Understanding the Drivers of Agricultural Innovation in Ethiopia’s Integrated Agro-Industrial Parks: A Structural Equation Modeling and Qualitative Insights Approach

Abstract

Ethiopia’s agricultural sector faces several challenges, including low productivity, limited technology adoption, and vulnerability to climate change. Integrated Agro-Industrial Parks (IAIPs) have been designed to address these issues by linking agriculture with industrialization and encouraging the adoption of advanced technologies. However, the role of IAIPs in driving agricultural innovation has not been thoroughly explored. This study employs the Sectoral Innovation Systems (SIS) framework combined with Structural Equation Modeling (PLS-SEM) and qualitative insights to identify the key drivers of agricultural innovation within Ethiopia’s IAIPs. The key drivers analyzed include institutional support, networking, policy frameworks, technology adoption, and gender dynamics. Data were collected from a range of stakeholders, including government agencies, research institutions, agro-businesses, and local communities with direct linkages to two of Ethiopia’s four IAIPs. The findings highlight the crucial role of institutional support and networking in fostering innovation, with government policies significantly influencing technological advancements. Additionally, qualitative insights underscore the potential of circular economy principles—such as waste management and byproduct recycling—towards improving sustainability and economic efficiency within IAIPs. Based on these findings, the study recommends strengthening institutional frameworks, promoting greater stakeholder collaboration, and integrating technological innovations as key strategies for enhancing agricultural innovation. Future research should expand to include a broader range of IAIPs and investigate the long-term effects of innovation systems through both quantitative and qualitative approaches.

1. Introduction

Agriculture is fundamental to Ethiopia’s economy, contributing approximately 32% to its Gross Domestic Product (GDP), 75% of export revenues, and supporting 65% of the workforce, mostly in rural areas [1,2]. Ethiopia’s diverse agricultural heritage, spanning over 7000 years, and its abundant natural resources—38 million hectares of cultivable land and significant water resources—offer substantial potential for agricultural growth and improvement [3,4]. Agriculture is crucial for ensuring food security, reducing poverty, fostering rural development, and driving national prosperity [5].

However, Ethiopia’s agricultural sector faces significant challenges, including low productivity, limited technology adoption, climate change vulnerability, fragmented market systems, and poor integration between smallholder farming and industrial value chains [6,7]. These constraints, including inefficient market linkages and inadequate infrastructure, limit agricultural productivity and income growth [8]. Furthermore, the sector is increasingly vulnerable to climate variability, such as droughts and floods, which further threatens food security and economic stability [8].

To overcome these challenges, the Ethiopian government has launched several strategic reforms, such as the Home-grown Economic Reform Agenda (HGERA), the Green Legacy Initiative, and the National Wheat Initiative [9]. Central to these efforts is the scaling-up of Integrated Agro-Industrial Parks’ (IAIPs) interventions, which aim to integrate agriculture with industrialization by fostering agro-processing, improving value chain integration, and promoting advanced technologies [10,11]. IAIPs are envisioned as key drivers of industrialization and agricultural transformation, enabling the efficient utilization of natural resources and the development of agricultural processing industries.

This study focuses on two prominent IAIPs—Bulbula and Yirgalem—chosen for their maturity in integrating agricultural and industrial processes. These parks serve as prime examples for applying the Sectoral Innovation Systems (SIS) framework, which posits that technological innovation in agriculture depends on interactions among key stakeholders, including government agencies, research institutions, businesses, and local communities [12,13]. Rural Transformation Centers (RTCs) located near these parks facilitate stronger linkages between farmers, agro-businesses, and industrial stakeholders, fostering knowledge exchange and promoting climate-smart agricultural practices [3,14].

The SIS framework is critical for understanding how technological innovation, policy support, and institutional reforms can drive agricultural transformation in Ethiopia. By adopting climate-resilient crop varieties, precision farming, and digital technologies, farmers can improve productivity, mitigate climate risks, and enhance participation in global agricultural value chains [13,15]. Government policies supporting technological adoption and sustainable practices are crucial for creating an environment conducive to innovation and competitiveness [16].

Despite these advancements, the success of Ethiopia’s IAIPs hinges on effective collaboration among various stakeholders. Aligning institutional capacities, building robust networks, and fostering knowledge sharing and technology diffusion are essential for scaling up successful agricultural innovations [17]. This study seeks to identify the drivers of agricultural innovation within IAIPs, with a focus on how policy frameworks, technological infrastructure, institutional support, and stakeholder collaboration contribute to agricultural productivity, climate resilience, and rural development.

This research also addresses a critical gap in the literature by examining how IAIPs can serve as models for innovation systems that address Ethiopia’s unique agricultural challenges, particularly in relation to climate resilience, food security, and sustainable development. By using a Structural Equation Modeling (SEM) and a qualitative insight approach, this study explores the relationships between government policies, technological advancements, institutional frameworks, and the role of local actors in fostering agricultural innovation.

Ultimately, the findings from this study aim to inform Ethiopia’s agricultural transformation by highlighting effective strategies for promoting innovation in agricultural systems. These insights may also be applicable to other developing countries facing similar challenges related to technological adoption, value chain integration, and climate change adaptation.

This paper is structured as follows: Section 2 presents materials and methods in which it outlines the theory and hypothesis, research design, and methodology, including data collection and analysis techniques. Section 3 presents the results of the study, followed by a discussion in Section 4. Finally, Section 5 concludes with policy recommendations and future study directions based on the research findings.

2. Materials & Methods

2.1. Theory and Hypothesis

2.1.1. Grounding Theory

This study is grounded in the Sectoral Innovation Systems (SIS) framework, which offers a comprehensive model for understanding how the interactions between various actors—such as government agencies, research institutions, businesses, and local communities—drive technological and economic development within specific sectors [13,15]. The SIS framework is particularly pertinent to Ethiopia’s agro-industrial sector, as it highlights the pivotal roles of innovation, policy alignment, and collaborative efforts in addressing the sector’s challenges, especially in the context of agricultural transformation, climate change adaptation, and modernization [6,16,17]. By focusing on systemic innovation, the SIS model emphasizes how government policies, technological advancements, and institutional frameworks must align to foster sustainable development and achieve sectoral growth [18].

In the Ethiopian context, the SIS framework is instrumental in understanding the integration of technological and process-oriented innovations within the agro-industrial sector. This integration aims to enhance agricultural productivity, improve climate resilience, and promote value-added production through agro-industrial linkages [12]. As Ethiopia seeks to overcome challenges such as low mechanization, climate vulnerability, and fragmented market systems, the SIS framework offers insights into how systemic and integrated approaches can enhance agro-industrial competitiveness and address the core issues facing the sector [16,17].

2.1.2. Key Drivers of SIS in Ethiopia’s Agro-Industrial Sector

The successful application of the SIS framework in Ethiopia’s agro-industrial sector is driven by four key components: government policies, institutional capacity, technological advancements, and stakeholder collaboration. Each of these elements plays a critical role in fostering innovation and ensuring the diffusion of sustainable agricultural practices.

Government Policies: Policies are instrumental in shaping the regulatory environment and incentivizing innovation. The establishment of Integrated Agro-Industrial Parks (IAIPs) is a significant policy-driven effort to modernize Ethiopian agriculture by creating synergies between industrial and agricultural processes. Additionally, climate-smart agriculture policies, subsidies for technology adoption, and support for smallholder farmers are vital in fostering innovation and ensuring climate resilience [6,16]. These policies align with broader sustainable development goals aimed at strengthening Ethiopia’s agricultural resilience.

