Online freelancing and impact sourcing: Examining the inclusive development potential of online service work in the Philippines

Online freelancing and impact sourcing have in recent years emerged as new models for offshore service delivery. Both have the potential of spreading the gains of online service work. Based on empirical research in the Philippines, this article examines how both models integrate outlying areas and more marginalized workers in international networks of online service delivery. The different models of information and communication technologies (ICT)‐enabled service delivery were observed to rely on the same pool of labor, thereby limiting the broader distribution of its gains. The article concludes that ICT4D research can benefit from an inclusive development lens when examining the beneficiaries and users of new (information) technologies and their longer‐term prospects for income generation.

A comparison of the two models shows that they emerged from diametrically opposite orientations. Whereas CSR motivates impact-sourcing initiatives, labor arbitrage through low-cost labor sourcing is central to online freelancing digital labor platforms (Beerepoot & Lambregts, 2015).
Labor arbitrage has intensified the North-South divide where work mainly comes from the Global North, and online freelancers are based in the Global South (Berg et al., 2018;. Unlike the community orientation of impact sourcing (Kannothra et al., 2018), digital labor platforms are criticized for the lack of social protection and uneven distribution of profits for freelancers (Berg et al., 2018;Wood et al., 2019). Despite this, Carmel et al. (2016) describe a developmental overlap between impact sourcing and online freelancing by arguing that they share aspects of pro-poor, market-based, and poverty reduction strategies.
From the perspective of potential developmental gains, both impact sourcing and digital labor platforms are manifestations of what is regarded as ICT for development (ICT4D) 2.0 1 (Heeks, 2014(Heeks, , 2018Walsham, 2017). They exemplify the evolving orientation in research and policymaking of ICT4D from an initial focus on the availability of ICTs for helping the poor in becoming users of digital content (Avgerou, 2010;Heeks, 2009), through the uptake of ICTs as a tool for income generation such as online work (Heeks, 2009(Heeks, , 2014, to a focus on the impact of ICT on social and economic development goals (Heeks, 2009(Heeks, , 2014(Heeks, , 2018. 2 ICTs overcome the social, cultural, and physical barriers that exclude marginalized groups from participating in the labor market (Everest Group, 2014;Rockefeller Foundation, 2011). However, the means through which the poor can utilize ICTs for income generation requires examination for a deeper understanding of the development gains of ICTs.
In this article, digital labor platforms and impact sourcing are compared with regard to the opportunities they provide for spreading the gains of online work. Economic benefits are reviewed along two dimensions: the geographical spread to more peripheral areas and the extent to which economically marginalized workers are included in international networks of digital service delivery. From a business perspective, financial gains emanate from the ambition to create shared value (CSV) (Porter & Kramer, 2011) in the community where a firm operates. From a community perspective, the financial gains of the deployment of ICTs should contribute to inclusive development (Hickey et al., 2015). However, information system (IS) literature reveals an intrinsic bias toward the business perspective when analyzing online service work (Babin et al., 2019;Nicholson et al., 2016). For example, in impact sourcing, Carmel et al. (2016) argue that most research has focused on only two stakeholders-the impactsourcing operations and their clients, and not the workers. This leaves the developmental value of this type of employment relatively unexplored. Sein et al. (2019) note that ICT4D research often lacks an explicit development perspective, while Avgerou (2010Avgerou ( , 2017 opposes the causal claims about ICT and development. ICT4D research can benefit from an inclusive development lens in which including excluded people and utilizing their capabilities is pivotal (Johnson & Andersen, 2012). As de Leth and Ros-Tonen (2021) argue, inclusive development theory helps operationalize societal value and make the inherent tension between business objectives and societal value (as in CSV) more transparent. This article argues that research in ICT4D can benefit from an inclusive development lens when examining the beneficiaries and users of new (information) technologies and their longer-term employment prospects.
