2 The UNITAR Plastic Model
Summary
The UNITAR model provides a harmonised structure for estimating plastic put-on-market and waste generation across countries using trade and production data. However, the quality of the results is limited by partial product code coverage, and uncertainty stemming from broad plastic-content assumptions, and quality data issues. Despite these gaps, the model delivers actionable insights—revealing both insights on plastic volume trends and key anomalies in specific sectors such as transportation which need to be addressed for accurate results.
2.1 A harmonised approach based on trade and production statistics
The plastic model in development by UNITAR aims to support the creation of harmonised national inventories by building on existing trade and production data. Such trade and production databases already capture a large share of plastic flows—both from domestic production and imports. For example, in Denmark, two-thirds of plastic in products is imported and one-third is produced domestically (Gravgård et al., 2023).
Leveraging this data allows countries to estimate apparent plastic consumption—the total amount of plastic entering an economy, calculated as domestic production plus imports minus exports. The model is particularly useful for estimating earlier life cycle stages,
National statistics and legislation have historically focused on what is collected and treated, leaving gaps in the understanding of market inflows, in-use stocks, and uncollected plastic—whether due to environmental leakage or collection that fall outside official plastic collection statistics.
However, trade and production databases are not designed to track plastic specifically and often miss plastics embedded in composite goods (Barrowclough et al., 2020).
These include plastics used in vehicles, electronics, furniture, and textiles – categories where plastic is present but not always reported as a distinct material (UNITAR & UNEP, 2025).
2.2 Key methodological challenges of the UNITAR model
To better understand how the UNITAR model performs in practice, we must first examine the key methodological challenges that affect its accuracy and comparability. These challenges include data limitations, generalised assumptions, and structural gaps in how trade and production statistics represent plastic flows.
Plastic fractions are based on assumptions for broad product categories
Trade and production databases were not built to track plastic specifically. Even though there is a category for plastics in the trade system (Chapter 39 in the HS codes) (World Customs Organization [WCO], 2002), trade or product codes often group plastic and non-plastic items together. To work around this, estimates—e.g., 10% plastic in vehicles—are drawn from literature or one-off studies and carry high uncertainty, especially for complex categories like textiles and construction materials. Both previous national and Nordic analyses have noted that default values are often outdated or poorly documented, limiting the reliability of such calculations (Berge et al., 2023). Moreover, plastic content can change significantly over time due to technological shifts or policy influence—for instance, the transition to electric vehicles affects material composition. This underlines the need for regular revision of plastic coefficients to reflect changing product designs and usage patterns.
Packaging is often excluded
Trade statistics typically record the net weight of goods excluding packaging (Berge et al., 2023). This means plastic packaging used to ship or contain products is not accounted for in import/export weight data.
It should be noted however, empty packaging, which is the product in itself is accounted for through their HS/CN codes.
Inconsistent units make comparisons difficult
While many trade records use kilograms, others rely on item counts, litres, or other category-specific units (Berge et al., 2023). This increases the risk of misclassification or missing weights, which in turn adds uncertainty.
A UN trade analysis found that issues with trade data, such as missing physical weight values, are relatively common and require targeted imputation. Each of these conversions adds to uncertainty in the final material flow figures (Zhang et al., 2022).
Polymer-specific data is limited
If the aim is to break down flows by polymer type (PE, PP, PET, etc.), trade/production data seldom provide that detail except for basic resins. Finished goods and articles are typically not labelled by polymer in official stats. Analysts must either restrict analysis to broad groups (all plastics by sector) or apply assumed polymer splits within product categories (e.g. using market research data to say packaging is X% PE, Y% PP, etc.). This can introduce error if the assumed polymer mix differs from reality. The Danish preliminary MFA reported that “only few data were available on polymer composition of selected plastic flows” (Pivnenko et al., 2019).
Systematic national validation is needed for robust results
Adopting the UNITAR trade-and-production-based model provides a structured path to harmonised plastic flow accounts using existing Nordic data. Yet its reliability depends on refining key assumptions, improving data inputs, and transparently managing uncertainty. Nordic countries can assist in this process by aligning plastic content coefficients, addressing packaging gaps, and contributing to a shared regional approach.
With such efforts, the model can provide a robust foundation for cross-country comparisons—while also highlighting where complementary methods, like micro-level waste composition studies, are needed.
