The general objective of the study is to examine the effect of entrepreneurial leadership on employee creativity and innovative work behaviour in African SMEs through the dual mediating roles of entrepreneurial intention and entrepreneurial self-efficacy. Specifically, the study seeks to:

1) Examine the direct effect of entrepreneurial leadership on employee creativity and innovative work behaviour.

2) Assess the effect of entrepreneurial leadership on entrepreneurial intention and entrepreneurial self-efficacy.

3) Determine the effect of entrepreneurial intention on employee creativity.

4) Determine the effect of entrepreneurial self-efficacy on innovative work behaviour.

5) Examine the mediating role of entrepreneurial intention in the relationship between entrepreneurial leadership and employee creativity.

6) Examine the mediating role of entrepreneurial self-efficacy in the relationship between entrepreneurial leadership and innovative work behaviour.

The study makes three contributions. Theoretically, it extends social cognitive theory and the entrepreneurial leadership perspective by demonstrating how leadership operates through two agentic motivational states to shape distinct innovation outcomes, thereby enriching the nomological network surrounding entrepreneurial leadership ([9]; [37]). In terms of contribution to existing knowledge, it provides rare empirical evidence from African SMEs and separates the ideation pathway from the implementation pathway, clarifying that intention is more proximal to creativity while self-efficacy is more proximal to innovative work behaviour ([27]; [2]). With respect to policy implications, the findings inform enterprise development agencies, business support institutions, and policy makers seeking to embed innovation capacity within the SME sector as a vehicle for inclusive growth and employment generation.

The remainder of the paper is organised as follows. Section two presents the theoretical framework, reviews relevant literature, and develops the hypotheses and conceptual model. Section three describes the methodology. Section four reports the analysis and results. Section five discusses the findings and the theoretical and managerial implications. Section six presents limitations, future research directions, and conclusions.

The study is anchored on social cognitive theory propounded by [9], complemented by the entrepreneurial leadership perspective advanced by [37] and the componential theory of creativity developed by [5]. Social cognitive theory proposes that human behaviour is the product of a continuous reciprocal interaction among personal factors, environmental influences, and behaviour itself, a process Bandura termed triadic reciprocal determinism ([9]). Within this logic, the leadership context functions as an environmental stimulus, entrepreneurial intention and self-efficacy operate as personal cognitive states, and creativity and innovative work behaviour represent the behavioural outcomes. The theory is highly relevant to the present study because it explains why and how exposure to entrepreneurial leadership cultivates the motivational beliefs that drive employees to generate and implement novel ideas ([32]; [38]). Self-efficacy, a central mechanism in the theory, captures the confidence that enables employees to attempt and persist with creative tasks despite uncertainty ([10]; [41]). The entrepreneurial leadership perspective justifies the focus on a leadership style that explicitly models opportunity seeking behaviour, while the componential theory of creativity explains the conditions under which intrinsic motivation and domain skills converge to produce creative output ([5]; [27]). Together these perspectives provide a coherent and theoretically defensible foundation for the proposed model.

Entrepreneurial leaders foster a climate of psychological safety, autonomy, and opportunity orientation that encourages employees to explore unconventional solutions ([8]; [28]). Empirical evidence consistently indicates that entrepreneurial leadership stimulates creative behaviour by signalling that experimentation is valued and failure is tolerated ([11]; [27]). However, a minority of studies report weaker or context dependent effects, suggesting that the relationship may be sensitive to organisational resources and national culture ([19]; [22]). On balance, the weight of evidence supports a positive association in entrepreneurial settings such as SMEs. Accordingly, the following hypothesis is proposed:

H1. Entrepreneurial leadership has a positive and significant effect on employee creativity.

Beyond ideation, entrepreneurial leadership encourages the championing and implementation of ideas that characterise innovative work behaviour ([2]; [8]). Leaders who model proactiveness and risk taking provide both the inspiration and the resources that employees require to convert ideas into practice ([31]; [30]). While most studies confirm a robust positive relationship, some scholars caution that implementation depends additionally on structural enablers that leadership alone cannot supply ([19]; [35]). The preponderance of evidence nonetheless favours a direct positive effect. Thus:

H2. Entrepreneurial leadership has a positive and significant effect on innovative work behaviour.

