Playing with Numbers: A Methodological Critique of the Social Enterprise Growth Myth

Abstract

Social enterprise is a contested concept which has become a site for policy intervention in many countries. In the UK the government has invested significant resources into social enterprise infrastructure, partly to increase the capacity of social enterprises to deliver or replace public services. Government publications show the number of social enterprises to have increased from 5,300 to 62,000 over a five-year period. This paper explores the myth of social enterprise growth in the UK through a methodological critique of the four government data sources used to construct and legitimise this myth. Particular attention is paid to how political decisions influence the construction of evidence. We find that growth is mainly attributable to political decisions to reinterpret key elements of the social enterprise definition and to include new organisational types in sampling frames.

Key Words:

• Critique
• legitimacy
• official statistics
• social enterprise
• social entrepreneurship

Introduction

It is relatively unproblematic to state that academic use of the labels, social entrepreneurship and social enterprise, is increasingly common (Defourny and Nyssens 2010, Bacq and Janssen 2011). On a related note, it is also somewhat unproblematic to state that governments are paying increased attention to social enterprise as a policy solution to various social problems (Kerlin 2009). But social enterprise is a contested concept whose meaning is politically, culturally, historically and geographically variable (Kerlin 2009, Teasdale 2012). While policy actions can be legitimised by the use of indicators or statistics (Kingdon 1984, Kendall 2000, Rehn 2008), these statistics often conceal the social construction of empirical categories behind the veneer of scientific evidence (Smagorinsky 1995). It would therefore become somewhat problematic if academics were to unquestioningly accept claims of dramatic growth in the number of social enterprises as truth.

The United Kingdom (UK) is seen as having the most developed institutional support structure for social entrepreneurship/enterprise in the world (Nicholls 2010a). In 2002 the Government's Department for Trade and Industry (DTI) set out a loose definition of social enterprise designed to capture the wide variety of different organisational types then associated with the label:

A social enterprise is a business with primarily social objectives, whose surpluses are principally reinvested for that purpose in the business or in the community, rather than being driven by the need to maximise profit for shareholders and owners. (Department for Trade and Industry 2002, p. 8)

This definition allowed considerable scope for subsequent reinterpretation by government as to what is and is not a social enterprise. The UK collects some of the most detailed official statistics pertaining to social enterprise in the world (Teasdale et al. 2011). These statistics appear to point to phenomenal growth in social enterprises since 2003, suggesting that the UK government policy has been highly successful (Nicholls 2010b). However the symbolic use of evidence can promulgate myths and create meaning far beyond any intrinsic value of the data (Brown 1994), and be used to enrol others in particular courses of action (Latour 1987). In the UK, official statistics pointing to this phenomenal growth have been used to justify government policies, including restructuring of the National Health Service, and wider moves to rebalance the welfare mix toward the private sector and social enterprises (Teasdale et al. 2012). Indeed governments from across the world have sought to learn from the ‘successful’ UK social enterprise experience (Young 2011).

Until now academics have had no reason to doubt these official statistics, leading some respected commentators to refer to dramatic growth in the number of social enterprises (see for example Haugh and Kitson 2007, Chell et al. 2010). But in this paper, we demonstrate (through analysis of the different data sources used to construct the myth of social enterprise growth in the UK) that growth can primarily be explained by political modification of the way in which data is collected and reported. This enables the myth of social enterprise growth to be presented as truth by political actors to further their own ideological agendas.

The paper proceeds as follows. First we draw upon academic literature to understand how policy makers construct and symbolically present evidence to support their chosen policy solutions. Next we introduce the reader to the contested concept of social enterprise, paying particular attention to the fluidity of the concept across context and time. Drawing upon the UK experience we show that social enterprise has been able to change shape to fit prevailing ideologies. We explore how official statistics pertaining to social enterprise growth have been used to legitimise political strategies, and how claims of growth in the numbers of social enterprises have been perpetuated in academic literature. In the empirical section of the paper we methodologically critique state sponsored surveys of social enterprise. First, we show that growth is primarily a consequence of changing the ways in which the statistics were constructed, and political decisions to include different organisational types as social enterprises. Second, we show that official statistics, which undermine the social enterprise growth myth, have been concealed. In our conclusion we return to the more general literature to argue that where concepts are contested there is considerable scope for governments to select evidence to further their own purposes. We then set out the more specific implications for social enterprise researchers and practitioners.

