Emotional Labour and the Autonomy of Dependent Self-Employed Workers: The Limitations of Digital Managerial Control in the Home Credit Sector

Abstract

Changes to the labour process in the home credit sector have exposed the industry’s agency workforce to increased levels of digital managerial control through the introduction of lending applications and algorithmic decision-making techniques. This article highlights the heterogeneous nature of the impact of digitalisation on the labour process and worker autonomy – specifically, in terms of workers’ engagement in unquantified emotional labour. By considering the limitations of digital control in relation to qualitative elements of the labour process, it becomes evident that emotional labour has the scope to be a source of autonomy for dependent self-employed workers when set against a backdrop of heightened digital control. This article therefore contributes to ongoing labour process debates surrounding digitalisation, quantified workers and digital managerial control.

Introduction: Home credit and the work of collection agents

During the financial year 2018–19, 22% of the UK population received below 60% of the national median income, placing 14.5 million people within a low-income bracket (Gov.uk, 2020). Individuals living in low-income households often have poor credit scores and are therefore unable to access mainstream financial services, leading many to turn to alternative, sub-prime, high-cost credit options (Financial Conduct Authority (FCA), 2017; Leyshon et al., 2006). Home credit (also known as home-collected credit or doorstep lending) is one form of high-cost consumer finance, typically using sub-contracted workers, commonly known as ‘agents’ or ‘collection agents’, who offer small loans on behalf of a lending company and then collect regular ‘manageable’ repayment amounts from borrowers’ homes over an agreed number of weeks. In 2016, the mean home credit loan value was £770, with the home credit borrower having an average income of £15,500 and a credit score of 41 out of 100 – a credit score below 50 is considered ‘sub-prime’ (FCA, 2017: 38).

Such high-cost consumer lending models are established worldwide, with areas of operation including the United States (Prager, 2014), the Republic of Ireland (Byrne et al., 2007) and post-communist European states (Burton, 2017). The home credit model also exists in India, Mexico and some South American states. However, home credit has a particularly rich history in the United Kingdom, especially within established working-class communities. Home credit in the UK can be traced back to 1880, when Joshua Kelley Waddilove sought to provide affordable credit to low-income families in West Yorkshire – founding Provident Clothing and Supply Co. Ltd. In the UK, most home credit provision and borrowing can be found in the post-industrial heartlands, including areas in the North of England and the central belt of Scotland. Since 2014, authorised home credit companies in the UK have been regulated by the Financial Conduct Authority (FCA), with the FCA’s latest figures estimating that 1.6 million people in the UK have outstanding home credit debt (FCA, 2017).

Within the myriad of high-cost consumer credit options, including payday loans, catalogue credit, retail finance, rent-to-own schemes, logbook loans and guarantor loans, agent home visits are unique to the home credit model of lending. However, the home-visit element of the home credit model often leads to the conflation of home credit agents’ work with the work of debt collectors and bailiffs (Collard and Kempson, 2005). While agents do collect repayments from borrowers, their role also involves selling and service provision, differing significantly to the work of debt collectors. Moreover, agent relationships with home credit borrowers commonly extend beyond one initial loan, meaning it is in the agents’ commercial interest to facilitate positive service interactions with their borrowers (Leyshon et al., 2006; Rowlingson, 1994). Previous research into home credit suggests that borrowers value the relationship with their agent and are often reluctant to switch lending company, although, when coupled with the takeover of many small home credit companies by four big market players in recent years, this may limit consumer choice (Brooker and Whyley, 2005; Collard and Kempson, 2005).

Over the last decade there has been limited empirical research focused on the home credit sector despite considerable changes in the market, including the appointment of the FCA as the industry’s regulatory body and significant digitalisation. Further, there is a dearth of literature on the home credit labour process and the work of its (dependent) self-employed workforce. This article provides exploration of home credit, focusing on home credit agents’ engagement in emotional labour and the significance of these work activities within the labour process. In contrast to Hochschild’s (1983) debt collectors, who adopt a guarded and depersonalised approach during interactions with debtors, extant literature suggests that home credit agents for the most part engage in friendly interactions with borrowers (e.g. Byrne et al., 2007; Rowlingson, 1994). It is acknowledged that home credit itself is problematic, particularly in terms of the cost of credit and related consumer issues (see Appleyard et al., 2016; FCA, 2017), but these matters are beyond the direct scope of this article, which focuses on the experiences of the industry’s often overlooked workforce.