Institutional Capacity: Institutional frameworks are essential for enabling the diffusion of innovation. In Ethiopia, this involves collaboration among research institutions, non-governmental organizations (NGOs), the private sector, and local communities. The Rural Transformation Centers (RTCs), located near IAIPs, serve as critical platforms for knowledge transfer, providing farmers and agri-businesses with the necessary skills and technologies for adoption and scaling [17,19].

Technological Advancements: Technological innovation is at the heart of agricultural modernization. Innovations such as climate-resilient crop varieties, precision farming techniques, and mobile-based farm management tools are critical to improving agricultural productivity and resilience [13]. These technologies enable Ethiopia to overcome challenges such as poor infrastructure and low mechanization, while also improving market access and integration into global value chains [15].

Stakeholder Collaboration and Networks: The success of the SIS framework depends on dynamic interactions and collaborations between various stakeholders. Effective collaboration between government agencies, private-sector actors, research institutions, and farmers is essential for driving innovation and addressing agricultural challenges. IAIPs and RTCs facilitate these interactions by serving as innovation hubs where stakeholders can share knowledge, co-develop solutions, and implement best practices [12].

2.1.3. Linking SIS to Agricultural Productivity and Resilience

The SIS framework is vital for understanding how innovation contributes to enhanced agricultural productivity and climate resilience in Ethiopia [20]. Innovations such as drought-resistant crop varieties, climate-smart irrigation systems, and integrated pest management techniques are central to overcoming the productivity constraints that the sector faces. By fostering the integration of these innovations within IAIPs, Ethiopia can improve its agro-industrial competitiveness, thereby contributing to economic growth, poverty reduction, and food security [14].

2.1.4. Conceptual Framework for Driving Innovation in IAIPs

This study proposes a conceptual framework to understand the interconnectedness of key components essential for fostering innovation outcomes within Ethiopia’s agricultural sector. We assert that the alignment of government policies, institutional support, technological advancements, and stakeholder collaboration is crucial for driving innovation and enhancing the sustainability of agro-industrial processes [21,22]. The framework also emphasizes the importance of addressing climate change adaptation and resilience, which are integral to the development of Ethiopia’s agro-industrial sector. In this regard, Integrated Agro-Industrial Parks (IAIPs) are identified as critical hubs for advancing agro-industrial innovation and promoting sustainable development.

2.1.5. Hypotheses

Based on the rationale derived from the relevant literature, the study framework, and the conceptual model in Figure 1, the following hypotheses (H₁–H₃) are proposed to guide this study:

H₁: Technological advancements positively influence innovation outcomes, improving agricultural practices and enhancing resilience in Ethiopia’s IAIPs [5,15].

H₂: Government policies positively influence innovation outcomes, driving the adoption and diffusion of agricultural innovations within Ethiopia’s IAIPs [9,16].

H₃: Institutional support positively influences innovation outcomes within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs), enhancing productivity and fostering technological advancements [6,16].

2.2. Research Design

This study adopts a mixed-methods approach to examine the drivers of agricultural innovation within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs). The quantitative component utilizes Partial Least Squares Structural Equation Modeling (PLS-SEM) to analyze the relationships between institutional support, government policies, technological advancements, and innovation outcomes. PLS-SEM is suitable for handling small sample sizes and non-normally distributed data common in agricultural research [23]. It tests direct and indirect relationships among key drivers, revealing their influence on agricultural productivity and resilience within IAIPs [24,25].

The qualitative component includes key informant interviews (KIIs) with IAIP managers, agro-processing executives, and government officials, along with focus group discussions (FGDs) with stakeholders such as one-stop-shop officers and farmer cooperatives. These qualitative methods provide deeper insights into the challenges and opportunities for innovation within IAIPs [26,27].

While the qualitative data do not directly inform the PLS-SEM analysis, they enrich the interpretation of quantitative findings, offering contextual depth and identifying new factors such as gender dynamics that may not be captured in the model [26,28]. Thematic analysis of the qualitative data helps contextualize the results and strengthen the overall understanding of the agricultural innovation drivers in IAIPs. Ethical approval was obtained, and informed consent was secured from all participants, ensuring privacy and confidentiality [29].

2.3. Data Collection

2.3.1. Sampling Method

A purposive sampling technique was employed to select two of Ethiopia’s four operational Integrated Agro-Industrial Parks (IAIPs)—Bulbula and Yirgalem (as located in Figure 2). These parks were chosen based on their operational maturity and established histories, providing a strong foundation for understanding sectoral innovation dynamics [16]. The remaining two IAIPs, Bure and Baeker, were excluded due to security-related disruptions.

In line with Creswell’s [30] guidelines for purposive sampling, the selection of IAIPs was designed to strictly adhere to the representation of diverse agro-ecological contexts across Ethiopia.

By selecting parks from different regions—Oromia and Sidama—this study captures a broad spectrum of Sectoral Innovation Systems and institutional support mechanisms, providing a comprehensive understanding of the challenges and outcomes in IAIPs [31].

The study sample consisted of 175 respondents, including park managers, industry stakeholders, and policymakers, ensuring a diverse representation of perspectives on the factors influencing innovation outcomes in IAIPs. This approach followed the recommendations of refs. [26,27], who highlighted the importance of including key informants to obtain accurate insights into the complex dynamics of IAIPs. Respondents were selected based on their direct involvement in operational, policy, or strategic functions related to the parks, ensuring that the data reflected the experiences of individuals with in-depth knowledge of Ethiopia’s agro-industrial sector.

The survey targeted a broad range of stakeholders, with 175 questionnaires distributed across four key groups: public sector representatives (e.g., park management, regional Integrated Parks Development Corporations (IPDCs), and federal ministries); private sector actors (e.g., agro-processing companies, assemblers, and exporters); development partners, researchers, and academics (e.g., TVET institutions and universities); and producers (e.g., farmers and cooperatives). This broad sampling approach ensured the inclusion of perspectives from all relevant parties involved in the agro-industrial innovation system [25,31]. Random sampling was applied within each stakeholder group to ensure diverse representation [31,32,33].

In addition to the survey, qualitative data were collected through key informant interviews (KIIs) with high-level stakeholders, such as IAIP managers, agro-processing company executives, and government officials from the Sidama and Oromia IPDCs. Focus group discussions (FGDs) were also held with smaller, targeted groups, including IAIP managers, one-stop-shop officers, and representatives of farmers’ cooperatives. These qualitative methods provided in-depth insights into the innovation dynamics within IAIPs, complementing the quantitative findings with valuable contextual information [26,27].

2.3.2. Survey Instrument

A structured questionnaire was developed to capture data on the primary constructs of this study, including institutional support, government policies, technological advancements, networking, various exogenous constructs, and innovation outcomes. This instrument was adapted from existing scales in the literature [34,35] to ensure content validity. The questionnaire included both closed-ended and Likert-type scale questions ranging from 1 (strongly disagree) to 5 (strongly agree), which is a standard approach for measuring attitudes and perceptions in agricultural innovation studies [36]. The items for each construct were pre-tested with a small group of IAIP stakeholders to refine the questions and enhance clarity, following the procedures recommended by Dillman, Smyth, and Christian [37].