The Symbolic Role of Evidence in the Policy Making Process

Policy solutions need to be considered legitimate by legislators for them to become accepted (Kingdon 1984). Evidence has an important role to play in demonstrating the pragmatic legitimacy of policy solutions. For example, Kendall (2000), writing at a time when evidence-based policy-making was a central tenet of New Labour political discourse in the UK, argued that an important influence in bringing the third sector to the attention of policy makers was the new availability of a critical mass of indicators in the late 1990s. These indicators provided pragmatic legitimacy to policy measures which closely fitted the changing norms and values of a new government with an expressed ideological belief in the third way – conceptualised by Giddens (1998) as beyond market liberalism and socialism.

Evidence can also be used to confer the moral legitimacy of policy solutions. As Kingdon (1984) recognises, being able to persuade policy makers that a policy solution is ideologically correct may be crucial in convincing legislators of its legitimacy. Using the example of Lysenkoism, Rehn (2008) illustrates this point to hint at the dangers of academics uncritically accepting the prevailing modes of thought in management studies. Lysenko persuaded Soviet legislators to adopt a particular type of agricultural science by presenting it as based upon practical science, while constantly shaping narratives to fit the prevailing (Soviet) language and ideology. In this he was always one step ahead of critical academics, as before they could react to a proposition, Lysenko had developed a new narrative (Rehn 2008). The implication then for the policy making process is clear; ideas which are able to fit prevailing ideologies face a greater chance of being accepted by legislators. And ideas which are able to constantly change shape in response to a shifting ideological environment are likely to have longevity as policy solutions.

Drawing on the early work of Goffman (1959), institutional theorists focusing on the micro level of the organisation have recognised that one way of influencing legitimacy is the use of symbols to convey meanings beyond their intrinsic value (Zott and Huy 2007). Symbolic action can include narrative accounts (Burke 1966). It can also involve the use of material symbols, such as business plans to make a new venture appear legitimate to funders (Zott and Huy 2007), by demonstrating that an organisation conforms to the wider institutional norms (Zimmerman and Zeitz 2002). But symbolic action is culturally contingent, and organisational (and political) actors must use symbols that conform to the expectations of their audiences (Goffman 1959).

Just as organisational actors can use evidence symbolically to legitimise their actions so too can policy actors. Brown (1994) demonstrates how evidence can be suppressed to protect legitimacy, or manipulated to create legitimacy. Effectively the selective presentation of evidence is a form of symbolic action which can be used to portray legitimacy. Evidence is socially constructed (Berger and Luckman 1966) and researchers are able to manipulate findings by careful construction of empirical categories (Smagorinsky 1995). Rather than there being a single claim to truth there are often multiple and competing claims. So just as organisations are able to construct and present evidence which fits their pre-determined goals (Brown 1994), policy makers are also able to choose from a selection of counter claims, each supported by empirical evidence (Campbell 1985). And just as organisational actors can use evidence as symbolic proof of organisational legitimacy, policy makers can choose, or even construct, evidence which demonstrates the legitimacy of their policy solutions.

Social Enterprise

Social enterprise is a concept which is variably interpreted according to historical, geographical, political, social, cultural and economic factors (Diochon and Anderson 2009, Kerlin 2009, Bacq and Janssen 2011, Teasdale 2012). At its broadest level, social enterprise involves the use of market-based strategies to achieve social goals (Kerlin 2009). What is meant by the term social is often concealed, which hides the contestation played out between competing factions (Parkinson and Howorth 2008, Lyon and Sepulveda 2009, Dey and Steyaert 2010).

Social enterprise has become particularly popular among academics, practitioners and policy makers across the world as a potential solution to a range of policy problems. These include, but are not limited to, a potential solution to area-based deprivation (Blackburn and Ram 2006); an alternative vehicle for the delivery of publicly funded services (Simmons 2008); a more effective means of international development (Peredo and Chrisman 2006); an additional revenue raising stream for nonprofits (Dees 1998), or a potential alternative to winner takes all capitalism (Westall 2001, Amin et al. 2002). It is of course unlikely that any single organisation is able to achieve all this. It would appear that like Lysenkoism, social enterprise is an idea which is constantly able to change shape in response to a shifting ideological environment.

This is particularly apparent in the UK where social enterprise has been constantly reshaped by practitioners and policymakers to fit shifting social and political environments (Teasdale 2012). Although the term had recently come to prominence in the US to refer to nonprofits’ earned income strategies, and in some parts of Europe to refer to co-operatives, it had not entered common usage in the UK before the late 1990s. A loose alliance of influential practitioners from the co-operative and community development movements, sympathetic Labour politicians, and members of left leaning think tanks adopted the concept of social enterprise as part of an ideological aim to radically change the ways of organising markets (Westall 2001, Grenier 2009, Ridley-Duff and Bull 2011). To attract further political support they presented social enterprise as an organisational form of the third way and as a solution to the policy problem of area-based deprivation (Grenier 2009, Teasdale 2012).

