Labor control in the gig economy: Evidence from Uber in China

Abstract

The rapid development of digital platform businesses has facilitated the expansion of gig work in China and elsewhere in recent years. Now that IT-powered platforms have been used in part to free the capital from taking employer responsibilities, the capital’s toolkit for labor control has been significantly limited. Drawing on qualitative field research supplemented by quantitative data on Uber in China, this article provides a novel empirical account of the labor control of digital platforms, and more importantly, their effects on different types of workers. The authors have identified three crucial strategies that Uber has devised to control its drivers’ labor process: an incentive pay system, a customer evaluation system, and flexible work arrangements. These strategies will, however, demonstrate significant effects on drivers’ working hours and income only when we consider the different motivations of Uber drivers. Specifically, the working efforts of those who drive for Uber as their only source of income are responsive to incentive pay schemes and a platform’s evaluation system, but are not as responsive to work flexibility. The exact opposite is the case for drivers who have other jobs and sources of income.

Keywords

China digital platform gig work labor control labor process

Introduction

Numerous digital platforms have emerged in recent years offering a wealth of different services. In an idealistic sense, economists believe such platforms can realize an instant match between supply and demand in a given market, minimizing the transaction costs through these IT-powered platforms (Schmidt, 2017). These industries have therefore attracted a wealth of investment from global capital that speculates on their profitability in the long run. In the labor market, these platforms have accentuated the emergence of what is being called a ‘gig economy’, having attracted a large number of workers to engage in various forms of flexible work. The number of online platform-based workers has increased rapidly. In the United States alone, 0.5% of all workers in 2015 were found to have engaged in some form of platform-based work (Katz and Krueger, 2016). A similar gig economy has emerged in China. Specifically, China’s State Information Center (2019) estimated that 75 million people in China had offered work through digital platforms in 2018.

In granting workers work flexibility, platform companies do not give direct commands to arrange the working hours and schedules of their workers. And a worker is not committed to any specific platform, but can switch platforms to take other work tasks by simply opening another app. In light of this, we ask how any single platform can secure quality service and a sufficient and stable labor supply without such traditional labor control tools. In other words, if the current trend of ‘Uberization’ of work has fundamentally restructured labor–capital relations in many traditional sectors, what alternative and innovative strategies have platforms developed in order to control their workers? Unfortunately, while there seems to be consensus amongst scholars that alternative control tactics have indeed been developed, the toolkit for platform companies regarding labor control remains something of a mystery. Some legal scholars have examined how digital platforms organize their labor process and control their workers, and have noted strategies that include the use of ‘big data’, mobile phone apps as well as customer evaluation (Aloisi, 2016; De Stefano, 2016; Veen et al., 2019). Less is known, however, regarding how the workers themselves react to different platforms’ strategies. It is one thing to devise innovative strategies that seek to control labor, but it is another to actually win workers’ consent through the application of such tactics.

This article provides a novel empirical account of the labor control strategies of digital platforms and their effects, using Uber in China as an example. In particular, it focuses on an important taxonomy of Uber drivers – namely whether Uber driving constitutes their sole source of income – and examines the effects that Uber’s labor control strategies may have on these different types of workers. Since the existing literature on the subject does not offer much of a theoretical basis to derive specific hypotheses for testing, a qualitative inductive approach is utilized for this study to conduct exploratory research and identify the important strategies that Uber uses to control its drivers. The authors have also constructed a major dataset with the help of Uber. Some of the data from the dataset are used in this article to show the effects of Uber’s labor control strategies over drivers’ working hours and income.

Relevant literature

The theme of labor control has been gaining limited but growing scholarly attention in recent research on digital platform-based economy. This pertains to two theoretical perspectives. The first is the legal perspective. Labor and employment law scholars recommend examining managerial tactics in order to determine the existence of employment relationships within gig work. They have focused their efforts on the specific control tactics of digital platforms but have often overlooked the effects of these strategies on workers. The second is the labor process perspective. Derived from Marxist thinking, labor process theory provides great insights into understanding class subsumption and conflict between labor and capital in modern workplaces. But scholars have just started to apply the labor process theory to the new platform-based economy and have not generated much empirical work.

Legal perspective

The labor control of capital has become one of the key features that define employment relationships in the modern legal regimes for labor and employment. ‘Employees agree to be economically dependent on their employers by relinquishing control over many aspects of their work lives (and, to some extent, their economic futures) and, in return, employers must provide workers with a degree of economic security’ (Harris and Krueger, 2015: 7). In other words, labor control and employer responsibility are two sides of the same coin. In the United States, for instance, various laws have factored the theme of control – that is, whether the employer maintains control over pay, time, work manners, means of production, and other work-related issues – in their determination tests for employee status. Through such tests, workers are divided into two categories, namely, independent contractors and employees. The user firms of independent contractors are exempted from many legal employer obligations, including minimum wages, social insurance contribution, and refraining from employment discrimination (Harris and Krueger, 2015). In China, where this study was conducted, the courts draw on the Labor Ministry’s 2005 ‘Notice on Various Issues Regarding the Determination of Employment Relationships’. It constitutes one of the three defining features of employment relationships in this regulation if the employer disciplines, manages, compensates, and schedules work for a worker.

In its latest developments, capital has begun proactively restructuring the labor process by utilizing digital platforms in traditional industries in order to avoid assuming employer responsibilities and to achieve cost-effectiveness on a larger scale (Friedman, 2014). However, this process of informalization has greatly limited capital’s toolkit for labor control. In particular, through such new systems, continuous employment relationships have been fragmented into small task assignments. Work tasks are assigned and scheduled via virtual networks and smartphone apps. In this process, a feeling of ‘bosslessness’ has emerged amongst workers, as this process provides these workers with the illusion that they are instructed rather by IT algorithms (Steinmetz, 2015). Not only are the actual employers hidden, but in many ways the workers themselves are hidden as well. In the rhetoric of the gig economy, terms such as ‘share’, ‘task’, ‘help’, and ‘service’ are now used in place of words like ‘work’, ‘job’, and ‘employee’. Workers have thus become invisible, and are seen by their customers as a simple extension of IT equipment, virtual platforms, and cellphone apps (De Stefano, 2016).

In light of this, scholars have engaged in debate regarding whether it should be lawful to regulate platform-based work within the traditional employment relations (vs independent contracting) framework (Aloisi, 2016; Cunningham-Parmeter, 2016; Finkin, 2016; Scott and Brown, 2017) or within innovative paradigms that reach beyond these schemes (Cherry and Aloisi, 2017; Harris and Krueger, 2015; Stewart and Stanford, 2017). But a consensus has been reached among these scholars that while employers and managers have become invisible in a sense as a result of these recent developments, capital’s control over the labor process has been at best obscured, rather than relinquished, and surplus value has become similarly made invisible. In many ways, argues Aloisi (2016), platform companies have demonstrated features as employers rather than just media or databases.

