4. A critical review of existing statistical sources on digital platform employment

This chapter is a review of existing sources and metrics to measure digital platform employment, limited to online and location-based services mediated by digital labour platforms. It therefore generally excludes digital platforms whose objective is selling or renting goods and assets, unless differently specified. The measurement of internal digital platforms employment (i.e. platform workers who are engaged by the digital platform as employees) is also generally outside the scope of this review.1

Due to the lack of internationally agreed definition of digital platform work and employment, the terminology used in the reviewed papers is not harmonised. When discussing the findings from the reviewed sources, the current chapter reports for completeness also the original terms used, either in the main text or in footnotes. Information in the tables is also based on the original terminology.

The objective of this chapter is to: i) review what measurement initiatives on digital platform employment have been undertaken so far2; ii) identify the lessons learnt from these initiatives; iii) understand the pros and cons of the various statistical vehicles for answering to different policy issues.

Since the emergence of digital platform employment, there have been several attempts to estimate the number of digital platform workers. Initial attempts made use of existing data sources, combined with strong assumptions. A number of surveys conducted by both researchers and private agencies followed, with government agencies having sponsored some of the research. Since then, official statistical agencies of OECD Members have begun to introduce questions on digital platform workers into Labour Force Surveys (LFSs) and Internet Usage Surveys. Lastly, big data or administrative data, such as social security or tax data, have been used to estimate the number of digital platform workers.

The chapter looks at the attempts to measure digital platform employment by private agencies and official statistical agencies through surveys; highlights innovative uses of data; and concludes by discussing advantages and disadvantages associated with the different measurement methods.

Researchers have commonly used surveys to estimate the number of digital platform workers, though with wide variation in estimates. Surveys carried out by non-official organisations are presented first (summarised in Table 4.1), as chronologically have preceded surveys carried out by national statistical agencies (summarised in Table 4.2).

In the United States, (Katz and Krueger, 2016[1]) aimed to meet the lack of official statistics by conducting a version of the Bureau of Labor Statistics’ (BLS) Contingent Workers Survey (CWS) and found that 0.5% of the workforce identified customers through an online intermediary3. In line with existing labour market statistics, the survey referred to work done in the past week, although they used a different sampling method. In contrast, the PEW Research Centre used a broader definition of digital platform worker (including those who engage in digital platform employment as a secondary job) and a longer reference period (looking at those who engaged in digital platform employment in previous 12 months) and found that 8% of US working age adults were digital platform workers (Pew Research Center, 2016[2]). Several attempts have also been made to estimate the number of digital platform workers in Europe.

For the United Kingdom, the CIPD (a representative body for British Human Resource professionals) used an online survey and concluded that 4% of British adults had engaged in digital platform employment in the past 12 months in 2016 (CIPD, 2017[3]). Despite using a broader definition (of gigs, including work found using a digital platform), a slightly lower prevalence was provided by the Royal Society for the Encouragement of Arts, Manufactures and Commerce, for the share of British adults who tried gig work of some form, 3.1% (Balaram, Warden and Wallace-Stephens, 2017[4]). Using a definition of “gig economy” limited to including digital labour platforms only, both as main and secondary source of income, an online survey in Great Britain (Lepanjuuri K., 2018[5]) found that 4.4% of the population had “worked in the gig economy” in the 12 months previous to the survey. To correct for potential selection bias due to carrying out the survey online, the panel also included members responding by telephone. Huws et al. (2019[6]) found that 5.2% of the population in the United Kingdom had worked at least once a week for digital platforms in 2016, and that this share doubled to 9.4% in 2019.

In Germany, Bonin and Rinne (2017[7]) used a telephone survey to estimate that 2.9% of adults at some point in the past had engaged in digital platform employment. Evidence from this survey showed that respondents often misunderstand the definition of digital platform employment, and tend to classify online activities, such as job search websites, as digital platform employment. As a high number of respondents could not name the digital platform they were working for, or named platforms not related to labour platforms, the researchers corrected the share of real digital platform workers (“crowd workers”) to 0.85% of adults.

In France, Le Ludec et al. (2019[8]) used a combination of three methods to estimate that about 320 000 workers (about 0.8% of the working population) are registered in digital platforms mediating offer and demand of “micro-work”. The latter is a specific subset of digital platform employment, where workers are engaged to carry out “micro-tasks”, i.e. small independent units of larger tasks which are to be carried out independently, often remunerated with small amounts of money (ILO, 2018[9]). The authors selected the main micro-work platforms operating in France and used the results of Digital Platform Labour (DipLab) survey to apply a specific “capture-recapture” method.4

Two Scandinavian surveys highlight the importance of choice of question (see Annex A2). In a telephone survey, Alsos et al., (2017[10]) found that 0.5% to 1% of Norwegian working age adults have used a digital platform (including also platforms for renting accommodation, such as AirBnB) to earn income in the past 12 months. They found that questions asked over the phone gave more accurate responses than online surveys, as does mentioning specific digital platforms. An earlier survey carried out in the country among 1,525 Norwegian adults had found higher estimates: 10% of respondents indicated they had done work for a platform at some point and 2% said they performed platform work on a weekly basis (Jesnes et al., 2016[11]). The importance of specifying whether an individual provided, or merely offered a service, is highlighted in a report for Government of Sweden, which found that although 4% of Swedish working age adults searched for work via a digital platform, only 2.5% were successful (SOU, 2017[12]).

There have been several cross-country studies of digital platform workers. McKinsey Global Institute conducted an online survey of 8 000 workers across six countries (the United States, the United Kingdom, Germany, Sweden, France, and Spain) and found approximately 1.5% of respondents have earned income via digital labour platforms in the pooled sample (Manyika et al., 2016[13]).

Huws et al. (2019[6]) estimated the share of digital platform workers based on online surveys carried out in 13 European countries5 between 2016 and 2019, either as an addition to an existing omnibus survey or as a standalone survey. Data collected through samples of about 2 000 respondents in each country led to estimates of the number of regular (at least weekly) digital platform workers ranging from 4.9% of the working population in Sweden and the Netherlands in 2016 to 28.5% in the Czech Republic in 2019. However, differences in the age ranges in the samples limit cross-country comparability of this study. Estimates of the prevalence of digital platform employment from this survey are higher than those found in other surveys. This may derive from selection bias and overrepresentation of online workers among the respondents, particularly those used to perform micro-task work, such as filling online surveys. In addition, the effect of paying the respondents to answer the survey may add a bias. To assess potential selection biases in online surveys, the authors carried out companion offline surveys in two countries: a face-to-face survey in the United Kingdom and a telephone survey in Switzerland. Although the two UK surveys returned similar results, those carried out in Switzerland by telephone yielded lower estimates of digital platform workers (1.6% of total population aged 15 to 89 years) than those measured through the online survey by the authors.

