University of Birmingham The Goldilocks Question: What Size is 'Just Right' for Social Care Providers?

Here, we apply the ‘Goldilocks’ question to social care: what size of care provider is ‘just right’? Empirical research to date has struggled to find evidence for an optimal size for public service providers, although policymakers remain keen to suggest that size is a key aspect of organisational performance. The article makes an innovative contribution to this literature, drawing on empirical research with care providers and people who use their services in England. Findings from 143 interviews with people using different-sized care services suggest that micro-organisations (employing five staff or fewer) achieve better outcomes for their cost base than larger organisations, although our study is necessarily exploratory rather than statistically definitive. The salience of size in a social care setting provides a basis for hypothesising that organisational size may be more significant in relation to care than it has been found to be in broader public management literature, though research with larger and more robust samples is needed.


Introduction
The 'Goldilocks' 1 question has been applied to a range of settings, from bowls of porridge to the size of the universe (Davies, 2007). Here, we apply it to the provision of care and support: what size of provider is 'just right'? Are care services best delivered through small, locally based services that are 'close to the user' or through larger organisations with greater economies of scale? These are essential questions, both in the UK and internationally, as care provision becomes more marketised and large chains, backed with private equity, establish a stronger presence (Burns et al, 2016).
In the public management literature, there is often an assumption that 'smaller is better' because small organisations may be more flexible and innovative, achieving better outcomes for workers and consumers (Schumacher, 2011;Locality, 2014), although large organisations are also known to have advantages in terms of adaptability and durability (Harrison, 1997). As 'person-centredness' becomes a guiding principle of health and care services not just in the UK, but internationally (World Health Organization, 2015), and a growing number of countries introduce self-directed approaches into care (Glasby and Dickinson, 2009;Grit and de Bont, 2010;Alakeson, 2010;Christensen and Pilling, 2014;Dickinson et al, 2014), it may be that small local organisations are best placed to deliver this person-centred support (Bickerstaff, 2013). However, it is also possible that the countervailing pressures on care budgets at a time of rising demand will incentivise larger organisations offering economies of scale and deploying a flexible range of skills and resources without compromising care outcomes (Bubb and Michell, 2009). Within English social care, it is possible to find strong support for the principles of small-scale provision (HM Government, 2007). However, for many local authorities, the financial context of social care has been a countervailing force, encouraging a reliance on framework contracts with a small number of large care providers who satisfy the need to meet growing demand on a reduced cost base (Rubery et al, 2013: 423).
This article presents findings from the first major study to explore whether the size of social care providers affects the quality and cost of the support provided. It is based on a two-year Economic and Social Research Council (ESRC)-funded study examining the contribution of micro-enterprises to social care in England. Here, micro-enterprises are defined as organisations that employ five staff or fewer. The research compares these organisations with small (6-25 employees), medium (26-99 employees) and large (over 100 employees) providers of comparable care services to understand better the relationship between organisational size and performance. The article sets out different approaches to defining size, findings from existing studies and the methods used in the research before presenting findings on three measures: care outcomes, person-centredness and value for money. The final section of the article considers the sustainability of small care providers within local care markets.
While there are many different dimensions to the quality of care services, there are two main reasons for focusing on size as a potentially key variable. First, the international literature on public service performance and improvement is underdeveloped and inconsistent on this issue, making it difficult for policymakers, public service commissioners and service providers themselves to know what to do for the best. According to Boyne's (2003a: 383) meta-review (described in more detail later), 'The results of eighteen studies that have tested for a linear relationship between organizational size and performance … offer little comfort to the advocates The Goldilocks question 67 of large or small organizations'. While more resources and better management can improve performance, the review concludes that changes to other variables (including size) are 'largely a shot in the dark and could equally lead to poorer rather than better performance' (Boyne, 2003a: 390).