In sum, the legal perspective has generated a series of important studies that seek to decipher the business models of platforms, and examined how platforms like Uber and Amazon Mechanical Turk organize and control the labor processes through various tactics. But more knowledge is needed regarding labor subjectivity, namely how workers react to these strategies.

Labor process perspective

Labor process theory has a Marxist tradition of looking at how capital controls labor in its organization of work and production. This originates from the classic Marxian premise that only labor produces value such that capital has to control the labor process in order to extract surplus value – typically understood as profit (Marx, 1867 [1990]). Early labor process scholars also somewhat overlooked labor subjectivity. Their scholarship focused on economic and disciplinary tools that management had devised for the purpose of labor control. These include the Taylorist scientific management (Braverman, 1974), organizational bureaucracies and technologies (Edwards, 1979), subcontracting (Clawson, 1980; Littler, 1982), and practices of granting workers limited but responsible autonomy (Friedman, 1977).

This early focus on control tactics has also been criticized for being inattentive to labor subjectivity. In other words, workers should be taken as proactive participants in the labor process rather than just subject matters (Burawoy, 1985; Thompson, 1990). Two relatively recent developments in industrial sociology are relevant here. The first looked at workers’ reactions to the control of labor, in particular resistance (or nonresistance) at work. Burawoy’s (1985) politics of production theory provided an evolutionary account for that. He argued that in the capitalist world the control of labor has historically transformed in format. It started with despotism, a system that draws on coercive and disciplinary measures as well as workers’ dependence on individual employers. Thereafter, management gradually adopted hegemonic control that seeks to win workers’ consent. Specifically, as the state institutionalized labor-protective policies as well as social security systems, workers were no longer dependent on their individual employers. Capital has to therefore develop new strategies for labor control at the workplace in order to prompt workers to voluntarily subscribe to capitalist accumulation (Burawoy, 1979).

This hegemonic transformation is reflected in modern workplaces through the increasing use of self-managing teams, in place of the traditional Weberian bureaucratic organization of work (Barker, 1993), the adoption of flexible work arrangements in place of strict controls on the time and place of work (Rau and Hyland, 2002), as well as the development of various media for employees to make their voice heard in both union and nonunion settings (Batt et al., 2002).

As capital accumulation became increasingly intensified in the service sector, the notion of emotional labor emerged to denote a social process where capital has imposed on the labor process as new forms of managerial control (Hochschild, 1983). This scholarship factors customer–worker relations in their analysis of the labor process in various often feminized service settings, including call centers (Mirchandani, 2012), hotels (Otis, 2012), and home-based care work (England, 2005). In these workplaces, workers are required to effectively manage their own feelings and manipulate displays beyond natural conditions in order to achieve desirable outcomes, in particular customer satisfaction. Emotional labor has become a popular instrument for studying labor control in service work since the notion was coined (Steinberg and Figart, 1999).

In general, labor process theory has provided important analytical constructs for studying labor process in the capitalist world, but empirical works have by and large focused on waged workers within formal organizations, but overlooked the current gig economy (Gandini, 2018)¹ which often has nonstandard forms of work organization, obscured employment relationships, and flexible arrangements of time and space (Cappeli and Keller, 2013). Also, labor process scholars tend to focus on the point of conflict, and spend great effort in explaining workers’ resistance (or nonresistance, or rather, consent). In the gig economy with more flexible work arrangements and less commitment between platforms and workers, we argue that workers’ consent to a digital economy is embodied in the time they commit themselves to the platform. In other words, workers with grievances most often would just offer fewer hours or simply switch to another platform instead of going on strike. In this study, the authors therefore use working hours and income as opposed to labor conflict to measure the effects of platforms’ control tactics.

That being said, we found that labor process theory maintains great power in explaining labor control in the digital economy. The notions including disciplinary control, consent-giving, and emotional labor are relevant for theorizing our own findings about Uber, to which we now turn.

Methods

For this study, the Chinese branch of Uber Technologies Inc. was investigated as a significant case for understanding digital platform-based work. After its launch in China in February 2014, Uber rapidly expanded to many major Chinese cities. By the end of April 2016, 48.9 million users and over 5 million drivers in China had registered on the platform. Uber is the inventor of the current global business model for digital platform-based ride-hailing services, and the term ‘Uber’ has even become synonymous with ‘platform economy’. Didi Chuxing Technology Co., formally Uber’s biggest competitor in the Chinese market, purchased Uber’s Chinese business in August 2016. Before the acquisition, Didi and many other market players (e.g. Yidao and Kuaidi) engaged in very similar business models. Therefore, Uber can be regarded as a prototypical case for understanding the labor processes that exist in the platform-based ride-hailing industry, and can also be regarded as an important case for better understanding work relations in a gig economy as a whole.

For this study, exploratory field research was conducted from January 2015 to April 2016. The qualitative data that were collected can be divided into three parts. First are the interviews. A total of 120 interviews were conducted with Uber drivers as well as Uber managers. Key informants hailed from the Beijing headquarters of Uber’s branch in China. We interviewed them first in order to gain systematic understanding of Uber’s business strategies and tactics for driver control. These interviewees then introduced their subsidiaries in Beijing, Chengdu, Guangzhou, and Hefei, where further interviews were conducted with both drivers and local managers. In the drivers’ interviews we asked about their work experience with Uber and possibly competing platforms, and how they reacted to the platforms’ control tactics; and in the local managers’ interviews we focused on the interaction between drivers and the platform as these interviewees functioned to bridge the platform and drivers in the real (offline) world.

The second part of our qualitative data was collected through observations. In addition to the nonparticipant observations made as passengers on a regular basis in all these cities during the fieldwork, one researcher also engaged directly in participant observation, working as a driver in Beijing under Uber from January to April 2016. The observation was geared towards gaining firsthand experience of Uber driving, as well as direct information regarding managerial practices of Uber in China. It therefore constituted a major source of information for qualitative research. Finally, note that throughout 2015 and 2016, supplementary documents available online, as well as internal documents provided by Uber, were also collected. These materials provided more precise information on the platform’s weekly market-subsidy policies, internal rules for managing drivers, and various marketing strategies, which served as an important source of triangulation for the data obtained through the interviews and observations. All the qualitative data collected were coded and analyzed through standard procedures of an inductive exploratory research with the help of Atlas.ti.

To show the effects of Uber’s labor control strategies, the authors also drew on a major dataset the authors have established. A major survey of Uber drivers was conducted first. With Uber’s assistance, questionnaires were distributed to drivers in nine major Chinese cities. These questionnaires were sent out via cellphone messages in May 2016. A total of 15,484 responses were received. In this study’s analyses, 3061 cases were excluded due to missing values in key variables. In the end, a sample of 12,423 valid responses was acquired. The survey data were then matched with the internal data provided by Uber regarding working hours and weekly income, among others. The survey data and the internal data together constitute a novel quantitative dataset on the labor and employment of gig economy workers in China. Given the space constraint, only a limited portion of this dataset is utilized in the present study to supplement the qualitative findings.