The concept was eagerly taken on board by a Labour government, ideologically committed to a third way beyond state socialism and free market capitalism. In 1999, social enterprise was first presented as a response to area-based deprivation (HM Treasury 1999, Blackburn and Ram 2006). Between 1999 and 2001, social enterprise (also) became seen as a solution to the failure of markets to distribute goods and services equitably (Teasdale 2012). By 2002, social enterprise was (additionally) portrayed as a policy solution to the failure of the state to deliver public services that were responsive to consumers (Simmons 2008). In 2006, social enterprise became presented as a solution to the failure of the third sector to scale-up (Office of the Third Sector 2006). Since 2008, social enterprise has been particularly emphasised as a vehicle through which public services can be spun-off allowing greater democratic ownership and control (Blond 2009). Each extension of the concept has involved re-articulation through drawing on different claims and evidence. At different times and across different contexts, policy actors have repackaged elements from different narratives to bolster their own legitimacy. For example, when legitimising the use of social enterprises to deliver public services, they were presented as: innovative and entrepreneurial, enabling access to public services and improving outcomes for the hardest to help (Office of the Third Sector 2009). This was rather different to the emphasis in 1999 on social enterprises’ (supposed) ability to build social capital through democratic engagement and hence contribute to neighbourhood renewal (HM Treasury 1999). So the fluidity of the concept permits considerable scope to select positive attributes in order to legitimise policy action.

The Myth of Social Enterprise Growth

The first government-sponsored estimate of the number of social enterprises in the UK, published in 2003, estimated a total of 5,300 (ECOTEC 2003). A mapping study undertaken for the Department of Trade and Industry (DTI) in 2004 estimated there were 15,000 social enterprises in the UK (IFF Research 2005), although the methodology was described in The Economist as ‘slightly alarming’ by an ‘expert’ rumoured to be the editor of this journal (The Economist, 24th November 2005). By 2005, the Annual Survey of Small Businesses (ASBS) was used to claim there were ‘at least’ 55,000 social enterprises in the UK, leading commentators relying on this data to claim that ‘a new characteristic of the third sector under New Labour has been the increase in the number of social enterprises’ (Haugh and Kitson 2007, p. 975). In 2008, official government publications based on the 2007 ASBS revised this to an estimated 62,000 social enterprises.

The statistics showing growth in social enterprise are regularly used in government publications, often combined with narratives of heroic success and a list of positive attributes (see for example, Cabinet Office 2006, the Government's Social Enterprise Action Plan). The apparent growth from 55,000 to 62,000 social enterprises led to a Cabinet Office spokeswoman stating that:

We do believe that the number of social enterprises is increasing and therefore this new estimate is consistent with feedback from our partners that the sector is growing, which is great news. (cited in Hampson 2009)

The same spokeswoman symbolically managed the myth of social enterprise growth by associating the statistics to other public bodies to portray legitimacy.

We have strengthened the methodology following advice from the Office of National Statistics. (Hampson 2009)

The statistics are regularly used by politicians and practitioners to legitimise their past and future actions, even where the relevance of the figure to the argument being made is not obvious. For example, in 2010, the Secretary of State for Health referred to the 62,000 social enterprises when announcing plans to ‘create the largest and most vibrant biggest social enterprise sector in the world’ through empowering ‘millions of public sector workers to become their own boss and help[ing] them to deliver better services’ (Department of Health 2010).

These estimates have been widely cited not just in policy and practitioner literature, but also academic publications (including, but not limited to: Haugh and Kitson 2007, Spear et al. 2009, Bridgstock et al. 2010, Mawson 2010), and the myth of social enterprise growth has been presented as absolute fact:

In the UK, social enterprises have undergone high growth rates with high levels of success in pursuit of their aims. (Chell et al. 2010, p. 486)

It is perhaps surprising that the UK government estimates of 62,000 social enterprises have been so widely accepted given that academics and policy makers have not been able to agree upon a common definition of social enterprise (Peattie and Morley 2008). However in previous articles the authors of this paper have also used these government figures. Our own (self) critical reflections suggest that just as practitioners and policy makers have used the statistics to legitimise their actions, so have academics sought to legitimise the field of social enterprise research by symbolically highlighting its importance using evidence [1]. Put simply, we all need to demonstrate that our area of research is important when seeking to win research grants or to publish in high quality journals. Perhaps now that the field of social enterprise research has become relatively well-established, it is time to take a step backwards and explore the methodological foundations upon which our claims of growth are constructed.

