The measurement of social impacts in rural social enterprises: a systematic literature review and future research implications

ABSTRACT

Rural social enterprises (RSEs) represent an emerging actor in rural and local socio-economic development. The study of RSEs recognizes the importance of place-based actions for development. Social impacts of RSEs have been touted, particularly in filling roles in the rural context that are underperformed by governments and private actors. However, RSEs’ social impact measurement remains emerging. This review confirms that measurement of RSEs’ social impact, and its attribution to RSEs’ interventions, remain underdeveloped and lacking in both rigour and consistency. Solutions are proposed that contribute to improved methodological approaches applicable to rural regions and appropriate for related interventions confronting spatial disparities in rural development.

KEYWORDS:

• rural social enterprise
• social impacts
• social impact measurement
• social entrepreneurship
• rural areas
• local development
• rural development

JEL CLASSIFICATIONS:

• L31
• P25
• R58

1. INTRODUCTION

Social entrepreneurship, a field that spans disciplines (Bruin & Teasdale, 2019; Morris et al., 2020; Saebi et al., 2019) is emerging in regional studies and economic geography to address roles that place-based actions (cf. Horlings, 2015) play within and for their regions (Steiner & Teasdale, 2019). Social enterprises (SEs) that are focused on rural development (thus part of mainstream rural businesses (Norris, 2020), hereafter rural social enterprises (RSEs),¹ have the potential to contribute to solving intractable challenges such as poverty and inequality, which are disproportionately faced in rural areas (Breau & Saillant, 2016; World Bank, 2021). RSEs deliver products and services in rural areas using market-based approaches (Musinguzi et al., 2022aa, 2023; Steiner & Teasdale, 2019; van Twuijver et al., 2020). Programmatically, they are a development strategy that reduces rural poverty and inequality through their social impacts (Eversole et al., 2014; Musinguzi et al., 2022a, 2022bb, 2023; Olmedo & O’Shaughnessy, 2022; O’Shaughnessy et al., 2022).

In development, rurality (Pike et al., 2010) commonly poses challenges related to inadequate resources and service supply (Steiner & Teasdale, 2019; Ward & Brown, 2009). Restricted product and service provision negatively affects rural dwellers and leads to their disproportionate exposure to grand challenges such as high poverty rates and inequality (World Bank, 2021). These factors make rural areas undesirable for the operations of mainstream private and government service providers. It is thus vital to focus on ‘rural’ towards achieving local and regional development (Leeuwen, 2019; Ward & Brown, 2009).

Cavanaugh and Breau (2017) remark that ‘regional scientists have yet to focus much attention on understanding the changing dynamics of inequality across rural regions’. Martin (2021, p. 153) argues that scholars conducting regional studies should ‘take a much more explicitly progressive stand and strive through their theoretical, empirical and public engagement for fair and equitable regional and local outcomes’. However, rural areas have received little attention in social entrepreneurship (Steiner et al., 2019; Steiner & Teasdale, 2019; Weerakoon, 2021). Muñoz (2010) set out a geography-oriented research agenda for social entrepreneurship, featuring spatially oriented social entrepreneurship studies. However, this field is multidisciplinary, and there persists limited available synthesis by systematic literature review (SLR) of studies of RSEs (see van Twuijver et al., 2020, for an exception featuring European RSEs), particularly of social impacts and their measurement.

Social impacts, and their measurement, in social entrepreneurship remain underdeveloped theoretically and empirically (Hertel et al., 2020; Organisation for Economic Co-operation and Development (OECD), 2015; Rawhouser et al., 2019). Initiatives towards theoretical/conceptual understanding of social impact measurement have appeared (e.g., Arena et al., 2015; Bagnoli & Megali, 2011; Grieco, 2015; Hertel et al., 2020; Irene et al., 2016), as have efforts in empirical social impact measurement (e.g., Rawhouser et al., 2019). However, the existing theoretical/conceptual frameworks lack an explicitly rural approach, notably on the definition of social impacts, while received empirical work does not address the measurement of these impacts and associated methods.

Many published social entrepreneurship studies report positive social impacts, generating concerns about potential bias towards ‘success stories’ (Dees et al., 2008, Amin et al., 2002, cited in Muñoz, 2010). This is resounded by Vázquez-Maguirre et al. (2018, p. 327) who note that social impact as a ‘concept is encumbered with the pragmatism of administrative theory, characterized by the documentation of only good practices or successful business cases’. Thus, such positive narratives that lead to assuming that SEs generate positive social impacts on their beneficiaries’ livelihoods should be further assessed’ (Vázquez-Maguirre et al., 2018). Davies et al. (2019, p. 1619) note that ‘much of the current SE literature has adopted a positive management frame in which advantageous values, virtues and impacts are proselytized’ rather than a critical analysis of the field (Dey & Steyaert, 2012). Thus, understanding and measuring SEs’ social impact could enable cognizance of the importance and contribution of SEs in development, and indeed could contribute to the establishment of SEs’ legitimacy (Molecke & Pinkse, 2020; Sarpong & Davies, 2014). Without such understanding, there exists a danger of overhyping SEs and their being discarded in the future as a development fad (Lyon, 2009) in general terms and particularly within the local and rural development context which is a focus for this study.

This paper responds to these gaps in the literature with a SLR. It has the following objectives: (1) to map extant RSE studies measuring social impact, and their measurement approaches; and (2) to propose future social impact measurement approaches in RSEs, based on gaps identified.

To the authors’ knowledge, it is the first SLR addressing these subjects. It extends to 21 empirical studies published in the period 1987–2019, and introduces a fresh conceptualization, understanding and application of social impact measurement in rural development. It advocates methodological approaches such as: (1) a definition of social impact, through applying relevant existing frameworks within RSE studies; (2) identifying the appropriate operational level for social impact measurement; and (3) identifying rigorous analytic techniques, specifically for attribution. We attempt to advance methodological rigour in RSEs’ social impact measurement, and thus advance both scholarship and management of RSEs. We also answer Harrison et al.’s (2019) call to extend the conceptual and methodological boundaries of regional studies as well as Martin’s (2021, p. 154) agenda of advocating methods for articulating ‘an unassailable case for reducing spatial disparities’ particularly in rural areas.

The remainder of the paper is structured as follows. Section 2 highlights key existing literature focussing on social impact measurement. Section 3 presents the SLR approach that employs the preferred reporting items for systematic reviews and meta-analyses and the sample, phenomenon of interest, design, evaluation, research type frameworks for assembly and analysis of material. Section 4 presents the results. Section 5 discusses the findings reflecting on four methodological themes and future research directions. Section 6 concludes the paper.

