Red flags but no yellow cards: The ‘WelCond’ research on welfare conditionality and unemployed people

Peter Dwyer, Lisa Scullion, Katy Jones, Jenny McNeill & Alastair B. R. Stewart, The Impacts of Welfare Conditionality: Sanctions, Support and Behavioural Change. Policy Press. 2022. pp. 218. £80 (hbk). ISBN: 978-1447320111. £26.99 (pbk). ISBN: 978-1447343738. £26.99 (ebk). ISBN: 978-1447343745

Behavioural conditions have always been attached to the receipt of social security benefits for the UK's unemployed. These conditions have increased in number and scope in recent decades, alongside heavier sanctions for non-compliance, as part of a trend across OECD countries (Watts & Fitzpatrick, 2018). The UK Welfare Reform Act 2012 was a landmark in this process. Its “work search requirement” obliges claimants of unemployment benefit – now Universal Credit (UC), previously Jobseeker's Allowance (JSA) – to take “all reasonable action” and “any particular action specified by the Secretary of State, for the purpose of obtaining work” (s. 17). In practice this means that a Department for Work and Pensions (DWP) ‘Work Coach’ can now compel an unemployed UC claimant, under threat of a financial penalty, to apply for a job of their choosing. Those who leave a job voluntarily and then apply for UC can now also face a “sanction” (s. 49).

Heralding the 2012 Act, its architect, the then Secretary of State for Work and Pensions Iain Duncan Smith (2010), argued in a speech that “reinforced conditionality” to ensure claimants “take reasonable offers of work” was necessitated by the benefit system's regrettable drift towards one that “Beveridge warned against”, in which “idleness” had become “institutionalised” (see Beveridge, 1942, p. 58). In the same speech, he claimed that in the 2000s British companies had been “unable to get British people to fill” some job vacancies, so “workers from overseas stepped in”. Around this time, Duncan Smith referred to a television documentary in which some unemployed benefit claimants would not get “on a bus” to a nearby city to widen their job search (BBC, 2010). The issues Duncan Smith referred to may be persisting; there were 5.6 million UK people of working age on out-of-work benefits in December 2023, just below the record high of 5.9 million in 1993, despite 0.9 million unfilled job vacancies (see Andrews, 2024).

This review article focuses on the £2 million Economic and Social Research Council (ESRC)-funded ‘Welfare Conditionality: Sanctions, Support and Behaviour Change’ research programme (often abbreviated to ‘WelCond’). Led by Peter Dwyer, a Professor of Social Policy at the University of York, and involving academics from six universities, it ran from 2013 to 2018, just as the 2012 Act's new policies were being slowly rolled out across the country. The project's main purpose was to test the effectiveness of the new policies in relation to their intended aim of changing claimants' behaviour. The project's recently published flagship book, The Impacts of Welfare Conditionality, brings together the WelCond project's main findings and conclusions. While the project's sample size was large enough to enable the inclusion of meaningfully sized groups of all social security claimant categories affected by conditionality, this review article focuses on its coverage of unemployed people who would have previously claimed JSA. I argue that important research questions were left unanswered. Furthermore, because Dwyer and his team are against welfare conditionality in principle, they offer no meaningful policy recommendations to the newly elected Labour government, which is almost certain to retain the policy of compelling unemployed working-age state benefit claimants to seek employment.

Section 2 provides an introductory description of the dominant paradigm in UK social policy academia, and it explains how this has guided the study of welfare conditionality and unemployed people. Section 3 describes the WelCond programme's aims, research design, findings and conclusions, as reported in The Impacts of Welfare Conditionality. Section 4 provides a critique of the book. Finally, Section 5 consolidates the main arguments and makes suggestions for improving academic debate about unemployment and welfare conditionality.

