Skip to Content
SustainabilitySustainability
  • Article
  • Open Access

5 February 2026

18 Pages

Research on Supply Chain Performance Evaluation of Geographical Indication Agricultural Products a Case Study of Tea Categories

and
1
School of Management, Jiangsu University, Zhenjiang 212013, China
2
School of Intellectual Property, Jiangsu University, Zhenjiang 212013, China
*
Author to whom correspondence should be addressed.

Abstract

The high brand premium of geographical indication (GI) tea has not been efficiently converted into widespread economic benefits through its supply chain. The current performance evaluation system is confronted with a dual predicament: first, the strong external environment (such as policy support and industrial agglomeration) interference is hard to isolate, making it impossible to distinguish between “environmental advantages” and “true management levels”; second, the general agricultural indicators fail to capture the output essence of GIs centered on “brand value”. Therefore, this study constructs an evaluation framework integrating methodological and indicator innovations. Methodologically, a three-stage DEA model is adopted to eliminate the influence of exogenous environments and random noises, precisely measuring the “pure management efficiency” of the supply chain. Indicatively, common variables are abandoned, and a customized system is established with logistics facilities, production area, and regional digital investment as inputs, and brand reputation, value, and income as outputs. Based on the panel data of twelve representative tea GIs from 2021 to 2024, the study finds that the following: (1) The “pure management efficiency” of the supply chain is the key factor influencing performance evaluation. (2) “Diseconomies of scale” are the main structural bottleneck restricting performance improvement rather than technological backwardness. (3) Solving the above-mentioned management efficiency problems, especially resolving “diseconomies of scale”, is the micro foundation for achieving sustainable industrial development. This research not only provides methodological support and empirical evidence for the refined management and sustainable development of the geographical indication agricultural product supply chain, but also has significant practical significance for promoting the quality and efficiency improvement of the tea industry and facilitating the sustainable development of related agriculture.

1. Introduction

Geographical indication products of Chinese tea, such as West Lake Longjing and Yunnan Pu’er tea, are the crystallization of natural endowments and human craftsmanship, carrying extremely high brand value and market expectations. However, their industrial development faces a common predicament: the high brand reputation has not been efficiently and stably converted into actual benefits for all participants in the supply chain through the supply chain system. Problems such as “emphasizing certification over management”, insufficient supply chain collaboration, and imbalance between scale and quality highlight the blockage in the value realization mechanism.
To accurately diagnose and unblock this blockage, it is first necessary to effectively evaluate the supply chain performance. However, the existing evaluation system faces fundamental challenges in this regard. This is due to the dual characteristics of geographical indication products: their value is rooted in unique geographical environments and also depends on human management and collaboration. Therefore, the supply chain performance of geographical indication products is essentially the efficiency of brand value creation and transmission, rather than a simple extension of traditional agricultural production efficiency. Using general evaluation indicators for ordinary agricultural products (such as yield per mu and fertilizer input) will seriously mislead the judgment, as these indicators cannot reflect the core output of brand premium at all. At the same time, strong external environmental factors (such as local government’s special support and regional industrial clusters) will systematically mask or distort the true performance, making it difficult for traditional methods to answer a key question: To what extent does the performance of a geographical indication brand result from its own management capabilities and to what extent does it benefit or suffer from the external environment?
To overcome these difficulties, this study has made dual innovations in methodology and indicator systems. Methodologically, the article introduces a three-stage DEA model, whose core advantage lies in its ability to strictly quantify and eliminate the interference of environmental factors and random noise through Stochastic Frontier Analysis, thereby achieving an accurate measurement of the “pure management efficiency” of the supply chain. In terms of indicator construction, this paper has made a disruptive design: the input end focuses on the core elements crucial to brand value construction (such as logistics infrastructure for quality assurance, area of origin reflecting scarcity, and the regional digital investment for characterizing the efficiency of information synergy); the output end directly uses brand reputation, brand value, and brand income as the ultimate measures, directly targeting the market-oriented goals of geographical indication strategies.
Based on this innovative framework, this study aims to systematically answer three progressive core questions:
After eliminating environmental advantages and disadvantages, what is the true (management) performance level of the supply chains of major Chinese tea geographical indications?
Is the key bottleneck restricting their performance “pure technical efficiency” insufficiency or “scale efficiency” diseconomy?
For different types of efficiency, how should differentiated optimization paths be formulated?
By “tailoring” evaluation scales for geographical indication products and “purifying” the evaluation environment, this study aims to go beyond traditional analysis and provide a scientific diagnostic tool. Its conclusions can not only provide a new perspective for the academic community to understand the value realization mechanism of geographical indications but also offer a direct and reliable decision-making basis for the government to implement precise industrial policies and for industry players to identify key shortcomings, thereby promoting geographical indication tea brands from “fame” to “profit” and achieving sustainable upgrading of the value chain.
In the process of writing this article, a structural framework of “from macro to micro and then back to macro” was adopted to systematically review the relevant literature. Firstly, the general characteristics and unique attributes of the supply chain of geographical indication agricultural products were defined. Secondly, the special challenges and research status of the tea supply chain were analyzed. Then, the evolution of the theory and methods of supply chain performance evaluation was systematically reviewed. Finally, the application value and implications of the three-stage DEA model in the agricultural field were deeply explored, thus constructing a complete logical chain from “problem identification” to “method support”. The literature search was based on core keywords such as “geographical indication agricultural products”, “tea supply chain”, “supply chain performance evaluation”, “three-stage DEA”, “blockchain”, “digital traceability”, and “sustainable development”. The screening criteria focused on the relevance of the literature to the theme, the authority of the source (core journals and cutting-edge achievements), and the timeliness of publication (balancing classic research and the latest progress). Cross-disciplinary experiences from other supply chain fields such as fruits and medicine were also referred to, to ensure that the review is comprehensive, in-depth, and forward-looking.
This study selects representative leading geographical indication products of tea, such as West Lake Longjing and Anji White Tea, as case studies, focusing on the relevant data from 2021 to 2024. The data mainly come from official statistical yearbooks released by the government, industry reports, and annual research reports from authoritative institutions like the China Agricultural Brand Research Center, aiming to ensure the reliability and timeliness of the empirical analysis.