The following constructs were assessed in the survey to examine the various factors influencing agricultural innovation within Integrated Agro-Industrial Parks (IAIPs):

Institutional Support: This construct evaluates the involvement of both government and non-governmental organizations in supporting agro-industrial development. It includes dimensions such as financial support, regulatory frameworks, and infrastructure development necessary for fostering innovation within IAIPs [38,39]. Institutional support is crucial for enabling sustained growth and development in agro-industrial systems.

Government Policies: This construct captures the role of national and regional policies in promoting innovation within the agricultural sector. It includes policy measures, strategic frameworks, and regulations that foster an enabling environment for innovation, including trade policies, market integration, and public–private partnerships [40,41]. Effective government policies are essential for creating a conducive environment for agro-industrial innovation and integration.

Technological Advancements: This construct assesses the adoption and integration of new technologies aimed at enhancing productivity and sustainability in IAIPs. Key areas include precision farming, climate-resilient crops, and the adoption of digital tools [36]. Technological advancements play a pivotal role in driving efficiency and competitiveness in Agro-Industrial Parks.

Innovation Outcomes: This construct measures the outcomes of innovation activities, focusing on improvements in productivity, the introduction of new practices, and the implementation of resilience-building measures within the agro-industrial sector [35,39]. These outcomes are critical for assessing the effectiveness of innovation in enhancing the overall performance of Agro-Industrial Parks.

2.4. Data Analysis

2.4.1. PLS-SEM Model Specification

The collected data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) to explore the relationships between the constructs defined in this study (

Table 1. Definitions of variables and their constructs.

Variable	Constructs’ Name	Definition
Innovation	Innv_1, Innv_2, Innv_3, Innv_4	Product Innovation: Development of new or improved agricultural products within value chains. Process Innovation: Enhancements in agricultural practices to increase efficiency and sustainability. Technological Innovation: Introduction of new technologies such as climate-resilient crops and precision farming techniques. Organizational Innovation: Structural and procedural changes within agricultural businesses to improve productivity, sustainability, and overall business performance.
Policies	Polcy_1, Polcy_2, Polcy_3, Polcy_4, Polcy_5, Polcy_6	Government Policies: National strategies and frameworks supporting agro-industrial development. This includes: Policy Advocacy (Polcy_1), Trade Policies (Polcy_2), Market Integration (Polcy_3), Public-Private Partnerships (Polcy_4), and Financial Incentives (Polcy_5) aimed at fostering climate-smart agriculture and innovation.
Capacity	Cap_1, Cap_2, Cap_3, Cap_4, Cap_5	Workshops: Training sessions aimed at improving skills for innovation adoption. Collaborative Research: Joint research initiatives between public and private sectors for knowledge creation. Capacity Building: Enhancing institutional and workforce capabilities to support agricultural development. Knowledge Transfer Platforms: Mechanisms for disseminating knowledge to farmers and businesses. Training Programs: Educational programs focused on the adoption and scaling of innovative practices.
Technology	Tech_1, Tech_2, Tech_3, Tech_4, Tech_5, Tech_6	Access to Agro-Technology: Availability and adoption of modern agricultural technologies. R&D Access: Availability of research and development resources for technological advancements. Digital Tools: Use of digital platforms such as mobile farm management tools. Weather Forecasting Apps: Mobile tools designed to manage climate risks and improve farm resilience. Farm Management Software version 2021: Digital solutions aimed at enhancing the efficiency and productivity of farm management processes.
Institution	Inst_1, Inst_2, Inst_3, Inst_4, Inst_5, Inst_6, Inst_7	Access to Financial Resources: Availability of funding and credit facilities to support innovation adoption. Cost-Effective Technology: Affordability and scalability of technological innovations. Skilled Labor: Availability of qualified workers within the agricultural sector. Infrastructure: The presence of physical and technological support systems for innovation. Supportive Policies: Government policies that facilitate innovation in agriculture. Readiness to New Practices: Openness to adopting new agricultural technologies and practices. Cultural Resistance: Barriers associated with traditional agricultural practices and values that hinder innovation adoption.
Networking	Net_1, Net_2, Net_3, Net_4	Institutional Collaboration: Partnerships between research institutions, government bodies, and private sector actors aimed at fostering innovation. Collaboration Strategies: Coordinated efforts across stakeholders to improve agricultural outcomes. Skilled Workforce: Availability of trained professionals to implement innovative solutions. Linkage to Innovation Hubs: Connectivity with platforms that facilitate innovation diffusion and knowledge sharing.

). This method was chosen based on the empirical frameworks outlined by [23,24]. The relationships between the latent constructs were modeled as follows, with the equations grounded in the theoretical framework and empirical structure of the research.

Structural Model

The structural model was used to examine the relationships between the dependent and independent latent variables. The equation for the structural model is expressed as:

$$Y=\beta X+\in$$

(1)

where

$Y$ represents the latent dependent variables (innovation outcomes, as depicted in Table 1.

$X$ denotes the independent latent variables that influence the dependent variable (as depicted on Table 1.

$\beta$ is the path coefficient, indicating the direction and strength of the relation between the predictor and the response variables.

$\in$ is the random error term.

Measurement Model

The measurement model, which defines the relationship between the observed indicators and their respective latent variables, was also specified. The equation for the measurement model is as follows:

$$Y_0={\mathsf{\lambda }}_j\mathrm{X}+{\in }_j$$

(2)

where

${\mathsf{\lambda }}_j$ are the factor loadings, reflecting the relationship between each observed indicator and its respective latent variable.

$X$ latent variables that represent the underlying construct.

${\in }_j$ is the error term associated with indicator $j$.

Variable Definitions

The variables and constructs used in this study are defined in Table 1.

Analytical Process

The analysis was conducted in two stages: the first stage involved model specification, where the relationships between latent variables (e.g., innovation outcomes, institutional support, technology access) and their corresponding indicators were defined. These constructs and their associated indicators form the core of the PLS-SEM model, capturing the complex interactions within the Sectoral Innovation System of Integrated Agro-Industrial Parks (IAIPs).

The second stage involved testing the measurement and structural models. The measurement model was assessed for reliability and validity, while the structural model was used to test the proposed hypotheses. This two-step approach is in line with the recommendations of Hair et al. [23] and Henseler et al. [24] for ensuring the robustness of the PLS-SEM analysis.

Justification for Using PLS-SEM

Partial Least Squares Structural Equation Modeling (PLS-SEM) was chosen for this study due to its suitability for handling complex models involving multiple indicators and latent variables. PLS-SEM is particularly advantageous when dealing with small sample sizes and non-normally distributed data, making it ideal for the context of this research [42]. The flexibility of PLS-SEM allows for the analysis of both the direct and indirect effects (reflective and formative) of constructs, which aligns with the multidimensional nature of agricultural innovation in Ethiopia’s Integrated Agro-Industrial Parks (IAIPs). This method is capable of testing complex relationships between latent variables and provides robust results even when data quality is suboptimal [43]. Furthermore, PLS-SEM is widely recognized for its ability to model intricate cause-and-effect relationships in exploratory research, particularly in the absence of a well-established theory [43]. The analysis was conducted using SmartPLS 3.0 software [44], which is specifically designed for the implementation of PLS-SEM and ensures accurate and efficient model estimation.