It is well established that the academic discipline of Social Policy, where discussion about welfare conditionality and unemployment often takes place, has long been heavily dominated by the political left (Deacon, 2002; Dunn, 2014; Welshman, 2012). Alan Deacon (2002), a former Journal of Social Policy editor, noted that Social Policy was dominated by a close-knit group of high-ranking left-wing academics who were heavily involved in political campaigns. This group espoused most, though not all, of the views of Social Policy's founding father, Richard Titmuss, who was Professor of Social Administration (as the subject was then called) at the London School of Economics until his death in 1973. Deacon (2002, pp. 22–6) suggested that what he called the “quasi-Titmuss paradigm and school” had dominated the discipline since the 1980s. This ‘paradigm’ fits the landmark description of the term by Kuhn (1962), as it contains assumptions that guide research design and academic debate, while arguments that challenge it and empirical evidence that contradicts it are marginalised and discredited. Its assumptions include the following: that welfare provision should be generous, universally available, and, of most relevance here, unconditional. Furthermore, it assumes that poverty should be defined and measured purely as relative to national average material living conditions – an idea traceable to Karl Marx's writing (see Dunn, 2022). This carries the implication that those on state unemployment benefits, whose incomes usually place them below the conventional relative poverty line (set at 60 per cent of median income), are victims not only of unemployment but also of poverty.

Importantly, quasi-Titmuss school accounts strongly emphasise ‘structure’ (economic, social, and political institutions and processes) at the expense of ‘agency’ (the individual's capacity to exercise meaningful choice). Twentieth-century UK social scientists understandably drew inspiration from the work of influential US sociologist C. Wright Mills (1959, p. 9), who chose the issue of unemployment to illustrate his view that public issues were often wrongly assumed to be a “private trouble” by those who, because they lacked what he called a “sociological imagination”, downplayed the importance of structural forces. In the UK, under the mentoring of Titmuss, Adrian Sinfield (1968) presented evidence supporting the view that unemployment was a structural problem. By the 1980s, mainstream UK Social Policy writing had extended Titmuss's aversion to blaming poverty on subcultures or individuals' dispositions into a “denial of agency”, which “precluded any discussion of such factors” (Deacon, 2002, p. 14). Those who mentioned not only structures but also diverse human traits when explaining unemployment were castigated for ‘victim blaming’. For example, Dee Cook (1997, p. 112) accused Lawrence Mead, who is widely considered to be the guru of 1990s US welfare reform, of “victim blaming” in an Institute of Economic Affairs publication edited by Deacon (1997). Deacon (2002) later noted that mainstream Social Policy authors did not respond in detail to Mead (1982, 1986) or the other highly influential US conservative author Charles Murray (1984) because their work was considered beyond the pale.

The quasi-Titmuss school and paradigm is personified by Peter Townsend, the celebrated socialist writer on inequality (see Deacon, 2002, p. 30). Townsend routinely failed to acknowledge any merit in his political opponents' arguments. Referring to what he called “liberal custodians of inequality” who disagreed with him about the meaning of poverty and who made mention of individual or subcultural characteristics when accounting for why some people and not others were below the relative poverty line, he stated “it is important not to mince words with this traditional school of thought. It is as destructive as it is unscientific” (Townsend, 1993, p. 4). Consistent with Deacon's (2002) analysis, Ruth Levitas (2005, pp. 9–12) cited Townsend as the key figure in what she called the ‘RED’ (a telling abbreviation of ‘redistribution’) discourse on social exclusion, which sees reducing inequality as a panacea for eradicating problems endured by relatively deprived and marginalised citizens.

The quasi-Titmuss paradigm/school's assumptions and the ‘RED’ discourse remain identifiable in much twenty-first-century UK Social Policy writing about unemployment and social security. Shildrick et al.’s Poverty and Insecurity (2012) is now discussed in some detail, as it is both a good illustrative example of the paradigm and school and an important precursor to Dwyer et al.’s research on unemployment and labour market behaviour. Shildrick et al. presented findings from 60 in-depth interviews on Teesside, England, with people aged 30 to 60 who had moved to and from low-status employment and reliance on out-of-work state benefits. When it won the British Academy's Peter Townsend prize, one of the judges – Professor Sara Arber – said it was “in the tradition of Peter Townsend's writing” (Teesside University, 2013). She was not wrong. A key conclusion was that, given that the 60 “loved” working (Shildrick et al., 2012, pp. 8, 136), “the total demand for labour determines employment, and that is essentially the entire story” (p. 36). Hence, the findings seemingly helped justify the quasi-Titmuss school/paradigm's remarkably heavy emphasis on the role of structural forces in determining individuals' outcomes.