2. Literature Review

2.1. The Particularity and Research Status of Geographical Indication (GI) Agricultural Product Supply Chains

GI agricultural product supply chains are distinctly differentiated from ordinary agricultural product supply chains by their unique “geographical indication” attribute. Their core characteristics lie in regional exclusivity, quality uniqueness, and multi-stakeholder diversity, which impose higher demands on supply chain origin management, quality control, and collaborative governance [1,2]. Studies have shown that their healthy development depends on effective interest linkage and collaboration mechanisms, requiring the realization of holistic empowerment through multi-agent collaboration such as “government leadership + regional micro-autonomous organization co-governance” and “traditional tea farmers + new business entities” [3]. “Chain leader” enterprises can construct a win–win symbiotic system by extending industrial chains, optimizing supply chains, enhancing value chains, and linking interest chains [4,5]. Multi-source data fusion technology can improve the accuracy of tea quality detection and further ensure tea product safety [6]. Meanwhile, digital traceability technologies such as blockchain, by addressing information asymmetry, have emerged as effective governance tools for preventing “free-riding” behavior and maintaining brand credibility [7,8]. The application experience of blockchain in fresh fruit supply chains can also facilitate digital traceability in tea supply chains and resolve the issue of information tampering in origin management and quality control [9].
For tea-specific GI agricultural products, their supply chains additionally face special challenges including high preservation requirements, strong traditional craftsmanship, and high brand cultural added value. Tea supply chains can draw on the evaluation experience of the Jiangyin fruit and vegetable supply chain to optimize circulation link control and supervision [10]. The deep integration of the digital economy is driving their transformation toward intelligence; big data, the Internet of Things (IoT), and other technologies provide new pathways for precise positioning of tea brands, intelligent supply chain management, and market expansion [11,12]. The application of fruit-friendly environmental coating technology in tea preservation can be explored to help solve tea preservation problems and enhance supply chain sustainability [13]. Cold chain intelligent detection technology suitable for gel polysaccharide-based fresh products can also be applied to tea cold chain quality control to achieve real-time monitoring of tea quality [14]. However, their transformation also faces practical difficulties such as high technical barriers and lagging standardization [12]. Especially in the current context of accelerating the construction of a unified national market, efforts should be made to break down local protectionism and market segmentation [15].
International research on the supply chain management of geographical indications has developed a multi-dimensional perspective covering legal protection models, governance mechanisms, and technological applications. Globally, three major management systems have emerged, represented by France (focusing on appellation of origin and industry public rights management), the United States (relying on trademark law and market self-regulation), and Japan (a hybrid model). The differences among them reflect distinct intellectual property protection concepts and industrial governance logics [16]. In recent years, the academic community has increasingly focused on the global governance and sustainable development of GI supply chains. For instance, in 2022, the Food and Agriculture Organization of the United Nations (FAO) and the International Center for Agricultural Research in the Dry Areas (CIRAD) jointly convened over 200 scholars, policymakers, and practitioners from around the world to discuss the latest advancements in GIs in areas such as public and private management, market forces, sustainable development, and intellectual property rights. The outcomes of this event were compiled into the book A Global Perspective on Geographical Indications, which was published in 2025 and systematically reviewed the cutting-edge trends in global GI research [17].

2.2. Literature Review on the Theory and Methodology Evolution of Supply Chain Performance Evaluation

The theoretical and methodological system of supply chain performance evaluation has undergone a notable evolutionary process, manifesting primarily in three dimensions: the expansion of evaluation dimensions, the innovation of evaluation methods, and the deepening of research on supply chain typologies.
Evolution of Evaluation Dimensions: Research has trended toward shifting from traditional financial indicators to a multi-dimensional integrated framework. For instance, green supply chain performance evaluation in the pharmaceutical industry has evolved into a comprehensive system encompassing drug greenness, green marketing, operational processes, quality and safety, and economic efficiency [18]. In agricultural product supply chains, collaborative governance evaluation can incorporate the impact of grassroots leaders on cross-stakeholder collaboration efficiency [19]. Agricultural product supply chain indicator systems need to integrate dimensions such as production, business processes, logistics and consumer engagement, plot characteristics, and the alignment between regional industrial structure and supply chain “greenness” [20,21,22,23]. Additionally, business process evaluation for agricultural supply chains can reference factors influencing farmers’ vertical cooperation to assess inter-entity collaboration efficiency [24]. Extending the traditional balanced scorecard (BSC), some studies have developed a six-dimensional performance evaluation framework for agricultural product supply chains [25].
Innovation of Evaluation Methods: Methodological advancements have transitioned from single-method approaches to hybrid modeling. Early research relied on subjective weighting methods such as the Analytic Hierarchy Process (AHP) and Fuzzy Comprehensive Evaluation (FCE) [26]. Subsequent developments include integrated models like the PCA-DEA-TOPSIS three-stage framework [21] and hybrid models combining gray clustering with FCE [25]. Modern models increasingly incorporate digital transformation indicators to more accurately reflect supply chain efficiency in the digital era [27]. Concurrently, intelligent algorithms such as neural networks have gained widespread application [28,29,30].
Deepening of Typology-Specific Research: Evaluation methods for specialized supply chains exhibit high specificity and target orientation. For green supply chains, efficiency evaluation commonly employs the three-stage DEA method (Liao et al., 2022 [31]) or combinations of DEA with principal component analysis (PCA) (Fang et al., 2017 [32]; Cao et al., 2017 [33]). For new retail enterprises, performance evaluation requires adapting the BSC to model-specific characteristics [34]. Logistics service supply chains, by contrast, utilize the extension evaluation method based on BSC [35].