In particular, this study uses working hours (and income, which is highly contingent on hours) to evaluate the effects of labor control. The reasons are as follows. According to Marx, capital seeks to maximize surplus value in two ways. One is to extend working time or increase work intensity, which generates what Marx called absolute surplus value. The other is to shorten socially necessary labor time and extend surplus labor time (i.e. increasing productivity) through technological advances, which increases relative surplus value in turn. In the contemporary gig economy, IT-powered platforms enable capital to reduce transaction costs and market segmentation through efficient matching between service providers and consumers. Such technological advances have significantly increased the relative surplus value generated by labor. While these strategies have been applied to all workers at the platform level, they do not exhibit variation across workers. Therefore, there is a focus on the generation of absolute surplus value for this study, and drivers’ average weekly working hours are used to measure the effects of Uber’s labor control. Another reason to examine the length of working time as a proxy for the effectiveness of labor control is the simple fact that Uber always benefits when more drivers are on the road. It is clearly Uber’s goal to encourage longer working times for its drivers, which means faster customer response times at very low cost to the platform.

It is worth noting that Uber’s willingness to facilitate this study’s research was the result of a collaborative project between the authors and the Chinese branch of Uber for evaluating Uber’s impact on the Chinese labor market. It was mutually agreed that the authors would be permitted to use the data from the project to carry out wholly independent research, without any censorship or interference on the part of Uber. The collaborators at Uber in China respected the data collection processes, which were by and large developed and conducted by the authors.

Who are driving for Uber?

Before delving into Uber’s labor control strategies, we answer the question, who are driving for Uber? As noted, Uber drivers have been reduced in many existing studies to one undifferentiated group of subjects who are coopted in the new business model. An important contribution of our study is to uncover the features of this group of people and, more importantly, their different motivations for Uber driving.

Specifically, the authors found that the work motivations of those who drive for Uber as their only source of income (called sole-source drivers hereafter) significantly differ from the motivations of those who have other sources of income (called multiple-job drivers hereafter). Table 1 reports the various demographic features of the Uber drivers included in the study’s sample pool. About half of both sole-source and multiple-job drivers are less than 35 years of age, have received a middle school education at most, and have one child. Of the sole-source drivers, 64% do not have a local household registration (namely hukou) certificate (meaning migrant workers) and 59% of them work in first-tier cities (namely Beijing, Shanghai, Guangzhou, and Shenzhen), whereas only slightly more than half of the multiple-job drivers are in the same situations. Sole-source drivers have longer periods of driving experience (9.86 years, on average) as well as longer periods of service under Uber (137.99 days, on average) than their multiple-job driver counterparts (9.72 years and 123.13 days, respectively). Finally, t-test results show that all of these demographic features, with the exception of length of driving experience, are significantly different between sole-source and multiple-job drivers.

Table 1.

Demographic features of sole-source and multiple-job Uber drivers.

Variable	Category	Sole-source drivers		Multiple-job drivers		P-value (t-test)
Variable	Category	N	Percentage	N	Percentage	P-value (t-test)
Age	<30	1763	26.97	1987	22.22	<0.001
	30–34	1497	22.90	2156	24.11
	35–39	1164	17.80	1976	22.10
	40–45	1010	15.45	1513	16.92
	>45	1104	16.89	1310	14.65
Education	Primary school	1580	24.17	1059	11.84	<0.001
	Middle school	3603	55.11	3842	42.97
	High school	1017	15.56	2475	27.68
	Bachelor’s degree and above	338	5.17	1566	17.51
Number of children	0	1451	22.19	1938	21.67	<0.001
	1	2988	45.70	4796	53.63
	2	1834	28.05	1985	22.20
	≥3	265	4.05	223	2.49
Hukou (household registration)	Without local hukou	4189	64.07	4353	48.68	<0.001
Hukou (household registration)	With local hukou	2349	35.93	4589	51.32	<0.001
Length of driving (year)		6528	9.86 (mean)	8919	9.72 (mean)	0.163
Length of service (day)		5470	137.99 (mean)	7281	123.13 (mean)	<0.001
City	Non-first-tier cities	2679	40.98	4566	51.06	<0.001
	First-tier cities	3859	59.02	4376	48.94	<0.001

The two types of drivers have very different motivations for Uber driving. Through interviews, it was found that multiple-job Uber drivers are motivated to work on the platform for various reasons. Some of them see Uber driving as a way of socializing or killing time, some seek to ‘subsidize the expenditures at the gas pump’, while others are simply curious. These drivers’ work motivation demonstrates a regional variation within China. Most multiple-job drivers interviewed in Beijing were white-collar workers who worked the day shift for various businesses and government institutions. Their formal jobs were well paid. They would drive for Uber just before and after their formal work, typically working to subsidize their family income.

In Guangzhou, where small businesses and entrepreneurships are better developed, the authors came across more salesmen and entrepreneurs. They had to drive around on a regular basis to visit their clients, and would take some Uber passengers along the way (if possible) as a convenient way to supplement their formal, but likely unstable, income. In addition, it has been reported in the news that many drivers will offer Uber rides with their luxury cars, just to make friends or simply as part of their lavish lifestyle.² Platform companies are likely manipulating media to propagate such lifestyles behind the scenes. This last example is certainly not common, but it shows how the concept of the sharing economy has penetrated the nation’s upper classes, further encouraging the participation of multiple-job drivers from the middle class who tend to admire and imitate the lifestyles of the elite. In any case, these multiple-job drivers own reliable cars that are useful for more than just working for Uber. These drivers tend to be better educated, and know the city’s roads better than sole-source drivers, as over half of them hold local hukou as noted earlier.

In contrast, sole-source drivers have very different motivations. Many of these drivers formerly drove unlicensed cabs, referred to as ‘black cabs’ by licensed cab drivers and customers. Now on Uber, they say that they ‘no longer have to hide’. Indeed, the question of whether drivers working under Uber are simply black cab drivers had remained legally unresolved until the Chinese central government finally legitimized the platform-based ride-hailing industry in its July 2016 regulatory legislation. In order to initially stabilize their workforces, both Uber and Didi agreed to refund any fines issued for drivers on their platform if they were caught by any government agency seeking black cabs.

