Abstract
This article considers whether unbounded times and spaces of work are systematically associated with self-employment. In contrast to analyses that frame the spatial and temporal location of work as signifying autonomy or freedom, it posits that self-employment is produced by, and then reproduces, constraints on and preferences about spatio–temporal organization at both occupational and individual level. Using data from 5 years of the U.K. Labour Force Survey (2013-2017), the article takes a novel approach in the quantitative analysis of self-employment by conducting intraoccupational analysis within each of four relatively homogenous occupational groups: hairdressers, shopkeepers, art workers, and accountants. Analysis shows that (a) at population-level self-employment is strongly associated with both spatial and temporal unboundedness; (b) these effects are stronger for women than men; (c) in intraoccupational analyses, gender, alongside other sociodemographic measures, is largely nonsignificant, suggesting that the relationship between these and self-employment is primarily produced by differences associated with occupational segregation; and (d) the association between self-employment and different types of spatio–temporal unboundedness varies markedly by occupation. The article points to the importance of occupation and the spatio–temporal organization of concrete work activity in understanding the reproduction of self-employment. It concludes, therefore, that spatio–temporal unboundedness should be considered as a feature, or structural component, of self-employment, not a choice or by-product.
Introduction
As Karl Marx noted a century and a half ago, battles to set the temporal limits of employment, to carve out and defend nonworking time (weekends; holidays) and to reduce the length of the working day, have been central to class conflict and, therefore, the development of capitalist labor relations (Marx, 1867). Similarly, the spatial delineation of places of work and the separation of these from extra-work (and domestic) life have been quintessential aspects of industrial capitalism and its imaginings (Hareven, 1991) and underscored gendered and racialized divisions of labor (Ekinsmyth, 2014).
Yet if wage-labor produces and is produced within historically constructed spatio–temporal boundaries, self-employment is subject to different social, historical, and individual pressures and different spatio–temporal constraints. For instance, a characteristic of self-employment, notwithstanding the myriad of forms it takes (Bögenhold, this issue; Bögenhold & Klinglmair, 2016), is a twofold lacuna, the absence of direct labor relationships and of associated regulatory constraints. Alongside that, self-employment is characterized by individualized responsibility, risk and precarity. This article explores the extent to which self-employment is also characterized by less spatially and temporally demarcated work, asking to what extent unbounded spaces and times are a general feature of self-employment as opposed to a quirk or choice.
Advocates for self-employment suggest that it provides workers with freedom and autonomy, enabling them to choose where and when to work (Dellot, 2014). In this version, work during unsocial times, at home, on the move, or at variable hours of the day is evidence of personal preference. Yet qualitative studies have over and again shown self-employment status produces pressures on individuals to tolerate greater unpredictability in their working times (Hardy & Sanders, 2015) and spaces (Cohen, 2010); and that unbounded work produces tensions for self-employed workers, who frequently respond by attempting to rebound working times, places, and social relations (Mustafa & Gold, 2013; Myrie & Daly, 2009). This article suggests, additionally, that by stepping back from prevalent discourses of independence and entrepreneurialism (Valdez, 2015), we can identify the ways that spatio–temporal unboundedness is located within broader contexts, namely occupational structures and individuals’ extra-work lives. These contexts produce sets of preferences and constraints that are both constitutive of, and reinforced by, self-employed status.
Later sections of the article employ secondary quantitative data from the U.K. Labour Force Survey to chart the relationship between spatio–temporal unboundedness and self-employment, initially within the whole working population, but latter, within four relatively homogeneous occupational groupings. The analysis shows that (a) at population-level self-employment is strongly associated with both spatial and temporal unboundedness; (b) these effects are stronger for women than men; (c) in intraoccupational analysis, gender and other sociodemographic measures are largely nonsignificant, suggesting that relationships between these and self-employment are primarily produced by differences associated with occupational segregation; and (d) the association between self-employment and different types of spatio–temporal unboundedness varies markedly by occupation. Based on this analysis, the article suggests that spatio–temporal unboundedness is not simply a contingent outcome of self-employment, but rather is intimately related to its (re)production: as such it is a feature, not a bug.
(Re)producing Self-Employment
Self-employment rates have increased across most Organisation for Economic Co-operation and Development countries (Meager, 2016). In the United Kingdom, in the period since the financial crisis of 2008, rises in self-employment have far outpaced employment, with self-employment now accounting for about 15% of total employment. Yet average self-employment income has fallen (Office for National Statistics, 2014), exacerbating the concerns of policy makers for self-employed workers who typically lack access to pension, maternity, or sickness benefit and therefore occupy a precarious financial position (D’Arcy & Gardiner, 2014).