2.3. Application and Implications of the Three-Stage DEA Model in Agriculture and Related Fields

The three-stage Data Envelopment Analysis (DEA) model has been widely applied in efficiency evaluations within agriculture and related fields due to its ability to effectively eliminate the interference of environmental factors and random errors. In research on agricultural production efficiency, this method has been extensively used for evaluating regional agricultural efficiency [36,37] and for studying agricultural land use efficiency and agricultural land ecological efficiency [38]. Results indicate that scale efficiency is a key factor constraining comprehensive efficiency. When applying the three-stage DEA to evaluate agricultural production efficiency, factors such as farmers’ willingness to purchase crop insurance, the application level of agricultural machinery coordination technology, and institutional pressure can be incorporated as environmental variables [39,40,41].
In the field of cold chain logistics, research has found significant regional disparities in the efficiency of cold chain logistics for fresh agricultural products, with environmental factors exerting a substantial influence on efficiency [42,43,44]. Regarding policy evaluation, studies have employed the three-stage DEA model combined with the Tobit model to analyze the impact of the minimum grain purchase price policy on the efficiency of the grain processing industry [45] and the effect of socialized agricultural machinery services on farmland production efficiency [46]. In three-stage DEA policy assessments, indicators related to climate change can be included to isolate their interference with industrial efficiency [47]. In enterprise efficiency research, this method has been applied to evaluate the financing efficiency of cross-border e-commerce enterprises involved in agriculture [48] and to analyze the operational efficiency of vertical fresh food e-commerce platforms [49].
Furthermore, in the domain of agricultural product circulation, research has revealed the significant impact of environmental factors on circulation efficiency [50]. When using the three-stage DEA model to evaluate agricultural product circulation efficiency, characteristics of regional trade networks can be treated as environmental variables [51].
These studies provide important methodological insights for this paper: the three-stage DEA model can effectively address the problem of environmental factor interference in traditional efficiency evaluations, making it particularly suitable for complex systems like the geographical indication (GI) agricultural product supply chain; its combined use with the Tobit model offers a powerful tool for analyzing factors influencing efficiency; and existing research provides significant reference points for this study in terms of selecting environmental variables and constructing the model.

3. Indicator Selection and Data Sources

3.1. Performance Evaluation Indicator System for General Agricultural Product Supply Chains

Through a systematic review and synthesis of 26 scholarly articles focused on agricultural product supply chain performance evaluation, six high-frequency dimensions were identified, encompassing a total of 12 general performance evaluation indicators for agricultural product supply chains (Table 1).
Table 1. Performance evaluation indicators for general agricultural product supply chains.

3.2. Performance Evaluation Index System for Geographical Indication (GI) Agricultural Product Supply Chains

The selection of indicators for constructing a performance evaluation system for GI agricultural product supply chains must profoundly embody the unique value connotations and industrial attributes inherent to such products. Unlike the generic frameworks commonly applied to ordinary agricultural product evaluation—which prioritize universal agricultural production elements (e.g., basic natural resources, production materials, and technology–human inputs) and focus outputs on directly quantifiable material dimensions (economic yield, product quality, and ecological benefits)—the core competitiveness of GI tea products does not derive from general agricultural production efficiency. Instead, their value is rooted in specific regional, cultural, and historical heritage [52] and ultimately materialized through market brand value. Thus, the core of their performance evaluation lies in measuring the efficiency of converting regional uniqueness into brand premium and market reputation [53].
Based on the above considerations, the indicator system of this research was specifically constructed. In terms of input indicators, the common production material input indicators such as chemical fertilizers and pesticides were abandoned, and instead, the core elements crucial to the formation of brand value were emphasized, such as logistics infrastructure that guarantees quality and timeliness, the area of origin that reflects the scarcity of raw materials and scale effects, and the regional digital investment for characterizing the efficiency of information synergy. In terms of output indicators, the traditional economic output and quality inspection indicators were abandoned, and the three dimensions of brand reputation, brand value, and brand income were directly adopted to accurately capture the market recognition and excess profits that the geographical indication strategy ultimately pursues. The setting of environmental variables also focuses on macro governance factors such as industrial policies and cluster effects, aiming to strip away the objective impact of the external environment on management efficiency (Table 2).
Table 2. Performance evaluation indicator system and data sources for geographical indication (GI) agricultural product supply chains.

4. Research Methodology

The classic DEA model only considers the relationship between input and output indicators, and fails to eliminate the influence of external environmental factors and random errors on the efficiency value. The three-stage DEA method integrates the advantages of classic DEA and SFA, which can not only better measure the innovation efficiency under multiple inputs and multiple outputs, but also eliminate the influence of external environmental factors and random errors on innovation efficiency, making the obtained efficiency value more accurately reflect the innovation level of enterprises. This article conducts a three-stage DEA operation using DEAP 2.1 and Frontier 4.1.
Stage 1: Given that this study focuses on the ability of the supply chain to maximize brand value with given inputs, the output-oriented BCC model is adopted in the first stage to calculate the initial efficiency of each DMU. This model considers variable returns to scale (VRS), and its linear programming form is as follows:
max   ϕ s . t .   ∑ j = 1 n λ j x i j ≤ x i 0 ,   i = 1 , … , n ∑ j = 1 n λ j y r j ≥ ϕ y r 0 ,   r = 1 , … , s ∑ j = 1 n λ j = 1 ,   λ j ≥ 0 ,   j = 1 , … , n
Here, x i j and y i j , respectively, represent the i-th input and the r-th output of the j -th DMU; λ j is the weight coefficient; ϕ is the output efficiency value, satisfying ϕ ≥ 1 , and its reciprocal 1 / ϕ is the traditional comprehensive technical efficiency value (ranging between 0 and 1). The efficiency values calculated in this stage are mixed with management inefficiency, favorable environmental factors, and random noise.
To correct the above deviations, in the second stage, the impact of environmental variables on efficiency is quantified through regression models (such as Stochastic Frontier Analysis, SFA), and random errors are eliminated. Specifically, for the slack variables calculated in the first stage, a regression equation is constructed:
s i j = f ( z j ; β i ) + v i j + u i j ,   v i j ∼ N ( 0 , σ v i 2 ) ,   u i j ∼ N + ( 0 , σ u i 2 )
In the formula, f ( z j ; β i ) represents the influence of environmental factors on the slack value (usually set as a linear form β i z j ); v ^ i j is the random error term, capturing statistical noise; u_ij represents managerial inefficiency. After separating the estimated environmental influence factor f ( z j ; β i ) and the estimated random noise v ^ i j through maximum likelihood estimation, all DMUs are placed under the same, most unfavorable external environment and luck conditions:
x i j * = x i j + max j { f ^ ( z j ; β ^ i ) } − f ^ ( z j ; β ^ i ) + max j { v ^ i j } − v ^ i j
This adjustment “costs” both the environmental advantage (manifested as a smaller f ^ ( z j ; β ^ i ) ) and good luck (manifested as a negative v ^ i j ) as additional virtual inputs, thereby obtaining the purified input data x i j * , which only reflects the input level controllable by management.
Substitute the adjusted input data x i j * from the second stage and the original output data y rj into the output-oriented BCC model described in the first stage for calculation again:
max   ϕ * s . t .   ∑ j = 1 n λ j x i j * ≤ x i 0 * ,   i = 1 , … , m ∑ j = 1 n λ j y r j ≥ ϕ * y r 0 ,   r = 1 , … , s ∑ j = 1 n λ j = 1 ,   λ j ≥ 0

