Prescripted living: gender stereotypes and data-based surveillance in the UK welfare state

Introduction

The UK welfare benefits system has undergone significant changes since it was introduced in its modern form after World War II. Over more than 70 years, successive governments have shifted their focus away from universal support, and placed more and more demands on claimants to prove that they are ‘deserving’. In this paper, I will examine the operation of the UK welfare benefits system, and in particular the ways that it utilises both gender stereotyping and surveillance, thus harming both individuals and society as a whole.

In the first section, I will consider how international law views gender and gender stereotypes: in section 2, I will examine how, since the 1940s, the welfare system in the UK has operated based on stereotypes and assumptions, particularly about the roles of poor women within families. In the third section, I will track the development of surveillance in the UK welfare benefits system, and examine how increased conditionality of welfare benefits—distribution based on behaviour, not need—has used surveillance to track compliance and impose sanctions on those who do not behave according to requirements.

In Section 4, I will argue that surveillance—in particular data-based surveillance—and gender stereotyping act to reinforce each other, creating a vicious cycle in which the surveilled are incentivised to conform, and the non-conforming are increasingly surveilled. I will focus in this section on the ways that gender stereotypes and data-based surveillance act to categorise and to control individuals. Finally, in Section 5, I will argue that the increased use of surveillance in the welfare system in the UK increases the extent to which welfare claimants are monitored for compliance with gender norms and stereotypes, risking a vicious cycle in which claimants, fearing discrimination and denial of benefits, are increasingly coerced into compliance, and suggest how human rights can frame and support resistance.

Gender and gender stereotyping in international human rights law

Defining gender

In this article, I will define gender as the cultural and social meanings ascribed as a result of the legal, social and/or self-determined classification into one or more identities. This definition is based on human rights law, specifically by the UN Committee on the Elimination of All Forms of Discrimination Against Women (CEDAW Committee): “socially constructed identities, attributes and roles,” based on whether someone is (or is not) male or female, and the cultural and social meaning ascribed to people as a result (General Recommendation No.28, 2010, para. 5).

However, it is important to note that the CEDAW Committee’s definition is binary, considering only two genders: male and female. While the Committee recognises that a wide range of factors, including cultural and social factors, can have an impact on gender, they neglect to recognise that not everyone fits neatly within a binary gender model. Historically many cultures recognised—and continue to recognise—non-binary genders in society,1 several countries now allow for legal recognition of genders apart from male and female,2 and still more people living outside of these countries identify as non-binary or other gender identities, even if these are not recognised in law or widely understood in society. Consequently, the way in which genders outside of the binary are—or are not—recognised in law, in society, or at an individual level in different contexts remains relevant to this analysis.

Defining stereotypes

A stereotype is a “belief about the characteristics of groups of individuals”(Stangor, 2000, p. 1): gender stereotypes, therefore, are beliefs about the characteristics of individuals based on their gender. Gender stereotypes are culturally and temporally specific, but the existence of gender stereotypes—sets of beliefs about individuals, based on their gender identity or their social or legal gender—persists across social sectors, across cultures, and across time.3

As Stangor has pointed out, while there may be some element of truth to stereotypes, they frequently overgeneralise: not every member of a particular group possesses a particular characteristic (Stangor, 2000, p. 7), but the nature of stereotypes as categorisation tools means that they may be treated as though they do. This categorisation may be harmful: decisions made on the basis of gender stereotypes may be inaccurate in harmful ways. For example, as Cook and Cusack point out, if a woman is denied a job as a firefighter on the basis that ‘women are weak’, even if she is herself capable of passing any strength test requirement for the job (Cook & Cusack, 2010, pp. 15-16).

This is the case even when stereotypes are deployed in a way which may appear protective, for example in labour laws that prevent some jobs from being done by women, which are often rationalised as protecting women’s reproductive health. The CEDAW Committee has repeatedly noted that such laws in fact do not do so, and act instead to restrict employment and economic opportunities on the basis of stereotyping women as mothers (Holtmaat, 2012, pp. 157-8).

In human rights law, stereotypes are harmful when they act in such a way as to deny people their human rights (Brems & Timmer, 2016, p. 4), which include the right to be free from discrimination (ICESCR, 1966; ICCPR, 1966; CEDAW, 1979). The Committee which oversees the International Covenant on Economic, Social and Cultural Rights (ICESCR) has further recognised that stereotypes can underpin discrimination (General Comment No. 20, 2009, para. 20).