The Political Construction of a Social Enterprise Growth Myth

As discussed earlier, social enterprise first entered the UK policy landscape in the late 1990s. In 1999, the Neighbourhood Renewal Unit produced Enterprise and Exclusion which drew upon narrative profiles of successful social enterprises to claim that:

Social enterprises can make a positive contribution to the regeneration of deprived areas by helping to provide employment, goods, services and more generally build social capital. (HM Treasury 1999, p. 112)

A number of policy measures aimed at developing social enterprise activity were announced, and an indicator of the success of government policy was to be the level of social enterprise activity (HM Treasury 1999).

Following concerted lobbying by the social enterprise movement, a Social Enterprise Unit was established within the then Department for Trade and Industry (DTI) in 2001 (Grenier 2009, Bland 2010). The Secretary of State for Trade and Industry, Patricia Hewitt, hosted a social enterprise seminar at which an ‘immediate need’ for ‘better data’ and ‘more mapping of the sector’ if the ‘potential of social enterprise’ was to be realised (Kenyon and Crellin 2001). Consequentially a social enterprise mapping working group involving civil servants, high profile practitioners and academics was created to pursue this need. This led to the government definition of social enterprise reproduced again here:

A social enterprise is a business with primarily social objectives, whose surpluses are principally reinvested for that purpose in the business or in the community, rather than being driven by the need to maximise profit for shareholders and owners. (Department for Trade and Industry 2002, p. 8)

Three elements are crucial to this definition:

1.	First, ‘a business’ implies that social enterprises are different to traditional charities as they trade in the marketplace.
2.	Second, ‘primarily social objectives’ distinguish social enterprises from traditional businesses.
3.	Third, ‘surpluses are principally reinvested ’ (to meet their social objectives) rather than ‘being driven by the need to maximise profits for shareholders and owners’.

Over the next decade, four state-sponsored studies operationalised this definition as an empirical category in order to produce statistics pertaining to social enterprise scale and growth (see Table 1).

Table 1 Shifting interpretation of key elements of the social enterprise definition

			Interpretation of key elements
Survey undertaken	Data source	Sampling Frame	a business	primarily social objectives	reinvestment of surplus	Estimated social enterprise population
1999–2003	ECOTEC, 2003	Existing local surveys	50% of income derived from trading	Governance and ownership structure based on democratic participation	Surplus can only be distributed to stakeholders as profit sharing or used for the benefit of the community	‘up to 5,300’
2004	IFF Research, 2005	Companies Limited by Guarantee and Industrial and Provident Societies	25% of income derived through trading	The primary purpose is to pursue a social goal (or) to make profit for owners, partners and shareholders?	Asked whether profit is mainly distributed between owners, partners and shareholders; (or) mainly reinvested in the organisation or community, to further social goals?	‘around 15,000’
2005	ASBS dataset 2005	All enterprises with less than 250 employees	25% of income derived through trading	Asks if organisation is a close fit with the DTI definition	organisations which do not pay more than 50% of profit to owners/shareholders	55,000
2007	ASBS dataset 2007	All enterprises with less than 250 employees	25% of income derived through trading	Asks if organisation is a close fit with the DTI definition	organisations which do not pay more than 50% of profit to owners/shareholders	70,000 (62,000 figure based on rolling average)
2009/10	NSTSO dataset 2010	Third sector organisations	50% of income derived through trading	Asks if organisation is a close fit with the DTI definition	Asks whether (all) surpluses are used to further social and/or environmental aims	8,507

In the following sections we methodologically critique these studies, in turn, paying particular attention to political contestation over how to construct the empirical category of social enterprise and how the decisions taken have influenced evidence on the growth of social enterprise in the UK. We show that the social enterprise growth myth can be almost wholly explained by shifting interpretations of the key elements of the social enterprise definition, and related decisions as to which types of organisation make up the population from which social enterprises are drawn.

2003: Up to 5,300 Social Enterprises? Guidance on Mapping Social Enterprises

In 2003, the DTI commissioned ECOTEC, an economic development consultancy, to examine how different mapping exercises had been carried out; to determine whether the DTI definition of social enterprise could be retrospectively applied to these surveys; and to provide guidance as to how to operationalise the DTI definition for future surveys.

