5. Measurement recommendations
This chapter describes the statistical initiatives of different institutions that undertook measurement of digital platform employment, in order to assess it and provide recommendations for the future. As most of these experiments do not rely on established methodologies, this chapter is a contribution to build a shared measurement framework based on common vocabulary, common definitions and a common understanding of the phenomenon. In practice, a template was provided to different members of the Technical Expert Group in order to harmonise their contributions. Statistical experiments are described along the following dimensions: i) original purpose of the initiative; ii) reference population and sampling; iii) other relevant survey features (e.g. time references, data collection mode); iv) implied operational definition; v) obtained goals and lessons learned. Finally, the chapter distils some recommendations drawn from these experiments that should guide future statistical endeavours in this area.
Digital platforms are a multi-domain phenomenon, which can be observed either from the point of view of the worker, the client or the platform itself. High-quality statistics on digital platforms can inform about their impact on the labour market evolution and on its economic and social aspects. This has an implication on the sectoral structure of the economy but also on the technological evolution, on the production innovation process and recently also on the cultural evolution, with new terms entered in the vocabularies. During the COVID-19 pandemic, these platforms entered into own lives, helping in delivery of goods for confined populations and in working from home for those who were living under restrictions.
This variety of dimensions is discussed in Chapter 4, which describes different information sources according to different information needs and different domain of analysis. Information needs are driven by different policy interests and concerns: tax authorities, employment ministries, national accounts analysts and others will propose different operational definitions of the phenomenon and will use different sources accordingly.
Chapter 5 aims to be more practical, looking at actual measurement experiences. This chapter builds on evidence drawn from different testing exercises to generate standard questionnaires and methodologies for various types of surveys or other sources.
Its objective is to describe the initiatives from different institutions that undertook some actual measurement on the digital platform employment (DPE), in order to learn from these experiences and provide recommendations for the future. Most of the experiences discussed in the chapter are still experimental and do not rely on established methodologies, this is a contribution to build a shared measurement framework based on common vocabulary, common definitions and even a common understanding of the phenomenon.
Different members of the Technical Expert Group contributors have contributed to this chapter, bringing their experiences on generating information on DPE from different sources. In order to harmonize their contributions, and to facilitate comparisons, a common template was provided to authors, asking them to describe:
Original purpose of the initiative: Investigating which are the objectives and underlying questions the measurement experience aims at answer. For example, if the interest is in the quantitative dimension of the phenomenon, how many persons involved in DPE are in the population? If the interest is on the qualitative labour market features then the questions can be which are the DPE persons’ characteristics or which are their background features (demographic, education, previous experiences)? If the interest is on the social aspects of the market then the questions can be which are their working conditions? Or which is the degree of formality/informality of the working arrangements? If the interest is on the work itself then the question is which are the features of the job (driving, delivery, services to household, freelance such as program coding/translation)? If the interest is on the technical and innovation features then the question can be which is the type of the platforms (providing online or on-location work) and how many are they?
Reference population and sampling: Providing a description of the statistical units (that can be households, individuals, enterprises, platforms). According to the population, different sampling techniques can be applied and should be here clarified. The coverture is also interesting: does the survey reach the entire population with uniform inclusion probability or has the sample been extracted in order to focus one particular sub-population? The dimension of the sample of course also determine the statistical significance of the measure.
Other relevant survey features: Any other relevant feature such as time references, data collection mode (Personal, telephonic, web based, …) and any other methodological choices. This is particular important to compare different sources: yearly data is necessarily different from data referring to a single week.
Implied operational definition: General and operational (often only implicit) definitions of the analysed concepts. With, if possible, a description of the consequences of the practical definition chosen. For example it may be that some parts of the phenomenon are excluded, that some platforms are not considered, or other limitations due to the boundary built around the definitions. This should be linked to the conceptual framework as described in Chapter 3. The main questions are:
1. does the operational definition include goods and/or services (for example only include the provision of services but not production of goods)?
2. is it limited to some specific types of digital platforms (for example restricted to digital labour platforms only, with the exclusion of other specific types of platforms such as Airbnb, YouTube, Instagram, or includes only platforms with direct intermediation i.e. triangular relationship)?
3. is it restricted to a specific employment status, for example own-account workers only excluding employees or volunteer work?
4. is it limited by any other implicit or explicit restriction (only telephonic apps, only web browser platforms, only national based platforms, …) and definition used (is there a definition for algorithm, for platform, for employment (ILO strict concept or larger)?
Obtained goals and lessons learned: Possibly including some figures, some policy consequences already actuated (if relevant), a list of do and do not for the future and a short presentation of lessons learned about a better understanding of the DPE market or at least on how to measure it. This is also the template used to present the measurement experiences reviewed below. The last section distils some statistical recommendations from these experiences that, in the TEG’s view, should guide future statistical endeavours in this area.
Labour Force Surveys (LFSs) collect information on the supply side of the labour market, i.e. from worker’s perspective. Sampling units are either households or individuals, but the aim is always to provide information on individuals belonging to the (non-institutional) resident population. Given the large reference population, LFSs are well suited to capturing general dynamics of participation rather than to provide detailed information on small phenomena: even if their sample is large, the probability to include in the sample enough observations on small phenomena is low. While LFSs are a potentially good solution to estimate the total number of digital platform employed persons, they lack in statistical precision when studying particular characteristics of sub-groups of workers in DPE. Generally, LFSs are fit for quantitative analysis but if a qualitative analysis is needed then a smaller survey, with a more focused reference population, can provide more details. Several countries experimented (or are experimenting) the use of the LFS in the DPE domain.
US Bureau of Labour Statistics
Original purpose
The Current Population Survey (CPS) is the US monthly Labour Force Survey from which the official unemployment rate, alongside with other measures about the work force, are derived. In May 2017, BLS added four questions about digital platform employment (DPE) to the periodic CPS supplement on workers in “contingent and alternative work arrangements” (CWS).
The purpose of these four DPE questions was to measure the number of people employed as digital platform workers in the United States. By virtue of the DPE questions being asked as part of a supplement to monthly CPS, the questions also were designed to obtain demographic information and job characteristics of digital platform workers. For example, dependent on the estimated number and sample size, the gender, race, ethnicity, marital status, immigration status and educational attainment of digital platform workers could be quantified and compared to those of non-digital platform workers. The industry, occupation, usual and actual hours worked, multiple job holding status and earnings of digital platform workers could also be examined. Using information collected in other parts of the CWS module, information about whether digital platform workers have health insurance, the source of health insurance for those who have it, and job search activities could also be explored.
The original intention of these supplementary questions was to determine whether digital platform workers engaged in this type of work as their primary job or as secondary work. The DPE questions also were designed to distinguish between location-based platform workers (where tasks are performed offline in real physical locations) and online web-based platform workers (where tasks are performed online and remotely by workers). No other information about platform work (such as the names of the platforms used, the platform’s work rules, or how workers obtain assignments from the platform) was collected through these questions.
Reference population and sampling
The reference population of the monthly CPS is the US non-institutional population age 16 and older. Because the DPE questions were part of a supplement to the CPS, the sample design was the same as that of the CPS. The CPS uses a multi-stage, stratified random sample of approximately 72 000 housing units from 852 sample areas throughout the United States. Use of this reference population also allows estimates to be in accord with other US labour force measures.
Other relevant survey features
The CPS is typically conducted in the week containing the 19th of the month, and asks respondents about their activities in the prior week. To DPE questions were asked to individuals who had been classified as employed during the week of May 7- 13, 2017. As with the monthly CPS, the DPE questions were asked by Census Bureau interviewers using a computerized questionnaire. These interviews were conducted using a combination of in-person and phone interviews. Phone interviews were conducted either from interviewers’ residences or from centralized facilities.
Implied operational definition
The questions included in the CWS module of the CPS were:
Introduction - I now have a few questions related to how the Internet and mobile apps have led to new types of work arrangements. I will ask first about tasks that are done in-person and then about tasks that are done entirely online.
Q1 Some people find short, IN-PERSON tasks or jobs through companies that connect them directly with customers using a website or mobile app. These companies also coordinate payment for the service through the app or website.
For example, using your own car to drive people from one place to another, delivering something, or doing someone’s household tasks or errands.
Does this describe ANY work (you/NAME) did LAST WEEK?
Q1a Was that for (your/NAME’s) (job/(main job, (your/NAME’s) second job)) or (other) additional work for pay?1
Q2 Some people select short, ONLINE tasks or projects through companies that maintain lists that are accessed through an app or a website. These tasks are done entirely online, and the companies coordinate payment for the work.
For example, data entry, translating text, web or software development, or graphic design.
Does this describe ANY work (you/NAME) did LAST WEEK?
Q2a Was that for (your/NAME’s) (job/(main job, (your/NAME’s) second job)) or (other) additional work for pay?
The questions about DPE workers in the CPS supplement operationally defined DPE workers as workers who obtained tasks, job or projects through an online platform, and whose customers used the same online platform to pay for the services provided. This operational definition of DPE workers encompasses both location-based platform workers (e.g. rideshare drivers, delivery workers, and home service providers) and online web-based platform workers (e.g. web developers, translators, and content mediators). To be counted as a DPE worker in the CPS supplement, individuals must be employed. People involved in volunteer work, unpaid trainee work, and own use production work are therefore excluded, even if their work was done through or on a digital platform. All types of employed individuals can be DPE workers, except unpaid family workers. Self-employed (own account workers) can be classified as DPE workers regardless of whether they have employees or are incorporated businesses.
The CPS definition of DPE workers, by virtue of the questions wording and their restriction to those who are employed, only includes workers using digital labour platforms. By asking about tasks or jobs that people obtain through an app or website, this definition implicitly excludes individuals with income generated from the selling of new or used goods online. This definition also implicitly excludes those who generate income from the rental of property and activities associated with generating this income. However, direct questions to screen out these activities were not explicitly asked - it should be noted that these exclusions from the definition of DPE workers did not preclude from being counted as employed individuals doing activities associated with the production of goods to sell and the rental of property.