Privilege—relative power within the matrix of domination (Collins, 2009)—may offer freedom from gender stereotypes. The consequences for non-conforming behaviour are less serious for people who already possess comparative freedom to move about the world. This has been noted in the international human rights system. For example, the CEDAW Committee has noted that in the context of disaster response, “The categorization of women and girls as passive “vulnerable groups” in need of protection from the impacts of disasters is a negative gender stereotype” and that this approach fails to recognise women’s contributions in disaster preparation and response (General Recommendation No.37, 2018, para. 7). Similarly, the Inter-American Court of Human Rights has addressed gender stereotypes in two cases: the Cotton Field case in 2009 and the Atala Riffo case in 2012. In both cases, the Court specifically noted the reinforced obligations of the state to protect people in vulnerable situations, and/or who experience structural discrimination (Undurraga, 2016, pp. 77–80).

International human rights law has specifically addressed gender stereotypes and the harm that they can do, primarily through the CEDAW Committee General Recommendations. These include underpinning harmful practices (Joint General Recommendation No. 31 and General Comment No. 18, 2014, para. 17), contributing to violence and abuse against older women (General Recommendation No. 27, 2010, para. 16), and excluding women from post-conflict and peacebuilding efforts (General Recommendation No. 30, 2013, para. 43). The CEDAW Committee has stated, as a result of this potential for harm, that states have the obligation to address gender-based stereotypes, at individual, institutional and legal levels (General Recommendation No. 25, 2004, paras 7, 27).

Gender stereotyping in the UK welfare system

In this section, I will examine the ways in which the UK welfare system has historically relied on gender stereotyping in order to assess who should receive state welfare assistance. I argue that this reliance on assumptions and stereotypes about gender discriminates against those who do not conform to stereotypical norms. In particular, the UK welfare system has been designed to perpetuate the norm that women should be supported by a male breadwinner, with state assistance only when such support is impossible to obtain.

In the UK, the post-war Labour government under Clement Attlee (who was elected in 1945) are usually credited with the creation of the modern welfare state (Page, 2016, p. 126), much of which was based on the 1942 Beveridge Report. The reforms—including the 1946 National Insurance Act, providing for security from loss of wages, and the 1948 National Assistance Act, providing for security for those unable to work—were centred on ideas of universality and destigmatisation. In theory, support was available to everyone.

In practice, however, the institutions responsible for distributing welfare benefits were opposed to taking on responsibilities that they believed were family responsibilities (V. Noble, 2009, p. 46). Welfare policies in this period rested on stereotypes that families were stable units over time, and that men and women had very different family roles: men were full-time workers while women were full-time carers (Haux, 2016, p. 411). Men, in other words, were considered to be breadwinners, earning a ‘family wage’, while their female partners worked in the home (Parton, 2014, pp. 17–18). As a result, however, welfare policies in the post-war welfare state were designed to incentivise men to work and disincentivise women from working (Fredman, 1997, p. 84), and were focused on the family as a unit, assuming that benefits received by men would also benefit women and children (Fredman, 1997, pp. 18–30).

This ‘male breadwinner’ model had its closest analogue in the family lives of middle-class married couples at the start of the 20^th century (Lewis, 2000, p. 83) and the extent to which families in the UK in the 1950s conformed to this stereotype was debatable (Haux, 2016, p. 411). Nonetheless, welfare policies were based on the assumption of a male provider. One of the main reforms, the introduction of social security against loss of wages (as part of the National Insurance Act 1946) was mandatory for men and for single women. Married women, however, were not automatic participants, lost all credit from contributions made before their marriage, and received lower benefits. The policy assumed that they would always be supported by their husbands and therefore needed no protection (V. Noble, 2009, pp. 3–4).

Over the course of the second half of the 20^th century, social security began to be based more on an ‘adult worker’ or ‘citizen worker’ model, which increasingly treated women as economically independent individuals (Hunter, 2016, p. 93). This was to some extent linked to changing patterns of family life—for example, in the early 1970s, 8% of families were headed by a lone parent: this figure had risen to 25% by the early 2010s (Haux, 2016, p. 411). At the same time, there was a recognition that roles within families, including caring responsibilities, were shifting. The 1980 Social Security Act allowed (via complicated calculations) for the possibility of either partner in a relationship to claim non-contributory benefits, and from 1988 couples were allowed to choose which partner claimed these benefits (Fredman, 1997, p. 171).