Entrepreneurial leaders serve as salient role models whose behaviour shapes the entrepreneurial intentions of their followers through vicarious learning and verbal persuasion ([10]; [33]). Exposure to opportunity focused leadership strengthens employees’ commitment to pursue value creating activity, an effect documented across diverse settings ([41]; [38]). Similarly, entrepreneurial leadership builds entrepreneurial self-efficacy by providing mastery experiences, encouragement, and supportive feedback that raise employees’ confidence in their own capabilities ([32]; [8]). Although a few studies find that the efficacy effect is partly contingent on prior experience, the dominant view supports positive direct effects ([11]; [28]). Hence:

H3. Entrepreneurial leadership has a positive and significant effect on entrepreneurial intention.

H4. Entrepreneurial leadership has a positive and significant effect on entrepreneurial self-efficacy.

Entrepreneurial intention reflects a goal directed motivational state that channels cognitive and emotional energy towards opportunity seeking, a disposition closely aligned with the generation of novel ideas ([26]; [33]). Employees with strong entrepreneurial intention actively scan their environment for problems to solve and possibilities to exploit, behaviours that underpin creativity ([38]; [27]). While some scholars argue that intention is more closely tied to venture creation than to within job creativity, recent evidence affirms its relevance to creative output in established firms ([41]; [30]). Therefore:

H5. Entrepreneurial intention has a positive and significant effect on employee creativity.

Self-efficacy is among the most consistent predictors of innovative work behaviour because confident employees are more willing to champion ideas and absorb the risks associated with implementation ([10]; [2]). Employees who believe in their entrepreneurial capabilities persist through obstacles and mobilise support for novel initiatives ([32]; [30]). Although a small number of studies report attenuated effects under highly bureaucratic conditions, the consensus supports a strong positive relationship ([11]; [38]). Accordingly:

H6. Entrepreneurial self-efficacy has a positive and significant effect on innovative work behaviour.

Social cognitive theory implies that the influence of environmental stimuli on behaviour is transmitted through personal cognitive states ([9]; [32]). On this basis, entrepreneurial leadership is expected to shape creativity indirectly by first elevating entrepreneurial intention, and to shape innovative work behaviour indirectly by first strengthening self-efficacy ([41]; [38]). Mediation evidence in adjacent domains supports such indirect transmission, although the strength of mediation varies with context and measurement ([33]; [27]). Building on the preceding direct hypotheses, the study proposes two mediating effects:

H7. Entrepreneurial intention mediates the relationship between entrepreneurial leadership and employee creativity.

H8. Entrepreneurial self-efficacy mediates the relationship between entrepreneurial leadership and innovative work behaviour.

Rationale for the Matched Mediation Pathways

The model deliberately pairs entrepreneurial intention with employee creativity and entrepreneurial self-efficacy with innovative work behaviour, rather than specifying crossed paths in which intention also predicts innovative work behaviour and self-efficacy also predicts creativity. This specification is theory driven and follows from the distinct functions that the two cognitive states perform within social cognitive theory and the intention to action sequence ([9]; [26]). Entrepreneurial intention is a deliberative, goal directed motivational state concerned with whether an employee is willing to pursue opportunities and value creating ideas. Because creativity is fundamentally an ideation construct, that is, the generation of novel and useful ideas, it is most proximal to the motivational, opportunity scanning orientation that intention represents ([5]; [33]; [27]). Entrepreneurial self-efficacy, by contrast, is an agentic capability belief concerned with whether an employee is confident of executing entrepreneurial tasks under risk and uncertainty. Because innovative work behaviour is fundamentally an implementation construct that requires employees to champion, mobilise support for, and realise ideas in the face of resistance and possible failure, it is most proximal to the volitional, persistence enabling function that self-efficacy represents ([10]; [32]; [2]). In other words, the deliberative mechanism is matched to the ideation outcome and the agentic mechanism is matched to the enactment outcome, which mirrors the conceptual progression from idea generation to idea implementation. The two crossed paths were therefore not hypothesised because they are less strongly grounded in this theoretical logic; the study acknowledges them as plausible secondary relationships and returns to them as an avenue for future research in Section 6.1.