The requirement that customers pay through the same platform used to connect workers with customers further restricted the CPS measure of DPE workers to workers where digital platforms provide at least some direct intermediation (triangular relationships). The CPS definition does not make any other restriction based on the type of intermediation platforms implement, while permitting platforms to have varying degrees of control over how workers are assigned work (algorithmically or not), the price customers are charged, and other forms of direct intermediation. This same requirement (that customers pay through the same platform) implies that workers conducting a job search online for non-DPE jobs (e.g. posting resumes online, reviewing online job postings or conducting online job interviews) and workers advertising for customers online on their own (e.g., using an electronic bulletin board such as Craigslist) are excluded from the CPS definition of DPE workers.
Workers conducting a job search online for non-DPE jobs (e.g., posting resumes online, reviewing online job postings or conducting online job interviews) and workers advertising for customers online on their own (e.g., using an electronic bulletin board such as Craigslist) are excluded from the CPS definition of DPE workers by the requirement that customers pay through the same platform used to connect to workers.
Key results
Based on the data collected through these questions, BLS released estimates about the number and characteristics of digital platform workers. Doing so, however, required modification of some of BLS’s standard procedures. In particular, after extensive review of the collected data, BLS determined that the new questions did not work as intended and included a large number of incorrect “yes” answers. To eliminate these “false positives”, BLS manually recoded answers using additional information collected in the CPS survey. This recoding reduced the number of “yes” answers by approximately two-thirds. BLS is confident that the recoded data provides a better picture of location-based and online web-based platforms workers. In the interest of transparency both the data as originally collected and recoded were released to the public.
BLS analysis of its recoded data indicates:
In May 2017, there were 1.6 million digital platform workers in the United States, accounting for approximately 1% of total employment.
Digital platform employment was fairly evenly divided between location-based and online web-based platforms. Location-based platform workers accounted for 0.6% of total employment, while online web-based platform workers accounted for 0.5% of total employment.
Digital platform workers were slightly more likely to be men than women, reflecting the fact that, overall, a higher percentage of the employed in the United States were men.
17% of digital platform workers were Black, higher than their share of overall US employment (12%). By contrast, 75% of platform workers were White, slightly lower than their share of workers overall (79%). Hispanics made up 16% of platform workers, and Asians accounted for 6% of platform workers, close to their shares of total US employment.
Black people were overrepresented among location-based platform workers (23%), while White people were overrepresented among online web-based platforms workers (84%).
Digital platform workers were more likely than workers overall to work part time. (Part-time workers are defined as those who usually work less than 35 hours per week at all jobs combined.)
Additional information about the characteristics of digital platform workers can be found on the BLS website https://www.bls.gov/cps/electronically-mediated-employment.htm. This website refers to digital platform employment as “electronically-mediated-employment”; it also refers to “location-based” as “in person” and to “online web-based” as “online”.
Lessons learned
Based on a long-standing tradition, BLS does not use the names of companies in its survey questions, particularly in emerging or rapidly changing industries. BLS avoids the use of company names both to avoid focusing respondents on just a few, particular companies and to ensure that the questions are durable across time (or even between when the questions are developed and when they are fielded). To distinguish between people in various categories, BLS relies on asking questions about characteristics that distinguish the group of interest. This is the approach that was used with the DPE questions. However, due to other constraints, only 4 questions could be used to identify DPE workers.
Both the nature of platform work and the constraints of only having 4 questions contributed to the questions not working as intended. Specifically, the questions used in the CWS to identify location-based and online web-based platform workers were too complicated. Due to the length of the question, it appeared that at least some respondents did not hear the part of the question related to customers paying through the platform. The complexity of the questions also caused some respondents to focus on the examples, rather than on the characteristics of platform work embedded in the question. Additionally, if respondents hesitated, interviewers sometimes repeated only the examples.
Another issue with the questions, is that respondents interpreted them too broadly. This broad interpretation was related both to the widespread use of apps, computers, and websites by people in all types of work and to the complexity of the questions. This broad interpretation occurred for workers who were connected to customers through a website or app as part of their work but were not paid through it (such as customer service workers or fast-food preparers), workers who used mapping apps to obtain routing directions to customers (such as gravel deliverers) and workers who used computers as part of their work (such as receptionist who schedule medical appointments using a computer and university lecturers teaching online). Respondents with these broad interpretations appeared to have not heard the part of the question asking if customers paid through the same app or website that connected workers to them.2
Despite the difficulties with the questions, BLS will continue to use questions asking whether people have certain work characteristics associated with being platform workers to identify platform workers in the future. BLS will also not use the names of companies as it ensures the largest coverage at a point in time, is most likely to maintain durability of the measurement across time and provides the most flexibility for data users. Several do’s and don’ts have been learned about following this approach, however.
Switzerland – Federal Statistical Office
Original purpose
In 2019, Switzerland conducted an LFS ad-hoc module on “internet-mediated platform work”. The main objective was to estimate the number of internet-mediated platform workers in the country, with additional questions on the reasons for this type of work, on how long the person has been providing these services, whether they take place as a main, second or additional job, the regularity and number of hours worked, the income generated, and the name of the platform or app. All types of services were considered (renting out accommodation, taxi services, sale of goods, and other services). The names of platforms or apps were mentioned as examples in the filter questions.
Reference population and sampling
The reference population of the module covered 15 to 89 year-olds from the permanent resident population. No specific sub-population was targeted and the questions were asked to all persons independently of their labour market status. The total annual sample of the module was around 12 000 persons which were chosen by random selection at the end of the first wave of the LFS (one-third of first wave, two-thirds being already moduled).
Other relevant survey features
In general, the questions on the internet-mediated platform services referred to the last 12 months preceding the interview. In order to have information on work done in the reference week, an additional question was used to find out if the person had provided one of the named services in the past week. All interviews were conducted in CATI.
Implied operational definition
In the Swiss module, every type of internet-mediated platform service was considered, including renting out accommodation via Airbnb or selling goods. Four filter questions allowed distinguishing between platform work in the narrower sense and platform services as a broader concept, including those selling goods and renting out accommodation. The following criteria had to be fulfilled for platform work or platform services:
Internet-mediated platform work:
- The person providing the service is connected to the client via an internet platform or app.
- Payment is usually made via the internet platform or app.
- The vast majority of the services provided consist of work (e.g. cleaning) and not of assets (e.g. renting out accommodation).
All areas of activity that can be processed via an internet platform or app are included, such as taxi services, cleaning, food delivery services, transport and delivery of goods, tradesperson services, programming, translation, data and text entry, and web and graphic design.
Other internet-mediated platform services:
Online rental of accommodation (rooms, apartments, and houses)
- The person providing the service is connected to the client via an internet platform or app.
- Payment is usually made via the internet platform or app.
Online sales via an internet platform
- The person providing the service is connected to the client via an internet platform or app.
- The goods sold must have been deliberately collected, bought or produced for the purpose of resale.
- Payment is not usually made via the internet platform or app.
Key results
The analysis of the Swiss module showed that in 2019, 0.4% of the population said they had carried out work via internet-mediated platforms in the past 12 months (taxi services: 0.1%; other services such as programming, food delivery, cleaning, etc.: 0.3%). At 0.6%, a slightly larger percentage of the population had rented out accommodation through an internet platform, and 0.8% of the population had sold via an internet platform goods that were especially collected, purchased or produced with the aim of reselling them.
Lessons learned
As this form of work is not very common in Switzerland, it is difficult to collect data on these specific types of activities and to formulate questions in a way that reflects the conceptual issues at stake and that is understood correctly by the respondent. There was confusion between providing and using a service via a platform / app and consequently a certain number of “false positives”. Plausibility checks based on the hours worked, income, named platform, reason for activity, but also the interviewers’ comments were very important in the data production process.
Based on these first experiences, Switzerland reflected on potential improvements of questions. One suggestion would be to add, after a positive answer to the filter questions, a question on payment and commission:
-“Have you been paid / will you be paid for this service?” Answer categories: via the platform, app / by the client / won’t be paid
-“Does the internet platform / app take a commission for the service* you offer on it?” Answer categories: yes / no.
*renting out an accommodation / taxi services / sale of goods / other service
A second suggestion concerns the sale of goods to better capture the persons who acquired goods with the specific aim of reselling them and to generate income. If the respondent says “yes” to the filter question on sales of good, there would be a follow-up question:
“Why did you sell these goods?” Answer categories: I didn’t need them anymore / I collected, bought, or produced them with the specific aim of reselling them
These additional check questions would help to exclude a significant number of false positives, i.e. persons answering “no” to payment and commission; and persons selling goods because they didn’t need them anymore.
Chile – National Statistics Institute
Original purpose
Since January 2020, the National Employment Survey (ENE, Spanish acronym) has incorporated a new series of updates in order to strengthen its production process and more accurately reflect the national labour market. These improvements include expanding the questionnaire to include new dimensions of analysis in order to capture new trends in the Chilean labour market.
Through these processes, the ENE began to capture, measure, and analyse employment work that uses mobile applications or digital platforms. This kind of work is a recent phenomenon that has been growing at an accelerated pace throughout the world, and it grew further during the COVID-19 pandemic. Thus, National Statistics Institute (INE, Spanish acronym) can provide information for more research into the nature and trends of employment work in Chile, enabling a response to such questions as “How many people are employed in work obtained through the use of digital platforms?”, “What are their demographic characteristics?”, “In which economic sectors are these workers employed?”, “What is the status of the population employed on digital platform work?”, “How do such workers access social welfare benefits?”, and “Which platforms did they use to obtain work?”