This article also explores the lending company management’s use of digital control methods and the impact on agents’ labour process. The introduction of a digital application, onto which agents must record much of their work activity, is a key feature of managerial digital control over agents in this case. Further, decision-making regarding borrowers and lending – formerly heavily influenced by agents – is now largely made by algorithmic calculations of affordability. Yet, not all work activities can be quantified or automated within digital systems, given their complex, contextual constructions and lack of universal solutions (Autor, 2015; Pettersen, 2018). Thus, the role of dependent self-employed workers’ emotional labour is examined alongside an exploration of the viability of the human agent within an increasingly digitalised home credit model.

This article contributes to ongoing labour process debates surrounding algorithmic automation, the quantification of workers and digital managerial control (e.g. Veen et al., 2019; Wood et al., 2019). The article reports on the findings from interviews with 75 individuals working in the home credit sector and highlights the heterogeneous nature of the impact of digitalisation on work activities and worker autonomy, specifically, workers’ engagement in unquantified emotional labour. This current work also questions whether emotional labour, frequently cited as a potentially harmful practice for workers (Diefendorff and Richard, 2003), has the scope to be a source of autonomy when set against a backdrop of heightened digital managerial control mechanisms.

Digital managerial control in dependent self-employment

Historically, the role of technology within the labour process is central to both Marx and Braverman’s (1974) explanations of the degradation of work and the power of capitalism over the subjugation of worker autonomy. From a Marxist position, the development of capitalism has been driven by technological expansion, with technology evolving to increasingly exploit, enslave and alienate workers (Spencer, 2018). Yet, this arguably partial perspective fails to acknowledge the fact that emergent digital systems are both differentially deployed and differentially experienced by workers, depending on the sector and employment conditions (Rosenblat and Stark, 2016).

In particular, those that work under precarious conditions are most likely to be subject to exploitation at the hands of technology. For the purpose of cost savings and efficiency, several sectors have generated precarity by replacing permanent formal employment with fixed-term, temporary or (in)dependently contracted (self-employed) work; for example, in direct selling (Moisander et al., 2018), parcel delivery (Moore and Newsome, 2018) and fitness centres (Harvey et al., 2017). Traditional self-employment typically refers to freelance workers with multiple clients, strategic input and/or management responsibility (Moore and Newsome, 2018). By contrast, dependent self-employed workers are usually contracted to work exclusively for one firm on a long-term basis, which creates high levels of both economic and personal dependence on the contracting organisation (Muehlberger, 2007).

Extant research suggests that dependent self-employed workers are increasingly constrained by heightened managerial surveillance and control over work allocation and practice, resulting in a reduction of individual worker autonomy (Moore and Newsome, 2018). It is often the case that contracting organisations closely monitor workers, not only through meetings with supervisors, but increasingly via digital and algorithmic technologies (Rosenblat and Stark, 2016; Wood et al., 2019). These technologies enable management to control working practices, homogenise work and conduct surveillance of the workforce (Evans and Kitchin, 2018; Howcroft and Bergvall-Kåreborn, 2019). Workers across industries are targeted for quantification by managerial forces in pursuit of precise control mechanisms and value extraction (Moore, 2018), meaning the scope for any dependent self-employed workforce to have meaningful levels of autonomy is considered highly questionable (Harvey et al., 2017).

Current writing on dependent self-employment focuses on gig or platform work, where people undertake work that is mediated via digital platforms or apps, which includes the well-cited examples of Uber drivers, Helpling cleaners and Deliveroo agents. Such models ensure that there is a strong degree of reliance by the worker on the contracting company for the production of work opportunities. While some meaningful comparisons between the modern home credit agent and gig workers can be made, there are also important distinctions that make the discussion of home credit agents – as a unique group of workers – both important and interesting. Unlike most gig or platform workers, home credit agents are tied to a particular organisation and unable to undertake work for multiple credit providers. Also distinguishing home credit operators from the majority of gig workers (and even some workers in traditional forms of self-employment), agents cultivate lasting relationships with borrowers through regular interactions over prolonged periods of time (see Appleyard et al., 2016; Leyshon et al., 2006). Though there is a certain similarity between the agents’ mode of dependent self-employment and the typical gig worker, in terms of heightened organisational input resulting from the introduction of algorithmic controls, there is one important difference. With a century’s worth of history of human interaction in the home credit sector, it may be a step too far to replace a person-dominated relationship with a digitally led interaction.

Emotional labour: Unquantified work(?)