Drawing on entrepreneurial leadership theory and the preceding empirical arguments, Figure 1 presents the conceptual framework of the study. The framework positions entrepreneurial leadership as the exogenous construct that directly influences employee creativity and innovative work behaviour among SMEs in Africa. It further proposes entrepreneurial intention and entrepreneurial self-efficacy as dual mediating mechanisms through which entrepreneurial leadership translates into creativity and innovation related outcomes.

The framework specifies eight hypotheses. First, entrepreneurial leadership is expected to directly enhance employee creativity (H1) and innovative work behaviour (H2). Second, entrepreneurial leadership is proposed to strengthen entrepreneurial intention (H3) and entrepreneurial self-efficacy (H4). Third, entrepreneurial intention is expected to improve employee creativity (H5), while entrepreneurial self-efficacy is expected to enhance innovative work behaviour (H6). Finally, entrepreneurial intention is proposed to mediate the relationship between entrepreneurial leadership and employee creativity (H7), whereas entrepreneurial self-efficacy is proposed to mediate the relationship between entrepreneurial leadership and innovative work behaviour (H8). Because Figure 1 represents the hypothesised model rather than estimated output, the paths are labelled with their hypothesis numbers (H1 to H8) and do not display fitted coefficients; the estimated coefficients are reported later in Figure 2 and Figure 3 and in Table 7 and Table 8.

Figure 1. Conceptual framework of the study, showing the hypothesised direct and indirect paths (H1 to H8). Source: Authors’ construct (2026).

Figure 2. PLS algorithm output showing item loadings, standardised path coefficients, and R² values. The values shown match Table 3, Table 6, and Table 7. Source: Authors’ construct (2026).

Figure 3. Structural model with standardised path coefficients and t-values. The values shown match Table 7. Source: Authors’ construct (2026).

The study adopted a quantitative research approach because the objectives required the objective measurement of constructs and the statistical testing of hypothesised relationships among them ([40]; [15]). A quantitative approach is appropriate when a study seeks to verify theory through the examination of relationships among measurable variables in a large sample ([13]; [16]). A cross-sectional survey design was employed, whereby data were collected at a single point in time from a defined sample of respondents ([36]; [40]). This design was justified on the grounds of efficiency, cost effectiveness, and its suitability for capturing the perceptions of a dispersed population of SME employees within a feasible timeframe ([15]; [29]). The choice of design was further informed by the deductive logic of the study, which moves from established theory to the empirical testing of derived hypotheses ([13]).

The target population comprised full time employees of registered SMEs operating in the Greater Accra region of Ghana, a setting that hosts a dense concentration of the country’s formal enterprise activity ([1]; [7]). The sampling frame was constructed by consolidating two administrative sources, namely the regional register of micro, small, and medium enterprises maintained by the Ghana Enterprises Agency and the business operating permit databases of the metropolitan and municipal assemblies in the region. After removing duplicate entries, firms that no longer operated, and enterprises outside the SME employment size band, the consolidated frame yielded 2418 active registered SMEs. To make the stratified procedure transparent, the frame was divided into three sector strata that reflect the structure of formal SME activity in the region, namely manufacturing, trade and commerce, and services.