2. SOCIAL IMPACTS OF SEs AND THEIR MEASUREMENT

Social impact of SEs is referred to variously as social value, public value, social performance, social returns, social output, environmental performance, triple bottom line, social performance, environmental impact, social accounting, non-financial performance, and others (Hertel et al., 2020; Rawhouser et al., 2019). This inconsistent nomenclature impedes academic advancement (Rawhouser et al., 2019). Despite disagreement on the nomenclature, and on the definition and measurement of SEs’ social impact (Hertel et al., 2020; Ormiston, 2019; Rawhouser et al., 2019), as explored further below, most scholars concur that understanding social impact is essential to advancing social entrepreneurship in academia and in practice (Barraket & Yousefpour, 2013; Grieco, 2015; Hertel et al., 2020; Rawhouser et al., 2019).

Measurement of social impact is variously interpreted (OECD, 2021). Vanclay (2003, p. 6) notes that it ‘includes the processes of analysing, monitoring and managing the intended and unintended social consequences, both positive and negative, of planned interventions and any social change processes invoked by those interventions’. It is beset with definitional inconsistencies even in established fields such as international development (Belcher & Palenberg, 2018) and evaluation (Vo & Christie, 2018). The OECD (2021, p. 28), states that in measuring social impacts, the following changes should be considered: ‘the effects achieved by others (alternative attribution), those that would have happened anyway (deadweight), potential negative consequences (displacement), and sustainability over time (drop-off)’.

Research in the impact evaluation field is converging towards defining social impacts in terms of changes that are brought about by actions or effects which are produced by a particular intervention, regarding the extent of achievement of an intervention’s objectives (Ebrahim, 2020, cited by Hertel et al., 2020). Disagreement on a definition of SEs’ social impact (Hertel et al., 2020; Rawhouser et al., 2019) persists: Hertel et al. (2020, p. 7) note that an emerging focus interprets social impact as ‘intended, positive effects on a target audience’. Exceptions include: Vanclay’s (2003) definition that encompasses both the intended and unintended social consequences of interventions; Pärenson (2011) who emphasizes the inclusion of both positive and negative impacts, and Nicholls (2018) who considers social impact to be the effects on targeted beneficiaries which can be both positive and negative. Amongst such variety in definition and nomenclature, the OECD (2015) notes that social impact measurement by SEs themselves is not currently widespread because they have limited resources (both financial and human) suited to the task. Further to this, SE practitioners note that pursuing social impact measurement can be unrealistic and confusing (Adams et al., 2017; McCreless & Trelstad, 2012, cited in Hertel et al., 2020). There are other reasons for choosing to measure social impact or not in SEs beyond limitations imposed by resources and these include alternative priorities for some social entrepreneurs (e.g., Barraket & Anderson, 2010; Siqueira et al., 2021), past experiences with social impact measurement (e.g., Ormiston, 2022), the rationale for measurement such as measuring performance improvement (van Rijn et al., 2021), and to obtain financial resources from new funders or meet the compliance demands of existing ones (Barraket & Anderson, 2010; Barraket & Yousefpour, 2013; Nguyen et al., 2015; van Rijn et al., 2021).

To facilitate understanding of social impact measurement in RSEs, we need to understand how SEs have previously been framed in relation to their function and disciplinarity. This provides bases for methodological elements to consider in the measurement of social impacts. Academics in the social entrepreneurship field have conceptualized SEs as development actors or as having a development model, intervention or programmatic base (e.g., Cieslik, 2016; Venot, 2016) aimed at regional/rural development in RSEs (e.g., Eversole et al., 2014; Musinguzi et al., 2022a, 2022b). It is noted that SEs combine existing development ideas/models and reframe them in new and unique ways for addressing contemporary and complex challenges of development (e.g., Chandra, 2018). SEs’ development basis does not however justify ‘positive framing’ (Dart, 2004; Venot, 2016) in the absence of critical understanding of SEs’ social impact creation.

Given SEs’ promise in contributing to development, and in the case of RSEs rural and regional development, development literature offers insights for social impact evaluation methods. The basis of impact measurement from development literature would ideally be at the core (e.g., Abadie & Cattaneo, 2018; Barnett et al., 2020; Duflo et al., 2008; Vinod & Chindarkar, 2019). This includes attribution – impact evaluation requires differentiating programme effects from other confounders (factors), and addressing selection bias (Abadie & Cattaneo, 2018; Barnett et al., 2020; Duflo et al., 2008; Vinod & Chindarkar, 2019). The emerging SE literature on social impact evaluation strongly supports this attribution emphasis (e.g., Caló et al., 2021; Rawhouser et al., 2019). Development literature also advocates the explicit use of a counterfactual. Development literature divides impact evaluation approaches into ex ante and ex post (Khandker et al., 2010, cited in Barnett et al., 2020). Whichever the case, the emphasis is on ‘what would have happened to programme beneficiaries/participants if they had not been involved’. Reporting results of programmes’ achievement according to stated objectives is insufficient to determine whether it indeed created these achievements. This is because, organizations including RSEs are involved in implementing particular interventions/programmes with targeted clients/beneficiaries based on the objectives/ aims of the intervention and the needs of the target group. The associated selection bias then requires statistical techniques which put into context the observed and unobserved variables that affect self-selection (e.g., Abadie & Cattaneo, 2018; Duflo et al., 2008; Vinod & Chindarkar, 2019).

Approaches taken to measurement of social impact in the social entrepreneurship context include mixed method, quantitative and qualitative approaches. Notwithstanding the foregoing’s tendency toward quantitative approaches, social entrepreneurship literature features mixed method approaches (both qualitative and quantitative techniques) as the most feasible (e.g., Caló et al., 2021). This is because SEs’ core purpose is mission achievement (Grieco, 2015; OECD, 2015; Peattie & Morley, 2008) which can be best measured using both qualitative and quantitative methods.

Quantitative approaches applied in social impact measurement include the use of experiments or quasi experimental methods (e.g., Caló et al., 2021; Wry & Haugh, 2018). These apply techniques such as correlation analysis, propensity score matching and multivariate difference-in-difference (DiD) models. Qualitative approaches are also applied widely, including realist evaluation employing identified context–mechanism–outcome (CMO) configurations for explaining why particular causal links materialize or not in the logic model or Theory of Change (ToC) (e.g., Caló et al., 2019, 2021). This has been applied in combination with a quasi-experimental method for attribution purposes (Caló et al., 2021).

