Carbon flow map: A substance flow analysis of carbon in the Dutch economy

1 Carbon flow map: A substance flow analysis of carbon in the Dutch economy Authors: Juraj Petrík, Gert Jan Kramer, Vivian S.C. Tunn, Li Shen Copernicus Institute of Sustainable Development, Utrecht University, Vening Meinesz Building A, Princetonlaan 8a, 3584 CB Utrecht, Netherlands Corresponding author(s) email: [email protected], [email protected] Abstract Carbon is a fundamental component of our economy, as well as a key element of greenhouse gases such as carbon dioxide and methane. Tracking carbon flows captures energy, material, and climate impact, and it enables a direct examination of the link between material and energy transitions, indicating the progress towards circular carbon and the net-zero target. This study presents a novel method for mapping carbon flows at the macro level using national accounts data, while also enabling granular analysis of carbon flows across different sectors. Applied to the Netherlands as a case study, the results reveal that the Dutch economy strongly relies on imports and exports of carbon, with net imports accounting for 131 Mt C (82% of total carbon supply) and net exports for 89 Mt C (83% of total carbon demand) in the year 2020. Furthermore, over 80% of the carbon supply is fossil-based. Therefore, to meet the Dutch government’s 2050 targets of climate neutrality and full circularity, i.e. relying solely on biogenic and other circular carbon sources, significant shifts in carbon supply and demand are urgently needed. Keywords: circular economy, bioeconomy, material transition, fossil fuels, climate neutrality, Netherlands 2 1. Introduction Carbon is a fundamental element connecting energy, materials, and climate, as it is a building block of various economic inputs and outputs, including fuels, chemicals, plastics, biomass, and other manufactured materials, as well as an element constituting greenhouse gases (GHGs) such as carbon dioxide (CO₂) and methane (CH4) (vom Berg et al. 2023; Delahaye and Mosterd 2024). Rising concentrations of anthropogenic carbon in the atmosphere, largely driven by fossil fuel use, have intensified the greenhouse effect and accelerated climate change, posing risks to ecosystems, biodiversity, and human societies (Delahaye and Mosterd 2024; Hanemaaijer et al. 2023; PBL 2024). At the same time, today’s economic model, based on a linear “take-make-use-dispose” approach, has accelerated resource depletion, energy demand, and waste generation (Mayer et al. 2019; Tan and Lamers 2021). This unsustainable trajectory has resulted in the exponential exploitation of planetary resources, and current projections suggest that by 2050, nearly all planetary boundaries will be exceeded (Bachmann et al. 2023; van Vuuren et al. 2025). Two agendas, interconnected yet often addressed separately, aim to tackle these challenges: the transition to a circular economy (CE) and achieving climate neutrality (PBL 2024). CE strategies focus on increasing resource efficiency by reducing material inputs, extending product lifetimes, maximising reuse and high-quality recycling, and substituting fossil feedstocks with renewable or secondary resources (Hanemaaijer et al. 2023). The Dutch government, for example, has set an ambitious goal of achieving a fully circular economy by 2050, with an interim objective of halving the use of primary abiotic raw materials by 2030 compared to 2016 levels (Hanemaaijer et al. 2023; IenW 2023). In parallel, the Dutch Climate Agreement addresses national climate neutrality targets, prioritising reductions in GHG emissions across sectors such as electricity generation, industry, transport, the built environment, and agriculture and land use (EZK 2019). Although both agendas share the overarching aim of reducing dependence on fossil carbon, the link between them is not always made explicit (PBL 2024). Research on material flows has become a key tool for exploring this link. Efforts to map material flows began in the 1990s with the development of economy-wide Material Flow Accounting (EW-MFA), which is based on national accounts data. EW-MFA has since become a widely recognised and internationally standardised method for describing how economies import materials and energy and release waste to the environment (Baars, Rajaeifar, and Heidrich 2022; Chen et al. 2022; Delahaye, Tunn, and Tukker 2023). However, EW-MFA focuses on overall material inputs and outputs for the economy as a whole and does not provide details at the industry or sector level (Kovanda 2019).zTo address these limitations, more detailed approaches were developed, such as physical supply-use and input-output tables at national and global scales (Delahaye, Tunn, et al. 2023; Kovanda 2019). Circularity indicators, such as resource efficiency, substitution rates, the circular material use rate, and the share of biobased resources, are often derived from these material flow data (Delahaye, Tunn, et al. 2023; Hanemaaijer et al. 2023). Studies have been carried out at global (Haas et al. 2015; Krausmann et al. 2009, 2018), regional (Mayer et al. 2019), and national levels (Delahaye, Tunn, et al. 2023; Kovanda 2021), helping to track the drivers of resource use and environmental impacts. In addition, researchers have conducted targeted material flow studies for specific resources. 