Timebanking and the co‐production of preventive social care with adults: what can we learn from the challenges of implementing person‐to‐person timebanks in England? Health and Social Care in the Community, Timebanking and the co-production of preventive social care with adults; what can we learn from the challenges of implementing person-to-person timebanks in England? research methods used at each case study exploring timebank context and process

This paper explores the potential contribution of timebanking, an innovative volunteering scheme, to the co-production of preventive social care with adults in England. Interest in volunteering in social care has increased as one proposed solution to the international crisis of a rising demand for services in juxtaposition with decreased resources. Volunteering has been particularly promoted in preventive services that prevent or delay care needs arising. Despite sustained interest in volunteering and co-production in social care, little is known about how theory translates into practice. Reporting implementation data from a Realistic Evaluation of six case studies in England, this paper explores one volunteering scheme, timebanking. The research explores how timebanks were working, what contribution they can make to adult social care, and whether they are an example of co-production. Data collected included interviews, focus groups or open question responses on surveys from 84 timebank members, and semi-structured interviews with 13 timebank staff. Each timebank was visited at least twice, and all timebank activity was analysed for a period of 12 months. Data were triangulated to improve reliability. The research found that in practice, timebanks were not working as described in theory, there were small numbers of person-to-person exchanges and some timebanks had abandoned this exchange model. Timebanks faced significant implementation challenges including managing risk and safeguarding and the associated bureaucracy, a paternalistic professional culture and the complexity of the timebank mechanism which required adequate resources. Lessons for timebanks are identified, as well as transferable lessons about co-production and volunteering in social care if such schemes are to be successful in the future.


| BACKG ROU N D
Preventive services and the use of volunteers are linked solutions to address a resource crisis in social care in England; prevention reduces demand, whilst volunteers are thought to provide additional resources (Bottery, Ward & Feeney, 2019). Prevention in social care encompasses services that prevent or delay support needs arising, and include low-level support services that help people remain independent and improve well-being. Prevention was made a requirement for social services in England in the 2014 Care Act, but chronic resource shortages have undermined the translation to practice (Bottery, Ward, & Fenney, 2019). One proposed solution is to use low-cost support, such as communities, unpaid family carers or volunteers (McCall et al., 2020).
The community and voluntary sector has long formed a part of delivering social care services in the mixed economy of welfare in England, but the focus intensified since the 'Big Society' policy of the Coalition Government in 2010 (Dickinson, Allen, Alcock, Macmillan, & Glasby, 2012), which promoted increased community involvement to support public services (Alcock, 2010). Although the Big Society narrative has 'quietly faded' (Gibson, 2015), encouraging volunteering in social care has continued (Cameron, Johnson, Willis, Lloyd, & Smith, 2020), but despite sustained interest there has been little exploration of how volunteering in social care works in practice (Dickinson et al., 2012;Mountain, Gossage-Worrall, Cattan, & Bowling, 2017). This paper explores one such scheme, timebanks, which claim to deliver preventive social care by providing low-level services such as practical support with cleaning and shopping, otherwise unavailable through statutory services, through a mutual aid network (Timebanking UK, 2017).

| Volunteering in social care in England
Volunteers are currently supporting statutory, private and voluntary sector delivered social care in England through providing peer support, befriending, respite care, counselling, support with administration and fundraising, outreach and advocacy (Naylor, Mundle, Weaks, & Buck, 2013;Paylor, 2011). Usually, a volunteer is recruited and managed by a host such as a voluntary sector organisation or state funded service (McCall et al., 2020). Volunteering is widely credited with positive outcomes for both the provider and receiver of support including improved physical and mental health, improved quality of life and reduced isolation (Naylor et al., 2013;Smith, Drennan, Mackenzie & Greenwood, 2018). However, a systematic review found little robust evidence about the effectiveness of volunteers in social care and more research is needed to explore volunteering in practice (Mountain et al., 2017). Boyle, Crilly, and Malby (2017) found that for volunteering to be effective, solutions need to be found for a range of barriers which include recruitment, retainment, training and supervision. Recent research that has explored volunteering in social care found that rather than saving organisation's money, sufficient time and resources were required to supervise and support volunteers, and protect vulnerable service users Smith, Drennan, Mackenzie, & Greenwood, 2018).