This article is based on empirical research among impact-sourcing workers and online freelancers in the Philippines. The Philippines is a lower-middle-income country with high unemployment and underemployment (Ofreneo, 2015) and is one of the foremost providers of freelance labor on global digital labor platforms (Beerepoot & Lambregts, 2015). Freelance labor and online service delivery, primarily by large-scale business process outsourcing (BPO) companies, offer English-speaking College graduates employment opportunities (Kleibert, 2015). Recently some impact-sourcing initiatives have been established there. Furthermore, the government has introduced policies toward enhancing the participation of Filipinos in digital labor platforms (Bayudan-Dacuycuy et al., 2020). Multiple international policy reports written on ICT-enabled work (cf. ADB, 2021;ILO, 2021;UNCTAD, 2019) illustrate the development promise expected from this type of employment.
The balance of the article is as follows. Section 2 explains how impact sourcing and digital labor fit within the evolving state of offshore service delivery and the current debates on ICT4D. Section 3 describes the methodology of this study. Section 4 analyses the geographical spread of both forms of digital work in the Philippines. Section 5 provides a profile of impact-sourcing workers and online freelancers and their longer-term prospects for employment. Finally, Section 6 provides the concluding remarks to this article.

| EVOLVING LANDSCAPE OF ICT-ENABLED SERVICE DELIVERY
The past decade has seen many optimistic reports on the "service revolution" (Ghani & Kharas, 2010;Ghani & O'Connell, 2014). The service revolution claims that through innovations in ICT and the increasing tradability of services, rapid income growth, job creation, gender equality, and poverty reduction can be achieved in developing countries. The services revolution has opened many more niches for countries through which they can leverage the global economy (Ghani & Kharas, 2010). Emergent global services production networks assimilate (semi)peripheral places into the global economy and offer new economic opportunities to their economic agents (Lambregts et al., 2016). Most prominent examples of new economic opportunities provided include the range of BPO services such as low-value-added voice-based services (call centers), back-office services (data entry, transcription), and high-value services such as legal services, financial analytics, and engineering design, which require domain 1 Information and Communication Technology for Development. expertise and specific skills (Lambregts et al., 2016). The rise of the BPO sector in countries in the Global South is part of a broader and longer trend of companies outsourcing non-core activities to specialist service providers (Lacity & Willcocks, 2006;Porter, 1998). Innovation in ICTs has made the international dimension feasible. Lu et al. (2015) summarize the trends in offshore outsourcing of services as shown in Table 1. To their original categorization of three waves, we added impact sourcing as a mode of service delivery and added more detail on the characteristics of each wave. Outsourcing of services evolved from a domestic to a global offshore process and from being driven by large firms with large volume contracts to be equally accessible to smaller firms and individuals through online freelance work. The emergence of impact sourcing within this framework illustrates how the mainstream services outsourcing sector has adopted social responsibility objectives.
Whereas internationalization of services production has been mainly driven by larger multinational companies, digital labor platforms make outsourcing and offshoring of work accessible to smaller companies and individuals (Beerepoot & Lambregts, 2015;Lehdonvirta et al., 2019).
Thus, by enabling the outsourcing of "small tasks" or microwork with a short duration and often limited compensation, digital labor platforms bring service outsourcing within reach of new actors. With internet connectivity spreading at an unprecedented rate (Graham & Mann, 2013), and the prices for personal computers and internet use dropping in most parts of the world, online work via digital labor platforms is within reach of an increasing number of people in developing countries (Kässi & Lehdonvirta, 2018).
What differentiates an impact-sourcing venture from traditional digital service delivery is considering the relative "need" of the respective communities for suitable employment opportunities (Bulloch & Long, 2012;Kannothra et al., 2018). Social enterprises employ and train people who have limited opportunities for decent employment to provide information-based services to domestic and international clients (Madon & Sharanappa, 2013). As such, impact sourcing aims to look beyond the common source of supply for traditional outsourcing to provide higher-income employment and access to new income opportunities to individuals who may not otherwise be employed in this sector (Bulloch & Long, 2012). Various case studies describe social enterprises and CSR initiatives that provide employment outside of core locations for service outsourcing by targeting new workers groups (Heeks & Arun, 2010;Madon & Sharanappa, 2013;Sandeep & Ravishankar, 2018). Prominent and frequently cited examples of impact-sourcing initiatives include Samasource, Digital Divide Data, and CloudFactory (Kannothra et al., 2018;Melia, 2020;Nicholson et al., 2016).