Reference population
The ENE of the INE is the principal survey on the labour force in Chile, and it has publishable information since 1986 to date. The ENE is aligned with the recommendations of the International Conference of Labour Statisticians of the International Labour Organization. The universe of the ENE consists of all persons who reside in occupied private dwellings within the national territory and who are covered by the sampling frame of the survey called MMV 2017.3 Within this universe, the target population of the ENE is defined as the working-age population (i.e., all persons aged fifteen and over) who habitually reside in occupied private households in Chile.
The universe of the National Employment Survey (ENE) consists of all persons who reside in occupied private dwellings within the national territory and who are covered by the sampling frame of the survey called MMV 2017.4 Within this universe, the target population of the ENE is defined as the working-age population (i.e., all persons aged fifteen and over) who habitually reside in occupied private households in Chile.
Sample dimension
Because the digital platform work continues to be a phenomenon in constant change and development, information on as many workers as possible should be captured. For this reason, the question on work obtained through digital platforms in the main activity5 is directed to all occupied persons, except for persons classified as “Contributing family workers” in consideration of the nature of such work. Information on work obtained through digital platforms is also studied among workers who state that they engage in a secondary activity, in order to describe such activities.
Other relevant survey features
The ENE sample is spread over three months, constituting moving quarters, and it is then divided into subsamples that are repeated every three months for a maximum period of one and a half years (six rounds of visits). Thus, the ENE captures information for each monthly subsample, but it processes, analyses, and publishes information for moving quarters (INE, 2021[1]).
The information is collected on a mobile capture device (i.e., by CAPI6) in an application designed especially for the ENE. In addition, information is gathered through a mixed methodology that includes face-to-face interviews and telephone surveys.
Implied operational definition
The ENE defines digital platform work as follows: “work that uses a mobile application or digital platform to offer goods or services by exclusively or predominantly using media that involve remote contact with customers, either through the internet (digital platform) or by cell phone (mobile application)” (INE, 2022[2]). In short, platform workers are those who use a digital platform or mobile application to mediate a service or exchange goods.
Because expert international organizations have not yet officially defined this recent and rapidly developing phenomenon, which has a strong impact on employment, published statistics are considered experimental within the framework of the labour market statistics published by INE. The questions included in the ENE questionnaire to capture digital platform work are the following:
The structure of the questions is as follows: first, the interviewer asks a dichotomous question on whether this type of technological tool is used while permitting partial response. Next, if affirmative, the interviewer asks the name of the digital platform, “Which one?”. However, the fields intended to identify the digital platform do not have any type of validation during the collection process, thus the open nature of the question may result in errors in the information provided by the respondent. Therefore, the response is subject to cleaning, validation, and imputation.
Once available, the information on the mobile application or digital platform is cleaned. This process consists of standardizing the open text record and eliminating texts determined to be “erroneous”. These texts include names of digital platforms that do not meet the accepted definition of digital platform work. After the texts are cleaned, only those persons who replied with “valid” texts are included in the estimation. To define a text as valid, a dictionary of digital platforms was created after an exhaustive review of the digital platforms in Chile. The dictionary is constantly being updated in order to reflect the evolution of digital platforms.
After, the question of digital platform work in main activity will be imputed according to the description of the question on employment and activity. This procedure consists of reviewing the text to identify the occupational group, the activity of the company, and the name of the company that pays the person's income, searching in the descriptions of these questions when the employed person states that they obtain work through one of the following mobile applications: Uber, Cabify, Didi, Beat, Cornershop, Rappi, Uber Eats, or Pedidos Ya7. The final variable is published in the database called “plataformas_digitales”. The value of the variable is 1 if it represents those who actually have a digital platform work in their primary occupation. The variable “pd_especifique” represents in which digital platform work. After the process, the variables on secondary occupations of digital platform work, are published as “sda_pd” and “sda_pd_especifique”.
Consequently, we can observe that the conceptual scope of the questions of digital platforms work can be classified by type of production, type of digital platform or by employment status. This can be seen in Figure 5.3.
Obtained goals and lessons learned
For the moving quarter April-May-June Q2 2022, it is estimated that there were 205 741 employed persons working on their main activity on digital platforms, which represents 2.3% of the total employed population8. Of these 205 741 employed persons, 31 376 were foreign nationals, or 15.3% of the total number of employed persons on digital platform work.
Source: National Statistics Institute of Chile, ENE (INE, 2022[2]), Separata técnica n°4: Nuevas dimensiones de análisis.
Breaking down statistics by gender, there were 108 630 employed men who work on digital platform, which represents 2.1% of the total number of employed men and 52.8% of the employees on digital platform. In contrast, an estimated 97 111 employed women obtained work on digital platforms, which represents 2.6% of the total number of employed women and 47.2% of the employees on digital platform. For the last moving quarter for which data is available (April–June 2022), 5.3% of the informally work on digital platforms, which amounts to 126,441 persons. In contrast, 1.2% of the formally employed population work on digital platform (45 027 persons). Also, most platform workers are independent (78.1% of digital platform workers that represents 160 713 persons).
When disaggregated by economic activity, the estimate of employed persons who works on digital platform shows that “wholesale and retail commerce”, represented 28.8% of platform workers. For this group, the following platforms and applications were the most prevalent: Facebook, Instagram, and Whatsapp. Following is “transport and storage”, which includes vehicle drivers, and it makes up 19.5% of platform workers (40 025 persons) during the period. For this group, the application Uber is the most important. Finally, 33 130 persons were categorized as working in “manufacturing”, which represents 16.1% of platform workers. The activities of this group include the manufacture of textiles and the processing of some food and personal care products.
Source: National Statistics Institute of Chile, ENE (INE, 2022[2]), Separata técnica n°4: Nuevas dimensiones de análisis.
In Chile, Law 21.431, about digital platform work, was enacted in March 2022, and it comes into force on September 2022. This one establishes norms and it regulates the relationship between platform workers (both dependent and independent) and the digital platform company. The Law defines what is to be understood as a digital platform company and digital platform worker. It describes the type of employment contract between dependent and independent digital platform workers, specifying the employer's duty to protect workers, to normalize working hours and remuneration or fees, and respect platform workers' right to access social welfare benefits and their right to disconnection among other contractual obligations.
The law is narrower in its definition of digital platform work than ENE's definition, and therefore information based on the law constitutes a subset of what is captured in the survey. ENE estimate employed persons, and the resulting figures may therefore differ from the number of jobs available on digital platforms because a worker may be registered on more than one platform at the same time. The measurement of the number of digital platform workers has been important in approximating the number of persons who will be affected by the law upon its implementation. The ability to characterize this subgroup of employed persons has provided a framework for the application of Law 21.431, and it has enabled the monitoring of the evolution of the characteristics and working conditions of this subgroup over time.
Finally, the analysis and monitoring of the employed population on digital platform is a new topic in labour force surveys, and information on this subgroup is frequently requested. The ENE has incorporated the observation of this subgroup since 2020 by using a data processing methodology based on text mining, which allows validating and cleaning data according to the definition of a dictionary of digital platforms.
These estimates are defined as experimental statistics because they still show room for improvement, and they have not yet achieved sufficient maturity to be included in the list of official statistics. It should be noted that for the last moving quarter for which data is available (Q2 2022), the total number of employed persons on digital platform reached 205,740 persons, equivalent to 2.3% of the total employed population.
Manpower Research and Statistics Department, Singapore
Original purpose
Singapore’s Labour Force Supplementary Survey on Own Account Workers provides data on the prevalence of own account workers9 engaged in online platform work. Dedicated questions in the supplementary survey allow examining the demographic profile of own account workers engaged in on-line platform work, the occupations they worked in and, the types of online matching platforms used. Beyond this, the survey also seeks to understand the motivations for taking on platform work, whether the workers involved were doing this as their main job or on the side, and the challenges they faced.
Reference population and sampling
The module considered here is a supplement to the Comprehensive Labour Force Survey, which covers all private households in Singapore. The survey sample is selected based on a stratified random design with proportional allocation.
All respondents aged 15 years and over who indicated in the Comprehensive Labour Force Survey that they did own account work in the past year were questioned through the supplementary survey, which asks for detailed information on their experiences in own account work and whether online matching platforms were utilised. Approximately 4 200 persons responded to this supplementary survey, with results grossed up to the resident population using multiple estimation factors. The use of a random sampling methodology and a high response rate of at least 85% help ensure that findings are representative of the general population.
Other relevant survey features
The reference period of a year is used to enable a more accurate sensing of the prevalence of own account and platform work, given the ad hoc and transient nature of such work arrangements. Survey responses were mainly collected through internet submissions or telephone interviews. Where needed, face-to-face interviews were also conducted.
Implied operational definition
Platform workers in Singapore’s context refer to own account workers who provided paid labour services via online matching platforms. Such labour sharing platforms serve as intermediaries to match or connect buyers with workers who take up piecemeal or assignment-based work. These platforms could be either websites or mobile applications, covering services such as ride-hailing, goods/food delivery, creative work etc. Other types of platforms (e.g. capital-sharing or e-commerce platforms) are not included, as labour is not the main traded good in those cases.
Notwithstanding this, the survey separately captures a wider measure of own account workers who utilised any online channels to obtain business. Besides the aforementioned online matching platforms, this include (but not limited to) the use of e-commerce websites and social media platforms.
Some platform workers could be doing such work for their livelihood, while others do platform work on the side, e.g. multiple job holders with a full-time employee job, students, homemakers and retirees. Such a differentiation provides better insights to aid in more targeted policy interventions.
Singapore adopts the same definition of “employment” as that stipulated in the ILO resolution10 During the reference period, employed persons either (i) worked for one hour or more either for pay or profit; or (ii) have a job or business to return to but are temporarily absent because of illness, injury, breakdown of machinery at workplace, labour management dispute or other reasons.