| Timebanks and co-production
One form of potentially low-cost volunteering promoted in social care is timebanking, an innovative tool where people are paid in time to volunteer to support members of their community, which they can then use to buy support for themselves, functioning as a mutual aid network (Cahn, 2004). The nature of support is varied and ranges from sharing a skill such as baking or IT skills, to practical help such as moving furniture or helping with shopping. Each hour can either be spent or banked for future need. Timebanks have been implemented with different aims across various international contexts (Gregory, 2015). One approach is to fund timebanks attached to statutory, private or voluntary sector services to create networks that support their delivery. This is the dominant approach of those seen in England (Boyle & Bird, 2014;Gregory, 2015), Japan (Hayashi, 2012) and the USA (Cahn, 1986(Cahn, , 2004 and has been considered a form of co-production, defined as user involvement in delivering health and social care outcomes (Needham & Carr, 2009).
Advocates of timebanks claim that they offer additional benefits to volunteering through operating within this co-production framework, which has become an important theme in contemporary social care policy (Flemig & Osborne, 2019). Co-production is thought to improve user outcomes through an asset or strengths-based approach (Cahn, 2004). Volunteers in social care usually operate within What is known about this topic • The government in England aims to increase volunteering in social care, particularly in preventive, low level services.
• Timebanks are one form of volunteering and are claimed to offer additional benefits such as co-production.
• Little is known about how timebanks can deliver social care in practice

What this paper adds
• Person-to-person timebanking is an attractive idea that promises co-production. However, it is difficult to make timebanks work in England.
• Timebanking exchanges were limited by time consuming regulation, a professional paternalistic culture, and the complexity of the timebanking mechanism which required adequate resources.
• There are transferable lessons about challenges facing volunteering in social care, such as the importance of providing sufficient resources to support volunteering schemes. a 'gift model' where those in need are supported in a paternalistic relationship that mirrors professional support (Bransford, 2011;Duffy, 2010Duffy, , 2013. This has been criticised for being disempowering, creating a dependency on the care provider and eroding capacity for self-help. Timebanking offers an alternative; a reciprocal exchange breaks down the dichotomy of provider and receiver by encouraging participants to support each other, which acts as a 'multiplier' of well-being outcomes for participants who are no longer only defined by their support needs (Timebanking UK, 2017). Co-production approaches are considered to require less resources when compared to traditional approaches to social care (Bovaird, Flemig, Loeffler, & Osbourne, 2019;Needham & Carr, 2009). In theory, a co-production approach to volunteering offers a reduced cost model; advocates of timebanks claim that they form a less resource intensive, citizencontrolled community scheme that functions as a mutual aid network, without the need for professional intervention and associated costs (Evans, Hills, & Orme, 2012).
The potential for timebanks to deliver low-cost community co-produced, preventive social care services have led them to be described as 'an idea whose time has come' (Cahn & Gray, 2015). In England, timebanks have been supported as a mechanism for people to provide and access practical support, thereby mobilising communities to deliver preventive services not currently funded through statutory services (Boyle & Bird, 2014;Ryan-Collins, Stephens, & Coote, 2008;Timebanking UK, 2017). Timebanks experienced a wave of popularity and were included in the government vision for transforming adult social care in England (Department of Health, 2012), as an intervention to build community capacity (Public Health England, 2015) and are included in the Social Care Institute for Excellence (SCIE) Prevention Library (2020). Timebanking UK (2017), a national umbrella organisation established in 1998, promotes timebanking as a tool to transform social care through mutual aid.