A prime motivation for firms to get involved in impact sourcing (or outsource work to impact-sourcing units) is to engage in CSR activities that are closely related to their core competencies (Oprins & Beerepoot, 2019). Here they follow Porter and Kramer's (2011) principle of creating shared value, which connects societal and economic progress. Nevertheless, the success of impact-sourcing units in generating societal value is limited by some critical operational challenges in running them. The main weakness of many impact-sourcing initiatives is that they are based mainly on the traditional model of office-based service delivery carried out from a remote location or involving new groups of underprivileged workers. This requires substantial transaction costs in training workers and building the necessary infrastructure to operate. Remote locations are less equipped with the infrastructure necessary to support a scalable center that can compete with international players (Bulloch & Long, 2012), T A B L E 1 Trends in outsourcing and offshoring of services

| Positioning digital labor platforms and impact sourcing in ICT4D
Impact sourcing and digital labor platforms can be seen as recent trends in service offshoring, but when focusing on their potential beneficiaries, they can equally be seen as manifestations of ICT4D. ICT4D has emerged as a vibrant field for investigation in the past decades (Graham, 2019;Heeks, 2020;Walsham, 2017). Sein et al. (2019, p. 9) argue that "ICT4D lies in the intersection between ICT, development and the transformative process by which ICT may lead to development." De et al. (2018, p. 63) define ICT4D research as seeking "to examine the social and economic changes in developing countries brought about by the deployment and use of ICT." Initially a subfield of IS research, a range of academic disciplines (development studies, anthropology, and geography) has since explored the use and application of ICT4D. ICT4D initially focused on how ICTs (particularly the Internet) could be applied in developing countries (Heeks, 2009;Nicholson et al., 2016) and primarily addressed the access of the poor to ICT. From this initial ICT4D 1.0 (Heeks, 2009), the attention shifted to how ICTs can be utilized for income generation by the poor.
ICT4D 2.0 focuses on the new applications and business models (as in this case, impact sourcing and digital labor platforms) that can use the growing ICT base of mobile phones, telecenters, and the like, to create employment (Heeks, 2009 (Heeks, 2013), the bottom of the pyramid becomes a source for the delivery of online work. However, this requires reframing the poor, not as passive consumers of technological innovation (as under ICT4D 1.0) but as active producers and innovators. Heeks argues that the, sharp end of impact sourcing lies with the suppliers: The base-of-the pyramid workers who deliver the IT services and who typically live in developing countries. One has to move beyond the slum dweller or peasant farmer stereotype to understand this group. Substantial numbers of the poor have high school diplomas or college, even university, degrees. (Heeks, 2013, p. 24) This exemplifies that ICT4D research should increasingly focus on precisely who are the beneficiaries and users of new technologies (Arora, 2019) and the impact that it has on them. Consequently, ICT4D research can benefit from an inclusive development lens for a better operationalization of societal value.
Despite the ambiguous use of the term inclusive development (Hickey et al., 2015), inclusive development focuses not only on the distribution of social and material benefits across social groups and categories but also on the structural factors that cause and sustain exclusion and marginalization of vulnerable groups in society (van Gent, 2017). This justifies the choice for geographical spread and inclusion of marginalized communities as key dimensions for analyzing the societal gains of online service work. Other dimensions of inclusive development involve passive and active inclusion (Andersen & Andersen, 2017). Whereas ICT4D 1.0 is oriented toward passive inclusion (enabling access to ICTs), ICT4D 2.0 should be regarded as active inclusion, effectively utilizing ICTs for income generation. As stated by UNCTAD (2019, p.121), inclusive development gains from the digital economy will not be realized automatically from simply expanding access to affordable broadband connectivity.