Other qualitative approaches include contribution analysis (CA) (Befani & Mayne, 2014; Delahais & Toulemonde, 2012) and process tracing (PT) (Trampusch & Palier, 2016) that provide attribution of impacts to an organization. In terms of data, these social impact evaluation methods are based on narrative causal statements elicited from the beneficiaries/participants of interventions and other relevant actors without use of a control group (BetterEvaluation, 2016). The attribution logic appears through the beneficiaries’ own accounts of the casual mechanisms as analysed through the lens of CA and PT (Befani & Mayne, 2014; Punton & Welle, 2015a, 2015b).

CA is a process-based evaluation technique developed by Mayne as a response to the issues which arose in the assessment of cause and effect of complex interventions extending to policies, programmes, services or other interventions for which experimental designs are infeasible (Mayne, 2001). Rather, a causal chain is established alongside the relative influence of exogenous factors, in the ToC model (e.g., Befani & Mayne, 2014; Delahais & Toulemonde, 2012). CA acknowledges that there are many and complex processes at play in order to achieve an impact or outcome and thus helps in the estimation of organizations’ contribution.

PT is a qualitative method in the social sciences in which probability tests are used to assess evidence strength in specified causal relationships within a case without using a control. It helps to establish confidence in how and why an effect of an intervention occurred, hence going beyond most statistical evaluation designs (Punton & Welle, 2015a, 2015b). PT helps in understanding both the evidence and its strength (Punton & Katharina Welle, 2015a, 2015b). Some authors advise integration of these methods in conducting impact evaluation studies (Befani & Mayne, 2014). In the context of RSEs, CA and PT could enable a thorough understanding of the sequence and causal steps RSEs’ interventions have gone through to achieve the intended outcomes/social impacts. Implementation of such methods is enhanced by the use of a logic model, or ToC tools. These facilitate understanding of social impacts as proposed by the OECD (2021, cited in Hertel et al., 2020), Ebrahim (2020) and Wry and Haugh (2018), and provide conceptual clarity. They facilitate the identification of relevant indicators and cause-and-effect pathways for interventions (OECD, 2021; Hertel et al., 2020; Wry & Haugh, 2018). They are commonly used in the field of evaluation (Clark et al., 2014; Ebrahim, 2020) and are receiving attention in social entrepreneurship (Hertel et al., 2020; Wry & Haugh, 2018). A logic model for a RSE outlines linkages amongst five basic components (input, activities, outputs, outcomes and impact) (cf. Clark et al., 2014; So & Staskevicius, 2015). The ToC, on the other hand, is more detailed and goes beyond the linear representation of a programme to elaborate how and why the desired change is expected to happen (Clark & Anderson, 2004). For RSE management and social impact measurement purposes, these tools can illustrate how an intervention or project should work. They offer several other functions when used in RSE social impact evaluation, such as easing goal articulation, providing better understanding of the RSE intervention or project including how goals can be achieved, guidance in planning, design and execution of the social impact measurement processes and establishing the correct scope of the activities internally.

The measurement of social impacts requires an understanding of what to measure in terms of social impact indicators and this is aided by a well-developed logic model/ToC. Existing performance frameworks in social entrepreneurship provide indicators (also termed dependent variables in the field of economics) commonly used for measuring social impacts on their beneficiaries/clients (independent variables in economics). Common dependent variables include: percentage of clients finding a permanent job in work integration SEs; poverty reduction; contributions to local associations; net income changes; job creation; increased skills or knowledge; promotion of democracy and civic engagement; citizen engagement; democratization and political advocacy; and women’s empowerment (e.g., Bagnoli & Megali, 2011; Irene et al., 2016; Penna, 2011; Slaper & Hall, 2011). In the context of RSEs, there are currently no explicit indicators. Emerging literature (e.g., Musinguzi et al., 2022) note the following:

increased independence of participants; reduced isolation of participants due to increased capability to travel through rural transport provision; improved access to health and related care services; economic resilience through job creation; addressing rural market failure through promoting local commodity marketing; stimulation of voluntary and collaborative community culture; supporting and building skills amongst young people; environmental education; promoting sustainable energy and agricultural production etc.

Other available information sources include the RSE Hub established by Inspiralba Ltd,² which lists information on RSEs and what they have achieved. Other RSE-related indicators are available from the Global Reporting Initiative (GRI), which contains general and sector-specific indicators (GRI, 2021).

Given the many possible social impact indicators of SEs, their reduction could be vital in cases where comparison between many RSEs is a concern. To this end, multidimensional indices are emerging such as Hertel et al.’s (2020) civic wealth index. There are existing multidimensional indices such as the multidimensional poverty index (MPI) (Alkire & Santos, 2013) from development and economics fields that could be adapted to measure poverty reduction as an indicator for social impact in RSEs.

The social impact indicators mentioned in the context of RSEs link to rural livelihoods and well-being, which in turn connects to an existing overarching approach within the development literature: the rural livelihoods or sustainable livelihoods framework (DfID, 1999; Scoones, 2015). This framework has an emerging linkage to rural social entrepreneurship studies (Laeis & Lemke, 2016; Masukujjaman et al., 2016). Borrowing of theories and/or concepts and frameworks from other fields to enrich an emerging field such as (rural) social entrepreneurship has been encouraged by other researchers such as Vo and Christie (2018) regarding research between evaluation and impact measurement and Bruin and Teasdale (2019) and Haugh (2012) on social entrepreneurship.

3. METHODOLOGY

Researchers have undertaken a number of SLRs on aspects of social entrepreneurship, and SEs. SLRs have been preferred by scholars to narrative literature reviews, as they overcome researcher bias through comprehensive search and analysis strategies such as cross-referencing between researchers, extensive search of scientific databases and use of a transparent exclusion/inclusion criteria (Phillips et al., 2015; Tranfield et al., 2003). For instance, Gupta et al.’s (2020, p. 211) SLR identifies 21 earlier SLRs. There are also a number of related SLRs in the social entrepreneurship field (e.g., Aliaga-Isla & Huybrechts, 2018; Bansal et al., 2019; Bozhikin et al., 2019; Buratti et al., 2022; Persaud & Bayon, 2019; Phillips et al., 2015; Rawhouser et al., 2019; Roy et al., 2014; Saebi et al., 2019; Short et al., 2009; Stephan & Drencheva, 2017; Suchowerska et al., 2019; Tan et al., 2019; van Lunenburg et al., 2020; van Twuijver et al., 2020). Out of this number of SLRs, just two have to date explicitly covered issues related to RSEs, specifically as a solution to contemporary rural development challenges globally (Buratti et al., 2022) and in Europe (van Twuijver et al., 2020). These papers mention some impacts/contributions of RSEs but do not address issues considered in this study. Only one study, Rawhouser et al. (2019) reviews social impact measurement but these authors note that the data used in their SLR are mostly drawn from papers from developed or industrialized countries. The review also does not conduct a detailed analysis of social impact measurement methods from empirical studies, nor does it have an explicit focus on rural and regional development aspects.