The minimum sample size was determined using the [24] formula, which recommends a minimum of 384 respondents for populations exceeding one hundred thousand, a threshold also consistent with the requirements of PLS-SEM for the present model complexity ([29]; [15]). A proportionate stratified random sampling procedure was then applied in two stages. In the first stage, firms were selected at random within each stratum using random numbers applied to the frame. In the second stage, eligible full time employees were selected at random within each sampled firm. In total, 150 firms were sampled and 450 questionnaires were distributed across the three strata. The allocation was as follows: manufacturing comprised 38 firms with 110 questionnaires distributed and 95 retained; trade and commerce comprised 52 firms with 150 questionnaires distributed and 131 retained; and services comprised 60 firms with 190 questionnaires distributed and 162 retained. Of the 450 questionnaires distributed, 401 were retrieved and 388 were retained after data screening, yielding an effective response rate of 86.2 per cent and a final sectoral distribution of 95 manufacturing, 131 trade and commerce, and 162 service respondents. Non response bias was assessed by comparing the responses of early and late respondents on the key study variables, and no statistically significant differences emerged, indicating that non response bias was not a material concern ([6]; [36]).

All constructs were measured with multiple item reflective scales adapted from previously validated instruments and anchored on a five-point Likert scale ranging from one, denoting strongly disagree, to five, denoting strongly agree ([15]). Entrepreneurial leadership was measured with six items adapted from [37]. Entrepreneurial intention was measured with five items adapted from [26]. Entrepreneurial self-efficacy was measured with five items adapted from [32]. Employee creativity was measured with five items adapted from [5] and later operationalisations, while innovative work behaviour was measured with six items adapted from [21] and [2]. Item wording was contextually adapted to the Ghanaian SME setting through expert review and a pilot study, and all 27 items were retained after the measurement assessment because each met the recommended reliability and validity thresholds. The full source details, the number of items retained per construct, and a sample item for each scale are provided in Appendix. Gender, age, organisational tenure, and education were included as control variables given their documented influence on creative and innovative outcomes ([30]; [27]); their treatment in the structural model is reported in Section 4.5.

Data were collected through self-administered structured questionnaires distributed both in person and through electronic forms over a period of ten weeks. Trained research assistants supported respondents to minimise misinterpretation and item non completion. Participation was entirely voluntary, informed consent was obtained from every respondent, and anonymity and confidentiality were guaranteed throughout the process ([36]; [40]). The study protocol observed established ethical principles for research involving human participants, and respondents were assured that the data would be used solely for academic purposes and were free to withdraw at any stage without penalty.

Because the data were obtained from a single source using a self report instrument, common method bias was a potential concern and was addressed through both procedural and statistical remedies ([36]; [23]). Procedurally, respondent anonymity was assured, item wording was simplified, and the predictor and criterion items were separated within the questionnaire. Statistically, Harman’s single factor test was conducted, and the first unrotated factor accounted for 31.4 per cent of the variance, which is below the 50 per cent threshold. In addition, the full collinearity variance inflation factors derived from the structural model were all below 3.3, confirming that common method bias did not threaten the validity of the findings ([23]; [15]).

Data Analysis

Partial least squares structural equation modelling (PLS-SEM) was preferred over covariance based structural equation modelling because the study is prediction oriented and aims to maximise the explained variance of the endogenous innovation outcomes within a relatively complex model that contains two mediating mechanisms, while remaining robust to potential non normality in the data ([15]; [39]).

Table 1 reports the demographic characteristics of the 388 respondents. The sample was reasonably balanced by gender, with a moderate concentration of younger employees and those with shorter tenure, which is consistent with the demographic profile of the African SME workforce ([7]).

Table 1. Demographic characteristics of respondents (N = 388).

Source: Field Survey (2026).

The results show that males represented 54.1% of the sample, while females accounted for 45.9%, suggesting a reasonably balanced gender distribution. Most respondents were between 30 and 39 years old (37.9%), followed by those aged 18 to 29 years (34.0%). This indicates that the sample was dominated by relatively young and economically active employees. In terms of tenure, the largest proportion had worked for 2 to 5 years (36.1%), while 29.9% had less than 2 years of work experience. This suggests that many respondents were still in the early or developing stages of their employment careers. Regarding education, most respondents held a master degree (37.9%), followed by bachelor degree holders (30.2%), indicating that the sample was fairly educated and capable of providing informed responses on entrepreneurial leadership, creativity, self-efficacy, intention, and innovative work behaviour.