To address the sporadic nature of own account or digital platform work, Singapore’s survey used a longer reference period of a year. In addition, the survey sieves out persons who did such work on a regular basis during the year, as such persons likely have higher work attachment.
Key results
The survey enabled a better understanding on the prevalence and nature of platform work in Singapore. At the most general level, the data show that, despite the emergence of online matching platforms, own account work has not displaced regular employment. The percentage engaged in own account work has remained fairly stable, at about 8% to 10% of all employed residents over the last decade.
With respect to digital platform employment per se, the latest 2020 survey found that about 3% or 73 500 of Singapore’s resident workforce was engaged in platform work as their primary job in the year. They were mainly involved in the transportation of goods and passengers. The majority engaged in platform work by choice, because of the flexibility and freedom they enjoy.
The top occupations in platform work were private-hire car drivers, taxi drivers, and car and light goods vehicle drivers. Work arrangements of platform workers can resemble those of employees. Platform companies often set the price of their product, determine the assignment of jobs to workers, and manage how the workers perform, including imposing penalties and suspensions.
Lessons learned
Platform workers’ contracts with platform companies are not employer-employee contract of services. This means platform workers do not have the statutory provisions that employees have, such as work injury compensation, union representation and mandatory social security contributions made by the employer. Because of this, the government is looking into strengthening protections for platform workers, specifically delivery workers, private-hire car drivers and taxi drivers, and ensuring a more balanced relationship between intermediary platform companies and platform workers. Singapore’s Prime Minister announced at the 2021 National Day Rally that an Advisory Committee will be convened to study this issue.
European Commission – Eurostat
Original purpose
In 2019 Eurostat launched a Task Force on Digital Platform Employment (DPE), with the purpose to answer to the increasing pressure for comparable information on gig/collaborative economy. Statistical measurement are of paramount importance to design informed policies and the main stakeholders are precisely the policy Directorates General of the European Commission. One of the goals provided in the Political guidelines for the European Commission 2019-2024, presented by President von der Leyen, is “improving the labour conditions of platform workers notably by focusing on skills and education”. EU-action is recognised as necessary in order to protect platform workers and reduce the risk of regulatory fragmentation across member states.
The Task Force mandate has included the development of technical and methodological elements plus a set of variables/questions, which have been piloted in the EU-LFS within 2022 by the volunteered Member States to evaluate the results of this pilot exercise and to come with a revised proposal, in view of a full implementation in the EU-LFS. The pilot survey on DPE is ongoing and no results are available at present, but 2023 results will be used in terms of developing the final EU LFS framework for the measurement, including the definitions of the phenomena. The implementation of DPE module within the EU LFS for regular production depends on the pilot survey results and the final discussion taken by the Eurostat LAMAS LFS Working Group.
The technical/methodological proposal includes: defining the objective of the information to be collected, for example capturing additional persons in employment (as first or second job) who find work through platforms and/or having an overall number of platform workers; defining the labour market characteristics of interest of these platform workers: working conditions, professional status, quality of the work, required skills, etc.; defining the target phenomena (what are platform works? which platforms?); defining the reference population and the reference period.
Reference population and sampling
The reference population is all population within 15 and 64 years old, with respect to the population of the EU-LFS all persons aged 65 and more are excluded. Constraints to the survey are imposed by the EU-LFS nature. The focus should be on:
the activities constituting employment (work for pay or profit, for more than one hour in a reference week);
the counting of heads (not transactions, volumes or revenue);
the activities where a website or an app plays an integral role in connecting workers to customers.
The main challenges are two: to identify the platform work within the employed people already detected in the LFS (as part of first or second job) and to identify employed people non-detected in the LFS who actually perform paid platform work.
The Task Force suggests that the pilot is asked to a sub-sample of the standard LFS (with the indication to be placed at the end of the questionnaire in order not to interfere in any way with the results of the current LFS). Dimension of the sub-sample should be at least, in each quarter of 2022, 1/10th of the national quarterly sample, in order to provide significant figures on the small phenomenon of the DPE. The information on actual sample size is not yet available when writing this Handbook.
Other relevant survey features
All features of the survey should be identical, or as similar as possible, to those of the EU-LFS, in order to produce comparable results.
Implied operational definition
The first step is to define “employment” in the context of DPE. This definition should remain into the LFS framework, implying that “employment” refers to the usual criteria: at least one hour of work for pay or profit in a reference week. Nevertheless, since the time horizon of a survey on DPE, being the DPE a rare phenomenon at the moment, cannot be the single LFS reference week, it has been decided to somehow extend the reference period of the pilot: the first set of questions refers to the last 12 months ending with the reference week of the LFS. Once the number of the DPE workers has been estimated the following sets of questions, on their characteristics, refer to the last month and to the reference week in order to make the link with the core LFS.
Consequently, the criteria for defining “employment” in the context of DPE is proposed to be:
To have worked for pay or profit in tasks/activities organised through an internet platform or a phone app, for at least one hour in at least one week, during the reference period.
Persons with very short spells of DPE work, specifically never spending more than one hour in one week for one year on paid work, are excluded, it may be the case of some micro tasks workers. This should be a very rare phenomenon and its loss is compensated by the comparison advantage to link with the LFS definition of employment. Moreover this definition, while sticking on the LFS standard, enlarges the reference period to have more robust estimation.
Any task/activity that can be considered as “employment” in the LFS, i.e. production of goods or provision of services but also time spent in searching for clients or in setting up the working activity, should be considered as DPE when the other criteria are fulfilled. It includes in particular, work for pay and profit (1) for providing taxi or transport services including driving clients, delivery of food for restaurants, or transport or delivery of any kind of goods or similar, (2) for providing actual services in view of renting out a house, a room or any other accommodation, (3) for selling any good produced or bought with the intention of selling it online, and (4) for providing other kinds of services or work, among others: cleaning, handiwork, programming and coding, online support or checks for online content, translation, data or text entry, web or graphic design, medical services, creating contents such as videos or texts (with the purpose of earning money or other benefits).
For all tasks/activities listed here above, as far as the matching between the provider and the client is done throughout an internet platform or a phone app, the task/activity can be considered as DPE.
Concerning the definition of a “platform”, the OECD approach (based on work from the European Commission and others, used in its publication of March 2019 ‘Measuring the digital transformation: a roadmap for the future’) is followed:
An online platform is a digital service that facilitates interactions between two or more distinct but interdependent sets of users (whether firms of individuals) who interact through the service via the internet
Then, to define a work organised through a digital platform is necessary. It has been decided to take into consideration only work organisations involving three separate agents. On the vertex of the triangle are the three agents:
The provider, in this context this is the supply side of the labour market and it is our target (the employed person);
The client, in this context this is the demand side of the labour market (it may be an individual or a legal person);
The platform: an internet platform or a phone app with the purpose to facilitate the match between the provider and the client.
It is important that the three agents are distinct. If the platform and the provider coincide, for example if the provider owns the platform like a producer that sells its goods on its own website, this is not anymore a triangular work organisation and the work cannot be classified as a DPE. Another example is the case in which the client and the platform coincide, the provider works through the platform owned by the client, as in case of teleworking. Again, this cannot be classified as DPE. Both examples show standard two agents relationships, while we are interested in new situations in which the market is characterised by the presence of a third agent. This can result in a smoother market where the platform plays a facilitation role in the matching of supply and demand, but also in market perturbed by the dominant position of the platform.
It is also important that all three sides of the triangle represent an actual connection between vertexes. If the provider has no relationship with the client but only with (receive instructions, get paid, report to, etc.) the platform and the client also has only relationships (a commercial contract or other) with the platform, then there are two different traditional two-agent relationship instead of a triangular one. The decisions taken may exclude some kind of bilateral DPE, and this endanger the identification of some DPE workers in an employee position, but their number has been considered small with respect the high risk to improperly include many people simply in telework.
It has to be noted that for classifying a respondent in employment or not, the focus should be on the provider (i.e. the respondent); characteristics of the platform and of the client should not be taken into consideration. As far as the provider works for pay or profit, there is employment. Platform and client can be based in the country of interest or not, they can be individual or legal entities, etc. The provider must be resident in the country of interest and must be a person to be counted as “employed” or “not employed”.
The ‘use of ICT’ survey is a household, or individual, survey aimed to investigate the diffusion and the trends in usage of new technologies such as the Internet or tools such as computers11. Information needs to be covered are, among others:
access to and use of ICTs (including tools: computers, mobile phones, others)
use of the internet and other electronic networks for different purposes (e-commerce for example),
Focusing on the individuals it covers the supply side, the DPE worker point of view, and the client side, when the client are individuals, such in case of food delivery or taxi services. It may be difficult to distinguish the provider and the client, both answering that they use the platforms. Survey designers should put attention on clearly wording the questions, especially when the data collection mode is web based, with no help from an interviewer. A section on DPE should be explicitly provided and ad hoc designed, among the other information sections of the survey.
It can cover the DPE not covered by the LFS, for example for very occasional providers, which do not even consider themselves as in employment, and can be useful to investigate particular aspects such as the skills of the providers and details on time spent on the platforms.
Canadian Internet Use Survey - Statistics Canada
Original purpose
The 2018 and 2020 Canadian Internet Use Survey (CIUS) covered digital labour platform work through its module on online work. Among other things, the module aimed to capture participation, activities and money earned from online work. There was the possibility of getting some information on the features of the platform workers through questions on education, Aboriginal identify, and other labour market activities.
In CIUS 2018, a filter question asked about whether the respondent had used the Internet to earn income in the previous 12 months. For those who did, a follow-up question asked about whether the income earned was the main source of income or an additional source of income. Finally, a third question asked the method through which the income was earned, with respondents having to choose between seven categories. Each category included examples of platforms used for earning income online.