Our SLR uses an adapted version of PRISMA (Moher et al., 2009 in Krupoderova & Portnov, 2020). The PRISMA is a protocol that facilitates the understanding and appraisal of methods for review (Moher et al., 2009). This approach is commonly applied in medical science reviews but it is increasingly being applied in business and management including social entrepreneurship (Aliaga-Isla & Huybrechts, 2018) with an emerging application in Regional Studies, Regional Science (e.g., Krupoderova & Portnov, 2020). The PRISMA framework provides a checklist that facilitates preparation and reporting of a rigorous protocol for a SLR (Moher et al., 2009). It generates comprehensive and replicable search results, that is, identification of a range of studies of interest for a particular topic for determining the known and unknown for a given period (Denyer & Tranfield, 2009) and thus is advantageous when compared with a random search of information which can be selective regarding thematic coverage and resulting summary findings.

We developed and employed a template (see Table A1 in Appendix A in the supplemental data online) for consistent data extraction, with variables used obtained from: (1) authors’ judgement on suitability based on the objectives of the SLR; and (2) existing analytical tools used for SLR synthesis (e.g., Sample, Phenomenon of Interest, Design, Evaluation, Research type framework – SPIDER; Cooke et al., 2012) that was adapted (Table 1). The SPIDER is a framework suitable for use in review studies of qualitative and mixed method research designs (Cooke et al., 2012). It assists researchers in mapping ‘the primary dimensions of research designs that involve qualitative or mixed research methods’ (Suchowerska et al., 2019, p. 7).

Table 1. Theory and method applied in the studies.

			Research design
Reference	Phenomenon of interest	What was evaluated	Research type	Data collection method	Sampling strategy	Sample	Theory/concept
Caló et al. (2019)	The impact of an SE-led activity on beneficiaries	Stakeholders’ perceptions of how SEs produce positive health outcomes	Qualitative (with a comparison group) (2)^a	(In-depth semi-structured) Interviews	Purposive	68 beneficiaries, service providers and external stakeholders	Realistic evaluation theory and Theory of Change (ToC)
Steiner and Teasdale (2019)	The prevalence of SEs in rural areas and their contribution to rural development	Perceptions of SE stakeholders about the contribution of SEs to economic resilience	Qualitative (descriptive without any comparison group) (2)	Interviews	Purposive	11 stakeholders	Implicit: Rural development and social entrepreneurship
Franzidis (2018)	The business model of an SE and its importance to stakeholders	Mechanisms through which SE can disseminate value among its stakeholders for equity purposes.	Qualitative (without a comparison group) (1)	Interviews, field observation, photographic documentation and document review	Purposive	Sample number not specified but it includes business founders, managers and SE employees	Implicit: social entrepreneurship
Barstow et al. (2016)	The importance of SE intervention to solving rural energy and water problems	Use of improved cookstoves and advanced water filters by the poorest quarter of households	Quantitative (cross-sectional with a comparison group) (1)	Survey	Random	198,319, i.e., 99,515 and 98,804 community members in the first and second rounds of the survey respectively	Implicit: community development
Butler and Lobley (2016)	Efficacy of SE in offering training and skills development in rural areas	Participants’ perceptions of social and economic impacts of training programmes	Mixed method (longitudinal) (3)	Interviews (telephone) and a survey	Purposive	50 participants	Implicit: performance measurement
Cieslik (2016)	How rural Burundi’s agrarian communities accommodate and benefit from the SE model	Participant’s perceptions of the importance of SE	Mixed method (without a comparison group) (9)	Interviews, focus group discussion and document review	Purposive	Nine communities; senior staff, field workers, three group leaders and group members	Moral economy
Farmer et al. (2016)	The mechanisms of well-being generation from an SE	The SE participants’ well-being impacts as a result of ‘being inside’ an SE	Qualitative (ethnographic without a comparison group) (1)	Mental mapping and (walking) interviews	Purposive	13 beneficiaries	Spaces of well-being and therapeutic assemblage
Vázquez-Maguirre et al. (2016)	The SE factors that enable the empowerment of women and influence local community development	The worldviews of local Indigenous communities	Qualitative (descriptive without any comparison group) (1)	(In depth semi-structured) Interviews	Purposive	70 participants; current and former SE, suppliers, clients, community members and government	Empowerment
Mohanan et al. (2016)	The impacts of social franchising in the health sector	The healthcare outcomes from appropriate treatment for childhood diarrhoea and pneumonia for children aged five and below	Quantitative (quasi experimental with a comparison group) (1)	Survey (household)	Random	67,950 children below five years (36,315 children living in 21,646 households in 2011; and 31,635 children in 21,367 households in 2014)	Impact evaluation and programme ToC
Munoz et al. (2015)	The conceptualization of SEs as spaces of well-being	The SE’s participants perceptions of well-being benefits from the SE	Qualitative (ethnographic without a comparison group) (1)	Ethnographic observation, focus group discussion and (walking) Interviews	Purposive	Three paid staff and 21 volunteers	Spaces of well-being and therapeutic assemblage
Spencer et al. (2016)	Effectiveness of an SE	The Indigenous people’s perceptions of impacts obtained from participating in an SE within their community	Qualitative (descriptive without any comparison group) (1)	(Semi-structured) Interviews, participant observation and Yarns	Snowball	24 participants (15 staff and management and 9 community stakeholders)	Social entrepreneurship (Indigenous)
Holt and Littlewood (2015)	Measuring SEs’ social impacts	The perceptions of SE stakeholders’ impacts from SEs	Qualitative (without a comparison group) (20)	Interviews and observation	Purposive	20 SEs	Hybrid organization theory
Kabeer and Sulaiman (2015)	Impacts of SE’s social mobilization and livelihood improvement	Member’s perceptions of Nigera Kori’s (NK) political, economic and social impacts	Quantitative (quasi experimental with a comparative group) (1)	Survey (semi-structured), interviews and focus group discussion	Random	500 participants (250 SE members and 250 non-members)	Realistic evaluation theory and ToC concept
Bonny and Rajendran (2013)	The conditions under which women’s empowerment is related to value creation in self-help groups (SHGs)	The relationship between women empowerment and value creation with in SHGs	Quantitative (correlational analysis without a comparison group) (50)	Participant observation and (key informant) interviews	Proportionate	50 women SHGs from 22 public–private partnership models of women entrepreneurship development	Implicit: women empowerment
Sakata and Prideaux (2013)	The impacts of an SE within community-based ecotourism (CBT)	Indigenous community members’ perceptions of the impact of CBT based SE on their livelihoods	Qualitative (descriptive without a comparison group) (1)	Interviews and focus group discussion	Purposive	Six stakeholders (guest house owner, manager), village councillor, dance group leader, tour guide leader and school head	Social exchange theory and community agency
Tobias et al. (2013)	How entrepreneurship may catalyse prosperity and peace in poverty and conflict-stricken areas	The interrelationships between poverty and conflict from participating rural community members	Quantitative (cross-sectional without a comparison group) (1)	Survey	Convenient	239 coffee workers	Entrepreneurship
McKague and Tinsley (2012)	The SE rural distribution model as a possibility of combining market-based solutions to poverty reduction	Stakeholders’ perceptions of the impacts of the SE rural distribution model.	Mixed method (cross-sectional and longitudinal (on a few variables) without a comparison group) (1)	Participant observation, field observation, (face to face) interviews and document review	Purposive	25 rural sales women, Bata employees and CARE staff	Implicit: social entrepreneurship (SE business model)
Pless and Appel (2012)	How an SE can realize a vision of an equitable and sustainable society	The SE stakeholders’ impact perceptions accruing from their participation in the SE’s activities	Qualitative (1)	Desk research (existing reports and papers), workshops, focus group discussion, interviews and field visit	Purposive	109, i.e., 37 community members, 14 senior management and other staff members,11 local village level staff, 45 regional managers and two external stakeholders	Social entrepreneurship and community development
O’Shaughnessy et al. (2011)	The role of SE in improving rural transport	Rural local residents’ perceptions of impacts derived from transportation services of an SE	Mixed method (mainly descriptive without a comparison group) (1)	(In-depth) interviews	Purposive	Sample not specified but it includes selected passengers and key stakeholders	Social entrepreneurship
Lapeyre (2010)	The contribution of community-based tourism enterprises (CBTEs) in rural areas	Perceived impacts of poverty alleviation and community empowerment by the SE on rural households’ livelihoods	Qualitative (without a comparison group) (1)	Survey, semi-structured interviews, focus group discussion and participatory observation	Purposive	16 tour guides (number of household members or other stakeholders not provided)	Global commodity chain
Torri (2010)	Community-based enterprises (CBEs) as an alternative model for enhancing marginalized rural livelihoods	The marginalized rural participants’ perceived impacts of enhanced livelihoods and socio-economic development from SE intervention	Qualitative (descriptive without any comparison group) (1)	Interviews and focus group discussion	Snowball and random	36, i.e., 12 staff, four field coordinators and 20 community members	Community entrepreneurship