Table 2 presents the means, standard deviations, and latent variable correlations for the study constructs.

Table 2. Means, standard deviations, and correlations among study constructs.

Note. N = 388. EL = entrepreneurial leadership; EI = entrepreneurial intention; ESE = entrepreneurial self-efficacy; EC = employee creativity; IWB = innovative work behaviour. All correlations are significant at p < 0.01.

The mean scores ranged from 3.71 to 3.92, indicating that respondents generally reported favourable perceptions of entrepreneurial leadership, entrepreneurial intention, entrepreneurial self-efficacy, employee creativity, and innovative work behaviour. Entrepreneurial leadership recorded the highest mean score (M = 3.92, SD = 0.71), followed by entrepreneurial self-efficacy (M = 3.85, SD = 0.66) and innovative work behaviour (M = 3.80, SD = 0.70). Employee creativity recorded the lowest mean score (M = 3.71, SD = 0.74), although it remained above the midpoint of the five-point scale. All constructs recorded mean scores above the scale midpoint, indicating generally favourable perceptions, while the correlations were positive, moderate in magnitude, and statistically significant, providing preliminary support for the hypothesised relationships and revealing no evidence of excessive collinearity ([15]; [39]).

The correlation results show positive and statistically significant associations among all constructs at p < 0.01. The strongest relationship was observed between employee creativity and innovative work behaviour (r = 0.58), followed by entrepreneurial self-efficacy and innovative work behaviour (r = 0.56), entrepreneurial leadership and entrepreneurial self-efficacy (r = 0.55), and entrepreneurial intention and employee creativity (r = 0.54). These results provide preliminary support for the proposed theoretical relationships. Importantly, all correlation coefficients were below the conservative threshold of 0.85, suggesting that multicollinearity was not a serious concern and that the constructs were empirically distinct ([15]; [39]).

The measurement model was evaluated for indicator reliability, internal consistency reliability, and convergent validity before the structural relationships were examined ([15]; [39]). As reported in Table 3, all outer loadings exceeded the recommended threshold of 0.70, and the variance inflation factors were well below the conservative cut off of 5, and indeed below 3.3, indicating the absence of problematic collinearity among indicators. Composite reliability and Cronbach’s alpha values for every construct surpassed 0.70, demonstrating satisfactory internal consistency, while the average variance extracted for each construct exceeded 0.50, confirming convergent validity ([15]; [12]). Figure 2 displays the PLS algorithm output, in which the item loadings, the standardised path coefficients, and the R² values correspond exactly to the values reported in Table 3, Table 6 and Table 7.

Table 3. Measurement model: Outer loadings, collinearity, and reliability.

Note. α = Cronbach’s alpha; CR = composite reliability; AVE = average variance extracted; VIF = variance inflation factor. All loadings significant at p < 0.001.

Discriminant validity, which establishes that each construct is empirically distinct from the others, was assessed using both the [14] criterion and the Heterotrait-Monotrait ratio of correlations ([18]; [15]). As shown in Table 4, the square root of the average variance extracted for each construct, presented on the diagonal, exceeded its correlations with all other constructs, satisfying the Fornell and Larcker criterion. Table 5 reports the Heterotrait-Monotrait ratios, all of which fell below the conservative threshold of 0.85, providing robust confirmation of discriminant validity ([18]; [12]).

Table 4. Discriminant validity: Fornell-Larcker criterion.

Note. Bold diagonal values are the square roots of the average variance extracted. Off-diagonal values are latent variable correlations.

Table 5. Discriminant validity: Heterotrait-Monotrait ratio (HTMT).

Note. All HTMT values are below the 0.85 threshold, confirming discriminant validity.

