Monetizing Food Waste and Loss Externalities in National Food Supply Chains: A Systems Analytics Framework

Abstract

Reducing food loss and waste (FLW) is a global priority under UN SDG 12.3, yet Taiwan has lacked stage-specific FLW data and systematic valuation of its environmental and economic implications. This study addresses these gaps by integrating localized FLW estimates from the APEC-FLOWS database with an enhanced analytical framework—the Environmentally Extended Input–Output Valuation (EEIO-V) model. The EEIO-V extends conventional input–output analysis by monetizing multiple environmental burdens, including greenhouse gases, air pollutants, wastewater, and solid waste, thereby linking FLW reduction to tangible economic benefits and policy design. The simulations reveal substantial differences in environmental cost reductions across supply chain stages, with downstream interventions delivering the largest benefits, particularly in reducing air pollution and greenhouse gases. By contrast, upstream measures contribute relatively smaller improvements. These findings highlight the novelty of EEIO-V in bridging environmental valuation with system-level FLW analysis, and they provide actionable insights for designing cost-effective, stage-specific strategies that prioritize downstream interventions to advance Taiwan’s sustainability and policy goals.

1. Introduction: Background and Research Objectives

Food loss and waste (FLW) has emerged as a central concern in global sustainability discourse, reflecting its deep interconnections with food security, social equity, and environmental integrity. FLW is closely linked to the efficiency of energy and resource use, land and water consumption, and greenhouse gas (GHG) emissions, placing it at the nexus of several planetary challenges [1,2,3]. According to the estimation of Food and Agriculture Organization of the United Nations (FAO), approximately one-third of all food produced globally—over 1.3 billion tons annually—is lost or wasted across various stages of the supply chain [4]. The embedded environmental and economic costs of this inefficiency are substantial [5,6], undermining the effectiveness of global food systems and exacerbating climate change and natural resource degradation [7].

In response, the United Nations formally adopted Sustainable Development Goal (SDG) 12.3 in 2015 as part of the global sustainability agenda. SDG 12.3 calls for halving per capita food waste at the retail and consumer levels and reducing food losses along production and supply chains by 2030 [8]. This target has since served as a strategic benchmark for national governments in restructuring food systems and enhancing resource efficiency. Moreover, it has catalyzed growing engagement from businesses, cities, and civil society in developing FLW mitigation plans and sustainability actions.

As the governance of FLW becomes increasingly institutionalized and policy-driven, the need for systematic assessment of the environmental and economic impacts of FLW management strategies has emerged as a critical interdisciplinary research agenda. For instance, a study conducted in Australia revealed that the production of food ultimately lost or wasted accounted for 9% of the nation’s total water use and 6% of its greenhouse gas (GHG) emissions [9]. In the United States, simulation-based assessments estimated that halving national FLW could reduce the environmental impacts of the food system by 8–10% [10]. On a broader scale, another study quantified the nutritional and environmental consequences of FLW across 121 countries and 20 composite regions. From 2004 to 2014, FLW in these areas increased by 25%, with rising food imports exacerbating both losses and environmental footprints in low-income exporting countries [11].

To capture the intersectoral structure of modern food supply chains (FSC) and assess the full-system environmental implications, many of these studies employ the Environmentally Extended Input–Output (EEIO) model. EEIO models are particularly well suited for system-level analyses because they account for inter-industry linkages and can incorporate multiple impact categories—including GHGs, water use, air pollution, and waste generation—depending on the policy focus [12,13,14,15,16,17].

However, existing EEIO-based research on FLW remains largely limited to physical quantification and seldom includes monetized valuation of environmental externalities. Without translating environmental burdens into economic costs, it becomes difficult to incorporate such results into fiscal policy frameworks, incentive design, or cost–benefit comparisons [18]. As a result, while EEIO models offer valuable insights into pollution hotspots and structural sources of environmental pressure, their direct applicability to policymaking remains constrained by the absence of monetary valuation.

Taiwan has also gradually incorporated FLW into its broader sustainability governance framework. The national government has signaled its intention to align with SDG 12.3 by establishing a 2030 reduction target [19], and relevant ministries have launched initiatives such as food banks, surplus food redistribution, school-based education programs, and consumer behavior campaigns [20]. However, critical data gaps persist in Taiwan’s FLW governance landscape. There is currently no comprehensive inventory of the physical quantities of FLW, no disaggregated estimates of losses and waste by supply chain stage, and no systematic analysis of the environmental impacts associated with FLW.

However, Taiwan currently faces a critical research gap regarding FLW due to the lack of comprehensive data on the quantity of losses, stage-wise distribution across the supply chain, and the actual reduction potential of management interventions. Specifically, there is no clear differentiation between food loss and waste across stages, and systematic assessments of their environmental impacts are missing. Moreover, no existing studies have integrated FLW management measures into policy simulation scenarios or linked them to quantitative evaluations of environmental externalities. As a result, the benefits of such measures cannot be readily monetized, making it difficult to translate research outcomes into actionable policy designs or to support cross-sector stakeholder engagement.

Although several initiatives such as food banks, redistribution programs, and consumer awareness campaigns have been launched, Taiwan still lacks an integrated institutional framework for FLW governance. Policies remain fragmented across ministries, and there is no unified statistical system that can monitor stage-specific FLW while connecting management interventions with quantitative environmental assessments. This gap not only constrains evidence-based policymaking but also highlights the need for systematic data integration aligned with international approaches such as those developed by FAO and APEC-FLOWS.

In response to the aforementioned context and research gap, this study aims to develop the Taiwan Environmentally Extended Input–Output Valuation model (EEIO-V), based on Taiwan’s 63-sector input–output table. By converting environmental impacts—including greenhouse gas emissions, wastewater, air pollution, and solid waste—into monetized environmental costs, the model serves as a systematic analytical tool for policy benefit evaluation and cross-sector stakeholder communication. Compared to conventional EEIO approaches, the EEIO-V model offers a key innovation by integrating both environmental burden quantification and external cost monetization, allowing the benefits of FLW management strategies to be explicitly linked to overall societal resource savings and the reduction in environmental governance costs.

Specifically, the primary objective of this study is to construct a system analysis methodology and apply it to evaluate the expected environmental outcomes and benefits of Taiwan’s FLW management measures. Additionally, this study lays a foundation for future expansion of FLW-related research. On one hand, it enables further refinement through integration with spatial data, physical logistics information, and food category classifications to support more granular modeling. On the other hand, by incorporating policy instruments such as carbon pricing, external cost internalization, and regulatory impact analysis (RIA), the EEIO-V framework can also function as a decision support system (DSS) for the development of FLW governance strategies at the national or urban level.

Overall, this study contributes to the literature by addressing the current void in full-chain FLW mitigation potential and externality valuation, while showcasing the methodological applicability and policy relevance of the EEIO-V approach through a Taiwan case study. The insights generated may inform sustainable food system governance and circular economy transitions more broadly.