Second, there are long-standing literatures and debates within care services themselves around the extent to which size matters, with almost diametrically opposing views and trends. Whereas the Gershon (2004) review of public sector efficiency led to a series of mergers in back-office functions and regional approaches to procurement, advocates of the personalisation agenda have emphasised the importance of highly individualised approaches and micro-commissioning (Glasby and Littlechild, 2016). While findings from England's health and social care regulator, the Care Quality Commission (CQC, 2017), suggest that smaller services caring for fewer people are rated better than larger services, the trend within the care home industry has been one of an ongoing concentration of the market, with a small number of very large companies dominating provision (see, eg, Holden, 2002;Laing and Buisson, 2010;Institute of Public Care, 2014). Similarly, there have been moves towards very small, community-based services for people with learning disabilities, alongside policies that are believed to be creating pressure for the recreation of larger, more institutional forms of care (Jackson, 2017). As but one example of this, the Winterbourne View care scandal caused national consternation, not just because of the abuse that was uncovered, but also because of the extent of regionally based, out-of-area, institutional provision that was also revealed. As the government response argued: Winterbourne View also exposed some wider issues in the care system. There are far too many people with learning disabilities or autism staying too long in hospital or residential homes, and even though many are receiving good care in these settings, many should not be there and could lead happier lives elsewhere. This practice must end. We should no more tolerate people being placed in inappropriate care settings than we would people receiving the wrong cancer treatment. (DH, 2012: 5) In the same way, the UK National Health Service (NHS) has witnessed ongoing debates about optimum size, ranging from attempts to create smaller-scale, more local services versus policies to create more regionally focused economies of scale, with many recent NHS reforms seeming to fall somewhere on a spectrum between 'small is beautiful' and 'bigger is better.' As a very stark reminder of the importance of the issues at stake (dating back to the very creation of the NHS in 1946-48), the founder of the NHS, Aneurin Bevan, famously observed that: Many of the hospitals are too small -very much too small. About 70 per cent have less than 100 beds, and over 30 per cent have less than 30. No one can possibly pretend that hospitals so small can provide general hospital treatment. There is a tendency in some quarters to defend the very small hospital on the ground of its localism and intimacy, and for other rather imponderable reasons of that sort, but everybody knows today that if a hospital is to be efficient it must provide a number of specialised services. Although I am not myself a devotee of bigness for bigness sake, I would rather be kept alive Jon Glasby et al 68 in the efficient if cold altruism of a large hospital than expire in a gush of warm sympathy in a small one. (Bevan, 1946) Against this background, the present research is believed to be the first major study of size as a key variable in organisational performance within a social care setting (although see the Discussion section later for the extent to which size may also be a proxy for other key elements of service quality).

Size as a variable
The relationship between size and performance has had a high salience in the international public management literature since the emergence of new public management (NPM)-type approaches 30 years ago. Whereas traditional Weberian theories of bureaucracy prized the economies of scale and coordination offered by large organisations, NPM approaches were inspired by public choice theory and saw small organisations as more results-focused and responsive to the public (Hood, 1991). While public management fashions have ebbed and flowed, international reviews of public service providers have found a highly contingent relationship between size and performance (Sheaff et al, 2003;Peckham et al, 2005). Boyne, in a meta-analysis of sources of public service improvement, looked at linear and non-linear models of the relationship, concluding from 23 studies that organisational size is not a significant determinant of performance: 'almost as many tests indicate that performance at first falls with size and then eventually rises as indicate the reverse.… Thus whether reformers are better advised to break up large public agencies or amalgamate small ones remains unclear' (Boyne, 2003a: 385). Although based within the UK, Boyne's review is truly international, including 54 studies from the US, six from the UK, two from Sweden and one each from Australia, Finland and Holland, and also looked across sectors (including education, health care, housing, police and firefighting). Interestingly, none of the studies were of social care.
In the care sector, the previous two decades in the UK can be characterised as periods in which care services have been proceeding in the direction of larger service providers. The care management reforms in social care in the early 1990s (see, eg, Means et al, 2008) that facilitated the outsourcing of care services to the private and third sectors have, over time, led to a dominance of private sector providers, offering domiciliary care (help with washing, dressing and food preparation) or residential care (Skills for Care, 2015). Data from the social care national skills council demonstrate the shape of the sector, with the vast majority of providers still concentrated at the smaller end of the sector: 35% of organisations have fewer than five full-time equivalent members of staff, whereas only 6% have more than one hundred (Skills for Care, 2015). However, very large providers (with over 250 full-time equivalent staff) -which are only 2% of all care providers -employ almost 45% of the care workforce (Skills for Care, 2015). The CQC State of care report in 2015 reported that, in a highly challenging financial context, medium-sized providers were closing and growth was coming from new large providers (CQC, 2015: 17).
There is some evidence of an inverse relationship between size and performance in the care sector, including CQC (2015, 2017) data on the relatively poor performance of large care homes, Skills for Care (2013) data highlighting high staff turnover in large organisations, and concern from the Public Accounts Committee (2011) about The Goldilocks question 69 the viability of large care chains. Research on residential care for people with learning disabilities has also drawn attention to the poor quality of large-scale provision (ODPM, 2003;Gordon et al, 2010). However, there is little empirical evidence that explores the mechanisms through which scale and performance intersect. Literature that links care outcomes to organisational size is largely theoretical. There is a well-developed literature that focuses on the interpersonal nature of care, and the importance of sustained relationships, which draws on theoretical insights linked to an ethic of care and makes an implicit or explicit link between small-scale care, continuity and good outcomes (Barnes, 2012;Lewis and West, 2014;Barnes et al, 2015). The growth of large-scale private care providers, particularly in the domiciliary sector, is critically reviewed in this literature for utilising a model of care that is antithetical to the ethic of care (Barnes, 2012). Similarly, advocates of self-directed support draw on practical critiques of current service models and theoretical insights around social justice and citizenship to argue that decisions about people's care should be made as close as possible to the individual concerned, ideally by the person themselves or, if this is not possible, someone who knows them well and really cares about them (see, eg, Glasby and Littlechild, 2016;Duffy, 2016). For Needham (2015), personalisation is premised on the value of the small scale because it requires interpersonal care and strong relationships that might be difficult to achieve at large scale with large staff teams.