These self-employment rises have prompted attention, with much of the focus on “new” types of self-employment, namely forms of subcontractual labor associated with online algorithmically driven platforms such as Uber, Etsy, Deliveroo, and Mechanical Turk, variously identified as “crowd work” (Huws & Joyce, 2016), “gig work” (De Stefano, 2015; Graham, Hjorth, & Lehdonvirta, 2017), “the platform economy” (Forde et al., 2017) or “sharing economy” (Rahim, Lepanjuuri, & Piggott, 2017). Yet if increasingly large numbers of workers receive a small income working with these platforms, a much smaller number depend on them to regularly top-up earnings and a yet smaller number rely on platform work for their full income (Rahim et al., 2017). Moreover, as Lockey (2018) notes, “the two debates—rising self-employment and the emergence of the platform economy—are regularly elided, [but] there is almost no evidence that connects the two trends.” Rather, analysis of self-employment reveals considerable continuity—for instance, although Uber has changed the conditions of self-employment for taxi drivers, this was an occupation with already high rates of self-employment and significant subcontracting. Therefore, the main effect of online platforms has been to change the mode of third-party control over a proportion of self-employed workers: an important change for the workers concerned, but relatively minor in terms of net self-employment or even its occupational clustering. Instead, empirical studies have found that, “those who became self-employed post-financial crisis are not markedly different from those who were already self-employed” (D’Arcy & Gardiner, 2014, p. 40). Continuities occur, in part, because self-employment is a social relationship that is produced and then reproduced. Cross-sectional analysis of the currently self-employed is, therefore, as much an insight into the structural locations of self-employed reproduction (or how self-employment is sustained), as an insight into the production of new self-employment. Indeed, postrecession U.K. self-employment rate increases are due to fewer people exiting self-employment, rather than increased entry (Office for National Statistics, 2014).
Time, Space, and Self-Employment
The (re)production of self-employment is both individual and structural. At individual-level reproduction of self-employment depends on individual (or household) resources, constraints, options, and preferences. At a societal level, the reproduction of self-employment involves the reproduction of those socioeconomic niches in which self-employed activity can be sustained, for instance, where micro-businesses can compete against or are welcomed by large scale capital. As discussed below, in both sites of reproduction, spatio–temporal unboundedness is key.
Individual Spatio–Temporal Pressures and the (Re)production of Self-Employment
Social historians and feminist sociologists have long argued that work and extra-work (which includes, but extends beyond, the domestic and familial) are not wholly separate spheres, but rather involve interconnected and interdeterminate sets of social, but also spatial and temporal relations (cf. Glucksmann, 2005; Hareven, 1993). In line with this, there is considerable evidence that self-employment is understood by those engaged in it as a strategy to deal with extra-work demands and, especially, to manage work–family conflict (Myrie & Daly, 2009). For instance, a Swedish study found that a majority of self-employed men and a large majority of self-employed women gave “family/lifestyle” motives for becoming self-employed (Johansson Sevä & Öun, 2015). One of the reasons for this is that self-employment appears to offer possibilities for a “spatial reconfiguration” of work (Ekinsmyth, 2014; Halford, 2005) and, associated with this, increased temporal flexibility. Thus, studies find that women, especially new mothers, look to home-based self-employment to facilitate the management of competing demands (Berke, 2003; Carrigan & Duberley, 2013; Ekinsmyth, 2014; Lewis, Harris, Morrison, & Ho, 2015; Luckman, 2015). This understanding is most explicit where self-employed women self-identify as “mumpreneurs” (Ekinsmyth, 2011), a framing of work activity that foregrounds domestic status.
There is, however, a considerable gulf between the ideals of self-employment and its lived reality. Where self-employment produces individualized responsibility and workers who are “always on” it becomes difficult to separate from work during “nonworking” times (Hilbrecht & Lero, 2014). The absence of formal spatial or temporal boundaries can therefore exacerbate the pressures of work (Mustafa & Gold, 2013; Myrie & Daly, 2009). In these contexts, “the notion that women will achieve work-life balance by entering self-employment” is undermined by “the reality of the hours and commitment that self-employment and business ownership demands” (Forson, 2013, p. 467). Thus, self-employment does not alleviate work–family tensions (Johansson Sevä & Öun, 2015). Rather, as Luckman (2015, p. 155) points out, self-employment “can operate as a twenty-first century poorhouse . . . including many people working hours far in excess of those they desire for family-friendliness.” Craig, Powell, and Cortis (2012) find that, indeed, self-employed mothers do not work shorter hours, but they are able to engage in “time-shifting,” for example, moving their working hours so that they work fewer daytime hours and more at atypical times (evenings or weekends). Primarily, however, in their study self-employment facilitates the juggling of tasks, because it involves a spatial realignment and homeworking.
Self-employment is also disproportionately undertaken by workers nearing the end of their working lives (D’Arcy & Gardiner, 2014; Mallett & Wapshott, 2015; Small, 2012; Tomlinson & Colgan, 2014). Typically framed as a “pull” effect—that older workers have built up sufficient capital to start businesses and are motivated by a desire for self-realization and independence—qualitative studies also reveal “push” reasons that are both spatial and temporal. Older workers may find the labor market inhospitable, especially work within corporately regulated workspaces requiring standards of youthful aesthetics (Tomlinson & Colgan, 2014), something which self-employed work within domestic spaces may allow them to circumvent (Cohen, 2010). Additionally, older workers increasingly face renewed need for “flexible” temporalities to manage care for aging parents and partners alongside grandchildren (McKie, Biese, & Jyrkinen, 2013).