After establishing the reliability and validity of the measurement model, the structural model was assessed to examine explanatory power, predictive relevance, and the hypothesised path relationships. The model produced an SRMR value of 0.056, which is below the recommended threshold of 0.08, indicating an acceptable model fit ([17]; [15]). As shown in Table 6, the R² values ranged from 0.27 to 0.44, suggesting that the model explained a meaningful proportion of variance in the endogenous constructs. Specifically, entrepreneurial leadership explained 27% of the variance in entrepreneurial intention and 30% of the variance in entrepreneurial self-efficacy. The model further explained 41% of the variance in employee creativity and 44% of the variance in innovative work behaviour. These values indicate moderate explanatory power. In addition, all Q² values were greater than zero, confirming that the model had adequate predictive relevance for all endogenous constructs ([42]; [39]).

Table 6. Model fit, coefficient of determination, and predictive relevance.

Note. SRMR = 0.056. Q² values greater than zero indicate predictive relevance.

The structural relationships were tested using the bootstrapping procedure with 5000 subsamples to generate standard errors, t-values, and bias corrected confidence intervals ([15]; [43]). Table 7 summarises the results for the six direct hypotheses, and Figure 3 depicts the structural model with the corresponding path coefficients and t-values.

Table 7. Structural model results: Direct effects (Bootstrapping, 5000 subsamples).

Note. β = standardized path coefficient; SE = standard error; BCI = bias corrected confidence interval; f² = effect size. EL = entrepreneurial leadership; EI = entrepreneurial intention; ESE = entrepreneurial self-efficacy; EC = employee creativity; IWB = innovative work behaviour.

The results in Table 7 show that all six direct paths were positive and statistically significant. Entrepreneurial leadership had a significant positive effect on employee creativity, β = 0.24, SE = 0.06, t = 4.00, p < 0.001, 95% BCI [0.13, 0.36], supporting H1. Similarly, entrepreneurial leadership significantly predicted innovative work behaviour, β = 0.21, SE = 0.06, t = 3.50, p < 0.001, 95% BCI [0.10, 0.33], supporting H2.

Entrepreneurial leadership also had strong positive effects on the two mediating constructs. It significantly predicted entrepreneurial intention, β = 0.52, SE = 0.05, t = 10.40, p < 0.001, 95% BCI [0.42, 0.61], supporting H3. It also significantly predicted entrepreneurial self-efficacy, β = 0.55, SE = 0.04, t = 13.75, p < 0.001, 95% BCI [0.46, 0.63], supporting H4. Furthermore, entrepreneurial intention significantly predicted employee creativity, β = 0.38, SE = 0.06, t = 6.33, p < 0.001, 95% BCI [0.27, 0.49], supporting H5. Entrepreneurial self-efficacy also significantly predicted innovative work behaviour, β = 0.41, SE = 0.05, t = 8.20, p < 0.001, 95% BCI [0.31, 0.51], supporting H6.

The effect size values further show that entrepreneurial leadership had a large effect on entrepreneurial intention, f² = 0.37, and entrepreneurial self-efficacy, f² = 0.43. The effects of entrepreneurial intention on employee creativity, f² = 0.18, and entrepreneurial self-efficacy on innovative work behaviour, f² = 0.21, were moderate. The direct effects of entrepreneurial leadership on employee creativity, f² = 0.07, and innovative work behaviour, f² = 0.06, were small but statistically meaningful. Figure 3 presents the structural model with standardised path coefficients and t-values. All six direct paths were positive and statistically significant, and every bias corrected confidence interval excluded zero, providing support for hypotheses H1 to H6 ([15]; [39]).

To rule out alternative explanations, the four control variables, namely gender, age, organisational tenure, and education, were entered into the structural model as additional predictors of employee creativity and innovative work behaviour. None of the control paths reached statistical significance: for employee creativity, gender (β = 0.05, p = 0.28), age (β = 0.06, p = 0.21), tenure (β = 0.03, p = 0.49), and education (β = 0.06, p = 0.19); and for innovative work behaviour, gender (β = 0.04, p = 0.33), age (β = 0.05, p = 0.27), tenure (β = 0.04, p = 0.38), and education (β = 0.05, p = 0.24). Their inclusion did not materially change the magnitude or significance of the focal path coefficients reported in Table 7, indicating that the hypothesised relationships are robust to these controls. For parsimony, the controls were therefore retained in the estimated model but are not displayed in Figure 2 and Figure 3, which present the focal theoretical paths.