The relevance of an inclusive development lens is exemplified by the current knowledge gap on who the beneficiaries are of both forms of digital labor and what long term opportunities they afford for these individuals (Carmel et al., 2016). Research reveals that a problem with impactsourcing initiatives is that the workers included are not the most remote, excluded or the poorest (Heeks & Arun, 2010;Malik et al., 2016). Madon and Ranjini (2019) observe that impact-sourcing service providers are reluctant to set up or start operations in rural areas due to infrastructural issues such as access to good education, health care, and transportation, preventing the deployment of experienced staff to rural areas.
Concerning digital labor,  argue how more empirical detail is required on who benefits from online freelancing, the knowledge spillovers, and impacts on local economies and communities. At the micro (individual) level, ICT4D research predominately analyses the impact on economic well-being (Heeks, 2018(Heeks, , 2020Walsham, 2017). Carmel et al. (2016) define the impact of impact sourcing as improving livelihoods, building skills and capacity, and increasing incomes. Heeks (2020) conceptualizes micro-level impact as the accumulation of financial capital, livelihood assets, and capabilities. As argued by Heeks (2020, p. 2), the micro-level (individual) impacts of ICTenabled employment "represent very small drops in a very large development bucket." Despite the difficulty of linking micro-level impacts to macro-level transformations, the focus on digital work in recent reports by the Asian Development Bank (2021) and International Labour Organization (2021) illustrates the broader transformative effect expected from this type of employment. Viewing inclusive development as a process that occurs when social and material benefits are equitably distributed across social groups in society (Hickey et al., 2015), the question is how these impacts reach targeted groups or translate into broader community impacts.

| RESEARCH METHODOLOGY
This study is based on an inductive research approach (Bryman, 2012), which involved collecting qualitative data to understand how online freelancing and impact sourcing provide new employment opportunities. The explorative approach was deemed most appropriate given the lack of previous inquiry into both domains and limited knowledge on the accessibility of respondents.
In addition, to study the geographical distribution of the online freelancing sector across the Philippines, a spatial analysis (using ArcMap 10.1) was conducted among Filipino freelancers working via Upwork, the largest global digital labor platform. Data on the places of residence of Filipino freelancers involved in administrative support or web development were collected from Upwork profile pages. Based on probability sampling (Bryman, 2012), 459 online freelancers who had worked at least 100 h and were active on Upwork during the previous 6 months when the research was conducted in 2015 were included. Once having collected the data, GIS was used for geocoding. Shapefiles of the Philippines, derived from Global Administrative Areas, version 2.0, served as reference data (GADM, 2015).
Semi-structured interviews served to gain insight into the social profile of online freelancers and impact-sourcing workers in the Philippines.
The interviews provided insight into the actual workings of the sector, the (dis)advantages of this type of work and their social profile. Initially, the  (Bryman, 2012, p. 418, 424). These sampling techniques are considered appropriate when the researcher is experiencing difficulties accessing members of the target population (Babbie, 2011), such as for reaching online freelancers who generally work from home. Purposive sampling provides access to current as well as former online freelancers. Former freelancers were included to know the reasons for departure and new employment challenges and opportunities.
A total of 30 Filipino online freelancers working via Upwork were interviewed in August-September 2015 (Table 2). Likewise, purposive sampling (Bryman, 2012, p. 418) was used to ensure the variety of the impact-sourcing sample. For studying impact-sourcing workers, we gained access to the largest impact-sourcing initiative in the Philippines, which is based in Tanjay in Negros Oriental (Central Philippines). First, they were selected based on the amount of time they worked for this company. A preference was given to those who had been with the company for over a year, although the sample also includes some workers who were only recently offered a contract. Second, both workers who work during the daytime and night-time were included. Third, an effort was made to include impact workers from all three work units. Fourth, respondents were selected based on their willingness to contribute to the study. We interviewed 30 workers of a staff complement of 116. These interviews were complemented with four interviews with staff members of impact-sourcing ventures, which gave an understanding of the motivations behind the initiatives, their location choice, and the long-term prospects of their venture.