Wider kin relations also play a role in self-employment entry and in the spatio–temporal organization of this work. Kin can offer financial, instrumental, and emotional support, and enable the self-employed worker to meet variable demands. For instance, employed partners may provide income to stabilize the early days of self-employment (Hilbrecht & Lero, 2014). Kin resources may be especially important to immigrant and ethnic minority self-employed workers, who face obstacles in accessing formal institutions, such as banks or state agencies and may rely instead on support from spouses, but also children and more distant kin, sometimes supplemented by members of the coethnic community. Where such kin support involves access to cheap (or free) labor, it enables the self-employed to accommodate long working hours and gain competitive advantage in occupational niches in which cheap labor is essential and capital costs low (Cobas & DeOllos, 1989; Phizacklea & Ram, 1996; Ram & Edwards, 2003; Waldinger, Aldrich, & Ward, 1990). Both the type and quantity of kin support available (or sought) is, however, gendered in ways that tend to advantage self-employed men (Hilbrecht & Lero, 2014; Marlow, Henry, & Carter, 2009; Valdez, 2016).
Structural Spatio–Temporal Pressures and the (Re)production of Self-Employment
Discussion of self-employment is often surprisingly abstract, with self-employment framed as if it involves a single set of activities. Yet self-employment involves concrete work activity undertaken within specific occupations. This occupational context is important for both the (re)production of self-employment and the spatio–temporal location of self-employed work. In the postrecession period, D’Arcy and Gardiner (2014) find relative continuity of self-employment rates within occupations, despite change in overall rates, highlighting the extent to which self-employment is an occupational phenomenon. Occupation also appears in individual rationales for self-employment entry. For instance, in a U.K. study, Dawson, Henley, and Latreille (2014, pp. 809-810) find that the “nature of the occupation” was the second most commonly given reason for becoming self-employed, after “to be independent.” This suggests that for large numbers of workers self-employment is a career stage within their occupational trajectory rather than a radical break. Dawson et al. (2014) also find that regional variation in self-employment rates are largely produced by differences in occupational concentration. Similarly, Loscocco and Bird (2012) show that occupational sector determines both the relative success of self-employment, and the ways it is organized and gendered.
If these studies point to the relevance of sector and occupation, they tend to operationalize these using broad categories. For instance, Loscocco and Bird (2012) employ categories as overarching as “personal services” (including a range of activities from dry cleaning to personal training), Dawson et al. (2014) use just six sectoral categories and D’Arcy and Gardiner (2014) nine occupational categories. Problems with this approach become apparent when we note the concentration of self-employment within a few much narrower occupational settings. For instance, 2011 U.K. Census data shows that over half of all male self-employed work occurs in just 23 out of a list of 369 occupations. Exactly the same number of occupations (23), albeit a slightly different list, accounts for half of all female self-employment (Office for National Statistics, 2013, author’s own calculations). Thus, within the vast majority of occupations there is very little self-employed work. Conversely, other occupations have extremely high rates of self-employment and/or are sufficiently large occupations to account for a large part of all self-employment. For example, the single occupational category, “hairdressers and barbers” accounts for over 6% of all female U.K. self-employment. Given this degree of concentration, consideration of more fine-grained and better specified occupational groups is important.
Concrete work tasks are not infinitely mutable. This means that within any sociohistorical and technological context occupations set the parameters within which space and time are negotiated. Whereas some work activity must be performed at a specified time/place, other activities can be moved or time-shifted or, alternatively, may require temporal and/or spatial variability or a temporal duration outside the normal working day. For instance, a classic analysis of the development of capitalism drew attention to the ways in which the factory physically concentrated workers and required their presence during fixed temporal periods (Thompson, 1967). In contemporary capitalism, there continue to be many types of work activity with spatial constraints imposed by a requirement for proximity to others or to work with large scale machinery—for instance, a teacher’s work requires she be present at school during the school day. In contrast, some work activity, for instance, taxi driving, is less bounded and working hours or spaces less easily concentrated (Cohen, 2010), while other work, most obviously occupations reliant on mobile technologically, might not necessarily be unbounded, but may be able to be performed across a wide variety of places and atypical temporalities (Hislop et al., 2015).
It has been suggested elsewhere (Cohen, 2011; Cohen & Wolkowitz, 2017) that particular forms of occupationally rooted temporal unboundedness make it unprofitable for large firm sectoral dominance and may therefore establish a fertile terrain for self-employment. Which is to say, that much self-employment occurs within the interstices of the labor market and is focused on activities that are difficult to coral into the spatio–temporal requirements of profitable direct employment. This includes those work tasks that cannot be concentrated into a given period or that produce a mismatch between time at work (the time during which workers must be available to carry out tasks) and time working (those moments when labor is employed in profit-making work tasks). It also includes work activity that is spatially dispersed so as to produce temporal and spatial hiatuses (Cohen, 2010) or where the times and spaces of work activity are persistently unpredictable. To the extent that self-employment thrives in these contexts it is in large part because the self-employed worker is willing to “count” her time differently to an employee and, for instance, is willing to discount nonworking time at work. This is, of course, facilitated when time “at work” is not spatially demarcated from extra-work life.