| Evidence for timebanks and co-production
Despite interest in timebanking in social care there is as yet little evidence of its effectiveness. Two international evidence reviews that explored timebanking in health (Lee, Burgess, Kuhn, Cowan, & Lafortune, 2020) and social care (Naughton-Doe, 2015) demonstrated a shortage of good quality research studies; the majority of the research considers one case study, with unclear or unreliable methods. Significantly, there is little discussion about how timebanks facilitate co-production (Clement et al., 2016). There is however, evidence of implementation challenges for timebanks that specifically aimed to support social care services. In the USA, evaluators of a timebank that aimed to provide support to elderly social service users through building mutual aid networks concluded that problems with recruiting members, safeguarding vulnerable participants and organising exchanges meant that timebanking was difficult and time consuming for staff (Dentzer, 2003;Feder, Howard, & Schalon, 1992;RWJF, 2000).A scheme in Japan that aimed to provide care for elderly people faced similar problems; organising timebanks proved challenging and members struggled to exchange services due to being frail or vulnerable (Hayashi, 2012;Nakagawa et al., 2011). In England, an early evaluation reported that contrary with timebanks supplying additional resources, timebanks were resource intensive and vulnerable to funding shortages (Seyfang & Smith, 2002). A preoccupation with risk and safeguarding had limited the capacity of timebanks in England to create exchanges and build relationships (New Economics Foundation, 2001;Panther, 2012).
These challenges faced by timebanks are also experienced by coproduction approaches in social care; a review of the evidence found that barriers included a risk-averse approach in services prevented professionals giving up control, and co-production approaches being more resource intensive than expected (Flemig & Osbourne, 2019;Needham & Carr, 2009;Voorberg, Bekkers, & Tummers, 2015). To undertake a detailed assessment of how timebanks are working and their outcomes, Lee et al. (2020) called for further robust evaluation.
With the likely continued interest in volunteering in social care, and co-production, this paper presents research from 2012-2015 that explored the potential contribution of timebanks and identified transferable future lessons for similar schemes.

| RE S E ARCH ME THODS
This paper explores data about timebank implementation collected during a Realistic Evaluation of timebanks in England (Naughton-Doe, 2015). Realistic Evaluations typically collect data about intervention contexts, mechanisms, outputs and outcomes, to generate context specific theories about how interventions work, in what circumstances, for whom and why (Pawson & Tilley, 2004). This paper reports findings relating to timebank implementation from six case studies for the purpose of generating theories about how timebanks were implemented in different contexts and the challenges that they faced. The research questions addressed in this paper are: 1. How are timebanks in England working? 2. Are they facilitating co-production?
3. What can they contribute to preventive social care?

| Recruitment and sample
Six timebanks were recruited as case studies following an advertisement in a national umbrella organisation, Timebanking UK.
Interested timebanks applied to be part of the study and the final sample of six timebanks was selected to represent a range of contexts, including a GP surgery and a timebank that worked with older people. The sample differed in size, age and location, and included well-established and start-up timebanks (see Table 1). At each timebank, staff were interviewed, including time bank 'brokers', who directly organised activity and exchanges, and timebank managers, who supervised them. In total, 21 semi-structured interviews were conducted with 10 brokers and 3 timebank managers both in person and on the phone (see Table 2). Timebank members (people who had signed up to the timebank) were recruited either through convenience, advertisement or random sampling, depending on timebank capacity (see Table 2) and 83 timebank members shared their experiences: 36 timebank members were interviewed, 15 participated in two focus groups and 32 gave feedback in open questions on a survey (see Table 3).

| Methods
At each case study, a range of methods were used to explore timebank context and process (see Table 3). Interviews conducted with staff and members were supplemented with observational data and informal conversations recorded in field notes. Each timebank was visited at least twice, e-mail exchanges were saved and nine events were observed. Internal documents such as funding applications, policies and procedures, website content and reports for funders were reviewed. Twelve months of output data, including the number and type of exchanges, were collected from Time Online, the Timebanking UK trading software.

| Analysis
Data were first prepared for analysis; sections of interviews with staff and members that were relevant to timebank implementation were transcribed verbatim. Transcriptions, field notes and notes from document analysis were then entered onto Nvivo, coded and context-specific theories about how timebanking was working and why. Four key elements of timebank context that appeared to influence timebank implementation and activity were identified (see Table 4).

| Research ethics
Risks to participants were minimal as the interviews focussed on timebank activity rather than sensitive or personal topics. The biggest risk identified for the timebanks was reputational damage. All organisations and participants were required to give informed written consent and their details were anonymised to preserve confidentiality. Ethical approval was granted by the University of Bristol School for Policy Studies Ethics committee in 2011.

TA B L E 3
Interviews and surveys with timebank members at each case study Timebank code Sampling method Activity levels of members in the sample Interview method (n = 51)

TB1
Random sample of active members (n = 10) One member was very active, the remainder had engaged occasionally.

