of COVID-19 on U.S. Employment and Hours: Real-Time Estimates with Homebase Data.

Lafrance-Cooke, A., & Leung, D. (2024). New businesses since the beginning of the COVID-19 pandemic. Economic and Social Reports, Statistics Canada Catalogue no. 36-28-0001.

LaRochelle-Côté, S., & Gilmore, J. (2009). Canada’s employment downturn. Statistics Canada Catalogue no. 75-001-X.

Lemieux, T., Milligan, K., Schirle, T., & Skuterud, M. (2020). Initial impacts of the COVID-19 pandemic on the Canadian labour market. Canadian Public Policy, 46(S1), pp.S55-S65.

Leonard, P.S.J., Emery, J.C.H., & McDonald, J.T. (2017). Push or pull into self employment? Evidence from longitudinal Canadian tax data. New Brunswick Research Data Centre, 21.

Leung, D., & Liu, H. (2022). The Canada Emergency Wage Subsidy program and business survival and growth during the COVID-19 pandemic in Canada. Economic and Social Reports, Statistics Canada Catalogue no. 36-28-0001.

Leung, D., & Robinson, C. (2011). The Dynamics of Male Self-employment in Canada: Comparing the 1990s to the 2000s. Available at SSRN 1972265.

Lin, Z., Picot, G., & Compton, J. (2000). The entry and exit dynamics of self-employment in Canada. Small business economics, 15, pp.105-125.

Liu, H., & Zhang, Y. (2025). Update on the Entry in and Exit of Self-employment and Business Ownership in Canada. Analytical Studies Branch Research Paper Series, Statistics Canada.

Nichols, L. (2023). Working through the Unknowns: Canadian Youth’s Experience of Employment during the COVID-19 Pandemic. Canadian Journal of Family and Youth/Le Journal Canadien de Famille et de la Jeunesse, 15(3), pp.113-129.

Schoar, A. (2010). “The divide between subsistence and transformational entrepreneurship.” In Innovation Policy and the Economy, ed. J. Lerner and S. Stern, p. 57–81. Cambridge, Massachusetts: National Bureau of Economic Research.

Schumpeter, J.A. (1942). Capitalism, Socialism, and Democracy. New York: Harper

Slade, M.E. (2022). Many losers and a few winners: The impact of COVID‐19 on Canadian industries and regions. Canadian Journal of Economics/Revue canadienne d’économique, 55, 282–307.

Statistics Canada. (2020). Canadian Survey on Business Conditions: Impact of COVID-19 on businesses in Canada, March 2020. The Daily, 1–4. Statistics Canada Catalogue no. 11-001-X.

Tam, S., Sood, S., & Johnston, C. (2020a). Impact of COVID-19 on businesses majority-owned by visible minorities, third quarter of 2020. StatCan COVID-19: Data to Insights for a Better Canada, Statistics Canada Catalogue no. 45-28-0001.

Tam, S., Sood, S., & Johnston, C. (2020b). Impact of COVID-19 on small businesses in Canada, third quarter 2020. StatCan COVID-19: Data to Insights for a Better Canada, Statistics Canada Catalogue no. 45-28-0001.

Wan, L. (2017). Literature Review about the Influence Factors of Self-employment. American Journal of Industrial and Business Management, 7(02), p.79.