Yet some individuals damaged their employability by taking heroin (pp. 85–88) or accumulating criminal convictions (pp. 85, 86, 92), and some left “jobs voluntarily or were sacked” (p. 195). The authors repeatedly explained such reported behaviours as the constrained choices of powerless individuals near the bottom of an unequal society. It is not known whether those from similar social backgrounds who behaved differently were more likely to have longer employment records, because there was no comparison of the two groups. The research project on which the book was based also included interviews with ten local employers and 13 employees of local agencies who work with unemployed adults. However, their accounts, which included the view that many young adults showed little interest in entering and retaining low-status jobs, are virtually absent from the book. While the 60 other interviewees' accounts are accepted uncritically, the authors, in a way reminiscent of Townsend's dismissive attitude towards the ‘liberal custodians of inequality’, suggested such views are symptomatic of those respondents' “deeply embedded … ideological stance on worklessness” (p. 75). Similarly, an earlier book, also co-authored by Robert MacDonald, presented the accounts of young ‘socially excluded’ adults on Teesside while ignoring agency worker interviewees' views. MacDonald and Marsh (2005, p. 44) acknowledged that their decision reflected “a political preference to prioritise the accounts of young people themselves”.

Shildrick et al. (2012) did not comment on the possible implications of their wholesale dismissal of the views of agency employees and employers. This is remarkable, given the noted strengths and limitations of accounts provided by both them and unemployed benefit claimants (see Dunn, 2014, pp. 172–4 for a more detailed discussion). Most importantly, these staff sometimes exhibit prejudices and misapprehensions but tend to possess detailed knowledge of their clients; while claimants can draw upon all relevant circumstances they face, they sometimes fear that being candid runs a risk of being reported to the DWP. The view that significant numbers on Teesside were unwilling to undertake low-status jobs is consistent with a glut of evidence showing that UK employers tend to regard economic migrants as more willing than the UK-born to undertake such jobs (see Fitzgerald & Smoczyński, 2017). Quantitative work has also found that migrants' in-work absenteeism rate is lower than that of UK-born workers (Dawson et al., 2018).

Remarkably, the aversion of mainstream Social Policy authors to ‘victim blaming’ can lead them to draw different conclusions depending on which ‘victim’ group they are attempting to shield from blame. Unlike those studying benefit claimants, authors who specialise in ‘race’ and ethnicity often accept there is a serious gap between the employment commitments of migrants and non-migrants. For example, Gary Craig (2008, p. 232) observed that Britain was “happy to accept workers from elsewhere, to help fill the low-paid, dirty gaps in the labour market that the majority (usually white) residents are unwilling to take on”. My own research on people's actual choices between claimant unemployment and low-status jobs (Dunn, 2014) also exposed a diversity of dispositions and a widespread dislike of low-status employment that is usually overlooked in Social Policy accounts. Yet Tracy Shildrick remains convinced that the agency workers and employers in her Teesside study were wrong and, therefore, that increased welfare conditionality was unnecessary. In her more recent book, she states that “where people can possibly find work – even that of very poor quality and pay – they will choose paid work over ‘welfare’” (Shildrick, 2018, p. 67); this generalisation reminds me of a comment by the late Frank Field, a moderate Labour member of the UK parliament who was chair of the House of Commons Work and Pensions Committee. Field (1997, p. 61) wrote that left-wing Social Policy academics often imply that benefit claimants are “immune from the faults of laziness and dishonesty”.

Section 3 introduces Dwyer et al.’s book, before section 4 argues that the quasi-Titmuss paradigm is strongly identifiable in it.

The book begins by contrasting the positions of those who are for and against welfare conditionality. It notes that “those who oppose the use of welfare conditionality” regard it as “unfair, intentionally punitive, ineffective in promoting paid employment and personal responsibility and likely to exacerbate the poverty and social exclusion of already disadvantaged populations” (p. 1). Project leader Peter Dwyer (1998) is referred to alongside six other authors who express this view. Their position is contrasted with that of four “advocates of welfare conditionality” who assert that “sanctions and support are both a fair and effective way to move people off welfare and into work” (p. 1). The four listed are Lawrence Mead (1982), former prime minister Tony Blair (2006), Iain Duncan Smith (2014) and me (Dunn, 2014). Then the two core aims of the project are described; they are “First, to develop an empirically and theoretically informed understanding of the role of welfare conditionality in promoting and sustaining behaviour change … [and] … Second, to consider the particular circumstances, if any, in which the use of conditionality may, or may not, be considered ethically justified” (p. 2).