In this sense, size may be a potential proxy for other features of care that service users are known to value. However, even here, we know little about what impact size has. For example, a very small organisation with only a few staff may be better than a large multinational company at providing continuity of care. However, it is equally possible that a small service provider could have so few staff, relative to the level of need, that continuity is problematic, and that a larger organisation with greater economies of scale might be better placed to match the right staff with the right people using services (eg with a small team able to work with a small number of service users and thus provide a greater degree of continuity if someone is ill or on holiday). Thus, size seems to be a key -if poorly understood -potential variable, and the current study seeks to address a previous lack of evidence on the role of size when organising and delivering social care services.

Care on a small scale
The micro-enterprise care sector is increasingly visible as such enterprises are expected to contribute to three strands of English welfare policy: the personalisation of public services, particularly care services (DH, 2010a); government plans to develop the world's largest social enterprise sector as part of what was, for a time, called the 'Big Society' (Alcock, 2010;Cabinet Office, 2010;DH, 2010b); and the need to make care services viable in a context of austerity (Power, 2014). The Department of Health and Social Care (previously known as the Department of Health or 'DH') has been keen to promote micro-providers of social care, based (it appears) on a perception that small organisations are 'closer to the user', and therefore more responsive than larger ones (DH, 2010a). This fits into the personalisation agenda, implemented by successive UK governments, which promotes person-centred care for people with support needs (HM Government, 2007;Fisher et al, 2012). Personalisation is particularly associated with devolved funding: eligible people being given an allocation Jon Glasby et al 70 of money to cover their support (a 'personal budget'), which they can either take as cash (a 'direct payment') or have managed on their behalf (Roulstone and Morgan, 2009;Needham, 2011).
As the DH explains: The providers of micro (very small) social care and support services are independent of any larger or parent organisation and truly stand alone in a sector of work that is complex, challenging and demanding. Many providers deliver the service themselves, alone or as a couple, and employ no staff to help them. Others employ a small number of staff or work with volunteers or members of their extended family in order to deliver the service. For the purposes of our work we have defined micro providers as those working with no more than 5 paid or unpaid full-time equivalent workers. Most providers of micro services are happy to provide services on a very small scale. This may be because they are committed to supporting one or two individuals; because they believe they can better retain control of their enterprise if it remains small; because they equate small scale with high quality and user led or because they want to work from home and in their community. The majority are not aiming to develop their enterprise in order to support more people or to expand into a different area. (DH and NAAPS, 2009: 12) Although not always the case, some micro-enterprises may be run by people who have cared for a family member with experience of using services, by people who have previously used services themselves or by people frustrated at working in more traditional services and seeking greater flexibility to deliver better care. Irrespective of this, micro-enterprises tend to be characterised by the belief that very small services can be more flexible, individual and caring. According to the 'Small Good Stuff' website, for example, micro-enterprise is based on the assumption that: • People want care and support at times and in ways that suit them.
• Local people helping other local people is good for everyone and for communities.
• Very small organisations can offer great care and be very imaginative and responsive (see https://www.smallgoodstuff.co.uk/; for further definitions and examples of the ethos of micro-enterprise in action; see also Box 1) The Care Act 2014 gave English local authorities a duty to act as market shapers, developing a diverse local market of care providers so that people can have choice about what services to use. As part of this, and the broader personalisation agenda, a number of local authorities have been working to increase the number and viability of micro-enterprises delivering care services in their area. Proponents of micro-forms of service delivery argue that very small care organisations outperform larger care providers in a number of ways, drawing on vignettes of successful cases to support their claims. In our study, we used these claims to help guide the research and investigated a number of themes, which we list as the following hypotheses: Hypothesis 1: Micro-enterprises deliver more valued outcomes for users than larger organisations (NAAPS, 2010).

The Goldilocks question 71
Hypothesis 2: Micro-enterprises are better than larger organisations at delivering services that are personalised to the individual (HM Government,

Box 1: What are micro-enterprises (Needham et al, 2015, p 4)
Micro-enterprises (also called micro-providers) are very small organisations delivering social care services that employ five or fewer staff (full time equivalent). They are usually independent of any larger organisation and are offered by a range of people and organisations in the community, including people who are disabled or need support themselves... Micro-enterprises vary widely from each other. Some micro-providers employ staff (even ten or more on a part-time basis) or work with volunteers, whilst others are sole traders, working on their own. Some are set up as social enterprises (including Community Interest Companies) or charities; others are limited companies. Our evidence indicates that those who set up micro-enterprises generally aim to make enough out of their venture only to pay the wages of those involved and subsequently, they can be classed as social enterprises or social businesses...

Examples of Micro-Enterprises include:
Micro Domiciliary Service: Full Lives was set up by Janet who worked within local authority care services, and wanted the opportunity to provide a more flexible offer. It now has four members of staff, supporting three people. The work varies between personal care in the home, and support to access activities outside the home. Janet says, 'Because we are a small company we can be more flexible, at the hours people want. We don't have a lot of clients so we get to know the people we work with. You can build up strong relationships.' Micro Day Support: Pam runs a day service with six part-time staff. Their small size allows them to support the social integration of people with learning disabilities, including Pam's daughter. Pam explains 'I set up Woodlands because day centres were closing … and there wasn't anything [else].' Woodlands is based in a semi-rural community, and aims to connect people to the community: 'our members go out and get recognised by the shopkeepers and people and they develop relationships with folks in the community,' says Pam.