The remainder of this article is organized as follows: Section 2 and Section 3 outline the EEIO-V model’s conceptual design and data structure; Section 4 presents the policy-relevant scenario simulation design and analyzes changes in environmental impacts and costs across multiple sectors and pollutant categories; Section 5 discusses the strategic and managerial implications of the findings; and Section 6 concludes with key takeaways and future research directions.

2. Methodology: Environmentally Extended Input–Output Valuation Model (EEIO-V)

To systematically evaluate the environmental and economic impacts of FLW management strategies, this study constructs the Taiwan EEIO-V model, tailored to the structure of Taiwan’s national economy. The EEIO-V model builds upon the conventional EEIO approach by incorporating an environmental valuation module that monetizes key environmental impacts. This integration enables a unified analytical framework that simultaneously captures both physical reductions in environmental burdens and their corresponding economic benefits, while also allowing for comparative scenario analysis. By transforming physical environmental impacts into monetary terms, the EEIO-V model enhances the interpretability of results for policy applications. It supports policy prioritization, cost–benefit analysis, and the design of cross-sectoral mitigation strategies under sustainability objectives. The relationship between the conventional input–output model, the EEIO model, and the EEIO-V model developed in this study is outlined as Figure 1.

2.1. Incorporating Environmental Impacts: The EEIO Framework

The core component of the EEIO-V model constructed in this study is the EEIO model. From a technical perspective, the EEIO framework represents a combined application of the traditional input–output (IO) model and environmental impact coefficients [21]. More specifically, the conventional IO model, which has been under development since the 1930s, depicts the inter-industry production and consumption relationships within a macroeconomic system [22]. The EEIO model extends this by incorporating the environmental byproducts generated during production processes, enabling researchers to simultaneously observe both economic and environmental impacts in macroeconomic or sectoral analyses [23,24]. Since its emergence in the 1970s, the EEIO approach has undergone multi-regional expansions, database integrations, and software tool development. Today, it is regarded as an indispensable methodology for assessing environmental impacts in macroeconomic analysis [25].

The conventional IO model captures the inter-industry relationships that characterize the structure and functioning of an economy [26]. Within the context of life cycle analysis (LCA), IO models are particularly valuable because they reflect the interactions between upstream and downstream supply chains and the broader economy. As such, the IO framework is considered a practical tool for conducting life cycle inventory (LCI) assessments at the aggregated industrial level [27]. For this reason, IO-based approaches are often referred to as economic input–output life cycle assessment models (EIO-LCA).

In the EEIO extension, various environmental burdens—such as air pollutants, greenhouse gas (GHG) emissions, and other by-products—are linked to sectoral production activities. By coupling these environmental stressors with the economic production structure embedded in the IO model, the EEIO framework systematically accounts for the environmental externalities associated with each unit of output [28].

The conventional IO model is constructed based on an interindustry transaction table, which details the flow of goods and services across sectors. Each column in the table represents the production structure of a specific sector, indicating the value of intermediate inputs sourced from other industries, payments to production factors (such as labor and capital), and indirect taxes. Each row reflects the consumption pattern of a sector’s output—how much is used as intermediate inputs in other sectors versus how much is allocated to final demand components, such as household consumption, investment, government spending, and net exports.

Let F denote the total final demand across the four main components. The economic transaction relationships that describe sectoral production and consumption behaviors can be expressed mathematically as (1):

$$\begin{array}{l}{X}_1=Z_{11}+Z_{12}+Z_{13}+\cdots +Z_{1n}+F_1\\ X_2=Z_{21}+Z_{22}+Z_{23}+\cdots +Z_{2n}+F_2\\ ⋮\\ X_n=Z_{n1}+Z_{n2}+Z_{n3}+\dots +Z_{nn}+F_n\end{array}$$

(1)

Using matrix notation, Equation (1) can be reformulated as (2):

$$\left[\begin{array}{c}{X}_1\\ ⋮\\ X_n\end{array}\right]=\left[\begin{array}{ccc}{a}_{11}& \cdots & a_{1n}\\ ⋮& \ddots & ⋮\\ a_{n1}& \cdots & a_{nn}\end{array}\right]\times \left[\begin{array}{c}{X}_1\\ ⋮\\ X_n\end{array}\right]+\left[\begin{array}{c}{F}_1\\ ⋮\\ F_n\end{array}\right]; a_{ij}={\displaystyle \raisebox{1ex}{$Z_{ij}$}\!\left/ \!\raisebox{-1ex}{$X_j$}\right.}$$

(2)

Rearranging terms yields (3):

$$\left[\begin{array}{ccc}1-a_{11}& ⋮& -a_{1n}\\ ⋮& \ddots & ⋮\\ -a_{n1}& ⋮& {1-a}_{nn}\end{array}\right]\times \left[\begin{array}{c}{X}_1\\ ⋮\\ X_n\end{array}\right]=\left[\begin{array}{c}{F}_1\\ ⋮\\ F_n\end{array}\right]$$

(3)

Further simplified using matrix notation as shown in (4) and (5):

$$\left(I-A\right)X=FX$$

(4)

$$X={\left(I-A\right)}^{-1}F$$

(5)

where I is the identity matrix, and A is the matrix composed of technical coefficients a(ij). The term (I − A)⁻¹ represents the matrix of direct plus indirect requirements, also known as the Leontief inverse matrix, which captures the total inter-industry dependencies.

By applying (5), one can determine the impact of a unit change in production or consumption in a specific sector on the output of all other sectors in the economy, due to interlinked upstream and downstream relationships. However, the conventional IO model only captures economic variables and does not account for the environmental impacts associated with production processes. To incorporate these aspects, the model must be extended into an EEIO framework, as conceptualized in Equation (6) [29,30].

$$E=R·X=R·{\left(I-A\right)}^{-1}F$$

(6)

In (6), E represents the vector of environmental impacts generated by economic activities, while R is a diagonal matrix capturing the sector-specific environmental intensities, that is, the amount of environmental impact (potentially across multiple categories) generated per unit of economic output in each sector.

2.2. Integration of Environmental Valuation into EEIO: The EEIO-V Model

Conventional EEIO models are characterized by their ability to quantify the environmental impacts associated with economic activities. In contrast, the EEIO-V model developed in this study advances the conventional framework by incorporating the monetization of environmental impacts. This enhancement enables the evaluation of policy or management measures from a cost–benefit perspective, facilitating integration into decision-making processes [31]. The conceptual structure of the EEIO-V model can be represented as (7):

$$EV=EV·E=EV·R·X=EV·R·{\left(I-A\right)}^{-1}F$$

(7)

In (7), EV denotes the vector of monetized environmental values resulting from economic activities, capturing the environmental costs or benefits incurred as externalities.

Conceptually, the EV vector can be understood as a bridge that links physical environmental impacts to their corresponding monetary values. While a conventional EEIO model quantifies the amounts of pollutants or emissions generated, the EV vector allows these burdens to be expressed in monetary terms as external costs. For example, instead of only reporting that a given sector produces a certain volume of PM_2.5 or greenhouse gases, the EV framework enables us to represent these impacts in terms of their associated economic damages, such as public health costs or climate-related damages. This monetization feature makes the results more intuitive and directly comparable across different environmental categories, which is precisely what distinguishes the EEIO-V model from traditional EEIO approaches.