The main findings from the qualitative longitudinal project are as follows: No fewer than 186 respondents reported being ‘sanctioned’ (p. 119), many of whom felt wronged, sometimes because their punishment was for a relatively minor offence such as being late for a meeting with their Work Coach. The present system of welfare conditionality and sanctioning was “routinely ineffective in facilitating people's movements off social security benefits and entry into, or progression within, the paid labour market over time” (p. 157). There was much “work–welfare recycling”, which means that interviewees moved back and forth from insecure employment to claimant unemployment. The authors note that “occasional, sustained movements off welfare and into work were evident, but these were extremely rare” (p. 157). The structure of “local labour markets” determined opportunities, so “respondents' ability to access or increase their paid work routinely had little to do with their behaviour”. Conditionality and sanctioning were generally “experienced as unnecessary” by interviewees (p. 16) as they were already committed to seeking employment.

The researchers noted that, rather than reintegrate them into employment, the financial sanctions were far more likely to push claimants further towards illness, criminality and destitution (see particularly pages 129–30). Furthermore, the authors identified what they call “counterproductive compliance” with strict rules on job-search behaviour leading some claimants to spend 35 hours per week in futile job-search activity to satisfy their Work Coach's demands purely to avoid a sanction, rather than use their time to do what they believed would maximise their chances of entering employment (pp. 125–8).

The second key aim of the project (see above) is the subject of chapter 7: ‘Ethical Debates’ (pp. 137–54). Four contrasting “ethical” positions” on “citizenship rights” and “the state's right to intervene” in people's lives are identified. These are “contractualism, paternalism, mutualism and (unconditional) entitlement” (p.137). In support of ‘(unconditional) entitlement’, Dwyer (1998) is cited as expressing the egalitarian view that welfare conditionality is unjustifiable because it is only imposed on those “who are experiencing poverty while largely ignoring the irresponsibility and inactivity of wealthier citizens” (p. 142). In contrast, interviewees generally agreed with the basic principle of welfare conditionality that has long underpinned UK social security policy – that, in return for state benefits, those capable of work and of working age have a responsibility to seek employment (pp. 147–52). The ‘(unconditional) entitlement’ ethical position is most fully endorsed by the authors in the last two sentences of the book's final chapter: “Welfare conditionality is counterproductive, ineffective and unethical. It is therefore time to end the obsession with behaviour change and focus on promoting meaningful employment support, genuine social security and greater equality” (p. 160).

Dwyer and his colleagues accept that their opposition to welfare conditionality in principle is out of step with public opinion (as evidenced by their own respondents) when they lament that “it would be over-optimistic to predict the demise of welfare conditionality” (p. 156). While these authors' opposition to conditionality does not invalidate their work, it is unhelpful that they do not suggest any ideas for improving a policy which they accept is here to stay. More specifically, they say nothing about how claimants' concerns might be balanced with those of taxpayers; rather, the reference to “wealthier citizens” (see section 3) implies an egalitarian view that those who express concern about taxpayers' possible frustrations should redirect their critical gaze towards structural inequalities.

Yet much of the time the authors do not openly acknowledge their own ideological position. Discussion of the principles surrounding welfare conditionality is couched as a discussion about ‘ethics’. Unlike Economic Affairs, a journal that openly states in every printed edition that it is “broadly inspired by classical liberalism”, the WelCond researchers tend to present themselves as unbiased, scientific experts. However, their partiality became visible when Peter Dwyer appeared before the House of Commons Work and Pensions Committee on 20 June 2018.ⁱ Dwyer spoke out against WelCond respondents being unexpectedly ‘sanctioned’ for a first rule breach. In response, Conservative member of parliament Nigel Mills suggested a ‘yellow card’ could be issued for a first offence, with only a second offence incurring a financial penalty. This exposed Dwyer's view as one based on principle rather than any evidence his team had uncovered; appearing ruffled by Mills' suggestion, he retorted that he was against sanctioning under any circumstances.