In Europe, cross-country surveys have been undertaken by Eurobarometer and the European Commission. A Eurobarometer poll estimates the number of adults who provided a service using a digital platform in 2016 (and updated in 2018), including digital labour platforms, car sharing and digital platforms to rent accommodation. This survey highlighted wide variation across countries in the number of workers having offered their services through a digital platform at least once, ranging from 16% in France to less than 1% in Malta in 2016.6 The study also highlighted the importance of choosing an appropriate reference time, as those who regularly supply a service are a small fraction of those who do so occasionally (Eurobarometer, 2016[14]; Eurobarometer, 2018[15]).

Findings from the European Commission’s Joint Research Centre Collaborative Economy and Employment (COLLEEM) pilot survey conducted in 2017 in 14 EU Member States and repeated in 2018 across 16 EU Member States7 (both fielded by the Public Policy and Management Institute) are described in Chapter 1 of this Handbook. According to COLLEEM, the share of adults who provided services via online platforms monthly (digital labour platforms only) was 11% in the 16 countries surveyed in 2018, slightly higher than in 2017 (9.5%). Estimates from COLLEEM are affected by some methodological limits. The survey was conducted online among frequent Internet users, thus leading to potential self-selection bias, particularly of those providing professional services online. Potential self-selection bias was corrected for by using weights for education, employment status, and frequency of Internet use (based on Eurostat’s LFS and ICT survey) when reporting results for the adult population as a whole. However, bias in this survey may remain (Pesole et al., 2018[16]); (Urzì Brancati, Pesole and Fernández-Macías, 2020[17]); (Piasna and Drahokoupil, 2019[18]).

To overcome potential biases of paid, opt-in online surveys, Piasna and Drahokoupil (2019[18]), collected data on digital platform workers in five central and eastern European countries (Bulgaria, Hungary, Latvia, Poland and Slovakia) through the ETUI Internet and Platform Work Survey, using stratified random sampling of the entire population and face-to-face interviews. The respondents were not remunerated for their participation in the survey. Based on more than 4 700 respondents, they found that a lower share of adults engaged in monthly digital platform employment8 than previous estimates, with proportion of 0.4% in Poland, 0.8% in Latvia,1.1% in Slovakia, 1.% in Bulgaria and 3% in Hungary. More regular digital platform employment (at least weekly) ranges from 0.4% in Poland and Slovakia and 0.5% in Latvia, to 0.8% in Bulgaria and 1.9% in Hungary.

The reviewed studies show the importance of choice of the survey mode and its impact on survey’s results (Box 4.1). These considerations are also applicable to surveys carried out by official organisations.

Beyond estimating the prevalence of digital platform workers, (Urzì Brancati, Pesole and Fernández-Macías, 2020[17]) also included questions aimed at better understanding their working conditions. There is a growing body of literature focusing on specific aspects of digital platform employment, such as legal work arrangements. While these studies do not provide information on the size of digital platform employment, they could allow improving questions in surveys administered for measurement purposes.

ILO (2018[9]) provides one of the first comparative studies of working conditions of micro-task workers around the world. It is based on an ILO survey covering 3 500 workers in 75 countries and working on five major globally operating micro-task platforms. This was supplemented with in-depth, follow-up interviews with a random sample of workers. The report analyses the working conditions on these micro-task platforms, including pay rates, work availability and intensity, social protection coverage and work–life balance. Drawing on surveys and interviews with about 12 000 workers and representatives of 85 businesses, ILO (2021[20]) examines working conditions, patterns of work and income, access to social protection, association and collective bargaining rights of digital platform workers operating in online web-based and location-based platforms around the world.

In Belgium, the food delivery platform Deliveroo employed workers through an intermediary company in 2016-2018 (SMart). Based on the administrative data provided by SMart, Drahokoupil and Piasna (2019[21]) analysed data on riders active from September 2016 to April 2017 and administered a survey to these riders. They analysed workers’ characteristics, patterns of work and pay, motivation for engaging in digital platform employment, as well as their perceived benefits and disadvantages of cessation of the Deliveroo-SMart contractual agreement.

Existing labour statistics, such as those produced by LFSs, have difficulties in tracking digital platform workers. Such surveys focus on a worker’s primary job and can be unreliable in their coverage of secondary jobs and self-employment, and do not capture the diversity of employment contracts (Bernhardt and Thomason, 2017[25]); (Abraham et al., 2018[26]). This causes difficulties if digital platform workers already have a stable job and use digital platform employment to complement their income. Therefore, it is necessary to develop new questions for surveys. Recently, questions have been included in LFSs in Canada, Denmark, Finland, Singapore, Switzerland and the United States (Table 4.2). Italy also included a specific module on “gig workers” in its LFS in 2021 (ISTAT, 2021[27]).

In the United States, the Bureau of Labor Statistics (BLS) reinstated in 2017 the Contingent Work Survey (CWS) – a supplement to the nation’s monthly LFS -, which had been discontinued in 2005. In 2017, the BLS introduced two new questions on “electronically-mediated work”, with a view of measuring participation in the platform economy. The interviews were conducted by telephone and used a ‘last week’ reference period. While 3.3% of respondents (out of 46 000 people interviewed) answered positively to the situations described as electronically mediated work, a number of false positive answers were detected and in the recoded data; overall, only 1% the workforce was classified as working through an online intermediary.

Finland introduced in 2017 a question in the LFS to estimate the number of people aged 15 to 74 who had earned an income through digital platforms in the previous year (Finland, 2017[28]). Results from about 43 000 respondents showed that 0.3% of adults had earned more than 25% of their income from digital platforms. The question refers to a limited number of specific digital platforms, including some non-labour digital platforms, such as AirBnb and national digital platforms for selling second-hand goods. Pilot tests before the running of the survey had shown that respondents lacked understanding of what should be considered within the scope of digital platform employment and income (Sutela, 2018[29]).

Denmark also included specific examples of digital labour platforms and digital platforms for renting accommodation in three questions on digital platforms added to the 2017 LFS. The large-scale survey involved 18 000 randomly selected Danish citizens aged 15–74 years, interviewed using a combination of web survey and phone interviews. The survey concluded that only 1% of the workforce had earned income from platform mediated work in the last 12 months (Ilsøe and Larsen, 2020[30]).

The specific module on “Internet-mediated platform work” added to the 2019 LFS in Switzerland (Swiss Federal Statistical Office (SFSO), 2020[31]) also showed the importance of addressing cognitive biases when formulating the questions. Implementation of this module showed that plausibility checks are very important; these checks were based on hours worked, income, named platforms and interviewer’s additional comments, in order to control for false positive. Results from about 11 500 respondents showed that 1.6% of the population aged 15 to 89 provide platform services in Switzerland including renting out accommodation and sale of goods (without these two, digital platform employment amounts to 0.4% of total population).

As an annual supplement to its LFS, Singapore also included questions to capture the prevalence of own account workers who engaged in digital platform employment. This referred to digital platforms that serve as intermediaries to connect buyers with workers who take up piecemeal or assignment-based work. Results showed that in 2020, 3.6% of the workforce were regular own account workers who took up work via online matching platforms, either as their main job or on the side, over a one-year reference period. With the growth of ride-hailing and item delivery apps, most of the workers who utilised such digital platforms were providing services related to the transportation of goods and passengers.