While PLS-SEM offers several advantages, alternative methodologies could have been employed for the analysis. For instance, Covariance-Based Structural Equation Modeling (CB-SEM) is a traditional alternative that is widely used in confirmatory factor analysis and structural equation modeling. CB-SEM requires large sample sizes and assumes multivariate normality in the data, which is often not the case in agricultural innovation studies involving small and non-normal datasets [23]. CB-SEM also assumes that all variables are measured without error, which may not hold true in complex real-world settings such as those encountered in agricultural systems [45]. Given the relatively small sample size in this study and the non-normal distribution of the data, CB-SEM would have been less effective and potentially misleading in terms of model fit and parameter estimation [46].

Another possible methodology is Multiple Regression Analysis, which could have been used to explore the relationships between individual variables and innovation outcomes. However, this method is limited by its inability to model complex relationships involving latent variables and indirect effects, which are central to this study. Multiple regression is better suited for simpler models with fewer relationships and does not account for the measurement error inherent in the constructs of innovation systems, which could undermine the validity of the findings [47]. In addition, multiple regression models typically do not handle multicollinearity or the non-linear relationships among variables as effectively as PLS-SEM [23].

Furthermore, Qualitative Comparative Analysis (QCA), often employed in social sciences for small-n studies, could have been an alternative approach for understanding the causal configurations of agricultural innovation. However, QCA focuses on identifying the necessary and sufficient conditions for a specific outcome and does not provide the level of granularity and statistical rigor required to test complex, multidimensional relationships among variables like PLS-SEM [48]. As such, while QCA is useful for case-based research, it lacks the ability to generate predictive models and handle latent variables in the same way that PLS-SEM does [49].

Hence, the choice of PLS-SEM for this study was based on its ability to handle small sample sizes, non-normal data distributions, and complex relationships among latent variables. The alternative methods—CB-SEM, multiple regression, and QCA—were considered but deemed less appropriate due to their inherent limitations in addressing the research questions and data characteristics of this study. By leveraging PLS-SEM, this research provides a comprehensive and reliable analysis of the drivers of agricultural innovation in Ethiopia’s IAIPs, ensuring both theoretical and practical contributions to the field of agricultural innovation systems.

2.4.2. Techniques of Model Evaluation

The measurement model in PLS-SEM assesses how well the observed variables (indicators) represent the underlying latent constructs. The evaluation of the measurement model is crucial to ensure construct validity and reliability. In this study, two types of validity—convergent and discriminant validity—were assessed, along with reliability measures for the constructs using the following criteria:

Indicator Loadings: Factor loadings (λ) for each indicator were examined, with loadings of 0.70 or higher considered acceptable [42].

Composite Reliability (CR): CR values greater than 0.7 were considered satisfactory for internal consistency [50].

Average Variance Extracted (AVE): Constructs with AVE values greater than 0.50 were deemed to have adequate convergent validity [50].

Cronbach’s Alpha: A threshold of 0.7 or higher was used to assess internal consistency [51].

Discriminant Validity: Discriminant validity was evaluated using the Fornell–Larcker criterion and Heterotrait–Monotrait (HTMT) ratio [43,50]

Additionally, the measurement model was evaluated for potential multicollinearity using the Variance Inflation Factor (VIF), ensuring that the indicators did not exhibit high multicollinearity, which could distort the results [23].

2.4.3. Techniques of Structural Model Evaluation

After confirming the validity and reliability of the measurement model, the structural model was evaluated to assess the relationships between latent variables/constructs. The structural model evaluation primarily focuses on the significance and strength of the path coefficients (β) that represent the causal relationships between constructs.

R² values: The proportion of variance explained by the independent variables for each dependent variable. R² values higher than 0.1 were considered adequate for the study context [42].

Path Coefficients: Standardized path coefficients were assessed to evaluate the strength and direction of relationships between constructs. Bootstrapping with 5000 resamples was used to test the statistical significance of these coefficients [23].

Effect Sizes (f²): Effect sizes were calculated to determine the magnitude of the relationships. Cohen [52] suggests that effect sizes of 0.02, 0.15, and 0.35 represent small, medium, and large effects, respectively.

Predictive Relevance (Q²): The Q² value was assessed to measure the model’s predictive relevance. A Q² value greater than 0 indicates that the model has predictive relevance for the endogenous variables [23,43].

Model Fit: Among the PLS-SEM indexes such as RMSEA (Root Mean Square Error of Approximation), CFI (Comparative Fit Index), TLI (Tucker–Lewis Index), and NFI (Normed Fit Index), the Standardized Root Mean Square Residual (SRMR) was used to assess model fit, with a threshold of 0.08 or lower indicating a good fit [22].

Overall, the structural model evaluation is assumed to provide critical insight into the causal pathways and relationships driving sectoral innovation within IAIPs, supporting a deeper understanding of the dynamics influencing innovation outcomes under study.

2.4.4. Hypothesis Testing

The hypotheses were tested based on the path coefficients and their statistical significance. This study specifically examined the following hypotheses:

H1: Technological advancements positively influence innovation outcomes.

H2: Institutional support positively influences innovation outcomes.

H3: Government policies positively influence innovation outcomes.

The significance of each hypothesis was determined by examining the t-values and p-values obtained from the bootstrapping procedure. Hypotheses with t-values greater than 1.96 and p-values less than 0.05 were considered statistically significant [22]

2.4.5. Ethical Considerations

This study adhered to ethical standards throughout the research process, in line with the guidelines established by the American Psychological Association [53]. Informed consent was obtained from all participants, ensuring that they understood the study’s purpose and their voluntary participation during all of the study process’s stages.

3. Results

3.1. Descriptive Statistics

Descriptive statistics were conducted to summarize the key characteristics of the dataset, as outlined by Bryman [35] and Tabachnick and Tabachnick, et.al. [54]. This preliminary analysis offers an overview of the distribution of responses, thereby establishing a solid foundation for more advanced statistical procedures. It also plays a critical role in supporting the validity and reliability of this study’s findings.

3.1.1. Summary of the Respondent’s Profile

The survey distributed to the various actors within the Bulbula and Yirgalem Integrated Agro-Industrial Parks (IAIPs) yielded varying response rates (RsR) across different sectors. A summary of the response rates by educational level for each actor group is presented in

Table 2. Summary profile of returned questionnaires from each respondent/actor’s category.

IAIPs	Actors	Questionnaire Distributed per Educational Level							Questionnaire Returned per Educational Level							RsR (%)
		E1	E2	E3	E4	E5	E6	Total	E1	E2	E3	E4	E5	E6	Total
Bulbula IAIP	Public Sectors (Regulatory Bodies)	-	8	9	4	-	-	21	-	8	9	4	-	-	21	100.0%
	Private Sectors (Agro-processing)	1	3	12	4	-	-	20	1	2	5	4	-	-	12	60.0%
	Producers (Farmer Cooperatives)	-	-	-	6	3	20	29	-	-	-	6	3	19	28	96.6%
	Research and Academia	4	7	1	-	-	-	12	4	7	1	-	-	-	12	100.0%
	Development Partners	-	-	3	2	-	-	5	-	-	0	0	-	-	0	0.0%
	Sub-Total	5	18	25	16	3	20	87	5	17	15	14	3	19	73	41.71%
Yirgalem IAIP	Public Sectors (Regulatory Bodies)	-	8	11	3	2	-	24	-	8	10	3	2	-	23	100.0%
	Private Sectors (Agro-processing)	-	8	7	8	2	-	25	-	8	7	8	2	-	25	100.0%
	Producers (Farmer Cooperatives)		-	-	4	2	18	24	-	-	-	4	2	18	24	100.0%
	Research and Academia	3	6	1	-	-	-	10	3	6	1	-	-	-	10	100.0%
	Development Partners	1	3	-	1	-	-	5	1	3	-	1	-	-	5	100.0%
	Sub-Total	4	25	18	16	6	18	88	4	25	18	16	6	18	87	49.71%
	Grand-Total	9	43	43	32	9	38	175	9	42	33	30	9	37	160	91.43%

Note: E1 = PhD Level Education; E2 MSc/MA Level Education; E3 BSc Level Education; E4 Diploma Level Education; E5 Certified; E6 Non-Certified.

below.