Micro Accommodation: 'Our House' is a micro-enterprise providing shared accommodation to men with learning disabilities. The owner provides low level support beyond the landlord role, including social support and activities, as one tenant explains: 'He does things out of his way to get you out the house. He'll … come round once a week and just check in with us … and he's very protective as well, because say if you wanted a mobile contract … he goes "right, well give me about two days, I'll look it up", he'll come back with a better deal.' Micro Support in the Home: Barbara works on her own, providing help in the home to 14 people in her local area. The support she provides is very flexible, from preparing food to cleaning out cupboards and taking people to the doctors or to concerts. Hypothesis 3: Micro-enterprises deliver better value for money than larger organisations (NAAPS, 2010): 'Micro-enterprises can often offer lower cost services, particularly when a highly tailored solution is desired, because they have few management and overhead costs' (NHS Confederation, 2012).
The research examined how 17 micro-enterprises performed on these measures, compared with 10 small, medium and large organisations, in three case-study sites in England, speaking to 143 people who work in or use these services. To our knowledge, this is the first study to compare the costs and outcomes of different-sized care organisations utilising a validated outcome measure, the Adult Social Care Outcomes Toolkit (ASCOT). Micro-enterprise, in particular, is under-researched as a form of service delivery. The over-representation of micro-enterprises in our sample reflects the dual aims of the research: to better understand the micro-enterprise sector; and to undertake comparative research with small, medium and large care providers. We acknowledge that this sample cannot provide definitive statistical results for testing each of the hypotheses. Our aim is to make a start in this debate, providing evidence where we can, but noting that this research is largely exploratory and needs to be confirmed (or not) with larger samples drawn in a more representative manner. The focus on size and performance is crucial for current policymakers, commissioners, service providers and service users. It comes at a time when the advent of personal budgets is enabling individuals with support needs to make more meaningful choices about who provides their care, and when there seems to be uncertainty in policy circles about whether smaller forms of provision are cheaper or more expensive than larger ones. We are, of course, aware that a sample of this size cannot yield definitive statistical tests with high degrees of 'certainty'. Nevertheless, we draw on evidence of different kinds to help secure validity and present new evidence into the lacunae that currently exist.

The research study
The research project was a multi-site evaluation of care organisations, comparing micro-enterprises (with five staff or fewer) with small (6-25 employees), medium (26-99 employees) and large (over 100 employees) organisations across the statutory, charitable and private sectors. While several different definitions of micro-enterprises and micro-businesses exist (based on different staff numbers or annual turnover), we used a definition of care micro-enterprises taken from publications by the DH and Think Local, Act Personal (DH and NAAPS, 2009). Micro-enterprises are defined as local organisations with five or fewer paid or unpaid workers, which are set up to meet the needs of an individual or small group and are independent of any other organisation (Community Catalysts, 2011). In the case of large organisations, we measured size according to the number of staff working in a particular local authority, rather than across a national chain. To allow comparability with private sector providers, the research only included organisations with a trading income, i.e. that charge for their goods or services; this excluded some statutory and third sector providers.
Mixed-methods (qualitative and quantitative) research was carried out with 17 micro-care providers, comparing them with four small, four medium and two large providers. Case-study organisations were located in three areas of England. Working in three case-study sites allowed comparative analysis without spreading the research work too thinly (Yin, 2013). The sites were selected: (1) to differ from each other in their regional/demographic profiles; and (2) in areas with a known network of micro-enterprises, as advised by Community Catalysts, the national umbrella body for micro-enterprises.
Research was focused on organisations providing care services for older people and/or adults with learning disabilities. We expected these to fall under one of three headings: domiciliary care (ie personal care provided in the home of the person needing the service), residential care and group day services. Initial research in our three localities showed, however, that many micro-enterprises offered what we called 'one-to-one support'. This was a flexible service, covering a wide range of tasks in the home and community. It differs from a personal assistant (PA) role in that the micro-enterprises providing it were supporting a number of people in this way and were not necessarily providing personal care, whereas a PA usually works with only one person, and will often provide personal care as well. One-to-one support is well represented in the micro-enterprise sample, whereas the limited provision of residential care in the micro-enterprise sector meant that we only included one residential microenterprise (and one larger comparator). The sample is shown in Table 1.
We note that the sample is quite diverse, covering different sectors and groups of service users, as well as different scales of operation. This is not the kind of sample that would be drawn if the sole aim were to compare the effect of size on outcomes and costs. Instead, our research had a number of objectives. This means that the results cannot be regarded as statistically definitive, but are informed hypotheses that need to be followed up.