The following section outlines the methodology used to test the pilot version of the UNITAR model in a Nordic context.
2.3 Pilot application methodology
To test the UNITAR framework across the Nordic countries, we added the relevant trade and production data to the pilot version of the UNITAR plastic model (as of May 2025). For a detailed description of the method for adding the trade and production data see Appendix A.
Only a subset of codes is currently included
Of the approximately 620 HS codes and 240 unique CPC codes–defined in the UNITAR report (UNITAR & UNEP, 2025) to be most relevant for estimating plastic flows
The full list of CPC and HS codes are available in Statistical guideline for measuring flows of plastic throughout the life cycle (UNITAR & UNEP, 2025).
Based on a comparison with the appendix of Statistical Guideline for Measuring Flows of Plastic Throughout the Life Cycle (UNITAR & UNEP, 2025), the model’s trade data currently includes approximately: 40% of packaging codes (P1), 94% of transportation codes (P2), 100% of building and construction codes (P3), 0% of electrical and electronic equipment codes (P4), 58% of consumer and institutional product codes (P5), 7% of apparel and textile furnishing codes (P7), and 50% of “Other” codes (P8). Industrial machinery (P6) is not covered in either the pilot model or the report, which notes that “Further detailing [is] need[ed] to be developed” (p. 141).
Based on a comparison with the appendix of Statistical Guideline for Measuring Flows of Plastic Throughout the Life Cycle (UNITAR & UNEP, 2025), the model’s production data currently includes approximately: 55% of packaging codes (P1), 100% of transportation codes (P2), 88% of building and construction codes (P3), 0% of electrical and electronic equipment codes (P4), 81% of consumer and institutional product codes (P5), 15% of apparel and textile furnishing codes (P7), and 50% of “Other” codes (P8). Industrial machinery (P6) is not covered in either the pilot model or the report. The model also has 6 codes that are not found in the draft report appendix.
As a result, this analysis does not include a large part of the relevant plastic flows. The total volumes shown in this report are therefore likely lower than the actual amounts—but it is not currently possible to say by how much. This limitation also prevents direct comparison with national studies such as the Swedish material flow analysis.
These limitations are important to keep in mind when interpreting the results and visualisations in this report. The findings reflect a technical pilot test of the UNITAR model, which is still under development—not a complete or fully representative estimate of national plastic flows.
Icelandic data excluded
Due to the nature of the Icelandic trade data structure for their API and time constraints of the informants Icelandic data has not been included in the analysis.
No extensive data validation
The dataset used in this pilot implementation has not undergone full-scale data validation or cleaning. Unlike the rigorous estimation procedures applied in WEEE modelling by Van Straalen et al. (2016), this dataset has only been subject to limited manual review. Apparent outliers were assessed manually by the authors on a case-by-case basis: single-year extreme anomalies were either adjusted to better reflect surrounding trends when deemed implausible or reported separately to maintain visibility in the graphs.
A comparison with national data (e.g. from SSB in Norway) indicates that some of the extreme spikes found in Eurostat data likely stem from reporting inconsistencies or unit conversion errors rather than actual shifts in plastic flows. A detailed description of this can be found in Appendix D.
Where multi-year anomalies occur, these values have been visualised separately (see Figures 6 and 8) rather than being smoothed or removed. This to preserve the approach of reflecting the current challenges of the draft model and the aim to preserve transparency in raw data behaviour over time. Nevertheless, deviations in trade units or weight fields may affect result quality. More detailed information on data quality issues is provided in Appendix D.
Interpreting the results
This modelling process provides a harmonised—though partial—view of plastic flows in the Nordic economies. The results illustrate how the draft UNITAR plastic model can be applied using available trade and production data.
The findings should be interpreted as indicative outputs from an exploratory application of the model—not as validated or complete estimates of national plastic flows. The methodological limitations outlined in the section Key methodological challenges of the UNITAR model should also be kept in mind when reviewing the results.
2.4 Plastic Put-on-Market Results
Summary
The draft UNITAR model offers a harmonised method for estimating plastic put-on-market (POM) volumes across the Nordic countries. The pilot reveals a consistent trend of steady growth in POM volumes, with packaging as the dominant category. However, the results also highlight data anomalies—particularly related to specific product codes and unit mismatches.