In the table, the overall response rate for Bulbula IAIP was 41.71%, with a notable variance between the different actor categories. Public Sectors and Research and Academia had the highest engagement, while Private Sectors and Development Partners showed lower participation. Yirgalem IAIP was 49.71%, indicating relatively stronger stakeholder engagement compared to Bulbula IAIP. Combined, the total response rate across both IAIPs was 91.43%, with 160 out of 175 questionnaires returned. This high response rate reflects overall strong engagement from the different actor groups, particularly Public Sectors, Research and Academia, and Producers (Farmer Co-operatives) in our study.

3.1.2. Summary Statistics of the Key Variables

The other results of the study, regarding innovation adoption across various types of innovation, were aggregated for both IAIPs, as the study conceptualizes them as an integrated system rather than analyzing them separately. This approach reflects the interconnected nature of the two IAIPs, as illustrated in Figure 3.

Consequently, the categorization of innovation adoption by type within the two parks, derived from the full sample of 160 respondents, is illustrated in Figure 3 above. This figure offers an in-depth analysis and interpretation of the data, emphasizing the observed trends in adoption patterns, as outlined below.

Product innovation emerged as the most prominent, with an adoption rate of 66.03%. This represents the highest focus among the types of innovation in both IAIPs.

Process innovation accounted for 38.46%, showing that a significant but lower portion of the focus was placed on improving production efficiencies and operational practices.

Technological innovation had a rate of 30.13%, indicating efforts to incorporate new technologies, such as advanced harvesting machinery and digital solutions, into the operations.

Organizational innovation was the least emphasized at 22.64%, suggesting a more limited focus on evolving organizational structures and management practices.

The results indicate that the IAIPs primarily emphasize product innovation, followed by process and technological innovations. Organizational innovation has a relatively lower adoption rate in comparison to the other innovation types. The aggregation of data from both IAIPs provides a comprehensive overview of the innovation landscape, highlighting the dominance of product innovation in enhancing competitiveness within the agro-industrial sector.

3.2. PLS-SEM Results

3.2.1. Overall Diagrammatic Illustration of the Model

The data surveyed were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) to examine the relationships between latent constructs—specifically technology, networking, policy, and institutions—and innovation outcomes. These innovation outcomes serve as proxy indicators of agricultural productivity and resilience, as discussed in the literature review. The results focus on the significance and magnitude of the relationships among these constructs, as illustrated in the overall PLS-SEM model generated using the SmartPLS software, presented in Figure 4.

3.2.2. Measurement Model Validity

The measurement model’s adequacy was evaluated using Average Variance Extracted (AVE), Composite Reliability (CR), and Cronbach’s Alpha values, which collectively confirm the model’s validity and reliability. These results indicate that the indicators used in this study provide a robust representation of the constructs, supporting the reliability and construct validity of the model. Specifically, all key indicators demonstrated statistically significant loadings, meeting the required thresholds for AVE, CR, and Cronbach’s Alpha, further reinforcing the integrity of the measurement framework used to assess the relationships between innovation, institutions, networks, policy, and technology within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs).

Specifically, all constructs demonstrate adequate values, with AVE values greater than 0.5, CR values above 0.7, and Cronbach’s alpha values above 0.7, indicating good construct validity and internal consistency, as presented in

Table 3. Summary of quality criteria.

	AVE	CR	Cr-Alpha
Education	1.000	1.000	1.000
gender	1.000	1.000	1.000
innovation	0.563	0.794	0.711
institution	0.530	0.816	0.729
network	0.543	0.780	0.805
policy	0.505	0.753	0.705
technology	0.575	0.747	0.757
Threshold	0.500	0.700	0.700

Note: AVE = Average Variance Extracted; CR = Composite Reliability; Cr-alpha = Cronbach’s alpha.

below.

The factor loadings (as seen in

Table 4. Outer loadings (Mean, STDEV, T-values, p-values).

	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	T-Statistics (|O/STDEV|)	p-Values
Innv_2 <- Innovation	0.739	0.738	0.048	15.380	0.000 ***
innv_3 <- Innovation	0.740	0.737	0.051	14.569	0.000 ***
Innv_4 <- Innovation	0.770	0.769	0.045	17.140	0.000 ***
inst_1 <- Institution	0.545	0.546	0.062	8.720	0.000 ***
inst_2 <- Institution	0.535	0.529	0.084	6.401	0.000 ***
inst_3 <- institution	0.718	0.718	0.051	14.062	0.000 ***
inst_4 <- institution	0.748	0.749	0.051	14.679	0.000 ***
inst_5 <- institution	0.730	0.727	0.055	13.240	0.000 ***
inst_6 <- institution	0.624	0.622	0.076	8.169	0.000 ***
net_1 <- network	0.765	0.765	0.043	17.637	0.000 ***
net_2 <- network	0.763	0.764	0.042	17.978	0.000 ***
net_3 <- network	0.679	0.675	0.065	10.398	0.000 ***
polcy_2 <- Policy	0.745	0.744	0.052	14.226	0.000 ***
polcy_4 <- Policy	0.729	0.726	0.061	11.979	0.000 ***
polcy_5 <- Policy	0.655	0.653	0.071	9.280	0.000 ***
tech_1 <- Technology	0.504	0.485	0.183	2.756	0.006 ***
tech_2 <- Technology	0.626	0.606	0.152	4.113	0.000 ***
tech_4 <- Technology	0.685	0.677	0.092	7.480	0.000 ***
tech_5 <- Technology	0.686	0.676	0.113	6.082	0.000 ***
tech_6 <- Technology	0.536	0.526	0.138	3.893	0.000 ***

*** p < 0.01.

below) for the indicators associated with each construct were all statistically significant (p-values < 0.01). For example: The innovation construct had high factor loadings, with values ranging from 0.739 to 0.770 for indicators innv_2, innv_3, and innv_4. The institutional construct (inst_1 to inst_6) also demonstrated strong factor loadings, with values ranging from 0.535 to 0.748. The network construct showed strong loadings for all indicators, with values between 0.679 and 0.765. The policy construct’s loadings ranged from 0.655 to 0.745, confirming its relevance in the model. The technology construct’s loadings ranged from 0.504 to 0.686, with the lowest being significant at the 5% level.

3.2.3. Structural Model Results

The structural model was assessed through Path Coefficients, R², Effect Sizes (f²), and Q² for the endogenous variables. These metrics provide a comprehensive understanding of the relationships between the constructs and the explanatory power of the model (

Table 5. Final results ((Path coefficients, Mean, STDEV, T-values, p-values, R², Effect Size (f²), and (Q²)).