When sectoral economic activities change due to specific drivers or interventions—such as policy implementation—the corresponding variation in environmental costs or benefits can be estimated using (8).

$$∆EV=EV·R·{\left(I-A\right)}^{-1}∆F$$

(8)

To further clarify the methodological contribution of this study, it is important to note that the EEIO-V model represents an advancement of the conventional Environmentally- EEIO framework. While traditional EEIO analyses primarily focus on quantifying environmental burdens in physical terms, the EEIO-V model integrates monetary valuation to convert multiple environmental impacts into economic cost equivalents. This enhancement enables a more comprehensive assessment of the environmental and economic benefits associated with different mitigation strategies.

To the best of our knowledge, no existing studies on FLW have incorporated such valuation mechanisms within an EEIO framework. For contextual reference, Appendix A 

Table A1. Summary of selected studies that apply conventional EEIO models to FLW topics.

Authors	Year	Region	Focus	Summary
Osei-Owusu et al. [7]	2023	Europe	50% FLW reduction potential	Halving Europe’s FLW yields major CO2e, water, and land savings; regional variations call for tailored policies.
Reutter et al. [9]	2017	Australia	Environmental and socio-economic impacts of food waste using EEIO	EEIO analysis shows Australian food waste accounts for 9% of water use and 6% of GHGs, demonstrating the method’s suitability for integrated impact assessment.
Read et al. [10]	2020	United States	Stage-specific environmental benefits of 50% FLW reduction	EEIO analysis shows the greatest environmental gains from halving FLW occur in foodservice, processing, and households, guiding targeted reduction strategies.
Gatto et al. [11]	2024	Global (121 countries, 20 regions)	Quantification of FLW and its nutritional and environmental impacts	FLW rose 25% (2004–2014), worsening nutrition and footprints; reducing overconsumption in high-income regions can benefit exporting low-income countries.
Cederberg et al. [12]	2019	Sweden	Footprint indicators of food consumption impacts using MRIO (EXIOBASE3)	Swedish food consumption exerts significant overseas agrochemical and carbon footprints, highlighting the need for better data and global responsibility.
Usubiaga et al. [16]	2017	EU-28	Environmental benefits of consumer food waste reduction targets	Meeting the EU Roadmap goal to halve edible food waste could reduce GHGs, land, water, and material footprints by 2–7%, mainly from household actions.
Salemdeeb et al. [17]	2018	UK	Environmental assessment of food waste treatment methods under decarbonization scenarios	A hybrid IO-LCA shows composting performs best overall, with AD next; capital goods significantly affect impacts, highlighting hybrid methods’ value.
Canning et al. [45]	2020	United States	Relationship between dietary changes and natural resource use	Aligning U.S. diets with USDA guidelines could improve nutrition and resource efficiency, though water use may involve trade-offs for certain foods.

provides a brief overview of selected international studies that apply conventional EEIO models to FLW-related topics, illustrating the current research landscape and highlighting the unique methodological contribution of this study.

2.3. Key Features of the Taiwan EEIO-V Model

Summary of Key Features of the Taiwan EEIO-V Model as follows:

• Reflecting the industrial structure of 2021
  This study is based on the most recently released input–output table compiled by Taiwan’s DGBAS in 2024, which reflects the industrial structure of the year 2021 [32].
• Model construction based on a 63-Sector classification
  A major advantage of the EEIO-V model lies in its ability to incorporate environmental emissions from production processes and to monetize the associated impacts. Given that environmental emission data are compiled from multiple statistical databases, the model adopts a 63-sector classification as published by DGBAS. This approach balances economic representativeness and the availability of sector-specific environmental data.
• Simulation based on the domestic product transaction structure
  While conventional IO models allow for various assumptions regarding the linkage effects of imported goods on domestic industries, the EEIO-V model can also accommodate such flexibility. However, this study focuses on the environmental impacts and monetized externalities arising specifically from domestic food production activities. Therefore, the simulation is primarily based on Taiwan’s Domestic Product Transaction Table (D-Table), which captures the structure of transactions involving domestically produced goods.

2.4. Simulation Process

The simulation process in this study follows a structured sequence to ensure consistency and transparency in applying the EEIO-V framework. The first step involves establishing the baseline EEIO-V model, which links Taiwan’s 63-sector input–output structure with environmental satellite accounts and monetization coefficients. This provides the reference system of economic activities, environmental burdens, and their monetized external costs.

The second step designs alternative scenarios of food loss and waste (FLW) reduction in line with SDG 12.3. Each scenario specifies which stage of the food supply chain is targeted for intervention (e.g., agricultural production, processing, distribution/retail, or food services). For each case, the assumed reductions are mapped onto the corresponding sectors within the input–output framework.

The third step integrates these scenario-specific changes into the baseline model. Adjustments are made to either intermediate input requirements or final demand values, depending on the stage of intervention. The updated economic flows are then used to calculate the resulting changes in environmental burdens across the entire supply chain.

Finally, the monetization step translates these environmental changes into external cost estimates using the EV vector. This allows direct comparison of the economic significance of reductions across different pollutant categories and FLW stages. By following this sequential process, the EEIO-V model provides a coherent simulation framework that connects scenario design, system-wide economic adjustments, environmental outcomes, and monetary valuation.

3. Data Sources and Processing of Taiwan EEIO-V Model

3.1. Data Sources and Processing

• Industry-level input–output model
  The core structure of the model is based on Taiwan’s official input–output tables, published every five years by the Directorate-General of Budget, Accounting and Statistics (DGBAS). The input–output framework enables the simulation of how changes in final demand for goods and services in specific economic sectors affect interlinked production values across the broader economy.
• Environmental impacts inventory
  The model integrates environmental satellite data corresponding to key pollutant categories associated with economic production. Currently, it covers four major types of environmental impacts: air pollutants, solid waste, wastewater, and greenhouse gas (GHG) emissions. Emissions coefficients for each sector are compiled using multiple national data sources, including the Green National Income Accounts, the Energy Balance Sheets, and the National Greenhouse Gas Inventory, the latter of which is used to estimate GHG emissions from energy consumption across sectors (see

  Table 1. Data sources and processing for environmental impact categories.

  Category	Item	Data Source and Processing
  Air Pollution	PM2.5, SOx, NOx	Derived from Green National Income Account Compilation Report (Air Pollution Emission Ledger) 1
  Wastewater	Agriculture, Industry, Service Sector	Derived from Green National Income Account Compilation Report (Wastewater Emission Ledger) 1
  Solid Waste	Agriculture, Industry, Service Sector	Derived from Green National Income Account Compilation Report (Solid Waste Emission Ledger) 1
  Greenhouse Gases	Seven GHGs converted to CO2e	Fuel combustion emissions: Estimated using data from the Energy Balance Sheet 2 Non-fuel combustion emissions: Taken from the National GHG Inventory 3

  ¹ Source: [33]; ² Source: [34]; ³ Source: [35].