5. Data Analysis and Results Discussion

5.1. Stage 1 DEA Results Analysis

The initial Stage 1 DEA incorporated both environmental variables and random disturbances. The DEA model was constructed using the input–output data mentioned earlier, processed with Deap 2.1 software. The results are presented in the table below.
Table 3 and Table 4 present the unadjusted initial DEA results, reflecting significant differentiation and dynamic evolution in the supply chain performance of various brands. Among them, comprehensive efficiency represents the comprehensive technical efficiency: it measures the overall ability of a decision-making unit to minimize input or maximize output under the current technological level. Pure technical efficiency represents the pure efficiency of a decision-making unit in terms of resource utilization at the management and technical levels after stripping out the scale factor. Scale efficiency represents the proximity of the actual scale of a decision-making unit to the optimal production scale, which is obtained by the ratio of comprehensive technical efficiency to pure technical efficiency. Fuding White Tea and others have maintained “full efficiency” for four consecutive years, setting a high-efficiency benchmark, while Xinyang Maojian and Wuyi Dahongpao have efficiency values consistently below 0.9, indicating that there is still room for continuous improvement in their supply chains.
Table 3. Stage 1 DEA results (2021–2022).
Table 4. Stage 1 DEA results (2023–2024).
The composition of efficiency reveals key bottlenecks: most low-efficiency products (such as Lu’an Guapian) are simultaneously constrained by technical inefficiency and scale inefficiency. A typical phenomenon is that Lu’an Guapian achieved the best efficiency in 2021 but dropped to 0.823 the following year and has been in a state of decreasing returns to scale for three consecutive years. This suggests that its scale expansion may have exceeded its management capacity and market carrying capacity, leading to resource redundancy and efficiency decline.
Dynamic trends highlight management challenges: Anji White Tea and Dongting Biluochun, among others, have reached the efficient frontier through optimization in 2023–2024, demonstrating the effectiveness of management improvements. However, the number of overall efficient units has significantly increased in the later period, which may imply a “slack” in the frontier. This raises the question: Is this a general improvement in efficiency or is it due to overly strong benchmarks or environmental factors? This precisely underscores the necessity of conducting a subsequent second stage to eliminate the influence of environmental and random noise, in order to reveal the true “pure management efficiency”.

5.2. Analysis of Second-Stage SFA Results

In the second stage, based on the input slack variables identified in the first stage, Stochastic Frontier Analysis (SFA) was employed to separate environmental factors from random noise. Four input slack regression models were constructed, and the results obtained are as follows:
To eliminate external influences, we conducted an SFA regression analysis on the input slack variables of the first stage (Table 5). The analysis based on the SFA regression results of the second stage indicates that environmental factors have a significant and complex impact on the input allocation of the geographical indication tea supply chain. The γ values of the three input equations all exceed 0.9 (0.91, 0.94, and 0.95, respectively), and the LR tests are all significant. This statistically confirms that managerial inefficiency is the main cause of input redundancy and also validates the necessity and rationality of using a three-stage DEA model to eliminate the interference of environmental and random noise. Specifically, environmental variables show heterogeneous influence directions: for instance, environmental variable 1 (such as industrial agglomeration) has positive coefficients for input 1 and input 2 (496.88 and 28,605.83, respectively), suggesting that its development may increase the redundancy or demand for these resources; however, it shows a saving effect for input 3 (coefficient of −0.53). In contrast, environmental variable 2 (such as policy intensity) has a negative impact on input 1 (coefficient of −88.12), indicating its potential to promote resource conservation. Additionally, the degree of random shock faced by different inputs varies significantly, with the σ2 value of input 2 being extremely high, indicating that it is most disturbed by measurement errors or accidental factors. These findings collectively suggest that without identifying and eliminating the unequal impact of environmental endowments through this stage model, the initial efficiency assessment will be mixed with some exogenous disturbances, thus failing to accurately reveal the true management level of the supply chain. Therefore, this result provides a crucial statistical basis for subsequent input adjustment and the final calculation of pure “management efficiency”.
Table 5. Stage 2 slack regression results.

