Voice in context: An international comparative study of employee experience with voice in small and medium enterprises

Abstract The study of employee voice in small and medium enterprises (SMEs) across national contexts remains under-theorised and under-studied. This paper uses Kaufman’s integrative model of employee voice, and an exploratory study of 30 interviews with employees in non-unionised SMEs in the United Kingdom, Thailand and Nigeria, to compare the employee experience with voice, and the impact of this experience on voice behaviour at work. Findings show that the interaction between the external institutional context and internal SME context (organisational configuration, governance structure and internal contingencies in the employment relationship) impacts employee voice agency, the perceived levels of voice and, ultimately, employee voice behaviour. The paper contributes to employee voice theory by offering an analysis of voice determinants on voice behaviour specific to non-unionised SMEs from an international comparative employee perspective, presents these in an initial framework and explains how employees experience voice in small workplaces.


Introduction
Smaller businesses, the 'life-blood' of most economies providing employment and fostering innovation, are characterised by informality in their employment relationships (Gilman et al., 2015). Voice can bring a greater sense of job influence, and improve job satisfaction, organisational commitment and reciprocal trust in managers (Timming, 2012) and is, therefore, worth studying in the small businesses that are the majority of nearly every economy. However, in many developing economies voice Employee voice; voice determinants; comparative HRM; SMEs; UK; Thailand; Nigeria is viewed as a western concept aimed at promoting individualism (in the absence of trade union representation) and hardly consistent with collectivist cultures. Beyond trade union representation, the concept of voice has attracted very limited discussion in labour relations research in regions such as Asia or Africa, although Pyman et al. (2016), for example, offer somewhat generalised emerging voice constructs in developing economies, though these are not compared to voice in European countries (Johnstone & Ackers, 2015;Timming, 2007). International comparisons of voice in small and medium enterprises (SMEs) are, therefore, valuable.
Human resource management (HRM) in SMEs varies according to the external and internal organisational context (Bacon & Hoque, 2005;Gilman & Raby, 2013;Harney & Dundon, 2007;Wilkinson et al., 2007). Gilman et al. (2015) argue for the complexity of the SME context and consider the interaction between resources (human and social capital) and constraints (product market, labour market and strategic orientation) in shaping voice. However, other determining contextual factors of voice in SMEs remain largely under-researched and under-theorised (Marlow, 2006;. We, therefore, need research to capture both national differences and organisational contingencies (Gilman & Raby, 2013) to explain the variance in voice in smaller organisations across international settings. We build upon Gilman's et al. (2015) argument for the need to look at the interaction between voice determinants and how these shape voice in SMEs and focus our work on the following question: how does the macro-external context (institutional, socio-economic, cultural) interplay with the meso-organisational (organisational configuration, governance structure) and micro-individual contexts (internal contingencies in the employment relationship) to affect voice in non-unionised SMEs? Central to our investigation is the employee experience with voice in small settings and how the interaction of these macro-meso-micro factors may affect their voice behaviour.
We frame our paper within an inclusive definition of employee voice (Wilkinson & Fay, 2011) and conceptualise it as providing an opportunity for employees to raise issues that concern them, but also incorporate the discretionary communication of ideas or opinions about work-related issues. We adopt this inclusive definition because we want to study voice as a mechanism for sharing ideas at work, not just as a mechanism for expressing dissatisfaction or complaint.
Our definition requires an equally inclusive theoretical frame, so we utilise Kaufman's (2015) integrative model of employee voice. This is an appropriate analytical approach for conducting an empirical examination of the strategic features of voice in SMEs because it is broad enough to help us capture heterogeneity -although it does not directly explore the interaction of key macro-meso-micro level determinants within non-unionised SME contexts. This deficit, in part, motivates our qualitative context-sensitive research lens and inductive study (Eisenhardt & Graebner, 2007) which aims to capture these nuances regarding how voice changes from context to context. We follow a similar approach to Gilman et al. (2015) who utilised a priori frameworks to qualitatively draw out some determinants of employee voice in SMEs using Kaufman's comprehensive and integrative framework of voice determinants.
We explore voice in a sample of employees working in non-unionised SMEs in the United Kingdom (UK), Thailand and Nigeria because we want to understand and explain similarities and differences in employee experiences with voice between these three contexts. We chose these contexts because of the variation in their institutional bases, which makes the comparison interesting and the analysis insightful. The UK is categorised as a liberal market economy in the comparative capitalisms literature (Amable, 2003;Hall & Soskice, 2001). Thailand and Nigeria were ignored in the early literature, which focused on developed countries, but Thailand has been classified as sharing a Southeast Asian capitalist system (Andriesse & van Westen, 2009), while Nigeria is seen as evidencing a fragmented/segmented institutional form of capitalism (Amaeshi & Amao, 2009). We chose countries that do not share many commonalities in their institutional (or, indeed, cultural) settings, yet each of these countries plays a significant economic role in their respective regions: the UK ranks second in Gross Domestic Product (GDP) in Europe (Eurostat, 2018), Thailand is the second largest economy in Southeast Asia and Nigeria ranks first in GDP in Africa. The three countries are also experiencing significant economic and political developments, even before the COVID-19 pandemic: the UK is faced with the political and economic consequences of Brexit (Bulmer & Quaglia, 2018); Thailand is the only country in South East Asia to have escaped colonial rule but, over the years, civilian prime ministers have been unsuccessful at reining in the military as a result of the armed forces' close association with the monarchy (Chambers & Waitoolkiat, 2016); Nigeria has made major strides in its political development but still has significant work to do in improving national, state and local governance in line with principles of procedural legitimacy for sustaining its democratic underpinnings.
The paper contributes to employee voice theory by offering an analysis of voice determinants specific to non-unionised SMEs from an international comparative employee perspective, presents these in a framework and explains how employees experience voice in small workplaces. Kaufman's (2015) integrative employment relations (ER) model of employee voice is more comprehensive than similar frameworks (e.g. Mowbray et al., 2015) and provides a critique of how to incorporate contextual determinants at multiple levels of analysis, helping to avoid conceptual 'blind spots' that can occur with other models. These have been criticised for their tendency to focus on individualistic, psychological explanations of how and why employees decide to 'voice-up' , whilst paying less attention to institutional, HRM, industrial relations (IR) and ER contexts. The psychological explanations are relevant, but we concur with Kaufman that omitting alternative explanations of voice determinants often results in a deficient and de-contextualised understanding of the relevant factors.