However, the norm of the ‘male breadwinner’ persisted in welfare policy, even if not explicitly articulated. In the 1980s and 1990s, family policy shifted from being based on marriage to being based on parenthood, through an increased focus on ‘parental responsibilities’ (Lewis, 2000, pp. 90–91). At the same time, under a Conservative government, the universalist and destigmatising approach of the post-war welfare system was replaced by rhetoric of a welfare state ‘in crisis’ (V. Noble, 2009, p. 144). This ‘crisis’ included blaming lone parents (who were largely women) for their “unnecessary dependence” on the government: in practice, as Virginia Noble has argued, a failure to secure support from a male breadwinner (V. Noble, 2009, p. 144).

New Labour, which came to power in 1997, continued to blame the post-war welfare state for increasing ‘dependency’ on welfare, and focused their policies on getting claimants (back) into the labour market (V. Noble, 2009, pp. 146–147). In particular, they viewed returning lone parents to work as a key solution to child poverty (Grabham & Smith, 2010, p. 82). 90% of single parents were women: policies designed to encourage them to work were based on assumptions that they did not want to work (Grabham & Smith, 2010, p. 85) or did not recognise the value of work (Millar, 2019, p. 88).

This stereotype that lone parents—most of whom were women—were dependents, continued into the Coalition government elected in 2010, and the Conservative government which has been in power since 2015, both of which pursued austerity measures in welfare policy. A speech in 2016 by the then Secretary for Work and Pensions blamed the “low expectations” of lone parents for them not being in work (Green, 2016). Lewis has argued that lone mothers have three possible sources of income: the state, the labour market, and men (Lewis, 2000, p. 95). Since 2010, the Coalition and then Conservative UK governments continued the trend of encouraging—or coercing—lone mothers to move from state income to employment income. In practice, most lone mothers do want to return to work, but at a time that makes sense for themselves and their families, under circumstances that makes paid work viable (Grabham & Smith, 2010, p. 85). However, welfare policies in the UK ignore the actual experiences of lone mothers seeking work, and instead assume that work is always beneficial.

At the same time, the introduction of Universal Credit—a combined benefit replacing multiple individual benefits, based on a household means test and paid to a single bank account—risks returning partnered women in heterosexual relationships to dependence on men, and undermines women’s financial independence (Hunter, 2016, p. 96). For women in a cohabiting heterosexual relationship, the Universal Credit regime means that it is not always financially beneficial for them to take on work (Haux, 2016, p. 415). Almost 80 years after William Beveridge wrote his 1942 report—the “blueprint for the modern welfare state” (V. Noble, 2009, p. 3)—the British welfare state has returned to considering women benefits claimants—regardless of their actual situation—as dependents of a male breadwinner. In other words, the welfare benefits system continues to consider women not as independent actors, but instead based on stereotypes about their familial and social roles.

Surveillance in the UK welfare benefits system

In this section, I examine the ways in which the UK welfare benefits system has increasingly relied on surveillance to assess welfare claimants and monitor their compliance with welfare conditions.

Defining surveillance

In this article, I use David Lyon’s definition of surveillance: “a focused attention to personal details aimed at exerting an influence over or managing the objects of the data" (Lyon, 2002, p. 242). In other words, surveillance has two components: the act of observation, and the intent to use that observation to effect change. Surveillance, defined in this way, does not require technology, but technology makes surveillance easier.

David Lyon et al. note that surveillance has become generalised: instead of target-based surveillance (in which an individual of interest was identified and then surveilled), information is now collected on as many people as possible. The authors attribute this not only to technical advances—including the increasing availability of large data sets and computational power—but also to a shift in priorities: “a confluence of factors make surveillance often appear as the most appealing way to advance any number of institutional agendas” (Lyon et al., 2012, pp. 2–3).

As a consequence, data-based surveillance is now widespread. Data is collected on a large range of people: and the target for surveillance emerges from data analysis (Eubanks, 2018, p. 122). Anyone who is in the data set—or who shares any of the same characteristics—can be observed, scrutinised and judged, as part of what John Cheney-Lippold has called ‘dataveillance’ (Cheney-Lippold, 2017).

The development of welfare surveillance and dataveillance in the UK

The post-war welfare state aimed to destigmatise the claiming of welfare benefits. This included, for example, abolishing the household means tests required by the 1934 Unemployment Act, which was considered intrusive and humiliating for claimants (V. Noble, 2009, p. 17). Nonetheless, welfare claimants continued to be subjected to surveillance. The National Assistance Board, established in 1948 to oversee the provision of assistance for those not eligible for National Insurance, included a special investigative unit which had 60 staff by 1959: some of its investigations resulted in criminal prosecutions, but far more in denied or withdrawn claims for assistance (V. Noble, 2009, pp. 64–65).