The research was undertaken in 2013 alongside 17 additional co-researchers, recruited from the three localities on the basis of their experience of using care services (or of supporting someone who used them). The aim of their involvement was to improve the validity of the research, gaining insights for the design of research materials, and to develop enhanced rapport with interviewees (INVOLVE, 2013). Co-researchers were involved in all stages of the research, including the design of interview questions, conducting interviews, the analysis of transcripts and dissemination activities. Involvement of the co-researchers was evaluated by a team of researchers who were not part of the main study. An initial scoping exercise of care provision in the three sites was used to identify the case-study organisations. Based on data from the national Skills for Care database of care providers, commissioners in the three local authorities and three local microenterprise coordinators working for Community Catalysts, case-study organisations were selected on the basis that they met the following purposive sampling criteria: • A mixture of micro and small/medium/large providers. • Coverage of the main types of care provision (domiciliary, group day activities, residential, one-to-one support). • Supporting older people and people with learning disabilities. • Mix of local authority, private and third sector providers. • Trading organisations, offering a paid-for service.
Organisations were contacted and asked to participate in the research by taking part in a staff interview and linking us to people who used their service and their family carers. A total of 143 people were interviewed for the study. This included 32 staff members from the 27 organisations, usually the manager and/or the person who had set up the organisation (in some interviews, there was more than one person present; hence, the n is higher than 27). Staff interview questions were open-ended, asking about why they had set up the organisation, what services it provided, how much they charged, what relationship they had with the local authority and what, if anything, they saw as distinctive about the support. A total of 106 service users and carers were interviewed about their care provider; they were asked questions including why they selected the care organisation, what relationships they had with staff, how much the provision was tailored to their needs and what they perceived to be the strengths and weaknesses of the service. This included 30 older people and 49 people with a learning disability (or an impairment such as autism). We also interviewed some of their carers (27 in total), where the carer indicated that they would like to contribute. Participants in the interviews included a mix of self-funders and people funded by local authorities, including some direct payment-holders.
The interviews were semi-structured and guided by a schedule that was developed in collaboration with the co-researchers, who led the interviews. All participants were given an easy-read information sheet about the project and asked to sign a consent form before the interview started. People who did not have capacity to consent were not included in the project, as advised by the national Social Care Research Ethics Committee that provided ethical approval for the research. At the end of each interview, participants were asked to complete a short survey covering two domains of ASCOT. ASCOT is a validated tool, which is used by the DH in its annual survey of adult social care, providing comparable national data for our project. The ASCOT survey has eight domains, allowing people to score the quality of service they receive in various aspects relating to care (Netten et al, 2011). In pilot interviews, we found that asking about all eight domains overloaded respondents at the end of an interview and often required us to ask about aspects of care that were not relevant to their service (eg accommodation). We limited the survey to two of its eight domains (focused on 'choice and control' and on how people spend their time), which were relevant for all the different types of provider organisations. Since the national data are disaggregated by domain, we were still able to compare our findings with the national survey data for these two domains. While we recognise The Goldilocks question 75 that using the full ASCOT tool may have allowed us to generate data around qualityadjusted life years (QALYs), we did not feel that it was ethical to insist on this when our piloting suggested that it was too burdensome for participants.
Interviews were audio-recorded and later transcribed. Interview notes and transcripts were uploaded into QSR-NVivo 10 for coding by the authors. An initial coding tree was developed deductively from the hypotheses, reflecting the primacy of organisational size as an explanatory variable, and each transcript was coded thematically (Miles et al, 2014). A second phase of analysis was undertaken, deriving new codes inductively based on patterns in the data using the framework analysis approach (Ritchie and Spencer, 2002). To enhance inter-coder reliability, the first and second stages of the coding began as collective coding exercises, working with printed-out transcripts, and only later moved to individual coding on NVivo (Elliot et al, 2013). A third stage of coding identified quotes for inclusion in written outputs on the basis that they exemplified patterns found in the first and second stages of the analysis.
ASCOT survey responses were analysed using IBM SPSS v22 to generate descriptive statistics and crosstabs in order to compare people's care outcomes on two of the ASCOT domains, and their expected outcomes in these areas in the absence of a service. The data were analysed to identify any particular quality issues (codes within appropriate and plausible ranges). The main analysis was conducted by comparing key outcomes (the dependent variables) by size of provider (the independent variable) and other confounding variables relating to the service users. Findings were then compared with the most recent DH national Adult Social Care Survey (for 2013/14, being broadly contemporary with our own fieldwork), which also uses ASCOT. Value for money was measured by comparing prices (cost per hour for an individual) with ASCOT survey findings and user/carer interview data. For the quantitative analysis, the sample size is relatively small, and any comparisons need to take account of the other ways in which the providers differ, including the type of services provided and the kinds of service users served. Later analysis (see Table  3) provides some evidence for similarities between care workers in different sizes of employer (we have no indication from our qualitative work that these differed significantly between participating organisations). Work at this kind of scale cannot provide definitive quantitative conclusions. Where results indicated apparently quite large differences, we investigated the effects of other variables. The sample was not selected to be a random sample of the population, but was based on a range of practical and theoretical considerations. We are aware that this limits the robustness of comparisons, but we believe the text that follows is clear about the limitations of this part of the analysis. We also believe that the overall credibility of the research is enhanced by our use of a combination of different research methods, rather than relying on only one type of approach to the data analysis.