The full list of the methods to earn money on the internet were the following:
online bulletin board for physical goods (e.g., Etsy, Kijiji, Ebay);
online bulletin board for services (e.g., Kijiji, Craigslist);
platform-based peer-to-peer services (e.g., Uber, Airbnb, AskforTask);
online freelancing (e.g., Upwork, Freelancer, Catalant, Proz, Fiverr);
crowd-based microwork (e.g., Amazon Mechanical Turk, Cloudflower);
advertisement-based income (e.g., income earned through YouTube or personal blogs);
In 2020, the questions about activities to earn money online were asked differently, with each type of activities presented as a separate question. Nine types of activities done online to earn money were presented, compared to seven in 2018.12
Reference population and sampling
The target population of the CIUS was all persons 15 years of age and older living in the ten provinces of Canada. It excluded full-time residents of institutions.
The sample was based on a stratified design employing probability sampling. The stratification was done at the province level. The survey used a two-stage sampling design. The sampling units were telephone numbers grouped and linked to the same address. The final stage units were individuals within the identified households.
Information was collected from one randomly selected household member aged 15 or older, and proxy responses was not permitted.
In 2018, a field sample of approximatively 33 000 units was used on the basis of an expected response rate of about 50%. The size of the sample was chosen based on the requirements to report survey results by key sociodemographic groups, e.g. province. In 2020, the sample was increased to 44 800.
Other relevant survey features
The data collection method was web-based, with follow-ups for non-response by phone.
Implied operational definition
The concept of online work in CIUS was not limited to specific types of digital platforms. Bulletin boards (e.g. Kijiji, Etsy) were included, as well as social media (e.g. YouTube) and more capital-intensive labour platforms (e.g. Uber, AirBnb).
In addition to including categories about selling services, CIUS included one category specifically about selling goods online, with no explicit mention about whether goods sold were used or new. In 2020, for example, the category related to selling goods was restricted to “physical goods that you built or created”, which likely led some respondents to exclude ‘used goods’.
Questions on online work were asked to all respondents, i.e. internet users, irrespective of their employment status. In the 2018 CIUS, definitions were provided for some concepts, including online bulletin board, platform-based peer-to-peer services and platform. These definitions were not provided in the 2020 version. Labour platform work was measured through participation in any of the money-earning online activities listed in the questionnaire.
Obtained goals and learned lessons
In 2018, the concept of using the Internet to earn income may have been measured inaccurately due to question wording and questionnaire flows. The filter question asked whether a respondent used the Internet to earn income. Respondents could have been interpreted this as solely online work, and not thought of Internet as a facilitator, to organize and receive payments for services (i.e. platform-based peer-to-peer services), particularly on-location services. This may have caused an underrepresentation of peer-to-peer service workers in the CIUS.
As well, the “include” statement (“Include money made through online bulletin boards”) in the filter question was the only include statement for the question. It is possible that after reading the statement, some respondents may have not thought about all of the other methods of earning money online.
Finally, the question on whether income earned online was the main source of income was only asked when a respondent said ‘yes’ to earning income online. Without clicking yes to the filter question, respondents may have thought of “income” as a sole money source and deterred people from thinking about minimal amounts earned online.
Business surveys can provide information on the labour market from the demand side. Usually the sampling units are enterprises, although not always all the enterprises of a country are in reference population. Enterprises can include digital platforms (or a digital platform can be part of their core business) or businesses that are clients of platforms.
In both cases this data source typically identify economic units that relies on digital platforms, this kind of entities are more in relation to the OECD concept of digital economy, as discussed in Chapter 2, even if they can also provide information on digital work if asked about their providers. Moreover, if for example sole traders are included then business surveys could be used to also get a direct estimate of part of the digital platform employment.
If the aim is to survey platforms then a register of platforms should be available to build the sample, but this is not always the case: such a register is not yet common in most countries and constructing it presents several difficulties, such as the cross-border feature of many platforms. It is much easier to collect information on business as clients, those who demand the labour factor to the platform. Business surveys can also be used to collect information on existing (and operating in the country) platforms, with the purpose of building the platforms register.
The French statistical office (INSEE) has included some questions on use of platforms into its existing business survey. These questions aimed to cover both business clients and platforms themselves, with no clear distinction between them.
France – INSEE
Original purpose
Every 4 years, INSEE conducts a three-wave survey (called “SINE”) on a sample of newly created or reactivated enterprises. For each sample, information is collected at three moments of company’s life:
the profile of the creator and the conditions for the start-up of the enterprise;
the development of the enterprise's activity, its type of clientele, its co-operation with other enterprises;
changes to employee numbers (hiring, dismissals, occasional staff);
Since 2018, all waves of SINE have included questions on the use of DP. In 2018 (first wave of enterprises created in 2018), respondents were asked whether they were working with DP in the first year of activity and, in this case, whether it was the company’s main source of turnover. The main objective of these questions was to estimate the number of new enterprises working with DP and how important such platform was for their income.13
All types of services and all kinds of platforms were considered, the names of the platforms were not asked. Some examples were given in brackets (ride-sharing services, home delivery, personal services, business services or consulting) to help respondents.
More recently, additional questions about DP were added in the 2021 survey, concerning all types of companies ("auto entrepreneur", sole proprietorship, and other company) started in 2018 and interviewed 3 years after their creation. This survey is still in progress. These questions ask about satisfaction in terms of income generate, terms of contract, access to the market provided by the DP, and freedom in the organization of work (hours, subordination, etc.).
Reference population and sampling
The reference population of SINE covers all the entrepreneurs who created or reactivated their enterprise (excluding agricultural activities) during the 1st semester of the survey year. Stratified random sampling was used, with the reference population stratified by type of enterprise (natural person or legal entity), regions (NUTS level 1) and economic sectors (NACE level 1).
The size of samples was determined to estimate the longevity of the enterprises with enough accuracy five years after creation. The sample for the 2014 survey covered 40 000 entrepreneurs and 40 000 "auto entrepreneurs" who created their enterprise during the first semester of 2014, which corresponds to a sampling rate of 32%. The sample for the 2018 survey covered 24 000 entrepreneurs and 56 000 "auto entrepreneurs" who created their enterprise during the first semester of 2018. This sample of 80 000 enterprises corresponds to a sampling rate of 24 % among the enterprises created in the first semester of the year. For the next cohort (2022), it is planned to sample between 34 000 and 40 000 enterprises of each type (auto-entrepreneurs and others).
Other relevant survey features
The SINE survey programme was launched in 1994. Until now, six generations of start-ups have been observed: 1994, 1998, 2002, 2006, 2010, and 2014. From 2010, the survey incorporates "auto entrepreneurs". The seventh generation, 2018, is currently being sampled.
Source: INSEE, Système d'Information sur les Nouvelles Entreprises, https://www.insee.fr/fr/information/2044890.
Questions about DP were first introduced in both questionnaires in 2018; these questions were then added in the "auto entrepreneur" questionnaire of the third wave of the survey concerning firms created in 2014 and observed (in 2019) five years after set-up. The questionnaire for the second wave of the 2018 generation is in progress and will include additional questions about companies’ satisfaction with digital platform.
The data were collected by post mail until 2019/2020 survey, and by internet and post mail since 2021. Due to sampling methods, INSEE releases results in percentages (rather than level) and for groups containing at least 20 companies in their sample.
Implied operational definition
The SINE survey considers every type of digital platform used for contact with clients, either for selling goods or providing services, including accommodations rental, meal delivery, passengers transport. There were no explicit filters (e.g. YouTube, Instagram were not clearly excluded) but some examples of digital platforms were given in the questionnaire:
No explicit definition of DP was provided in the questionnaire. The main question on DP included is: “Do you work through one or several digital platforms to reach clients (e.g. passenger transport, home delivery, home services dedicated to households, consultancy)?”.14 The question was asked only to own-account workers (creators of an enterprise, either as a natural person or a legal entity). In most cases, the question concerns users of DP but creators of a digital platform themselves (digital platform being considered as an enterprise) are likely to be sampled in SINE (although probably very rarely).
Key results
Results are already available for two cohorts of firms, while those for a new cohort will be available in 2022. Considering companies created in 2018 and it their first year of activity:
16 % of the "auto entrepreneurs" work with DP (2 out of 3 among those in the transportation sector). For three-quarters of them, DP is the main source of turnover. In the transportation sector, two-thirds of the "auto entrepreneurs" work with DP; for 90% of them, the DP is the main source of turnover, while half of them created their business especially to work with DP.
Among the other companies, 10 % work with DP. For half of them, the DP is the main source of turnover. In the transportation sector, 36 % of the new enterprises work with DP, and for 75 % of them the DP is the main source of turnover. A large share of companies in this sector is constituted with taxi companies or ride-sharing companies. Entrepreneurs working with DP are also overrepresented in the accommodation and food service activities (16 %).
DP remain important after the company creation: Among auto-entrepreneurs with companies created in 2014, SINE data show that, by 2019, five years after their set-up. 5 % of the auto-entrepreneurs still in activity use DP. Among them, 54 % use DP as their main source of turnover.15,16
Tax registers can be used as an information source on digital platforms either if a list of platform operating in the country is available (with the problem to identify cross-border platforms) or if the information about working for a platform (or being one, in case of a file of enterprises) is present in the tax register. The main limit of the tax register approach is that is very closely linked to the national legislation, and its feasibility depends on it. This, however, also causes severe harmonization problems: even though the derived information can be very useful at national level, international comparisons are nearly impossible.
Tax register data on digital platforms is available in Belgium following enactment of a specific law in 2016, defining a specific tax regime for incomes of enterprises in the “sharing economy”. Statbel, the Belgian statistical office, uses information on the tax situation of individual workers in the sharing economy for producing experimental statistics on the number of platform workers and their earnings. While not all informal work arrangements are covered by these registers, which can led to an underestimation of the number of platform workers and on the total income generated by the platforms, the introduction of a specific tax regime for these firms aimed to limit as far as possible the area of informal employment in DPE.