Another type of sole-source drivers is that of migrant workers who have moved to China’s big cities from other provinces. These people will self-identify as workers. Although they do not necessarily consider themselves as working for Uber, they will often refer to their profession as one of being an Uber driver. With the emergence of the platform-based ride-hailing industry, many of these workers quit their previous jobs, buying a car no better than required by the lowest standards of the platform, registering an Uber account, and coming to big cities like Beijing and Guangzhou to start a new life as a chauffeur. They earned considerable sums of income in the earlier stages of Uber driving, when the platforms were in intense competition and thus heavily subsidized. The authors found that these differently motivated drivers react very differently to the platform’s control strategies, to which we now turn.

The labor control strategies of Uber

Through exploratory field research, we have identified three strategies that Uber used to attract, motivate, and retain drivers, namely the incentive pay system, the customer evaluation system, and the flexible work arrangements. These tactics respectively pertain to the themes of economic control, emotional labor, and consent-giving derived from labor process theory. This section examines these strategies in detail.

Economic control: The incentive pay system

Uber’s first tool for labor control was its incentive pay system. This system had two major components. The first was the ride fare, from which Uber used to deduct a 20% commission in 2016. And Uber used part of that revenue to subsidize a very complex bonus system – the second component – which was geared towards both motivating and managing its drivers. Uber had developed a fare system that covered different types of vehicles. By 15 September 2016, the fare for the lowest class of vehicles (which comprises the majority of all vehicles operating under Uber) increased from 1.50 to 1.80 Chinese renminbi yuan per kilometer (RMB yuan/km) and from 0.25 to 0.35 yuan/minute. These numbers were still significantly lower than the respective 2.30 yuan/km and 0.46 to 0.92 yuan/minute fares paid for traditional cabs in Beijing.

What made Uber driving a rewarding line of work was the various bonuses the platform offers its drivers. Management in contemporary business organizations often use incentive pay to encourage high performance (Gerhart et al., 2009). The most straightforward example of this is piece rates, which function as a useful tool for labor control in manufacturing as well as many other industries. Piece rates on the one hand encourage high productivity (Lazear, 2000), and on the other hand reduce labor–capital conflict by gaining worker consent (Burawoy, 1979). Cabs essentially function as a piece-rated service. However, in this study it was found that Uber had devised a much more sophisticated incentive pay system. Specifically, this system was designated to encourage not only a larger labor supply, but also work schedules that were structured according to market fluctuations.

It is important for a platform to actively manipulate drivers’ online time in order to accommodate the relatively high ride demands that accrue during peak hours. In Beijing during early 2016 for instance, Uber offered drivers a peak-hour bonus. Moreover, when a driver within a given peak period failed to make a certain amount of money, an amount that Uber had earlier guaranteed, the company paid out the difference to ensure this guarantee. This bonus was offered to a driver on four conditions. The first was that the driver worked online for at least 45 minutes and completed at least one ride in every single hour during the specified peak period. The second was that the driver scored a customer rating above 4.7 out of 5. The third was that the driver achieved a ride-completion rate higher than 45%. The fourth and final condition was that the driver had completed at least 10 trips throughout the preceding week. As a result, although drivers had autonomy in determining their working time in principle, they often found themselves involved in some routine work schedule that was expected by the platform in order to ensure that bonuses could be made now and in the future.

Furthermore, it was found that most workers were enthusiastic about winning these bonuses. The income received from subsidies and bonuses could possibly account for up to 90% of an Uber driver’s total monthly income, according to an estimate from Uber’s management. Note, however, that this instance represented an exceptional case in 2016, when Uber was involved in the most intensive stage of its initial (and ultimately unsuccessful) competition with its Chinese rival, Didi. On average, however, such bonuses still accounted for 30–50% of an Uber driver’s total income from the platform. Indeed, both Uber and Didi intentionally set fares that were considerably lower than those for traditional cabs, in order to strengthen the power they held over drivers in their control through bonuses. In addition, this strategy also functioned to undercut business that used to go to taxis.

Yet another type of bonus was aimed at encouraging better service. This was linked to the platform’s customer-based driver-evaluation system. If a driver achieved a ride completion (accepting the request and completing the ride) rate higher than 60%, scored a total average customer rating above 4.8, and completed more than 80 trips in a week, the platform would issue a bonus of 80% of the total fares that the driver earned in that week, up to 2000 yuan. Moreover, the top 100 drivers in a given city, in terms of weekly rides, would receive an additional 400 yuan each.

In addition, various other bonuses were presented by Uber on occasion. For example, a newly registered driver would typically during the first week of work get a bonus of 200, 500, or 800 yuan if the new driver completed 5, 10, or 15 trips, respectively. Uber sometimes also subsidized nonpeak hours in order to reduce the occasionally stark disparity between driver income during peak hours and the income during nonpeak hours.

This wide portfolio of bonuses – which was somewhat unique to China during late 2015 and early 2016 – served Uber’s goals in labor process control. As a result, drivers ended up involved in something of a game, whereby they often found themselves somewhat unconsciously taking more rides. Such bonuses and subsidies constituted a major proportion of an Uber driver’s daily income, and in order to win these various bonuses, a driver would voluntarily extend online time and increase the number of rides offered beyond their own personal preferences otherwise. More importantly, drivers would maneuver their work schedules to accommodate the rocketing demands of peak hours. While a traditional cab driver’s primary task at the beginning of a work day is earning back the daily rent paid for the cab and its service license, it was found in this study that ‘completing today’s tasks for bonuses’ very often explained an Uber driver’s primary work goal. One of the authors once took an Uber ride around 12:00 a.m. Before arriving at the destination, the researcher was asked by the driver to terminate the ride in advance. The driver promised that the researcher would still be sent to the destination, but that the driver had to procedurally finish the ride on the app, in order to accumulate one more ride before the next day. This was because the driver was only one more ride away from achieving the next level of bonuses. To be exact, the driver told the researcher:

The platform has a very tricky bonus system. You almost have to work for over 15 hours in order to receive a decent bonus. From morning to night, you have to accumulate rides to reach Uber’s standards, and it will be recalculated after 12:00 a.m. on the last day of the week. You have to reach that standard in order to receive the weekly bonus, with no exceptions for any single week … Our fares are much lower than those for taxis, and we bought our own cars. If I do not get these bonuses, it is not worth it. (Interview, April 2016)

It is worth noting here that the rules for receiving bonuses, as implemented by the platform, changed constantly over time as an intentional business strategy for tackling market fluctuations and competition. Thus, drivers also had to adjust their work schedules to accommodate these continuously changing rules. No matter how these systems for receiving bonuses changed, however, Uber made the rules transparent to all drivers. Although some drivers were at times ‘confused about how the payment is calculated’, the transparency of these rules gave drivers a feeling of ‘fairness’. It is also important to note that Uber served as the administrator of the platform’s system for pay as well as its pay standards, whereas neither the drivers nor legislation had a direct say in that regard. Once workers were involved in the ‘game’, however, they stopped questioning the rules of the game, which invariably won the platform labor consent anyway.