3 Plastics, for example, have been mapped in detail across the EU countries, covering production, consumption, waste, and recycling (Amadei, Rigamonti, and Sala 2023; Eriksen et al. 2020; Hsu, Domenech, and McDowall 2021; Lobelle et al. 2024). Similarly, the chemical sector, a major industrial energy user and GHG emitter, has been analysed through global mass flow linking fossil feedstocks to chemical products (Levi and Cullen 2018). Yet, despite carbon’s dual role as both an essential material and the main driver of climate change, little attention has been given to explicitly mapping carbon flows at the macro level. To address this gap, the concept of a ‘circular carbon economy’ has been proposed, with the central aim of decoupling industrial processes, particularly in carbon-intensive sectors such as chemicals and materials, from fossil carbon sources. By integrating material and climate agendas, the concept highlights opportunities to replace fossil carbon with biogenic, recycled, or captured carbon that remains within economic loops (Moser et al. 2022; Newman et al. 2023; Tan and Lamers 2021; Vidal et al. 2024; Zeilerbauer et al. 2024). Yet only a few studies have attempted to systematically trace carbon flows on a national or macro scale (Gunnarsson 2022; Ma et al. 2012). Some scholars have instead approached the issue through carbon footprint analysis based on input–output tables (Andrieu et al. 2024; Tukker, Wood, and Schmidt 2020). In the Netherlands, the only comparable effort has been the carbon accounting monitor compiled by Statistics Netherlands (Centraal Bureau voor Statistiek, CBS) (Delahaye and Mosterd 2024). However, the CBS report distinguishes presented carbon flows only by origin (fossil or biogenic) and provides limited detail on methodology, sectoral flows, intersector connections, or the types of carbon-containing products involved. This work aims to address existing gaps by presenting a novel method for mapping carbon flows at the macro level, referred to as the carbon flow map (CFM), which utilises national accounts data and provides detailed granularity across different sectors and carbon-product categories. By tracing carbon inflows, internal transfers, and outflows across the socioeconomic system, the framework aims to provide a detailed and coherent view of carbon’s role in the economy and its environmental implications. We apply this approach to the Netherlands with four main objectives: (1) to evaluate the national economy’s dependence on carbon imports and exports, (2) to quantify societal carbon consumption, (3) to trace circular carbon flows, and (4) to identify potential sources of carbon that could be utilised in future circular strategies. The resulting analysis will provide a detailed, policy-relevant perspective on the role of carbon in the Dutch economy, offering insights that are directly applicable to both circular economy initiatives and climate policy. 4 2. Materials and methods This study proposes a methodological framework, referred to as the carbon flow map (CFM), that integrates existing statistical data on material flows, converts them to carbon flows, and quantifies the approximate substance flow of carbon across different sectors in the national economy. The following sections outline the framework in detail. 2.1 Geographical and temporal scope, data description The geographical scope of our CFM analysis focused on the Netherlands as a case study. The Netherlands has a well-developed monitoring of physical material flows, which is carried out through the Material Flow Monitor (MFM), a comprehensive framework developed by CBS, based on concepts and definitions of the System of Environmental-Economic Accounting (United Nations 2014). The data sources used in the MFM include monetary flow data based on national accounts supply and use tables, as well as physical flow data from Physical Energy Flow Accounts, international trade statistics, agriculture statistics, waste statistics, material flow accounts, and air emission accounts. Thus, the MFM integrates statistical data from monetary sources, which are converted into physical material flows, along with data from physical sources, to create a consistent and economy-wide representation of material flows in the Dutch economy. A detailed methodological description of the MFM framework is provided by Delahaye, Tunn, et al. (2023). The outcome of the MFM framework is a dataset, structured as physical supply and use tables (P-SUT), which systematically maps the supply and use (in million kilograms per year) of approximately 400 product categories, 16 waste categories, 10 natural resources, and CO₂ emissions across approximately 130 economic sectors. It also covers the supply and use in key categories such as imports, domestic extraction, consumption by households and public institutions, exports, accumulation, and the environment. In general, the supply table reports carbon sources entering the system, such as imports and domestic extraction, and the use table details the destinations of carbon, including exports, stock changes, and emissions to the environment. Both tables provide data on the supply and use of raw materials, products, waste and emissions by economic sectors, households, and public institutions. Our analysis was based on data from the MFM, with the temporal scope set to the year 2020, chosen for the completeness and consistency of the available MFM dataset (available upon request from CBS). Although 2020 was impacted by the COVID-19 pandemic, we did not expect significant disruptions in the outcomes of our analysis. This is because, as the results later demonstrate, carbon flows in the Netherlands were largely driven by fossil fuels, with existing trends showing a gradual decline in their use even before the pandemic. Therefore, we expect these trends to have had a more significant influence on overall carbon flows than the temporary effects of the pandemic. 