Several national statistical offices of OECD Member States have conducted pilot surveys to measure the number of consumers and workers using digital labour platforms (Table 4.2). Initial attempts focused on use of digital platforms by consumers and were included in ICT usage surveys (such as those of Eurostat). More recently, questions asking whether participants have engaged in digital platform employment have been included in Internet use surveys in Canada, the United States, and in an EU-wide survey ran in 2018 and 2019. While the available estimates are not comparable across countries, they show a variety of approaches to dealing with the issues of providing definitions to questionnaire respondents, and setting appropriate reference periods. In addition to cross country differences, there are also substantial differences with surveys done by private organisations (see above). Some of the differences in estimates of platform use are due to differences in methodologies and definitions between countries and over time.

The Canada Internet Use Survey included a detailed module on Online Work in 2018 and in 2020. The 2018 results show that, among Internet users, 8% use Internet to earn income. Among them, 14.1% earned income using online freelancing, and 6.1% through platform-based peer-to-peer services.

The US Computer and Internet Use Supplement (CIUS), which is compiled as a supplement to the CPS, includes a question on online work, asking about own services offered for sale via the Internet. Estimates referring to November 2019 show a prevalence of 7.6% among Internet users, up from 6% in November 2017.

Eurostat inserted two questions in the Community Survey on ICT Usage in Households and by Individuals in 2018 and 2019. At the European level, results were not published as considered not reliable due to the small sample size and to limited respondents’ understanding of the concept of digital platform employment. However, Slovenia and Switzerland published some results, which confirm that only a tiny share of the population obtained paid work by using an intermediary website or apps. For example, in 2019, the share was 2.1% in Switzerland (among individuals aged 15 and more) and 0.5% in Slovenia 2019 (among individuals aged 16 to 74). An accurate measurement of digital platform employment through the ICT Survey would require a small ad-hoc module with several questions, so that respondents can have an appropriate understanding. The Eurostat ICT survey currently does not have this space, as it is aimed of surveying a number of other topics. Furthermore, estimates show that a small number of digital platform workers are likely to be included in each sample, making it difficult to gain high quality statistics of digital platform workers via this type of official survey.

In Australia, (McDonald et al., 2019[32]) carried out for the Victorian government an online survey of more than 14 000 adults to enquire about the extent and nature of digital platform employment9 across the country. The survey found that 7.1% of survey respondents worked through a digital platform or had done so in the previous year.10 Based on the findings from the survey, the Victorian government released a report on the “on-demand workforce” – of which platform work is considered a subset – (The State of Victoria, 2020[33]) highlighting digital platform workers’ conditions and offering recommendations for improvement.

In France, the National Institute for Statistics INSEE (Richet Damien, 2020[34]) surveyed individual entrepreneurs who had newly registered as “micro-entrepreneurs” in 2018. The Information system on new enterprises-survey of micro-entrepreneurs (Système d’information sur les nouvelles entreprises (Sine) – enquête Micro-entrepreneurs) allows to survey at regular intervals 56 000 new micro-entrepreneurs in France, to follow the developments for a new generation of enterprises. The survey found that one in six (16%) of them worked via a digital platform, with this percentage as high as two thirds for micro-entrepreneurs in the transport sector. About one third of new micro-entrepreneurs working through a digital platform – more than half of those in the transport sector – declared having created the enterprise specifically to this end. The Information system on new enterprises-survey (Sine) still asks this question in the following surveys (2019, 2021, 2022) and extended the scope, not only aiming at “micro-entrepreneurs” but also all newly created enterprises.

In Italy, the National Institute of Public Policy Analysis Innovation (INAPP), added a module on the gig economy to its 2018 survey (Participation, Labour, Unemployment, Survey, INAPP-PLUS). The survey, covering 45 000 adults and administered by telephone, found that 0.45% of Italians (about 213 000 people) offered services through labour-mediating digital platforms in the year before the survey (Cirillo, Guarascio and Scicchitano, 2019[35]). An earlier web-based survey, based on a sample of 15 000 respondents, estimated that a higher share of the population engaged in digital platform employment11 (2.6% of the working population) (Boeri et al., 2018[22]), although it used a different reference period (the week before the survey).

Other official agencies have considered digital platform workers as a subset of the broader category of “informal workers”. In the United States, the Federal Reserve's 2019 Survey of Household Economics and Decision-making (SHED), included a section on Gig Economy, including childcare, house cleaning and ride sharing. The survey – which counted on over 12 200 responses from a representative sample of the adult population – found that overall 17% of adults engaged in some form of gig work in the previous month, although only 13% of them found customers and received payments through an app or digital platform (Board, 2020[36]). In Canada, a study based on the Bank of Canada’s Canadian Survey of Consumer Expectations (Kostyshyna and Luu, 2019[37]) estimated that 18% of respondents had carried out informal work, with about 35% of them using websites and/or mobile platforms in the course of doing this work. However, the small sample limited the representativeness of this study (Sung-Hee, Liu and Ostrovsky, 2019[38]).

When asking whether a person is a digital platform worker it is necessary that respondents have the same understanding of digital platform employment, and that the definition captures the wide variety of activities that can be done through digital platforms, while setting the boundaries with those that should not be considered within it. The United Kingdom’s ONS explicitly referred to finding work on a ‘digital platform’ in its pilot survey, but many respondents poorly understood the term. Other statistical agencies have taken the approach of providing a definition of digital platform employment, giving examples of digital platforms, or restricting their questions to a narrow range of digital platforms, such as ride-hailing (Annex 4.A). In addition, both the ordering of questions and use of probing questions can affect results (Abraham and Amaya, 2018[39]).

Both the US Bureau of Labor Statistics (in the 2017 CWS) and McDonald et al. (2019[32]) (in the survey carried out in Australia) included a detailed description of digital platform employment. While such detailed description is appropriate for an occasional survey focusing specifically on contingent workers, it is likely to be cumbersome if included in a regular survey, such as monthly or quarterly LFSs.

Although the CWS does not explicitly mention digital platforms, its question refers to finding work (performed in-person) “through companies that connect [workers] directly with customers using a website or mobile app”. Therefore, the description is robust to whether or not respondents consider themselves to be self-employed or an employee of the platform. In addition, the description states that the app or website coordinates payment for the service. The description aims to reduce the possibility that respondents, when answering this question, could include capital intensive services (such as providing accommodation) by referring to “short tasks or jobs”, although respondents may differ in their understanding of what is considered a short duration of time, and may exclude freelancing. Finally, the CWS description gives the example of providing transport, household chores or online work, but does not refer to specific digital platforms. However, many respondents poorly understood the definition, answering “yes” even if they merely made use of a computer or mobile app in their job. After recoding the data (e.g. by removing obviously incorrect responses, including hairstylists that said they worked entirely online), the estimated number of digital platform workers was reduced from 3.3% to 1% (Bureau of Labor Statistics, 2018[40]).