This section has suggested that self-employment is occupationally concentrated and patterned. It has pointed to differences in the temporal and spatial organization of work activity in different occupations and proposed that occupations that involve unbounded temporal and spatial work activity creates contexts in which self-employment may be found, specifically because those contexts provide scope for the self-employed to develop coping strategies that are not open to larger scale capital.
Method
Data
Data are from the U.K. Labour Force Survey. The U.K. Labour Force Survey is conducted with a representative sample of U.K. households four times a year. It employs a panel design with one fifth of the sample replaced each quarter. Respondents therefore typically remain in the sample for five consecutive quarters. Each quarterly data set contains about 90,000 respondents. Because it involves a large sample, it is also ideal for exploring relatively small subgroups in the population such as the self-employed or homeworkers and has been used to explore within-occupation variation (Friedman, Laurison, & Macmillan, 2017).
Data from 5 years of the U.K. Labour Force Survey (2013-2017) were merged. In each case, the January to March (winter) quarter was selected. Wave five data were omitted because respondents in this wave were first included the winter of the previous year. This avoids any respondent being double sampled. The initial combined data set included 391,036 respondents. Of these, 185,925 had a currently recorded occupation and were included in further analysis.
The combined data are analyzed in two ways. First, the data are used to identify population-level relationships between unbounded work and self-employment. A secondary question in this analysis is whether and to what extent this is gendered. Or, do women do self-employment in a way that involves greater or lesser boundary breaching than do men? The second set of analyses explore equivalent questions, but do so by exploring work within specified occupations.
Occupational Groups
Four occupational groups were selected to explore in greater depth—accountants, hairdressers, shopkeepers, and art workers. These occupations were selected on the basis of (a) size—sufficient respondents in a single occupation or closely aligned set of occupations to conduct statistical analyses; (b) rates of own-account and employer self-employment—a reasonably large number of both types of self-employed worker; (c) diversity—selected occupational groups were varied, with different status, qualifications, sociodemographic profiles, and different types of work activity. 1 Hairdressing was chosen as the most common form of female self-employment. It involves workers in a mix of “body work” (involving the touch and the manipulation of customers’ bodies) and emotional labor (Cohen & Wolkowitz, 2017). In the United Kingdom, it does not require accreditation. In contrast, accountancy work is professionally accredited, typically involves considerable desk-based work and extensive use of information and communication technologies, making this a “technologically dependent” form of work (Hislop et al., 2015). Shopkeeping is the archetypical “petit-bourgeois” activity, and one of the most common forms of both male and female self-employment, as well as an occupation associated with immigrant self-employment (Villares-Varela, Ram, & Jones, 2018). The occupational group “art workers” as constructed here is a little more diverse, but encompasses artists; authors, writers, and translators; actors, entertainers, and presenters; dancers and choreographers; musicians; arts officers, producers, and directors; and photographers, audio visual, and broadcasting equipment operators. These workers perform “creative work” (Florida, 2002) 2 in a sector renowned for poor employment practices and the exploitation of workers’ internal commitment and motivations (Gill & Pratt, 2008; Hesmondhalgh & Baker, 2013). The four occupational groups shown are not intended to “represent” all self-employment. Rather, by deploying a more fine-grained understanding of occupation than found in previous analysis and then conducting within-occupation analysis, we can highlight variability and continuity. We can also ask whether self-employment and its spatio–temporal relations appears as a different phenomenon when analyzed at occupation, rather than population, level and how this varies across quite different occupations. Table 1 shows the unweighted sample sizes for each of these occupations for each year in the sample. These occupation-specific samples range from 1,156 to 3,026. Combined, these four occupational groups comprise over 4% of the total labor force: a small, but nontrivial, share of all employment.
Unweighted Sample Size of Occupational Groups, U.K. LFS 2013-2017 (Waves 1-4 Only).
Note. U.K. LFS = U.K. Labour Force Survey.
Art workers includes SOC2010 categories 3411-3417: “Artists”; “authors, writers, and translators”; “actors, entertainers, and presenters”; “dancers and choreographers”; “musicians”; “arts officers, producers, and directors”; “photographers, AV and Broadcasting equipment operators.” bAccountants includes SOC2010 category 2421: “Chartered and certified accountants.” cHairdressers includes SOC2010 categories 6221: “Hairdressers and barbers” and 1253: “Hairdressing and beauty salon managers and proprietors.” dShopkeepers includes SOC2010 category 1190: “Managers and directors in retail and wholesale” and 1254: “Shopkeepers and wholesale/retail dealers.” eAll workers includes all respondents with nonmissing information on occupation.
Variables
Work relations is measured by three categories: employee, self-employed own-account worker and self-employed, employing others. As Table 2 shows self-employment accounts for about 15% of the working population, with this primarily taking the form of own-account work. This is much higher in some of the selected occupations, however. Thus, a majority of hairdressers and art workers are self-employed (51% and 69%, respectively) as compared with a large minority of shopkeepers (33%) and a smaller minority of accountants (20%). Among art workers, only about 3% of the self-employed are employers. In contrast, about 43% of self-employed shopkeepers are employers. This highlights how self-employment clusters within particular occupations, and that the opportunity to move from own-account to employer self-employment is unequally distributed across occupations.
