Selecting an Appropriate Mechanism for Developing Iran’s Hazelnut Value Chain: An Analytical Comparison of Expert Opinions and Artificial Intelligence Tools



Extended Abstract

Objective: Agricultural value chains play a central role in enhancing productivity, improving producer incomes, strengthening market access, and increasing the competitiveness of agri-food systems. Despite Iran’s considerable potential for hazelnut production, the sector remains constrained by fragmented production structures, limited processing and storage capacity, weak coordination among value-chain actors, insufficient market integration, and the continued influence of informal intermediaries. These structural and institutional deficiencies restrict value creation, weaken bargaining power, and reduce the efficiency of product flows from production to final markets. Against this background, the present study aimed to identify and prioritize the most appropriate business model for the development of Iran’s hazelnut value chain by comparing four alternative governance configurations: the orchestrator, market maker, layered, and integrated models. The study further examined the extent to which expert-based priorities converged with or diverged from assessments generated through artificial intelligence tools.

Methods: This applied study employed a descriptive-survey design combined with a multi-criteria decision-making framework. Data were collected from 19 experts with specialized experience in hazelnut production, research, policymaking, and agricultural value-chain development. The assessment framework comprised four main criteria—economic, socio-cultural, environmental, and structural—and 21 associated sub-criteria. Expert judgments were elicited through structured pairwise-comparison questionnaires and analyzed using the Grey Analytic Hierarchy Process (GAHP), enabling the incorporation of uncertainty into the weighting and prioritization process. Consistency ratios were calculated to evaluate the internal coherence of expert judgments. To complement the expert-based analysis, a standardized comparative assessment involving eight artificial intelligence tools was conducted using an identical decision structure and prompt format. The four business models were ranked under equivalent decision conditions, and agreement with the expert-derived ranking was examined using Spearman’s rank correlation coefficient, while the stability of repeated rankings was assessed using Kendall’s coefficient of concordance. The reasoning underlying the AI-generated rankings was also coded according to the economic, socio-cultural, environmental, and structural dimensions of the decision framework.

Results: The economic criterion received the highest relative importance (55.87%), followed by the socio-cultural (19.69%), environmental (16.20%), and structural (8.24%) dimensions. Within the economic criterion, profitability (23.30%), chain financing (17.01%), and productivity (16.00%) emerged as the three most influential sub-criteria. The final synthesis of the GAHP results ranked the market maker model first, with an overall weight of 42.87%, followed by the layered (23.97%), integrated (18.51%), and orchestrator (14.61%) models. Decomposition of the final weights showed that the economic dimension made the largest contribution to the superiority of the market maker model, while its relatively strong performance across the socio-cultural, environmental, and structural dimensions reinforced its overall position. The comparative AI analysis revealed substantial heterogeneity across tools; however, the market maker model retained the highest aggregate ranking across the combined AI evaluations. The strongest rank-based agreement with expert judgments was observed for some tools, whereas others consistently favored integrated or orchestrator configurations. The thematic analysis further indicated that coordination among value-chain actors, organization of small-scale producers, processing and standardization, and export and branding capabilities represented the most recurrent areas of convergence between expert reasoning and AI-generated assessments.Results: The economic criterion received the highest relative importance (55.87%), followed by the socio-cultural (19.69%), environmental (16.20%), and structural (8.24%) dimensions. Within the economic criterion, profitability (23.30%), chain financing (17.01%), and productivity (16.00%) emerged as the three most influential sub-criteria. The final synthesis of the GAHP results ranked the market maker model first, with an overall weight of 42.87%, followed by the layered (23.97%), integrated (18.51%), and orchestrator (14.61%) models. Decomposition of the final weights showed that the economic dimension made the largest contribution to the superiority of the market maker model, while its relatively strong performance across the socio-cultural, environmental, and structural dimensions reinforced its overall position. The comparative AI analysis revealed substantial heterogeneity across tools; however, the market maker model retained the highest aggregate ranking across the combined AI evaluations. The strongest rank-based agreement with expert judgments was observed for some tools, whereas others consistently favored integrated or orchestrator configurations. The thematic analysis further indicated that coordination among value-chain actors, organization of small-scale producers, processing and standardization, and export and branding capabilities represented the most recurrent areas of convergence between expert reasoning and AI-generated assessments.