2.2 Approach to developing CFM This section describes an approach to developing CFM, based on the conversion of material flows from the Dutch P-SUT for 2020. The main steps of the approach are summarised in Table 1 and elaborated in the following sections. A full list of the MFM product categories and sectors, 5 their classification within the CFM model scope, and the applied carbon conversion factors is provided in the Supplementary Materials. Table 1. Overview of steps to develop the carbon flow map (CFM). Step no. Step Description Step 1 CFM scope definition The CFM scope was defined by selecting the sectors and carbon flow categories to be monitored. The resulting scope is shown in Figure 1, with the selected carbon flow categories listed in Table 2. Step 2 Aligning MFM classifications with CFM scope All items reported in the MFM’s P-SUT, i.e. product categories, physical extensions (e.g. waste categories, CO2 emissions, etc.), and sectors, were assigned to carbon flow categories and sector clusters (defined in Step 1) to align with the CFM scope. Step 3 Conversion of material to carbon flows For each product category from MFM’s P-SUT, a carbon conversion factor (CCF) was developed to convert material quantities into carbon quantities. The MFM’s P-SUT were then transformed into the carbon supply and use tables and aggregated by the carbon flow categories and sector clusters. Step 4 CFM modelling The resulting data from Step 3 were used to reconstruct carbon flows in the Netherlands for 2020, in line with the defined CFM scope. Additional modelling was carried out to estimate carbon flows not directly monitored by the MFM. 2.2.1 Step 1: CFM scope definition Although the MFM’s P-SUT offer detailed information on the supply and use of materials within individual sectors, their overall structure did not fully support our objectives for tracing carbon flows, particularly between sector clusters. Therefore, we developed a tailored scope for the CFM model to address these limitations, enabling the quantification and visualisation of carbon flows across sectors and providing enhanced insights to better monitor carbon circularity. Figure 1a outlines the scope of the CFM, defining its accounting boundaries and highlighting the key system components along with the carbon flows connecting them. In the figure, system components are represented as boxes, with arrows indicating carbon flows between them. The system components are further classified into four categories: sources, transformation sectors, destinations, and functional elements. The definitions of these components and carbon flows are as follows: • Sources represent carbon inflows into the Dutch socioeconomic system via imports or domestic extraction. Furthermore, the model accounts for additional source system components such as ‘recovered carbon’, i.e. carbon recovered and recycled from transformation sectors, and ‘stocks(-)’, i.e. carbon supplied from short-term stocks and calculated as the net decrease in existing stock levels (e.g. reductions in national fuel reserves). • Transformation sectors are where carbon is processed or consumed and, thus, transformed from one carrier to another (e.g. from crude oil into refinery products, or fuels into carbon emissions). The CFM model recognises three clusters of transformation sectors: (1) agriculture, industry and other sectors, further referred to as the ‘economic sectors cluster', (2) waste management, and (3) households and public institutions. Furthermore, to enhance interpretability of the economic sectors cluster, 6 we further divided it into ten sub-clusters: (1) raw materials extraction, (2) manufacture of coke and refined petroleum products, (3) electricity, gas and steam supply, (4) transport, (5) manufacture of chemicals/pharmaceuticals – intermediate, (6) manufacture of chemicals/pharmaceuticals – end, (7) agriculture, forestry and fishing, (8) manufacture of food products, beverages and tobacco products, (9) other manufacturing sectors, and (10) retail, services and other non-manufacturing sectors. In general, these high-level economic sector sub-clusters align with those used in statistical classifications, as reported, for example, by Delahaye, Tunn, et al. (2023). However, waste management, classified in statistics under the economic sectors, was in our case treated as a distinct cluster due to the subsequent highlighting of recovered and recycled carbon flows originating from this cluster. Lastly, the overall scope of our model specified fewer individual economic sector sub-clusters than in statistics, as some manufacturing sub-clusters were aggregated where additional detail was not required for the purposes of the CFM analysis. • Destinations represent points where carbon exits the socioeconomic system, including exports, emissions to the atmosphere, and litter dispersed into the environment. Furthermore, stocks were also accounted for as destinations. We distinguished three types of destination-like stocks where carbon was stored: (1) short-term stocks, which reflect net increases in national reserves (i.e. 