  ).
• Monetization of Environmental Impacts
  In addition to quantifying changes in physical pollution, the model incorporates monetization factors to convert environmental changes into economic terms. This feature enables the estimation of external environmental costs or benefits under different policy scenarios. The monetary values used to quantify environmental externalities are derived from established valuation studies and official guidelines, as summarized in the corresponding data table (see

  Table 2. Data sources and processing of monetization indicators for environmental impacts.

  Category	Data Source and Processing
  Air Pollution	Based on the Benefit per ton (BPT) method developed by the US EPA, with parameters localized for Taiwan 1
  Wastewater	Converted using the Wastewater Damage Ledger in the Green National Income Report 2
  Solid Waste	Converted using the Solid Waste Damage Ledger in the Green National Income Report 2
  Greenhouse Gases	Adopted the 2021 Social Cost of Carbon (SCC) estimate published by the US EPA 3

  ¹ For a detailed explanation, please refer to the following section; ² Source: [33]; ³ Source: [36].

  ).

Firstly, the computation of Benefit per Ton (BPT) associated with air pollution involves a three-step process: (1) simulating the relationship between pollutant emissions and corresponding ambient concentration changes; (2) estimating the impact of concentration changes on health risks—typically measured through changes in specific health outcomes such as mortality or morbidity rates; and (3) monetizing these health risks [18]. Health damages may include medical expenditures or the value of premature mortality. Among these, the economic valuation of mortality risk generally exceeds that of healthcare costs. The monetary valuation of life loss is often used as a proxy for the social cost of air pollution or the benefits of pollution abatement. Empirically, the Value of a Statistical Life (VSL) is commonly employed to monetize mortality risk. The formula for calculating BPT in terms of pollution reduction is expressed as follows:

$${BPT}_j=(∆I_j\times {VSL}_t)/{AP}_j$$

(9)

Here, AP_j denotes the emission volume of pollutant j in the year 2021, where j = PM_2.5, SOx, or NOx. ΔI_j represents the change in mortality risk attributable to the variation in ambient concentration of pollutant j. VSL_t denotes the Taiwan-specific value of a statistical life (VSL) in year t.

The VSL used in this study is transferred from prior research. The most recent Taiwan-specific VSL estimate available is from [36], which was derived using Taiwanese survey data and based on 2014 wage levels. In this study, we apply a benefit transfer approach to adjust 2014 VSL estimate to 2021 monetary terms. Since the VSL in [37] was estimated using the hedonic wage method, differences in average wage levels across years directly influence the VSL outcome. Therefore, Equation (10) serves to adjust for temporal differences in wage levels, which correspond to changes in the VSL.

$${VSL}_{2021}={VSL}_{2014}\times \left\{1+{\displaystyle \frac{{\in }_w\times (W_{2021}-W_{2014})/W_{2021}}{100}}\right\}$$

(10)

The parameter sources related to the BPT calculation are summarized in

Table 3. Parameter sources related to the BPT calculation.

Parameter	Definition
$∆I_j$	Change in mortality risk due to variation in pollutant j concentration 1
${AP}_j$	Air pollution emission volume of pollutant j in the year 2021 1
${VSL}_{2014}$	Estimated value of statistical life (VSL) for Taiwan in 2014 2
${\in }_{w,2014}$	Wage elasticity of labor supply in Taiwan 3
$W_{2014} \& W_{2021}$	Average monthly earnings in US$ in 2014 and 2021 4

¹ As estimated by [18]; ² Source: [37]; ³ The results from [37] are used to compute the elasticity results used in this study; ⁴ Source: [38].

.

Based on the aforementioned parameters and the BPT methodology, the unit damage costs (BPT) for PM_2.5, SOx, and NOx in 2021 monetary values were estimated at 2.19 million USD/t, 0.31 million USD/t, and 0.11 million USD/t, respectively.

Therefore, the monetization indicator for GHGs emissions was derived from the 2023 Social Cost of Carbon (SCC) estimates published by [36]. Meanwhile, the monetization factors for wastewater and solid waste were sourced from Taiwan’s Green National Income Account Compilation Report [32]. To ensure consistency with the model’s reference year, this study adopted the valuation results for 2021. A summary of these monetization coefficients is provided in

Table 4. Monetization Factors for Greenhouse Gases, Solid Waste, and Wastewater ¹.

Category	Monetization Results
Wastewater	Agriculture: 59.9 USD/t; Industry: 7438.4 USD/t; Services: 4364.1 USD/t
Solid Waste	Agriculture: 59.9 USD/t; Industry: 61.5 USD/t; Services: 54.9 USD/t
Greenhouse Gases	197 USD/tCO2e (2021 value, based on a 2.5% discount rate)

¹ USD values were converted using Taiwan’s average exchange rate in 2021 (1 USD = 28.02 TWD); Source: [39].

.

3.2. Sources of Uncertainty and Their Implications

The EEIO-V model developed in this study integrates the traditional EEIO framework with monetized environmental indicators. Thus, the discussion of data uncertainty can be approached from these two dimensions.

From the EEIO perspective, the model is primarily constructed using Taiwan’s input–output tables and various environmental datasets. As previously mentioned, both the input–output data and environmental impact data are derived from official government sources in Taiwan, including the Green National Income Report, Energy Balance Sheet, and National GHG Inventory. These official statistics are based on survey methods, and data uncertainty largely stems from the quality of data collected during the reporting and survey processes. However, these sources do not provide explicit information or analysis on data uncertainty, which precludes quantitative uncertainty analysis of the EEIO model itself. This limitation is also commonly noted in the literature on traditional EEIO modeling [10].

On the other hand, a key feature of the EEIO-V model is its integration of monetized environmental indicators, allowing environmental burdens generated throughout the production and consumption processes to be expressed in monetary terms—as environmental costs. The monetized indicators used in this study—covering wastewater, waste, air pollutants, and greenhouse gases—were derived from different sources. Some are based on official statistical data, while others were calculated using value transfer methods from existing literature. The sources and nature of uncertainty thus vary by category.

For wastewater and solid waste, the monetary values were directly taken from Taiwan’s Green National Income Account. As these values are based on government reports that do not include uncertainty analysis, the corresponding monetary estimates are treated as deterministic values, subject only to potential uncertainties arising from the original survey process.

For greenhouse gas emissions, this study adopted the Social Cost of Carbon (SCC) values published by the U.S. Environmental Protection Agency (US EPA) in 2021 [36]. The SCC was estimated using state-of-the-art Integrated Assessment Models (IAMs), which combine socio-economic, emission, climate, and damage modules. As a result, SCC estimation incorporates a wide range of parameter uncertainties [36]. To capture these uncertainties without making the result overly diffuse and less practical for analysis, the EPA adopted a certainty-equivalent value approach. This method does not provide a simple point estimate; instead, it reflects the expected marginal damage of greenhouse gas emissions, accounting for all quantifiable uncertainties (e.g., socio-economic pathways, climate physics, sea-level rise parameters, and damage function parameters). The report states that this approach is intended to both reflect uncertainty and offer a manageable reference value for policy analysis. Hence, the SCC values used in this study represent a synthesis of multiple uncertainties already embedded in the original estimation process.