TB2
Random sample of active members (n = 7) All members had interreacted with the timebank in the year Face-to-face semi-structured interviews (n = 7) No surveys were collected TB3 Members recruited via advertisement (n = 7) The group interviewed were very active members who regularly attended group activities Focus group (n = 7) n = 1

TB4
Members recruited via advertisement (n = 8) The majority exchanged several times a year Focus group (n = 8) n = 1

TB5
Random sample of all members (n = 9) Majority were casual users who had exchanged once or twice, one had never traded and one was a regular trader.
Face-to-face semi-structured interviews (n = 6) Telephone semi-structured interviews (n = 3) Members recruited via the broker (n = 10) Half were active and the other half exchanged once or twice Face-to-face semi-structured interviews (n = 10) n = 19

| FINDING S
This section will first explore how timebanks were implemented before reviewing some implementation challenges.

| How timebanks worked in practice: similarities and differences
Timebanks are presented to offer a community co-produced model of volunteering which differs from those managed by a professional (Boyle & Bird, 2014). However, a common approach across all timebanks was that they were organised such that all the activities were Comparative analysis identified four themes where timebanks differed that appeared to influenced the amount and type of activity timebanks produce: these were, how they were funded, whether they had targets, how many staff they had, and their approach to managing risk and safeguarding (see Table 4) Timebanks that received funding from external funders usually had targets such as the number of members they had to recruit, or the number of hours of activity they were expected to create, whereas those who received funding internally from their host organisation, did not. Some timebanks had more staff resources than others which determined how much time was available to support activity. Whilst all timebanks had policies in place to protect their members, some timebanks had a rigid, formal approach to risk assessment and sometimes actively supervised members during exchanges; these timebanks were risk-averse. Two timebanks had a more positive risk-taking approach -they managed risks using flexible, informal guidelines and rarely supervised member exchanges.

| Timebank activity, members and exchanges
Timebanking UK (2017) claimed that timebanks could be a mutual aid network that could deliver low level services in social care through person-to-person exchanges. The analysis of data from Time Online, the timebank software, found that in practice, person-to-person exchanges accounted for less than a quarter of activity across all timebanks, and five percent or less at three case studies (see Table 5). Some timebanks organised lots of activity, but across all timebanks there were only a small number of person-toperson exchanges which accounted for an average of 5 hr of activity a week. Two timebanks (TB2 and TB3) had higher proportions of person-to-person exchanges, and this was due to brokers investing a significant effort into facilitating a small amount of exchanges, whereas other timebanks had mostly abandoned the timebank mechanism. Rather than being especially innovative, the majority of hours earnt and spent by timebank members either involved volunteering at a local organisation (e.g. with administration), or by attending courses (e.g. cooking or crafting) and group activities (e.g. quilt making, social events or trips) organised by the timebank.
Timebanks also involved a small number of people, with an average of 51 active individual members where active was defined as people who participated at least once in the preceding 12 months (see Table 5). This was different to the higher numbers of people reported to be members by timebanks during recruitment to the study (see Table 1), which ranged up to 1,000 members. One reasons for this gulf between the number of members and active members was that other than at TB2, all timebanks included any person who had ever signed up to the timebank in their number of members, even where they might have never exchanged. TB5 with 1,000 members had been open to the longest at 15 years and so had 1,000 members, but only 66 active individuals, though this number was likely to be higher as the timebank had stopped using Time Online systematically. Another reason for the higher number TA B L E 5 Timebank members and person-to-person exchanges over a 12-month period Timebank members also generally gave far more than they received; brokers reported it was difficult to encourage people to 'spend' their accrued hours. Consequently, rather than facilitating reciprocal exchanges, the timebank relationships were quite similar to more traditional one-directional volunteering.
Whilst there were some examples of a few members providing shopping, dog walking or even a visit following being discharged from hospital, the timebanks in this study did not match up to the low cost, self-sustaining mutual aid model in the way described by advocates of timebanking such as Boyle and Bird (2014) and Timebanking UK (2017): timebanks were mostly broker controlled and were organising limited amounts of person-to-person exchanges. The remainder of this paper will focus on exploring why timebanks struggled to organise person-to-person exchanges in practice.