'stocks(+)'), (2) long-term stocks (i.e. ‘accumulation’), where carbon products are stored for more than one year, and (3) landfilled waste. It should be noted that, although the accumulation destination is classified as a distinct category from households and public institutions, it principally represents materials accumulated in these sectors. • Functional elements are auxiliary components, such as ‘carbon supply’ and ‘carbon demand’, introduced to support the structure and interpretation of the CFM by grouping specific flows. • Flows are represented by carbon-containing products grouped into relevant categories. We defined eleven main carbon flow categories to be monitored within the CMF model, covering all carbon-containing products, as presented in Table 2. In some cases, these categories were further subdivided to support more detailed interpretation and analysis. Each arrow in Figure 1a, specified by a flow code (e.g. S1), thus represents the total carbon flow between system components, reflecting the combined sum of values across eleven monitored carbon flow categories. 7 (a) (b) Figure 1. Carbon flow map (CFM) conceptual model: (a) overall scope of the CFM; (b) detailed scope of the CFM’s waste management. Boxes represent system components, and arrows represent carbon flows between them. 8 Table 2. Overview of the CFM’s carbon flow categories, the basis of their carbon conversion factors (CCFs) and estimated data uncertainties. Carbon flow category Carbon flow subcategory Included MFM product categories CCF (dry matter) estimation based on Estimated data uncertainty Uncertainty cause Hard coal and lignite n/a hard coal, lignite emission factor very low statistical errors Crude oil n/a raw crude oil, petroleum oils emission factor very low statistical errors Natural gas n/a natural gas emission factor very low statistical errors Refinery products n/a C1 – C4 hydrocarbons, fuels, blast furnace gas emission factor very low statistical errors Chemicals n/a organic chemicals, fertilisers, polymers, paints, household chemicals, pharmaceuticals molecular structure low statistical errors; representing chemical groups by their principal compound Biomass Raw, agricultural and other biomass crops, woody biomass, livestock, fish and seafood, animal products elemental analysis; average values for biomass applied moderate statistical errors; varying significant moisture content; average CCF applied Biomass for energy solid and liquid biofuels, biogas elemental analysis; average values for biomass applied moderate statistical errors; varying significant moisture content; average CCFs applied Food products and animal feed Food products processed food products (i.e. from raw agricultural biomass), beverages elemental analysis; average values for biomass applied moderate statistical errors; varying significant moisture content; average CCFs applied Animal feed fodder elemental analysis; average values for biomass applied moderate statistical errors; varying significant moisture content; average CCFs applied Other materials Minerals limestone, peat molecular structure; elemental analysis high statistical errors; estimated composition Rubber and plastic products tyres, plastic packaging, plastic products for end-use molecular structure low statistical errors; representing chemical groups by their principal compound Mixed materials textiles, wood, paper, construction materials, asphalt, steel, appliances, furniture, automotive elemental analysis; estimated product composition high statistical errors; estimated CCFs for some cases (e.g. appliances, furniture, etc.) Carbon emissions n/a carbon dioxide, methane molecular structure very low statistical errors Waste n/a various waste flows, e.g. paper, plastics, organic, mixed, etc. estimated waste composition high statistical errors; varying moisture and carbon content Recycled materials n/a recycled plastics, paper, etc. see plastics, paper, etc. moderate statistical errors; product variability; average CCFs applied 9 2.2.2 Step 2: Aligning MFM classifications with CFM scope To align MFM’s product category and sector classifications with the CFM scope, we systematically reviewed approximately 400 product categories and 130 sectors from the P-SUT. Each product category was assigned to one of eleven defined CFM carbon flow categories (as described in Table 2). For example, gasoline and diesel were assigned under refinery products, alcohols and aromatic compounds under chemicals, etc. The carbon emission flow category included both carbon dioxide and methane emissions, covering not only process-related emissions but also those from livestock and human respiration, as these are reported in statistics for balancing purposes. Some carbon flow categories were further divided into sub-categories for better interpretation of results (e.g. biomass for energy purposes was distinguished from other biomass types). The carbon flow sub-category ‘mixed materials’ included a wide variety of products with diverse structures and carbon contents (e.g. steel, textiles, furniture, appliances, etc.). Despite this diversity, these carbon flows were aggregated into a single category, as representing them individually would produce numerous minor flows that are relatively small compared to other categories, and they would be difficult to distinguish in a visualisation. Similarly, each sector from P-SUT was assigned to sector clusters included in the CFM scope (see Figure 1a). Sectors assigned to the economic sector cluster were further organised into ten sub-clusters, as described above. The households and public institutions cluster, where carbon is predominantly consumed, included sectors such as households, government buildings, education, and healthcare, as well as public spaces like libraries, theatres and museums. Besides households, these entities primarily provide services, and their operations result in