Regarding air pollution, this study estimated localized Benefit per Ton (BPT) values based on methodologies in existing literature [18]. The core parameter in calculating BPT is the Value of Statistical Life (VSL), which was estimated using a value transfer method based on wage-risk tradeoffs. The original VSL source relied on the hedonic wage method, which infers individuals’ willingness to pay for mortality risk reductions by examining wage differentials across industries with varying occupational risks [37]. In empirical estimation, the coefficient on fatal risk is the critical variable for computing VSL. The statistical uncertainty of this coefficient is typically presented as a confidence interval, allowing for an indirect measure of VSL uncertainty. Based on the source used in this study, the 90% confidence interval (at a 10% significance level) for VSL was estimated to be ±11.3%, which is also used as the uncertainty range for the BPT values in this study.

In summary, given the limitations and availability of data, the quantifiable uncertainty in the EEIO-V model primarily pertains to the monetized values for air pollutants (i.e., BPT), with an estimated uncertainty range of ±11.3% for pollutants such as PM_2.5, SOx, and NOx. Other environmental indicators in the model rely on deterministic statistical data, and uncertainties are mostly rooted in the original data collection and reporting processes.

4. Scenario Simulation and Analysis

4.1. Alignment of the FSC with the Input–Output Table

This study synthesizes and adapts classifications of the FSC stages based on prior literature [10,40,41,42,43,44,45]. We categorize the national FSC into four stages: (1) agricultural production, (2) food processing, (3) distribution/retail, and (4) accommodation and foodservice.

• The agricultural production stage encompasses primary activities related to food raw material production, including crop farming, fisheries, and livestock operations.
• The food processing stage refers to the transformation of raw materials into consumable food products.

These two stages together represent the upstream segment of the FSC.

• The distribution/retail stage covers wholesale and retail trade of food products, as well as associated transportation and storage activities.
• The accommodation and foodservice includes hospitality-related services such as restaurants and lodging establishments involved in food provision.

These latter two stages constitute the downstream segment of the supply chain.

To integrate the FSC with the structure of the EEIO-V model—comprising 63 industrial sectors based on Taiwan’s input–output accounts—we established a mapping between the supply chain stages and relevant economic sectors. The correspondence is summarized in the table below.

It is important to note that not all activities within the corresponding sectors are directly attributable to the FSC, particularly in downstream stages. For instance, retail services encompass not only food-related transactions but also sales of non-food goods; the same applies to transportation and warehousing services. To more accurately reflect the environmental impacts and monetized outcomes attributable to the FSC, it is necessary to disaggregate the environmental burdens of these downstream activities based on their food-related share. In practice, this study applies the sales flow structure recorded in the input–output table to estimate the proportion of each relevant sector’s output that is dedicated to food supply. These proportions are then used as weighting factors to allocate the environmental impacts and corresponding monetized values accordingly.

For example, “Land Transportation” corresponds to the 38th sector in Taiwan’s 63-sector input–output table. The horizontal row associated with Sector 38 records the value of land transportation services consumed by all other commodity-producing sectors. Based on the summary provided in

Table 5. Data sources and processing for environmental impact categories ¹.

FSC Stage	Description of Activities	Commodity Sector Codes
Agricultural production	Crop production, fishery activities, and livestock production	1,2,3
Food processing	Food processing, tobacco manufacturing, and beverage manufacturing	6,7
Distribution/retail	Wholesale, retail, land and water transportation, delivery services, and warehousing services	36,37,38,39,40,41,42
Accommodation and foodservice	Accommodation services, food services	43,44

¹ Corresponding sector codes of the 63-sector EEIO-V model.

, which identifies the commodity sectors involved in each stage of Taiwan’s FSC, we can extract from the input–output table the total value of land transportation services provided to FSC-related sectors. This value is then divided by the total annual revenue of the land transportation sector, yielding the proportion of land transportation services consumed by the FSC. This proportion serves as the weighting factor described earlier, which is used to allocate the environmental impacts and corresponding monetized costs of specific sectors to the FSC.

4.2. Baseline Assessment of Environmental Impacts and Costs in Taiwan’s FSC

To establish a reference point for subsequent scenario simulations, this study first evaluates the environmental impacts and associated costs of Taiwan’s FSC within the broader economic system for the year 2021. The analysis focuses on three dimensions: (1) identifying pollution hotspots across the supply chain stages, (2) comparing environmental burdens relative to economic output, and (3) examining the composition of pollutant categories. This baseline assessment aims to clarify the key sources of environmental externalities, quantify their intensity, and inform policy prioritization.

4.2.1. Environmental Impact Contributions Across Supply Chain Stages

Based on the EEIO-V model estimations, Taiwan’s total economic output in 2021 across the 63-sector input–output framework reached approximately USD 1.62 trillion. Among these, sectors associated with the four stages of the FSC—agricultural production, food processing, distribution/retail, and foodservice—accounted for roughly USD 97.56 billion, representing 6% of total output. However, the environmental impacts attributable to the FSC were disproportionately high. For instance, food-related activities accounted for 16% of total PM_2.5 emissions, 43.9% of wastewater discharge, and 5.4% of greenhouse gas (GHG) emissions, all exceeding their economic share.

A closer examination of the contributions across different supply chain stages reveals that agricultural production is the primary environmental hotspot (Figure 2). Although this stage contributed only 1.4% to total output, it was responsible for 10.5% of PM_2.5 emissions, 37.7% of wastewater, and 13% of solid waste, highlighting the pollution-intensity of burning, fertilization, irrigation, and initial on-site processing activities in agriculture. Food processing also showed substantial contributions to GHG emissions (1.4%) and solid waste (13%). In contrast, the distribution/retail and foodservice stages had relatively lower environmental burdens but still contributed non-negligibly to PM_2.5 and GHG emissions, suggesting that end-use consumption and service delivery entail notable externalities.

These results illustrate that different supply chain stages play distinct roles across pollution categories. If carbon mitigation is prioritized, policy focus should shift toward the processing and consumption stages. Conversely, addressing water resource pressures would require targeting the agricultural sector. This stage-specific impact profile also supports the development of pollutant-specific strategies, such as promoting water reuse in agriculture, improving energy efficiency in food processing, and encouraging low-carbon dietary transitions—each aligned with corresponding environmental hotspots.

4.2.2. Environmental Cost Distribution Analysis

This study further analyzes the environmental costs associated with each stage of Taiwan’s food supply chain in the baseline year, using a monetized valuation approach, the results are illustrated in Figure 3. Overall, Taiwan’s economic activities generated a total environmental cost of approximately USD 163 billion in that year. Of this, the four stages related to the food supply chain accounted for about USD 15.56 billion, or 9.5% of the national total—significantly higher than their economic value share of 6%. This indicates that the food supply chain as a whole is a high environmental-burden sector, facing considerable pressure in terms of pollution control and environmental governance.