Far shorter questions have been included in other surveys, such as the LFS of Denmark, though it is questionable whether they convey to respondents a clear understanding of digital platform employment. The Danish survey asks whether respondents earned money by “performing work done through websites or apps” (Ilsøe and Madsen, 2017[41]). In the 2018 Eurostat ICT Usage in Households and by Individuals Survey, Eurostat referred to “intermediary” websites or apps. However, it is questionable whether all respondents would have the same understanding of the term intermediary. Although Eurostat does not say the work must be performed through the app or website, the survey explicitly excludes employment agencies. However, robustness checks (such as asking participants to name the digital platform which they work with) have shown that respondents poorly understood the question, which led Eurostat to decide not to publish the results.

Several surveys offer greater clarity by asking separate questions for digital platforms offering goods and services and for those mediating labour. The Canadian Internet Use Survey mentions six categories of digital platforms from which respondents can choose. The US Federal Reserve’s Survey of Households Economics and Decision-making (SHED) similarly offers six categories of activities. While the category “driving or ride-sharing” also mentions examples of digital platforms mediating this job, for the category “other paid personal tasks, such as deliveries” it is ambiguous whether a respondent would include services mediated by a digital platform. Likewise, a respondent may not include physically delivered services, such as handiwork, within the category “paid tasks online”. The Swiss LFS in 2019 had four filter questions for respondents to choose between renting out accommodation, providing taxi services, selling goods, or providing other services. The Danish LFS asks a separate question to those who earned money ‘performing work’ and those who rented property, while the Canadian LFS refers specifically to ride services and private accommodation services (to the exclusion of all other digital platforms). Both the United States CIUS Supplement and Statistics Finland do not distinguish between digital platforms renting accommodation and those mediating labour.

As discussed in Chapter 2, a number of different policy objectives and user needs might call for measurement of digital platform work and employment. In order to meet the range of different objectives, flexibility is needed to adjust the conceptual boundaries depending on the specific area of interest.

Most official surveys name specific examples of digital platforms to aid respondents understand what digital platforms are. The most common example of a digital platform mentioned by LFS is Uber, which is mentioned by the Canadian, Danish, Finnish and Swiss surveys. Among the surveys that do not offer an example, the French LFS combines both platforms and businesses that direct customers to the worker (“intermediary”, including digital platforms) (Insee, 2018[42]) while the US Bureau of Labor Statistics offers a detailed description.

There are also several minor differences in question wording between surveys; experience from Sweden’s State Public Reports (SOU) suggests that this can have a large effect on the estimated number of digital platform workers (SOU, 2017[12]). These include asking if the respondent offered, or provided, a service; whether the question is broad enough to include those who engage in occasional digital platform employment for secondary income; and the chosen reference period.

Almost all surveys ask whether the worker provided a service, implying the worker completed a commercial transaction. However, the US CIUS asks whether a service was offered for sale (rather than provided), without specifying whether a transaction was completed or not. Similarly, the Canadian LFS asks whether the respondent ‘offered’ a service (and not necessarily ‘provided’ it) and does not mention the earning of income, meaning the survey could include those who offered a service for charitable reasons, and did not complete a commercial transaction.

Labour force statistics have traditionally focused on a worker’s main job. However, digital platform employment offers workers the flexibility to earn additional income, without becoming the respondent’s ‘main job’. Only the French LFS excludes those who engage in digital platform employment as a secondary job (by means of a series of filter questions). In contrast the US Fed only include secondary income, while the US Bureau of Labour Statistics, the 2018 Canadian Internet Use Survey, and the 2018 Eurostat ICT Usage Survey asks the respondent to specify whether the work done was as a workers main job, or to gain additional income. Likewise, the Swiss LFS ad-hoc module asks to specify whether the service provided was as part of the main, second or an additional job.

A related problem in comparing estimates of the number of digital platform workers with other categories of employment is the reference period used. LFSs typically ask for a respondent employment status in the past reference week. However, only the Bureau of Labor Statistics (CWS) asks whether the respondent performed digital platform employment in the last week. In contrast, surveys such as the Canadian, Danish, and Finnish LFSs refer to the past 12 months. The use of a longer reference period can greatly increase the estimated number of digital platform workers. Using a longer reference period also increases the share of occasional digital platform workers among all digital platform workers. Therefore, asking whether a respondent engaged in digital platform employment in the past 12 months as filter question, and then whether they engaged in digital platform employment in the past week can ensure comparability with the LFS employment count, and capture the larger number of irregular digital platform workers. This approach is taken in the Swiss LFS ad-hoc module. However, it can also be argued that the number of hours is more relevant than the frequency someone works on a digital platform (Pesole et al., 2018[16]).

Although official surveys are likely to be the best tool to estimate the total number of digital platform workers and their characteristics, the relatively small overall number of digital platform workers means that sample sizes are too small to provide quality information and to allow analysis at a more detailed level (e.g. by socio-demographic variables). In addition, such surveys cannot provide information on past trends in digital platform employment. Alternative sources, such as administrative data or data provided by digital platforms may usefully complement the information gained from official surveys.

Administrative data can overcome the problem of small sample size, reduce the burden on data providers and the cost of data collection. However, as administrative data are not collected for statistical purposes, they may have problems of timeliness, relevance, and accuracy (Office for National Statistics (UK), 2016[43]). In addition, due to a lack of definition and to ambiguities in the regulation of digital labour platforms, they may be omitted from some datasets. For example, ride hailing apps blur the lines between street hailing of a cab and pre-booking a chauffeur, and many apps take advantage of loopholes in existing labour market regulation (Broecke, 2018[44]). The tendency of digital platforms to locate in such blurred regulatory boundaries creates obstacles to the use of administrative data. For example, in Italy digital platform workers often lack formal contractual agreements (Cirillo, Guarascio and Scicchitano, 2019[35]) and almost half of the digital platforms are not formally registered at the National Institute for Social Security (INPS, 2018[45]). In addition, the source of income may not be identifiable (if for instance is reported from self-employed activity without further breakdown), or workers may not provide information on this type of activity, if they engage in digital platform employment as a secondary job or as a hobby. The cross-border nature of digital platforms further increases challenges to capture this type of employment, as workers may not report work done for a digital platform located in another country. Lastly, as systems of administration differ across countries, comparability is limited.

Administrative data have offered insights into contingent workers (such as employees who occasionally perform secondary work to earn additional income), though only a few studies distinguish digital platform workers from the broader group of non-standard workers.

In the United States, Collins et al. (2019[46]) used micro administrative tax data from the Internal Revenue Service (IRS) to explore the role of gig work mediated by digital platforms. In particular, they looked at tax data filed by self-employed individuals working for firms or performing independent contract work intermediated by firms. They refer to these arrangements - a subset of the broader gig economy - as the "online platform economy" for labour (labour OPE). They found that the share of workers with OPE income was approximately 1% of the workforce in 2016. Consistently with other sources, the results show that digital platform employment is mainly a secondary job to provide for a complementary income. Collins et al. (2019[46]) also included data on the number of digital platform workers by State in 2016. Moe, Parrott and Rochford (2020[47]) updated the data for New York State, by relating the annual growth in the number of these workers to the growth in the average number of for-hire vehicle trips in New York City, mainly supplied by drivers working for Uber and Lyft. The study estimated that there are about 150 000 digital platform workers in New York, representing about 1.6% of the State’s workforce.