associated carbon flows through material consumption on the input side, and waste generation and emissions on the output side. A similar function can be observed in service sectors. However, the key difference is that these sectors primarily provide services for profit through commercial activities or combine public and private functions (e.g. sports facilities). As such, service sectors were categorised under the economic sectors cluster, specifically within the sub-cluster of retail, services and other non-manufacturing sectors. Regarding transportrelated carbon flows, the data in P-SUT did not distinguish between personal, public, and freight transport in all cases. As a result, we grouped all transport-related flows under the subsector ‘transport’. However, P-SUT reported fuel consumption within the households category, which suggests that personal passenger car transport is accounted for under the households and public institutions cluster. 2.2.3 Step 3: Conversion of material to carbon flows Carbon conversion factors (CCFs) were developed for product categories in the P-SUT to convert material quantities into corresponding carbon flows. However, several product categories were excluded from this process for the following reasons: (1) product category had reported zero value for the year 2020 (i.e. zero carbon flow), (2) the product category was included in the dataset for statistical balancing, making it impossible to accurately estimate its carbon content, or (3) the product category did not contain carbon (e.g. material flow of aluminium). 16 (a) (b) Figure 3. Detailed view on selected parts of the carbon flow map (CFM) for the Netherlands in 2020: (a) breakdown of the economic sectors cluster (only flows larger than 0.5 Mt C); (b) zoom-in on the waste management cluster. Note that numbers may not always add up to the total due to rounding. 17 4. Discussion 4.1 Understanding circularity and carbon neutrality through carbon flows Our analysis of Dutch carbon flows revealed two critical insights. First, the economy was predominantly fossil-based, with natural gas, crude oil and their derivatives making up a significant portion of the total carbon supply. Second, the Dutch economy was also heavily reliant on both carbon imports and exports, accounting for 83% and 82% of the total supply and demand, respectively. By examining carbon types, we found that approximately 19.5% of the total carbon supply was biogenic, while fossil carbon accounted for 80.1%, and mineral carbon for just 0.4%. In the total carbon demand, biogenic carbon represented 16.0%, fossil carbon 83.9%, and mineral carbon made up a very minor share of 0.1%. Biogenic carbon contributed approximately 17% of the total emissions, and its share in trade flows was similarly low (16.3% of net imports, 12.8% of net exports). In the two most carbon-intensive sectors, refining and intermediate chemicals production, biogenic carbon inputs were particularly low (just around 1 Mt C each, i.e. 1.6% and 3.6% of sectoral totals, respectively). These results underscore the limited role of biogenic carbon in the Dutch industry and the economy’s strong reliance on fossil resources, highlighting the challenges of advancing circularity and meeting the climate neutrality targets set by the Dutch government. This empirical reality also underscores a fundamental conceptual challenge: assessing circularity through carbon flows requires a precise definition of ‘circular carbon’. Previous studies showed a lack of consensus on defining the term circular carbon. Zeilerbauer et al. (2024) pointed out that the term is used inconsistently among scholars, most often in the context of carbon capture and utilisation (CCU) technologies and the recirculation of CO2 emissions, or alternatively within waste management. Other examples include Tan and Lamers (2021) focusing solely on biogenic carbon, or Meys et al. (2021) using the term in the context of achieving net-zero emissions from plastics. Haas et al. (2015) did not specifically focus on circular carbon, but discussed what ‘circular’ means for material flows of biomass. They noted that treating biomass as circular assumes it is produced renewably, which is often not the case. Applied to carbon, this implies that biogenic carbon cannot be considered circular unless its production is renewable. Moser et al. (2022) suggested that the concept of circular carbon can be used to merge two separate domains: (1) material resources and retention of carbon within the loop, and (2) climate change mitigation through the reduction of CO2 via the transition to sustainable energy systems. Yet, authors noted that a uniform definition has not yet been established. Building on these varying perspectives, we distinguish two complementary ways of defining circular carbon for the purpose of our analysis. 1. Narrow definition: circular carbon is limited to carbon recovered from domestic transformation sectors and carbon captured for utilisation (although the latter is negligible in this study), thereby avoiding release into the atmosphere. Based on this definition, only 3% of the total carbon supply would be considered circular in the Netherlands. 