Descriptive Statistics for Workers in Selected Occupational Groups and for All Workers, U.K. LFS 2013-2017 (Percentages).
Note. U.K. LFS = U.K. Labour Force Survey. These data are weighted to represent population by using PWT11 (2013 data), PWT14 (2014 data), and PWT17 (2015-2017 data). All years, Waves 1-4 only (Wave 5 excluded). Only those providing occupation data (SOC10M) included.
Three types of unbounded temporality are explored: duration, schedule, and variability. Duration is measured using the variable “Usual weekly hours.” Analysis of self-employed work has highlighted that it may involve inordinately long hours, but also marginal, short hours. Thus, the variable is transformed into three categories: “normal” hours are measured here as weekly hours between 30 and 45 per week. 3 Over half of the working population reports working these hours. Short hours (1-29 per week) are reported by 24% of the population and long hours (46+ hours per week) are reported by 19%. There is some variation by occupation, with hairstylists and, to a lesser extent, art workers notable for their short hours (worked by 42% and 31%, respectively) and shopkeepers for their long hours (worked by 38%).
Unbounded schedule here refers to work being performed during a culturally nonworking day—the weekend. The variable is coded one if respondents work on either or both weekend days. A fifth of all workers report working weekends. When this is examined by occupation, we see that the majority of each of hairdressers and shopkeepers, but only 3% of accountants, do weekend work.
Unbounded variability is measured by a variable that asks respondents whether their “weekly hours tend to vary,” with response categories Yes and No. Over a third of workers report variable hours. Given the current focus on the negative effects of zero-hour and variable-hour contracts, this is perhaps noteworthy. The variable does not, however, distinguish variability initiated at the worker’s discretion (a positive form of flexibility) from variability that is the product of unpredictability (a negative impact). Notably, more than two thirds of art workers report that they work variable hours, a rate almost twice that found among hairdressers. If this variability were purely the product of workers’ preferences, it is unlikely interoccupational differences would be that high.
Unbounded Space is explored using a single variable that asks workers to specify whether they work in their “own home,” “the same grounds or building” as home, “different places with home as a base” or in a “Separate from home.” 4 The large majority of workers work “separate from home.” This varies a little by occupation, with 21% of art workers and 8% of accountants working at home and 33% of art workers and 13% of hairdressers working in different places (heretofore described as “mobile”; Brown & O’Hara, 2003; Cohen, 2010). Because relatively few workers state that they work in the “same grounds or building,” future analyses merge this category with those who work from their own home. Both working at home and mobile work involve the breaking down of boundaries, but mobile work is likely to radically increase unpredictability, exposing workers to unfamiliar environments and increasing the difficulty with which even familiar work tasks are performed.
Analysis Strategy
The analysis below investigates whether self-employment is associated with work that is spatio–temporally unbounded. Second, it looks at how this relationship, between spatio–temporal unboundedness and self-employment, is gendered and third, in a separate set of analyses, how the relationship between spatio–temporal unboundedness and self-employment varies by occupation. Multinomial regression is used to explore those factors underpinning the (re)production of self-employment. In all analyses, own-account and employer self-employment are differentiated from employee status (the reference category). This follows previous studies in which important differences were found between own-account and employer self-employed (cf. Johansson Sevä & Öun, 2015) and reflects the different temporal and spatial possibilities open to sole workers and those who employ others. In all analyses, both sociodemographic and temporal and spatial arrangements are employed as independent variables. Gender is employed as an interaction term to examine the gendering of identified processes. The second set of regression analyses are carried out separately on each of the four selected occupational groups. Lower sample size, and less variability make it impossible to interact gender with as many spatio–temporal measures in these analyses.
The use of spatio–temporal arrangements as independent, rather than dependent, variables is relatively unorthodox: the standard assumptions are that the times of work follow from employment status. Yet, as discussed above, if we consider the individual and structural bases for the (re)production of self-employment the reverse argument emerges: that self-employed status is the product of spatio–temporal constraints or preferences. For instance, self-employment may represent an attempt to experience greater “freedom” (something that often is implicitly spatial or temporal); it may be sought to manage work–life conflict; and, at a structural level, self-employment may occur within the interstices of the labor market focused on activities, which are difficult to squeeze into the temporal-spatial requirements of profitable direct employment.
Findings
Demographics
Many of the sociodemographic features identified previously are found to impact the odds of being self-employed, either working as an own-account worker or employing others (Table 3). Male workers have odds over three times that of female workers of being self-employed own-account workers and just under three times that of women of being self-employed, employing other. Additionally, older workers, workers not born in the United Kingdom and workers who do not report health problems all have raised odds of being self-employed (either own-account workers or employers). Marriage is a little more complicated. Being married decreases the likelihood of own-account self-employment, but only for men. Conversely, being married increases the odds of both men’s and women’s employer self-employment, but the increase is bigger for women than men. This suggests that the benefits of shared household resources, for supporting individual self-employment are especially important for women in larger scale (employer) self-employment. Having a child younger than 4 years slightly increases the odds of own-account self-employment for both men and women (a 15% increase). It also increases the odds of employer-self-employment, especially for women. Although not a very strong effect, this is in line with previous studies that suggest that self-employment may be a way of managing child care, especially for mothers.