In Canada, Sung-Hee, Liu and Ostrovsky (2019[38]) introduced a definition of gig work specific to the way work arrangements are reported in the Canadian tax system and estimated the size of the gig economy using various Canadian administrative sources. They also examined the characteristics of gig workers by linking administrative data to 2016 Census of Population microdata. The study found that, from 2005 to 2016, the percentage of gig workers in Canada rose from 5.5% to 8.2%. However, their definition of gig workers is not limited to individuals working through digital platforms.

Partnerships with digital platforms have the potential to improve administrative data sources. For example, the Estonian Tax and Customs Board (ETCB) has reached an agreement with two ride-sharing platforms to share their data with the ETCB. However, drivers must first give consent to share their data, which can lead to selection bias. Denmark is developing a digital solution for declaring income arising from the sharing economy. The Mexican Tax Administration (SAT) has reached an agreement whereby drivers must be officially certified before registering with a platform (OECD, 2018[48]). In France, since 2019 digital platforms are obliged to report the annual gross income an individual earns on the platform to the tax authorities, while in Belgium platforms are obliged to both withhold taxes and report information to the tax authorities (HM Revenue and Customs, 2018[49]; European Commission, 2017[50]). As countries are developing reporting systems to obtain income data from platforms, there may be benefits to harmonise reporting systems at EU level, so to reduce the reporting burden for platforms that operate cross-jurisdictionally and increase compliance (Ogembo and Lehdonvirta, 2020[51]). An additional aspect that should be considered is that legislation may apply only to digital platforms formally registered in the country. While digital platform providing in-person services most of the times are registered in the local business register, the same doesn’t apply for those mediating fully digital services.

The use of some alternative large datasets can also provide useful insights into the characteristics of platform workers. Harris and Krueger (2015[52]) estimated the number of US platform workers to be 0.4% of total employment by using data on the number of Uber drivers, and scaling this by the total number of Google searches for a list of 26 labour platforms (relative to the number of Google searches for Uber). The same method was used to estimate that as few as 0.05% of EU employees were active platform workers at the end of 2015 (Groen and Maselli, 2016[53]).

Using data from the bank accounts of those who received payments from digital platforms, economists at JP Morgan Chase investigated the characteristics of digital platform workers using data on 39 million Chase checking accounts (Farrell and Greig, 2016[54]; Farrell, Greig and Hamoudi, 2018[55]). In line with other studies, they found that approximately 1% of workers (twice the level of early 2016) used a digital platform, earning an average of under USD 800 per month, with the earnings of those using transportation apps having fallen by half since 2013. There is also a high rate of workers entering and leaving the sector. Such high churn highlights the need for an appropriate reference period when comparing the numbers of digital platform workers with other employment sectors. Koustas (2019[56]) used a transaction-level dataset from a large financial aggregator and bill-paying application to analyse how household balance sheets evolve when starting a “gig economy job”. Based on data for about 25 000 workers from 10 popular digital platforms, the study found that entry into gig work is generally preceded by a decline in non-gig income.

The use of web-scraping can also be used to assess trends in parts of the digital platform labour market. The Online Labour Index (OLI) measures the utilisation of digital platforms mediating online labour over time across countries and occupations; although it does not give an estimate of the absolute number of digital platform workers, it does capture trends. The index is based on tracking all projects and tasks posted on a sample of platforms, using an application-programming interface (API) and web-scrapping. The index is limited to platforms through which buyers and sellers of labour or services transact fully digitally: the worker and employer are matched digitally, the payment is conducted digitally via the platform, and the result of the work is delivered digitally. The samples include the top five platforms for which it was possible to collect data over time and which accounted for at least 70% of all traffic to online labour platforms (according to Alexa’s figures) (Kässi and Lehdonvirta, 2018[57]). The current sample is limited to English-language platforms.

However, data provided by platforms can have similar problems to administrative data (as the number of registered users could be higher than the number of actual users) (Office for National Statistics (UK), 2016[43]). Additionally, methods like web scrapping raise some concerns regarding data protection and statistical/research ethics. Therefore, such data can only complement rather than replace surveys.

The rich data on earnings and hours worked by digital platforms can also serve as a resource to look at general labour market issues, beyond estimating the size of digital platform employment. This is highlighted by the study of (Cook et al., 2018[58]) who used data on over a million drivers to examine the gender wage-gap and decomposed it into its main components, such as women being less willing to work anti-social hours (perhaps due to home duties or a lack of safety in picking up passengers late at night).

To date several methods have been used to measure the number and characteristics of digital platform workers, although differences in definitions and methodologies limit their comparability. These methods serve different purposes and each of them has its own strengths and weaknesses (see Table 4.3 for a summary). The choice of method depends on the research objectives, the resources available, and the trade-offs faced by statistical agencies or researchers.

A first overarching observation is that measuring the same concept of digital platform employment across national and international surveys is key for internal and international comparisons. As shown in this review, the terminology and the definitions are not harmonised across countries.

For surveys, a key problem is how to ensure that respondents understand the meaning of digital platform employment. To gain consistent statistics over time it is necessary that respondents to questionnaires have a similar understanding of the question in each period. Although giving named examples of digital platforms to respondents is an easy way to convey the meaning of digital platform employment, this can be problematic as different digital platforms enter or exit the market. Providing a clear definition of digital platform along with examples is important to ensure that respondents understand the question. However, this should not lead to overly long introductory text, as this would increase the propensity of respondent to ignore this text (Montagnier, P.; Ek, I., 2021[59]).

The overall importance of the topic of digital platform workers to a survey affects the appropriate amount of space devoted to formulating an easily understandable question. However, rather than give a detailed definition of digital platform workers, consideration should be given to asking a series of short questions concerning different elements of digital platform employment, with the interviewer or subsequent analysis then determining whether the respondent should be considered as a digital platform worker or not. Filter questions can also be used to determine the nature of the work conducted, such as whether the service was provided online or delivered physically. This approach has the advantage of ensuring the survey is robust to changes in traditional employment, such as firms using apps to roster workers’ hours.

Next to the definition and clarification of the survey object, attention should also be devoted to the survey mode, as it can affect results by introducing coverage and measurement biases (see Box 4.1). While online surveys may be suitable to measure digital platform workers, they may not be representative of the overall population. Telephone or face-to-face surveys, however, may not be able to reach out to those digital platform workers who are not in national phone registers, or who are not available at the times that surveys are carried out. While evidence suggests that respondents are more honest when answering self-administered questionnaires, interviewer-administered surveys may yield higher quality results, as interviewers can correct inconsistencies in respondents’ answers. Cost and time are also relevant factors to consider. Face-to-face surveys tend to be more costly and take a longer time horizon to be realised than online surveys. Accordingly, if budgets are limited or results are required quickly, the online mode might be the preferred one.