Theoretical framework
For Kaufman, determinants of employee voice are intrinsically multi-dimensional and intersectional. His integrative model has interrelated dynamic elements and begins by examining the fundamental role played by the external environment within which organisations are embedded. The organisations' configuration and governance structures form a conceptual 'roof ' under which organisational dimensions, such as the employment relationship, are housed. Organisational configurations are inherently diverse, as they provide the infrastructure within which voice is expressed (or withheld), heard (or silenced), acted upon (or ignored). The governance structure refers to the management system and political structure for making key strategic decisions about policies and practices, exercising authority and influence over employee terms and conditions and choosing who (and what) to reward, punish, or exclude.
Internal contingencies are specific factors internal to the organisation and the employee. Organisational contingencies include, for example, managerial quality, organisational culture, production technology and levels of technological infrastructure in the firm. Employee contingencies include factors such as the workforce's knowledge, skills and abilities, demographics and psycho-social attitudes towards work. The next section provides the contextual background of our study.

The external, institutional, socio-economic and cultural context in the UK, Thailand and Nigeria
In the comparative capitalisms literature (Amable, 2003;Hall & Soskice, 2001), the UK is invariably categorised as a liberal market economy characterised by limited employment protection, extensive external labour markets, easy recourse to flexible working patterns, limited active employment policy by government, defensive union strategies and decentralisation of wage bargaining. Voice is, if contested, widespread (Brewster et al., 2015), at least in larger organisations. In 2017, SMEs accounted for 99% of all private sector businesses and offered 60% of all private sector employment (Federation of Small Businesses, 2020). After the 2008 global financial crisis, the UK experienced the slowest recovery in levels of output since the 1920s and only managed to reach pre-crisis levels in the second quarter of 2013 (Jackson et al., 2019). More recently, issues around Brexit and its economic impact on businesses have concerned politicians (Slater, 2016).
Thailand has been classified as sharing a Southeast Asian capitalist system, characterised by employment protection within large corporations, labour-market dualism, seniority-based wages, cooperative industrial relations, and decentralisation of wage bargaining (Andriesse & van Westen, 2009). Thailand has experienced acute political conflict since the 1970s when it transitioned from a military-dominated authoritarianism to a constitutional parliamentary regime (Brown, 2016). Since then, competing regimes attempted to structure the institutions regulating workplace participation (Connors, 2009). Thai employment relations are seen as protecting the interests of the state and marginalising those of labour. Self-employment has acted as a shock absorber during frequent economic crises, such as the 1997 Asian financial crisis and the 2008 global financial crisis, and a lack of longevity in employment makes unionisation difficult (Charoenloet, 2015). SMEs constitute 99% of all enterprises in the Thai economy and provide 73% of all employment (Charoenrat & Harvie, 2017), playing a pivotal role in the country's economic development.
The institutional context of most African economies has been described as fragmented or segmented (Amaeshi & Amao, 2009). Nigeria is characterised by a dual economy; oil production accounts for over 90% of export earnings. The rest of the economy demonstrates a developing African model; around 30% of GDP comes from agriculture while the manufacturing sector is limited (Afangideh, 2012). The drastic fall in oil prices in 2014 meant a severe rationalisation of foreign reserves which had a negative impact on SMEs (Njoku, 2016). However, Nigeria is a country with increasing business opportunities: after decades of military governments, there is now a democratic system; and the political environment is gradually settling into place. Corruption remains widespread and is a serious obstacle to economic growth, despite the government's long-term efforts to tackle it (Fagbadebo, 2007). Self-employment and the urban informal economy are growing (ILO, 2017), but Nigeria is characterised by a deficit in employment opportunities, labour rights, and systems that promote quality work (Otobo, 2007). SMEs account for 90% of businesses and provide 50% of employment and industrial output and 95% of the formal manufacturing activity (Me, 2018).
In terms of culture, Nigeria and Thailand can be characterised as 'tight cultures' , less likely to have democratic processes or economic freedom (Gelfand et al., 2011). Social order in Nigeria and Thailand consists of an elite class with many working for the military, whereas the lower class includes people like farmers who largely see the world through the lens of threat to job security or medical care (Gelfand, 2018). Rules in these societies are largely prescribed and formalised, but do not apply equally to all social classes (Li & Gelfand, 2019). Employees in tight cultures prefer autonomous leaders who do not rely on others because they view them as strong leaders who make their own quick decisions (Aktas et al., 2016). In tight cultures, like in Thailand and Nigeria, organisations are more formal and disciplined, focused on order, obedience, rules, formality and hierarchy. Criticism is taboo because work is an integral part of national identity (Li & Gelfand, 2019).
On the contrary, in 'loose cultures' , such as the UK one, social equality, tolerance, creativity and receptiveness to change are key features, while rules are generally respected by society, although there is a general lack of formality (Gelfand et al., 2006). Loose cultures are more likely to allow employees to practice self-management have more freedom and adopt a risk-taking attitude, while idea sharing for innovation is a wide-spread practice (Chua et al., 2015).