An enduring and distinctive feature of the home credit model of lending is the strength of relationships between agents and borrowers, built on foundations of commonality, community and shared identity which often results in ‘friendliness’ and a high degree of trust between the two parties (e.g. Byrne et al., 2007; Collard and Kempson, 2005). The need for agents to manage their interactions with borrowers in an affable manner, to ensure regular repayments and the retainment of reliable borrowers has been well documented (Leyshon et al., 2006). Further, agents have traditionally held a significant degree of autonomy over lending decisions, utilising a combination of intuition and past experiences to assess the creditworthiness of potential borrowers and the likelihood of repayment (Byrne et al., 2007; Kempson et al., 2009; Leyshon et al., 2006). As with other groups of predominantly service-led occupations, responsibility for the generation of revenue income frequently places significant emotional demands on home credit agents (c.f. Ikeler, 2016; Jenkins et al., 2010).

Emotional labour, a concept developed by Hochschild (1983), describes forms of work where workers are required, either explicitly or implicitly, to manage their outward emotional displays. Emotional labour differs from ‘emotion work’, which is seen to be undertaken by individuals during all forms of social interaction (Cohen, 2010). Emotional labour on the other hand is the commodified version of emotion work, undertaken by workers in the labour market in exchange for payment (Cohen, 2010; Hochschild, 1983). Hochschild (1983: 147) outlines three features of ‘jobs that call for emotional labour’, all of which characterise (to varying degrees) the role of home credit agents:

First, they require face-to-face or voice-to-voice contact with the public. Second, they require the worker to produce an emotional state in another person – gratitude or fear, for example. Third, they allow the employer, through training and supervision, to exercise a degree of control over the emotional activities . . . [of workers].

Hochschild (1983) argues that there are ‘feeling rules’ attached to jobs that require emotional labour. Feeling rules are ‘guidelines for the assessment of fits and misfits between feeling and situation’; these rules ‘tend to be latent and resistant to formal codification’ (Hochschild, 1979: 566). Hochschild (1983: 137) posits that within service provision, workers perform as either the ‘toe’ or the ‘heel’; those who deliver the service act as the ‘toe’, while those who collect payment for services act as the ‘heel’. The feeling rules attached to these seemingly polarised functions are presented as a dichotomy by Hochschild (1983), who explains that flight attendants are expected to feel trust and goodwill, while the debt collector is expected to feel distrust and sometimes bad will.

However, unlike Hochschild’s flight attendants and debt collectors, who operate as the ‘toe’/service provider and ‘heel’/payment collector, respectively, the home credit agent’s job encompasses both service delivery and payment collection roles – meaning agents may have to navigate a complex array of feeling rules. Workers may induce or inhibit certain feelings or emotional displays dependent on the social norms – or feeling rules – of their situation (Hochschild, 1979). Throughout the working day, an individual may deploy different emotional displays, either simultaneously or in combination (Bolton, 2000; Gabriel and Diefendorff, 2015). Hochschild (1979: 565–566) explains that, similarly to the rules of social interaction in general, there is room within the ‘zone’ of a feeling rule for ‘motion and play’, which implies scope for actors’ interpretation.

Many organisations seek to ‘suppress, hide or manage’ workers’ feelings in line with organisational goals (Bolton, 2000: 156). In some sectors, management interference has diminished service workers’ autonomy by reducing the required complexity of their emotional labour (e.g. Curley and Royle, 2013; Ikeler, 2016). Yet, Bolton (2000) argues that a narrow organisational focus on prescribed, profit-seeking forms of emotion management neglects recognition of the emotional labour that workers undertake for the benefit of themselves and others, including customers. Moreover, O’Donohoe and Turley (2006) suggest that emotional labour is incompatible with managerial attempts at systematisation and control. Allowing service workers a high level of autonomy over their emotional displays, with minimal managerial interference, may allow positive emotional labour to prosper (O’Donohoe and Turley, 2006). Management attempts to manipulate workers’ emotional labour may therefore be counterproductive to organisational goals (O’Donohoe and Turley, 2006).

Workers who interact with customers directly may, nevertheless, be pushed to engage in emotional labour by normative managerial control mechanisms that promote customer focus (see Gandini, 2019; Rosenblat and Stark, 2016; Veen et al., 2019). Brook (2009) stresses that emotional labour should not be detached from its exploitative nature and human cost. The extent to which emotional labour is detrimental (or not) to individual well-being is highly dependent on the social conditions of its performance (Humphrey et al., 2015). Service workers who supervise their own interactions with customers and related emotional labour face stronger demands on their emotional capacity (Wouters, 1989). These demands are particularly evident in the context of dependent self-employed work; for example, in direct selling, where workers are encouraged by organisations to ‘reconstitute themselves and their lives as “enterprises”’, in pursuit of individual autonomy (Moisander et al., 2018: 392).