ECOTEC analysed 33 small-scale regional studies undertaken between 1999 and 2003. The studies used different approaches to collecting data, ranging from ‘top-down’ analysis of publicly available data from the Financial Services Authority to ‘bottom-up’ approaches approximating to snowball sampling. Definitions of social enterprise varied, although most studies required core criteria of: more than 50% of income from trading; explicit social aims; and social ownership defined by ECOTEC as:

Autonomous organisations with a governance and ownership structure based on participation by stakeholder groups (users or clients, local community groups, etc.) or by trustees. Profits are distributed as profit sharing to stakeholders or used for the benefit of the community. (ECOTEC 2003, p. 25)

A ‘guesstimate’ of the number of social enterprises in the UK was derived from those existing studies which adopted ‘defining characteristics’ of a 50% trading level, and enabled a comparison with the wider business population. The proportion of social enterprises in the wider business population ranged between 0.2 and 0.3% in these studies. Applying this ratio to the population of businesses in the UK provided a total ‘of up to 5,300’ (ECOTEC 2003).

Guidance for future mapping studies included developing ‘tests’ based on the DTI definition. Social enterprises should be formally registered to imply a degree of permanence. They should derive at least 50% of income through trading. A test that identifies whether an organisation has primarily social objectives and whether in pursuit of these objectives it principally reinvests surpluses in the business or in the community was also recommended. Ideally future work would identify this key element through examining legal forms and memorandums of association to ensure that core values are embedded in the legal constitution. However this was recognised as time consuming. As a consequence it was recommended that a first national mapping study should focus on Industrial and Provident Societies and Companies Limited by Guarantee (CLG). In addition to these constituting the large majority of social enterprises captured in existing studies, these legal forms guaranteed social ownership [2].

2004: 15,000 Social Enterprises? A Survey of Social Enterprises Across the UK

In 2004, the DTI commissioned IFF Research to conduct a mapping exercise of social enterprises in the UK. In this survey an organisation was considered a social enterprise if:

1.	regular activities involve providing products or services in return for payment – i.e. the enterprise has some form of trading activity
2.	at least 25 per cent of funding is generated from trading – i.e. through direct monetary exchange for the provision of goods and services
3.	the primary purpose is to provide social/environmental goals – i.e. rather than purely for profit
4.	they principally reinvest profit/surplus into the organisation or community to further social/environmental goals – i.e. at least 51 per cent of profit/surplus (IFF Research 2005, p. 7–8).

Including all those organisations with more than 25% trading income (instead of 50%) was ostensibly to capture newer social enterprises moving towards a 50% trading threshold (IFF Research 2005). That the revised approach might also capture organisations moving in the opposite directions as well as those which would never achieve a 50% trading threshold was seemingly not considered. Of course it should not be overlooked that a higher estimate of the social enterprise population would legitimise the government's large investment in the social enterprise support infrastructure, and more general commitment to develop the social enterprise sector (DTI 2002).

The sampling approach followed the ECOTEC guidance by focusing on CLG, a large proportion of which were also charities, and Industrial and Provident Society (IPS), a legal form commonly used by cooperatives. From a sample frame of 62,500 organisations, 8,401 were interviewed and 3,446 of these were identified as social enterprises. The survey was used to provide an estimate of around 15,000 social enterprises or 1.2% of all enterprises in the UK. It was estimated that they contributed £18 billion to GDP, and employed 475,000 paid staff, of whom two-thirds were full time.

Much of the apparent increase in the number of social enterprises from the earlier study is likely to be explained by the loosening of the trading threshold from 50 to 25%. Also, it should be remembered that the earlier study generalised from studies undertaken prior to 2003, some of which had included democratic participation as a defining criteria of social enterprise (ECOTEC 2003).

2007: 62,000 Social Enterprises? The Annual Small Business Surveys (ASBS)

In 2004/05, the DTI added questions to the long-running Annual Small Business Surveys in order to estimate what proportion of mainstream businesses were social enterprises. This approach had been recommended by ECOTEC in order to capture social businesses and charities with a Company Limited by Share legal form, and the new legal form of Community Interest Company (ECOTEC 2003). The survey was conducted by IFF Research with academic input in the analysis and reporting (Atkinson et al. 2006).

The survey used a self-classification system for social enterprises. For a business to be classified as a social enterprise they should:

1.	think of themselves as a social enterprise (Q37 in the 2006/07 survey)
2.	not pay more than 50 per cent of profits to owners/shareholders (Q36)
3.	generate more than 25 per cent of income from traded goods/services (Q34A)
4.	think that they are a very good fit with the DTI definition (Q38) (IFF Research 2008).

