Executive Summary

The rapid expansion of “digital labor platforms” – those that serve as a mediator between service providers and customers – is seen as a promising engine for employment, empowerment, productivity improvements, and poverty-alleviation (World Bank 2022). Across 8 countries in sub-Saharan Africa, Johnson et al. (2020) document growth between 38%-52% in the number of digital labor platforms, and the number will likely continue to grow. Digital labor platforms may be especially promising for providing employment opportunities in low and middle-income countries where the vast majority of the population is engaged in informal and casual labor (e.g., over 80% of the workforce is informal in Kenya, Ghana, and India; JobAlliance 2023, ILO 2017). 

In general, gig work is defined as a form of self-employment characterized by short-term contract (or non-contract) work, freelancing, and self-employment, and is paid per task rather than per hour of work (US Bureau of Labor Statistics, 2018). The main focus of the literature review is digital labor gig work – a subset of all gig work – in which short-term employment jobs are found through a digital platform. Using big data and algorithms, the platforms are digital interfaces that mediate supply and demand between the service provider, setting prices for services that incorporate market conditions, location of workers and customers, ratings of workers, and other factors (Berg, 2016, Vallas & Schor, 2020). 

Despite the increasing prevalence of digital gig work, there remain significant knowledge gaps about how digital platforms affect the workers. The goal of this literature review is to examine and aggregate the current literature on low-capital, location-based digital gig workers’ experience with digital labor platforms in low and middle-income countries (LMICs), bring out recurring themes, highlight differences and points of disagreement between studies, and form an initial narrative of the digital economy landscape across various countries and sectors. While our goal was to cover all major sectors likely employing lower-income populations in digital labor platforms, which are (1) passenger drivers, (2) delivery drivers, (3) home services, and (4) domestic work, the vast majority of empirical descriptive evidence is available for digital drivers. No descriptive evidence was found on home service workers. Gray literature¹, organization and government reports, and white papers are included due to the paucity of academic and rigorously conducted studies. The focus of the review is on experiences working for the platforms and welfare implications on the worker and does not include consumer experiences.²

The narrative of the review is driven by the available empirical evidence from LMICs; however, selected findings from high-income countries (HICs) – where platform work originated – are included for comparative purposes. The contrasts are drawn between these settings to highlight common patterns and divergence, which would ultimately inform how similar or different policy responses should be. 

The following unique features characterize the platforms. 

• They operate as intermediaries, facilitating interactions between different parties rather than offering goods or services directly. 
• The transactions are mediated by algorithms relying on access to large volumes of historical and current data. 
• Service pricing within these platforms is often but not always dynamic, adjusting in real-time to meet the ebb and flow of supply and demand. 
• Entrants into new markets implement diverse and aggressive strategies to capture the full market due to network effects – i.e., each customer increases the volume of work for service providers, and each service provider increases the quality of services offered. Additionally, higher transaction volume directly improves matching algorithms, further improving service quality. 

As a result of this aggressive market structure, digital platform markets are often monopolistic (or monopsonistic) or oligopolistic (or oligopsonistic), characterized by one or a few dominant players.

The literature in LMICs includes a few quantitative large-N studies. However, the vast majority of these studies are not derived using representative samples, so conclusions reached in this literature review may be subject to bias arising from the non-representativeness of samples. It is likely that studies that do not use administrative data from companies as sampling frames over-represent full-time workers and under-represent part-time workers since full-time workers spend more hours driving; however, there is no empirical support for that, and the direction of bias is unknown. The literature and conclusions of the review are dominated by studies of driving sectors, likely because driving apps were some of the earliest entrants in the market. 

Key Themes:

• Demographics:
  • In LMICs, digital app workers are younger, more educated, and more likely to live and work in urban environments than the general population. In the current research, the vast majority of workers are male, but this finding is driven by a high prevalence of driving gig work and overrepresentation of the driving sector in the literature.

• Inclusion of women/poorer populations: 
  • Digital labor platforms are generally gender-segregated and align with gender norms and traditional sectors that men/women occupy. Women face similar barriers to accessing platform work as in traditional employment, such as gender role norms and expectations, lower access to capital (e.g., smartphones), concerns about personal safety, and more hazardous working conditions. There is a documented gender-pay gap in the e-hailing sector in the US, which is driven by men being able to take advantage of surge pricing and drive at hours that are compensated at higher rates. It is unlikely that the platforms are reaching the most vulnerable and urban poor in LMICs as workers since access to digital platform gig work is conditional on having a smartphone³, penetration of which is still far from universal even in urban areas. However, there are major efforts to increase smartphone penetration in LMICs, which will lower entry barriers (e.g., USAID Digital Economy Initiative)