Yet, writers have argued that these social and emotional interactions are not easily programmed into digital managerial systems (Autor, 2015; Pettersen, 2018). Activities that require workers to undertake complex problem-solving, apply common sense or interact with other people, present immense challenges to the application of algorithmic logics, given that tacit human judgement, emotions and feelings are near impossible to quantify (Autor, 2015; Frey and Osborne, 2017). As such, there is uncertainty surrounding the impact of digital controls on the labour process in sectors where workers engage in emotional, longer-term encounters.

There are a number of examples of work that may be useful in informing understanding of the emotional labour undertaken by home credit agents. On the face of it, the obvious parallel is with Hochschild’s (1983) work and her focus on debt collectors, yet – as previously noted – the debt collectors’ modus operandi was centred on an emotional display of intimidation, rather than the development of friendly relationships that is common in the work of home credit agents (Collard and Kempson, 2005; Leyshon et al., 2006). Perhaps more relevant, therefore, is Cohen’s (2010) work on hairdressers. As with the agent–client relationship, the stylist–customer relationship extends over decades and generations, and is distinct from the anonymous, distant interactions featured in work on emotions and the gig economy – typified by short-termism and the fungibility of workers (Flanagan, 2019).

Yet, where the role of home credit agents clearly differs from that of hairdressers is in the nature of the labour process, in particular the rapid introduction of digital technologies. The central focus of this article is on the generation of an understanding of whether agents’ emotional investment in the borrower relationship is affected by the introduction of digital technology within a long-term, work-based emotional encounter.

Hence, the aim of this article is to understand the impact of digitalisation on emotional work activities as identified in the following research questions:

• RQ1: What impact does digitalisation have on the labour process and the autonomy of home credit agents?

• RQ2: What is the impact of heightened digital managerial control on the emotional labour and autonomy of home credit agents?

Methodology

This article is based on work from a larger project exploring the home credit industry and the experiences and practices of its workers, focusing on one of the key market competitors in the home credit industry. Semi-structured interviews were used to capture the experiences and perceptions of the interviewees. The interview questions were varied in terms of format, allowing discussion of participants’ work histories and exploration of the role of managers and agents, impact and experiences of digitalisation, relationships with borrowers and the broad demographic profiles of both the agents and the lending company’s customer base (home credit borrowers). All interviews were conducted face-to-face by the authors who adhered to semi-structured interview principles throughout (Creswell, 2014).

Access to participants was negotiated through company gatekeepers and all interviewees participated voluntarily. Interviews with agents took place in regional offices where employed managers are based. Agents meet with their line manager regularly to discuss sales, collections and other business matters; the authors interviewed on days when agents would normally visit the office to minimise disruption to participants’ usual working patterns. Most interviews were conducted with individual participants, although seven managers were interviewed in pairs or threes for expediency. In total, 71 semi-structured interviews were conducted between May and September 2018, with 75 participants working across 10 locations, split between the North of England and the central belt of Scotland. Interviews varied in length from 25 minutes to well over 100 minutes, averaging 62 minutes per interview. All interviews were audio-recorded and transcribed verbatim.

The majority of participants were aged between 30 and 60 years old, although there were some in their 20s and others post-retirement. Participants were predominantly white and British. Home credit is typically a female-dominated industry, particularly in terms of its collection agents (Rowlingson, 1994). The participants in this study included 40 women and 35 men, with women accounting for 30 out of the 43 agents interviewed, but only 10 of the 32 managers interviewed. Women accounted for over 65% of agents across the lending company as a whole between 2015 and 2018. Managers from all levels of the lending company’s management structure were interviewed, from local line managers to regional directors and board-level executives. Strikingly, responses to interview questions were broadly consistent across different levels of management, particularly in discussing the importance of agents’ relationships with borrowers. This may perhaps be attributed to many of the lending company management having experienced work as an agent prior to their appointment to managerial positions.

Detailed data analysis was completed using thematic analysis. Themes were largely generated inductively, informed by the interview data and secondary data analysis undertaken prior to, during and after data collection. NVivo 12.2 was used to organise codes. A set of codes were agreed upon between authors, having jointly reviewed field notes and interview transcripts, although some codes were added or altered over the course of the coding period (Saldaña, 2013).