For informed consent (Bryman, 2012), guarding respondents' privacy, and to refrain from deception, all respondents were informed about the content and aim of the research before the interviews commenced. Through a participant consent form, they were reassured that their contribution was anonymous and strictly confidential. Moreover, through the consent form, they understood that they could stop the interview at any time and were free to decline answers to questions. Furthermore, they were informed that after the interview, they could withdraw their contribution without explanation. Respondents were explicitly asked for permission to record the interview. None of the respondents insisted on stopping the interview, declined to answer questions, or withdrew their contribution afterwards. All respondents allowed the interview to be recorded. The data management plan stipulated that the interview data was securely stored throughout the entire research cycle.
Interview data were inductively coded, deriving codes from the data and analyzing the interview transcripts iteratively. First-order codes included, for example, "location," "income," "level of education," "economic status," and "years of employment." A coding tree (see Figure 1) was developed to assign these first-order codes to axial codes (Saldaña, 2013) that represent the key dimensions of inclusive development (geographical spread, social inclusion, and long-term prospects).

| GEOGRAPHICAL SPREAD OF IMPACT SOURCING AND DIGITAL LABOR IN THE PHILIPPINES
The Philippines has long interested researchers as a "development puzzle" (Balisacan & Hill, 2002;Medella et al., 1995), particularly for its limited success in industrial development (Ofreneo, 2015) and the financial dependence on remittances of its overseas labor migrants (Boquet, 2017).
After decades of low overall economic growth relative to its neighboring countries, the country has benefited from the phenomenal growth of its information technology-enabled offshore service sector since the early 2000s. The presence of a well-educated workforce that is readily conversant in the English language has made the country one of the primary beneficiaries of the international relocation of tradable service activities (Kleibert, 2015;Lambregts et al., 2016). From a mere 3.000 workers in 2000 (Ofreneo, 2015), currently, an estimated 1.3 million workers are active in this sector (IBPAP, 2020), with voice-based services (call centers) and back-office services (data processing) being the largest subsectors.
For new college graduates, online service employment offers an alternative to overseas labor migration (Soriano & Cabañes, 2020), with around 70% of jobs situated in the greater Metro Manila area (see IBPAP, 2020). However, smaller provincial cities have also become part of international networks in service delivery (Kleibert, 2015).
The emergence of impact-sourcing initiatives and the popularity of global digital labor platforms among Filipinos cannot be separated from the rise of the offshore service sector. The skills required for work in back offices or call centers resemble those necessary for work via digital tiative. The global provider encouraged a local BPO firm located in Makati (Metro Manila) to establish a "rural BPO," guaranteeing that it would outsource work to the company. As a potential place for the location, they required one that had high-quality electricity, telecommunications, and Internet infrastructure. Moreover, they insisted on proximity to colleges and universities to serve as feeder pools for workers rather than targeting more disadvantaged groups. In line with the CSV approach, their representative motivated their decisions as follows.
We really wanted a location that was farther out [where] they [referring to highly-educated inhabitants] had no means of being gainfully employed in their area as much as they would have to go someplace else. (respondent #36, October 3, 2015) Although many smaller provincial cities would have met these locational requirements, personal motivation eventually led to the selection of Tanjay, as it was the hometown of one of the initiators. Tanjay is a provincial city of approximately 80,000 inhabitants (PSA, 2015) located 30 km from the provincial capital Dumaguete. Its primary sources of income include sugarcane, farming, and fisheries. Tanjay accommodates two colleges whose graduates tend to leave the community to find jobs elsewhere in the Philippines or abroad because of the lack of local employment opportunities (respondent #36, October 3, 2015). The choice of Tanjay aligns with the location choices of impact-sourcing ventures studied by Madon and Sharanappa (2013), Malik et al. (2016), and Heeks et al. (2020).