In conclusion, the platform’s incentive pay system facilitated drivers’ consent to Uber’s demanding work relations. The results of descriptive analysis of the quantitative data demonstrate the effects of the platform’s control strategies. Table 2 shows workers’ weekly working hours on Uber over a dichotomy of their appreciation of the labor control tactics, while Table 3 shows workers’ weekly income they made on Uber over the dichotomy. As shown, those who appreciated the incentive pay system on average spent significantly more time (28.27 hours/week) and also made significantly more money (636.97 yuan/week) on Uber than those who did not appreciate the system (26.38 hours and 588.46 yuan).

Table 2.

Drivers’ weekly working hours and their appreciation of labor control strategies.

	Category	All drivers	Sole-source drivers	Multiple-job drivers
Do you think the platform’s incentive pay system is good?	Yes	28.27	38.35	20.52
	No	26.38	36.21	19.70
	P-value	<0.001	<0.001	0.015
Does the evaluation system incentivize you to serve passengers better?	Yes	28.22	38.23	20.36
	No	26.00	35.76	19.71
	P-value	<0.001	<0.001	0.042
Do you think working through the platform is more flexible?	Yes	28.74	37.14	21.03
	No	25.99	37.22	19.41
	P-value	<0.001	0.553	<0.001

Table 3.

Drivers’ weekly income (yuan) and their appreciation of labor control strategies.

	Category	All drivers	Sole-source drivers	Multiple-job drivers
Do you think the platform’s incentive pay system is good?	Yes	636.97	884.17	443.41
	No	588.46	842.03	412.71
	P-value	<0.001	0.026	0.006
Does the evaluation system incentivize you to serve passengers better?	Yes	638.22	894.72	432.99
	No	575.88	816.72	417.79
	P-value	<0.001	<0.001	0.102
Do you think working through the platform is more flexible?	Yes	674.44	896.91	467.10
	No	558.57	824.90	398.72
	P-value	<0.001	<0.001	<0.001

It is worth noting that after Didi’s acquisition of Uber’s Chinese business in 2016, the new Didi had moved to gradually reduce its bonuses for drivers until 90% of the bonuses were cut off.³ As Didi had become a market monopoly, market expansion had been deprioritized as a result, and the corporate focus had shifted more toward efforts to expand the company’s profit. As a result, incentive pay had become less relevant as a labor control strategy.

Finally, this economic control strategy had divergent effects on sole-source and multiple-job drivers. As noted, the sole-source drivers invested tremendously in the business of Uber. They invested their time, gave up their past careers, and even bought their means of production, the cars themselves. Because of this, the sole-source drivers were much more concerned than their multiple-job counterparts as to whether the bonus policies for the upcoming weeks would be fair or favorable for them, and whether the platform’s evaluation system was sound. These drivers often compared Uber with Didi when making such considerations. Most sole-source Uber drivers interviewed easily recalled Didi’s bonus policies on the spot, and all of them noted that they were ready to switch over to Didi if feelings of unfairness were to be fomented.

In contrast, the considerable income that multiple-job drivers earned from their formal jobs and the relatively low income they acquired from Uber made them less concerned over the platform’s incentive pay and the customer evaluation system. To multiple-job drivers these systems might be well-designed, but most of them would not end up receiving many bonuses, as they most often required great commitments throughout the days and weeks. As shown in Table 2, the incentive pay system’s association with working hours held for sole-source drivers. The t-test for the multiple-job driver group passed the significance test at the 0.05 level but not at the 0.001 level.

Emotional labor: The customer evaluation system

The second control strategy used by Uber was its customer evaluation system. Customer satisfaction is key to the success of a service industry, and because of this, customer satisfaction is often integrated into the managerial practices of many organizations. For Uber specifically, customers are granted the right to monitor workers’ performance (Rosenbalt and Stark, 2016). This system allows the platform to achieve de facto hierarchical control over these outsourced workers without instituting actual organizational hierarchies, in the sense that the workers are distributed into different levels, and can be disciplined in various ways based on their customer satisfaction feedback (Aloisi, 2016). At the end of each trip, the platform encourages passengers to rate their driver based on a five-star scale. The workers who achieve higher levels will receive better benefits as well as other work advantages. These ratings become important ‘capitals’ of workers in the gig economy. And the fact that they cannot be transferred across platforms has exacerbated a worker’s dependence on a specific platform (De Stefano, 2016).

However, we found that this customer evaluation system not only served to increase driver dependence, but also functioned to manipulate a driver’s labor process in three noticeable ways. First of all, a driver’s past customer evaluations often functioned as a precondition for any peak-hour or service-related bonuses. Drivers therefore devised various supplementary strategies to improve their service and entertain their passengers. For example, some drivers, at their own expense, provided passengers with paper tissues, bottles of water, Wi-Fi, as well as chargers. Drivers tried to talk with passengers, and served them in a service-oriented and sociable manner. They asked passengers whether they wanted the air conditioning in the car to be turned up or down, and what kind of music they preferred to listen to. Some drivers even tried to educate passengers on the consequences of their customer ratings, replete with stirring tales of past hardships. All in all, this process involved going far beyond one’s duty as a chauffeur, taxing drivers with additional emotional labor (Gandini, 2018) and handing a substantial amount of control over drivers’ working environment to their unwitting passengers.

Second, drivers’ working time was affected by this customer evaluation system. A driver’s total score was based on the last 500 rides that the driver had offered. The drivers who had a relatively low score would tend to increase the number of trips they offered in a day, hoping to eject their past poor ratings from the last 500 trips as soon as possible. As speculated upon by the drivers interviewed for this study and confirmed by Uber’s management thereafter, the platform factored a driver’s instant total score in its ride dispatching system, and higher ranked drivers would receive more ride calls with better, longer routes. Sometimes, the eligibility of a driver for all bonus packages depended on their customer evaluation results from the preceding week. If a passenger rated a driver one star, the driver would thus be forced to complete more five-star trips in order to offset the negative effect of that one rating as soon as possible. A good total score would not guarantee income security for a driver, but regardless, drivers had to be cautious about every single ride they provided in a given week.

Finally, Uber would also punish drivers directly based on their customer evaluation results. Low evaluation scores would disqualify drivers from certain bonuses, as noted – essentially a form of economic punishment. In addition, Uber would shut down a driver’s account if their scores dipped below a certain standard – a de facto outright termination. A customer could also appeal to Uber’s representatives to file complaints and the relevant driver would be open to variable penalties, which could build up to a lifetime ban from working with Uber once a complaint had been confirmed.