The first-order codes were semantic and coarse (Braun and Clarke, 2006), describing broad areas of discussion from the interview data, including ‘industry developments’, ‘agent–customer relationships’ and ‘the labour process’. The second-order codes were more focused and captured implicit ideas within the data (Braun and Clarke, 2006). ‘The labour process’ was broken down into second-order codes including ‘work activities’, ‘digital technologies’ and ‘emotional labour’. Every code was allocated a description in NVivo that identified the ideas housed under each coding label. For example, ‘emotional labour’ included data evidencing agents’ emotional displays and the agents’ role in producing emotional states in borrowers. ‘Digital technologies’ included revelations regarding electronic device use, the role of digital applications and evidence of digital managerial control.

Once coding was complete, the authors analysed the coded data for patterned meanings and actively generated themes (Braun and Clarke, 2006). The authors identified several key themes including ‘heightened industry regulation’, ‘digitalisation of the labour process’ and ‘the role of agents’ emotional labour’. These themes are explored in the remainder of this article.

Digitalisation and algorithmic management in home credit

• RQ1: What impact does digitalisation have on the labour process and the autonomy of home credit agents?

The home credit agents participating in this study were paid commission on the loan repayments collected from borrowers (who participants referred to as ‘customers’). Agents therefore rely on the generation of new borrowers and new loans to ensure the viability of their collection books and personal income. However, the nature of agents’ work has become more challenging in recent years. Over the last decade in particular, there have been significant changes in the home credit market. With the appointment of the FCA as the new industry regulator in 2014, there was the introduction of a price-cap, which limits the fees that borrowers can be charged for failing to make repayments on time. Yet, according to the senior managers interviewed, the price-cap does not directly impact the organisation because all of the fees are fixed at the outset of each loan agreement. However, the FCA also focuses heavily on improving consumer protection through their regulatory adjustments, which have included the introduction of more stringent, compulsory affordability checks and restrictions on lenders’ advertising practices – changes which directly impact on home credit providers. Such changes were broadly supported by management in the company, as demonstrated in the quote:

… the industry’s changed quite a bit. So maybe in years gone by, we could have gone out on the basis of, ‘I know them, I know the family’ . . . I’m happy to support that agent knowledge and experience, but now the industry and environment’s totally different . . . So fair play to the Office of Fair Trading and the FCA. It’s up to us, as a business, to make sure that we do have our house in order, and we’re treating [borrowers] the right way. (Manager, male)

To allow for the closer monitoring of lending practices and the conduct of agents in line with FCA requirements, the home credit company introduced algorithmic managerial control into a key area of the labour process – the assessment of borrowers’ creditworthiness and subsequent credit lending decision-making – ostensibly in the name of regulatory compliance. Management discussed how the new technology helps ensure compliance with both FCA regulations and company-level lending policies:

… now everything is done [using] technology [because] everything has to be proven, so that if somebody comes to check they can see that you have adhered to policy. That is where the technology comes in. (Manager, male)

Interview participants reported that the lending company’s utilisation of a credit score and background checking facility, alongside the introduction of a digital application (referred to by participants as the ‘app’), reduces the input of the subjective judgements required from the agent during the loan application process. The basic organisation of agents’ work remains largely unchanged, including the mechanisms for the identification for new borrowers. However, potential borrowers can also make an initial application via the company website and are then distributed to agents who arrange a home visit to conduct affordability checks (using the ‘app’). Nonetheless, most borrowers are still identified through word-of-mouth recommendations via existing borrowers or through leafleting (Kempson et al., 2009). Owing to FCA regulations, agents are no longer able to generate new business through speculative ‘door knocking’; however, the ‘app’ produces new business by automatically generating lists of existing borrowers who potentially qualify for new loans.

The benefit of this borrower list is the relative certainty of the creditworthiness of the borrowers. Indeed, the homogenisation of lending decisions brought about through the introduction of the digital lending application makes the loan approval process for both new and returning borrowers far more stringent but reduces the input of the agent. Instead of filling in numerous forms, the agents now record relevant borrower information in the ‘app’ on a phone or tablet. Recorded information includes evidence of weekly income as well as likely outgoings to assess borrowers’ creditworthiness:

… you press the button, and if it says processed, that’s [the affordability checks] passed, but if [borrowers] do not have enough disposable income, it will decline it . . . It would not be me that was refusing it, it would be the app refusing. (Agent, female)

Company managers recognise that the introduction of automated algorithmic measures via the digital lending application has significantly limited agents’ autonomy over lending decisions. Agents also report that a refusal of further borrowing from the lending application, under circumstances where an individual borrower would previously have qualified for a loan, can cause tensions with existing borrowers. Further, the restrictions imposed through algorithmic lending decisions have limited the pool of potential new borrowers available to agents.