The mediating hypotheses H7 and H8 were examined using the bias corrected bootstrapping approach recommended for the assessment of indirect effects in PLS-SEM ([34]; [15]). As reported in Table 8, both indirect effects were positive and statistically significant, and neither confidence interval included zero. The indirect effects equal the product of the corresponding direct paths reported in Table 7 (for H7, 0.52 × 0.38 = 0.198; for H8, 0.55 × 0.41 = 0.226), so the figures and tables reproduce the same mediation model. Because the direct effects of entrepreneurial leadership on the outcomes (H1 and H2) remained significant alongside the significant indirect effects, the results indicate complementary partial mediation in both cases ([15]; [34]). Entrepreneurial intention therefore partially mediated the relationship between entrepreneurial leadership and employee creativity, while entrepreneurial self-efficacy partially mediated the relationship between entrepreneurial leadership and innovative work behaviour.

Table 8. Mediation analysis: Specific indirect effects.

Note. All indirect effects significant at p < 0.001. Both direct effects remained significant, indicating complementary partial mediation.

As shown in Table 8, the indirect effect of entrepreneurial leadership on employee creativity through entrepreneurial intention was positive and statistically significant, β = 0.198, SE = 0.035, t = 5.66, p < 0.001, 95% BCI [0.13, 0.27]. Therefore, H7 was supported. This result indicates that entrepreneurial intention is an important psychological mechanism through which entrepreneurial leadership enhances employee creativity.

The indirect effect of entrepreneurial leadership on innovative work behaviour through entrepreneurial self-efficacy was also positive and statistically significant, β = 0.226, SE = 0.034, t = 6.65, p < 0.001, 95% BCI [0.16, 0.29], supporting H8. This finding suggests that entrepreneurial leadership strengthens employees’ confidence in their entrepreneurial capabilities, which subsequently improves their innovative work behaviour. Since the direct effects of entrepreneurial leadership on employee creativity and innovative work behaviour remained significant after including the mediators, the results confirm complementary partial mediation in both relationships ([34]; [15]).

This study contributes to theory in three ways. First, it extends social cognitive theory by demonstrating how entrepreneurial leadership functions as a contextual stimulus that activates entrepreneurial intention and entrepreneurial self-efficacy, which then shape employee creativity and innovative work behaviour. This confirms the relevance of cognitive and motivational mechanisms in explaining leadership effects on innovation ([9]; [32]).

Second, the study contributes to entrepreneurial leadership research by showing that entrepreneurial leadership is a significant antecedent of two distinct innovation outcomes: employee creativity and innovative work behaviour. By separating creativity from innovative work behaviour, the study clarifies the difference between idea generation and idea implementation, a distinction often overlooked in leadership and innovation studies ([2]; [27]).

Third, the study advances existing knowledge by modelling entrepreneurial intention and entrepreneurial self-efficacy as parallel mediators in one integrated framework. The findings suggest that entrepreneurial intention is more closely linked to creativity, while entrepreneurial self-efficacy is more closely linked to innovative work behaviour. This provides a more precise explanation of how entrepreneurial leadership influences different stages of the innovation process ([41]; [38]). The study also responds to calls for more context specific leadership research in African SMEs, where innovation is often constrained by limited resources and institutional challenges ([8]; [22]).

The findings provide important guidance for SME owners and managers. First, managers should develop entrepreneurial leadership behaviours such as opportunity seeking, proactiveness, calculated risk taking, autonomy support, and encouragement of experimentation. These behaviours can create a work climate that enables employees to think creatively and engage in innovation related activities ([28]; [35]).

Second, SMEs should deliberately strengthen employees’ entrepreneurial intention by encouraging initiative, recognising creative suggestions, involving employees in problem solving, and linking daily work tasks to value creation. Employees who see innovation as part of their role are more likely to generate useful ideas and contribute to organisational renewal ([33]; [27]).