A closer examination of the composition of environmental costs reveals varying contributions across different pollution categories (Figure 4). Several key findings are summarized below:

• Health damages from PM_2.5 represent the single largest component of total environmental costs, accounting for 40.6%. The FSC contributes 16% of this amount, underscoring the prominent exposure risks associated with agricultural production and foodservice activities.
• Greenhouse gas (GHG) emissions are the second-largest contributor, comprising 30% of the total environmental costs. The FSC accounts for 5.4% of this category, indicating its notable carbon footprint.
• Wastewater and solid waste are largely concentrated in the upstream stages of the food system—namely, primary agricultural production and food processing. While these two pollutants constitute 0.9% and 1.4% of total environmental costs, respectively, their shares within the FSC reach 43.9% and 14.9%, highlighting substantial opportunities for waste management and resource recovery interventions.

Figure 4. Distribution of Environmental Cost Categories by Stage of the FSC (in million USD).

Figure 4. Distribution of Environmental Cost Categories by Stage of the FSC (in million USD).

Overall, among the four stages of the FSC, agricultural production incurred the highest environmental costs, amounting to USD 8.45 billion, which accounts for 54.3% of the total environmental costs associated with the FSC. This highlights it as a priority hotspot for environmental cost reduction strategies. The accommodation and foodservice stage follows, with USD 3.75 billion (24.1%) in environmental costs, primarily attributed to food waste management, energy consumption, and end-user behaviors. These findings provide important insights for policymakers to tailor environmental management strategies based on the distinct cost structures of each supply chain stage.

4.2.3. Comparative Analysis of Environmental Costs and Economic Output

This study further contrasts the economic and environmental structures of Taiwan’s FSC by examining the share of total environmental costs and the intensity of externalities per unit of output (Figure 5). Although the FSC accounts for only 6% of total economic output, it contributes to 9.5% of total environmental costs (approximately USD 15.56 billion), indicating its classification as a high environmental intensity sector. Among the four stages, agricultural production (food raw materials) exhibits the most pronounced imbalance: it contributes merely 1.4% of total output yet is responsible for 5.2% of total environmental costs (USD 8.45 billion), the highest intensity across all stages. This asymmetry highlights the disproportionate externalities associated with agricultural activities, suggesting that in the absence of environmental cost internalization, the social costs of agricultural production are systematically underestimated.

The bubble chart analysis in Figure 5 enables a comparative examination of environmental costs and economic output across the four stages of Taiwan’s FSC. Key insights are summarized as follows:

• Agricultural Production: A High-Density Environmental Hotspot
  Located in the upper-right quadrant of the figure, Agricultural Production accounts for only 1.4% of economic output but contributes a disproportionately high 5.2% share of total environmental costs, amounting to USD 8.452 billion—the highest among all sectors. This indicates that agricultural production is a “low economic output–high environmental burden” sector, representing a typical high-pollution-intensity hotspot. As such, it should be prioritized as a key intervention target in environmental policy design.
• Accommodation and Foodservice: Second-Largest Pollution Scale with Moderate Efficiency
  Located near the center of the figure, the Accommodation and Foodservice sector accounts for 1.8% of total economic output but contributes 2.3% of the total environmental cost, amounting to USD 3.754 billion, making it the second-largest contributor to environmental costs among food-related sectors. This highlights the critical potential for food waste reduction interventions, particularly at the consumption end. Behavioral strategies—such as surplus food reduction programs and consumer education—should be emphasized to address this sector’s environmental burden.
• Food Processing: A Manufacturing Stage with Relatively High Eco-Efficiency
  Although the Food Processing sector contributes 2.3% of the total economic output, its share of environmental costs is only 1.2% (equivalent to USD 1.888 billion). This indicates that while this sector is not the lowest in terms of environmental burden, it demonstrates relatively higher eco-efficiency when compared to the economic value it generates.
• Distribution and Retail: Emphasis on Improving Eco-Efficiency
  The Distribution/Retail sector is positioned in the lower-left quadrant of the diagram, indicating both low economic output (0.5%) and low environmental cost share (0.9%), with the smallest total environmental cost among the four stages (USD 1.468 billion). Future management strategies for this stage should focus on enhancing economic output while maintaining effective control of environmental burdens, thereby improving its overall eco-efficiency.

4.3. Simulation of Environmental Impacts and Benefits Under Taiwan’s FLW Reduction Target

In Taiwan’s approach to FLW reduction, national planning generally aligns with the target set by UN SDG 12.3, which aims to: “by 2030, halve per capita global food waste at the retail and consumer levels and reduce food losses along production and supply chains, including post-harvest losses.” To establish clear management objectives, it is essential to first define the current status of FLW in Taiwan. However, official statistical data on food loss and waste at different supply chain stages is currently lacking, necessitating estimations.

This study adopts the APEC-FLOWS database to calculate Taiwan’s FLW in the baseline year of 2021. APEC-FLOWS is a research and information-sharing platform established under a multi-year program supported by the Asia-Pacific Economic Cooperation (APEC). The initiative aims to promote food loss and waste prevention and reduction across APEC economies through scientific, standardized quantification methodologies and databases, thereby encouraging collaboration and knowledge exchange between the public and private sectors [46]. Participating teams from various APEC economies have co-developed a harmonized FLW accounting methodology, based on a mass flow model, which enables estimation across 19 food categories and all stages of the food supply chain (FSC)—from production, processing, and distribution to consumption. This ensures cross-country comparability of FLW estimates.

Users can obtain several key datasets from the APEC-FLOWS database—including the Food Balance Sheet, Loss Ratios, and Allocation and Conversion Factors—to compute country- and year-specific FLW outcomes based on the standardized methodology.

This study applies the APEC-FLOWS datasets and methodology to aggregate production, consumption, and loss data for 19 food categories across the four FSC stages. Using this approach, we calculate Taiwan’s 2021 FLW quantities by stage, with results summarized in the following

Table 6. Estimated Loss Rates by Stage of the Food Supply Chain in Taiwan (2021) ¹.

FSC Stage	Loss and Waste (kt)	Production/Consumption (kt)	FLW Rate (%)
Agricultural production	1499.3	11,390.9	13.2%
Food processing	303.8	8207.2	3.7%
Distribution/Retail	512.9	16,731.0	3.1%
Accommodation and food-service	1379.5	16,218.1	8.5%

¹ The APEC-FLOWS database contains data on 19 major food categories. To focus the analysis on the stages of the FSC, this study aggregated the data by FSC stages and applied them in the EEIO-V model for simulation. Readers are interested in the detailed FLW results of these 19 food categories in Taiwan for 2021—including production volumes, losses and waste quantities, and FLW ratios—may refer to Appendix A Table A2. Source: Estimates were calculated based on the APEC-FLOWS methodology using original data from the database.

.

Subsequently, this study designs four distinct scenarios, each corresponding to a 50% reduction in food loss and waste (FLW) at one of the four stages of the food supply chain (FSC), in alignment with the reduction targets of SDG 12.3. Specifically, Scenario A assumes a 50% reduction in FLW during the agricultural production stage; Scenario B targets a 50% reduction during the food processing stage; Scenario C applies a 50% reduction during the distribution and retail stage; and Scenario D focuses on a 50% reduction during the accommodation and food service stage. Based on the simulation results from the EEIO-V model, the study examines the environmental impacts and cost differences arising from FLW reduction interventions at each individual FSC stage.