5.3. Stage 3 DEA Results Analysis

Based on the quantitative evidence from the second-stage SFA regression, the third stage of this study aims to “equalize” the initial inputs to eliminate the interference of environmental advantages and random noise. Specifically, this study will moderately increase the benchmark values of input 1 (such as logistics facilities) and input 2 (such as production area) according to the model estimates to remove the natural conveniences brought by location agglomeration. Conversely, for units where policy support (environmental variable 2) significantly promotes resource conservation (such as Fuding White Tea that received strong support), the benchmark value of input 1 will be correspondingly reduced to compensate for the “false” efficiency shown due to external policy support.
At the same time, the estimated random noise residuals will be used to correct abnormal data caused by accidental fluctuations. The purified input data after this adjustment will be re-substituted into the BCC model to ultimately calculate the “pure technical efficiency” that only reflects internal management levels, thereby providing a reliable basis for identifying real management shortcomings (such as the decreasing returns to scale problem of Lu’an Guapian) and formulating precise improvement policies (such as optimizing the resource allocation mechanism in highly agglomerated areas and strengthening targeted support for areas with low policy support).
Using the “clean” data adjusted in the second stage to reconduct the DEA, we obtained the results reflecting pure management efficiency (Table 6 and Table 7). Compared with the first stage, the main findings became clearer and more instructive for management:
Table 6. Stage 3 DEA results (2021–2022).
Table 7. Stage 3 DEA results (2023–2024).
The analysis revealed that the interference of the external environment on the initial assessment was made explicit. For instance, the outstanding performance (efficiency value of 1) of Anji White Tea and Dafo Longjing in the first stage of 2021 was found to be partially attributed to environmental advantages. After adjustment, their pure technical efficiency dropped to 0.874 and 0.904, respectively, highlighting the real room for improvement in management. Meanwhile, Fuding White Tea and West Lake Longjing maintained “dual-stage effectiveness” in multiple years, demonstrating that their superior efficiency is based on solid internal management rather than merely relying on external benefits.
More importantly, the adjusted data precisely identified the core shortcomings in supply chain management. Lu’an Guapian is a typical example: its overall efficiency remained persistently low, and in 2023–2024, it showed a structural imbalance with high pure technical efficiency (0.972, 0.938) but low-scale efficiency (0.796, 0.806). Combined with its continuous state of decreasing returns to scale, this clearly indicates that its supply chain issues are not due to technological backwardness but rather that its scale expansion has far exceeded its organizational management and resource coordination capabilities, leading to severe diseconomies of scale. Similarly, the generally low efficiency of products like Duyun Maojian and Anhua Dark Tea after adjustment reflects persistent real management inefficiency.
Looking longitudinally, only a few products could consistently maintain high efficiency. The efficiency of most products fluctuated significantly after stripping away environmental factors and did not show a stable improvement trend, confirming that enhancing internal management efficiency is more challenging than relying on external conditions. This provides a precise basis for subsequent targeted policy interventions, such as optimizing the scale structure of over-expanded production areas and strengthening process management in low-efficiency areas.

5.4. Robustness Check

To ensure the reliability of the research conclusions and avoid the impact on the robustness of efficiency estimation due to a small sample size, this study follows the classic method of Wilson (2000 [54]) and employs the Bootstrap–DEA technique for robustness testing. This method simulates the scenario of repeatedly drawing different samples from the same population through repeated sampling with replacement (this study sets B = 2000 times), thereby directly quantifying the bias in efficiency estimation and statistical uncertainty caused by the limited sample size. Specifically, this study conducts Bootstrap inference on the “pure management efficiency” values obtained after the third stage adjustment, aiming to assess whether the main conclusions remain statistically significant when considering sampling variability. The results of this test will be systematically presented below through the corrected efficiency values, confidence intervals, and summary statistics (Table 8 and Table 9).
Table 8. Comparison of efficiency values before and after correction and confidence intervals (2023–2024).
Table 9. Summary statistics of the overall efficiency of DMU correction.
Based on the robustness test results of Bootstrap–DEA, this study reaches the following conclusions: Firstly, the test confirmed that under a limited sample size, the traditional DEA estimation does indeed have a systematic “small sample overestimation” bias, as evidenced by the decline in the average overall efficiency after correction, and all decision-making units previously judged as completely efficient (efficiency value = 1) were adjusted to within the efficient frontier. This validates the necessity of statistical correction. However, more crucially, the correction did not undermine the empirical basis of the core findings of this study: for all decision-making units identified as having significant inefficiencies in the main text (such as Lu’an Guapian, Duyun Maojian, etc.), the upper bound of their Bootstrap confidence intervals was significantly lower than 1. This result provides solid statistical significance support for the assertion that “there is significant inefficiency in their supply chain management.” At the same time, the high consistency in efficiency rankings before and after correction indicates that the conclusions regarding the relative performance differences among various geographical indication products are robust. In summary, although the limited sample size has led to estimation biases in absolute efficiency levels, the main conclusions of this study regarding the pattern of management efficiency differences and the identification of key weaknesses are statistically robust.

6. Conclusions and Recommendations

This study employed a three-stage DEA model to conduct a four-year empirical analysis of the supply chain performance of 12 Chinese geographical indication tea products. By successively stripping away the interference of environmental factors such as industrial agglomeration and policy support, as well as random noise, the research precisely measured the “pure management efficiency” reflecting internal management levels. The empirical results showed that after eliminating environmental advantages, the efficiency evaluations of most products changed significantly, confirming the systematic impact of external conditions on performance assessment. For instance, the efficiency values of products like Anji White Tea dropped after adjustment, indicating that their previous high efficiency was partly due to environmental dividends, while products like Lu’an Guapian exposed persistent low efficiency issues, revealing the true existence of management deficiencies. This method effectively distinguished between “environmental advantages” and “management efforts”, providing a pure perspective for performance diagnosis.
The adjusted data precisely identified the key management bottlenecks within the supply chain, with structural characteristics being particularly prominent. For example, Lu’an Guapian consistently showed a significant state of decreasing returns to scale during the observation period, and in 2023–2024, its relatively high pure technical efficiency was in sharp contrast with an extremely low scale efficiency of less than 0.81. This clearly indicated that the core issue of its supply chain was not backward production technology, but that its scale expansion had far exceeded its organizational coordination and resource integration capabilities, leading to significant diseconomies of scale. Meanwhile, products like Xinyang Maojian and Duyun Maojian had long-term low comprehensive efficiency after adjustment, also pointing to widespread deficiencies in technology application and process management. These findings collectively suggest that the key blockage preventing brand premiums from converting into economic benefits lies in the insufficient management efficiency within the supply chain.
This study further reveals that the performance disparity of supply chains is closely related to the structural characteristics of their development models. Government-led models (such as Lu’an Guapian and Duyun Maojian) typically rely on policy and agglomeration advantages. Their efficiency values in the first stage are still acceptable (for instance, the comprehensive efficiency of Lu’an Guapian was 1 in 2021), but after excluding environmental factors, the “pure management efficiency” drops significantly (the comprehensive efficiency was only 0.756 in 2024), and they remain in a state of decreasing returns to scale (with scale efficiency as low as 0.806), highlighting the “scale diseconomy” predicament—that is, externally driven scale expansion exceeds internal management capabilities. In contrast, market-driven models (such as Fuding White Tea and West Lake Longjing) focus more on brand and operation, maintaining high efficiency even after environmental adjustments (with pure technical efficiency of 1 in multiple years for both), demonstrating stronger internal management resilience. However, this model may also be constrained in scale expansion due to insufficient synergy (for instance, the scale efficiency of some brands has not reached the optimal level). It is evident that the essence of efficiency differences lies in that “external empowerment” easily leads to a disconnection between scale and capability, while “endogenous management” is more conducive to maintaining technical efficiency. Therefore, policies should be differentiated: for the former, the focus should shift from “expanding scale” to “improving efficiency”, and for the latter, regional coordination and facility sharing should be strengthened to achieve a balanced development of scale and management.
The low “pure management efficiency” revealed by the study, especially structural contradictions such as “diseconomies of scale”, directly undermines the triple bottom line of sustainable development in the industry: economically, it leads to the dissipation of brand premiums and unstable incomes for tea farmers; socially, it weakens the foundation for regional collaboration and equitable development; environmentally, it implies the pressure of resource misallocation and extensive utilization. Therefore, improving supply chain performance and promoting sustainable development are essentially the same process. Future policy interventions must shift from general support to precisely enhancing “management efficiency”, by resolving diseconomies of scale and popularizing best management practices, to simultaneously achieve economic resilience, social equity, and ecological sustainability. This is precisely the core essence of sustainable development in the geographical indication industry.