The impact of organisational and employee characteristics on voice in the UK, Thailand and Nigeria
In SMEs, formal management practices are lacking, primarily due to resource constraints and most HRM activities are performed by the owner or general manager of the business (Harney & Dundon, 2007). Voice is usually informal, exhibiting variations in style that can be attributed to factors such as patterns of ownership, management skills and relationships with other firms (Bacon & Hoque, 2005).
The application of HRM in UK SMEs has witnessed fundamental changes over the past decades as the ER environment has changed in favour of promoting market flexibility, although institutional mechanisms have allowed firms to adopt a wide range of HRM practices (Gilman & Raby, 2013). HRM in Thai and Nigerian SMEs is affected by a blend of indigenous managerial behaviour, a colonial legacy in Africa, and a desire to mimic western management approaches (Dibben et al., 2017). In Thailand, there are two types of private sector firms: family enterprises and 'Thai-owned' corporations (Lawler & Atmiyanandana, 2003). The former are SMEs, relying on the conventional management practices of Chinese-style family enterprises (the majority are owned by Chinese nationals who have relocated to Thailand), with informal HRM based on personal relationships playing an important role in hiring, salaries and promotions. The latter are larger family enterprises or privatised state enterprises. In Nigeria, the key influences on HRM practice have been technological advances, the change from military to civilian government, competition from multinational companies, government legislation and economic conditions (Azolukwam & Perkins, 2009). HRM practices in SMEs lack strategic direction and follow a hybrid approach that moves between local cultural peculiarities and normative pressures from western-type HRM practices.
Factors such as the organisation's production technology, managerial quality and organisational culture influence voice at work (Kaufman, 2015). Research into non-union voice in SMEs in the UK has shown the influence of the organisational context, history of the firm, management attitudes towards voice and organisational culture (Dundon et al., 2005). Literature specific to voice in SMEs in Thailand and Nigeria is scarce, although the work of Emelifeonwu and Valk (2019) on voice in Nigerian multinationals provides an indication of how culture, characterised by respect for those in higher status positions, and employment insecurity, may also be factors limiting voice in Nigerian SMEs.
At the employee level, factors influencing decision-making and voice choice include workforce psycho-social attitudes towards work and management (Timming & Johnstone, 2015), knowledge, skills, abilities and demographics (Kaufman, 2015), although there is no specific literature on these issues in the three contexts under investigation.

Research methodology
We adopted a subjectivist ontological position and an interpretivist epistemological position because we wanted to capture employees' perceptions of workplace voice (Farndale et al., 2011). Such an approach enables us to understand human experiences from the perspective of organisational members (Bell et al., 2019). We used an exploratory, inductive, qualitative research design to collect data through in-depth semi-structured interviews (Wengraf, 2001) to obtain context-sensitive data (Cooke, 2018) that delve into people's experiences (Silverman, 2020) and fit the under-developed state of employee voice theory in smaller organisational settings (Gilman et al., 2015).
We interviewed 30 individuals working in different organisations, 10 in each of the UK, Thailand and Nigeria. Small-scale interview-based research (typically less than 20 interviews; Crouch & McKenzie, 2006) is common in exploratory studies (e.g. Bardoel, 2016;Perera et al., 2018) that seek to indicate rather than to conclude and is acceptable if research is intentionally conceptually generative (Crouch & McKenzie, 2006). In line with other small-scale qualitative studies (e.g. Timming, 2011) we recognise that only limited conclusions can be drawn from a sample of 30 interviews, however our aim is to explore voice in SMEs from the perspective of employees and invite further investigation in future studies. What is important is to have a rigorous procedure to data collection and analysis, which we explain next.