Importantly, The Impacts of Welfare Conditionality fails to acknowledge how the research team's ideological commitment might have influenced the way it investigated the issue and how it interpreted and presented its findings. The need to connect ideological positions to empirical findings when studying social security has long been stressed by Lawrence Mead (see Beem & Mead, 2005; Mead, 1988); yet Dwyer and his colleagues seem oblivious to this, despite mentioning Mead's publications on 16 separate occasions. Mead (1988) argues that your ideological position guides what behaviour you demand from unemployed benefit claimants. Conservatives like him think welfare claimants should consider applying for “all legal jobs” and, echoing Iain Duncan Smith's ‘bus’ comment, even apply for “far away” jobs (Mead, 1988, p. 48). Furthermore, Mead points out that ideology is important in determining who we hold responsible for an individual's employability (and hence his/her likely employment status); “liberal” (in the US political sense) commentators blame society if people have, for example, criminal convictions, addiction problems and limited educational experience/qualifications, whereas “conservatives tend to deny these workers deserve jobs” (1988, p. 50).

Mead's comments show the overlap between the quasi-Titmuss paradigm/school's emphasis on structure over agency in explaining social phenomena and a political philosophy that attributes blame on the same basis. Like Shildrick et al. (2012) (see section 2), Dwyer and his team exemplify both Mead's understanding of the ‘liberal’ position on unemployment and the quasi-Titmuss school's extremely heavy emphasis on structural explanations. They never mention the possibility of people leaving their local area if job opportunities are limited there, and they do not insist that people cease drug use to improve their employment chances (p. 109). Furthermore, they accept that suffering from depression is enough to justify permanent absence from employment (p. 74), despite much evidence indicating that employment can help tackle it (see Wilson & Finch, 2021). Dwyer et al. claim Mead (1986) sees “individual failings of benefit claimants” as a “key cause of unemployment” (p. 20), just because he discusses both structural constraints and individual traits; in fact, Mead blames social security systems' permissiveness for people's long-term reliance on out-of-work benefits (see Deacon, 2002, pp. 49–61). Thus, Dwyer and his fellow researchers ignore one plank of Mead's contribution while misrepresenting another.

Interviewees' accounts are uncritically accepted by Dwyer and colleagues, who do not discuss the likelihood that people's descriptions of their behaviour can reflect a desire to present themselves positively (see Hollway & Jefferson, 2012). Respondents do not appear to have been quizzed on whether they could have done more to escape unemployment, and there is no discussion of whether some might have entered employment if they had been less choosy in job search. Nor is there any discussion about possible diversity in their willingness to undertake low status jobs – despite such diversity being a key finding of my in-depth interviews (Dunn, 2014).

Indeed, authors of an earlier WelCond publication seemed to forget that a key aim of the 2012 Act is to stop unemployed claimants being too choosy in job search:

Readers of The Impacts of Welfare Conditionality might wonder how many of the shockingly high number of sanctioned claimants were punished for being too choosy. The authors do not address this. They instead devote much space to presenting claimants' accounts of scandalously unfair sanctioning decisions.

Dwyer and colleagues, like Stewart and Wright (2018), do not reflect on the possible implications of either the absence of Work Coaches from their study or the absence of focus group and stakeholder interview data from their presented findings. As was noted in section 2, favouring one kind of respondent over another can skew the pattern of findings and conclusions, yet Dwyer and his colleagues, like Shildrick et al. (2012), do not reflect on this. Nor do they discuss findings from published UK studies of the views of professionals who have unemployed clients. These studies include my interviews which exposed welfare-to-work professionals' widely held belief that many of their clients were too choosy in job search (Dunn, 2013), and Rahim et al.’s (2017) DWP study of the impact of the new conditionality policies on welfare claimants' behaviour, which included interviews with 124 Work Coaches. The absence of this discussion cannot be attributed to The Impacts of Welfare Conditionality focusing solely on presenting WelCond findings, as its bibliography includes over 500 items.

Discussion of Rahim et al.’s (2017) research would have enabled Dwyer and his fellow authors to compare their findings and methodological choices with those from a similar large-scale study. Like the Dwyer team, Rahim et al. (2017) found that some welfare claimants engaged in many hours of pointless job search just to avoid a possible sanction. However, in contrast to Dwyer et al., the DWP study found that the toughened conditionality “drove a number of positive behaviours”, such as “more diversity in jobs applied for”, which led to “entry into jobs that claimants would have previously ruled out” (Rahim et al., 2017, p. 63). This difference in findings and conclusions might be explained by the DWP report's clear recognition (in contrast to Dwyer et al.) that a key aim of UC was to “diversify the types of roles” claimants applied for (Rahim et al., 2017, p. 3). Rahim et al. (2017, p. 30) exposed varying levels of employment commitment among claimants:

Rahim et al.’s (2017) inclusion of 124 DWP Work Coaches also seems a plausible explanation of why they uncovered excessive choosiness while Dwyer et al. did not. Observing interactions between DWP staff and unemployed clients in a previous research job made me well aware of their concern that unemployed clients are too choosy; I remember one encounter in which a young unemployed woman said ‘really I want to be a dancer’, which was met with the immediate, sarcastically toned response ‘well I want to be an astronaut’.