Considering the relationship between production and consumption structures within the food supply chain (FSC), this study adopts the approach proposed in the existing literature [18]. It assumes that reductions in FLW at the agricultural production, food processing, and distribution/retail stages (corresponding to Scenarios A, B, and C) primarily affect the reduction in intermediate inputs—i.e., reductions in raw material inputs required during the production and service processes. In contrast, a reduction in FLW at the accommodation and food-service stage (Scenario D) is assumed to result in a decrease in final demand for food-related products, including primary agricultural products, processed foods, beverages, and tobacco. Specifically, for each scenario, the reductions in intermediate inputs or final demand are calculated based on Equation (11) and then incorporated into the EEIO-V model to simulate the corresponding environmental impacts and environmental costs.

$$d_{new}=d_{baseline}\ast (1+p_{FSC}\ast \left({\displaystyle \frac{1-W_{baseline}}{1-W_{baseline}\ast \left(1-r\right)}}-1\right))$$

(11)

In Equation (11), d_new represents the updated intermediate input coefficients or final demand values under each scenario, reflecting a 50% reduction in FLW. W_baseline denotes the baseline-year FLW rate for each product sector; r is the reduction target ratio set by SDG 12.3 (i.e., 50%); and P_FSC refers to the proportion of a specific product sector’s output that is used within the FSC. For example, agricultural products are assumed to be entirely allocated to the FSC, hence P_FSC = 1; whereas for sectors like warehousing or transportation, only a portion of their services is directed to the FSC, resulting in P_FSC < 1 [18].

Specifically, in Scenarios A, B, and C, Equation (11) is applied to calculate the proportional reduction in intermediate inputs across relevant product sectors (as listed in Table 5). These revised intermediate input coefficients are then used to update the EEIO-V model for further simulation analysis. For Scenario D, the equation is applied to estimate the change in final demand for food-related products, and the updated final demand vector is directly input into the EEIO-V model to conduct the simulation of environmental impacts and costs.

4.3.1. Analysis of Environmental Cost Reductions by FSC Stage

According to the simulation results, all four reduction scenarios (Scenarios A–D) demonstrate the potential to reduce environmental external costs associated with the food supply chain (FSC) through FLW reduction. However, the policy effectiveness varies significantly across scenarios (see Figure 6).

In terms of total environmental cost reduction, Scenario D—representing a 50% reduction in FLW in the accommodation and food-service stage—achieves the most substantial effect, reducing external costs by USD 464 million. This highlights the considerable benefit of prioritizing downstream FLW interventions. In contrast, Scenario B—targeting a 50% reduction in food processing stage FLW—yields the smallest reduction, with only USD 94.9 million in cost savings. Ranking the scenarios by effectiveness yields the order: Scenario D > C > A > B. These findings imply that, although the upstream stages (e.g., agriculture and primary production) tend to be pollution-intensive, targeting downstream stages (i.e., distribution/retail and accommodation/food-service) may offer greater potential for reducing environmental costs under Taiwan’s FSC structure.

A key advantage of the EEIO-V model lies in its ability to capture intersectoral interactions across the economic system. FLW reduction in a specific FSC stage triggers ripple effects through the production supply chain, indirectly influencing all four FSC stages. In this sense, the EEIO-V framework allows us to adopt a general equilibrium analysis perspective, enabling the comprehensive assessment of both direct and indirect impacts of FLW interventions and thereby informing more efficient policy design.

Nevertheless, it is important to note that the total environmental external cost associated with Taiwan’s FSC in the baseline year amounts to USD 15,562.5 million. The reduction achieved by each scenario ranges from 2.98% (Scenario D) to 0.61% (Scenario B), suggesting that FLW reduction alone provides only limited mitigation potential. To further reduce the environmental burden of the FSC, end-of-pipe pollution control measures remain essential. In summary, while FLW reduction serves as a meaningful and constructive strategy for mitigating FSC-related environmental impacts, it cannot substitute for traditional regulatory approaches in pollution control.

4.3.2. Analysis of Environmental Cost Reductions by Pollution Type

By examining the composition of reductions across six types of pollutants—PM_2.5, SOx, NOx, wastewater, solid waste, and GHGs—this study further analyzes the environmental improvement outcomes under different FLW reduction scenarios.

Figure 7 illustrates the reduction in environmental costs associated with each pollutant across the four scenarios, enabling a comparison of the environmental benefits stemming from FLW reduction at different FSC stages. Overall, the total environmental cost savings across the scenarios range from USD 94.9 million to USD 464 million, underscoring the substantial variation in effectiveness depending on which FSC stage is targeted.

First, PM_2.5 stands out as the pollutant with the greatest potential for cost reduction. If all four scenarios are implemented simultaneously (i.e., a 50% FLW reduction at all FSC stages), the combined benefit reaches USD 765.3 million—significantly higher than for any other pollutant. Notably, Scenario D (accommodation and food-service stage) alone accounts for USD 339.9 million of this amount, indicating that reducing food waste at the point of final consumption yields the most efficient environmental benefits for PM_2.5.

Next, for greenhouse gases (GHGs), the total reduction in environmental costs across all scenarios is USD 147.1 million. Scenarios C (distribution/retail) and D contribute the most—USD 68.7 million and USD 58.1 million, respectively—highlighting the importance of managing carbon emissions during downstream logistics, storage, and waste treatment processes.

In contrast, the overall benefits in terms of wastewater (USD 1.6 million) and solid waste (USD 24.4 million) reduction are relatively limited, with no single scenario demonstrating a dominant contribution. This implies that although FLW reduction can help ease some pressure on wastewater and solid waste systems, the associated unit environmental costs are comparatively low.

Furthermore, in terms of SOx, NOx, and GHGs, Scenario C yields the most substantial cost reductions (SOx: USD 26.2 million; NOx: USD 83.3 million; GHGs: USD 68.7 million). This suggests that the retail/distribution stage involves extensive logistics and storage operations, which are typically energy-intensive and reliant on fossil fuels, thereby exerting significant impacts on these environmental indicators.

In summary, from the perspective of total environmental cost reduction, Scenario D (USD 464 million) and Scenario C (USD 403.9 million) are the most effective intervention points. This reinforces the notion that downstream FLW management in the FSC offers the most substantial improvements in pollution mitigation and public health risk reduction. In contrast, while upstream reductions (Scenarios A and B) provide certain benefits, their overall effectiveness is more limited, emphasizing the importance and potential of end-of-pipe intervention strategies.

5. Policy and Management Implication

This study constructed an EEEIO-V model with monetization capability to simulate multiple food loss and waste (FLW) reduction scenarios and estimate corresponding changes in environmental impacts and costs. This analytical foundation not only provides Taiwan with a quantitative basis for implementing SDG 12.3 targets but also helps policymakers better understand the cost–benefit structures of various governance pathways. It serves as critical input for prioritizing policy interventions and allocating resources efficiently.