Sample selection
The sample of companies was selected purposefully (Patton, 2015); companies had to be of an SME size, although we did not place restrictions on other company characteristics, such as industry. The sample was obtained from the researchers' professional networks with the aim of capturing a mix of SMEs operating in the capital of each country (London, Bangkok, Abuja). For consistency purposes, we selected SMEs based in the largest urban areas in each country (capital cities), rather than SMEs operating in rural areas. In choosing the geographical location of our sample we were very aware that the three contexts were highly diverse; it is for this reason that we did not wish to exacerbate these differences by choosing SMEs based in a mix of urban and rural areas across the three contexts. Participant organisations employed between 10 and 209 employees (11 of them had more than 50 employees). None recognised a trade union. We sought to interview a broad range of participants in each context attempting to capture, where possible, some variety in position, age, gender and educational background (see Table 1).

Data collection and interview protocol
Interviews lasted between sixty and ninety minutes and were tape-recorded, with permission. The interviews in the UK and Nigeria were conducted in English and then transcribed, while the interviews in Thailand were conducted in Thai, translated to English and then transcribed. Participants were assured of confidentiality and anonymity. The interview questions were organised in six sections: 1. participants' demographic data (age; gender; position in organisation; tenure);  2. organisational information (industry/sector, organisational size, organisational structure, presence/role of HRM function in organisation); 3. voice mechanisms and practices (formal and informal voice mechanisms; trade union recognition; informal communication channels; employee involvement and participation practices); 4. factors promoting or inhibiting voice at work (perceptions and examples of feeling comfortable/uncomfortable discussing ideas or raising issues with line manager/boss; examples of cases where issues could not be openly discussed or complaints raised, frequency of occurrence and reasons for this; perception of organisational voice culture and processes; perceptions of freedom of expression at work; attitudes towards line manager/boss); 5. effects of the business environment on voice (how business conditions affect work security, management style, leadership style, people management policies/practices and voice); and 6. their reflections on their organisation and what they felt their organisation could do to improve its approach to employee voice.

Data analysis
All interviews were transcribed, and the resultant narrative data were analysed using thematic analysis, a method independent of research theory and epistemology (Braun & Clarke, 2006). This method is used to identify, analyse, interpret and report patterns (themes) within data (Roulston, 2001). We engaged in an inductive process of developing and refining a coding scheme by having each author independently engaged in the coding process (Braun & Clarke, 2006). The coding of each transcript commenced according to the original research questions which enabled us to provide an overview of common themes in relation to these questions (Elliot, 2018). Aside from this, we engaged in 'open coding' (Ezzy, 2002) where unexpected themes emerged from the data. After all transcripts had been coded, relevant texts were located within their respective thematic homes (Miles & Huberman, 2002). The main themes evolved from the initial analysis. A second wave involved identifying key sub-themes and patterns (Ritchie & Spencer, 2002); these were identified and highlighted, forming our new headings, supported by carefully chosen illustrative participant quotations. In short, we followed a process of analysis involving coding, data reduction, display, re-coding, and re-display (Miles & Huberman, 2002). The final common themes were agreed through discussion and these themes are presented in Table 2.

Validity and reliability
For qualitative research to be rigorously valid, the researcher's process should be transparent (Etherington, 2004) in the way that we have described our process in the sections above. Further, our design consideration demonstrates validity through a self-conscious research design, the justification of our sampling decision, the articulation of our data collection decisions, and the presentation and analysis of data that support our interpretations (Whittemore et al., 2001). Qualitative research must also be reliable (Richie & Lewis, 2008). While it may be difficult to achieve reliability from a measurement perspective in qualitative studies, interpretive work is empirically robust when supporting evidence is available, in the form of interview transcripts, coding sheets, thematic categorisation, participant profiles and reflective notes. In retaining interview notes, transcripts, and evidence of systematic coding and analysis, this research can, therefore, be also seen as rigorous.

Methodological limitations
None of the participants were drawn randomly, therefore, the sample is not representative, and generalisation of results is not possible (Timming,   . However, our objective was not to generalise findings, but rather to explore a new terrain and offer avenues for further research. In such a case, there is no methodological need for 'objectively' selecting the unit of investigation, but rather assurance that the site and unit of investigation are suitable for the issue under investigation (Diefenbach, 2009).