Whereas the previous survey had only sampled organisations with explicit social ownership, this survey included all registered businesses whatever their legal form. Rather than attempt to ensure that private businesses had a constitution which committed them to social aims and the reinvestment of surplus to meet those aims, the ASBS survey relied on respondents to answer test questions designed to ensure this criteria. However the test questions were also changed from the previous survey.

In the earlier IFF survey, respondents were explicitly asked if their organisation reinvested the majority of surplus for social or environmental purpose (IFF Research 2005). In the ASBS survey, the test question was modified such that organisations defined as social enterprises do not pay more than 50% of profits to owners/shareholders. Of course most businesses rarely pay more than 50% of profits as dividends. According to the ASBS data, only 18% of all SMEs paid more than 50% of profits to owners/shareholders in 2007.

To determine whether organisations’ objectives were primarily social, respondents were asked if they were a close fit with the DTI definition. This differs from the previous (IFF Research) survey which explicitly asked whether organisational objectives were to pursue a social (including environmental) goal (OR) to make profit for owners, partners and shareholders.

The data presented from the ASBS surveys has previously been superficial, providing a single overall published figure of a minimum of 55,000 social enterprises, rising in 2009 to a minimum of 62,000, based on a three year rolling average for 2005–07 (Hampson 2009).

Closer analysis of the 2006/07 survey data (see Table 2) yields the key finding that the overwhelming majority (89%) of the 151 organisations identified as social enterprises have a legal form that places no constraints on the distribution of profits to external shareholders (Companies Limited by Share, Partnerships or Sole Traders). Thus, these organisations would not have been considered social enterprises in the earlier surveys. The proportion of private, for personal profit enterprises included in the mythical figure of 62,000 social enterprises had not been disclosed prior to the analysis in this paper.

Table 2 Legal status by social enterprise (Annual Small Business Survey 2006/07)

	All SEs (0–250 employees)		SEs with 1 or more paid employees		All small and medium enterprises
	No.	%	No.	%	No.	%
Sole Proprietorship	262	56.1	49	32.4	4363	46.6
Private Limited Company	93	19.9	63	41.7	3172	33.9
Public Limited Company	9	1.9	2	1.3	40	0.4
Limited Partnership	1	0.2	1	0.7	33	0.4
Partnership	82	17.6	19	12.6	1682	17.9
Cooperative	0	0	0	0	11	0.1
Company Limited by Guarantee (CLG)	14	3	14	9.3	41	0.4
Friendly Society* (FS)	2	0.4	2	1.3	6	0.06
Community Interest Company (CIC)	4	0.9	1	0.7	4	0.04
Total	467	100	151	100	9362	100

A further political decision was made concerning the exclusion of the large number of enterprises (predominantly sole traders) not employing people but meeting the social enterprise tests. Including all respondents who met the test would increase the number of social enterprises to 234,000. This figure is notable by its absence in publications or policy documents. We suggest this figure was so far removed from earlier estimates that the legitimacy of the data would be questioned, if published.

The apparent growth from 15,000 social enterprises is more than explained by the political decision to include all businesses in the sampling frame. In the 2006/07 ASBS, only 0.7% of respondents were identified as social enterprises and had a legal structure implying social ownership and constraints on profit distribution. This analysis suggests that if private enterprises with the potential to distribute profit to shareholders were excluded, there would only be 8,000 social enterprises. To some extent this decision reflected wider political contestation around the social enterprise context. The survey was designed at a time when more commercially orientated social enterprises had significant influence within the DTI and prominent practitioners argued that it was acceptable to create private profit through social enterprise (Teasdale 2012).

The dramatic growth from 15,000 to 55,000 can largely be attributed to political decisions to count private for profit businesses as social enterprises then. However the growth from 55,000 to 62,000 between 2005 and 2007 requires further consideration. This increase might perceivably have been explained by an increase in newly registered organisations identifying as social enterprises. However closer analysis of the ASBS data reveals that only 5.3% of the social enterprises in the 2007 survey had been trading for three years or less. The growth therefore is more likely to be a consequence of a higher proportion of existing businesses claiming to be a good fit with the DTI definition of social enterprise. However the data are unable to tell us what extent this represents a change in behaviour of these businesses or a generalised perception that being identified as a social enterprise is socially desirable/morally legitimate.