Despite the apparent reduction of agents’ autonomy, agents did cite examples of where they were able to challenge algorithmically generated lending decisions by lodging a request with their company line manager to review the decision. In reality, this practice is rarely successful when the initial decision is a refusal for borrowing, but occasionally a regular and reliable borrower is allowed to take out a further loan on the recommendation of the agent – although management interference is required.

Some agents view the digital recording of lending decisions – as well as the collection of supporting evidence – as a measure of protection for themselves and their borrowers against regulatory breaches. Most agents believe that the ‘app’ mitigates the risk of human error on the part of the agent. Agents argue that home credit borrowers are now less likely to become over-burdened by unaffordable loans:

… well, on affordability . . . income and outgoings, say if that customer’s wanted five [hundred] and I thought, ‘Oh right, yeah, yeah, you can have five [hundred]’, that will show me on the app that they are not quite ready for £500. It is showing [the agent] responsible lending, it is saying to you, look we can’t do that for that customer, but we possibly could a few months down the line. (Agent, female)

Similar to agents, company managers argued that the digital application allows for the capturing of more accurate and complete information about both existing and potential borrowers:

… obviously when it comes to the agent doing the lending, they’ve got to capture so much information. If it was a case of without the technology, without the apps, when it was done on paper, they’re just writing down what they see. There’s no proof there . . . Whereas with the app, there’s actually an image of the wage slip or the benefit letter . . . So, it’s definitely better [for ensuring regulatory compliance]. (Manager, female)

For many agents, the digital loan approval system provides a buffer between them and borrowers by depersonalising elements of the lending process that require a judgement on borrowers’ creditworthiness and affordability. Agents indicated that they are able to inform borrowers that it is the technology that says ‘no’ rather than the agent’s personal judgement, in a bid to protect the carefully cultivated agent–borrower relationships. Nonetheless, the depersonalisation of lending decisions does not constitute a replacement for the human or the emotional element of agent–borrower interactions; if anything, agents implied that the company’s automated lending decisions have made the agent appear to be the ‘good cop’, even when loans were refused.

Unquantifiable work? Emotional labour and the limits of digital managerial control

• RQ2: What is the impact of heightened digital managerial control on the emotional labour and autonomy of home credit agents?

While the ‘app’ is valued by managers and agents, both parties also explained that agents’ knowledge of their borrowers’ circumstances is central in enabling both commercially sound and socially responsible decision-making. As one manager noted:

A lot of the agents seem to know what the customer can afford and because they are in and out of the house, they know their incomes and outgoings. (Manager, female)

There are clearly elements of the credit agents’ work that cannot be replicated or replaced by a digital application, specifically the awareness of domestic circumstances and values generated from the personal relationships developed between agents and their borrowers. Thus, effective agents employ emotional labour to gather knowledge about borrowers. Both agents and managers provide clear illustrations of the ways in which an agent’s relationship with a borrower can generate important insights that might otherwise be missed by the lending company and result in a borrower falling into arrears:

I would refuse a loan if I thought it wasn’t within [the borrower’s] capabilities to pay it, or if they were overstretching themselves. You get to know [them], customers tell you all their business. Because you’re in their house week in, week out, and you build up a close relationship with them. (Agent, female)

Relatedly, agents and managers discussed the importance of agents gaining and maintaining the confidence of their borrowers. Agents endeavour to influence the emotional state of their borrowers (see Hochschild, 1983: 147) to generate feelings of liking and trust, demonstrating how agents’ emotional labour plays a key role in the development of positive relationships between agents and borrowers:

It’s like any relationship, it takes time for [a customer] to trust you and for me too . . . to trust my customer . . . so going in once a week and having a chat off the record . . . I take the customer’s payment but then we like to chat about other things, you know? . . . I go into a customer’s house that I’ve been going into for 10 years, she has my lunch ready for me . . . that’s the kind of relationship that we’ve got. (Agent, female)

It is evident that the scope of some of the relationships between agents and their borrowers goes beyond the direct provision of home credit, with agents displaying compassion and kindness in choosing to provide support to borrowers beyond financial matters:

I would go to the chemist for people who couldn’t get out or going [sic] messages for people that are bedridden and things like that. (Agent, male)

… there is a real kind of bond there . . . they pick up prescriptions for them, they put up curtains . . . so they have, [agents] are part of the family on a lot of the occasions . . . that’s the fabulous part of it all. (Manager, female)

However, Leyshon et al. (2006: 177) argue that it would be a mistake to regard the relationships between agents and their borrowers as friendships, arguing instead that any ‘friendliness’ displayed by agents should be interpreted as ‘skilful interpersonal conduct’ grounded in commercial concerns. The data clearly indicate that successful emotion management on the part of agents can have commercial benefits for both the agent and the lending company. Yet, whatever the motivation behind agents’ emotion management, the results (whether commercial or otherwise) of this emotional labour cannot solely be accounted for by the digital application used in agents’ work activities. This mismatch between technology and the emotional nature of the job is evident from discussions with managers and agents about the initial lending process, but is perhaps most evident when participants discussed the agents’ role in repayment collections.