The influence of the external institutional, socio-economic and cultural context on employee voice in SMEs
Testimonies from participants reveal that the external environment can have a significant impact in determining the nature of employee voice behaviours. The economic context was of importance among Nigerian participants. The lack of a welfare state, the predominance of precarious work, unpredictable economic cycles, and high unemployment weighed on all participants in Nigerian SMEs. Concerns about such economic uncertainties contributed to a general wariness in expressing employee voice: Here, if you lose your job, there is no safety net, you are on your own… I'm not going to jeopardise this just because I might want to speak up to my boss.
For most Nigerian participants, the risk of severe economic impoverishment of family and self was a strong disincentive to engage in voice behaviours which posed a risk to job security. While some participants in Thailand expressed similar sentiments, there was an acknowledgment that the country had experienced improvements in social security provision, having transformed itself into a partial welfare state in recent years. As such, the existential threat to personal and family well-being due to a sudden loss of job, featured less among Thai participants.
However, in both countries, participants agreed that employment legislation did not encourage them to engage in employee voice. While both countries have constitutional provisions setting out minimum labour standards, protecting employees from unfair dismissal, and providing recourse to courts for employee grievances, participants were unaware of any legislation which stipulated a right to employee voice: Such laws might exist, but I have never heard of them… for the individual employee who is not in a union, I don't think we have many rights to voice (Participant 14,Thailand) By contrast, participants in the UK exhibited both less anxiety about the possibility of economic distress, and a greater awareness of, and trust in employment legislation: I may not be able to remember the exact name of the laws, but I do know that employees do have the right to voice their opinions and ideas to managers. And that you don't have to be in a trade union. (Participant 6,UK) The wider societal and cultural context was key for employee voice decisions. In Nigeria and Thailand, participant testimonies emphasised the social and cultural dominance of concepts such as collective duty/ obligation, group conformity and group loyalty above self-interest.
In Nigeria, participants spoke of the need 'not to rock the boat' and to avoid individual voice behaviours which breached collective cultural codes: [The] environment that the manager creates at my workplace makes me see the office as my extended family, and in Nigeria you don't bring trouble to the family, or you are seen as a bad person. (Participant 5,Nigeria) Thai participants were acutely aware of the need to constantly maintain in-group harmony, and the positive effects of exhibiting metta-karuna (love and kindness): In Thai society, we choose to be quiet because we don't want to create conflict in the group. (Participant 19,Thailand) Within these highly collective contexts, individual and independent voice was perceived as counter-productive to group harmony and cohesiveness. By contrast, participants in the UK appeared less constrained by such collectivist cultural concerns: If a person… always stay quiet, then we think they have no opinions, and nothing to contribute. Yes, people should weigh up what they say, and when to say it, but they should say something. (Participant 9, UK) Here we can see, the influence of the individualistic Anglo-Saxon cultural habitus within which British participant's real-world encounters take place. In such a context, the social and cultural emphasis on individual potentiality feeds through in a way that encourages voice behaviours.

Governance structure: The influence of SME leaders and managers on employee voice
For our participants, small structures and powerful owner-managers are central in shaping voice. SME size and organisational configuration accentuate the role of these CEOs: described variously as 'the emperor' , 'the all-powerful Oz' , and 'the Sun around which we all revolve' . For most participants this 'Sun-God' , is the central actor in their employment relationship; an actor who has a crucial role in shaping the culture.
Testimonies from Nigerian and Thai participants contrast with UK participants' testimonies in this regard. In both Thailand and Nigeria, employees accentuated the supremacy of the boss. While UK participants acknowledged the power of the boss, there was an awareness that employment legislation, and public opinion on the importance of workers' rights, were ingrained features of the national landscape, thus helping to frame a national consciousness around the need to limit excessive 'boss-power' and avoid worker exploitation. Nigerian participants, particularly, felt that there were few constraints on 'boss-power' . When the issue of employment legislation was raised it was often met with mild ridicule and considered to be largely rhetorical: … the boss of the firm is the boss; he gets to use his power; and is not afraid to show it. Even when laws say that workers should get paid on time, or should not be over-worked, or should have their problems addressed, the boss can override these. His power comes from the way we are brought up to respect power holders. (Participant 22,Nigeria) As such, in the Nigerian context, hierarchy is an important factor determining the extent to which (or even whether) employees speak up.