By the early 1980s, the political view of welfare had shifted from universal, community-based support to a ‘public burden’, which was placed on ‘society’, a group that was implicitly differentiated from welfare recipients (Mesher, 1981, pp. 119–120). The term ‘underclass’ began to be used by social and political analysts to refer to people who experienced long-term unemployment. While the idea of a group of people who were culturally and socially excluded from the working class was seen as unevidenced or even politically dangerous, the ‘underclass’ became a group of focus for both left-wing analysts—who saw a group ‘left behind’ by social and economic changes—and right-wing conservatives, who believed that the term encapsulated a group not only unable but unwilling to work (Welshman, 2013, ch. 8). The ‘underclass’ were seen as dependent on welfare benefits by choice.

Continuing the century-old rhetoric of distinguishing between the ‘deserving’ and the ‘undeserving’ poor (Garthwaite, 2011, p. 370), and influenced by (amongst others) right-wing writer Charles Murray (Welshman, 2013, ch. 8) (who would go on to co-write The Bell Curve), the UK government began to be increasingly concerned about abuse of the benefits system. As a result, identifying fraudulent benefit claimants became an increasing priority for the government (Mesher, 1981, p. 121).

At the same time, the rise of data processing techniques enabled increasing amounts of data about welfare applicants to be processed, with the stated aim of preventing fraud, in the 1980s (Simitis, 1987). Access to state support is by definition more important for people who are more marginalised: as a result, however, people who are relatively powerful are less likely to be subjected to surveillance and data collection by the state (Koskela, 2012, p. 52).

In the same period, the Conservative government (which came to power in 1979) introduced the idea of conditionality for welfare benefits: benefits which were distributed on the basis not only of need, but also of behaviour (Reeve, 2017). One of the first measures was the introduction of Jobsearch Diaries: records of job-searching activity which claimants had to fill out in order to be eligible for unemployment benefits. These measures gave front-line advisers legal backing to require specific actions from the unemployed, who faced sanctions if they did not comply (Fletcher & Wright, 2018, pp. 227–330).

This combination of increased attention to possible fraud, and increased surveillance of behaviours, continued into the New Labour years. Post-1997 policy continued to blame the post-war welfare state for increasing dependency (V. Noble, 2009, p. 146). The New Labour government focused on incentivising as many people as possible into work, including people—such as lone mothers—who had not previously been incentivised to work (Taylor, 2017, p. 5). The government also took measures to increase compliance requirements for lone mothers (Haux, 2012, p. 2).

The New Labour government spearheaded the use of technological tools, as part of its modernisation agenda. The focus on combating ‘social exclusion’—and getting the ‘socially excluded’ (back) into work—led to an increased effort to define ‘socially excluded’ groups more and more precisely, in order to target them for preventative measures and/or services, and thus to an increased effort to track and map individuals and households which might fall into this group (Pleace, 2007, pp. 947–948). In this period, the government also considered more invasive tools: in 2007, the government proposed (but did not implement) the use of phone-based lie detectors to assess benefit claimants and reduce fraud (V. Noble, 2009, p. 149).

The ongoing programme of increasing welfare conditionality was accelerated by the Coalition government, who came to power in 2010. This government continued promoting the rhetoric of ‘strivers and skivers’—the latest iteration of the idea of ‘deserving’ and ‘undeserving poor’—while pursuing an aggressively neoliberal agenda under which virtually no-one was considered ‘deserving’ of support (Reeve, 2017). The aftermath of the 2008 financial crisis was used to legitimise a programme of austerity measures, which included restructuring the welfare state (Wiggan, 2016, pp. 147–148), shifting from supporting unemployed people to access work to monitoring their compliance with behavioural rules (Wright et al., 2020, p. 291). New measures such as the Universal Jobmatch system allowed officials to remotely monitor the job-searching activity of claimants—and sanction them for not complying with requirements (Fletcher & Wright, 2018, p. 332).

Surveillance and conditionality has continued under the Conservative government, which was elected in 2015. At the same time, the increased focus in data collection and processing has allowed increased use of ‘dataveillance’ in the UK. As of 2017, Chichester Council was using software designed by a company, Xantura, to sort benefits claims—including Universal Credit applications—into low, medium or high-risk categories: the Council streamlined low-risk claims, and applied additional checks to high-risk ones (Chichester District Council, 2017).