2009: 8,507 Social Enterprises? The National Survey of Third Sector Organisations

While the figure of 62,000 social enterprises is widely quoted, a later study of organisations with explicit social ownership [The National Survey of Third Sector Organisations (NSTSO)] found only 8,507. This survey was commissioned by the Cabinet Office and was carried out by Ipsos MORI over 2008–09. To be defined as a social enterprise, an organisation had to self-identify as:

1.	Generating more than 50% of income from trading (i.e. the sale of goods and services);
2.	Reinvesting surplus to further social and environmental aims;
3.	Being a close fit with the DTI definition.

The sample frame of 170,552 Third Sector Organisations (TSOs) was constructed by Guidestar UK, and included registered charities, CLGs, IPSs and CICs (Ipsos MORI 2009). Companies Limited by Share would only be included if they were trading subsidiaries of registered charities. Thus, the sample frame considered exclusively of organisations which were legally obliged to have ‘primarily social objectives’ and restrictions on the ‘reinvestment of surplus’ embedded in their constitution. From this sample, a postal and a web survey were sent to 104,931 organisations with a response rate of 47%.

Analysis using Ipsos MORI's narrow definition of social enterprise, derived using all the above criteria, produces an estimate that 5% of TSOs can be classified as social enterprises, a total of 8,507 nationally. The question on reinvesting surplus was perhaps irrelevant, given that the sampling criteria were designed to capture only organisations which reinvested all surpluses in the organisation or to pursue social goals. However this question led to the 35% of organisations making a loss not being considered as social enterprises. When excluding this question on reinvesting, the number of organisations meeting the other tests is shown to be 16,361 (see Table 3), a number not dissimilar to the estimate of 15,000 social enterprises from 2004. The NSTSO dataset is deposited in the UK Data Archive but the estimate of the number of social enterprises has not been reported by the government, arguably because it does not support the myth of social enterprise growth.

Table 3 Analysis of 2008/09 National Survey of Third Sector Organisations questions related to social enterprise

	Percent of sample	Estimate of population
Involved in some trading activity	27.6	47,072
Have more than 50% of income from trading	12.5	21,344
Stated that they fit the social enterprise definition	48.1	82,046
Agree with definition and have more than 50% of income from trading	9.6	16,361
More than 50% trading, fit the social enterprise definition and use more than 50% of surplus for social and environmental goals	5.0	8,507

Of course the main reason for the apparent decline in the number of social enterprises was the decision to sample only third sector organisations, which by definition had explicit social ownership. But this could not easily be explained by the government. The survey was undertaken during a period when the Labour government was heavily promoting the third sector as a site for policy intervention, particularly as a means of delivering public services (Carmel and Harlock 2008, Di Domenico et al. 2009). As part of this process, departmental responsibility for social enterprise had shifted from the DTI to the new Cabinet Office of the Third Sector (OTS). The government claimed that:

social enterprises are part of the ‘third sector’, which encompasses all organisations which are nongovernmental, principally reinvest surpluses in the community or organisation and seek to deliver social or environmental benefits. (OTS 2006, p. 10)

If this were true, then how might the government explain that the phenomenal growth in social enterprises was a consequence of the inclusion of organisations not considered part of the third sector? It would appear that protecting legitimacy was to be achieved by the careful suppression of evidence.

Conclusion

Our methodological critique of the four government-sponsored surveys enables us to understand how the myth of social enterprise growth has been empirically constructed. All of the state sponsored surveys have attempted to estimate the numbers of organisations meeting the social enterprise definition, first specified by the DTI. While this definition has not changed, interpretation of the three key elements within it has altered dramatically. The first study undertaken by ECOTEC (2003) involved estimating from existing surveys what proportion of democratically controlled organisations, with restrictions on the distribution of surplus, derived at least half their income through trading. In the second study undertaken by IFF Research (2005), the criteria as to what constitutes a business was widened from achieving 50% to just 25% of income through trading. The apparently dramatic growth in social enterprises between 2004 and 2005 can be explained by changes to how the key element of ‘primarily social objectives’ was interpreted, from an emphasis on governance and ownership structures based on democratic participation to allowing respondents to decide whether they are a close fit with the DTI definition. At the same time, and perhaps crucially, the ‘reinvestment of surplus’ element has moved from a strict emphasis on surpluses only being distributed as stakeholder profit sharing (as followed by co-ops and mutuals) or used for the explicit benefit of the community, to not using more than 50% of surplus for personal profit. The reinterpretation of these elements allowed private businesses to be classified as social enterprises such that almost 90% of the mythical 62,000 social enterprises are organisations which would not have been considered social enterprises under the original interpretation of the definition. The most recent NSTSO allows an approximation as to how many social enterprises there are in the UK, meeting the 2004 interpretation of the definition. Our analysis suggests around 16,000 TSOs would be considered social enterprises using this approach. Counting for-profit organisations as social enterprises seemingly accounts for almost all of the apparent growth between 2003 and 2007.