Third, managers should build entrepreneurial self-efficacy through mentoring, coaching, skills development, feedback, and opportunities for employees to successfully handle innovation related tasks. Because self-efficacy strongly predicts innovative work behaviour, confidence building should be treated as a core managerial responsibility rather than a personal employee trait ([32]; [38]). In practice, SMEs may first stimulate employees’ intention to generate ideas and then strengthen their confidence to implement those ideas.

The findings have wider social and policy relevance. African SMEs are central to employment creation, livelihood resilience, poverty reduction, and inclusive economic growth. Strengthening their innovation capacity can therefore improve not only firm performance but also broader socioeconomic development ([1]; [7]). The study shows that innovation in SMEs does not depend only on finance or technology but also on leadership quality and employee motivation.

Policy makers, enterprise development agencies, and business support institutions should therefore complement financial and technical assistance with leadership development, innovation training, mentoring, and employee capability building. Such interventions are particularly important for resource constrained SMEs because entrepreneurial leadership, intention, and self-efficacy are relatively low cost but powerful drivers of creativity and innovative work behaviour.

This study has some limitations. First, the cross-sectional design limits the ability to draw strong causal conclusions. In particular, the mediation results should be interpreted with care: the indirect effects reported for H7 and H8 reflect statistical indirect associations estimated from data collected at a single point in time, rather than evidence of a temporal process in which leadership precedes the change in intention or self-efficacy, which in turn precedes the change in creativity or innovative work behaviour. Because all constructs were measured simultaneously, the analysis cannot establish the temporal ordering or the dynamic unfolding that the term mediation may imply. Future studies should therefore use longitudinal, time lagged, or experimental designs to examine how entrepreneurial leadership, intention, self-efficacy, creativity, and innovative work behaviour evolve and influence one another over time ([15]; [36]). Second, the use of self-reported data may raise concerns about common method bias, although procedural and statistical controls reduced this threat ([23]). Future research could combine employee responses with supervisor ratings or objective innovation indicators.

Third, the study focused on SMEs in Ghana, which may limit generalisability to other African countries and sectors. Comparative studies across African economies would provide stronger evidence on contextual differences and boundary conditions ([7]; [22]). Fourth, the model specified matched mediator to outcome pathways, linking entrepreneurial intention to creativity and entrepreneurial self-efficacy to innovative work behaviour, on theoretical grounds. Future research could formally test the two crossed paths, that is, the effect of entrepreneurial intention on innovative work behaviour and the effect of entrepreneurial self-efficacy on employee creativity, to establish whether the mechanisms are as outcome specific as the present theory suggests or whether they operate more broadly across the innovation process. Finally, future studies could extend the model by examining moderators such as organisational climate, perceived organisational support, digital readiness, learning culture, or environmental uncertainty. Additional mediators such as psychological empowerment, work engagement, and intrinsic motivation may also provide deeper insight into how entrepreneurial leadership shapes innovation.

This study examined entrepreneurial leadership as a driver of employee creativity and innovative work behaviour in African SMEs through the dual mediating roles of entrepreneurial intention and entrepreneurial self-efficacy. The findings confirm that entrepreneurial leadership directly enhances both creativity and innovative work behaviour. More importantly, the results show that entrepreneurial intention and entrepreneurial self-efficacy partially transmit these effects, indicating that leadership shapes innovation through employees’ motivational and cognitive states.

The study contributes to theory by extending social cognitive theory and entrepreneurial leadership research into the African SME context. It contributes to knowledge by distinguishing the creativity pathway from the innovative work behaviour pathway and by showing that entrepreneurial intention and self-efficacy perform complementary mediating roles. Practically, the study suggests that SME innovation can be improved by developing entrepreneurial leaders who build employee motivation, confidence, autonomy, and opportunity seeking behaviour. At the policy level, the findings show that strengthening leadership and employee capability offers an accessible route to improving SME innovation, resilience, employment generation, and inclusive growth in Africa.