18 2. Broad definition: circular carbon includes all carbon from the narrow definition and also biogenic carbon, reflecting the view that carbon sourced from biomass can form part of a closed-loop system. Under this broader definition, approximately 21% of the total carbon supply would be considered circular. However, this estimate is indicative as it does not account for the sustainability of the biomass used, which remains a critical factor in determining whether biogenic carbon can be regarded as circular. Regardless of the definition applied, both perspectives highlight the magnitude of the transition challenge, as achieving carbon circularity and climate neutrality will require reducing fossil carbon dependence while simultaneously developing sustainable alternatives, including sustainably sourced biomass, recycled materials, or captured and utilised CO₂. While reducing fossil fuel use in the energy sector is essential, eliminating fossil carbon entirely remains highly challenging due to the complexity of defossilizing material production systems. The recycling sector illustrates these limitations. (Klotz, Haupt, and Hellweg 2023) estimated that only up to 31% of plastic waste can be recovered as secondary material through the mechanical recycling route. In another example, pyrolysis, a promising technology for plastics recycling, suffers from inherent carbon losses, which limit its potential to achieve full carbon circularity and fully replace fossil carbon (Petrík et al. 2025). These examples suggest that a single technology cannot deliver sufficient carbon recovery on its own. To address this, Lange et al. (2024) examined a cascade recycling system that integrates multiple recycling technologies, including both mechanical recycling and pyrolysis. Their analysis suggested that, even under such a system, only 60–70% of the carbon currently derived from fossil naphtha in the plastics sector could be substituted with recycled carbon by 2050. According to our CFM results, plastics (and to some extent rubber) accounted for around 3.5 Mt C, or 3% of total carbon demand. This implies that 2.1-2.5 Mt C of plastics would need to be produced through such an optimised cascade recycling system. This figure far exceeds the 0.6 Mt C of recycled plastics from the waste management sector in our analysis, underscoring substantial challenges of scaling up recycling and ensuring feedstock availability. To increase circular carbon in material production, even after maximising recycling, fossil carbon supply must be substituted with alternative sources. Among these, converting captured CO₂ into products like naphtha, methanol, or CO₂-based polymers would be technologically viable, but deployment is limited by high energy demands and costs, and environmental benefits are dependent on reliance on the energy mix and end-of-life treatment scenarios (Gonella et al. 2025; Lange 2021). Biogenic carbon from sustainably sourced biomass may offer a more feasible option, provided it is available in sufficient volumes and at costs competitive with conventional fossil-based feedstocks such as naphtha (Lange 2021). It plays a key role not only in advancing circularity, as discussed above, but also in achieving climate neutrality. According to the IPCC, reaching netzero greenhouse gas (GHG) emissions requires balancing anthropogenic emissions with anthropogenic removals over a defined timeframe (Zhai et al. 2018). By capturing CO₂ during biomass growth and storing it in biomass-based products, biogenic carbon provides a continuous source of removals that can delay atmospheric release and offset residual fossil emissions. 19 However, meeting the biomass demand at scale presents significant challenges. According to the Sustainable Industry Lab (Akerboom et al. 2023), by 2050, an additional 20 Mt C in the form of biomass and residual flows will be needed in the Netherlands to support the circular chemical industry and biofuel production. In contrast, domestic extraction of biogenic carbon in 2020 was around 6 Mt C, most of which is already allocated to agriculture and food sectors. This gap suggests that meeting future demand will depend heavily on imports. As our CFM analysis showed, the Netherlands is already heavily dependent on imports due to its exportoriented industrial economy. As long as this model remains unchanged, achieving a fully sustainable and circular supply of carbon feedstocks will remain a systemic challenge. Addressing it may require not only a technological shift but also a critical reassessment of the Netherlands’ economic dependence on large-scale carbon-intensive exports. Additionally, scaling up biomass use for material applications may introduce environmental trade-offs, including increased pressure on land use, land use change, and water availability (Pawelzik et al. 2013; Zuiderveen et al. 2023). These constraints could limit the sustainable supply of biomass, particularly as demand grows across multiple sectors. In such scenarios, CO₂-based carbon is likely to play an increasingly important role in long-term circularity strategies for substituting fossil carbon. 4.2 Model uncertainties The Methods section outlined the uncertainty associated with carbon flow data, which may affect the accuracy of specific flow categories. However, as the results are dominated by fossil fuels (i.e. categories generally associated with low uncertainty), we consider the overall findings to be robust. A full uncertainty analysis would therefore be unlikely to alter the main conclusions. Moreover, conducting a comprehensive uncertainty assessment across the wide range of product categories included in our study would present a significant methodological challenge. Therefore, in this section, we focused on addressing uncertainties arising from the model structure. As shown in Figure 2 and Figure 3, several sectors showed imbalances, labelled as ‘balance gap’ flows. The largest of these, each about 2 Mt C, occur in (1) retail, services, and other nonmanufacturing sectors, (2) agriculture, forestry, and fishing, and (3) households and public institutions. For retail, services, and non-manufacturing sectors, and households and public institutions, these imbalances likely reflect the accumulation of materials in long-term stocks (e.g. buildings, roads and other infrastructure, durable goods) that were not fully captured in statistics. In agriculture, uncertainty may result from unreported internal reuse of residuals, such as manure applied on farms (Delahaye, Tunn, et al. 2023). Some flows were directly derived from the MFM datasets (e.g. imports, exports), while others, such as carbon supply and demand, recovered carbon, intersectoral carbon use, and waste flows, were calculated in our model and are therefore subject to higher uncertainty. Figure 4 shows a distinction between flows derived from statistics (lower uncertainty) and those calculated (higher uncertainty). Waste flows illustrate these challenges particularly well. While total waste imports and production quantities were accurately reported, the origins and destinations of waste streams 20 remain unclear, limiting the ability to track whether waste is reused, incinerated, or otherwise treated. To address these gaps, we applied estimates based on mass-balancing and, where possible, compared results with independent sources. For example, the Ministry of Infrastructure and Water Management (Rijkswaaterstaat 2022) reported that 1083 kt of imported waste was incinerated in 2020, corresponding to approximately 0.4–0.8 Mt C. Our model calculated 0.8 Mt C, thus within the reported range from the independent source. Furthermore, imported waste represents less than 1% of net imports, meaning that even significant uncertainty in this calculated flow would have only a minor effect and would not alter the overall conclusions. For other waste flows, however, no comparable data were available. This highlights the broader challenge of limited transparency and systematic underreporting in waste statistics, an issue also emphasised by Lobelle et al. (2024). Such gaps constrain the accuracy of carbon flow assessments and underscore the need for improved monitoring and reporting practices. 21 (a) (b) Figure 4. Relative uncertainty in carbon flows within the carbon flow map (CFM) for the Netherlands in 2020: (a) overall CFM; (b) zoom-in on the waste management cluster. Flows derived from the Material Flow Monitor (MFM) (light blue) are associated with lower uncertainty, whereas flows calculated by the model (grey) are associated with higher uncertainty. 22 4.3 Comparison with other studies To validate our results, we compared them with the Dutch carbon accounting monitor published by CBS (Delahaye and Mosterd 2024), as no other study has mapped macro-level carbon flows for the Netherlands. Given the study’s different scope, we focused only on comparable flows. We expect that the main deviation is related to using our own CCFs, developed in parallel with the ones used by CBS. The carbon flow values in CBS's report were also rounded, which may account for part of the deviation. Comparisons were made for imports, domestic extraction (fossil fuels and biomass), carbon supply, waste flows, recycled carbon, exports, emissions, and accumulation. Close alignment was observed for imports, exports, carbon supply, and emissions, with differences of only 0.8–2.4%. Larger discrepancies arose for domestically cultivated biomass (21.3% lower in our analysis) and recovered carbon (24.4% lower), which may reflect both rounding effects and a possible underestimation of biomass carbon content in our study. The largest divergence was observed for accumulation, where our estimate was 70% lower. This difference stems partly from a different scope, as we separated short-term stocks and landfill from accumulation. It may also reflect our earlier assumption that part of the balance gap from economic sectors, households, and public institutions is retained in stocks, but not accounted for under the accumulation category. Overall, the close alignment across most categories, despite methodological differences, demonstrates the robustness of our findings. 23 5. Conclusions In this study, we developed a novel method to map carbon flows at the macro level using national accounts data, distinguishing between product categories, including waste flows, and tracing these flows across different sectors and sub-sectors. Applied to the Netherlands, this approach enabled a detailed analysis of carbon-containing materials and their movement throughout the socioeconomic system. Carbon, as a tracer element, offers particular value in assessing circularity and climate neutrality because its mass balance is preserved across transformation processes. By quantifying and visualising carbon flows, including the share of biogenic carbon, the method provides a practical tool for monitoring progress while also deepening understanding of the interconnections between material and energy transitions, in which carbon plays a central role. Moreover, the carbon flow map offers policymakers, industry, and researchers a clear overview of carbon use across key sectors, supporting more informed decisions on strategies for circularity and climate policy. Beyond direct application, the method can also be used in future research to refine or test indicators that capture circular economy strategies and their environmental impacts. The results of the Dutch carbon flow map for 2020 showed that over 80% of the total carbon supply was