Research findings
Hypothesis 1: Do smaller services deliver more valued outcomes than larger services?
We explored the capacity for care organisations of different sizes to help people achieve valued outcomes using two ASCOT domains, focused on whether people felt that using the service gave them more opportunity to spend time doing what they value and enjoy, and more choice and control, than would be the case without the service. At a bivariate perspective, there were no reliable associations between the size of the provider and the relevant ASCOT outcomes data (see Tables 2a  and 2b). Those receiving care services from a micro-enterprise tended to report  being able to do more of the things that they value and enjoy: people using microenterprises achieved this 50% of the time, compared with 22% of those using small, medium and large providers. The micro-enterprise score here also compares well with the national Adult Social Care Survey 2013/14, in which only one third (33%) said that they could spend as much time as they wanted on the things they enjoy (HSCIC, 2014: 5). However, this difference was no longer statistically significant after controlling for the age and gender differences in service users between larger and smaller organisations (see Table 2b).
In terms of the second set of questions (which aimed to explore whether people had more choice and control as a result of receiving services from a given provider), the findings were inconclusive. The data showed that there were some signs that people using micro-enterprises had greater control over their lives than people using larger services, and that the care provider was making a difference to how much control they felt they had. These variations between providers are not large enough to have confidence in any statistical differences. Even so, the answers for the micro-providers in our study (41% having as much control as they would like) were higher than the national average of 33% (HSCIC, 2014: 4). Working with a small data set, we did not further disaggregate the data into types of service (eg domiciliary versus day services) or by type of user (eg older people versus younger disabled people) because of the unreliability of inferences drawn from such small subsamples, although the qualitative data discussed later point to some patterns by service type and user profile.
An outcomes orientation can be described as taking a 'black box' approach to how outcomes are achieved (Behn, 2003). We used the interview data to better understand the process through which the outcomes were achieved. In analysing the interview data, we used the category of being valued to mean that there had been a tangible improvement or benefit from using the service, which could be classified as an improved outcome (or, conversely, a worse outcome due to a problem with the service). However, it was clear from the interviews that, for many people, there was no real articulation of a particular outcome separate from the process by which the care was delivered. Lewis and West (2013: 4) affirm the importance of the quality of the process of care, arguing that it is not reducible to a set of outcome measures. This was particularly the case for the older people in our study, and for care delivered in the home. In the interviews with younger people or their family carers, there was more likely to be discussion of an end result (making new friends, building confidence, getting fit, finding a job), which was distinguishable from the support that made it happen.
Hypothesis 2: Are smaller services more personalised than larger services?
For the last decade in English social care services, there has been a strong push towards the process of care being personalised to the individual, and we used interview questions to probe the extent to which people felt that they received a person-centred service. Here, we found that the smallest organisations (our categories of 'micro' and 'small') performed better than the medium and large organisations, at least within the qualitative element of our data. For people receiving services, smallness was seen as synonymous with the personalisation of services around the individual: "If it gets bigger you lose some of the intimacy that you have now." (Older person, small domiciliary care) "I think it's quite nice it's small cos you're not gonna get loads of different people coming, are you?" (Older person, micro domiciliary care) "Because they are small, you see everyone, they come to the house, it's not just someone stuck behind a desk. They'll run things past us, let us know what's going on. They give us plenty of feedback, make sure we're satisfied with what they are up to." (Family carer, micro support in the home) The so-called 'time and task' model (whereby a set list of care activities is carried out at specific times of the day) is well known (and much criticised) as the dominant approach to personal care delivery, leading to very short care visits to undertake highly specified tasks (Leonard Cheshire Disability, 2013;Unison, 2014). This model was one that the staff in the smaller organisations defined themselves in opposition to: "Being a small company … you don't have to be so task-driven. This is the list of things that we're supposed to do on the care plan for whoever, but when you go in, if you can see that they're down, [then] they need you to sit and have a cup of tea with them." (Staff, small domiciliary care) The care sector is known to have high rates of staff turnover (Lewis and West, 2013), and people receiving care from larger care providers commented on the large numbers of different people coming into their home: Interviewer: "When [the care workers] come, is it normally -how many different people would you get? Is it normally the same one or two, or could it be a lot?" Respondent: "Oh, it could be dozens because there's that many." (Family carer, large domiciliary care) Although drawn from a small number of interviews with people using large agencies, this account substantiates the well-established profile of large care agencies in which pressures on unit costs lead to a low-paid, inadequately trained workforce, with high turnover rates (see, eg, EHRC, 2011; Lewis and West, 2013). We conducted some analysis of recent data on care workers from the Labour Force Survey (LFS) (n = 1,011; see Table 3). This did not find strong differences between care workers in different sectors, although there was some evidence that workers in smaller organisations (the smallest classification used in the LFS is 'up to 10' workers) were more likely than average to be undertaking unpaid overtime. With such data, it is not possible to know if this reflects exploitation or dedication on the part of the worker. We would also emphasise the 15% of care workers on zero-hours contracts, compared with 2.5% of workers outside the care sector, even though this did not vary systematically by size of care provider.