Multinomial Logit Regression of Work Relations on Demographic, Spatial, and Temporal Attributes, all Workers Listing an Occupation U.K. LFS 2013-2017: Waves 1 to 4 (Odds Ratios).
Note. U.K. LFS = U.K. Labour Force Survey; OR = odds ratio; CI = confidence interval; LR = likelihood ratio. Reference category: employee.
p < .1. *p < .05. **p < .01. ***p < .001.
Space
Unbounded space (mobile work and homeworking) is very strongly associated with self-employment. This is most true of own-account female self-employment, where being either a mobile worker or home-based multiplies workers’ odds of own-account self-employment over thirtyfold. The connection with employer self-employment is also strong, albeit a little less so: being mobile or home-based increases female workers’ odds of employer self-employment, by a multiple of five and eight, respectively. Overall, the findings here suggest that work that either requires, or work that can be performed in, unbounded space is much more likely to be performed by the self-employed. 5 Additionally, in all cases we see significant gender interactions which indicate that unbounded working spaces are somewhat less strongly associated with male self-employment. Thus, while men who are mobile for their job have increased odds of performing own-account self-employment (as compared with men with dedicated workplaces), these odds are half as great as for mobile women. Similarly, although working from home is strongly connected with male own-account and employer self-employment, the odds are 0.4 times as great as for home-based women. In other words, unbounded space is strongly associated with self-employment for both men and women, but the association is considerably stronger for women.
Temporality
The three types of unbounded temporality analyzed impact on the likelihood of self-employment in multiple ways. First, atypical temporal duration (short or long hours, as opposed to regular hours of 30-45 per week) increases the likelihood of own-account self-employment (by about 70% and 30% for short and long hours, respectively). Thus, the own-account self-employed may be in marginal employment (with low hours), but this type of self-employment also occurs where workers put in long shifts. Conversely, there is a negative association between working short hours and self-employment that involves employing others, whereas working long hours multiplies the odds of being in this status by 150%. Thus, long working hours increases the odds of self-employment, and especially of employer self-employment, but working short hours makes own-account self-employment more likely and employer self-employment less likely. The effects of temporal duration on the odds of being in own-account self-employment are significantly reduced for men, as opposed to women. Therefore, short or long hours of work are more strongly related to women’s than to men’s own-account self-employment. Turning to employer self-employment, however, there are no significant gender interactions with temporal duration.
Two other types of unbounded temporality are explored: schedule (weekend working) and predictability/flexibility (variable hours). Both types of unbounded temporality increase the odds of own-account and employer self-employment. As was the case with unbounded space and temporal duration, the impact of both weekend working and variable hours on the odds of self-employment is significantly less strong for men, as compared with women. The exception is the association of variable hours with own-account working, which is not gendered. Thus, both temporal variability and unsocial hours are strongly associated with own-account self-employment and, for women especially, self-employment is associated with weekend work, a form of “time-shifting” (Craig et al., 2012).
This analysis shows, therefore, that unbounded spaces (mobile and home) and times (temporal duration, schedule, and variability) are associated with both own-account and employer self-employment. To a large extent, this is true for men and for women. Where, however, gender differences exist, they show that unbounded times and spaces more strongly underpin women’s, than men’s, self-employment activity. This analysis cannot differentiate, however, whether this difference is due to differences in preference, or whether it relates to more structural differences in the spatio–temporal organization of the (different) types of work in which self-employed men and women engage. To explore this further, we need to control for some of those differences by doing within-occupation analysis.
Occupation
The second set of analyses (Table 4) again use multinomial logit to regress work relations (employee vs. own-account self-employed and employer self-employed) on workers’ demographic characteristics and on the organization of time and space. Now, however, separate analyses are done for each of the four selected occupational groups. This analysis includes all demographic, temporal, and spatial independent variables. Due to smaller Ns, and relatively small subgroups, especially among those involved in employer self-employment, it has not been possible to include all the gender interactions from Table 3. However, three gender interactions could be included in all four analyses (two demographic and one temporal). Other interactions were explored within those occupations where this was possible, but are not shown here.
Multinomial Logit Regression of Work Relations on Demographic, Spatial, and Temporal Attributes, for Four Occupational Groups U.K. LFS 2013-2017: Waves 1 to 4 (Odds Ratios).
Note. U.K. LFS = U.K. Labour Force Survey; LR = likelihood ratio. Reference category: employee.
p < .1. *p < .05. **p < .01. ***p < .001.