The customer evaluation system functioned as an effective tool for Uber to control their drivers, advertised under the misleading mantra that the ‘passengers are the best judge of service quality’. This passenger supervision occurred anywhere and at any time, and at low cost for Uber. Allowing customers to directly evaluate worker performance had also obscured the control the platform exerted over its drivers. Uber had become an invisible employer, and its drivers were kept from feeling supervised or managed directly by the platform. Driver complaints regarding any low ratings they might receive were invariably directed towards the passengers, as opposed to the platform itself. In extreme cases, this tension between drivers and passengers would boil over into confrontation, either right after a ride or later, through other means. When either one or both parties ended up taking a case to Uber directly, the company would almost always punish the driver, sometimes even compensating the passenger as well. Through this system, many potential labor–platform conflicts ended up neutralized.

In conclusion, this customer evaluation system facilitated drivers’ consent to Uber’s demanding work relations. We found that drivers who appreciated Uber’s customer evaluation system tended to spend more hours online as well as make more money per week, as shown in Tables 2 and 3. Given the fact that customer evaluation was strongly linked with the incentive pay system, sole-source drivers likewise were more sensitive to this control tactic than were multiple-job drivers. The t-test analysis on the effects of working hours did not pass the significance test at the 0.001 level for multiple-job drivers. And the analysis on the effects of income did not pass the significance test at the 0.05 level for the same group.

Consent making: Flexible work arrangements

Uber’s platform granted drivers certain nominal work flexibility and autonomy in exchange for their consent. The basic labor process is as follows. The app guided a driver through a flow of tasks. The system would dispatch a ride request to a certain driver close to the potential passenger. The driver who received the request had 15 seconds to press an ‘accept’ button on the app. Of course, drivers could also choose to decline if they did not want to accept the request for any reason, but too many declines would have adverse effects on their chances of future receipts of ride calls. The system would keep sending the call to more drivers until someone eventually took the offer. The driver who accepted the ride would be offered the option to communicate with the passenger(s) through a phone call or through instant messaging to further confirm or alter the pick-up location. After picking up the passenger(s), the system would recommend the best route for the ride. Once they arrived at their destination, payment would be completed through the app and the task of chauffeuring finished.

The platform appeared to grant drivers a high level of work flexibility and autonomy in several respects of this labor process. First, allowing drivers to take rides only when they had the time to do so increased work flexibility, and thus fostered greater feelings of personal freedom in drivers. In addition, these drivers owned and maintained their own means of production, namely their cars. This made drivers look and feel like entrepreneurs, as opposed to employees. Furthermore, even during a ride itself, the labor process was under the control of Uber only in an obscured sense. It was found that drivers working in this labor process did not consciously recognize that they were working for a platform. ‘Big data’ servers had significantly improved the efficiency of supply–demand matching, and rapidly expanding market demand had made it relatively easy to get a ride for an Uber driver during 2015–2016 in China.

This work flexibility in the gig economy is merely nominal, however. Digital platforms reduce the likelihood that workers will participate in collective action (Finkin, 2016), and workers have to face the arbitrary reductions in pay rates from platforms (Cockayne, 2016). Therefore, workers will be forced to extend their working hours, even though they are granted great freedom and flexibility in setting their own schedules on paper. Moreover, workers will often resort to continuous and long-time work schedules in order to maintain their rankings on such platforms. There is nothing freeing about the work on these platforms, concludes Aloisi (2016), and the nominal flexibility of these jobs has simply become a kind of solace.

In this study, it was found that Uber intentionally used this nominal flexibility to both attract and incentivize drivers. Uber propagated its business model using the tag line ‘You are your own boss. You decide when you want to drive, and for how long’. Uber advertised its platform-based work opportunities as characterized by freedom, flexibility, and discretion in order to attract more drivers. These findings indicate that the platform’s intention was to use this so-called flexibility to obscure the substantial control they held over the labor processes of their drivers. For these drivers, such platform-based work may be still seen as attractive though, as it offers a feeling of self-governance and autonomy, and thus these workers will voluntarily subject themselves to this new but insidious process of capitalist exploitation (Ettlinger, 2016).

It was found in interviews that most drivers did indeed maintain a feeling of working for themselves, and never regarded themselves as Uber employees explicitly. Many of the drivers did not even regard Uber driving as a formal occupation. They at least nominally reserved their right to decide when to get online and when to stay offline. And even during their time online, the otherwise monotonous driving work becomes a game of sorts, in which one simply ‘touches the screen and pushes the pedal down’, but forgets that one is still under the control of the platform via its app.

In sum, flexible work arrangements granted such drivers marginal autonomy in controlling their own schedules, but in return acquired drivers’ consent in the relations of production in Uber driving. This was how work flexibility began to serve as a more powerful medium for hegemonic control, to use Burawoy’s (1985) description. This is demonstrated by the results of the t-test analysis in Tables 2 and 3.

That being said, work motivation had a mediating role here too. Specifically, the multiple-job drivers would often show their appreciation for the flexibility of Uber driving, as they could only drive in their spare time. Conversely, the sole-source drivers did not care for the platform’s work flexibility as much as multiple-job drivers, as they would typically drive more than 14 hours a day anyway. As a result, the t-test analysis for the sole-source driver group did not pass the significance test at the 0.05 level, as shown in Table 2. These sole-source drivers were therefore subject to poor working conditions as a result of their inadvertent self-exploitation.

Not all drivers were satisfied with working for Uber, to be sure. As Uber and its major competitor Didi had gradually reduced the bonuses they offered since March 2016, drivers had started to complain correspondingly. Some drivers began to question the rules developed by the platform: ‘The 20% service fee charged by the platform is too much. This even exceeds the money paid by taxi drivers to their taxi companies.’ Some drivers also noticed their working hours rising continuously over time: ‘[The work is] so restraining that I have to keep driving since the early morning, and cannot get out of my car… I cannot earn money if I do not drive.’

Alarmingly, the ride-hailing industry has shown the potential to become a new epicenter for industrial conflict in China. According to the China Labor Bulletin’s Strike Map,⁴ digital platform-based drivers in Guangdong Province alone launched 16 strikes against arbitrary fare and bonus reductions, among other issues, from May 2015 to August 2018. Indeed, after a short wave of protests in late 2016, many drivers eventually decided to quit working in the sector. Regardless, little on the side of the platforms has changed since, and the drivers who stayed have ultimately accepted the rules, although many of them still expected a return to the golden days: ‘On the whole I’m satisfied with it. I just wish it would offer more bonuses.’

Discussion

While there is growing scholarly interest in the work and employment issues that have emerged with the gig economy, most scholarly debates have centered on whether the work relationship in online, platform-based gig economy work ought to be formally considered an employment relationship (Aloisi, 2016; Cherry and Aloisi, 2017; Cunningham-Parmeter, 2016; Finkin, 2016; Harris and Krueger, 2015; Scott and Brown, 2017; Stewart and Stanford, 2017). Scholars in this debate seem to agree that although the expansion of digital platform businesses has facilitated a global trend toward work informalization, capital has reformulated (rather than relinquished) its control over the labor process. This is consistent with classic Marxist dialectics. Wherever there is surplus value extraction, there is labor control.