The UN Special Rapporteur on Extreme Poverty and Human Rights issued a report after his visit to the UK, in which he stated his findings that all welfare applicants were treated as suspicious and “screened for potential wrongdoing in a system of total surveillance” (OHCHR, 2018). The UK is not alone in this trend: in his 2019 global report, the same Special Rapporteur warned of the risk of a “digital welfare dystopia” characterised by increased surveillance, increased welfare conditionality and increased punishment of beneficiaries who are deemed to “step out of line” (UN Special Rapporteur on Extreme Poverty and Human Rights, 2019, para. 77).

Surveillance, dataveillance, and gender stereotypes

In this section, I argue that surveillance—in particular data-based surveillance—and gender stereotyping act to reinforce each other, creating a vicious cycle in which the surveilled are incentivised to conform, and the non-conforming are increasingly surveilled. I focus in this section on the ways that gender stereotypes and data-based surveillance act to categorise and to control individuals.

Categorisation

Both gender stereotypes and data-based surveillance act as categorisation tools. They assign individuals to categories, and treat them according to their membership of this category. When decisions are made, membership of this category is considered, regardless of whether it is relevant for the decision, or accurate for the individual. For people who have been incorrectly classified—or who do not fit neatly into the existing classification—this can mean that they are subject to harmful outcomes from these decisions, such as (as discussed above) being excluded from employment.

The data that is held on an individual by a public sector agency in the UK is not limited to only that which they have provided through censuses or administrative data collection. Data sharing between public sector agencies gained traction in the early 21^st century as agencies sought to target specific groups more and more closely (Pleace, 2007, p. 948), and under the current Conservative government, there is an expectation—as well as legal provision, under the Digital Economy Act 2017—that local authority agencies providing services to so-called ‘Troubled Families’ will share data unless there is a strong reason not to (Ministry of Housing, Communities & Local Government, 2020, p. 1). There is also an increased interest in using data to create predictive models within local government (Centre for Data Ethics and Innovation, 2020, pp. 7–8): adding additional categorisation for individuals, not necessarily with their knowledge, let alone their consent.

Safiya Noble has argued that categorisation systems are not neutral: they prioritise certain kinds of information, and retain the power biases of the people who designed them (S. U. Noble, 2018, p. 137). As discussed above, categorisation systems also incorporate the social, cultural and historical contexts within which they have been conceptualised, designed, and deployed. The encoding of gendered categories in data and surveillance technologies can, in particular, act to further reify those categories. For example, technologies which aim to automatically categorise individuals as male or female systematically erase trans or non-binary people from their systems (Keyes, 2018, p. 3).

As well as acting alone, however, these systems act to mutually reinforce each other. The encoding of gendered categories in data and surveillance technologies can act to reify those categories. Understanding the parallels—as well as how gender stereotypes and data-based surveillance act together to reinforce inequality–offers up new possibilities for analysis, as well as opportunities for resistance.

Control

Both gender stereotypes and data-based surveillance operate as tools of control: they both normalise certain kinds of behaviours and existences and create a situation in which other kinds of behaviours are punished. In this way, they both act to reinforce and strengthen existing hierarchies of power. In addition, both data-based surveillance and gender stereotypes can be weaponised to control their targets, and to punish behaviour that is deemed ‘wrong.’

As discussed above, David Lyon includes in his definition of surveillance both the act of observation, and the intent to effect change. There is a clear parallel here with the two functions of gender stereotyping: surveillance plays a descriptive role, observing details about an individual—and a prescriptive one, aiming to influence or manage that individual. Anja Kovacs argues that through monitoring what individuals are doing or have done, surveillance exercises a disciplinary power through incentivising certain behaviours and discouraging others (Kovacs, 2017).

Gender stereotypes normalise some roles and behaviours, and implicitly define others as ‘gender non-conforming’—and therefore abnormal and suspect. Participation in society is restricted, to a greater or lesser degree, by the extent to which an individual is prepared to conform to a gender stereotype. This is not a homogenous experience: it is influenced by the extent to which an individual experiences other forms of intersecting discrimination, such as on the grounds of race, religion, class, or disability. Relative power within the matrix of domination (Collins, 2009) may protect an individual to some degree, but failure to conform may result in ridicule, discrimination, or even violence. As a form of social categorisation, stereotypes can therefore be used to protect social norms (Tajfel & Forgas, 2000, p. 58). An individual who does not conform to a stereotype pertaining to them: for example, the woman applying for the firefighter job who has the strength to do it, may be criticised (or worse) for being ‘unwomanly’: in other words, for failing to conform to the standards set for women.