In contrast to the whole labor-force analysis, intraoccupational analyses find that sociodemographic variables have few significant, or sizeable, effects on self-employment. It may be that the nonsignificance of sociodemographic variables here is, in part, due to smaller sample sizes in these analyses than in the labor-force analysis. As we will see, however, relatively strong spatio–temporal effects remain in all four within-occupation analyses. Additional analyses (not shown) in which self-employment within each occupation was regressed on sociodemographic variables only (without additional spatial or temporal variables) similarly identified very few significant effects. This suggests, therefore, that the sociodemographic effects identified in the whole labor-force analysis may in large part be occupational effects, something that is unsurprising given the gendered and racialized nature of many high-self-employment occupations. There are, however, a few exceptions among our four occupational cases, where sociodemographic characteristics appear to play a significant, and sizeable, role in within-occupational analysis. The first relates to shopkeepers. As the final columns of Table 4 shows, several sociodemographic effects on self-employment among shopkeepers persist even after controlling for spatio–temporal organization. These include a consistently significant effect for being born outside of the United Kingdom, highlighting the extent to which, among those working as retail or wholesale shopkeepers, the native born are more likely to work as employees and self-employment is associated with migrant status. This chimes with the various analyses of immigrant self-employment, which have focused on this occupation (cf. Phizacklea & Ram, 1996). A second strong sociodemographic effect is found in hairdressing, where with every year they age workers have odds 24% higher of becoming own-account self-employed and 49% higher of becoming employer self-employed. Given that this is a relatively young occupation this suggests that self-employment may be a career stage—wherein workers either leave the occupation or become self-employed as they age.
As already indicated, spatio–temporal effects are more consistently significant across the different occupations, and in some cases quite sizeable. As was the case at population level, spatial unboundedness is an important predictor of self-employment across occupations. This is especially the case for own-account work, although the effect strength of spatial unboundedness varies by occupation. For instance, mobile work has a very strong effect on own-account self-employment in hairdressing (multiplying the odds of own-account work by 220) and strong effects on own-account work in other occupations (multiplying the odds between 10 and 23 times). Mobile work is less strongly and less consistently associated with employer self-employment. Working at home has a consistently strong association with own-account work (multiplying the odds by a factor of between 12 and 27) and more varied relationship to employer self-employment. For example, among accountants, neither mobile work nor homeworking affect the odds of being in employer self-employment, but among art workers, mobile and homework each increase the likelihood of employer self-employment (fivefold and sevenfold, respectively). Thus, in some sectors, but not others, spatial unboundedness persists into employer self-employment. We can say, therefore, that spatial unboundedness (mobile and homework) are strongly and consistently associated with own-account work, despite occupational variation, but have inconsistent and weaker relationships with employer self-employment.
The story is much more mixed with respect to temporal unboundedness, with different types of unbounded temporality affecting differently on self-employed status across occupations. In accountancy, a professional occupation in which pressures to work long hours are widespread, we find that temporal duration is unconnected with self-employment (own-account or employer). In contrast, long hours are associated with employer self-employment for art workers, hairdressers, and shopkeepers and, to a much lesser extent, with own-account work in arts and shopkeeping. Short hours are positively associated with self-employment in the arts and shopkeeping, but negatively with own-account and employer hairdressing. Short hours in hairdressing, therefore, increases the likelihood of direct employment, whereas short hours in the arts increases the likelihood of self-employment. Since these are the two occupations in which short hours are most common this highlights that the precise ways in which temporal duration and self-employment are connected varies by sector. For instance, it is likely that within hairdressing, a relatively young and highly feminized sector, part-time work is a normal aspect of regular employee relationships, whereas short hours in the arts may more often be attributable to a lack of work—and self-employment ensures that this risk is assumed by the individual self-employed worker. In two of the four occupations, the interaction of gender and temporal variability is significant. In both cases (hairdressing and shopkeeping), the effect of variable hours on own-account self-employment is considerably less pronounced for men than women, suggesting that women’s self-employment is more closely associated with unbounded times in these occupations (although among shopkeepers, the main effect for temporal variability is insignificant, so it may be that in this case, it is male invariant temporality that is associated with own-account self-employment). Interestingly, the gender-variability interaction was not significant in Table 3, highlighting again the importance of occupational context.
Discussion
The foregoing analysis has shown that self-employment is much more likely to occur when work is performed in unbounded times and spaces. This study has considered unbounded temporality as multifaceted, and has included measures to assess temporal duration, schedule, and variability. All three dimensions of temporal unboundedness increase the likelihood of self-employment, albeit with slightly different effects on own-account versus employer self-employment. Unbounded spatial organization consistently and substantially increased the likelihood of self-employment. Since only some work can be performed in unbounded spaces—for instance, work involving heavy fixed machinery or the regular copresence of a group of coworkers and customers typically cannot—this finding identifies a potentially important underlying factor in the occupational concentration of self-employment. As noted above, spatial unboundedness is associated with boundary crossing and, where it involves mobility, may introduce increased social, environmental, and temporal unpredictability into working life. This highlights the extent to which self-employment regularly involves multilayered types of unpredictability and the associated multiple risks which must be borne by individual workers.