For respondents, the primary motivation to work for the company was that it provided them with the opportunity to work in a BPO, earn a relatively high income and live close to their families. According to one respondent, Impact-sourcing ventures are restricted in their geographic outreach, as digital connectivity and availability of human resources are critical for their establishment. In other words, as expressed by the business representatives, the location cannot be "too rural" to have a viable business model. The same ambiguousness was observed with advertising the initiative as an impact-sourcing venture, as being too rural would raise doubts on the quality of service delivery. Therefore, the two critical development dimensions of the initiative, namely, where it is located and who the beneficiaries are, is largely ignored when presenting the business case to clients.
Most studies on online freelancers have been limited to the international distribution of freelancers ILO, 2021). Figure 2 indicates the area of residence of our sample of 459 online freelancers within the Philippines. Interestingly, less than half of the sample (221 cases) is based in one of the traditional offshore outsourcing hubs in the Philippines, which demonstrates that online freelancing has a much more fine-grained distribution over the country than conventional offshore service delivery. Access to (reliable) Internet is a prerequisite, as the two islands notorious for unreliable Internet (Palawan and Mindoro) had no freelancers in the sample.
Like the mainstream offshore service industry (Kleibert, 2015), the more knowledge-intensive jobs were concentrated in Metro Manila, where a larger pool of higher educated labor is available. Web, mobile, and software development demands higher-skilled labor than administrative support, and a larger number of them are based in the country's main urban areas (Table 3).
Reliable internet connections and electricity provision were of key importance for interviewed freelancers. Respondents expressed that due to the fierce competition on digital labor platforms, they must be online every day or miss out on job opportunities. In some cases, this even meant that respondents moved to larger towns where stable Internet would be better guaranteed. According to one respondent:  (2015), classified as rural, the unstable internet connection and electricity provision in these areas is considered a complicating factor as it reduces the chances of acquiring online jobs. Moreover, it imposes limitations on the type of jobs one can conduct. As a result, online freelancing broadly functions as a spin-off model of the mainstream service outsourcing industry wherein internet connectivity and electricity supply play a key role in determining the location of BPO offices (Kleibert, 2015).

| WORKERS IN ONLINE FREELANCING AND IMPACT SOURCING: PROFILE AND LONG-TERM CAREER PERSPECTIVES
This section examines the profile of online freelancers and impact-sourcing workers and their longer-term perspectives for work in this sector.
Compared with workers in the mainstream offshore service sector, this gives an understanding of whether new beneficiaries are reachable. A central question for assessing the inclusive development potential of online freelancing and impact sourcing is whether it is accessible to workers who fall outside the traditional scope of the offshore service sector. The mainstream service offshore sector primarily employs young college graduates from a middle-class background (Beerepoot & Hendriks, 2013;Marasigan, 2016).
F I G U R E 2 Place of residence online freelancers (N = 459)

| Profile
The interviews with impact-sourcing workers and online freelancers began focusing on their socioeconomic background, employment history and number of years in the sector (see Table 4). In line with other studies that included profiling online freelancers (Berg et al., 2018;D'Cruz & Noronha, 2016;, many had a college education, which is a prerequisite when dealing with foreign clients in the English language. The educational backgrounds of impact-sourcing workers do not differ much from online freelancers, as most have a college education. The advanced levels of education of both groups indicate, in Heeks' (2013) terminology, that these workers could only be considered poor when a broad, ambiguous definition of the base of the pyramid is used. Although some respondents stated that their family has gone through financial hardships, none of them conveyed that they had grown up in extreme poverty, which was in line with their perceived middle-class status. These observations confirm the Everest Group (2014) view that impact sourcing is a means to tap into the educated, unemployed youth market. The respondents viewed their office as just another mainstream BPO service provider. They were not familiar with the term "impact sourcing," nor were they aware of their part in a multinational firm's CSR initiative.