Data-based surveillance may not necessarily operate in the state-organised, totalitarian manner envisioned in dystopian novels like Orwell’s 1984 and Zamyatin’s We. Modern surveillance actors include big tech companies. Shoshana Zuboff has argued that ‘surveillance capitalism’ seeks not only to profit from data collected about individuals by big tech companies, but also to limit the choices that individuals can make (Zuboff, 2019, p. 143). Hille Koskela notes that “interpersonal monitoring”—surveillance— has historically been used to enforce social norms (Koskela, 2012, p. 49). As surveillance technology has become more sophisticated, Lyon has argued that, “what is statistically or organizationally normal becomes the touchstone of what is right or at least appropriate" (Lyon, 2002, p. 249). In other words, data collection, for data-based surveillance, enables the formulation of prescriptive norms: with attendant consequences for those who are not ‘normal’.

Both gender stereotypes and data-based surveillance, therefore, can act as tools of control. With the help of both, certain kinds of behaviours are normalised and incentivised, while others are punished: sometimes with violence. Surveillance technologies and gender stereotypes both set limits on human actions and behaviours, and on ways of existing in the world. This can be through retaining power in the hands of those who already hold it, or through framing certain individuals as in need of ‘protection’ from others. In both cases, both gender stereotypes and data-based surveillance have the heaviest impact on those who are already marginalised.

Surveillance, stereotyping and discrimination for welfare claimants in the UK

In this section, I will argue that the increased use of surveillance in the welfare system in the UK multiplies the extent to which welfare claimants are monitored for compliance with gender norms and stereotypes, risking a vicious cycle in which claimants, fearing discrimination and denial of benefits, are increasingly coerced into compliance. I will examine how this cycle can be harmful at both individual and societal levels, and suggest how human rights can frame and support resistance.

While in this section I focus on the UK, it should be noted that the UK is not alone in using surveillance in its welfare provision in order to enforce certain norms and behaviours. In the US, for example, the US Aid to Dependent Children programme—created in 1935 as part of the New Deal—provided assistance to women with children. It was subject to eligibility requirements, which excluded women who were considered ‘employable’ or ‘sexually immoral’ (Eubanks, 2018, p. 29).

The impact of surveillance of welfare claimants in the UK has been documented both in how it acts to categorise people—implicitly as ‘deserving’ or ‘undeserving,’ as ‘striver’ or ‘shirker’—and has a disciplinary effect, as those surveilled change their behaviour in order to remain entitled to benefits (Wright et al., 2020). However, the categorisation of claimants as ‘deserving’ (or ‘undeserving’) of benefits relies on the extent to which they conform to stereotypes about gender. Over the past century, therefore, welfare surveillance has been used to enforce gender stereotyping: the closer an individual—and their circumstances—conform to these stereotypes, the more likely they are to receive welfare benefits.

As discussed above, early post-war welfare policy considered women as dependents of men. In the 1950s and 1960s, the National Assistance Board used a special investigative unit to surveil claimants. This unit focused on two areas: identifying men with unreported earnings, and women claimants who were cohabiting with a man. In this second category, investigations focused on proving a sexual relationship between the woman claimant and the man with whom she was cohabiting. While only marriage gave women legal rights to support from male partners, the National Assistance Board denied assistance to women cohabiting with a man, regardless of whether they were married (V. Noble, 2009, pp. 63–66). Racist and classist gender stereotypes were used to apply specific scrutiny to women who took in male lodgers, and to West Indian women, both of whom were deemed likely to be cohabiting with men and therefore ineligible for benefits (V. Noble, 2009, pp. 63–65).

By the 1980s, the increased welfare conditionality measures (described above) initially focused on white working-class men, but by the end of the 20^th century also targeted lone parents (who were, and are, mostly women) (Fletcher & Wright, 2018, p. 339). In the 1990s, under the Conservative government led by Major, lone mothers continued to be seen as a “moral problem” which could be addressed through withdrawing benefits and pressuring them to join the labour force (V. Noble, 2009, p. 6). The New Labour government, which increasingly focused policies on families with children, saw marriage as best for these families (Lewis, 2000, p. 99), but continued to incentivise work for lone mothers.

Successive governments targeted specific groups—for New Labour, the ‘socially excluded,’ while for the Coalition and Conservative governments, the ‘troubled families’—who were deemed to require a strong state response, in what Parton has characterised as an ‘authoritarian neoliberal’ state (Parton, 2014). Post-2010 welfare policy has focused on still further increased conditionality which effectively punishes certain behaviours and ways of life: these include a cap on benefits payments per household and a two-child limit for benefits supporting children (Millar, 2019, pp. 89–90).