5.1. Monetization Function of the Model: Supporting Policy Benefit Analysis

One of the methodological innovations of this study lies in translating environmental impact quantities into monetary environmental measurement, thereby aligning the EEIO model with the language of policy decision-making. Compared to traditional LCA analyses that only present carbon emissions, water use, or pollutant quantities, this model enables the evaluation of the “economic return” of different mitigation strategies. For example, among the four simulated scenarios, Scenario D—where FLW is reduced by 50% at the accommodation and food-service stage—achieves the largest reduction in environmental costs, whereas Scenario B—reducing FLW by 50% at the food processing stage—delivers the smallest benefit. These monetized results assist decision-makers in identifying high-cost-effectiveness (high-CP) intervention points and facilitate cross-ministerial communication and fiscal agency evaluations of the return on mitigation investments.

5.2. Differentiated Pollution Profiles: Aligning Management Policies with Environmental Goals

The simulation results reveal significant differences in environmental pollution reduction benefits across FLW reduction stages. Overall, reducing FLW at downstream stages (retail/distribution and accommodation and food-service) yields substantially greater environmental benefits compared to upstream stages (agricultural production and food processing). Based on these findings, this study proposes the following policy and management recommendations as follows.

(1)

Prioritize waste reduction at the food-service stage

The accommodation and food-service stage demonstrates the highest potential for environmental cost reduction. Cutting waste at this stage not only directly curbs end-consumer food waste but also indirectly drives reductions in upstream resource use and pollution loads across the entire supply chain. Therefore, policy resources should be primarily directed toward minimizing consumer-end food waste. Suggested measures include:

• Enhancing meal preparation management in restaurants and institutional catering,
• Promoting public engagement through “food-saving” civic actions and surplus food recovery initiatives,
• Establishing dynamic pricing and promotional systems for near-expiry products.

(2)

Improve efficiency in product preservation and distribution at the retail level

FLW in the retail/distribution stage also leads to substantial upstream resource losses, especially during packaging and transportation. Management strategies may include:

• Promoting smart logistics and cold chain systems,
• Encouraging retailers to adopt food-sharing platforms or resale channels,
• to reduce transport-related food loss and the environmental impact of fuel consumption.

(3)

Develop pollution-type-specific governance mechanisms

From a pollution composition perspective, air pollution and greenhouse gas emissions represent the main sources of environmental cost reductions. It is recommended that future FLW management policies be integrated into the national net-zero strategy and air pollution control frameworks. Doing so would enable FLW reduction efforts to contribute synergistically to Taiwan’s climate and air quality goals.

5.3. Cross-Sectoral Governance: Advancing Horizontal Integration and Resource Coordination

FLW governance involves multiple ministries, including agriculture, economics, environment, and health. However, Taiwan currently lacks a cross-agency institutional framework to coordinate FLW policies. This study reveals stage-specific disparities in contributions to total environmental costs. Recognizing these differences can guide the assignment of governance responsibilities and the strategic allocation of resources. Policy recommendations include: (1) establishing an inter-ministerial FLW governance platform to coordinate regulatory tools and funding; (2) introducing a department-specific environmental cost feedback mechanism to inform subsidies, taxation, and investment; and (3) developing modular policy instruments tailored to pollution types, enhancing both integration and flexibility.

5.4. Aligning with Global Trends and Addressing Local Challenges

Several countries have already implemented laws, quantitative targets, and incentive mechanisms to tackle FLW. By contrast, Taiwan’s policy framework remains fragmented and data deficient. This study suggests Taiwan should focus on:

(1)

Enhancing statistical data collection and establishing a practical roadmap for stage-specific FLW monitoring and simulation. Specifically, Taiwan could build on the experiences of the FAO Food Balance Sheet and the APEC-FLOWS methodology by gradually embedding FLW monitoring into existing agricultural and industrial statistical frameworks. For instance, the current food supply-demand balance statistics could be expanded to include surveys on production losses, processing losses, and retail/household waste. In the medium to long term, a cross-agency stage-specific FLW monitoring system could be established to ensure alignment with international standards and support domestic policy planning.

(2)

Developing benefit-based incentive mechanisms (e.g., subsidies linked to reductions in environmental costs).

(3)

Incorporating fiscal and legal instruments to encourage behavioral changes across sectors.

The EEIO-V model provides a robust quantitative tool and policy sandbox to support the development of science-based FLW governance strategies and facilitate Taiwan’s alignment with global sustainability practices.

6. Conclusions and Future Research Suggestions

This study develops an EEIO-V model to assess the environmental costs associated with food loss and waste (FLW) across different stages of Taiwan’s food supply chain. The model further simulates the potential environmental benefits and sectoral contributions under FLW reduction scenarios aligned with the United Nations’ SDG 12.3 target. By incorporating a monetization mechanism, the model translates multiple environmental impact categories into policy-relevant cost indicators, thereby offering concrete analytical support for governmental FLW management efforts.

The simulation results show that downstream FLW reduction strategies—specifically Scenario C (distribution/retail) and Scenario D (accommodation and food-service)—yield significantly higher potential in total environmental cost savings compared to upstream strategies (Scenarios A and B). This finding underscores the pivotal role of food distribution and final consumption in reducing the environmental burden of FLW. Moreover, by integrating a life cycle perspective with monetized results, the EEIO-V model can be further expanded for broader sustainability governance applications, such as resource recycling or total pollution cap management.

Nonetheless, this study also faces several limitations. First, the model is built on Taiwan’s 2021 industrial and environmental data, and more precise projections for 2030 would require the integration of structural adjustments or trend-based forecasting mechanisms. Second, while this study offers an initial sector-specific policy focus analysis, it lacks in-depth discussion on implementation costs, policy feasibility, and behavioral responses. Future research should bridge these gaps by incorporating institutional and social-behavioral dimensions.

In light of these observations, future research may proceed in the following directions:

• Model Function Expansion: Incorporate dynamic input–output structures and price elasticity modules to enable mid- to long-term policy simulation and dynamic assessments.
• Institutional Design Integration: Link the model with carbon pricing mechanisms, waste treatment fee structures, or green public procurement policies to formulate incentive-aligned, multi-instrument governance portfolios.
• Behavioral Linkages: Combine behavioral economics and social survey data to assess public acceptance and implementation feasibility of consumer-end reduction strategies.

It is worth noting that the EEIO-V model also holds strong potential for integration with dynamic policy instruments. For instance, it could be linked to carbon pricing schemes to reflect real-time variations in external costs across industries or be embedded into adaptive subsidy and tax mechanisms that adjust policy intensity based on pollutant types and temporal conditions. Moreover, EEIO-V can serve as a real-time information platform, providing monetized environmental cost estimates to support timely policy adjustments. These applications highlight that EEIO-V is not only a static analytical tool but also a feedback-oriented and adaptive decision-support system for policy governance.

Through these enhancements, the EEIO-V model is expected to evolve into a strategic tool supporting Taiwan’s FLW policy planning and cross-sectoral sustainability governance—contributing to the global agenda for improving food resource efficiency and ensuring environmental sustainability.