The more personalised care provided by the smaller organisations, compared to large care providers, seemed to stem from their greater continuity of front-line staff, The Goldilocks question 79 the greater autonomy of front-line staff to vary the service being offered and the high level of familiarity of managers with individual clients and staff. However, the account of smaller agencies as personalised, and of large agencies as alienating and impersonal, was tempered by a recognition that there can be disadvantages to a more intimate and personalised relationship. As one family carer put it, talking about her daughter: "[S]he doesn't like one person coming all the time cos, obviously, they can get an attachment, can't they? Then, if for whatever reason, they're not able to come, then, you know, you've got a fresh face, whereas if there's a few people bobbing in and out, they get to know that person." (Family carer, small domiciliary care) This is consistent with the literature on PAs (Leece and Peace, 2010) and offers a corrective to the assumption that people using services always want a close relationship with a single member of staff. Some interviewees also saw the benefit of larger organisations because of their broader staff base: "There's a real advantage to a smaller one but, of course, there's an advantage to the bigger one, that if somebody falls sick, there's always somebody else to step into her shoes' (older person, medium-sized support in the home).
A final modifying factor of the 'small = personalised'/'large = impersonalised' story is that it seemed to apply most intensely to the activities that took place in people's homes: domiciliary care services and other kinds of one-to-one support in the home. When it came to group and institutional settings -residential or day services -some interviewees argued that a larger scale enabled more diversity of provision and therefore better responsiveness to individualised need. As one family carer put it: "She also has mobility problems, restless legs and things, so they'll give her a massage.… In a smaller place, they wouldn't … have that level of expertise to do that level of care on an individual basis" (Family carer, large day activities). To assess the extent to which different sizes of provider offer value for money, we gathered price data from all the 27 organisations included in the research. Analysis of the pricing data from the small number of case-study organisations indicated that micro-enterprises were slightly cheaper in their average hourly rate than small, medium and large care providers (see Figure 1). Figure 1 looks at only two larger areas of service -personal care and support in the home -to increase the validity of the comparison (comparing 'like with like'), and also indicates the type of service user. The relatively small sample size means that we cannot generalise from this to say that micro-enterprises are always cheaper, and, indeed, there are some exceptions within the chart. However, it does indicate that economy-of-scale arguments do not necessarily hold within the care sector: the larger organisations in our study were rarely able to offer a service at a lower rate than a comparable micro-enterprise. There were a number of factors that also affected the level of costs. The hourly charge tended to vary with age, with services for those aged over 75 costing rather more than for other age groups. This could indicate that further research might focus on this important and growing group in particular. Services predominantly used by self-funders were more expensive than those used predominantly by people whose support was local authority-funded. The interviews showed that many of the micro-providers worked out of the manager's own home or from very small, shared office spaces, whereas the small, medium and large organisations had formal business premises.
Taken with the second hypothesis (that smaller services may be more personalised than larger ones), this suggests a lower-quality service. This finding does not necessarily mean that micro-enterprises will always be cheaper than larger organisations, however, because length of call also needs to be taken into account. When providing domiciliary care, the micro-enterprises (and, indeed, some of the small providers) would only undertake calls of 30 minutes or longer, making them more expensive to commission, per visit, than larger companies operating on a 15-minute-per-call basis. For some organisations, this was a matter of ethos, but also practicality given their small staff base and the difficulties of moving staff to a different location after 15 minutes. This would make the micro-enterprises more expensive if only a 15-minute call was deemed necessary; however, there has been significant national resistance to the notion of 15-minute care visits, with a growing consensus that they are unacceptable (Leonard Cheshire Disability, 2013;Walsh and Shutes, 2013;NICE, 2015).

What size is 'just right'?
The literature review highlighted gaps in what is known about size and scale as a variable that shapes organisational performance. The evidence gathered here is consistent with claims that smaller providers of care perform better than larger ones on the basis of care outcomes, person-centredness and value for money; however, our small and diverse sample means that we cannot claim definitive proof of such differences. Moreover, it is also possible that smaller providers could struggle in terms of economies of scale, and it is known that micro-businesses in general have a very high failure rate (NIESR, 2006). Boyne (2003a: 371), writing about organisational performance across public services, hypotheses that 'excessively small or large organizations may be less successful than their medium-sized counterparts'. In the care sector, we found a different pattern. It was the very smallest micro-enterprises and the largest providers (with 100+ staff) that seemed best adapted to providing care services in an era of severe cuts to care spending. At the smaller end, microenterprises had an advantage since the very limited time and capacity of a microenterprise imposed a limit on their activities (although some people were working seven days a week, including some evenings). The unpredictable flow of people using the service could also be managed by increasing or decreasing the micro-owner's own hours, albeit with financial implications. Some had informal arrangements with other micro-enterprises to cover each other's work in the event of holidays, sick leave or an oversupply of clients. Most spoke of how difficult it was to get on to local authority lists of providers, which was also a limit on growth and sustainability. Some of the small and medium providers reported the same problems, for example, in being unable to get on to the local authority's 'preferred provider' list, but also of being unable to compete with large providers on price. Large providers often work across local authority boundaries, with a staffing base large enough to achieve economies of scale. They can absorb large numbers of referrals, which remains alluring for local authorities despite national concerns about the service quality and long-term stability of large care companies offering a 'time and task' model of care (Leonard Cheshire Disability, 2013;Unison, 2014;NICE, 2015). Whereas the larger providers were receiving a regular flow of people funded by the local authority, the smaller organisations were more reliant on self-funders and direct payment-holders, and had to invest in marketing their services to raise awareness among potential users.