Second, and congruent with previous work (Craig et al., 2012; Ekinsmyth, 2011), the analysis has identified a strong relationship between unbounded spatio–temporal organization and gender. This is especially true of spatial unboundedness, with female own-account, but also employer, self-employment strongly associated with unbounded work spaces (homeworking or mobile work). The same association holds, albeit slightly less strongly or consistently, with respect to unbounded temporality. Whether this means that women are indeed embarking on self-employment that facilitates an idealized work–life integration and greater control and flexibility, or are forced to engage in self-employment in marginal spaces and at unpredictable times is not clear.
What this study has highlighted, however, is the importance of occupation. Taking a novel approach to quantitative analysis of self-employment, merged waves of the U.K. Labour Force Survey were used to construct relatively homogenous occupational groupings. This allowed analysis of within-occupation self-employment, where occupation was a relatively meaningful construct (“hairdresser” or “accountant” rather than “personal service worker” or “professional”). These within-occupational analyses show that there are different factors underpinning own-account and employer self-employment in different occupations. Sociodemographic factors have relatively little within-occupation impact on self-employment. In contrast, spatial organization remains hugely important, especially for own-account self-employment. Even here, however, there is occupational variation. For instance, accountants whose work involves unbounded space are more likely to be own-account self-employed, but there is no significant effect on their likelihood of being employers.
The argument, as laid out in Figure 1, is that self-employment is (re)produced in part in response to a set of spatio–temporal preferences and constraints. These preferences and constraints (Box 3) are generated in, at least, two ways. First, they are a product of the temporal and spatial structures of individuals’ extra-work life (Box 2). Second, they are produced by the spatio–temporal structures imposed by particular types of work (Box 1). In Figure 1, these relationships are represented by the arrows that go between Boxes 1 and 3 and between Boxes 2 and 3. These spatio–temporal organizational and individual preferences and constraints (Box 3) produce forms of spatio–temporal unboundedness and (re)produce self-employment.
Self-employment and spatio–temporal organization.
Although not shown on Figure 1, the question of whether organizational and individual preferences and constraints lead to self-employment is not pregiven but is likely to vary with institutional context. Thus, there may be reduced demand for unbounded workplace temporalities in societies with well-organized state-sponsored child care. Similarly, although types of work may involve temporal unpredictability and create problems for capital with respect to the efficient use of employee labor, the use of self-employment as a mechanism to shift the risk for such temporally variable work, may be less possible in societies with stronger employment protections (Hipp, Bernhardt, & Allmendinger, 2015; Kreide, 2003).
Conclusion
This article has explored the ways in which spatio–temporal variability underpins self-employment and how this is affected by the organizational contexts of work. This is not, however, intended as a negation of the argument that self-employment produces its own set of spatio–temporal pressures and possibilities. Rather, it is undoubtedly the case that the self-employed are sometimes more willing or are more strongly compelled than employees to work long hours, to vary their hours with varying demand, wait-out nonworking times without pay or to engage in work in otherwise nonwork spaces (Cohen, 2010; Hardy & Sanders, 2015). This aspect of self-employment is something that has been widely exploited in the development of new types of self-employment, most notably, as part of the “uberization” of work (Fleming, 2017). The argument made here, however, is that this temporal unboundedness is a feature not a bug of self-employment. As such it should not be seen as a choice or by-product of self-employed work, but rather as something that underpins the structural locations within which self-employed work is reproduced in capitalism. As this article has shown, both at population level, but even more, after controlling for occupation, a relatively large part of total variation in self-employment is indeed accounted for by spatio–temporal organization.
In the above analyses, shopkeepers were an interesting exception, with much stronger sociodemographic effects on the likelihood of being self-employed than found in the other three occupations examined. It is worth noting that shopkeeping is a type of self-employment that typically requires long hours or considerable labor input, but also relatively unskilled labor, as compared with the other occupations examined here. Shopkeeping is, therefore, the archetypal example of a self-employed activity in which the ability to command the (cheap) labor of oneself and others is critical, whether this is achieved on the basis of kin relations or by leveraging relationships among those with otherwise constrained labor market options, including immigrants (Edwards & Ram, 2006). This is a reminder, therefore, that not only does the relationship between self-employment and space and time vary by occupation, but so too does the types of sociodemographic context required to (re)produce oneself as self-employed.
This article has highlighted some of the ways with which space, time, and occupation intertwine with self-employment. However, it represents only the first steps in the development of this analysis. Future analyses could fruitfully look more closely at specific occupations and at more diverse occupations as well as different countries and time periods, to better specify the arguments made here.
Finally, while this article highlights the extent to which self-employment involves unbounded times and spaces the data do not differentiate whether this is experienced as freedom or uncertainty (Bögenhold & Klinglmair, 2016). I would suggest, however, that the fact that both spatial and temporal unboundedness appear to be part of how self-employment is structured, suggests that while individuals may find advantages in such unboundedness, it is neither a “choice” nor a quirk of self-employed work. In a context of rising self-employment, it may therefore become increasingly incumbent on self-employed workers and the agencies and campaigns that work with them, to develop mechanisms to spatially and temporally rebound work (Mustafa & Gold, 2013; Myrie & Daly, 2009).
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.