Belgium – Statbel
Original purpose
The Belgian Programme Law of 1 July 2016 introduced a specific tax regime for companies in the sharing economy. Platform companies are compelled by law to declare the income of its platform workers to the Belgian tax authority on an individual basis. The purpose of this tax regime was to provide fiscal advantages to platform work. By analogy with other administrative files, the new dataset containing the earnings of platform workers primarily serves an administrative/fiscal purpose. However, the Belgian statistics law provides the possibility for Statbel to receive and to use administrative data in order to compile anonymous statistics. For this purpose, a confidentiality contract was concluded between the Belgian tax authority and Statbel.
The Belgian tax register of the sharing economy includes the following five variables:
Reference population
Between 2016 and 2020, the tax authority attached three conditions to the application of the favourable tax regime for sharing economy enterprises:
the scheme was limited to services provided by a private individual to another private individual;
the platform company had to have official recognition, granted by the Belgian tax authority;
the favourable tax regime was only applicable if the gross income from platform work was below a predefined maximum amount. In 2020, this gross maximum amount equalled to EUR 6,340 on an annual basis.
In fiscal year 2019, 85 companies were granted recognition as platform companies. These 85 companies reported the income of 20 600 platform workers to the tax authority. From 2021 onwards, platform work is no longer limited to the recognised platform companies and the regulation regarding the maximum amount has also been dropped. All income from platform work must be reported to the tax authority via the same regulation, which should improve the completeness of the file.
Implied operational definition
The description of the reference population in the law lists the conditions that the tax authority attaches to the application of the favourable tax regime for platform work. First of all, the regulation only applies to services that a private individual provides to another private individual. The tax authority excludes the transaction of a good, including the rental of an accommodation, from its definition of platform work. However, there is an exception for transactions involves the provision of a service. The rental of an accommodation is for instance considered as platform work when cleaning of the room or a breakfast are provided.
The favourable tax regime is only applied if the gross income from platform work was below a predefined maximum amount. For 2020, this gross maximum amount was EUR 6 340 on an annual basis; when exceeding this amount, the tax authority considered the platform worker to be self-employed. Platform workers who earned more than the maximum amount were therefore not included in the tax register of the sharing economy, but in the regular tax register that collects professional income. The maximum amount also implied that platform work was always considered as an extra income and could not be the main activity.
Under the rules described above, platform work can be performed by both inactive persons (e.g. students or retired people) and employees or self-employed persons. An additional restriction applies for self-employed persons: the activity of platform work may not be their main activity; in other terms, a self-employed gardener may not perform platform work as a gardener but may provide another service within the sharing economy. Until 2020, the platform company had to have official recognition granted by the Belgian tax authority.
A new regulation applies since fiscal year 2021. Prior recognition as a platform company by the tax authority was abolished and the predefined maximum amount to earnings from DP no longer applies. All income from platform work has to be declared to the tax authority via the same regulation from 2021 onwards. The new regulation should improve the completeness of the specific register for platform workers.
Lessons learned
Statbel's experience with the use of tax registers in order to map out platform work is largely positive, thanks to the specific tax status of platform workers, which leads to a separate file dedicated to the sharing economy within the tax register. It is also important that the statistical institution has easy access to administrative data, which is facilitated in Belgium by the national statistical law. On the other hand, the use of administrative data also creates dependency and the data are not representative for the whole population. Main advantages and disadvantages that Statbel encounters when using the tax register include:
1. The specific tax regime for the sharing economy means that platform workers have their own file within the tax register. This allows platform workers to be easily identifiable while providing that Statbel an extensive and high-quality database. The tax register for the sharing economy contained individual data of 20 600 platform workers in 2019.
2. The tax register contains both a unique enterprise number for the platform company and a unique individual number for the platform worker. Based on both identifiers, Statbel can match the tax register to other databases, which offers multiple, relevant possibilities of analysis. For the platform company, couplings made with the statistical business register and VAT statistics (among others) provided information on economic activity (NACE nomenclature), size class, location and turnover. For platform workers, matching with other files within the tax register generated data on the status of the platform worker (employed, retired, student, etc.).
1. The tax authority determines the content of the register based on its tax needs. As a statistical institution, Statbel has therefore no say on aspects such as the available variables, concepts used, target population or timeliness. This creates a great dependency and may lead to Belgian definitions being different from international concepts.
2. Not all platform companies applied for official recognition, so that platform workers of non-recognised companies were considered as self-employed for tax purposes. In addition, given the predefined maximum income amount, the activity of platform work was never considered as the main work. The tax register is therefore not exhaustive, as some platform workers do not meet the conditions set by the tax authority.
3. Large platform companies often operate in Belgium from an office abroad. These companies are not included in Belgian statistics or databases, which means that the advantage of additional matchings with other databases is lost.
An ad-hoc survey, specifically focused on DPE, can gather very good information on thee more qualitative aspects of the platform work, such as working condition and income. The reference population can be targeted to capture the DPE phenomenon, with the sample adequately representing this population or at least covering it in a well-controlled way, thereby allowing to investigate even small or very small aspects of the phenomenon. As an example, the sub-population of platform workers can be the reference population, setting to one the probability of interviewing the persons object of the analysis, but losing the connections with the entire population. Planning an ad hoc survey requires good a priori information, for example a method to identify all platform workers to be included in the reference population. When this a priori is of poor quality, preventing the reference population to cover all interested persons, there will be high probability to have a biased, or at least self-selected sample. In the planning process of these ad hoc surveys, both the choice of the reference population and of sampling framework are of paramount importance.
An ad-hoc survey requires a huge amount of preparation, after which it can provide information not available otherwise. The Joint Research Centre of the European Commission (JRC) runs an ad-hoc survey in the context of the COLLEEM research project on collaborative economy.
European Commission – Joint Research Centre
Original purpose
The scope of the COLLEEM survey is broad and includes the following objectives:
Assess and quantify in an homogeneous way the prevalence of platform workers across selected EU countries and in Europe at large.
Provide a classification of platform workers on the basis of the frequency and intensity of platform work, and its relevance for individuals’ total income.
Profile platform workers in terms of their socio-demographic characteristic (e.g. age, gender, education; household composition; migrant status).
Identify the actual and perceived labour market status of platform workers (i.e. dependent employee or self-employed).
Identify the type of tasks that are provided via platforms (i.e. on-line professional, on-line non-professional, on-location)
Identify the motivations of workers for engaging in platform work.
Assess the work and employment conditions of platform workers.
Reference population and sampling
The reference population of each country included in the COLLEEM survey are all internet users aged 16 to 74. COLLEEM is a self-administered online panel survey. The survey followed a quota-based ‘non-probability sampling design’. Quotas of respondents were established to provide representative estimates by age groups (16 to 24, 25 to 54, and 55 to 74) and gender (male/ female). Targets of completed responses in each group were allocated proportionally to the size of each group in the total population of internet users aged 16-74 in each country.
COLLEEM used Eurostat’s most recent data on population by age and sex from the ‘European Labour Force Survey’ (LFS) for calculating the population in each age-gender category in each country. These figures were then multiplied by the proportion of internet users in each respective age-gender category (obtained from the Eurostat’s ‘Community survey on ICT usage in households and by individuals) to compute the population of internet users by age-gender groups.
The first pilot (COLLEEM I) wave was completed in 2017 gathering 32 389 responses from 14 member states. The COLLEEM II survey was conducted in 2018 gathering a total of 38 022 responses, from internet users aged between 16 and 74 years old, in 16 EU member states (Germany, France, Italy, Spain, Finland, the Netherlands, Sweden, Hungary, Slovakia, Romania, Croatia, Lithuania, Ireland, the Czech Republic, Portugal and the United Kingdom). COLLEEM aimed to gather a minimum of 2 300 responses per country. Countries were selected based on the following characteristics:
Internet availability, skills and use, clustered in 3 groups;
Geography, classified into ‘Nordic’, ‘Western Europe’, ‘Central and Eastern Europe’ and ‘Mediterranean’.
The survey was designed to capture the diversity of groups created by intersections of these three dimensions.
Other relevant survey features
The survey was administered using SurveyGizmo and disseminated to respondents of an online panel survey aggregator CINT (www.cint.com). A unique feature of the COLLEEM II survey is that it includes a longitudinal sub-sample, consisting of respondents who were interviewed in both 2017 and 2018, as CINT allows to reach the same individuals who have previously participated in a survey based on their unique identifiers. The longitudinal sub-sample consists of two parts: the first part is made up of re-invites, i.e. all people identified as platform workers in COLLEEM I and re-invited to take part in COLLEEM II; the second part consists of people (whether identified as platform workers or not) who were by chance selected again from the sampling frame to be part of the COLLEM survey. Taken together, the longitudinal sub-samples of COLLEEM II allow to analyse individual transitions in and out of platform work, as well as to check the robustness of the cross-sectional findings.
Implied operational definition
The broadest definition of platform workers used in the COLLEEM surveys refers to workers who have ever gained income from providing services via online platforms, where the platforms digitally match provider and client and ensure the payment, and where work is performed either on-line (location-independent) or on-location. Within this context, digital labour platforms are defined as digital networks that coordinate labour service transactions in an algorithmic way. It is also worth noting that the definition of platform work used in COLLEEM is restricted to labour services, thus excluding digital platform employment where the main purpose is selling goods, or to provide services such as renting out an apartment, where labour input is marginal. Definitions are identical in COLLEEM I and II.
In an effort to narrow the definition and to identify different categories of platform workers, COLLEEM II data divides platform workers into ‘sporadic’, ‘marginal’, ‘secondary’, and ‘main’ based on a combination of number of weekly working hours (less than 10, 10 to 19, more than 20) and contribution of the platform work to workers’ total income (less than 25%, from 25 to 50%, over 50%) (see Table 5.3).