The influence of SME organisational configuration, HRM, and voice systems on employee voice
Most SMEs had no dedicated HRM specialist, though some had a rudimentary HRM presence: "the woman who takes care of payroll, advertising jobs and arranging the interviews" (Participant 2, UK). Two organisations had a small HRM department, which provided a greater array of people management support (Participant 3, UK; Participant 13, Thailand). The testimonies of our participants generally indicate an absence of formal or structured voice mechanisms, with occasional exceptions in UK SMEs: A lot of communication happens via emails and meetings. Face to face communications too … various HRM policies facilitate voice … performance appraisals … suggestion schemes and employee attitude surveys. (Participant 9, UK).
Where voice systems do exist, even in rudimentary form, participants value them. There is more employee engagement where there are dedicated HRM specialists, voice mechanisms, autonomy and influence over task-related decisions. However, even in the few cases with formal systems, and in all others, the depth and scope of voice was largely restricted to operational, task-related issues. Participants' report limited input into strategic decisions on substantive matters: While I am consulted in decisions directly related to my work, for example, how should we organise our sales plan for the next term, that's about the extent of it. (Participant 2, UK) Participants used terms such as 'communication channels' , 'receiving information' , and 'sharing information' but never used the words 'consultation' , or 'negotiation' . Even formal voice mechanisms, where they exist, are heavily imbued with a unitarist ideology: The only time we have opportunity for voice, is when management calls meet-ings… Officially, we are encouraged to speak about problems we might have to our Head of Department at these meetings, but we know that some Heads do not want us to. (Participant 17,Thailand) Despite findings that formal voice mechanisms were in place in only seven out of thirty SMEs in this study, voice was not totally absent in the remaining twenty-three SMEs: I work in a relatively small firm, and OK there are no formal mechanisms, but we do talk to one another… If the senior partners need us to know something, they either just email us or call us in for a meeting. (Participant 6,UK) Participants saw the absence of formal voice structures and mechanisms as effectively 'closing off ' an avenue for the articulation of ideas and concerns. This was particularly felt by Nigerian and Thai participants, where the almost complete lack of formal opportunities for voice was met with a mixture of resigned dismay, and occasional anger. One Nigerian participant exclaimed: Forget about collective bargaining -here we are reduced to collective begging! (Participant 29,Nigeria) Similarly, a UK-based participant equates the lack of voice structures in her organisation to Oliver Twist daring to ask, "Please, Sir, can I have some more?": If there was a forum or committee or something like that I could feed into, it would be better. I'd feel less like a complainer, whining the whole time. (Participant 4,UK) Participants across the three countries shared the view that the lack of institutionally provided voice infrastructures forces employees into self-motivated acts of 'speaking-up' on issues that concern their welfare. This can be seen by management as a breach of the paternalist, family, unitarist culture. Such employees can be made to feel like dissidents, and risk managerial stigmatisation as 'trouble-makers': We are discontented with poorly upgraded technological resources, poor facilities, lack of renovations, unavailability of electricity, late payment of salaries, nepotism, favouritism… However, when I raised some gentle concerns and made some mild suggestions, I was advised by the boss that I was becoming too forward with my opinions, becoming an irritant, and creating problems… I will need the boss' reference if I move on, so I have stopped speaking up. (Participant 23,Nigeria) The organisational configuration of SMEs has a direct bearing on the extent of formal/direct voice mechanisms. A lack of structure has an impact on determining whether people speak up, and on what issues. It sends implicit, but powerful, messages about the value the SME places on voice which, as one participant commented, "was only for the big [organisations]" (Participant 10, UK). SMEs that offer minimalist voice mechanisms are often paid back with minimal voice, with the potential loss of innovative employee inputs. Among many participants in such situations, the belief is that: The organisation doesn't really want to hear your opinion or ideas, otherwise they' d set something up to channel it; so, if they can't be bothered to seek out my opinion, then why should I bother giving [it]? (Participant 27, Nigeria)

The interaction of internal contingencies in the SME employment relationship and their influence on employee voice
A key contingency is the boss's 'style' , or 'approachability' in listening to employees' opinions. Participants in Thailand felt an absolute necessity for 'friendly' and 'approachable' managers and bosses: Management internal contingencies may be of the utmost importance in determining whether employees 'speak up' , but our study also highlights the role played by internal employee contingencies such as their education, skills and confidence or self-efficacy. In general, those with university degrees, work experience, capabilities, and significant human and social capital, found it easier to express voice: Most of us here are university educated… and most definitely not afraid to speak up… [People] are well able to speak up for themselves and the managers will have to listen, or else people will walk. (Participant 9, UK) Conversely, those who held fewer qualifications and who possessed less marketable skills often displayed an awareness that their labour market 'replaceability' puts them at a disadvantage, especially on contentious issues in the workplace: I would be careful about which issues I discuss with the managers. I am a receptionist and they can easily find another receptionist if we don't get along or if I start complaining all the time. (Participant 8,UK) Nigerians and Thais displayed concerns about their (often) precarious position in a socio-economic context where lack of employment stability and security presents potentially existential threats to welfare. Voice did not belong to them -it belonged to others: My voice doesn't matter -because we are only maids… cleaning, I don't have any voice. I cannot express my feelings or opinion because they will not listen to me and they think we are so small. (Participant 14, Thailand)