The introduction of Universal Credit, discussed above, has been characterised as a system which moves entirely away from needs-based benefits, and instead seeks to exert control over behaviour and values. Millar and Bennett have argued that the system ignores the realities of low-income people, arguing that "Universal Credit seems designed to suit the people that ministers believe claimants should become, rather than starting from where they are now" (Millar & Bennett, 2017, p. 175).

Harmful to individuals

Gender stereotypes, and data-based surveillance, are both hard to avoid. But the extent to which an individual can evade many of the harms of both depends on their position in society. As with many other social forces, the harms of both act most strongly against people who are already in a situation of marginalisation.

The use of data-based surveillance in welfare has been documented to cause harm to individuals: Human Rights Watch has documented how Universal Credit calculations are based on data collected over time periods that do not match the periods in which people in work receive their pay, over- or under-estimating their income as a result and leaving claimants struggling when they receive less than what they are entitled to (Human Rights Watch, 2020). Data-based surveillance for Universal Credit is also ‘digital-by-default’, meaning that claimants who are unable to access the internet may be sanctioned for failing to comply with requirements that they are not aware of (Booth, 2019; OHCHR, 2018).

Evading data-based surveillance requires not only technical sophistication, but also a specific position in society: one in which an individual is not dependent on state support and its increasing levels of data collection and monitoring. This is difficult for many people, and impossible for others. Evading the harms of gender stereotyping also requires some power within the matrix of domination: a position in which a person can protect themselves from the consequences of not conforming to expectations based on their legal, social or self-identified gender. Individual rejection of gender stereotypes can put individuals at risk of harm from those who have an interest in maintaining the stereotypes: individual rejection of surveillance technologies can result in individuals being targeted for additional scrutiny, restricted in their activity, or punished as criminals.

Harmful to societal structures

I have described above how both gender stereotypes and data-based surveillance are tools of control: as a consequence, they both act to reinforce and strengthen existing hierarchies of power. This can be through retaining power in the hands of those who already hold it: or through framing certain individuals as in need of ‘protection’ from others.

The exact manifestations of gender stereotypes differ around the world and through time, but gender stereotypes are a remarkably consistent presence in different cultures. Seguino has argued that stereotypes and norms—including gender stereotypes—reflect underlying systems of power relations: these include those set up and maintained by institutions that contribute to “gender hierarchical attitudes” (Seguino, 2011). According to Cook and Cusack, gender stereotypes can also be deployed to subjugate individuals, or to maintain existing hierarchies of power (Cook & Cusack, 2010, p. 17). Gender stereotypes in particular frame women as subservient and inferior, which facilitates unequal power relations along gender lines (Holtmaat, 2004, p. xii).

An individual who fails to conform to prevailing gender stereotypes may be outright rejected by the society in which they live, and risks being coerced into one of the pre-determined ‘boxes’, regardless of the harm that may result. The unspoken result is a wider consequence to this failure to conform. As Holtmaat has argued, gender stereotypes may form a crucial part of national identities, particularly for women, whose bodies may form the site for these identities, and whose roles can be framed as “foundational” for a state (Holtmaat, 2013, pp. 117–118). Yuval-Davis argues further that gender relations are crucial for promulgating a “specific view about the meaning of the world and the nature of social order,” and that the ideas of a ‘proper man’ and ‘proper woman’ form an important part of this worldview (Yuval-Davis, 1997, p. 67). That is to say, failure to conform not only undermines the individual or their immediate community, but may undermine the nation state itself. Individuals who fail to conform may be ostracised, met with violence, or even killed, in the name of ‘protecting’ the nation (Yuval-Davis, 1997, p. 46).

Surveillance technologies, almost without exception, are controlled by those who already hold some form of power, including state actors and powerful corporate entities. Data-based surveillance, for example, requires data collection, storage and analysis capabilities which increasingly rest only in the hands of large technology companies. These companies may promote surveillance’s benefits to users, in the form of customer convenience, such as the provision of ‘personalised’ services. Benefits to the company, which are less public, include the creation of profitable data products, which can be sold to other actors (Zuboff, 2019, pp. 255-262). Profit accrues to those in power within these companies: which are not known for their diversity of staff, let alone their decision-makers (Myers West et al., 2019). The problems caused by homogenous groups building systems that impact on the lives of people who are not like them are well-documented (UNESCO & EQUALS Skills Coalition, 2019, p. 126).