Overall, it can be concluded that there is no perfect or ideal survey mode for digital platform employment surveys. All currently existing modes have specific advantages and disadvantages, and it needs to be decided on a case-by-case basis which mode is likely to result in the best outcome, that is which short-comings are acceptable against the specific information needs.

The choice of reference period will affect the type of workers captured by the survey. For researchers mainly interested in those who regularly engage in digital platform employment, asking whether someone performed such work in the reference week is appropriate. However, for those also wishing to capture occasional platform employment a longer time horizon is needed. Therefore, asking an additional question as to whether someone engaged in digital platform employment in the last 12 months may be appropriate, and would allow greater consistency with previous surveys.

When the objective is to ensure consistency with existing labour statistics, it is necessary to include questions on digital platform employment in the LFSs of national statistical offices, which ensures identical sampling frames and the same reference week (rather than a longer time horizon). This is likely, however, to give a lower quality estimate, as those who only perform this type of work occasionally are less likely to be captured.

The heterogeneity of labour services provided is a distinctive characteristic of digital platform employment, not normally found in traditional forms of labour provision. Therefore, careful consideration should also be given to the ordering and filtering of questions to ensure that it is clear about which episode of digital platform employment respondents are referring to when answering subsequent questions about the nature of the work or tasks they performed.

For researchers who are only interested in the use of a digital platform by a specific category of worker (such as the self-employed) it can be possible to use filter questions to identify the target group, and then phrase the question specific to that group (such as by asking the self-employed how they interact with customers). However, this approach comes at the cost of limiting data comparability with other surveys.

For researchers wishing to ensure cross-country comparability, the use of named digital platforms in survey questions may be problematic, as not all digital platforms may operate (or be equally known) in each country. The use of some existing big-data sources, such as used by Farrell et al. (2018[55]), can allow researchers to refine their research question as new digital platforms enter the market. Methodologies which rely on web-scraping may have problems of consistency over-time as digital platforms are added, or dropped, from the list of the ones that are monitored. These methods also raise some ethical issues. In addition, the potential use of administrative data is likely to be limited due to differences in administrative systems across countries. Therefore, the use of surveys is likely to be the best approach to gaining cross-country statistics.

Although LFSs may be the best option for those wishing to learn about the overall prevalence of digital platform employment, ICT Usage Surveys can be a better option for assessing technology usage and online behaviours. However, attempts to date have shown that this tool may not be the best vehicle to gain descriptive statistics, due to the small number of workers included in the sample. Time Use Surveys (TUSs) have the advantage of being able to capture platform work done for short period and as a secondary occupation, but to date they have not included questions to investigate this topic, and they also have the disadvantage of being conducted very unfrequently. Finally, income surveys are appropriate to examine whether individuals have earned a significant portion of their income from digital platform employment. Both types of surveys would require inclusion of additional questions in order to capture this phenomenon.

In conclusion, while the use of official surveys such as LFSs may give more accurate estimates on the overall prevalence of digital platform employment, problems of sample size reduce their suitability for gaining insights into the characteristics of digital platform workers. Even though the sample sizes of LFSs are typically very large, they will nevertheless lack statistical precision about characteristics of potentially small groups in the population such as digital platform workers. This is all the more true for ICT Usage Surveys, which have a smaller sample size than LFSs. Also, the nature of digital platform employment (task approach) is not that well compatible with the concepts underlying LFS. Therefore, other sources (such as ad hoc surveys, administrative datasets or big data) provide a useful complement. At present, the possibilities of using administrative data are limited, but these may increase as tax authorities develop data-sharing agreements with digital platforms. In addition, the use of online surveys can reduce costs (though possibly at the expense of reduced accuracy and sampling bias), allowing researchers to reach out to a larger number of respondents. Such approaches can complement official surveys, which can be used to test the overall accuracy of other approaches and to calibrate their results.

Based on this review, potential next steps should include the formulation of questions to be included in a range of official surveys (e.g. regular LFSs and ad-hoc modules within LFSs). It is also necessary to decide upon the most appropriate tool (and frequency) for addressing different facets of the phenomenon: for example, a short list of questions in core (monthly or quarterly) LFS questionnaires may be appropriate to monitor the evolution of digital platform employment over time. A longer list of questions in less frequent survey supplements (e.g. ad hoc modules in LFS, or TUSs or income survey supplements) on the other hand may be more appropriate to illustrate the variety and regularity of tasks performed by workers in digital platform employment and their characteristics and sources of income. Finally, more experimentation in terms of ordering of questions and use of prompting questions may be necessary before such questions are included in surveys. These points and additional methodological recommendations are further developed and discussed in Chapter 5.

The nature of work and its use of digital platforms are evolving rapidly. The frontiers between the various working arrangements and their legal status are blurring, and so are the workers’ perceptions of their occupations. This makes it difficult to accurately measure the evolution of digital platform employment. Although no optimal approach currently exists, this chapter suggests that a mixed approach, combining several measurement instruments (general population surveys, ad hoc surveys, administrative data, web scraping, etc.), is needed.


[30] A. StrømmenBakhtiar, &. (ed.) (2020), Digital platforms at work. Champagne or cocktail of risks?, Routledge, https://doi.org/10.4324/9780429293207.

[39] Abraham, K. and A. Amaya (2018), “Probing for Informal Work Activity”, NBER Working Paper 24880, https://www.nber.org/system/files/working_papers/w24880/w24880.pdf.

[26] Abraham, K. et al. (2018), “Measuring the Gig Economy: Current Knowledge and Open Issues”, NBER Working Paper 24950, https://www.nber.org/papers/w24950.

[10] Alsos, K. et al. (2017), “Når sjefen er en app”, Fafo-rapport, Vol. 2017/41, https://www.fafo.no/images/pub/2017/20649.pdf.

[4] Balaram, B., J. Warden and F. Wallace-Stephens (2017), Good Gigs: A fairer future for the UK’s gig economy, https://www.thersa.org/globalassets/pdfs/reports/rsa_good-gigs-fairer-gig-economy-report.pdf.

[25] Bernhardt, A. and S. Thomason (2017), What Do We Know About Gig Work in California? An Analysis of Independent Contracting, https://laborcenter.berkeley.edu/what-do-we-know-about-gig-work-in-california/.

[36] Board, F. (2020), Report on the Economic Well-Being of U.S. Households in 2019, Featuring Supplemental Data from April 2020, https://www.federalreserve.gov/publications/files/2019-report-economic-well-being-us-households-202005.pdf.

[22] Boeri et al. (2018), Social Protection for Independent Workers in the Digital Age, http://www.frdb.org/be/file/_scheda/files/01_Stephen_Machin.pdf.

[7] Bonin, H. and U. Rinne (2017), “Omnibusbefragung zur Verbesserung der Datenlage neuer Beschäftigungsformen”, IZA Research Report 80, http://ftp.iza.org/report_pdfs/iza_report_80.pdf.

[44] Broecke, S. (2018), Protecting workers from low pay in the future world of work: Are piece rate minimum wages part of the answer.