The Dwyer research team does not reflect on the implications of their project's lack of quantitative research for the pattern of findings it produced. Nor do they attempt to help fill gaps in our knowledge by referring to quantitative findings from other studies; for example, I found that longitudinal survey respondents who expressed negative attitudes towards low-status jobs spent significantly more time in unemployment over a 30-year period (Dunn, 2021). If the WelCond project had included its own quantitative analysis, it could have found out what percentage of those who arrived a little late for an appointment were sanctioned, or what percentage of rule breaches led to a sanction and how this compared with the percentage who were sanctioned despite not breaking any rules. Likewise, it could tell us how common it is for UC claimants to spend as much as 35 hours per week in job-search activity. Instead, the book highlights scandalous cases, which could be unrepresentative outliers, while failing to estimate their frequency.

Quantitative research can make comparisons between unemployed claimants' attitudes and behaviours and those of others. Table 1 presents findings from my analysis of all data gathered by British Social Attitudes surveys over the last 20 years about attitudes towards welfare conditionality and sanctioning; while these findings concur with Dwyer et al. that unemployed claimants are broadly committed to welfare conditionality, they are less likely than others to express that commitment. Findings like this perhaps imply the need for stricter social security rules, to ensure the overall UK population's behavioural standards are adhered to by unemployed benefit claimants. Regardless, the WelCond project's lack of statistically representative research and its lack of comparison between claimant and non-claimant groups are serious limitations on which Dwyer et al. do not reflect.

Dwyer et al.’s WelCond research project is worthy of some praise. It unearthed a perverse incentive problem whereby current policy can encourage claimants to waste considerable amounts of time engaging in futile job-search activity. Furthermore, it highlighted the effect that scandalous sanctioning decisions can have on claimants' well-being and their future employment chances. Nevertheless, because Dwyer and his colleagues support abolishing welfare conditionality – which, as they acknowledge, is highly unlikely to happen soon – they propose no ideas for reforming the present system. Allowing sanctioned claimants to have their benefits restored if they attend a full-time work placement might help avoid pushing benefit claimants further into the mire while appealing to the hard-working taxpayers often championed by more conservative authors. Indeed, this idea might garner support among mainstream politicians, including those in Keir Starmer's Labour government, which, at the time of writing, has not yet committed itself to specific changes regarding the nature of conditionality and the severity of sanctions. Another possible policy suggestion is to determine the level of job choosiness claimants are entitled to by examining their long-term employment/unemployment histories and their educational/training attainment levels.

Dwyer and colleagues do not discuss job choosiness, either in relation to their own findings or anyone else's. Indeed, at times it seems they do not grasp that guiding benefit claimants away from being excessively choosy in their job-search behaviour is an important policy aim. In contrast, Rahim et al.’s (2017) similar, large-scale DWP study investigated job choosiness and concluded that the post-2012 policies were successful in encouraging claimants to broaden their job searches. Like Shildrick et al.’s (2012) evidence-based book, The Impacts of Welfare Conditionality provides a classic example of what Deacon (2002) called the ‘quasi-Titmuss' paradigm's ‘denial of agency’: both books constantly imply that individuals have virtually no control over their employment status. Moreover, neither book comments on how the absence of either quantitative research findings or the views of those who deal with unemployed clients might have affected their overall pattern of findings and their conclusions. Perhaps most importantly, neither group of authors properly differentiates between their evidential and ideological objections to current policies.

Until mainstream/left-wing Social Policy authors properly engage with their critics' arguments and evidence, acknowledge the limitations of their own research, and reflect on how their methodological decisions might have impacted on their findings and conclusions, their work will rightly continue to be treated with caution by those who do not share their ideological position. The issues identified in this review article should serve as a red flag to policymakers, who should view these one-sided accounts with the utmost scepticism.