fossil-based. In terms of origin, 83% of the supply was imported, while the remainder came from domestic sources. Of the total supply, only about 3% was supplied from domestically recovered carbon, and 4% from domestically extracted biogenic carbon. This highlights both the fossil dependence and the largely linear structure of the current system. It also points to the structural interdependence between imports and exports. As long as the Netherlands remains an industrial hub exporting carbon-based products made from imported feedstocks, it will continue to rely heavily on external carbon sources, whether fossil, biogenic, or recycled. Circularity and climate neutrality goals must therefore be closely aligned with industrial strategy, balancing competitiveness with the need to secure sustainable carbon feedstocks. Meeting this challenge will require coordinated policy, innovation, and investment across the entire value chain. Although national statistics such as the MFM provide a comprehensive overview, they are often highly aggregated and difficult to validate, which may limit accuracy for certain carbon flow categories. Conversion from material to carbon further introduces uncertainty. Moreover, despite the central role of waste management in advancing circularity, current reporting provides insufficient detail on the collection, processing, and origins and destinations of waste streams. We therefore recommend improving the granularity and transparency of statistical reporting, particularly for waste-related flows. At present, only the Netherlands and Denmark maintain macro-level material flow monitors, and their scopes and categories differ (Statistics Denmark 2020). For future research, expanding such monitoring efforts to other EU member states and harmonising methodologies across countries would be crucial to enable consistent and comparable assessments of carbon circularity and its links to climate neutrality. 24 Acknowledgements The authors gratefully acknowledge Roel Delahaye and Redbad Mosterd for their guidance and support with the Material Flow Monitor data interpretation, and Monica van Leeuwen for insightful discussions and valuable feedback, which contributed to the development of this study. This study was conducted as part of the Institute for Sustainable Process Technology (ISPT) “Circular Carbon Innovation Programme,” under the project PROVE IT (Products of Value from various CO₂ sources, Enabled by Integrating Technologies). This research was funded by subsidies from the Netherlands Ministry of Economic Affairs (EZK) and the Ministry of Agriculture, Fisheries, Food Security and Nature (LNV), under the Mission-Driven Research, Development, and Innovation (MOOI) program. 25 References Akerboom, Sanne, Bert Bosman, Martijn Broekhof, Wouter Jongepier, Sascha Kersten, Sikke Klein, Gert Jan Kramer, Coby van der Linde, Marc Londo, and Anouk van Loon. 2023. Groene Keuzes Voor de Nederlandse Basisindustrie: Klimaatneutrale Productie in Een Circulaire Economie. The Sustainable Industry Lab. Amadei, Andrea Martino, Lucia Rigamonti, and Serenella Sala. 2023. “Exploring the EU Plastic Value Chain: A Material Flow Analysis.” Resources, Conservation and Recycling 197(June):107105. doi:10.1016/j.resconrec.2023.107105. Andrieu, Baptiste, Hugo Le Boulzec, Louis Delannoy, François Verzier, Guillaume Winter, Olivier Vidal, and Konstantin Stadler. 2024. “An Open-Access Web Application to Visualise Countries’ and Regions’ Carbon Footprints Using Sankey Diagrams.” Communications Earth and Environment 5(1):1–9. doi:10.1038/s43247-024-01378-8. Baars, Joris, Mohammad Ali Rajaeifar, and Oliver Heidrich. 2022. “Quo Vadis MFA? Integrated Material Flow Analysis to Support Material Efficiency.” Journal of Industrial Ecology 26(4):1487–1503. doi:10.1111/jiec.13288. Bachmann, Marvin, Christian Zibunas, Jan Hartmann, Victor Tulus, Sangwon Suh, Gonzalo Guillén-Gosálbez, and André Bardow. 2023. “Towards Circular Plastics within Planetary Boundaries.” Nature Sustainability 6(5):599–610. doi:10.1038/s41893-022-01054-9. vom Berg, Christopher, Michael Carus, Mathias Stratmann, and Lara Dammer. 2023. “Renewable Carbon as a Guiding Principle for Sustainable Carbon.” (September 2020). Chen, Chuke, Jianchuan Qi, Nan Li, Tiantian Ji, Heming Wang, Yuanyi Huang, Jing Guo, Xiaohui Lu, Ruru Han, Jianlimin Wei, and Wei Qiang Chen. 2022. “China Economy-Wide Material Flow Account Database from 1990 to 2020.” Scientific Data 9(1):1–15. doi:10.1038/s41597-022-01611-z. Delahaye, Roel, Maria-josé Linders, Redbad Mosterd, and Chantal Blom. 2023. Inventarisatie Monitoring Biogrondstoffen. Delahaye, Roel, and Redbad Mosterd. 2024. De Koolstofboekhouding van Nederland. Available at: https://seea.un.org/content/carbon-accounts-netherlands Delahaye, Roel, Vivian S. C. Tunn, and Arnold Tukker. 2023. “Developing a Material Flow Monitor for the Netherlands from National Statistical Data.” Journal of Industrial Ecology 27(2):408–22. doi:10.1111/jiec.13365. Eriksen, Marie Kampmann, Kostyantyn Pivnenko, Giorgia Faraca, Alessio Boldrin, and Thomas Fruergaard Astrup. 2020. “Dynamic Material Flow Analysis of PET, PE, and PP Flows in Europe: Evaluation of the Potential for Circular Economy.” Environmental Science and Technology 54(24):16166–75. doi:10.1021/acs.est.0c03435. EZK. 2019. National Climate Agreement-The Netherlands. FAO. 2015. Knowledge Reference for National Forest Assessments. Gonella, Sara, Mark A. J. Huijbregts, Heleen de Coninck, Vincent de Gooyert, and Steef V. Hanssen. 2025. “Greenhouse Gas Reduction Potential of Novel CO2-Derived Polylactic-