Clearly, this has implications for the current regulatory and financial climate, in which public sector austerity may force providers of commissioned services to meet the basic needs of people assessed as eligible for publicly funded care, potentially neglecting outcomes that arguably matter more to people using services (eg their sense of control and of doing things that they enjoy and value). Similarly, a care sector facing massive financial and demographic pressures, workforce shortages, and staff turnover has often tended to resort to tactics such as the use of zero-hours contracts, low pay and failing to remunerate staff for travel time (see, eg, Osborne and Duncan, 2016). It is possible that very large organisations may be better placed to survive such pressures, and that the very nature and ethos of micro-enterprises (see Box 1) could make it easier for them to ignore this development and deliver care in the way they feel is right. Certainly, Sheaff et al's (2003: 137) study of organisational factors and performance concludes that the success of an organisation is dependent on the fit between an organisation's strengths and weaknesses and its external environment. Thus, it can be argued that very small and very large care organisations offer the best fit for the current patterns of care commissioning, regardless of performance. Local authority commissioning continues to favour large providers; self-funding and direct payments can support some individualised commissioning, but this is a good fit only for the very smallest micro-enterprises that can cope with uncertain demand.

Conclusion
Organisations and organisational types thrive and decline on the basis of their performance, but also on the basis of their fit with prevailing norms. Boyne (2003b: 221) cites Gaertner and Ramayaran (1983: 97): 'an effective organization is one that is able to fashion accounts of itself and its activities in ways which constituencies find acceptable'. The support given by the DH to micro-enterprises suggests that they are advantaged by appearing to 'fit' well with the dominant narrative of personalisation and individualised commissioning.
The findings reported here are consistent with the view that the quality of service provided by small organisations is related to an affinity between being small and being personalised. The qualitative interview data in the study support the idea that small and micro-enterprises can deliver more personalised services, particularly in the home -even though size may be a proxy here for other elements of service provision, such as the ability of micro-enterprises to work alongside the people they support in order to achieve desired outcomes and to feel valued. At the same time, the ASCOT scores and pricing data indicated that micro-enterprises offer better value for money, but we acknowledge limitations in our data based on sample sizes and the use of particular ASCOT domains. The same patterns nevertheless emerge in both the qualitative and the more structured elements of our work. These benefits seem to be based on greater continuity of staff, greater staff autonomy and greater accessibility of managers, combined with the low overheads of operating at a microscale. However, stability can be a problem for micro-enterprises, particularly those that employ staff and need to have a relatively consistent financial turnover. In the current context, large providers look likely to continue to harvest local authority referrals, leaving micro-enterprises to pick up an uncertain residuum of self-funders and direct payment-holders. Growth in the number of direct payments could increase demand for micro-enterprises, as could a more sustained effort by local authorities to diversify local care markets. Both of these aspects were endorsed in legislation in the Care Act 2014, but progress in this appears slow as local authorities cope with the more immediate pressures of austerity (LGA, 2015).

The Goldilocks question 83
The scope of the research reported here -27 organisations and 143 interviews -is broad for a mainly qualitative study, although we are aware that it is not large enough to make claims about statistical generalisability and that we were unable to control for levels of need and dependency. We suggest that our findings are generalisable to theory in the way Popay et al (1998: 348-9) set out: 'the aim is to make logical generalizations to a theoretical understanding of a similar class of phenomena rather than probabilistic generalizations to a population'. Our analysis of micro-enterprises allows for 'the generation of the logical features of a type against which further cases can be examined with gradual evolution of our theoretical understanding' (Popay et al, 1998: 348-9). Through our analysis of the research data, we offer theoretical insights into the relationship between organisational size and care, and contribute to theoretical understandings of performance, innovation and personalisation in a care setting.
Two theories are offered here for testing in future research. The first is that the distinctive context of social care makes size a stronger determinant of performance than it has been found to be in other public services, where the relationship has been inconclusive. The intensity and intimacy of the support given, particularly in the home, requires consistency, flexibility and autonomy, features that, in our research, were most likely to be found in smaller organisations. The second is that the relationship between size and sustainability is non-linear in the social care context, with the very smallest and very largest organisations having market advantages that do not necessarily relate to the quality of care provided. In addition to their academic significance, these conclusions also offer important lessons for social care commissioners, who have a duty to shape the nature of local care markets and to ensure that assessed needs are met within very tight financial resources. While it is difficult to imagine an entire system relying on micro-enterprise, there seem very real benefits to ensuring that micro-enterprise is a key component in the overall mix, and that as many as possible of the barriers experienced by micro-enterprises are removed.