Note: ^aNumbers in parentheses indicate the number of SE cases involved in each study.

Our search strategy employed a comprehensive search of relevant databases.³ Databases were selected with advice from a University librarian as they are main sources for business, management and regional science information. Further to this, they are recognized internationally as academic sources that cover journals indexed for impact factor.

The search was confined to the period 1987–2019, sufficient to capture RSE studies given the field’s evolution (Short et al., 2009). The 1987 starting year was also used in a related SLR (Phillips et al., 2015), while the most recent related SLR uses the period 1996–2016 (Rawhouser et al., 2019). Our study deviates from most prior SLRs by its slightly longer time frame, and by its explicit rural focus.

Search terms⁴ featured variants on the names given to SEs/RSEs, for example, social entrepreneurship, social entrepreneurial organizations, social ventures, social entrepreneurial ventures, community-based enterprises plus other selected keywords related to rural/regional development (see Table A2 in Appendix A in the supplemental data online), and accessed only English language publications. We use peer review as an indicator of academic quality (Kraus et al., 2020), as well as SCImago Journal Rank and Scopus quartile rankings (Thananusak, 2019) (see Table A3 in Appendix A online). Following Heyvaert et al. (2013), our inclusion criteria after full text review required (1) clearly defined social impacts, and indicators used to measure them; and (2) a rural focus⁵ and targeting the conditions of rural populations. Studies that were non-empirical in nature were excluded and this occasioned a sharp fall in the total number of eligible articles (21) (Figure 1).

Figure 1. Systematic literature review flow diagram based on the adapted version of the PRISMA diagram.

Sources: Moher et al. (2009), cited by Krupoderova and Portnov (2020).

4. RESULTS

4.1. Academic domain, sector and location

The 21 articles included in this SLR are drawn from 18 journals. All journals in the sample published at least one article each, with the exception of Social Enterprise Journal from which four papers were drawn. The 18 journals cover a range of domains with 14 articles under business and management, but all the articles address the broad area of community development, relevant to regional/rural development. No captured study predated 2010 (see Table A4 in Appendix A in the supplemental data online). A peak of seven studies appears in 2016. The absence of studies published between 1987 and 2010 suggests recent emergence of empirical measurement of social impacts in RSEs.

The RSEs studied span the continents, confirming that (rural) social entrepreneurship is an international phenomenon (Short et al., 2009). Thirteen of our sample articles are located in developing countries (see Table A4 in Appendix A in the supplemental data online). This may reflect not that RSEs are more studied in such countries, but rather that most of the studied RSEs were originally non-governmental organizations (NGOs) or are NGO spinoffs as well as have been funded partly by external donors and funders (e.g., Barstow et al., 2016; Bonny & Rajendran, 2013; Cieslik, 2016; Kabeer & Sulaiman, 2015; McKague & Tinsley, 2012; Mohanan et al., 2016; Torri, 2010). A further consequence of this genesis for RSEs is that they are likely to conduct social impact measurement in order to comply with their external stakeholders’ expectations, as noted in extant literature (e.g., Barraket & Yousefpour, 2013; Nguyen et al., 2015).