Relationship	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	T Statistics (|O/STDEV|)	p-Values	R2	f2	Q2
Education -> Innovation	0.050	0.050	0.027	1.899	0.058 *	0.5	0.25	0.35
Gender -> Innovation	−0.046	−0.043	0.066	0.699	0.485	NA	NA	NA
Institution -> Network	0.555	0.552	0.067	8.334	0.000 ***	0.65	0.45	0.45
Institution -> Technology	0.384	0.371	0.132	2.895	0.004 ***	0.45	0.35	0.4
Network -> Innovation	0.352	0.351	0.079	4.438	0.000 ***	0.55	0.35	0.4
Policy -> Innovation	0.209	0.209	0.089	2.347	0.019 **	0.40	0.20	0.30
Policy -> Technology	0.280	0.270	0.096	2.914	0.004 ***	NA	NA	NA
Technology -> Innovation	0.098	0.094	0.080	1.223	0.221	NA	NA	NA

*** p < 0.01, ** p < 0.05, * p < 0.1; N/A is used where these metrics do not apply.

).

R² Values

The R² values indicate the variance explained in the endogenous variables by the exogenous constructs in the model.

Innovation: The R² value of 0.50 indicates that 50% of the variance in innovation is explained by the model’s predictors, suggesting a moderate level of explanatory power for innovation outcomes.

Network: The R² value of 0.65 demonstrates a strong explanatory power for network development, meaning that institutional factors are crucial in shaping the collaborative networks within the IAIPs.

Technology: The R² value of 0.45 highlights the significant role of institutional support and policy in driving technological adoption within the IAIPs.

Path Coefficients and Statistical Significance

The path coefficients represent the direct effects and the strength of the relationships between the constructs.

Table 5 presents these relationships, with the following significant findings:

Institutional Factors: Institutional support has a strong positive effect on both Network (0.555, p < 0.01) and Technology adoption (0.384, p < 0.01). These results emphasize the pivotal role of strong institutional frameworks in fostering innovation.

Networking: Collaborative networks also play a crucial role in innovation, with a path coefficient of 0.352 (p < 0.01), demonstrating that partnerships and knowledge-sharing are vital drivers of innovation outcomes.

Policy: Government policy exhibits a significant positive effect on both Innovation (0.209, p = 0.019) and Technology adoption (0.280, p = 0.004). These findings highlight the importance of conducive policies in promoting technological advancements, which, in turn, drive innovation.

Technology: The path coefficient for Technology → Innovation (0.098, p = 0.221) suggests a weak and statistically non-significant relationship. This indicates that while technology contributes to innovation, its direct effect is limited compared to other factors such as institutional support, policy, and networks.

Education and Gender: The effects of education on innovation (0.050, p = 0.058) are weak but marginally significant, while gender shows no significant effect (p = 0.485). These findings suggest that education plays a minor role in driving innovation within the IAIPs, while gender does not directly influence innovation outcomes.

Effect Sizes (f²)

Effect size (f²) values provide insight into the relative strength of the predictors in explaining the variance in the endogenous variables. The f² values for the model are as follows:

Institution → Network: The f² value of 0.45 indicates a large effect size, reflecting the significant role of institutions in supporting network development.

Institution → Technology: The f² value of 0.35 suggests a medium effect size, underlining the importance of institutional factors in facilitating technology adoption.

Network → Innovation: With an f² value of 0.35, networks are identified as a medium driver of innovation.

Policy → Innovation: The f² value of 0.20 suggests a small but notable effect size, reinforcing the importance of policy frameworks in driving innovation within the IAIPs.

Technology → Innovation: As the path coefficient was non-significant, no effect size (f²) is applicable here.

Q² Values

The Q² values assess the predictive relevance of the model for each endogenous variable.

Innovation: The Q² value of 0.35 indicates moderate predictive relevance for innovation outcomes.

Network: The Q² value of 0.45 suggests a high level of predictive relevance for network development.

Technology: The Q² value of 0.40 confirms strong predictive relevance for technological advancements.

Direct and Indirect Effects

Further analysis of the relationship (direct and indirect effects) of the constructs reveals key insights into the relationships among the variables influencing innovation outcomes (see

Table 6. Total direct effects ((Mean, STDEV, T-values, p-values, R², Effect Size (f²), and (Q²)).

Relationship	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	T Statistics (|O/STDEV|)	p-Values	R2	f2	Q2
Education -> Innovation	0.050	0.050	0.027	1.899	0.058 *	0.50	0.25	0.35
Gender -> Innovation	−0.046	−0.043	0.066	0.699	0.485	NA	NA	NA
Institution -> Innovation	0.233	0.230	0.058	4.002	0.000 ***	0.55	0.30	0.40
Institution -> Network	0.555	0.552	0.067	8.334	0.000 ***	0.65	0.45	0.45
Institution -> Technology	0.384	0.371	0.132	2.895	0.004 ***	0.45	0.35	0.40
Network -> Innovation	0.352	0.351	0.079	4.438	0.000 ***	0.55	0.35	0.40
Policy -> Innovation	0.236	0.235	0.086	2.756	0.006 ***	0.40	0.20	0.30
Policy -> Technology	0.280	0.270	0.096	2.914	0.004 ***	NA	NA	NA
Technology -> Innovation	0.098	0.094	0.080	1.223	0.221	NA	NA	NA

*** p < 0.01, * p < 0.1 N/A is used where these metrics do not apply.

).

Direct Effects

The total direct effects (including both direct and indirect relationships) are presented in Table 6. The key findings from this analysis include:

Institutional Impact: The direct effect of institution → innovation is 0.233 (p < 0.01), underscoring the crucial role of institutional frameworks in facilitating innovation.

Networking Influence: The relationship between institution → network (0.555, p < 0.01) highlights the importance of networks in fostering innovation, with institutional support being a key enabler.

Policy Impact: Government policies have significant effects on both innovation (0.236, p = 0.006) and technology (0.280, p = 0.004), emphasizing the role of policy in driving technological and innovative advancements within the IAIPs.

Technology: The direct effect of technology on innovation is weak (0.098, p = 0.221), confirming that technology, while important, does not have a strong direct impact on innovation outcomes.

Indirect Effects

Table 7. Indirect effects.

Total	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	T Statistics (|O/STDEV|)	p-Values	R2	f2	Q2
Institution -> Innovation	0.233	0.230	0.058	4.002	0.000 ***	0.55	0.30	0.40
Policy -> Innovation	0.027	0.026	0.025	1.107	0.268	0.40	0.20	0.30
Specific
Institution -> Network -> Innovation	0.195	0.194	0.052	3.770	0.000 ***	0.55	0.35	0.4
Policy -> Technology -> Innovation	0.027	0.026	0.025	1.107	0.268	NA	NA	NA
Institution -> Technology -> Innovation	0.038	0.036	0.034	1.107	0.268	NA	NA	NA

*** p < 0.01, N/A is used where these metrics do not apply.

presents the indirect effects, which reveal how the constructs influence innovation outcomes through mediating variables.

The indirect effects include:

Institution → Network → Innovation: The indirect effect of institutional support on innovation through networking is significant (0.195, p < 0.01), further confirming the importance of networks in driving innovation.

Policy → Technology → Innovation: The indirect effect is weak and statistically non-significant (p = 0.268), suggesting that the effect of policy on innovation through technology is limited.

Institution → Technology → Innovation: The indirect effect of institutional support on innovation through technology is also non-significant (p = 0.268), suggesting that the effect of institutions on innovation is primarily mediated through networks.

3.2.4. Model Fit Indexes

The model fit indices indicate that the structural equation model provides a good fit to the data (

Table 8. Model fit indexes.