[40] Bureau of Labor Statistics (2018), “Electronically mediated work: new questions in the Contingent Worker Supplement”, Monthly Labor Review, https://www.bls.gov/opub/mlr/2018/article/electronically-mediated-work-new-questions-in-the-contingent-worker-supplement.htm (accessed on  2018).

[3] CIPD (2017), To gig or not to gig? Stories from the modern economy, https://www.cipd.co.uk/Images/to-gig-or-not-to-gig_2017-stories-from-the-modern-economy_tcm18-18955.pdf.

[35] Cirillo, V., D. Guarascio and S. Scicchitano (2019), Platform workers in Italy: an empirical exploration on worker-level data, http://oa.inapp.org/xmlui/handle/123456789/516?show=full.

[46] Collins et al. (2019), Is Gig Work Replacing Traditional Employment? Evidence from Two Decades of Tax Returns, https://www.irs.gov/pub/irs-soi/19rpgigworkreplacingtraditionalemployment.pdf.

[58] Cook, C. et al. (2018), The Gender Earnings Gap in the Gig Economy: Evidence from over a Million Rideshare Drivers, https://www.nber.org/system/files/working_papers/w24732/w24732.pdf.

[21] Drahokoupil, J. and A. Piasna (2019), Work in the Platform Economy: Deliveroo Riders in Belgium and the SMart Arrangement, https://www.etui.org/publications/working-papers/work-in-the-platform-economy-deliveroo-riders-in-belgium-and-the-smart-arrangement.

[15] Eurobarometer (2018), “The use of the collaborative economy”, Flash Eurobarometer 467.

[14] Eurobarometer (2016), “The use of Collaborative Platforms”, Flash Eurobarometer 438, http://ec.europa.eu/COMMFrontOffice/publicopinion/index.cfm/ResultDoc/download/DocumentKy/72885.

[19] Eurofound (2019), Mapping the contours of the platform economy, https://www.eurofound.europa.eu/data/platform-economy/records/mapping-the-contours-of-the-platform-economy.

[50] European Commission (2017), Europe’s Digital Progress Report (EDPR) 2017 Country Profile Belgium.

[54] Farrell, D. and F. Greig (2016), Paychecks, Paydays, and the Online Platform Economy - Big Data on Income Volatility, https://www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/institute/pdf/jpmc-institute-volatility-2-report.pdf.

[55] Farrell, D., F. Greig and A. Hamoudi (2018), The Online Platform Economy in 2018: Drivers, Workers, Sellers, and Lessors, https://www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/institute/pdf/institute-ope-2018.pdf.

[28] Finland, S. (2017), Labour force survey 2017: platform jobs, https://www.stat.fi/til/tyti/2017/14/tyti_2017_14_2018-04-17_en.pdf.

[53] Groen, W. and I. Maselli (2016), “The Impact of the Collaborative Economy on the Labour Market”, CEPS Special Report 138, https://www.ceps.eu/wp-content/uploads/2016/06/SR138CollaborativeEconomy_0.pdf.

[52] Harris, S. and A. Krueger (2015), “A Proposal for Modernizing Labor Laws for Twenty-First-Century Work: The “Independent Worker”, Hamilton Project Discussion Paper, Vol. 2015/10.

[49] HM Revenue and Customs (2018), The role of online platforms in ensuring tax compliance by their users, https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/754206/The_role_of_online_platforms_summary_of_responses.pdf.

[6] Huws, U. et al. (2019), The Platformisation of Work in Europe. Results from research in 13 European countries., Foundation for European Progressive Studies, UNI Global Union and University of Hertfordshire, https://www.feps-europe.eu/attachments/publications/the%20platformisation%20of%20work%20in%20europe%20-%20final%20corrected.pdf.

[23] Huws, U., N. Spencer and S. Joyce (2016), Crowd Work in Europe: Preliminary resultsfrom a survey in the UK, Sweden, Germany, Austria and the Netherlands, http://researchprofiles.herts.ac.uk/portal/files/10749125/crowd_work_in_europe_draft_report_last_version.pdf.

[20] ILO (2021), World Employment and Social Outlook 2021: The role of digital labour platforms in transforming the world of work, International Labour Organization, https://www.ilo.org/global/research/global-reports/weso/2021/WCMS_771749/lang--en/index.htm.

[9] ILO (2018), Digital labour platforms and the future of work: Towards decent work in the online world, International Labour Office, https://www.ilo.org/wcmsp5/groups/public/---dgreports/---dcomm/---publ/documents/publication/wcms_645337.pdf.

[41] Ilsøe, A. and L. Madsen (2017), “Digital af arbejdsmark: Danskernes erfaring med digital automatisering og digitale [Digitalization of the labour market - digital automation and digital platforms in Denmark]”, FAOS Forskningsnotat 157, https://faos.ku.dk/publikationer/forskningsnotater/fnotater-2017/Fnotat_157_-_Digitalisering_af_arbejdsmarkedet.pdf.

[45] INPS (2018), XVII Rapporto Annuale, https://www.inps.it/dati-ricerche-e-bilanci/rapporti-annuali/xvii-rapporto-annuale.

[42] Insee (2018), Emploi, chômage, revenus du travail, Insee, https://www.insee.fr/fr/statistiques/fichier/3573876/ecrt18.pdf.

[27] ISTAT (2021), Rilevazione sulle forze lavoro - Questionario 2021, ISTAT, https://www.istat.it/it/files//2021/02/Questionario_FdL_2021.pdf.

[11] Jesnes, K. et al. (2016), Aktører og arbeid I delingsøkonomien.

[57] 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/.

[60] Katz, L. and A. Krueger (2019), “The Rise and Nature of Alternative Work Arrangements in the United States, 1995-2015”, ILR Review, Vol. 72/2, pp. 382-416.

[1] Katz, L. and A. Krueger (2016), “The rise and nature of alternative work arrangements in the United States, 1995-2015”, Working paper, Vol. 226, https://scholar.harvard.edu/files/lkatz/files/katz_krueger_cws_resubmit_clean.pdf.

[37] Kostyshyna, O. and C. Luu (2019), The Size and Characteristics of Informal (“Gig”) Work in Canada, https://www.bankofcanada.ca/wp-content/uploads/2019/02/san2019-6.pdf.

[56] Koustas, D. (2019), “What do big data tell us about why people take gig economy jobs?”, https://www.aeaweb.org/articles?id=10.1257/pandp.20191041.

[8] Le Ludec, C., T. Tubaro and A. Casilli (2019), Combien de personnes microtravaillent en France ? Estimer l’ampleur d’une nouvelle forme de travail, http://diplab.eu/wp-content/uploads/2019/02/WPi3-19-SES-02-LeLudec-Tubaro-Casilli.pdf. (accessed on  November 2020).

[5] Lepanjuuri K., W. (2018), The characteristics of those in the gig economy, United Kingdom Department for Business, Energy and Industrial Strategy, London., https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/687553/The_characteristics_of_those_in_the_gig_economy.pdf (accessed on  November 2020).