• Labor Supply:
  • Higher proportion of platform workers work full time in LMICs compared to the HICs, which is driven by entry barriers into some sectors (e.g., driving). A higher prevalence of “idle assets” (such as cars) in HICs lowers the barriers to entry into driving work, whereas, in LMICs, the choice of platform entry is often associated with significant investments. There is a higher prevalence of full-time drivers in some European countries with more stringent entry requirements (such as taxi licenses to operate). This trend does not hold universally in LMICs, and in some countries, there is an equivalently high presence of part-time drivers. This pattern may or may not hold for non-driving sectors where barriers to entry are lower. 
  • The population of drivers is highly heterogeneous, and those who engage part-time are very different from those who choose to work full-time. There is also high heterogeneity within the part-time worker population, even holding total hours of work constant, and those who work a few hours on the weekends and evenings likely have different profiles, economic circumstances, and goals compared to workers who work sporadically throughout the day. 
  • Workers who choose to work full-time usually work long hours (up to 70 hours/week) – a pattern predominant in the driving and delivery sectors. However, it’s important to note that workers in comparable offline professions are also known to work long hours, suggesting that the trend is not fully driven by the digital nature of work but rather reflects patterns in the broader sector. Most workers in LMICs join platform work to either earn  additional income or attain higher income compared to previous jobs and have a flexible working schedule – a feature especially important to women. In contrast to HICs, in LMICs, these jobs offer more formality compared to other similar offline gigs; workers in some countries report feeling like they are a part of the more formal labor market.

• Income: 
  • The evidence on how the earnings compare to minimum wage is mixed, but many studies find that gross earnings exceed minimum wages. That said, it is important to consider the various estimation challenges that are not consistently addressed in studies. A significant number of these studies benchmark gross earnings against minimum wages, a methodology that may not reflect the actual take-home pay when operating costs like fuel, maintenance, and equipment rental are factored in. Two studies that do account for operating expenses find that the net per-hour income is not universally higher than minimum wage and is similar to other jobs available to similar workers in other sections. 

• • The vast majority of studies report gig workers earning higher income compared to their previous jobs and select studies report that income is higher than similar offline occupations. However, the finding is somewhat tautological, as it simply mirrors the revealed preferences of current platform workers. These individuals are a part of the platform work market precisely because it offers a more suitable arrangement for them.
  • To our knowledge, there is only one causal study of the effects of gig work on income in LMICs. It shows null average effects of access to platform work but positive employment outcomes and increases in income for women whose skills closely match the jobs provided on the platform. This suggests that while platforms may not change economic circumstances for an average worker, there are identifiable benefits for workers with high job search costs.
  • In many cases, workers are still economically vulnerable due to either insufficient or fluctuating income

• Work Conditions:
  • There are mixed views on the importance of work flexibility – which is the main differentiating feature of the platform work compared to other jobs. There is a sub-population of workers who engage more on a full-time basis; for those individuals, flexibility is less important. Those who are working part-time highly value this aspect. Ultimately, the degree of value of flexibility depends on the motivation to join the platform.

• Income Mobility:
  • Given the limited evidence, it is challenging to conclude whether work in the digital labor economy fosters income mobility and contributes to long-term career growth. There is limited earning growth within the platform work, especially in jobs with little service differentiation, such as delivery and passenger driving. However, little is known about growth prospects and earnings differences for services with higher differentiation, such as domestic work and home services. There are some studies that document that some workers are pursuing higher education, suggesting that gig jobs may foster human capital development. Still, the extent of this phenomenon is not well documented. While platform work may improve customer service skills – potentially valuable for the broader labor market, it is currently unknown whether and to what extent these skills are transferable. Accumulation of savings – another important mechanism for income mobility – is unlikely, given that workers report feeling financially insecure.

• Financial Inclusion:
  • While there are a number of promising avenues for digital labor platforms to improve the financial inclusion of the workers, empirical evidence is very limited. There are a few theoretical ways in which engagement with platform work can facilitate wider financial inclusion: (1) in many contexts, having a bank account is a prerequisite to joining the platform, and in some cases, platforms help workers open accounts. This, in turn, may improve worker’s ability to receive interest-bearing savings accounts and increase access to loans; (2) one of the challenges of obtaining formal credit among workers is lack of proof of income. Platform income records can be used as a proof of income for securing loans; (3) platforms can directly provide financial products on the platform. Automatic, pre-committed saving deductions can facilitate consistent saving behavior.  Transactional and work pattern data can be used for generating digital credit scores and allow for differentiating creditworthy workers. The credit scores can be used to offer loans with more favorable terms. However, limited validation work has been done to verify that the platform data is sufficiently rich to generate creditworthiness; (4) Digital credit scores generated by the platform can be used in the wider ecosystem to allow for the extension of more credit in  other financial institutions. However, the success of broader financial inclusion largely depends on the integration and sharing of information between different players: banks, digital platforms, financial service providers, and credit bureaus. 
  • Platforms recently started collaborating with digital service providers to offer digital products to alleviate credit/saving constraints and improve cash management. These partnerships are mutually beneficial since platforms are interested in reducing worker churn, and financial providers are interested in increasing their customer base. The creation is challenging and requires iterative piloting to create products that fit the workers’ goals and needs. Limited empirical evidence suggests that the takeup is currently low, and platforms are not reaching their workers.