4.2. Studies’ methods, samples and theoretical basis

The most commonly used data collection methods (Table 1) include interviews (used in 11 studies), focus group discussions (six) and participant observation (six), followed by surveys (four), document reviews (two), ethnography (one) and workshops (one). All studies in the sample used a case study approach: 14 used a single case, two used two and the remaining five each used three, nine, 10, 20 and 50 cases, respectively. Further, 13 papers in our sample were qualitative in nature and applied qualitative techniques such as realist evaluation to measure social impact. Others conduct thematic and content analysis to describe respective RSEs’ social impacts. Fifteen studies explicitly state a theory and/or a concept and six more implicitly do so. Thirteen theories/concepts have been applied in situating the theoretical underpinning of the studies and they include the following with the number in parentheses indicating the number of studies applying that theory/concept: social entrepreneurship (six), ToC (three), impact evaluation (two), realist evaluation (two), spaces of well-being and therapeutic assemblage (two), empowerment (two), global commodity chain (one), community/rural development (two), hybrid organization (one), moral economy (one), social exchange/community agency (one), entrepreneurship (one) and community entrepreneurship (one) (Table 1).

Eight studies used quantitative methods for social impact measurement: correlation, propensity score matching (PSM) and DiD. Two studies used a quasi-experimental design to address selection bias, 17 studies used purposive (non-probability based) sampling, while in four it was random. These findings indicate that self-selection and attribution are in general not accounted for in social impact evaluation of RSEs. Of the eight studies that applied a quantitative approach to social impact measurement, five papers used solely quantitative methods while three used mixed methods.

4.3. Definition and terminology for social impact measurement

As Table 2 shows, social impact is defined in various ways across the studies included in this SLR, which concurs with extant literature regarding definitional issues (e.g., Grieco, 2015; Hertel et al., 2020; Rawhouser et al., 2019). Although this poses a challenge to researchers, we adhere to the term social impact, and its measurement (as in, e.g., Barraket & Yousefpour, 2013; Rawhouser et al., 2019; van Rijn et al., 2021), rather than creating further nomenclature controversy. With regard to the definition, an emerging agreement from our study is that social impact in this context refers to the outcomes that emanate from RSEs’ interventions. Although most of the included studies in this SLR do not measure negative or unintended outcomes, in defining social impact, RSE studies should recognize that social impacts can be intended/unintended as well as positive/negative (e.g., Nicholls, 2018; OECD, 2021; Pärenson, 2011; Vanclay, 2003).

Table 2. Terminology, definition and measurement of RSEs’ social impacts.

			Operationalization of SE impact
Reference	Terminology for SE impact	Definition of SE impact	Independent variable	Dependent variable	Social impact overall direction	Use of comparison group	Covariate adjustment method	What was measured
Caló et al. (2019)	Social outcomes	Improved health outcomes of SE participants	Participation in the SE’s programme	Perceptions/feelings of connectedness, inclusion and protection	Positive	Yes	None	Outcomes
Steiner and Teasdale (2019)	Impact	Outcomes from SE interventions	Participation in the SEs’ activities	Economic resilience Investments in the local areas Local service provision Community cohesion and added value	Positive	No	None	Outcomes
Franzidis (2018)	Impact	High-quality tourism product that disseminates social value among stakeholders	Participation in the SE’s livelihood programmes	Education Employment Local community support	Positive	No	None	Outputs and outcomes
Barstow et al. (2016)	Impact	Adoption of improved cook stoves and advanced water filters	Participation in the SE’s cook stoves and advanced water filters programme	Adoption of improved cook stoves and advanced water filters	Positive	Yes	None	Outputs and outcomes
Butler and Lobley (2016)	Social–economic impact	Changes in well-being of SE participants	Participation in training offered by the SEs	Well-being Return on investment	Minimal impacts	Yes	None	Outcomes
Cieslik (2016)	Benefits	Benefits which accrue from the SEs intervention	Participation in the SE’s (Project Lumière’s nine CPCs) activities	Education Poverty reduction Health outcomes Environment	Positive	No	None	Outcome
Farmer et al. (2016)	Benefits	Well-being impacts on people’s lives	Participation in the SE’s well-being programme	Health Well-being perceptions	Positive	No	None	Outcomes
Vázquez-Maguirre et al. (2016)	Benefits	Well-being of women	Participation in SE’s activities	Employment Provision of no-interest loans Political empowerment	Positive	No	None	Outcomes
Mohanan et al. (2016)	(Health) outcomes	Outcomes from the SE activity	Participation in the SE intervention	Prevalence of key children’s health outcomes Changes in parents’ healthcare practices for children	No impacts	Yes	Difference-in-difference	Outcomes and impacts
Spencer et al. (2016)	Social effectiveness (social performance)	Employment and social capital creation in the local community	Participation in the SE programmes and activities.	Social effectiveness; changes in beneficiaries’ livelihoods (job creation and readiness, income generation, and social capital	Positive	No	None	Outputs and outcomes
Holt and Littlewood (2015)	Impact	Social and economic changes in the lives of the SE participants	Participation in the different SEs’ activities/programmes	Income status Relational well-being benefits Community improvement	Mixed results	No	None	Outcomes
Kabeer and Sulaiman (2015)	Impact	Livelihoods improvement, political knowledge and local and national participation	Membership and participation in the SE activities	Improved livelihoods Improved knowledge of politics and policy Community participation	Positive	Yes	Propensity score matching	Impact
Munoz et al. (2015)	Benefits	Production of health and well-being benefits	Participation in the SE’s well-being programme	Health and well-being perceptions	Positive	No	None	Outcomes
Bonny and Rajendran (2013)	Social value	Value creation associated with empowerment of group members	Women's participation in SHGs as SEs	Social value creation – women empowerment)	Positive	No	None	Outcomes
Sakata and Prideaux (2013)	Impact	Community-defined positive/negative impacts of economic tourism	Participation in SE’s activities	Poverty reduction Social capital creation Economic impact	Positive	No	None	Outcomes
Tobias et al. (2013)	Social value	Poverty and conflict reduction	Participation in the SE programme	Wealth Quality of life Conflict	Positive	No	None	Outcomes
McKague and Tinsley (2012)	Impact	Improved livelihoods	Participation in the SE’s programme	Poverty reduction: Improved incomes Accessibility to socially beneficial goods	Positive	No	None	Outcomes
Pless and Appel (2012)	Impact	Improvement in health, restoration of dignity, empowerment of women and poverty reduction	Participation in SE programme activities	Health Education Livelihood improvement Environmental sustainability	Positive	No	None	Outputs and outcomes
Torri (2010)	Social–economic impact	Empowerment, capacity building and increased income	Participation in the SE’s activities	Increase in income Enhanced social status	Positive	No	None	Outcomes
(2011)	Impacts	Improving access to remote areas for marginalized rural users	Participation in the SE’s rural transportation services	Level of independence Perceptions of isolation Access to health and related care services	Positive	No	None	Outcomes
Lapeyre (2010)	Livelihood impacts	Improved livelihoods SE participants	Participation in SE intervention	Income status Stability in employment Capacity-building Social capital	Mixed results	No	None	Outcomes

4.4. Measurement of RSEs’ social impact

4.4.1. Independent and dependent variables

In the absence of studies’ explicit classification of variables employed in their analyses, we categorized them as independent and dependent. The independent variable common to all the 21 studies was participation in (including membership/beneficiary of) the RSE. The dependent variable (social impact) varied (Table 2).