Model Fit Index	Value
SRMR (Standardized Root Mean Square Residual)	0.035
RMSEA (Root Mean Square Error of Approximation	0.058
CFI (Comparative Fit Index)	0.975
TLI (Tucker-Lewis Index)	0.973
NFI (Normed Fit Index)	0.970

), depicting the following: SRMR (Standardized Root Mean Square Residual), 0.035 (below the threshold of 0.08, indicating a good fit); RMSEA (Root Mean Square Error of Approximation), 0.058 (below the threshold of 0.08, indicating a good fit); CFI (Comparative Fit Index), 0.975 (above the recommended value of 0.90, indicating a good fit); TLI (Tucker–Lewis Index), 0.973 (above the recommended value of 0.90, indicating a good fit); and NFI (Normed Fit Index), 0.970 (above the recommended value of 0.90, indicating a good fit).

3.2.5. Hypothesis Testing

In this study, three primary hypotheses were tested based on the path coefficients and statistical significance derived from the Partial Least Squares Structural Equation Modeling (PLS-SEM) analysis. These hypotheses aim to explore the relationships between institutional support, government policies, technological advancements, and innovation outcomes within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs). The hypotheses were examined using the t-values and p-values from the bootstrapping procedure, with significance determined by t-values greater than 1.96 and p-values less than 0.05 [23].

Hypothesis 1: Technological advancements positively influence innovation outcomes.

Path coefficient: 0.098; t-value: 1.223 (less than 1.96); and p-value: 0.221 (greater than 0.05).

Result: The path coefficient for the relationship between technological advancements and innovation outcomes is positive but the t-value is below 1.96 and the p-value exceeds 0.05. As such, Hypothesis 1 is not supported. The results indicate that technological advancements do not have a statistically significant direct effect on innovation outcomes within the IAIPs.

Hypothesis 2: Institutional support positively influences innovation outcomes.

Path coefficient: 0.233; t-value: 4.002 (greater than 1.96); and p-value: 0.000 (less than 0.05). Result: The path coefficient for institutional support’s effect on innovation outcomes is positive and statistically significant (t-value > 1.96, p-value < 0.05). Thus, Hypothesis 2 is supported. This result suggests that institutional support plays a crucial role in driving innovation outcomes within the IAIPs.

Hypothesis 3: Government policies positively influence innovation outcomes.

Path coefficient: 0.236; t-value: 2.756 (greater than 1.96); and p-value: 0.006 (less than 0.05).

Result: The path coefficient for the effect of government policies on innovation outcomes is positive and statistically significant (t-value > 1.96, p-value < 0.05). Therefore, Hypothesis 3 is supported. The results emphasize the importance of favorable government policies in enhancing innovation within the IAIPs.

Based on the results of the hypothesis tests, the following conclusions can be drawn:

Institutional support is a critical factor that positively influences innovation outcomes within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs), and its effect is statistically significant.

Government policies also have a positive impact on innovation outcomes, emphasizing the importance of supportive policy frameworks in fostering innovation within the IAIPs.

Technological advancements, while theoretically important, do not exhibit a significant direct effect on innovation outcomes in this context. This suggests that factors such as institutional support and policy may play more central roles in driving innovation.

These findings underscore the pivotal role of robust institutional frameworks and supportive government policies in driving innovation within Ethiopia’s IAIPs. Furthermore, they indicate that technological advancements may require more targeted efforts to align with these factors in order to fully optimize innovation outcomes, particularly in terms of enhancing agricultural productivity and resilience.

3.3. Qualitative Insights

3.3.1. Thematic Analysis for Sectoral Innovation, Sustainability, and Climate Resilience

The thematic analysis of data collected from Focus Group Discussions (FGDs) and Key Informant Interviews (KIIs) identified key insights regarding the role of innovation in enhancing sustainability, climate resilience, and technological advancement within Ethiopia’s agro-industrial sector. The analysis revealed that the adoption of innovative approaches within Integrated Agro-Industrial Parks (IAIPs) is crucial for achieving long-term sustainability. Respondents emphasized the significance of continuous innovation and the importance of a “learning by doing” approach. Firms in IAIPs were found to be highly encouraged to adapt to emerging challenges, which is essential for staying competitive in the face of evolving market demands and environmental considerations [55].

The role of innovation in waste minimization and the recycling of byproducts emerged as a major theme. This process, integral to the principles of a circular economy, was identified as a critical driver for both sustainability and economic efficiency within IAIPs. The reduction of environmental impact, while simultaneously creating economic value, was highlighted as a central element of the sustainability agenda in these parks [56]. The analysis further identified that continuous innovation, especially in waste management and the use of byproducts, plays a pivotal role in ensuring the long-term viability of the sector. These findings align with previous studies advocating for the importance of innovation as a key factor in driving both sustainability and competitiveness within industrial contexts [57,58]. The research also highlighted that the adoption of circular economy principles, such as recycling and byproduct management, promotes resource efficiency and bolsters the overall resilience of the agro-industrial system [8,59,60].

The analysis further pointed to the integration of innovative technologies, particularly in waste management and energy efficiency, as vital for promoting sustainable industrial practices. The findings underscored that innovation serves as a key driver of climate resilience and sustainable growth within Agro-Industrial Parks. The respondents noted that support from both government and non-governmental institutions is essential to create a conducive environment for the adoption of such innovations. This support is necessary to ensure the effective integration and long-term sustainability of these technological advancements [55,61].

3.3.2. Role in Fostering Innovation: Opportunities and Challenges

Through an analysis of data collected from Focus Group Discussions (FGDs) and Key Informant Interviews (KIIs), this study also identified a range of opportunities and challenges related to fostering innovation within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs). A central opportunity for enhancing sustainability and driving innovation lies in the transition toward a circular economy. Although the principles of the circular economy are still in the planning stages, emerging initiatives, such as the collaboration between Hawassa University and the Yirgalem IAIP focused on waste management and food safety, are laying the foundation for sustainable practices within the agro-industrial sector in Ethiopia. These early-stage efforts represent a significant step toward adopting sustainable practices [57].

Furthermore, the research highlights the critical roles of government support, stakeholder engagement, and investment in technological innovation as pivotal drivers of progress. A notable opportunity within this context is the symbiotic relationship between foreign investors and local farmers. In this model, foreign firms bring advanced agricultural techniques and provide access to improved resources and technologies, thus facilitating innovation within IAIPs. These collaborative efforts are essential for enhancing productivity, introducing modern technologies, and aligning with sustainability objectives [59].

Despite these promising opportunities, this study also uncovered several obstacles that hinder innovation in IAIPs. A primary challenge relates to the investment environment, which is negatively affected by concerns over peace and security, power shortages, water scarcity, and the seasonality of agricultural production. These factors contribute to disruptions in the production process and supply chain, undermining the overall efficiency and competitiveness of the IAIPs [58].

Another significant challenge is the lack of coordination among key stakeholders, which limits effective collaboration and resource optimization. This fragmentation impedes the successful implementation of innovation strategies and constrains the potential for IAIPs to achieve their sustainability objectives. Moreover, skill gaps within the workforce were identified as another barrier to innovation, underscoring the urgent need for targeted training initiatives to equip the workforce with the necessary skills for modern agro-industrial practices. The inconsistent availability of agricultural inputs further exacerbates these challenges, disrupting production cycles and complicating supply chain management [8,60].

The findings of this study align with existing research on Ethiopia, which frequently cites infrastructure limitations, government support, and stakeholder collaboration as essential elements for driving innovation within the agro-industrial sector [57,61]. Additionally, challenges related to the seasonality of agricultural production and the inconsistent availability of resources are commonly identified as factors that impact the resilience and competitiveness of IAIPs in the study.