| Managing risk and safeguarding
Brokers were asked in interviews why the number of person-toperson exchanges were so limited; the most prevalent explanation was the time-consuming process of managing risk and safeguarding.
A consequence of the intended users of timebanks being vulnerable people was that brokers understandably adopted the concerns of social care organisations to protect their members. Timebanks had to police check new members which involved an application to the police force to check that the individual had not committed an offence that precluded them from being suitable for volunteering; these checks are notoriously time consuming. Brokers also had to collect background references, and risk assess activity to comply with health and safety law in England and public liability insurance.
The associated paperwork took time, which limited the time that staff had to organise exchanges. One broker explained the process of risk assessment for arranging a dog walker: We would assess the house, assess the dog, assess the owner, have a chat, and once we found the walker, introduce the two of them together and go out for a walk with them as well just for a safety measure. We have a dog walking agreement, an off-lead agreement if the owner thinks it's appropriate for the dog to be off the lead, then they sign an agreement.
And we're sort of an intermediary if that dog walker can't make it on the day, they ring me, and I ring the person. (TB1) Risk and safeguarding management also limited what could be exchanged. For example, to comply with insurance or host organisation policies, some timebank brokers stopped activities they considered risky such as making food, offering childcare or DIY. For example, the broker at TB2 explained: I refused an exchange on the grounds that it was above one-story height. So, because falls from heights are so common, we have a little rule that you can only go one story high on a ladder within the timebank.
A consequence of the time-consuming and limiting process of risk management was that some brokers stopped organising person-to-person exchanges. Instead, they focussed on group activities which could bypass the need for police checks and lengthy risk assessments if the brokers supervised all activity.
In some cases, the concern of brokers to protect their members had created a culture of paternalism that went beyond risk assessment. Many brokers perceived that the vulnerable nature of the intended users of timebanks facilitated the need for additional support. For example, one broker commented: A lot of our participants need hand holding. Erm, and I don't think that's a bad thing … we work with people with mental health problems, learning difficulties, substance abuse problems … all that kind of stuff, a lot of people that do get referred to us are people who need that extra bit of support. (TB3) This resulted in an approach where brokers spent considerable time supporting members in ways that resembled a professional support worker role in social care. For example, some brokers supervised members at timebank exchanges rather than allowing members to exchange on their own. One broker explained: Every time two people meet each other for the first time, we will go along, and that does take up a lot of time … It's all about confidence, just that kind of reassurance that it's happening properly, you know, they've got to understand what they need to do. (TB6) This paternalistic 'hand holding' contravened the asset-based approach discussed in timebanking theory. The consequence was that timebanking exchanges became more labour intensive if brokers felt they had to support their members, and this limited the number of exchanges that could take place.

| Timebanking mechanism was complex and resource intensive
Person-to-person timebanking exchanges sound simple and spontaneous, but in practice, arranging two individuals to come together with a support need, the ability to help and the time Others reported that some requests, for personal care, such as hoovering, help with shopping or cleaning, were left unanswered; brokers perceived that this was because people did not want to do these tasks: Personal care isn't an attractive prospect is it-there is one lady … she was hoping to receive back some vacuuming as she is a wheel chair user … I put an advert in the newsletter asking if someone would vacuum, but nobody has. (TB5) Another issue was that many members did not request services, but only wanted to offer help. Without people asking for help, timebank activity slowed down, which the broker at TB2 referred to 'timebank constipation'. Where there were higher proportions of person-to-person exchanges, the broker reported a time-consuming process of 'engineering' exchanges to encourage members to participate.
The broker explained: When the timebank has been particularly quiet … I will say to someone who needs to earn credits, how about you bake a cake for someone? And I will say to someone with low credits, how about you have a cake? So I will engineer things, even though they might not have a birthday where they need a cake, I might try and engineer it. (TB2) This engineering process was laborious and it was not possible to organise many exchanges, which explains why TB2 was so small, but with a higher proportion of person-to-person exchanges.
All brokers agreed that person-to-person exchanges were resource intensive and activities at the timebank largely depended on how many staff the timebank had. One broker reflected: [ Brokers at TB2 and TB3 continued to prioritise organising person-to-person exchanges, but free from external targets, accepted that the amount of timebank activity would be limited. Timebanking UK (2017) claimed that timebanks are a transformative mechanism to build a mutual-aid network that delivers low cost, preventive social care (Boyle & Bird, 2014 ). This study did not find evidence of timebanks delivering this support in practice; most timebanks were not offering any low-level support such as cooking, shopping or transport. Person-to-person exchanges were time consuming to organise, resulting in some timebanks largely abandoning the mechanism, and others acknowledging that the time taken to facilitate exchanges resulted in a small amount of activity. Timebanks had small numbers of members and there were often insufficient or unanswered requests which resulted in the majority of members being inactive. These findings support those from studies in the USA (Dentzer, 2003;RWJF, 2000) and Japan (Hayashi, 2011;Nakagawa, Laratta, & Bovaird, 2013) that concluded that significant implementation challenges, including difficulties with recruiting active members, the time-consuming complexity of arranging exchanges, and the resulting resource demands, meant that timebanks could not be relied upon to deliver a cost-effective, consistent service in practice.