4.4.2. Attribution and causal inference

Rigorous measurement of social impacts of RSEs satisfies key methodological issues such as the use of a comparison/counterfactual and selection bias to attribute the social impacts to the RSEs. This can be achieved with both theory-based (e.g., PT and CT), quantitative (e.g., PSM) and DiD approaches. Thirteen qualitative studies in our sample do not use any comparison group nor a rigorous theory-based technique (that attributes social impacts to the RSE) when measuring the social impacts of RSEs. For the eight quantitative and mixed method studies, only three employ a comparison and two of these apply a technique to enable causal inference (Table 2).

4.4.3. Variables measured

We summarize studies’ measurement of activities, outputs, outcomes or impacts (see Table A5 in Appendix A in the supplemental data online for details). The most commonly studied social impact (17 studies) is poverty reduction in terms of well-being/livelihoods improvement. The identified overall sign of impact of the RSEs was positive in 17 studies. Two report mixed results and the remaining show no, or minimal, social impacts. The operational level at which social impacts are measured is uniform across studies, with 20 measuring social impacts at project/programme level while one combined many RSEs and analysed their impacts at organizational level. This kind of analysis masked detailed social impacts from different projects within the RSE studied. Notably, just one study attempted rigorously to measure social impacts following all the considerations for social impact measurement already described, while the remaining 20 reported a mixture of activities, outputs and outcomes (Table 2).

5. DISCUSSION AND IMPLICATIONS FOR FUTURE RESEARCH

As the first, to the authors’ knowledge, SLR oriented to regional/rural development-focused social entrepreneurial interventions’ social impact measurement, this study suggests that this scholarship is at an emerging stage. Further, we identify the need for conceptual clarity specifically in definition and in alignment with regional/rural studies and identification of approaches to enable rigorous social impact measurement. These approaches could contribute to clarity in the conceptual/underpinning of RSE social impact studies and provide reliable social impact measures essential for understanding and improving the effectiveness of regional/rural oriented interventions thus improving the practice besides stimulating appropriate research to progress regional/rural development research (Harrison et al., 2019) in the efforts of unravelling and reducing spatial disparities in welfare (Martin, 2021).

5.1. Social impact measurement in the context of rural studies

A substantial number of studies in this SLR employ theories/concepts to situate and structure their findings for conceptual clarity. A variety of theories or concepts are employed. This finding concurs with received SLRs on social entrepreneurship (e.g., Bozhikin et al., 2019; Gupta et al., 2020; van Twuijver et al., 2020). The finding that such a variety of theories or concepts are observed across a relatively small number of studies (21) is a further indication of the emergent nature of rural social entrepreneurship and more generally the social entrepreneurship discipline (Short et al., 2009). This study also identifies a number of RSE impact studies that do not explicitly employ a theory or key concept in which the research was anchored, which concurs with van Twuijver et al.’s (2020) SLR in European RSE studies which concludes that the theoretical lens of most studies in their SLR was not explicit. This result poses concerns about the rigour of some existing RSE studies, and progress towards the emergence of a distinct discipline. We argue that rural social entrepreneurship studies should indeed be anchored explicitly in a relevant theory/concept.

Despite the reported inconsistency in definition of RSEs’ social impacts, we identify an emerging definitional pattern that relates to livelihoods and well-being outcomes of the RSE target stakeholders. This strongly relates to RSEs’ overall goal: to improve the well-being/livelihoods of rural communities. We propose that this definition includes positive, negative, intended and unintended consequences of the RSEs’ interventions (e.g., Nicholls, 2018; OECD, 2021; Vanclay, 2003). Additionally, we suggest that RSE studies could also be anchored/situated in an overarching framework that is linked to regional and rural development, e.g., the rural livelihoods/sustainable livelihoods framework (DfID, 1999; Scoones, 2015). Further rationalization could lead to a heuristic framework that integrates social impact measurement with rural livelihoods.

5.2. Level at which social impacts are measured

It is vital to choose the level (intervention/project, organization or programme) at which to measure the social impacts of regional/rural development focused actions. For RSE practitioners and policymakers, the appropriate level is determined by the purpose of the analysis and the audience. However, all stakeholders in social impact measurement make rational decisions based on the resources available for data collection, analysis and reporting. As most (R)SEs are resource poor and do not prioritize social impact measurement (e.g., Barraket & Yousefpour, 2013; OECD, 2015; Siqueira et al., 2021), this consideration becomes even more important.

In most of our RSE sample studies, data collection/analysis of social impact is at project/programme level. This is vital because a RSE is in most cases involved in a variety of projects or programmes and measurement of social impact needs to consider the social impacts stemming from each. The project or programme level of measurement enables attribution (e.g., Kabeer & Sulaiman, 2015; Salazar et al., 2012). Its management advantage is its identification of project or programme-specific effects (both intended and unintended) at beneficiaries’ or clients’ levels, and this would enhance the operation of the SE to achieve its social mission which is the core reason for social impact measurement (OECD, 2015).

5.3. Selection of social impacts to measure

Selection first requires clarity on the conceptual basis of social impact and its alignment with the underpinnings of rural studies (see section 5.1 above). Second, it achieves clarity by the use of a ToC/logic model (Ebrahim, 2020; OECD, 2021, cited in Hertel et al., 2020; Wry & Haugh, 2018), which establishes linkages amongst five basic components (input, activities, outputs, outcomes and impact) of an intervention (OECD, 2021; Hertel et al., 2020; Wry & Haugh, 2018), and provides a basis for attribution.