A key challenge for any online platform business is how to secure sufficient labor supply and manipulate workers’ time online at low cost to themselves. This has become a particularly tricky challenge in the emergent gig economy, as the idea of independent contracting has limited the tools capital has traditionally utilized for labor control (e.g. various human resource management tactics, unions and collective bargaining agreements, and employment contracts). This study therefore sought to systemically examine the alternative strategies that have been devised by platforms to achieve labor control regardless, and even more importantly, the effects of these new tactics over labor.

Through qualitative research, the authors identified three important strategies that Uber, once an important digital platform business in China, had employed to attract, manage, and retain its drivers, while at the same time realizing labor control. With its incentive pay system, customer evaluation system, and flexible work arrangements, Uber was well-equipped to successfully extend its drivers’ working time.

These findings contribute to our understanding of the inner workings of the gig economy in several ways. The literature on service work, following emotional labor theory (Hochschild, 1983), provides insights into how the tangible presence of consumers has been introduced into workplaces as a labor control tool, which can have a significant impact over work relations in service industries. According to ethnographers engaged in various research settings, customer needs and preferences are taken into account in the design of service work processes in order to guarantee quality and sometimes personalized service (England, 2005; Mirchandani, 2012; Otis, 2012). In this study, it was found that a similar customer evaluation system and an intertwined incentive pay system had been institutionalized in the labor processes of Uber drivers. Here, customer evaluation was designated as a crucial labor control strategy, not only to encourage quality service, but also to extend and manipulate the working time of the platform’s drivers.

In the human resource studies literature, work flexibility is a well-developed construct. Scholars have examined various forms of work flexibility, such as ‘flextime’ (scheduling flexibility) and ‘flexplace’ (telecommuting) in organizational settings (Rau and Hyland, 2002). Some researchers regard such modern work arrangements as benevolent human resource management tactics for tackling conflicts in the work–family balance (Allen et al., 2013). Arguably, work flexibility is positively associated with employee satisfaction and work commitment (Azar et al., 2018), thereby functioning as a form of labor control strategy. Other scholars, however, have argued that work flexibility is a business strategy utilized to generate an informal sector, outside of formal organizational settings, as well as nonstandard employment relations, subsequently developing a precarious workforce (Kalleberg, 2003). All in all however, it has been concluded that informal work does objectively grant workers more autonomy (Millar, 2008). Moreover, workers have been found to sometimes value this flexibility, and to be willing to pay the price for it (Mas and Pallais, 2017). This study has found that in the informal sector, specifically the gig economy, online platform companies have been intentionally propagating and capitalizing on this desire for labor autonomy in order to attract and control their workers.

In practice, however, workers ended up hardly autonomous at all, and their work became less flexible over time. This was especially the case if Uber driving was the driver’s only source of income. Drivers had to commit to extremely long periods online, both in waiting for and taking on rides, in order to receive a realistic income. The huge gaps that existed between Uber fares and traditional cab fares would force drivers to ‘voluntarily’ subject themselves to the platform’s manipulation of their time and scheduling through the company’s use of incentive payments. The surprisingly large proportion that bonuses took up in a sole-source Uber driver’s total income illustrated the success that this strategy had achieved for the platform. Therefore, instead of granting autonomy, such flexibility had become at best a strategy for obscuring the employment relationship and extracting surplus value.

That being said, these different labor control strategies may not have equal effects in practice, and this study’s primary contribution to the field is its discovery of how considerable variation is exhibited across workers with different motivations for working in the gig economy. Past research has already deciphered Uber’s business model, and has by and large noticed how such online platforms use different tactics to control their workers (Aloisi, 2016; De Stefano, 2016; Rosenblat and Stark, 2016). Through examining their effects, however, it was found in this study that these labor control strategies are likely to demonstrate significant effects only when we consider the different motivations of such drivers. Here, it is suggested that there exists an important link between work motivation and worker behavior in response to the labor control practices of these platforms. In particular, work flexibility was found to be crucial for drivers with multiple jobs and sources of income who only drove for Uber in their spare time. This second category of drivers tended to offer more rides if they appreciated the time flexibility that Uber driving presented them. In contrast, sole-source drivers exhibited income dependence on Uber, and were thus more likely to be motivated by the so-called rules of the game. These drivers were likely to work more if they appreciated the platform’s incentive pay and customer evaluation system.

With the expansion of the gig economy on a global scale, debates among scholars, research institutes, as well as labor law practitioners have emerged regarding how to conceptualize the various forms of work relationships in a gig economy. Increasingly, research reports have recommended factoring in the theme of labor control when analyzing the nature of the work relations that are based on digital platforms (Cunningham-Parmeter, 2016; Harris and Krueger, 2015). The courts in many of the world’s nations will very often follow a similar rationale. This study indicates that, even though platforms will use the same set of control tactics to manage all its workers, the effects of these strategies may vary across workers with different motivations for working with said platform. Different people simply expect different things from engaging in gig economy work, such that policymaking and judicial decisions should not only draw on how platforms choose to control their workers, but also draw on how workers will react to these control mechanisms.

Conclusion

Drawing on qualitative field research supplemented by quantitative data, this study probed the labor process of digital platform-based cab drivers. Three labor control strategies that Uber in China employed in order to secure and obscure surplus value were identified as a result. Specifically, they were an incentive pay system, a customer evaluation system, and flexible work arrangements. The different effects that these strategies have on Uber drivers who are solely employed by the platform, and those with other regular forms of income as well, were also discussed. It was argued that multiple-job drivers’ appreciation of the platform’s work flexibility prompts them to spend more hours online, while sole-source drivers are motivated better by the platform’s incentive pay and customer evaluation systems.

Although Uber can be taken as an important case for investigating labor control in a gig economy, it is by no means a representative case for the entire digital platform-based sector. Future research should extend this study’s examination of these labor processes to other firms and industries in the gig economy, which may elicit variable situations regarding labor control. Quantitative analysis is also needed to examine the effects of such labor control strategies using data of better quality than that in this study. Finally, although this article has indicated that drivers’ complaints are by and large directed towards the passengers that give them bad ratings, and that these drivers seldom question the rules established by their work platform, labor conflict is by no means completely avoided through such scapegoating, as noted earlier. While this research was not geared towards capturing those moments of conflict specifically, future studies ought to examine more closely instances of the labor–capital conflict in the gig economy that are bound to form as the industry grows further.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Qingjun Wu is Associate Professor of Labor Relations at Renmin University of China’s School of Labor and Human Resources. His primary areas of expertise are the sociology of work and industrial relations. And his current research examines the impact of the digital economy over work and labor.

Hao Zhang is Assistant Professor of Labor Relations at Renmin University of China’s School of Labor and Human Resources. His research interests focus on the areas of employment relations, informal work, and skills development, with special attention to China.