7. Research Limitations and Prospects

It should be noted that the conclusions of this study should be examined within the following limitations: Firstly, due to the total quantity of geographical indication products, although the sample size (N = 48) has been corrected by statistical methods, the conclusions will be more universal if more product categories or a longer time series can be included in the future. Secondly, the inherent piecewise linear frontier assumption of the DEA model may not fully capture the complex nonlinear relationship between inputs and brand value output in the tea supply chain. Finally, for variables such as “regional digital input”, although the use of proxy indicators is a common practice, it is still difficult to completely avoid measurement errors. These limitations also point out the direction for future research. For example, the use of non-parametric frontiers or the integration of more detailed survey data can further deepen the understanding of the performance of geographical indication supply chains.

Author Contributions

Conceptualization, G.Z. and H.W.; Methodology, G.Z. and H.W.; Software, G.Z. and H.W.; Validation, G.Z. and H.W.; Formal analysis, G.Z. and H.W.; Investigation, G.Z. and H.W.; Resources, G.Z. and H.W.; Data curation, G.Z. and H.W.; Writing—original draft, G.Z. and H.W.; Writing—review & editing, G.Z. and H.W.; Visualization, G.Z. and H.W.; Supervision, G.Z.; Project administration, G.Z.; Funding acquisition, G.Z. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by The National Social Science Fund of China under Grant No.24FGLB016: “Research on the Derivation Mechanism and Brand Governance of Geographical Indications for Agricultural Products” and Jiangsu University Undergraduate Research Project Funding Program under Grant No. Y23C061: “Research on the Multi-Subject Synergistic Maintenance Mechanism of Agricultural Product Geographical Indications from the Perspective of Value Co-creation”.

Institutional Review Board Statement

Not applicable.

Data Availability Statement

The data presented in this study are openly available in [https://tjj.luan.gov.cn/public/6608391/10686480.html (accessed on 1 July 2025)].