[13] Manyika, J. et al. (2016), Independent work: Choice, necessity, and the gig economy, https://www.mckinsey.com/~/media/McKinsey/Featured%20Insights/Employment%20and%20Growth/Independent%20work%20Choice%20necessity%20and%20the%20gig%20economy/Independent-Work-Choice-necessity-and-the-gig-economy-Full-report.ashx.

[32] McDonald et al. (2019), Digital Platform Work in Australia: Prevalence, Nature and Impact, https://s3.ap-southeast-2.amazonaws.com/hdp.au.prod.app.vic-engage.files/7315/9254/1260/Digital_Platform_Work_in_Australia_-_Prevalence_Nature_and_Impact_-_November_2019.pdf.

[47] Moe, L., J. Parrott and J. Rochford (2020), The Magnitude of Low-Paid Gig and Independent Contract Work in New York State, https://static1.squarespace.com/static/53ee4f0be4b015b9c3690d84/t/5e424affd767af4f34c0d9a9/1581402883035/Feb112020_GigReport.pdf.

[59] Montagnier, P.; Ek, I. (2021), “AI measurement in ICT usage surveys: a review”, Digital Economy Papers No. 308, https://doi.org/10.1787/72cce754-en.

[24] OECD (2019), “Measuring platform mediated workers”, OECD Digital Economy Papers No. 282, https://doi.org/10.1787/170a14d9-en.

[48] OECD (2018), Tax Challenges Arising from Digitalisation – Interim Report 2018: Inclusive Framework on BEPS, OECD/G20 Base Erosion and Profit Shifting Project, OECD Publishing, Paris, https://doi.org/10.1787/9789264293083-en.

[43] Office for National Statistics (UK) (2016), The feasibility of measuring the sharing economy, https://www.ons.gov.uk/economy/economicoutputandproductivity/output/articles/thefeasibilityofmeasuringthesharingeconomy/2016-04-05.

[51] Ogembo, D. and V. Lehdonvirta (2020), “Taxing Earnings from the Platform Economy: An EU Digital Single Window for Income Data?”, British Tax Review, Vol. 1, pp. 82-101, https://eu2020-reader.bmas.de/en/new-work-human-centric-work/taxing-earnings-from-the-platform-economy-an-eu-digital-single-window-for-income-data%E2%80%8A1/.

[16] Pesole, A. et al. (2018), Platform Workers in Europe, Publications Office of the European Union, http://publications.jrc.ec.europa.eu/repository/bitstream/JRC112157/jrc112157_pubsy_platform_workers_in_europe_science_for_policy.pdf.

[2] Pew Research Center (2016), Gig Work, Online Selling and Home Sharing, http://www.pewinternet.org/2016/11/17/gig-work-online-selling-and-home-sharing/.

[18] Piasna, A. and J. Drahokoupil (2019), Digital labour in central and eastern Europe: evidence from the ETUI Internet and Platform Work Survey, European Trade Union Institute, https://www.etui.org/node/31491.

[34] Richet Damien, B. (2020), Micro-entrepreneurs immatriculés en 2018 : dans les transports, deux sur trois travaillent via une plateforme numérique, INSEE, https://www.insee.fr/fr/statistiques/fichier/version-html/4799082/IP1821.pdf.

[12] SOU (2017), Ett arbetsliv i förändring – hur påverkas ansvaret för arbetsmiljön?, https://www.regeringen.se/496173/contentassets/93df7ab18b704a8ab655080cb498dfd1/ett-arbetsliv-i-forandring--hur-paverkas-ansvaret-for-arbetsmiljon-sou-201724.

[38] Sung-Hee, J., H. Liu and Y. Ostrovsky (2019), Measuring the Gig Economy in Canada Using Administrative Data, Statistics Canada, https://www150.statcan.gc.ca/n1/pub/11f0019m/11f0019m2019025-eng.htm.

[29] Sutela, A. (2018), Platform jobs are here to stay – how to measure them?, http://www.stat.fi/tietotrendit/blogit/2018/platform-jobs-are-here-to-stay-how-to-measure-them/ (accessed on 11 December 2020).

[31] Swiss Federal Statistical Office (SFSO) (2020), Internet-mediated platform work is not very common in Switzerland, https://www.bfs.admin.ch/bfs/en/home/statistics/work-income/employment-working-hours/employed-persons/working-conditions/internet-platform-s.

[33] The State of Victoria (2020), Report of the Inquiry into the Victorian On-Demand Workforce, https://s3.ap-southeast-2.amazonaws.com/hdp.au.prod.app.vic-engage.files/4915/9469/1146/Report_of_the_Inquiry_into_the_Victorian_On-Demand_Workforce-reduced_size.pdf.

[17] Urzì Brancati, C., A. Pesole and E. Fernández-Macías (2020), New evidence on platform workers in Europe. Results from the second COLLEEM survey, Publications Office of the European Union, https://doi.org/10.2760/459278.

[61] Wikipedia (2020), Mark and recapture.


← 1. This chapter is mainly based on OECD (2019[24]).

← 2. The chapter includes studies whose aim is to estimate the size of digital platform employment drawing on quantitative methods, published by October 2020 in English (with the exception of a few studies in national languages). Although the chapter aimed at including as many available studies as possible, the evidence considered has to be intended as illustrative, rather than exhaustive.

← 3. This result was confirmed in (Katz and Krueger, 2019[60]), after the authors re-examined their results based on data from the CWS survey carried out in 2017, the RAND CWS 2015 survey and administrative tax data from the Internal Revenue Service (IRS) for 2000 to 2016. In line with (Farrell, Greig and Hamoudi, 2018[55]), they estimate that “only 0.5 percent to 1.5 percent of the workforce was engaged in online work for sample periods covering late 2015 to the end of 2017”.

← 4. Capture-recapture is a method commonly used in ecology to estimate an animal population's size where it is impractical to count every individual. A portion of the population is captured, marked, and released. Later, another portion will be captured and the number of marked individuals within the sample is counted. Since the number of marked individuals within the second sample should be proportional to the number of marked individuals in the whole population, an estimate of the total population size can be obtained by dividing the number of marked individuals by the proportion of marked individuals in the second sample (Wikipedia, 2020[61]).

← 5. Austria, Czech Republic, Estonia, Finland, France, Germany, Italy, the Netherlands, Slovenia, Spain, Sweden, Switzerland and the United Kingdom.

← 6. The French estimate fell to 11% in 2018, suggesting that understanding of the question by respondents changed over time.

← 7. Croatia, Czech Republic, Finland, France, Germany, Hungary, Ireland, Italy, Lithuania, the Netherlands, Portugal, Romania, Slovakia, Spain, Sweden and the United Kingdom.

← 8. “Platform work” in the original study.

← 9. “Platform mediated work” in the study.

← 10. The sample was constructed to be nationally representative according to gender, age and State/Territory and was administered by the Online Research Unit (ORU), an Australian-based online research panel provider.

← 11. “Gig-economy work” in the study.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD/ILO/European Union 2023

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at https://www.oecd.org/termsandconditions.