Voice agency, perceived levels of voice and voice behaviour
We consider voice agency as the degree to which participants were able to actively shape personal voice strategies, and the extent to which they could freely have a voice. This was heavily circumscribed by the external environment in which their experiences took place. Furthermore, governance structure, organisational configuration, HRM, voice systems and the nature of the employment relationship significantly shaped the nature, depth and scope of employee agency around voice.
Tensions, power, resistance and compliance were features of the organisational habitus of most participants. In most cases, we found employee voice agency to be highly contingent on the attitudes of SME managers. Generally, this resulted in strategies of employee silence. In rare situations where the organisational culture actively encouraged bottom-up voice, employees responded with open forms of speaking up. However, in most cases, employees adopted a partial voice strategy, where they chose to speak up on some issues, while remaining silent on others. Voice cultures were influenced by national social and cultural norms. Participants in the UK were generally found to possess stronger agency compared with those in Nigeria and Thailand. This is likely due to the higher individualist culture of the UK, the presence and acceptance of HRM in many of the British SMEs, and organisational cultures that encourage promotive voice.
The outcomes of these combinations of factors meant that most participants, particularly in Thailand and Nigeria, display a deeply felt belief that employee silence was often the best approach: What's the point when such outspokenness often leads to conflict? (Participant 23,Nigeria) The important thing is following the rules and getting your pay. (Participant 21,Nigeria) Participants in Thailand demonstrated an acute sensitivity to the idea of causing management to 'lose face' . Employees ensure that in expressing voice, or highlighting managerial mistakes, they avoid public embarrassment at all costs: If it was really necessary to speak, I would find a way to not make him feel embarrassed, or that it was his fault. Make it seem like it is my fault. (Participant 11,Thailand) However, it would be wrong to assume that in all Thai or Nigerian SMEs in this study employees worked under regimes of repressive silence. Rather, we see employees respond to conformist organisation cultures in innovative ways: amending strategies of silence by adopting discrete and non-confrontational, 'backchannel' methods: You can communicate up, but you have to be very subtle and not confront; you must think carefully of how you phrase your petition and concerns; who you raise it with, usually not collectively as this would be seen as insubordinate and could cause embarrassment to the chief. Whatever the case, I will be careful in putting my views. I will not disagree openly. It is a norm to speak but not to confront. (Participant 30, Nigeria).
So, in the general absence of formal voice mechanisms and structures in our Thai and Nigerian SMEs, we see participants employing informal, tactful and carefully choreographed voice strategies.

Discussion
Externally, in the presence of challenging conditions, employees may choose silence . Across the samples from the three countries, the issue of employment security was an important determinant arising from the institutional, socio-economic, context in each country. Those on zero-hours, temporary or part-time contracts displayed resistance to 'speaking up' , preferring to wait until their contractual status was more secure, keeping silent on issues of workplace conditions for fear of undermining their chances of achieving contractual stability in the future. However, some in this situation did 'speak up' in response to management invitations to develop innovative ways that might improve performance, making strategic choices to speak up in ways that, they hoped, would be seen as 'contributing value' , and endear them to management. Thailand and, to a lesser extent, Nigeria differed from the UK in their widely held fear that speaking up might engender organisational conflict, and possible negative management responses. National social and cultural norms affect voice behaviour, with UK participants exhibiting stronger voice agency compared to those in Nigeria and Thailand. The presence of a survival dimension in determining employee 'voicing-up' decisions was particularly evident in participant testimonies from Nigeria where there was no welfare state to fall back on in extremis, compared to those in the UK.
The SMEs we studied in the UK, Thailand and Nigeria share much in common in terms of organisational configuration and governance structures. Our analysis found that despite the diversity of sectors and geographical locations, all participants worked in organisations that had similar configurations which, consistent with their status as SMEs, were generally typified as having simple (rather than sophisticated) organisation designs and flattened (rather than hierarchical) structures. They were personified by a lack of divisional complexity and by command/ control (rather than participative) management styles. In SMEs, processes can be less formal given the nature of communication flows and the more flexible social setting (Wilkinson et al., 2007). HRM specialism in the organisation can be located within a distinctive HRM department or can more often be the responsibility of managers and owners (Bacon & Hoque, 2005). However, even where there were HRM specialists, their role was operational rather than strategic.
The organisational configuration, and its impact on determining employee voice in these organisations, did matter to participants and they did desire more voice and involvement opportunities. This contrasts with the findings of Gilman et al. (2015) where most employees in their study of UK SMEs did not express dissatisfaction or a need to be more involved in the organisation. Our study found that the organisational configuration of SMEs meant people management was operational in nature and devolved across different management levels. This opens the possibility of informal voice systems, but other internal contingencies tended to result, overall, in low levels of voice.
Governance structures, in nearly all cases, were characterised by a concentration of power in the hands of one person, usually the owner of the business. A handful of the larger SMEs also had a small number of departmental managers but, even there, their strategic role and influence was limited, constrained by the need to ultimately conform to the views and decisions of the owner. Across all three countries, an overriding theme was a unitarist perspective (Fox, 1966), based in the relationship between long-standing employees and managers. Such trust-based relationships, particularly in family firms, were strong and potent (Erdem & Atsan, 2015). Our participants approached voice with a mixture of caution, tact and prudence, to avoid marginalisation, or worse outcomes. A repeated theme throughout the interviews, in all countries, was that owners may seek voice aimed at delivering performance improvements, but can be indifferent, or at times openly hostile, to opinions that highlight deficiencies in the treatment or welfare of workers. This acts as a powerful deterrent to 'speaking-up' on such topics and helps construct a culture of norms and beliefs around what it means to be an employee with a voice, circumscribed by perceived management preferences, priorities and 'no-go' areas (Dundon & Gollan, 2007).
Voice systems do exist in SMEs but generally only with the purpose of information sharing. Voice systems tend to be unstructured, informal and sporadic. There is a pragmatic acceptance that voice structures and mechanisms will 'naturally' be underdeveloped compared with larger organisations (Sameer & Ozbilgin, 2014). The limited existence of formal voice mechanisms emphasised the importance of managers adopting an 'open door' approach, giving rise to 'informal voice' (Dundon et al., 2005). However, the success of informal voice depends on the manager's approachability and receptivity (Detert & Trevino, 2010).
We cannot discuss voice by examining employee and organisational contingencies separately -the process is one of ever dynamic, and mutually reinforcing, contingencies. These help to create the conditions that either hinder or encourage voice, and determine decisions to speak up, or stay silent. This leads to silence that is quiescent (self-protective based on fear) rather than silence that is acquiescent (disengaged) (Pinder & Harlos, 2001). Participants, especially in the Thai and Nigerian samples, often decided not to speak up "either because of fear of the consequences or because they have internalised the rules of the 'game'" (Wilkinson et al., 2018: 717).
Employee demographics emerged as relevant internal employee contingencies. Employee diversity intersected with organisation culture in ways that can, at times, result in forms of greater disadvantage, marginalisation and exclusion (Tatli & Ozbilgin, 2012). As Wilkinson et al. (2018: 717) noted, workers are diverse and "their opportunity or tendency to voice may be shaped by their gender, race, sexuality and personal perceptions in addition to institutional factors". Other employee contingencies we uncovered included employee length of service, employee capabilities, skills and qualifications, employee status in the organisation and the nature of the employment contract. Such employee contingencies shaped employee agency in exercising or withholding voice. The employee voice determinants in our analysis are presented in Figure 1.