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Yang, P.; Zhou, Y. Development Dilemmas and Optimization Strategies for Agricultural Product Geographical Indications. Gansu Soc. Sci. 2016, 3, 234–238. [Google Scholar]
  2. Quan, S.; Zhang, H.; Wei, H. The Chinese Exploration of Geographical Indication Protection and Management: Based on the Evolution and Innovation of Two Discourse Systems. J. South China Norm. Univ. (Soc. Sci. Ed.) 2025, 4, 19–33+205. [Google Scholar]
  3. Dang, G.; Guo, Y.; Zhang, L. Mechanisms and Paths of Geographical Indications Promoting the Construction of a Beautiful and Harmonious Countryside in Ethnic Mountainous Areas: An Analysis from the Perspective of New Endogenous Development. Chin. Rural Econ. 2024, 7, 96–118. [Google Scholar]
  4. Liu, X.; Zhou, W.; Sun, Y. Research on fair trade organization’s participation in supply chain poverty alleviation from the perspective of inclusive growth: Take the tea development of Guzhang county Pancao village as an example. J. Hunan Agric. Univ. (Soc. Sci.) 2021, 22, 18–26. [Google Scholar]
  5. Cai, J.; Xie, Y.; Li, W.; Xia, L. Research on the Mechanism of Specialized Agricultural “Chain Master” Enterprises Connecting and Empowering Farmers from the Perspective of Symbiosis. World Agric. 2025, 9, 104–115. [Google Scholar]
  6. Adade, S.Y.-S.S.; Lin, H.; Johnson, N.A.N.; Nunekpeku, X.; Aheto, J.H.; Ekumah, J.-N.; Kwadzokpui, B.A.; Teye, E.; Ahmad, W.; Chen, Q. Advanced food contaminant detection through multi-source data fusion: Strategies, applications, and future perspectives. Trends Food Sci. Technol. 2024, 156, 104851. [Google Scholar] [CrossRef] [Scilit]
  7. Dong, Y.; Qian, W. The Free-Rider Problem in the Construction of Region Public Brands of Agricultural Products: An Analysis Based on Governance Study of Digital Traceability, Transparency and Assurances Systems. China Rural Surv. 2022, 6, 142–162. [Google Scholar]
  8. Xing, B.; Yu, H.; Xu, D.; Luo, N.; Sun, C. Development and application of traceability system for black tea based on blockchain. J. Chin. Agric. Mech. 2022, 43, 133–138. [Google Scholar]
  9. Zhang, Y.; Chen, L.; Battino, M.; Farag, M.A.; Xiao, J.; Simal-Gandara, J.; Gao, H.; Jiang, W. Blockchain: An emerging novel technology to upgrade the current fresh fruit supply chain. Trends Food Sci. Technol. 2022, 124, 1–12. [Google Scholar] [CrossRef] [Scilit]
  10. Liu, B.; Xu, J.; Shi, G.; Huang, X. Evaluation and countermeasures for the secure supply of fruit and vegetable products in jiangyin, China. Agro Food Ind. Hi-Tech 2016, 27, 22–27. [Google Scholar]
  11. Zhang, X. Supply Chain Optimization Strategies for Zhejiang’s Tea Industry in the Context of the Digital Economy. Food Res. Dev. 2022, 43, 227–228. [Google Scholar]
  12. Zhang, X.; Bao, J.; Xu, S. Exploring Pathways for Agricultural Brand Development in China within the Context of the Digital Economy. Chin. J. Agric. Resour. Reg. Plan. 2025, 1–12. Available online: https://link.cnki.net/urlid/11.3513.S.20250313.1334.006 (accessed on 12 December 2025).
  13. Ammar, E.E.; Zou, X.; Ghosh, S.; Onyeaka, H.; Elmasry, S.A.; Alkeay, A.M.; Al-Farga, A.; Rady, H.A.; El-Shershaby, N.A.; Sallam, A.S. Fresh Futures: Cutting-Edge Eco-Friendly Coating Techniques for Fruits. J. Food Process. Preserv. 2025, 2025, 5201632. [Google Scholar] [CrossRef] [Scilit]
  14. Lu, S.; Zhang, M.; Xu, B.; Guo, Z. Intelligent quality control of gelatinous polysaccharide-based fresh products during cold chain logistics: A review. Food Biosci. 2024, 62, 105081. [Google Scholar] [CrossRef] [Scilit]
  15. Zhong, F. Coordinating Centralized and Decentralized Decision-Making to Accelerate the Construction of a Unified National Agricultural Product Market. Chin. Rural Econ. 2024, 1, 5–9. [Google Scholar]
  16. Chen, F.J.; Li, Z.G. International Agricultural Products Geographical Indication Management System and Experience for Reference. Jiangsu Agric. Sci. 2017, 45, 1–4. [Google Scholar]
  17. FAO & CIRAD. Global Perspectives on Geographical Indications: Crossed Views Between Researchers, Policy Makers and Practitioners; Food and Agriculture Organization of the United Nations: Rome, Italy, 2025. [Google Scholar]
  18. Jiang, S.; Wang, C. Study on Performance Indicator System of Green Supply Chain in Pharmaceutical Industry. China Pharm. 2021, 32, 1932–1937. [Google Scholar]
  19. Zhang, M.; Shi, H.; Williams, L.; Lighterness, P.; Li, M.; Khan, A.U. An Empirical Test of the Influence of Rural Leadership on the Willingness to Participate in Public Affairs from the Perspective of Leadership Identification. Agriculture 2023, 13, 1976. [Google Scholar] [CrossRef] [Scilit]
  20. Wang, H.; Sarkar, A.; Rahman, A.; Hossain, S.; Memon, W.H.; Qian, L. Research on the industrial upgrade of vegetable growers in Shaanxi: A cross-regional comparative analysis of experience reference. Agronomy 2021, 12, 38. [Google Scholar] [CrossRef] [Scilit]
  21. Ding, B.; Kong, L. Performance Evaluation of Agricultural Product Supply Chain under the Background of Agricultural-Commercial Interconnection. Manuf. Autom. 2023, 45, 85–90. [Google Scholar]
  22. Guo, Y.; Cui, M.; Xu, Z. Spatial characteristics of transfer plots and conservation tillage technology adoption: Evidence from a survey of four provinces in China. Agriculture 2023, 13, 1601. [Google Scholar] [CrossRef] [Scilit]
  23. Lu, S.; Sun, Z.; Huang, M. The impact of digital literacy on farmers’ pro-environmental behavior: An analysis with the Theory of Planned Behavior. Front. Sustain. Food Syst. 2024, 8, 1432184. [Google Scholar] [CrossRef] [Scilit]
  24. Li, W.; Liu, C.; Yang, Q.; You, Y.; Zhuo, Z.; Zuo, X. Factors influencing farmers’ vertical collaboration in the agri-chain guided by leading enterprises: A study of the table grape industry in China. Agriculture 2023, 13, 1915. [Google Scholar] [CrossRef] [Scilit]
  25. Wang, K.; Yang, Y. Grey Clustering-Fuzzy Comprehensive Model and Application of Performance Evaluation of Green Agricultural Product Supply Chain. Math. Pract. Theory 2020, 50, 111–119. [Google Scholar]
  26. Zhou, Y. Performance Evaluation of Agricultural Product Supply Chain Based on AHP-FCE Model. Stat. Decis. 2020, 36, 178–180. [Google Scholar]