Organisational actors can use evidence as symbolic proof of organisational legitimacy. Evidence can also be symbolically managed to legitimise policy action. Where policy decisions are made for ideological reasons, evidence which confers pragmatic legitimacy on these decisions can be a particularly powerful tool. For example, evidence showing productivity growth or declining unemployment is often used to highlight the success (or failure) of government initiatives. Scientific evidence has a symbolic value far beyond the intrinsic value of the data which conceals the social construction of empirical categories. This enables policy makers to construct evidence and use it to legitimise their actions through selective presentation. But while decisions made as how to construct Gross Domestic Product or unemployment statistics are highly political, (relative) consensus as to what is meant by unemployment or national output place limits on the scope for political manipulation. To some extent it may be that this is because unemployment and recession are policy problems. However like Lysenkoism, social enterprise is a policy solution which perhaps makes it easier to redefine according to prevailing ideologies. The wider contribution of this paper has been to show that where a policy solution such as social enterprise is highly contested, governments have more scope to select those aspects which fit its own ideological agenda and provide legitimacy to policy actions.

A wide range of organisational types lays claim to the social enterprise concept. Each has its own compelling stories to tell. This has enabled governments to selectively draw attributes from competing myths in order to create a loose policy construct of social enterprise. Concealing ideological underpinnings of pervasive belief in market based systems, and a desire to reduce the direct state provision of public services, this policy solution shifts over time and can seemingly be adapted to fit wider social and political contexts and respond to pressing social problems as they arise.

One myth accepted by most protagonists is that of social enterprise growth. It is here that those attempting to promote social enterprise as a policy solution want to show the increasing importance of social enterprise to bolster their own legitimacy, and as academics we are not immune from this. In particular the myth of social enterprise growth has enabled policy makers to lay claim to the success of their policies and to justify future policy action using social enterprise. But as this paper has shown, the myth of social enterprise growth can primarily be explained by political decisions to reinterpret key elements of the concept in response to shifting ideologies.

Even where growth has been measured using the same (ASBS) survey, this is not a consequence of new organisations being formed as social enterprises, but rather existing organisations claiming to meet the social enterprise definition. We would argue that as the language of social enterprise (and the DTI definition) has infiltrated popular discourse, more individuals are aware of the social desirability or normative legitimacy attached to self-identifying with this definition. Whether this growth reflects any real change in organisational behaviour is far from clear.

So the myth of social enterprise growth has been constructed through governments being able to manipulate survey methods, and to selectively present evidence supporting the myth while concealing contradictory evidence. The widely referred to figure of 62,000 social enterprises does not accurately represent the types of (socially owned) organisations which government policy is aimed at. Despite being aware of this, governments have used the 62,000 figure to justify the success of their policy measures. So just as organisations use evidence symbolically to portray legitimacy (Brown 1994), governments can manipulate perceptions through the selection of data collection tools and reporting strategies. This would seem a form of symbolic action, used to legitimate policy measures driven by ideology, particularly pertaining to the encouragement of TSOs to adopt more sustainable (market-orientated) funding mechanisms and to become involved in the delivery of public services.

Our study has direct implications for practitioners and researchers. For entrepreneurship research, there is a need to go beyond accepting that definitions can change over time due to political influence, and to recognise that evidence is a social construct. Research that explores the creation of such statistics rather than taking them as a singular truth is needed. For practice and policy actors there is a common need to challenge evidence based on loose definitions, and for those using such data, to be explicit on what they are collecting, what they are presenting and what they are excluding. With respect to social enterprise, this paper shows that understanding the concept is not possible without understanding how myths are constructed and how such myths influence (or are influenced by) ideologies in policy and practice.

Acknowledgements

The authors would like to thank the editor of this journal for his constructive comments. This paper derives from a programme of work being carried out at Birmingham and Southampton Universities as part of the Third Sector Research Centre. The support of the Economic and Social Research Council, the Office of Civil Society, and the Barrow Cadbury UK Trust is gratefully acknowledged.

Notes

Source: ASBS 2007 (weighted). Categories in bold usually associated with the ‘Third Sector’.

*Includes Community Benefit Societies.

Source: NSTSO 2009, unweighted responses = 48,939.

1. See Nicholls (2010b) for an excellent account of the legitimization of the social entrepreneurship field by different institutional actors.

2. For an overview of the different legal forms adopted by social enterprises in the UK see Smith and Teasdale (2012).