| D ISCUSS I ON
Timebanks have been claimed to be a way of facilitating co-production, which is thought to offer additional resources for social care (Evans et al., 2012) and create more benefits for service users when compared with traditional volunteering (Boyle & Bird, 2014). Timebanks in this study did not appear to be facilitating co-production due to barriers arising from a perceived necessity to protect their members. Far from being member-led, a risk-averse culture meant that brokers had the sole responsibility for establishing rules, arranging exchanges, risk assessment and safeguarding. Members could not access their own Time Online accounts or exchange without broker support. This was ironic because timebanks were originally conceived as schemes where people could engage in self-help without professional intervention (Cahn, 2004). Instead, the safeguarding policies recreated paternalistic professional dependency by devolving sole responsibility onto the broker and rather than challenge the professional-centric paternalistic model; timebanks seemed to replicate it. This study supports findings from Voorberg and et al., (2015) and Needham and Carr (2009)  advocates of timebanks have claimed that they would offer a less resource intensive co-production model of volunteering (Timebanking UK, 2017), it was apparent that timebanks were not immune to complications, and actually added a further layer of complexity through the person-to-person exchange mechanism. Cameron et al. (2020) has argued that one of the biggest challenges of using volunteers in social care is ensuring that there are sufficient protections for the vulnerable service users that social care services have a duty to protect; time brokers faced similar challenges.
Timebanks in social care have not yet lived up to their potential with few person-to-person exchanges taking place. If timebanks are to be successful, lessons can be learnt from the two timebanks that did facilitate higher proportions of person-to-person exchanges. What made these timebanks work was that free from external funding and related targets, they were less risk averse than the other timebanks, and more able to prioritise organising person-to-person exchanges, albeit in small numbers. Timebanks in other countries, such as New Zealand, have also had more success facilitating person-to-person exchanges because they have encouraged a less risk averse approach where members have greater control over their exchanges (McGuirk, 2017). If person-to-person exchanges are to become more widespread in England, one solution would be for timebanks to adopt this informal, positive approach to risk-taking approach. A new timebank system developed and launched by TBUK in 2014 has a function to allow members to manage their own timebank accounts, but at the time of this research, unsurprisingly, brokers were concerned about the risk implications. It is a difficult balance to manage risk without curtailing freedom, and this research has highlighted some challenges which question the appropriateness of timebanking in social care given the vulnerable nature of social care service users and the types of support required. The success of timebanks in New Zealand is in part because they were not implemented to deliver social care, but to build community relationships, and are therefore less limited by stringent regulations associated with protecting vulnerable members.

| Limitations of this study
The data in this study was collected

| Conclusion
The research explored timebanks as a tool for the co-production of preventive social care in England. With increasing interest in volunteering in social care, advocates of timebanks claimed they offered a value-added approach creating a co-produced mutual-aid network delivered with reduced resources compared to professionally managed schemes. In practice, timebanks in England were struggling to offer this due to implementation challenges including managing the safeguarding of vulnerable service users, and issues with translating timebanking theory into practice. As of 2020, only three of the timebanks in this study remain open, and two of those have abandoned the person-to-person mechanism in favour of other community development approaches. This research supports findings from the USA and Japan that real problems exist for timebanks providing mutual aid in social care. If timebanks, and other volunteering schemes in social care are to be successful, issues of safeguarding and providing adequate resources for managing the schemes must be addressed.

CO N FLI C T O F I NTE R E S T
The authors whose names are listed above have no affiliations with or involvement in any organisation or entity with any financial interest or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter discussed in this manuscript.