  27. Cui, Y.; Zhao, C.; Zhang, Q. Impact of digital transformation in agribusinesses on total factor productivity. Int. Food Agribus. Manag. Rev. 2024, 27, 843–857. [Google Scholar] [CrossRef] [Scilit]
  28. Feng, J.; Yuan, B.; Li, X.; Zhang, X. Progress of Neural Network in Supply Chain Management of Fresh Agricultural Products. Trans. Chin. Soc. Agric. Mach. 2019, 50, 366–373. [Google Scholar]
  29. Cao, Y. Research on Performance Evaluation of Agricultural Product Circulation Supply Chain Based on BP Neural Network. J. Commer. Econ. 2021, 11, 134–137. [Google Scholar]
  30. Fang, X.; Zhang, L.; Yan, X.; Dong, C. Intelligent Evaluation of Pharmaceutical Supply Chain Collaborative Performance Based on BAS-BP Neural Network. Comput. Appl. Softw. 2023, 40, 59–66+113. [Google Scholar]
  31. Liao, X.; Luo, X. Research on the Development Efficiency of Enterprise Green Supply Chain and Its Influencing Factors: An Empirical Analysis Based on Survey Data from National-Level New Areas. Friends Account. 2022, 6, 27–32. [Google Scholar]
  32. Fang, W.; Yang, B. Research on Evaluation of Enterprise Green Supply Chain Operation Efficiency Based on DEA Method. J. Ind. Technol. Econ. 2017, 36, 19–26. [Google Scholar]
  33. Cao, B.; Fan, Y. Research on Performance Evaluation of the Green Agricultural Products Supply Chain Based on DEA and Principal Component Analysis. Sci. Technol. Manag. Res. 2017, 37, 72–77. [Google Scholar]
  34. Chen, J.; Cheng, Y. Construction of Performance Evaluation System for Retail Enterprises Under New Retail Business Model. Commun. Financ. Account. 2023, 4, 89–95. [Google Scholar]
  35. Jia, P.; Dong, J. Extension Superiority Evaluation of Logistics Service Supply Chain Performance Indicators Based on BSC. Stat. Decis. 2018, 34, 44–48. [Google Scholar]
  36. Wang, L.; Yu, C.; Wang, M.; Ling, D. Study on Policy Effect and Spatial-Temporal Difference of Agricultural Production Efficiency in China—Empirical Analysis Based on Three-Stage DEA Model. Soft Sci. 2019, 33, 33–39. [Google Scholar]
  37. Cui, H.; Zhuo, W.; Yu, H.; Long, J.; Liu, Y. Calculation of agricultural production efficiency based on a three-stage Data Envelopment Analysis model and analysis of the spatial-temporal characteristics: An example from the Yangtze River Economic Belt. Chin. J. Eco-Agric. 2021, 29, 1243–1252. [Google Scholar]
  38. Wu, Z.; Lei, L. Study on Agricultural Land Ecological Efficiency Based on Three-Phase DEA Model: Taking Henan Province as an Example. Ecol. Econ. 2018, 34, 76–80. [Google Scholar]
  39. Islam, D.I.; Rahman, A.; Sarker, M.S.R.; Luo, J.; Liang, H. Factors affecting farmers’ willingness to adopt crop insurance to manage disaster risk: Evidence from Bangladesh. Int. Food Agribus. Manag. Rev. 2021, 24, 463–480. [Google Scholar] [CrossRef] [Scilit]
  40. Wang, B.; Du, X.; Wang, Y.; Mao, H. Multi-machine collaboration realization conditions and precise and efficient production mode of intelligent agricultural machinery. Int. J. Agric. Biol. Eng. 2024, 17, 27–36. [Google Scholar] [CrossRef] [Scilit]
  41. Osei, C.D.; Zhuang, J.; Adu, D. Impact of regulatory, normative and cognitive institutional pressures on rural agribusiness entrepreneurial opportunities and performance: Empirical evidence from Ghana. Int. Food Agribus. Manag. Rev. 2024, 27, 651–670. [Google Scholar] [CrossRef] [Scilit]
  42. Wan, F.; Qiu, Z. Evaluation of Fresh Agricultural Produce Cold Chain Logistics Efficiency Based on Three-Stage DEA. Sci. Technol. Ind. 2022, 22, 316–320. [Google Scholar]
  43. Qin, X.; Zhao, C. Study on Evaluation of Cold Chain Logistics Efficiency of Fresh Agricultural Products. Storage Process 2023, 23, 63–69. [Google Scholar]
  44. Zhang, L. Research on Spatial Differences in the Efficiency of Cold Chain Logistics for Fresh Agricultural Products in China Based on Three-Stage DEA. Supply Chain Manag. 2025, 6, 77–89. [Google Scholar]
  45. Yan, H.; Qiao, J.D. The Impact of Grain Minimum Purchase Price Policy on the Comprehensive Technical Efficiency of Grain Processing Industry: An Empirical Study Based on Three-Stage DEA and Tobit Model. Commer. Res. 2021, 4, 120–131. [Google Scholar]
  46. Lin, W.; Su, X.; Ma, Y.; Li, S.; Zeng, D. Impact of agricultural machinery socialization services on farmland production efficiency in our country: Empirical research based on three-stage DEA-Tobit models. J. Chin. Agric. Mech. 2022, 43, 239–248. [Google Scholar]
  47. Iqbal, B.; Alabbosh, K.F.; Jalal, A.; Suboktagin, S.; Elboughdiri, N. Sustainable food systems transformation in the face of climate change: Strategies, challenges, and policy implications. Food Sci. Biotechnol. 2025, 34, 871–883. [Google Scholar] [CrossRef] [Scilit]
  48. Pan, M. Research on Financing Efficiency of Cross-Border E-Commerce Enterprises Engaged in Agriculture Based on Three-Stage DEA-Tobit Model. Guangxi Econ. 2023, 41, 96–100. [Google Scholar]
  49. Tian, G.; Feng, Y. Is the E-commerce of Vertical Fresh Agro-Products Efficient?—An Empirical Research Based on Three-Stage DEA Model Considering Environmental Factors. J. Bus. Econ. Manag. 2018, 6, 27–33. [Google Scholar]
  50. Chen, W.; Zhang, X.; Zhou, R.; Liu, B. Study on Evaluation and Influencing Factors of Agricultural Products Circulation Efficiency in Heilongjiang Province Based on Three-Stage DEA Method. Bus. Econ. 2020, 5, 6–10+82. [Google Scholar]
  51. Alhussam, M.I.; Ren, J.; Yao, H.; Abu Risha, O. Food trade network and food Security: From the perspective of Belt and Road Initiative. Agriculture 2023, 13, 1571. [Google Scholar] [CrossRef] [Scilit]
  52. Zhang, J. A Comparative Study on the Effectiveness of Smart Transformation in the Supply Chains of Sichuan Pepper Oil Production Enterprises in Sichuan Province. Master’s Thesis, Xihua University, Chengdu, China, 2022. [Google Scholar]
  53. He, S.; Huang, J. Performance Evaluation of Geographical Indication Agricultural Product Supply Chain: A Typical Case Analysis of Gannan Navel Oranges Based on the “Farmers + Cooperatives + Supermarkets” Channel Model. World Agric. 2017, 10, 134–139. [Google Scholar]
  54. Simar, L.; Wilson, P. Statistical Inference in Nonparametric Frontier Models: The State of the Art. J. Product. Anal. 2000, 13, 49–78. [Google Scholar] [CrossRef] [Scilit]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.