Zhen Li received her Master of Labor Relations from Renmin University of China’s School of Labor and Human Resources in 2017 and is now the Compensation & Benefits Manager at China Central Television.

Kai Liu is Associate Professor of Social Security at Renmin University of China’s School of Labor and Human Resources. His primary areas of expertise are health policy and social policy in developing countries. And his ongoing project examines the effects of the digital economy on workers’ wellbeing, in particular their social insurance enrollment in China.

References

Allen

Johnson

Kiburz

, et al.(2013) Work–family conflict and flexible work arrangements: Deconstructing flexibility. Personnel Psychology 66(2): 345–376.

Aloisi

(2016) Commoditized workers: Case study research on labor law issues arising from a set of “on-demand/gig economy” platforms. Comparative Labor Law and Policy Journal 37(3): 620–653.

Azar

Khan

Van Eerde

(2018) Modelling linkages between flexible work arrangements’ use and organizational outcomes. Journal of Business Research 91: 134–143.

Barker

(1993) Tightening the iron cage: Concertive control in self-managing teams. Administrative Science Quarterly 38(3): 408–437.

Batt

Colvin

Keefe

(2002) Employee voice, human resource practices, and quit rates: Evidence from the telecommunications industry. Industrial and Labor Relations Review 55(4): 573–594.

Braverman

(1974) Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century, New York: Monthly Review Press.

Burawoy

(1979) Manufacturing Consent: Changes in the Labor Process under Monopoly Capitalism, Chicago, IL: University of Chicago Press.

Burawoy

(1985) The Politics of Production: Factory Regimes under Capitalism and Socialism, London: New Left.

Cappelli

Keller

(2013) Classifying work in the new economy. Academy of Management Review 38(4): 575–596.

10.

Cherry

Aloisi

(2017) Dependent contractors in the gig economy: A comparative approach. American University Law Review 66: 635–689.

11.

Clawson

(1980) Bureaucracy and the Labor Process, New York: Monthly Review Press.

12.

Cockayne

(2016) Sharing and neoliberal discourse: The economic function of sharing in the digital on-demand economy. Geoforum 77: 73–82.

13.

Cunningham-Parmeter

(2016) From Amazon to Uber: Defining employment in the modern economy. Boston University Law Review 96: 1673–1728.

14.

De Stefano V (2016) The Rise of the Just-in-Time Workforce: On-demand Work, Crowdwork and Labour Protection in the Gig-economy. Conditions of Work and Employment Series Working Paper No. 71. Geneva: ILO.

15.

Edwards

(1979) Contested Terrain, New York: Basic Books.

16.

England

(2005) Emerging theories of care work. Annual Review of Sociology 31: 381–399.

17.

Ettlinger

(2016) The governance of crowdsourcing: Rationalities of the new exploitation. Environment & Planning A 48(11): 2162–2180.

18.

Finkin

(2016) Beclouded work, beclouded workers in historical perspective. Comparative Labor Law & Policy Journal 37: 578–603.

19.

Friedman

(1977) Industry and Labour, London: Macmillan.

20.

Friedman

(2014) Workers without employers: Shadow corporations and the rise of the gig economy. Review of Keynesian Economics 2(2): 171–188.

21.

Gandini A (2018) Labour process theory and the gig economy. Human Relations. Epub ahead of print 25 September 2018. DOI: 10.1177/0018726718790002.

22.

Gerhart

Rynes

Fulmer

(2009) Pay and performance: Individuals, groups, and executives. Academy of Management Annals 3(1): 251–315.

23.

Harris SD and Krueger A (2015) A Proposal for Modernizing Labor Laws for Twenty-first-Century Work: The “Independent Worker”. The Hamilton Project Discussion Paper No. 2015-10, December.

24.

Hochschild

(1983) The Managed Heart: Commercialization of Human Feeling, Berkeley, CA; London: University of California Press.

25.

Kalleberg

(2003) Flexible firms and labor market segmentation: Effects of workplace restructuring on jobs and workers. Work and Occupations 30(2): 154–175.

26.

Katz L and Krueger A (2016) The Rise and Nature of Alternative Work Arrangements in the United States, 1995-2015. National Bureau of Economic Research Working Papers No. 22667, September.

27.

Lazear

(2000) Performance pay and productivity. American Economic Review 90(5): 1346–1361.

28.

Littler

(1982) The Development of the Labor Process in Capitalist Societies, London: Heinemann.

29.

Marx K ([1867] 1990) Capital, Volume I. Fowkes B (trans.). London: Penguin Books.

30.

Mas

Pallais

(2017) Valuing alternative work arrangements. American Economic Review 107(12): 3722–3759.

31.

Millar

(2008) Making trash into treasure: Struggles for autonomy on a Brazilian garbage dump. Anthropology of Work Review 29(2): 25–34.

32.

Mirchandani

(2012) Phone Clones: Authenticity Work in the Transnational Service Economy, Ithaca, NY: Cornell University Press.

33.

Otis

(2012) Markets and Bodies: Women, Service Work, and the Making of Inequality in China, Stanford, CA: Stanford University Press.

34.

Rau

Hyland

MAM

(2002) Role conflict and flexible work arrangements: The effects on applicant attraction. Personnel Psychology 55(1): 111–136.

35.

Rosenbalt

Stark

(2016) Algorithmic labor and information asymmetries: A case study of Uber’s drivers. International Journal of Communication 10: 3758–3784.

36.

Schmidt FA (2017) Digital Labour Markets in the Platform Economy: Mapping the Political Challenges of Crowd Work and Gig Work. Good Society-Social Democracy #2017 Plus Project Report, Friedrich-Ebert-Stiftung, Germany.

37.

Scott

Brown

(2017) Redefining and regulating the new sharing economy. University of Pennsylvania Journal of Business Law 19(3): 553–599.

38.

State Information Center (2019) Chinese sharing economy development report 2019. Available at: www.sic.gov.cn/News/557/9904.htm (accessed 19 April 2019).

39.

Steinberg

Figart

(1999) Emotional labor since the managed heart. The Annals of the American Academy of Political and Social Science 561(1): 8–26.

40.

Steinmetz

(2015) Is the on-demand economy taking workers for a ride? Time 186(5): 11–15.

41.

Stewart

Stanford

(2017) Regulating work in the gig economy: What are the options? The Economic and Labour Relations Review 28(3): 420–437.

42.

Thompson

(1990) Crawling from the wreckage: The labour process and the politics of production. In: Knights

Willmott

(eds) Labour Process Theory, London: Macmillan, pp. 95–124.

43.

Veen

Barratt

Goods

(2019) Platform-capital’s “app-etite” for control: A labour process analysis of food-delivery work in Australia. Work, Employment and Society. Epub ahead of print 25 March 2019. DOI: 10.1177/0950017019836911.