It is important to note that the model currently covers only a limited subset of plastic flows. As such, the results should be viewed as indicative outputs from an exploratory model application, not as conclusive estimates.
Sweden and Finland show stable increases in POM volumes, while Denmark follows a similar trend once outliers are excluded. Norway reports lower overall volumes but is notably affected by extreme spikes linked to offshore infrastructure.
This section presents the results of the pilot application of the UNITAR draft model for estimating plastic put-on-market (POM) amounts in the Nordic countries. It highlights both the strengths of the harmonised approach and the limitations that arise due to data availability, anomalies in trade/production statistics, and assumptions regarding plastic content.
Sweden: Stable profile dominated by packaging, with early volatility and outliers
In Figure 1 Sweden shows a relatively stable distribution of plastic put-on-market (POM) between 1995 and 2023. The modelled trend reveals notable volatility before 2011, with pronounced peaks in 1997 and 2010, and a sudden dip in 2003. Packaging is most consistently the largest category (approximately 20-40 percent of yearly total), especially from 2008 onward, where levels stabilise around 150,000–180,000 tonnes. Building and construction initially account for a large share of the annual total but later converge with transportation and consumer & institutional products. The “Other” category shows erratic behaviour, including a major spike in 2010. This increase comes from a few codes regarding glasses and other optical or protective equipment.
CPC codes:
- 48312 “Spectacles, goggles and the like, corrective, protective or other”
- 48313 “Frames and mountings for spectacles, goggles or the like”, and
- 48314 “Binoculars, monoculars and other optical telescopes; other astronomical instruments, except instruments for radio-astronomy; compound optical microscopes”
Figure 1 Plastic Put-on-Market Sweden
Finland: Gradual increase until 2018, followed by sharp rise driven by "Other"
Finland presents a steady increase in plastic put-on-market (POM) volumes from ca 150,000 tonnes in 1995 to just above 350,000 tonnes in 2018. There’s a noticeable sharp dip in 2007. From 2019 onwards, volumes surge to over 610,000 tonnes by 2022. The primary driver of this recent jump is a sharp increase in the ”other” category, contributing with an additional 150,000 tonnes per year. The spikes in Finland’s “Other” category are linked to the same production codes as in Sweden – mainly for glasses and other optical or protective equipment. This highlights the need to review the category’s validity, as well as possible unit mismatches or other reporting issues. There’s a 7 476 536-tonne spike for the transportation subcategory “Other vessels”
CPC codes:
- 49319 – ”Other vessels (including light-vessels, fire-floats, dredgers, floating cranes, floating docks, warships and lifeboats other than rowing boats), except floating or submersible drilling or production platforms”
- 49320 – “Floating or submersible drilling or production platforms”
Figure 2 Plastic Put-on-Market Finland
Denmark: Similar trend to Finland, with extreme outliers removed
For Denmark the increasing trend is similar to that of Finland–starting at above 140,000 tonnes in 1995 and peaking in 2021 at around 490,000 tonnes. Notably, the “Other” category is omitted from the graph due to extreme values in the data. From 2008 onward, annual volumes exceeded 1.3 million tonnes–five to ten times higher than totals in other years–before dropping back after 2012 (Figure 6).
For Denmark there’s also a 1 252 017-tonnes jump 1998 in the packaging category 36490 “Other articles for the conveyance or packing of goods, of plastics; stoppers, lids, caps and other closures, of plastics” that we have assumed to be a reporting error due to the irregular occurrence and extreme size.
Figure 3 Plastic Put-on-Market Denmark.
Other category is removed due to extreme values.
Figure 4 “Other” Plastic Put-on-Market Denmark
The “Other” category showing extreme values from a few codes regarding glasses and other optical or protective equipment between 2008-2012
Norway: Extreme transportation spikes and lower baseline levels
Norway stands out for both its lower baseline volumes and sharp anomalies. Unlike other countries, Norway’s issues relate primarily to the Transportation category. As seen in Figure 8 the transportation category shows extreme values for 2011-2016 (spikes of up to 200 million tonnes of plastic). This issue mainly comes from one subcategory “Other vessels”.