It is telling that 14 online freelancers and 25 impact-sourcing workers had previously worked as a service provider in the mainstream BPO industry. On average, the impact-sourcing workers had worked as mainstream service providers for 2 years and 9 months and, content-wise, did not consider their current job any different from their previous work. Nine impact-sourcing workers also had experience in online freelancing. This shows a clear overlap between the three domains of digital service delivery, for which experience in one domain helps for employment in the other. Online freelancers, especially, considered working in the mainstream BPO sector as a valuable experience. One online freelancer commented: The best thing to prepare for this kind of online job is to work first in the BPO industry. That's the best training ground that you can have. (respondent #29, September 29, 2015) When working in the mainstream BPO industry, freelancers improved their English language skills, were taught a variety of English accents, received classes on British, American, and Australian culture, and learned about customer service. Consequently, one of the respondents, the owner of one of the largest agencies in the online freelancing sector in the Philippines, only hires freelancers who have at least 2 years of work experience in a mainstream BPO.
However, the advantages of freelancing work were not enough to retain workers who quit online freelancing for various reasons. These reasons included being lonely working at home, not being entitled to health or social security benefits, or being unable to turn working online into a stable career, generating enough income.

| Long-term perspectives
Considering the advantages and disadvantages of both types of employment (see Table 5) is necessary to examine how workers perceive the long-term employment perspectives for assessing the inclusive development potential of both types of work. Mainstream BPO work is commonly considered to offer limited opportunities for career advancement (Beerepoot & Hendriks, 2013;Thite & Russell, 2010). The flat organizational structure of many firms and lack of variation in tasks would impede long-term employment opportunities.
Furthermore, respondents highlighted how age discrimination is a common practice in the Philippine labor market.
Here some respondents highlighted the advantage of working online. As one respondent optimistically claimed: Being an online freelancer, there's no discrimination when it comes to age. As long as you can click the mouse and you can still see your screen. (respondent #20, September 23, 2015) Similarly, others mentioned: If you know the tricks and trades of freelancing, it is somehow a stable job. (respondent #6, September 1, 2015) T A B L E 5 Work-related advantages and disadvantages

Impact sourcing Online freelancing
Work-related advantages -Work close home: not live far from home or even abroad (16); lower expenses on transport, food (and housing) (13); spend more time with family and friends (12) -Opportunities for upward career development (14) -Pleasant work environment (12) -Flexibility (11) -Wages are relatively high (11) -Spend more time with family (10) Work-related disadvantages -Wages are too low (11) -Fierce competition (9) -Work during the night-time (9) -Work in solitude (9) -Lack of passion for the job (5) -Workflow is unstable (8) -Jobs in the BPO industry are unstable (4) -Maintain a good work-life balance (7) T A B L E 4 Characteristics of impact sourcing workers and online freelancers Contrary to this optimism, many freelancers highlighted the everyday challenges of online work (D'Cruz & Noronha, 2016;, such as lack of social security, a distorted work-life balance, work in solitude and self-exploitation.
The average number of online hours worked, and per-hour income varies widely among freelancers. Whereas some reported working for 45 hours per week and earning $2.25 per hour, others said they work 60-80 hours per week for $16 per hour. One-third of the freelancers mentioned that the wages for online work are relatively high. As stated by one respondent: I had heard about the difference in net income compared to work in a call center or a day job in the Philippines. The net income is far more than I was getting before, even though it's just an entry position, like basic data entry or a coder, compared to a supervisor or a manager in a call center. (respondent #6, September 1, 2015) A notable challenge faced by online freelancers is that their job is not considered a "real job" (respondent #29, September 29, 2015;respondent #34, October 26, 2015;respondent #35, November 11, 2015;respondent #62, November 13, 2015). Two respondents who work online full-time were contemplating replacing freelancing with a full-time BPO job. As one of them said: If you work online, it won't be part of your resume. You won't have any certificate or something that'll prove that you've worked The impact-sourcing respondents work for 40 h per week, with gross monthly earnings ranging from 10,000 ($212) to 17,000 pesos ($360).