It is not only true that gender stereotypes and surveillance technologies reinforce each other: it is also true that in so doing, they exacerbate existing hierarchies and inequalities. Gender stereotypes can cause real harm to individuals: through denying recognition of an individual’s worth and dignity, and through denying fair allocation of resources (Cook & Cusack, 2010, p. 59). Surveillance technologies that reinforce and amplify these stereotypes only exacerbate this harm.

As discussed above, both gender stereotypes and the deployment of surveillance technologies are underpinned by structures that benefit those already in power. Those who benefit from gender stereotypes may be the same people who determine the design and deployment of surveillance technologies: those who are not subjected to surveillance in order to receive state support may be able to avoid widespread behaviour policing.

Resisting surveillance and resisting stereotypes

As I have argued in this article, there are parallels in the way that gender stereotypes and surveillance technologies operate in society. Both of these systems act to categorise individuals and to control behaviours. Both act to reinforce hierarchies and exacerbate inequalities in interoperable ways. These parallels, however, also offer opportunities for resistance. In the Netherlands, the SyRI risk assessment model was used to calculate the risk of benefits fraud, but a Dutch court found that the invasion of privacy was not justified and that it discriminated on the grounds of socioeconomic status and migrant status (Henley & Booth, 2020).

The methods developed by legal and feminist scholars like Cook and Cusack for identifying gender stereotypes and how they operate to discriminate, may also prove useful for identifying increasingly hidden and surreptitious surveillance and dataveillance methods. Understanding how gender stereotypes may cause harm—how they degrade, diminish, deny benefits to individuals, and impose burdens (Cook & Cusack, 2010, pp. 60–65)—may help to understand how surveillance technologies cause harm, and may provide a useful lens for interrogating these technologies.

At the same time, understanding how surveillance technologies operate and particularly how dataveillance is capturing more and more information about specific details of individual lives, can help to interrogate gender stereotypes. While it is true that many of these systems remain opaque, protected by intellectual property laws, more and more stakeholders are pushing for transparency in data systems. Understanding how these systems categorise and quantify human lives may help surface the ways that these categorisations act in other areas. Holding the systems to account may help hold the societies that build these systems to account in turn.

Conclusion

The welfare system in the UK has historically been designed to favour certain, stereotypical, family models. While the nature of these stereotypes has shifted since the establishment of the post-war welfare state, including increasingly recognising women as citizens in their own right, the post-2010 welfare reforms are designed for the circumstances of a stereotypical couple and their children, undermining women’s financial independence and ignoring the realities of life for people on low incomes. At the same time, increased conditionality for welfare benefits—which allocate benefits on the basis of compliance with behaviours, instead of according to need—has led to increasing surveillance of welfare claimants, increased use of data, data sharing and data-based surveillance to monitor claimants.

The combination of gender stereotyping and surveillance in the UK welfare state risks creating a vicious cycle, in which the categorisation and control dimensions of both stereotyping and surveillance reinforce each other. This risks coercing welfare claimants—by definition, people living on low incomes—into certain ‘accepted’ behaviours, and discriminating against those who do not conform. The increased conditionality of welfare benefits has already caused demonstrative harm to those who cannot or struggle to access Universal Credit. The coercive, surveillant nature of the welfare state risks cementing hierarchies of power which continue to stereotype and discriminate against low-income people, in particular low-income women who are expected to balance the demands of their disproportionate unpaid caring responsibilities as well as increasing requirements for job search activities.

Against this backdrop of coercion and conditionality, however, applying a human rights analysis—including recognition of the harms of gender stereotyping, as recognised by the CEDAW Committee—allows for the specifically gendered nature of the harm caused by surveillance and conditionality to welfare benefits claimants. UK law has already recognised the harm that can be caused by rigid assumptions in the welfare benefits system (Earnshaw, 2021). I argue that applying analysis of gender stereotyping can further identify—and combat—harms that are inherent in the current structure of the welfare benefits system in the UK, with the aim of ensuring that benefits are accessible for all who need them.

Acknowledgements

This paper benefited from discussion and feedback at the Forum Privatheit Workshop on Feminist Data Protection on 20 November 2019 in Berlin, as well as from feedback from Professor Lorna McGregor. I am grateful to the reviewers, Kirstie Ball and Tobias Matzner, as well as to the Special Issue editors and the managing editor, for their comments and suggestions which substantially strengthened this paper.