CPC codes:
- 49319 – ”Other vessels (including light-vessels, fire-floats, dredgers, floating cranes, floating docks, warships and lifeboats other than rowing boats), except floating or submersible drilling or production platforms”
- 49320 – “Floating or submersible drilling or production platforms”
Other notable findings are that the same dip in 2007 found in all other countries, after which follows a 10,055,649-tonne spike for packaging 2008
36410 – “Sacks and bags, of plastics”
CPC codes:
- 27912 – “Tulles and other net fabrics, except woven, knitted or crocheted fabrics; lace in the piece, in strips or in motifs”
- 27913 – “Embroidery in the piece, in strips or in motifs”
Apart from the spikes mentioned, Norway stands out with notably lower total put-on-market (POM) volumes – rising from just under 123,000 tonnes in 1995 to around 330,000 tonnes in 2021. These absolute values are lower than those reported for other Nordic countries, and far below the national estimates presented in Berge et al. (2023).
Berge et al. (2023) estimates the net plastic volumes (production + imports – exports) 570 000 tonnes for semi-finished products and 410 000 for primary plastic and finally around 330 000 tonnes of plastic from plastic containing products.
Figure 5 Plastic Put-on-Market Norway
Figure 6 Plastic Put-on-Market Norway – category “Transportation” showing extreme values between 2010-2017
Conclusion
The draft UNITAR model reveals broadly consistent trends in plastic put-on-market volumes across the Nordic countries, with packaging as the dominant category and gradual growth from the mid-1990s onward. However, the results also expose key limitations regarding the data quality uncertainty, reliance on assumed plastic shares for broad categories, and significant outliers tied to specific product codes (e.g. optical goods, offshore platforms). These issues underscore the challenges of a harmonised estimation model based on global trade and production codes (less detailed than their EU equivalents). Careful data validation, category-level adjustments, and expansion of code inclusion will be essential to fully realise the model’s comparative value.
2.5 Plastic waste generated
Summary
The waste generation modelling highlights the UNITAR framework’s potential to track not only input volumes, but also the long-term accumulation and point of treatment of plastic over time. By applying sector-specific lifetime assumptions, the model provides a dynamic view of how different plastic applications contribute to waste over time—enabling more targeted understanding of when and where waste arises. Despite limitations, the approach opens new avenues for forward-looking planning and shows how harmonised modelling can improve both policy timing and material management strategies.
The UNITAR model estimates plastic waste generation by applying product-specific lifetime assumptions to the previously modelled put-on-market (POM) volumes. This enables a time-distributed simulation of how plastic placed on the market becomes waste over subsequent years. It is important to note that these figures represent modelled, indicative outputs based on the POM results in the previous section—not actual collected waste volumes. They are therefore subject to the same limitations and challenges as the POM results, which should be considered when interpreting the findings.
Note on interpretation: The waste generation estimates presented in this section are based on previously modelled plastic put-on-market (POM) volumes, with the models product-specific lifetime assumptions applied. To improve realism and interpretability, certain one-year spikes identified as data artefacts in the POM stage—particularly those linked to “Other vessels” and optical equipment—have been excluded from the input data prior to waste modelling.
After 2023, total waste volumes begin to decline for all countries. This is not a forecasted real-world trend, but a result of the model’s scope: 2023 is the final year for which POM input data is available. Since no new inflows are added beyond that point, plastic waste generation gradually decreases as products from earlier years reach their end-of-life. A similar limitation affects the start of the graph: because the POM data series begins in 1995, the model does not capture end-of-life waste from long-lived plastic products placed on the market before that year, such as those used in transportation and construction.
This highlights both the utility and the limitations of the model: while it provides valuable insight into waste generation dynamics, it relies on estimations for years preceding available data and requires ongoing updates with new inputs to remain relevant beyond the current cut-off year.
Sweden
Sweden’s modelled plastic waste shows a rising trend from the mid-1990s, peaking around 2022. This pattern mirrors the historical growth in plastic placed on the market, with a time lag reflecting product lifespans. Packaging remains the dominant waste category throughout the period, followed by Consumer & Institutional Products and Transportation.
Note that as the put-on-market (POM) data starts in 1995, longer lifecycle plastics are not present at the beginning of the time series. After 2023, the modelled waste volumes decline due to the absence of further POM inputs.