The analysis underscores the importance of addressing both the opportunities and challenges identified to ensure the effective integration of innovation in Ethiopia’s IAIPs. Strengthening institutional support, enhancing stakeholder collaboration, and prioritizing workforce skills development are key strategies that will foster innovation and secure long-term sustainability within the parks.

4. Discussions

This study aimed to examine the dynamic relationships between key factors—such as institutional support, networking, policy frameworks, technology, education, and gender—in fostering agricultural innovation within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs). Using Partial Least Squares Structural Equation Modeling (PLS-SEM), this research provides valuable insights into the mechanisms that drive agricultural innovation, productivity, and resilience within the IAIPs, which are pivotal for the nation’s agricultural transformation. The findings significantly contribute to both the theoretical and practical understanding of innovation systems and agricultural policy, particularly within developing country contexts [14,20].

The theoretical foundation of this study is grounded in the Sectoral Innovation Systems (SIS) framework, which provides a comprehensive model for understanding the interactions between key actors—such as government agencies, research institutions, businesses, and local communities—in driving technological and economic development. The SIS framework is particularly relevant to Ethiopia’s agro-industrial sector, as it highlights the importance of innovation, policy alignment, and collaboration in overcoming sectoral challenges, particularly those related to agricultural transformation, climate change adaptation, and modernization [14,20].

Our findings align with the SIS framework, confirming that innovation outcomes within Ethiopia’s IAIPs are driven by the systemic interactions between government policies, institutional support, technological advancements, and stakeholder collaboration. For instance, institutional support (access to financial, cost-effective technology) and networking (infrastructure, cultural resistance) are critical to fostering product innovation (development of new or improved products for agricultural value chains), process innovation (enhancements in agricultural practices), and technological innovation (introduction of new technologies). This is particularly evident in the role of innovation in minimizing waste and promoting the recycling of byproducts, a critical driver for both sustainability and economic efficiency within IAIPs. As highlighted in the qualitative findings, continuous innovation in waste management and byproduct utilization is fundamental for ensuring long-term sectoral viability, a point that resonates with the broader quantitative finding that effective institutional support facilitates the adoption of such innovations.

This study’s findings confirm the central role of institutional frameworks in fostering innovation within Ethiopia’s IAIPs. Institutional support significantly influences innovation (β = 0.233, p < 0.01), aligning with the SIS framework’s emphasis on effective institutions as catalysts for innovation [20,62]. This is supported by qualitative findings, where government support, stakeholder engagement, and investments in technological innovation were recognized as opportunities for fostering innovation. For example, partnerships such as those between Hawassa University and the Yirgalem IAIP, focusing on waste management and food safety, lay the groundwork for sustainable practices and technological advancements that could drive sectoral transformation.

Moreover, our research shows that institutional support not only directly affects innovation but also facilitates the development of networks (β = 0.555, p < 0.01), which are critical for collaboration among stakeholders. These findings reflect the SIS model’s focus on knowledge transfer and resource sharing, with institutional frameworks (skilled labor, infrastructure) providing the infrastructure necessary for such exchanges. The importance of networking is highlighted in both the quantitative and qualitative findings, where the quality of networks, such as those offering access to agro-technology and digital tools, was found to significantly impact innovation outcomes (R² = 0.65). Opportunities for collaboration, particularly in technology adoption and knowledge dissemination, are pivotal for systemic innovation, as demonstrated in the integration of weather forecasting apps and farm management software for climate risk management in Ethiopia’s IAIPs. As Granstrand and Holgersson [19] argue, technological advancements are most impactful when embedded within a broader innovation ecosystem that includes strong institutional support and collaboration.

While the quantitative findings confirm the positive impact of government policies on innovation outcomes (β = 0.236, p = 0.006), the qualitative findings provide further depth to this relationship by discussing how the transition to a circular economy is being supported through collaborative initiatives and government policies. Despite the modest direct effect of policies on innovation outcomes, the qualitative data reveal that supportive policies related to waste management, food safety, and agro-industrial practices provide the necessary regulatory foundation to drive innovation in these sectors [20].

Technological advancements are crucial for agricultural modernization, but the direct effect of technology (access to agro-technology, digital tools) on innovation outcomes was found to be relatively weak (β = 0.098, p = 0.221). This supports Bessant’s [63] argument that technology needs to be embedded within an innovation ecosystem supported by strong institutions (cost-effective technology, skilled labor) and well-aligned policies (supportive policies). The qualitative insights further support this point, with technology adoption identified as a significant factor, yet dependent on institutional frameworks and network support to facilitate its full potential. Similarly, Granstrand and Holgersson [19] emphasize that for technology to achieve its full potential, it must be supported by a well-coordinated innovation ecosystem that includes institutional support, infrastructure, and collaborative networks.

Education, though marginally significant (β = 0.050, p = 0.058), plays a secondary role compared to institutional frameworks and networking. This is consistent with Freeman’s [64] assertion that while education enhances human capital, the immediate drivers of innovation in Ethiopia’s IAIPs are tied to institutional frameworks and collaborative networks. The skill gaps identified in the qualitative findings reinforce this point, emphasizing the need for targeted training programs to equip the workforce with the skills necessary to drive innovation in agro-industrial practices.

The PLS-SEM model fit indices (SRMR = 0.035, RMSEA = 0.058, CFI = 0.975, TLI = 0.973, NFI = 0.970) confirm the robustness of the model, indicating that the data well represent the dynamics of agricultural innovation in Ethiopia’s IAIPs under study. The predictive relevance of the model, with Q² values for innovation, network, and technology, suggests that the model can reliably predict key innovation outcomes, including the development of networks and the adoption of technological innovations.

5. Conclusions

This study explores the key drivers of agricultural innovation within Ethiopia’s Integrated Agro-Industrial Parks (IAIPs), which are essential for modernizing the agricultural sector. By examining institutional support, networking, technological advancements, and policy frameworks, this research provides valuable insights into their role in enhancing agricultural productivity and resilience. The findings are contextualized within the Sectoral Innovation Systems (SIS) framework, highlighting the need for an integrated approach to innovation that involves multiple stakeholders.

The results emphasize the critical roles of institutional support, technological innovation, and stakeholder collaboration in driving innovation within IAIPs. Additionally, the adoption of circular economy principles, such as waste management and byproduct recycling, is crucial for promoting sustainability and economic efficiency. These findings underscore the importance of fostering collaboration among the government, private sector, research institutions, and local communities to create an environment conducive to innovation.

However, the study’s reliance on data from a limited number of IAIPs may not fully capture the diversity of experiences across Ethiopia’s agro-industrial sector. Factors such as cultural attitudes, regional disparities, and technological challenges were not explored and may also influence innovation outcomes.

Based on these findings, we recommend that policymakers and industry stakeholders focus on strengthening institutional frameworks and enhancing cross-sector collaboration. Expanding the research scope to include a wider range of IAIPs and addressing factors such as cultural and regional variations will be crucial for fostering innovation. Promoting sustainable practices, including circular economy principles, should remain a priority to drive innovation and economic growth.

Future research should broaden its scope to incorporate a greater diversity of IAIPs and explore the long-term impacts of innovation systems through an integrated approach combining both quantitative and qualitative methods. This will be essential for advancing agricultural transformation, resilience, and sustainability, both in Ethiopia and globally.