Conclusion
We add to previously published frameworks on voice determinants (Kaufman, 2015) and on voice in SMEs (Gilman et al., 2015). We add to Kaufman's (2015) work by looking at the interaction of key macro-meso-micro level determinants and the impact they have on the employee experience with voice in non-unionised SMEs across three contexts. Like Kaufman, we found the key determinants of employee voice in SMEs to be the external environment, governance structure, organisational configuration and internal contingencies in the employment relationship. In SMEs, these elements interact in a unique way to influence employee voice agency, perceived levels of voice and ultimately voice behaviour. The external environment poses challenges to SMEs in the three contexts we studied. Weaker institutional settings, challenging socio-economic conditions and cultural variables limit voice. The SME governance structure and organisational configuration have a direct bearing on the extent of formal, direct voice. The owner/manager plays a pivotal role in the governance structure in shaping employees' agency in exercising or withholding voice. This sheds light on the effect of 'human and social capital' on voice (Gilman et al., 2015), particularly the pivotal role of owners/managers in these organisations. The presence or absence of a voice system also impacts on perceptions of the voice culture within the SME. This impacts employees' voice agency and ultimately voice behaviour.
We also reveal a dynamic relationship between internal contingencies in the employment relationship which further affect employee agency and voice behaviour. Voice is contingent on organisational signals and cultural messaging received from management, while employee length of service, capabilities, skills and qualifications, status in the organisation and the nature of the employment contract were also found to affect employee voice agency and behaviour.
The three contexts shared much in common in terms of voice determinants, with similarities in the way in which quality of leadership and management, extent of informality in practices, and the degree of employee voice agency, determine voice in these organisations. But we also note differences. Relative to the UK, the Nigerian and Thai contexts include: the prominence of hierarchy and the need for respect; a deep aversion to challenging authority; heightened employee sensitisation to potential negative repercussions of speaking up; stronger collective 'team-player' sensibilities; significant pressures towards conformity; and, especially in Thailand, a problematic conceptualisation around voice which results in forms of self-censorship and employee silence.
On a practical level, our proposed initial framework can guide SMEs in understanding the importance of the owner/manager in developing a voice culture within the organisation through engaging in participative management practices, as well as in capturing the importance of HRM and voice systems in enabling employees to have avenues for voice. Our initial framework also helps SME owners/managers consider how the overall external environment and internal contingencies of the employment relationship may limit employees' voice agency and behaviour. The consideration of such issues can enable small enterprises to adopt a more inclusive approach to voice in the workplace.
Our study is limited by the small number of cases in each context and by the number of countries we examined. In addition, our study was undertaken from an employee perspective and for this reason, we missed other possible determinants from an organisational/management perspective: e.g. product market, labour market and strategic orientation (Gilman et al., 2015). Further research on voice in SMEs could examine varying organisational sizes, a wider range of cultural settings, and could be of a diachronic nature offering a dynamic and process perspective on voice.

Disclosure statement
No potential conflict of interest was reported by the author(s).