Figure 7 Plastic waste generated Sweden
Finland
Plastic waste generation in Finland shows a consistent upward trend from 1995 to the late 2010s, reflecting the gradual increase in plastic put on the market. Waste volumes peak between 2020 and 2023, following the surge in put-on-market (POM) observed from 2019 onward.
The waste composition is dominated by Packaging and Consumer & Institutional Products, with a growing share of Transportation after 2010. The “Other” category also contributes more to recent years, corresponding to the increase in POM volumes attributed to optical equipment codes.
As in Sweden, long-lifecycle plastics are missing at the start of the time series, and modelled waste volumes decline after 2023 due to no further POM inputs.
Figure 8 Plastic waste generated Finland
Denmark
Denmark’s modelled plastic waste closely tracks its put-on-market (POM) profile, showing a steady rise from 1995 and peaking around 2021–2023. Packaging and Consumer & Institutional Products make up the largest shares, while other categories like Transportation gain from 2010 onwards.
Crucially, waste estimates are smoother than the underlying POM data because the “Other” category – which showed outlier spikes in Denmark’s POM results – has been excluded for affected years. This correction prevents exaggerated waste volumes from distorting the total.
As with Sweden and Finland, long-lifecycle plastics are missing at the start of the time series, and the modelled waste volumes decline after 2023 due to the absence of further POM inputs.
Figure 9 Plastic waste generated Denmark
Norway
Norway’s modelled plastic waste is the lowest among the four countries, aligning with its smaller put-on-market (POM) volumes. Waste levels increase gradually to around 230,000 tonnes by the early 2020s, driven primarily by Packaging and Consumer & Institutional Products.
However, the waste profile is also affected by Norway’s data anomalies—particularly from transportation. The “Other vessels” category, while partially removed from POM data, still leaves traces in the waste outputs for the years following 2011–2016. As a result, some fluctuations remain visible in the waste time series.
As with the other Nordic countries, long-lifecycle plastics are missing at the start of the time series, and the modelled waste volumes decline after 2023 due to the absence of further POM inputs.
Figure 10 Plastic waste generated Norway
2.6 Insights into polymer composition
Polymer-level outputs offer value by highlighting dominant material types and potential risk areas but should be treated as indicative trends. Beyond estimating total plastic flows, the UNITAR model also provides a first indicative breakdown of polymer types over time. This is done by assigning typical polymer compositions to each product group, based on literature and expert judgement (as described in the guideline’s annex) (UNITAR & UNEP, 2025). While approximate, this feature enables a harmonised view of material composition at both the put-on-market and waste stages.
More than a first-order indication of the plastic composition profile over time—valuable for understanding material dynamics—it also supports future analyses of recycling potential and chemical risk management, particularly for polymers associated with hazardous additives (e.g., PVC, PUR).
Figure 13 and 14 illustrates the polymer distribution in Sweden’s put-on-market, and waste generated, volumes respectively. Sweden is shown as a standalone example because of the extreme outliers in other countries and to highlight the indicative and exploratory nature of these results.
The model suggests that polyethylene (PE) and polypropylene (PP) dominate the flows, followed by smaller shares of PET, PVC, and other polymers—broadly consistent with global estimates. A similar pattern is followed by Plastics Europe’s data (Plastics Europe, 2024) which is quoted in Sweden’s first MFA, published in 2019, while the most recent MFAs do not offer plastic type breakdown.
However, these outputs rely on additional assumptions, such as fixed product-to-polymer mappings and constant shares over time and should be interpreted with caution. These limitations are explicitly acknowledged in the UNITAR guideline as a source of uncertainty (UNITAR & UNEP, 2025).
Figure 11 Polymer composition of plastic Put-on-Market (tonnes) – Sweden
Figure 12 Polymer composition of plastic waste generated (tonnes) – Sweden
2.7 Lessons Learned
Summary
The Nordic pilot confirmed the conceptual value of the UNITAR model but also revealed key barriers to its reliable application. While several challenges—such as confidentiality restrictions and unit inconsistencies—were already known from previous studies, the pilot provided practical insight into how these issues affect model performance in real-world settings. Addressing them is essential to ensure credible, comparable results across countries.