They explained that they earn less working in Tanjay than working in one of the bigger cities in the Philippines. However, working close to home saves on commuting, food, and housing (Table 5). Turnover of employees in the impact-sourcing unit is annually only 3%-5%, which is much lower than the 20%-80% range experienced in the mainstream BPO sector (Marasigan, 2016). As the managerial respondent stated: If they want to work in Tanjay Only 11 impact-sourcing respondents see themselves working in their current job for a longer-term. The job challenges include night-time employment, the routinized nature of work and the lack of career prospects (Table 5). Like the mainstream offshore service sector (Kleibert, 2015), impact sourcing does not overturn the historically rooted and dominant mindset wherein workers prefer employment overseas over working in the Philippines. Six respondents expressed the ambition to work abroad. One respondent said: I have two kids and one is already in preschool. I'm thinking about their future. That's why it came to my mind to go to Dubai and work there. (respondent #49, October 24, 2015) This is in line with the view of BPO work as a stage in their career and a stepping-stone to obtaining a job abroad (Beerepoot & Hendriks, 2013) while helping workers improve their language and communication skills. that rather than expanding the reach of ICT-enabled service work as a source of income generation, the existing BPO labor force benefits most from the new economic opportunities. There is a substantial overlap between the three modes of service delivery (mainstream BPO, online freelancing, and impact sourcing). The latter two are mainly derivatives of the first, and workers easily switch between them. Nevertheless, both models lead to a more fine-grained spread of activities and the inclusion of more peripheral places in networks of online service delivery leading to new income-earning opportunities in these places.
Both models have their limitations for contributing to inclusive development, as shown in Table 6, which calls for the investigation of hybrid forms of service delivery that combine the strengths of both models. In some developing countries, there are active policies to train young people for online work via digital labor platforms (Melia, 2020), bypassing impact-sourcing initiatives that face difficulties scaling up their operations. The question is whether such training can suffice for successful participation in a labor market with global competition. Although online freelancing can provide substantial financial gains at the individual level, its main limitation is the highly uneven distribution of gains among freelancers and lack of social security or secondary benefits (Berg et al., 2018;Lehdonvirta et al., 2019). A hybrid of both models would utilize the scale and easy mediation through digital labor platforms to connect shared value-oriented clients to online workers who receive fair compensation for their effort. Even with the emergence of hybrid models, the inclusive development potential of online labor at a community level (thus beyond individual financial gains) remains a subject for debate.
The requirement for international service delivery to communicate in English is an entry barrier for poorer (often more remote) communities to engage in this type of work, meaning that restrictions on who benefits from this type of employment will continue. Despite ambitions to CSV, the commercial logic behind impact-sourcing initiatives (Kannothra et al., 2018;Nicholson et al., 2016;Sandeep & Ravishankar, 2018) has led to development impacts that are often a by-product of profitable business (Arora, 2019). The result is that broader community impacts of both types of employment seldom get beyond anecdotal evidence of new restaurants and other local establishments that are developed as a result of the location of an impact-sourcing unit. Herewith the longstanding challenge within research on ICT4D of establishing a link between micro-and macro-level impacts (or longer-term impacts, see Madon & Ranjini, 2019) remains open for investigation. Policymakers have more general concerns about the development potential of offshore service-sector driven or the transformative potential of ICTs (Kleibert & Mann, 2020;Lambregts et al., 2016). Mainstream BPO and its derivatives online freelancing and impact sourcing can successfully provide new employment opportunities. However, the dynamics in online service delivery (such as automation, reshoring, and unbundling) (Baldwin, 2019) will mean ongoing challenges regarding where and who gains from them.

ACKNOWLEDGMENTS
This article benefitted from participation in the Development Implications of Digital Economies (DIODE) Strategic Research Network led by Richard Heeks. Furthermore, we would like to thank Elvis Melia, three reviewers and Malcolm Garbutt as overseeing editor for critical comments on a previous version of this article.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the authors. The data are not publicly available due to privacy and ethical restrictions.