We detect many dependent variables as measures of social impact. This creates difficulties in comparison of social impacts across sectors or contexts. Opportunities for improvement lie in measurement in similar contexts (such as rural settings), and the use of social impact variables that are indirect measures of human development or welfare (Salazar et al., 2012). Indicators drawn from the GRI contain appealing general and sector-specific indicators (GRI, 2021), as one example. The additional application of the ToC/logic model and the rural livelihoods framework would create an opportunity for comparison of social impacts. This would be strengthened by application of multidimensional indices, for example, multidimensional poverty index (Alkire & Santos, 2013) and Hertel et al.’s (2020) civic wealth index.

5.4. The measurement of social impacts

Because most of our sample’s studies are from business and management, they lean towards management research (Wry & Haugh, 2018). One study in our sample (Kabeer & Sulaiman, 2015) applies the logic model/ToC and attribution by way of a comparison group and PSM), exemplifying the feasibility of a mixed method approach. However, purely quantitative and qualitative approaches can also be applied in RSEs’ social impact measurement. Realist evaluation, which is a qualitative approach that involves identifying CMO was applied by a few of the studies in our SLR. Other qualitative techniques that can be applied to increase the rigour of social impact measurement, such as PT and CA, need to be applied together with a ToC/logic model to describe concisely the RSEs’ causal chain.

RSEs are involved in development through implementing particular interventions/programmes with targeted clients/beneficiaries (e.g., Cieslik, 2016; Kabeer & Sulaiman, 2015). In practice this introduces selection bias, and dealing with this requires random assignment of beneficiaries (treatment) to the interventions (Abadie & Cattaneo, 2018; Kabeer & Sulaiman, 2015; Vinod & Chindarkar, 2019). Rigorous social impact measurement also requires consideration of: (1) a comparison/control group to be compared with the treatment; and (2) a counterfactual (Abadie & Cattaneo, 2018; Kabeer & Sulaiman, 2015; Vinod & Chindarkar, 2019). Table A6 in Appendix A in the supplemental data online summarizes these identified key issues and suggests ways they can be dealt with.

6. CONTRIBUTION AND CONCLUSIONS

We respond to Harrison et al.’s (2019, p. 136) call for regional studies’ ‘approaches that, in combination, result in the pushing on (by creating), pushing off (by consolidating), pushing back (by critiquing) and pushing forward (by collectively constructing) the field’ by focussing on the ‘local and regional development perspective’ theme. Specifically, we ‘push back’ by critiquing RSE studies that have measured social impacts and identifying concepts/theories and methods which have been applied. We ‘push forward’ by suggesting a mix of approaches for social impact measurement suitable for (1) RSE practitioners and managers, (2) policymakers and (3) future researchers. In achieving these aims, we also contribute to ‘a progressive and melioristic regional studies’ agenda as called for by Martin (2021, p. 154) in order ‘to articulate an unassailable case for reducing spatial disparities in incomes, employment opportunities, health, education, housing, productivity, and well-being’ to achieve rural development.

Owing to the variation found in definitions of social impact and many variables used as indicators for its measurement in RSEs, we advocate the application of existing frameworks, such as the rural livelihoods framework, in the definition and classification of social impacts. We also advocate use of social impact measures related to human development/welfare, possibly drawing relevant indicators from the GRI, and the use of multidimensional construct measures. These steps contribute to comparability of results, both for practitioners and in the context of future quantitative meta-analyses.

Our review has also revealed minimal use of rigorous measurement methods for RSEs’ social impact. This constrains the extent to which objective analysis can proceed towards verifiable results with attribution to RSEs. Moreover, the mostly positive narratives of social impact could be rendered anecdotal. To measure effectively social impacts of RSEs, we advocate the use of a mixed method approach involving the RSEs’ ToC/logic model, and techniques (e.g., PSM) that enable attribution of social impacts to the RSEs. Rigorous theory-based impact evaluation methods such as PT and CT with the ToC/logic model could also be applied in cases of purely qualitative studies.

We find that empirical social impact measurement has been conducted primarily at project/programme level, and this offers opportunities to reduce ambiguity, and advance identification of project/programme-specific social impacts of the RSEs. At both design and implementation stages, this could lead to the improvement of the RSEs’ design and operation and thus contribute to regional/rural development.

The application of methods we suggest in this study introduces analytic rigour and enables attribution of the social impacts to RSEs. Without the use of rigorous social impact measurement methods, the resulting impacts could easily be rendered anecdotal and thus impede proper understanding and legitimizing RSEs as key partners within rural development besides missing opportunities that such findings could offer to relevant stakeholders e.g., policymakers, supporters, practitioners and researchers in the efforts of improving the effectiveness of such organizations for achieving rural development objectives. Most of the approaches described in this study are mostly from development and economics. Thus, given the emerging nature of social impact evaluation studies of social entrepreneurial organizations in the regional/rural context, practitioners, policymakers and researchers interested in rigorous social impact measurement should borrow from such fields because much can be gained from the application of diverse analytical approaches that have the potential to illuminate each other cumulatively for the advancement of sound and dependable knowledge i.e., theory and practice in this case on the social impacts of social entrepreneurial organizations particularly RSEs and other related interventions within rural development.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Additional information

Funding

The Centre for Agribusiness, University of New England (UNE) provided financial support for Open Access.

Notes

1 Social entrepreneurship still lacks a universally accepted definition of SEs (Morris et al., 2020), and consequently of RSEs. Based on the emerging rural social entrepreneurship literature (Musinguzi et al., 2022b; Steiner & Teasdale, 2019; van Twuijver et al., 2020), in this paper RSEs are defined as ‘organisations/enterprises with a social mission/goal of improving the well-being/creating positive social change for rural communities in form of social impacts achieved through the use of entrepreneurial/market based approaches’.

2 See https://ruralsehub.net/.

3 Databases searched include Web of Science, Scopus, ProQuest, EBSCO, JSTOR, Cambridge Core, Gale, Informit, Oxford Journals, Taylor & Francis, Wiley Online library, Science Direct, Business Source Complete, Emerald Insight and rural science/development (CAB abstracts). We exclude Google Scholar as a database for our SLR because it might not be reproducible (Kraus et al., 2020).

4 An example of a typical search string using a combination of key words from ProQuest is: ‘social enterprise’ AND impacts AND (Rural develop*) AND (at.exact (‘Article’) AND la.exact (‘ENG’) AND pd (19870112-20190331) AND PEER (yes)) AND pd (>19870101).

5 For studies that did not clearly state their geographic focus, we contacted authors for clarification.