Regarding the information on working tasks, COLLEEM elicits information on ten different types of tasks, combining locus of provision and skill level:
online clerical and data-entry tasks (e.g. customer services, data entry, transcription, and similar);
online professional services (e.g. accounting, legal, project management and similar);
online creative and multimedia work (e.g. animation, graphic design, photo editing and similar);
online sales and marketing support work (e.g. lead generation, posting ads, social media management, search engine optimisation and similar);
online software development and technology work (e.g. data science, game development, mobile development and similar);
online writing and translation work (e.g. article writing, copywriting, proofreading, translation and similar);
online micro tasks (e.g. object classification, tagging, content review, website feedback and similar);
interactive services (e.g. language teaching, interactive online lessons, interactive consultations and similar);
transportation and delivery services (e.g., driving, food delivery, moving services and similar);
on-location services (e.g. housekeeping, beauty services, on-location photography services and similar).
Key results
The COLLEEM survey was a pioneering attempt to measure in a homogeneous manner the prevalence of platform work across European countries. The survey has provided insights on the socio-demographic profile of platform workers, their motivations to enter into this type of activity, their actual and perceived professional status and their working conditions. Key results from COLLEEM were presented in Chapter 2.
Lessons learned
The experience of COLLEEM can provide useful insights for designing ad-hoc surveys on platform work in the future, always keeping in mind the limitation and potential source of bias of using on-line surveys for measuring platform work. In particular:
Gathering information on the frequency, hours and income generated through platform work is crucial to gain a more precise and policy-relevant measure of the phenomenon. As shown in COLLEEM II, it is important to classify platform workers on the basis of the frequency and intensity of platform work, and its relevance for individuals’ total income. Disentangling different types of platform workers according to these criteria allows to assess the share of workers who provide labour services through platforms as their main job (or at least frequently), their socio-demographic profile and perceived employment status – all aspects which may differ significantly from those of “sporadic” platform workers. This is particularly relevant for informing labour market policies aimed at regulating platform work.
Gathering information on the specific tasks carried out can provide insights on working conditions and on how platforms affect the nature of work. COLLEEM II survey asks to frequent (at least monthly) platform workers on which specific task they spent most of their working time, how long it usually took to complete such a task, how much they earn, and which platform they used to carry out the task the are referring to. At the aggregate level, collecting this type of information provides useful insights on the bundling and distribution of work tasks. Moreover, as more than 69% of platform workers report being paid on the basis of the performed task, collecting information at the task-level allows to directly link information on income with data on worked hours, working conditions, and the main used platforms. This is useful since platform workers often carry out different tasks through various platforms.
There are ways to address issues of representativeness and measurement error that are typical of online surveys. Self-administered online survey using non-probabilistic samples are useful in exploratory research aiming at identifying whether a specific and relatively rare characteristic – such as being a platform worker – exists in a population. However, whilst suited for the general aims of the COLLEEM, these types of surveys are affected by a number of possible errors, such as biased sample and reference population, unreliable responses, no knowledge of who is actually answering and under what conditions. In order to mitigate some of these issues, COLLEEM used a number of techniques to guide respondents throughout the questionnaire, including:
Finally, experience with COLLEM has shown that the key screener question, aimed at ascertaining whether respondents provide services via digital labour platforms, inevitably implies formulating complex and lengthy questions which are difficult to comprehend properly and are potent sources of respondent fatigue.
Self-administered online panel surveys may not be representative of the situation of disadvantaged groups. The fact that respondents are contacted and the information is collected entirely online, make it likely that some of the most disadvantaged forms of platform work (such as delivery or other low-skilled in-person services) are underrepresented in the survey. These more disadvantaged forms of platform work are often taken up by groups of the population that find it difficult to participate in the regular labour market (such as migrants). This implies that results on the overall working condition of platform workers may be biased, reflecting more the (better) conditions of online professional types of platform work than those of on-location personal services such as delivery or transport.
Conducting a pilot survey can prove useful to improve the representativeness of the actual survey. The COLLEEM pilot survey showed that reaching certain types of internet users, notably those with lower-than-average frequency of internet use and younger users with low formal education, can be very difficult. As a result, these groups remained rather underrepresented in the survey, requiring the use of larger sample weights and a loss of precision of the estimates. Therefore, in order to improve representativeness, and before making any weighting adjustments, COLEEM II focused on the panels which produced more representative responses than others in the first COLLEEM wave, targeting these priority panels during the fielding.
In light of the limitations of online panel surveys, and in a view to complementing existing evidence from COLLEEM I and II waves, the JRC is implementing a third wave of COLLEEM. This third wave will include the same questions for platform workers as COLLEEM I and II, but collected in the context of a face-to-face interview and with the respondents being randomly drawn from the population, using area sampling. Although only two countries will be covered (Spain and Germany), the sample will be considerably larger (between 3 000 and 4 000 cases per country) and of much better quality. This will provide information to assess the possible biases of the online panel approach used in previous COLLEEM waves, on top of providing the usual information. The third wave of COLLEEM will introduce a broader concept of platform work, understood as the use of digital networks for the coordination of work processes in all kinds of organisations and economic activities, and the related concept of algorithmic management. In other words, the third wave of the COLLEEM survey aims to broaden the concept of platform work, collecting information on the use of platform-like digital tools for the coordination of work processes in all sectors of economic activity (not only in digital labour platforms as done in COLLEEM I and II). The survey will also aim to gather information on how platform-type tools for coordinating work processes are impacting on work organisation, job quality and industrial relations.
European Commission – Eurofound
The rise of Web 2.0 technologies has facilitated the collection of data from internet users as a valuable source of information that can complement traditional sources such as surveys, expert opinions or data from stakeholders. The appeal of crowdsourced data stems from the lower costs of data collection and the potential for reaching a large number of respondents who would otherwise be inaccessible. However, these advantages need to be weighed against drawbacks around data quality, representativeness and consistency.
Crowdsourcing data are gathered by a public or private entity based on an explicit request for individual level users. Participation in data gathering exercises can be either voluntary or incentivised though a monetary or non-monetary incentive such as wage information of an employee relative to employees working in the same occupation, sector or jurisdiction. While active crowdsourcing implies a direct data request from platform users, the collection of user-generated content (UGC) is based on big data or web-scrapping and web-crawling tools to gather, sort and classify content. The latter approach was employed by (Kässi and Lehdonvirta, 2018[3]) who used big data to develop the Online Labour Index (OLI), an indicator of the size of the online platform economy. OLI uses the number of unique visitors and vacancies on a platform to measure change in the online labour market. A limitation of this approach is that OLI does not record the actual labour performed on platforms or information on working conditions but only the demand for labour. Another limitation of this approach is that the sample is limited to English language platforms.
Social media can also be used to reach smaller and geographically scattered cohorts of platform workers. Provided that stratification techniques are used to reduce bias, social media websites such as Facebook can be used to target platform workers and gather information though simple and short surveys. Facebook allows gathering geolocation data and behaviours specification – keywords such as expats, platform, rider - which allow for the survey to be targeted to relevant groups. Although appealing for its simplicity, this approach has several shortcomings stemming from the inherently biased sample of Facebook users.
The changing regulatory landscape around the platform economy can also facilitate data gathering exercises. In countries such as Belgium, Denmark, France or Estonia changes in the tax systems have introduced reporting systems that allow obtaining income data from platforms. While the data gathered through these sources can complement traditional surveys, their quality depends on the cooperation of platform companies as well as specific rules around income thresholds and reporting requirements (Eurofound, 2021[4]). For example, the Estonian system which is operational since 2017 operates on a voluntary basis and is used mainly by ride-sharing and rental platforms. While promising, current attempts to use big data to gather information about the scale of the platform economy are useful only in conjunction with traditional statistics.
The overview of statistical experiences presented above provides an overall picture of the situation: the variety of sources and approaches shows that there is not a single answer to the problem of measuring digital platform employment. Integration of information from different sources seems to be the optimal choice. The complexity of the measurement derives from the particularity of the DPE market: this is often a three agents market, with a provider (the worker), a client (that can be an enterprise or a household/individual) and an intermediary (the platform, which typically is an enterprise). This makes the diversity of sources an asset for statistical activities in this field.
While lessons learned, for each source, were already suggested by the TEG contributors in the specific sections above, some important general recommendations are provided below.
The labour force survey may be the most relevant source of information for the analysis, and for designing policy, of the labour market. It may evolve to be the best suited to provide quantitative information on the supply side of the labour market, i.e. from the workers’ perspective, and, having usually a high frequency, can promptly catch new dynamics on the market. It provides benchmarks for all other sources. Its main drawback is that it is not best suited in providing information about small or very small phenomena, like the DPE still is. Another problem of the LFS is the information on earnings, which is not always present or reliable.
The statistical experiences based on LFS described above highlight a range of differences in terms of whether questions on DPE are asked to all respondents or to subset thereof (e.g. those classified as “employed”, or own account workers), whether they include or exclude activities related to renting out capital goods (where labour services only have an auxiliary function); whether different typologies of digital platform employment should be distinguished; what reference period should be considered in order to classify workers as performing a digital platform activity (e.g. previous week or previous year); whether digital platforms should be limited to those that, in addition to matching workers and clients, also manage the payment between the two; whether specific digital platforms should be mentioned to guide respondents; and more. While no single answer to all these questions exists, and the best approach will partly depend on the number of questions that could be asked (i.e. few, in the case of questions included in the general questionnaire of LFS; potentially more, in the case of ad hoc or recurrent LFS modules), a number of general recommendations can be provided at this stage:
The LFS should be the tool of choice when it comes to attempting to measure the number of people involved in digital platform employment.