Although repayment amounts are tracked via the digital application and monitored by company management, the agents’ collection visits are otherwise outside the realm of digital managerial controls. Much of the activity within these visits constitutes emotional labour. Agents often reported spending extended periods of time engaging with borrowers, producing feelings of companionship:

… [borrowers] like the idea of you going to their house and talking to them, the older ones maybe don’t see anybody from day to day and they are quite happy for you to have a wee chat. I don’t just run in and run oot [sic]. My partner says if I kept my mouth shut, I would get on a lot better. I would be [home] a lot earlier . . . (Agent, female)

Agents often utilise emotional labour during collection rounds, particularly when borrowers are struggling to make repayments. However, in contrast to Hochschild’s (1983) debt collectors, who attempted to intimidate and pressure debtors into making repayments, the home credit agents and managers interviewed argue that more sympathetic emotional displays are widely adopted within the sector. This is what Hochschild (1983: 141) terms ‘the soft collect’: where borrowers are offered ‘the benefit of the doubt’, deploying understanding, tolerance and goodwill:

Even though [the borrower is] on a reduced payment, I can still get on with them well, kind of thing, because I know they’re going through a difficult time. (Agent, male)

Sometimes the customers can be more embarrassed if they’re missing a payment. I’ve had a couple of people that don’t want to [tell me], and I’m like, that, ‘Don’t be silly. Look, can you not make the payment this week? What’s up? Tell me. Don’t ever be scared to tell me.’ (Agent, female)

Agent visits are widely recognised as one of the defining features of home credit (Collard and Kempson, 2005; Rowlingson, 1994). It was suggested by interview participants that the strength of agent–borrower relationships facilitates higher rates of repayment than would be achieved without the input of agents and their engagement in emotional labour. Arguably, therefore, agents’ qualitative labour power – particularly their emotional labour – remains valuable, despite digitalisation, and is key to the maintenance of positive customer relations, which are sometimes threatened by negative lending decisions produced by the digital application:

… a lot of the time [the agent] can say [to the borrower . . .]: ‘Look, you’ve failed with [the app] but we could give you this or if you wait X amount of time this will be lower’. Again, you’ve got the, the human touch. (Manager, female)

Further, the highly social nature of agent–customer interactions as reported by participants cannot easily be programmed into digital systems (Pettersen, 2018). Tacit human knowledge, judgement and emotion remain beyond the scope of algorithmic logics (Autor, 2015; Frey and Osborne, 2017). It follows then – for the moment at least – that interpretation of complex feeling rules and the sympathetic management of relationships remain the realm of human agents in the home credit industry. This highlights one of the distinct limitations of digital systems that are designed to quantify and control workers, but which cannot account for the nuances of emotional labour (see Moore, 2018; Moore and Robinson, 2016).

Further, although Hochschild (1983) argues that appropriate emotional training and monitoring by employers is characteristic of jobs involving emotional labour, not all of the agents’ emotion management resulted from direct organisational supervision. Within this study, there were some agents who go ‘above and beyond’ the level of engagement in emotional labour required to satisfy the basic commercial dimensions of their relationships with borrowers. Descriptions from agents and managers support the findings of Byrne et al. (2007), who found that home credit agents become ‘family friends’ and are kept abreast of special occasions. As one participant from the study illustrated:

… [people] think we’re there to, sort of like a debt collector so to speak, but we’re not there, we’re actually part of the family . . . I get invited to weddings . . . you get invited to funerals, you get invited to christenings, you get invited to people’s graduations – because you’ve known these people for a long time, you know what I mean? So, [borrowers] treat you as part of the family, so they see you as a friend, a friend that can give you money [laughs]. (Agent, male)

Wouters (1989: 101) argues that workers who supervise their own emotional labour ‘obviously have jobs which put stronger demands on their emotion management’. It is clear that emotionally demanding work can be damaging to workers (Diefendorff and Richard, 2003; Hochschild, 1983). Yet, Wouters (1989) argues that preoccupation with the potential costs of emotional labour can limit understanding of the joy that jobs may bring to workers. Most agents view their relationships with borrowers as being genuine, thereby mitigating some of the potentially negative consequences of their undertaking of emotional labour (see Humphrey et al., 2015). Further, agent approaches to borrower interactions are varied, which suggests there are differing agent interpretations of the job’s latent ‘feeling rules’ around the soft collection approach. Agents hold a degree of individual autonomy over emotional elements of the labour process that fall outside of organisational prescription (Jenkins et al., 2010). One agent explained the variability in the nature of her relationships with her borrowers:

[There are some borrowers] that I don’t try and make conversation with, like, cos, I just think, errh, no [. . .] but the ones that I have brought over meself . . . or friends or family, whatever – my relationship with them, they’re absolutely brilliant. (Agent, female)

Agents’ discretion over the nature of their relationships with borrowers is partly a result of agents’ judgement (and expertise) over how and when to engage in emotional labour effectively. Effective judgement demonstrates agents’ maintenance of their autonomy over the emotional, human aspects of the home credit labour process. The role played by agents within the lending process is considered to be home credit’s distinguishing feature and ‘unique selling point’ within the high-cost consumer credit market (see Leyshon et al., 2006; Rowlingson, 1994). Further, participants reported that home credit borrowers usually build a stronger relationship with their agent than with the lending company, which sometimes culminates in a borrower following their agent to an alternative home credit provider when the agent makes a transition. Such reports serve to highlight the reliance of the lending company on their agents’ emotional labour in terms of customer retention levels:

… at the end of the day the customer doesn’t really . . . care about where the loan’s coming from . . . it’s who’s chapping the door to come and get it. So, for 10 years it’s always been me, so they’re happy with that. So, [. . .] this is an example – when I moved away from [different home credit company], and then new agents went into my old customers’ houses, you know, they didn’t pay them. They weren’t happy, they’re not interested. ‘I want [agent’s name]’, you know, because of that relationship that we’ve had over the years, just building up the trust. (Agent, female)

It is evident that home credit managers do not believe that directly monitoring and controlling qualitative elements of the labour process is desirable or compatible with achieving organisational goals (c.f. O’Donohoe and Turley, 2006). While Brook (2009) stresses that emotional labour is an exploitative feature of organisations’ search for improved performance through labour commodification, the findings from this study suggest that it may be more productive for managers to allow workers discretion over their emotional labour (O’Donohoe and Turley, 2006). Managerial control and organisational norms may assemble parameters of ‘acceptable’ emotion work, yet, it is ultimately the workers who calibrate the emotional agenda (Bolton and Boyd, 2003).

Conclusion

This article contributes to ongoing debate surrounding the complexities of the impact of digitalisation on the labour process by providing insight into the dynamics of individual worker autonomy and managerial control experienced by agents working in the home credit industry. The findings from this study demonstrate that particular work activities – namely those involving emotional labour that necessitate the interpretation of complex feeling rules through tacit human judgement – are largely incompatible with algorithmic logics and therefore present challenges to digital managerial control systems (Autor, 2015; Frey and Osborne, 2017; c.f. Jenkins et al., 2010). Indeed, the reach of digital controls appears to be limited when workers’ engagement in emotional labour represents a salient constitutive element of the labour process. Discretion over engagement in emotional labour can be a source of autonomy for workers when set against a background of enhanced digital managerial control. Further, the inherently social and emotional nature of home credit agents’ work means that complete automation of the agents’ role would be difficult within the current home credit model. Home credit companies remain reliant, for the moment, on (human) agents to facilitate commercially beneficial relationships with borrowers.

One limitation of this study is that the more nuanced effects of emotional labour on worker wellbeing cannot be determined. Further, the data only illustrate the experiences of agents and managers, and not the home credit company’s customer base (borrowers). Collecting borrower perspectives, as well as employing an observational or ethnographic research design, could provide deeper contextual insights into agents’ emotional labour. Future research should also consider the role of women in home credit and explore the differing expectations placed on male and female agents in terms of engagement in emotional labour.

While generalisations are problematic, findings from this study may be reflected in other forms of interactive work, particularly in sectors where dependent self-employment is increasingly the norm in terms of employment conditions. In particular, the analysis concerning worker autonomy over emotional labour may extend to other sectors (e.g. estate agency, retail, mainstream financial services), where digitalisation threatens to heighten managerial control over workers whose qualitative input has traditionally been a key element of the labour process.

Sadly, high-cost borrowing (including home credit) is a reality of modern Britain and reflects the nature of finance in working-class communities and the increasingly punitive nature of the benefits system. This article focuses on the home credit sector not as an endorsement of high-cost lending but rather to shed light on an industry which, while heavily reviled in some quarters, is strongly embedded within Britain’s post-industrial communities.