The pilot application of the UNITAR model in the Nordic context offered concrete insights into the practical challenges of harmonising plastic statistics, as previously noted by, for example, Berge et al. (2023). Although national data availability is relatively strong, the process exposed technical, methodological, and institutional barriers that must be addressed to enable robust and comparable plastic-flow modelling across countries.
Input data quality limits model reliability
Although the Nordic countries maintain detailed trade and production statistics, several recurring quality issues affect the accuracy of model results. These include:
- Confidentiality restrictions that suppress item-level production data.While confidentiality restrictions limit external access to detailed production data, national statistical agencies typically hold full access to these records. The model’s outputs are aggregated and do not necessarily compromise confidentiality (even though this must be evaluated case by case when unsuppressed data is used). This makes it both feasible and necessary for statistical offices to take an active role in the modelling process—enabling full data use while safeguarding disclosure.
- Inconsistent units or conversions, which could result in errors if data unit flags are not correct. Many of these issues are well-documented in statistical metadata and previous studies. However, the pilot confirmed how they distort plastic-flow modelling in practice
- Discrepancies between data obtained from national sources and from Eurostat.
Many of these issues are well-documented in statistical metadata and previous studies. However, the pilot confirmed how they distort plastic-flow modelling in practice. Each of these issues is explored in more depth in Appendix D, including case examples (e.g. Norway’s Eurostat vs. SSB mismatch).
Complexity in converting different units to tonnes
Transportation is a large contributor to plastic consumption and waste. It also represents a complex product category that illustrates the challenges of producing harmonised plastic consumption estimates in tonnes across all categories. One example is ship production, as reported in PRODCOM, where output is measured in either units or CGT (Compensated Gross Tonnage). CGT is not a physical weight but a work index that combines a vessel’s size and construction complexity into a single figure, allowing comparisons of labour effort between vessel types rather than their material content (see Compensated gross tonnage, Wikipedia). As a result, converting these CGT figures into tonnes of plastic requires additional assumptions about ship mass and typical plastic shares—data not available in PRODCOM itself.
Plastic content assumptions drive model sensitivity
Many plastic flow estimates depend not just on volume data, but on assumed plastic content per product. The pilot revealed that while some plastic-content coefficients can be considered reasonably accurate for common product groups (e.g. vehicles, clothing), others are less suitable for Nordic conditions — particularly in sectors where plastic content differ. The suitability of default coefficients varies by product group and national context. In general, assumptions for categories such as vehicles and clothing are likely to align reasonably well with Nordic conditions given their globalised markets.
For more detailed information see chapter 3 Status of monitoring plastic share in waste streams in the Nordics
In Norway, for example, default coefficients created large spikes in estimated plastic volumes for certain industrial products (e.g. drilling platforms). These artefacts reflect how a small number of outlier codes can distort results unless coefficients are tailored to national conditions.
A shared Nordic coefficient repository could help streamline future modelling efforts by coordinating updates across countries—especially for categories known to vary regionally, such as packaging, construction, WEEE and composites.
Harmonisation benefits depend on Nordic organisations working in a consistent way
The modelling work highlighted that statistical harmonisation is not only technical, but institutional. Differences exist between countries in:
- How they define the stages of plastic flowsFor example, Norway defines the plastic waste generated amount to be equal to the plastic waste collected volume – which implies a 100% recovery rate (Workshop participant, March 2024). In contrast, Sweden does not apply a fixed formula; according to the Swedish Environmental Protection Agency, national statistics are based on reported plastic waste from companies and municipalities, categorised by SNI codes (Swedish EPA representative, personal communication, 2024).
- What can be shared publicly by organisations due to legal constraints
- How much staff time and resources agencies can dedicate to maintaining and updating data
These differences limit the comparability and repeatability of national models unless coordinated structures are in place. The pilot underlines the value of a Nordic working group for plastics statistics to maintain assumptions, align classifications, and ensure a consistent interface with UNITAR reporting frameworks.
Conclusion
The UNITAR model is conceptually sound and broadly implementable. However, the Nordic pilot shows that reliable results depend on four enablers:
- Improved input data validation
- Updated and aligned plastic-content assumptions
- Institutional capacity and mandate for harmonisation
- Ongoing Nordic collaboration and shared model maintenance
Together, these are prerequisites for delivering robust, policy-relevant plastic-flow statistics—nationally and internationally.