People’s activities taking the form of an “employment relation” mediated by digital platforms should be clearly distinguished from activities referring to a broader notion of “work” and from those involving the rental of capital goods owned by survey respondents. While some NSOs may narrow their questions to DPE, those opting for a broader remit should do so in ways that allow to clearly distinguishing between the various activities.
While some NSOs may use a longer reference period to assess workers engagement with digital platforms, LFS questionnaire should always include questions about employment in the survey reference week, so as to allow measuring the incidence of DPE among all employed people.
In addition to basic breakdowns of DPE by demographic and other characteristics included in LFS, LFS should ask respondents on the “regular” or “occasional” nature of their DPE relation, based on either the number of hours worked or the earnings gained.
Digital platforms should meet the twin criteria of both intermediating between clients and service providers, and playing a role in managing the payment for the services provided. NSOs relying on broader definitions, should implement them is ways that allow narrowing the focus on the above definition.
Naming of specific digital platforms should be avoided when first asking questions on DPE, but could be used in follow-up questions.
The Information and Communication Technologies (ICT) Use survey on households and individuals is a natural source for information on DPE, and can offer information on the supply side and on the demand from private individual side. A section of the questionnaire, focusing explicitly on the DPE is necessary, since not all currently in use survey are studied with this purpose and confusion can arise by the fact that both, providers and clients ‘use’ the ICT tools and ad hoc questions should be studied to distinguish them. It can provide precious information on particular aspect such as time spent on work and needed skills.
Business surveys typically is the instrument for collecting data around the digital economy as outlined in chapter 2, however it can provide information on both, demand (those not covered by the ICT survey: the business demand) and supply side of the labour market, with insight on the individual income, which is a valuable complement to the LFS data. They may cover the platform themselves or the enterprises playing the role of clients in the market. In the first case, it will be difficult to reach all the platform operating on a country, especially for platforms based abroad, in the second case it is impossible to cover non-business clients. Particular important is to state the reference population of the survey, which may not always cover the entire enterprises population. To the extent possible, business surveys:
Should cover the universe of business units operating in a country, focusing on business’ demands for services provided by digital platforms.
Rely on a definition of digital platform that is closely aligned to that used by LFS, i.e. platforms that both mediate between clients and providers, and that intermediate the payment for the labour services provided.
Provide information on the quantitative importance of digital platforms for their turnover, as well as questions on their satisfaction with that relation.
Include common breakdowns of businesses (e.g. by size of their payroll, industry, annual turnover, ownership type) allowing to compare businesses’ reliance on digital platforms across different parts of the business community.
Tax registers, or in general other administrative registers, can provide information from both the platforms (when it is possible to identify them as tax payers) and from the workers (when it is possible to identify them as DPE workers). Information on income is the main item that is best covered by this source. However, coverage of tax registers will be limited to formal enterprises and workers, which is an important limit in some countries. Moreover, when low paid workers are exempted from tax declaration and/or payment, they are also excluded from the reference population.17 More generally, the reference population of tax registers and other administrative data will be affected by national legislation that usually has not statistical purpose, limiting the representativeness of the source and the cross-national comparability based on them of measures.
When the required information focuses on particular qualitative features of the work, such as working condition or working satisfaction, an ad hoc survey is the best choice. They can be specifically designed to focus on small or very small phenomena, ensuring a representative sample. While they require a big planning effort and dedicated resources, they provide information that is not feasible to collect with other sources, information which is often necessary to calibrate labour market policies focusing on relevant but small (or not yet big) phenomena.
Ad-hoc surveys such as COLLEEM have the advantage of typically covering a broad range of countries, hence providing comparable evidence on different aspects of DPE. Consideration should be given to including a small set of comparable questions in other non-official surveys on working conditions (e.g. EWCS, ISSP, etc.), while aligning concepts and definitions to those used in LFS.
Big data-based sources are in an early phase of investigation. They are promising but it is clear that they can just complement the other sources: by their nature it is difficult to plan their information coverage, this is more an output of the study than an input since the information extraction and its own definition proceed in parallel. Moreover, private ownership of the data makes them expensive and unreliable its use for official statistics. Also, continuity in time of the availability of this kind of data cannot be ensured. This makes this source more feasible, at the moment, for one-shot investigation on one particular aspect of digital platform employment, always in conjunction of traditional sources. Regular production of labour market indicators of DPE from big data sources does not seem a realistic option at present.
Whatever the data source, it is important to document the number of jobs exercised by digital platform employees and/or the frequency of DPE activities over the LFS reference period and over longer reference periods.
A further reflection should cover the role of well-designed policies for providing high quality statistical information. It is clear, for example, that tax legislation providing special regimes to DPE incomes helps in identifying the reference population and, more generally the boundaries of the phenomenon. Moreover, some countries have introduced obligations for big data producers or for labour digital platforms to provide their data to statistical authorities, and this is a real boost for high quality data production. This is an important area for future development.
Conversely, measures of digital platform employment should be independent of legislative changes that classify digital platform workers. Failing to ensure the independence between legislation and statistics would impact on both cross-sectional and time series estimates of DPE.
References
[4] Eurofound (2021), Initiatives to improve conditions for platform workers: Aims, methods, strengths and weaknesses, Publications Office of the European Union, Luxembourg.
[2] INE (2022), Separata técnica n°4: Nuevas dimensiones de análisis, https://www.ine.cl/docs/default-source/ocupacion-y-desocupacion/publicaciones-y-anuarios/separatas/tem%C3%A1ticas/2022-07-25_nota-t%C3%A9cnica-nuevas-dimensione.
[1] INE (2021), Documento Metodológico Encuesta Nacional de Empleo (ENE)., https://www.ine.cl/docs/default-source/ocupacion-y-desocupacion/metodologia/espanol/metodolog%C3%ADa-encuesta-nacional-de-empleo-ene-2020.pdf.
[3] Kässi, O. and V. Lehdonvirta (2018), “Online Labour Index: Measuring the Online Gig Economy for Policy and Research”, Technological Forecasting and Social Change, https://ilabour.oii.ox.ac.uk/new-publication-online-labour-index-measuring-the-online-gig-economy-for-policy-and-research/.
Notes
← 1. Note that names are used if the question is asked about people in the household other than the respondent (The CPS accepts proxy responses). If respondents answer “yes” to the in-person question (Q1) or online question (Q2), they are asked the follow-up “which job” question (Q1a or Q2a). The question wording and response option for "second job" only appears for people who were previously identified in the survey as having more than one job. For people with only one job, the question read "Was that for your job or additional work for pay?" People with only one job are asked whether the in-person platform work was for their job or additional work for pay.
← 2. For more information about the development of the questions, evaluation of the data, and recoding of answers, please refer to Electronically mediated work: new questions in the Contingent Worker Supplement . (https://www.bls.gov/opub/mlr/2018/article/electronically-mediated-work-new-questions-in-the-contingent-worker-supplement.htm#_edn12).
← 3. MMV 2017: The Sampling Frame of Dwellings (MMV, Spanish Acronym) is based on the results of the abbreviated Population and Housing Census of 2017 and the lists of dwellings associated with the 2016 Pre-Census. The MMV 2017 has been regularly updated during intercensal periods, using administrative records and, when necessary, an enumeration of prioritized geographic units.
← 4. MMV 2017: The Sampling Frame of Dwellings (MMV) is based on the results of the abbreviated Population and Housing Census of 2017 and the lists of dwellings associated with the 2016 Pre-Census. The MMV 2017 has been regularly updated during intercensal periods, using administrative records and, when necessary, an enumeration of prioritized geographic units.
← 5. The main activity is the activity in which the person worked the most hours in the reference week, regardless of whether the person engaged in another activity of shorter duration for which the person will be paid more.
← 6. CAPI: computer-assisted personal interview.
← 7. Only those applications and platforms that appear in the dictionary are searched for in the description because they are the most likely to appear. However, other applications and platforms may be included in the dictionary as they enter the Chilean labour market.
← 8. For official estimates, INE uses the Standard for Assessing the Quality of Estimates in Household Surveys to determine the reliability of an estimate.
← 9. Own account workers are persons who operate their own business without hiring any paid employees.
← 10. As adopted in the 19th International Conference of Labour Statisticians, Resolution concerning statistics of work, employment and labour underutilisation.
← 11. The ICT can also be designed to collect information from the enterprises side (ICT survey in business) but here the focus is on the households and individual survey.
← 12. 2018 questionnaire; 2020 questionnaire.
← 13. "Auto-entrepreneurs" (or “self-entrepreneur” in English, are self-employed entrepreneurs with a specific status including simplified formalities) were also asked whether working through DP was one of their main reasons to choose the "auto-entrepreneur" scheme. The same question was also asked in 2019 to auto-entrepreneurs who started their business in 2014, i.e. five years after creation. The question was not asked to other types of entrepreneurs.
← 14. “Travaillez-vous par l’intermédiaire d’une ou plusieurs plateformes numériques de mise en relation (exemple: VTC, livraison à domicile, services à la personne, services ou conseil aux entreprises) ».
← 15. Data from the 2022 survey (expected by early 2023), and referring to companies created in 2018 and observed three years after creation, will also cover: i) The year when the entrepreneur started working with DP; ii) its satisfaction level in terms of the turnover made by working with DP, the legal agreement between the entrepreneur and the DP, the access to the market through DP, and the degree of freedom in the organization and in work (hours, subordination, etc.) provided by DP.
← 16. In the 2022 survey, INSEE will add some questions on the digital platform module, asking new business owners whether they use a DP to get a better visibility, to benefit from a turnkey application, to expand their activity, the difficulty to access the market without DP. These items will be tested in February 2022 on a sample of new enterprises, with the possibility for the respondent to mention other reasons of using DP.
← 17. As an alternative or complement to the use of tax registers as a source of information on the income of digital platform workers, consideration should be given to including specific questions in official household income surveys.