Decomposition of Changes in Primary Abiotic Resources in the Netherlands, 1996-2022

This report provides a decomposition of changes in the use of primary abiotic resources in the Netherlands in the period 1996-2022. The reduction of the use of primary abiotic resources is a key CE policy target in the Netherlands: 50 percent reduction by 2030. We present an index decomposition analysis (IDA) of changes in primary abiotic material use. In this framework, there are four factors underlying changes in the resource consumption: substitution of abiotic for biotic materials, recycling of materials, improving resource efficiency and economic growth. We expect that the first three factors exert a dampening effect on primary abiotic materials consumption, whereas economic growth has an upward impact on this consumption.

In quantification of the IDA, we measured material use at first with domestic material consumption. Substitution takes place if the ratio of abiotic primary materials to the total of primary materials decreases. Recycling increases with a decrease in the ratio of total primary materials to total of both primary and secundary materials. Resource efficiency increases if the ratio of consumption of materials to GDP decreases. Finally, GDP growth reflects change in economic activity.

In the Figure we show our main result for the period 1996-2022. The orange line ('Abiotic resource consumption') shows the development of domestic primary abiotic resource consumption (the change in each year is the change relative to 1996). We can see an overall decline of 15 percent in abiotic resource consumption.

The other four lines reflect the contribution of the four factors to this change in resource consumption. If one would add up the contributions of all four factors, one would get the resulting change in resource consumption. Hence, if, for instance, only economic growth would determine the resource consumption and the other factors would not have exerted their dampening contributions, primary abiotic resource consumption would have increased with the extent of the contribution of economic growth.

The Figure shows that in the 26 years between 1996 and 2022, substitution, recycling and resource efficiency apparently go hand in hand with dematerialisation over time. These factors were mitigating the large upward pressure of economic activity. Increased substitution had a somewhat larger contribution to the decline of resource consumption than increased recycling. The recycling rate is already around 80 percent in the Netherlands. The contribution of substitution might be somewhat overrated as the underlying data include food and feed. But currently the biobased economy is still small, and thereby its impact possibly still to come. The driver that counteracts the effect of economic growth most is resource efficiency. However, this driver probably comprises more than real efficiency alone, including among other things the impact of economic sectoral shifts.

In order to show structural changes in the contribution of the factors, we also split the data into two sub-periods, one before 2009 and one after 2009. Then it becomes more clear that only in more recent years, substitution, recycling and improved efficiency apparently led to a substantial decrease of primary abiotic resource consumption. Then apparently more efforts were made to reduce materials use. In the period before 2009 the upward pressure of economic growth on material consumption was larger than in the period thereafter.

We emphasize that our results have to be interpreted carefully. The decomposition analysis does not present a causal link but can uncover correlations. There are issues with modeling, data, measurement and definitions. But the results shows the potential of a framework of factors underlying changes in primary abiotic resource consumption, and long time series data for an analysis providing insight in structural changes in substitution, recycling, efficiency, and economy over time.

Circular economy (CE) is an important policy theme in the pursuit of reduction of non-renewable natural resources use and the amount of waste leaving the economy. Policy makers utilize statistical data on physical stocks and flows of different types of materials, and indicators such as Domestic Material Consumption (DMC), Circular Material Use Rate (CMUR), and Resource Productivity. There is a growing need for more detailed information, analyses, and visualisations in order to unlock and disseminate the wealth of statistical information.

This report provides a decomposition of changes in the use of primary abiotic resources in the Netherlands over a long time period. The reduction of the use of primary abiotic resources is a key CE policy target in the Netherlands: 50 percent reduction by 2030. The analysis is an update of decomposition results reported by Delahaye et al. (2020, Section 3). ¹⁾ The update was made for the benefit of the ICER (Integral Circular Economy Report) publication in February 2025 by PBL Netherlands Environmental Assessment Agency. ²⁾

In the next sections, we present the index decomposition analysis (IDA) of primary abiotic resources use in the Netherlands. First, we describe the IDA model with drivers behind the use of primary abiotic resources, measured as DMC (Section 2). Data to quantify the drivers come from official statistical sources of Statistics Netherlands, allowing statistical agencies in other EU countries to apply the same methodology (Section 3). In Section 4, we present the results for the long time period 1996-2022, and also describe developments in two subperiods before and after 2009. Annex A provides results of a variant of measuring resource use, namely Domestic Material Input (DMI). The report concludes in Section 5 on the results of the IDA with discussion and recommendations.

Here we present a model for index decomposition analysis (IDA), which describes drivers behind changes in primary abiotic resources use.

Changes over time in resource consumption are supposed to be driven by: a) substitution of renewables by non-renewables, b) increased use of secondary materials (recycling), c) improved resource efficiency and d) economic growth. We measure resource use with Domestic Material Consumption. This is an international standard indicator, and equals domestic extraction plus imports minus exports.

More formally, in our IDA, the variable (primary abiotic resource consumption) is written as the product of the drivers underlying changes in this resources use:

\(Prim_{abiotic} \:=\:\frac{Prim\:abiotic}{Prim}\:\times\:\frac{Prim}{Total}\:\times\:\frac{Total}{GDP}\:\times\:GDP\)

of which:

Prim_abiotic: Primary abiotic resource consumption, including abiotic but excluding biotic and secondary materials. A Dutch policy target is to reduce the use of primary abiotic resources with 50 percent in 2030.

• Prim: Primary resource consumption, including both biotic and abiotic materials.
• Total: Total material consumption including secondary, biotic and abiotic materials
• GDP: Gross Domestic Product of the Dutch economy.

Hence the drivers are:

• Prim_abiotic / Prim: Substitution of abiotic resources by biotic materials (reflected by lowering the share of primary abiotic materials in total primary material use)
• Prim / Total: Recycling of secondary materials (represented by lowering the share of primary materials in total material use)
• Total / GDP : Resource efficiency (decreasing total material use per unit of value added)
• GDP: Economic activity level.

This equation is converted into an additive (dynamic) equation for the change in resource consumption decomposed into the contributions of changes in the respective drivers:

\(\Delta\: Prim_{abiotic} \:=\Delta\:\frac{Prim\:abiotic}{Prim}\:+\:\Delta\frac{Prim}{Total}\:+\:\Delta\:\frac{Total}{GDP}\:+\:\Delta\:GDP\)

where ∆ indicates a change between two points in time. This may be a change between two subsequent years t-1 and t, but can also be a total change over a longer time period. In our application, we calculated annual changes of the resource use and drivers. In order to calculate the total change over a longer time period, we add up all annual changes in the period under consideration. The resulting figure is a change on balance, or net change.

We expect that more substitution, more recycling and higher resource efficiency exert a downward pressure on primary abiotic resource consumption, whereas more economic activity has an upward impact on the resource consumption. Note that if we say that substitution, recycling and efficiency increase, the ratios which implicitly reflect them in the equation are decreasing. Note also that we quantify the contribution of the drivers, which does not equal the volume changes in the drivers themselves. For instance, if substitution increases, the ratio Prim_abiotic / Prim decreases by some amount, a change in the ratio itself. This has some downward impact, that is, the quantified contribution, on primary abiotic resource consumption. If all other drivers would not exist, the resource consumption would decrease by the amount of the contribution of this driver.

One of the factors in the IDA, resource efficiency (Total / GDP), has a complex relationship with domestic primary abiotic resource use. This factor may comprise not only real improvements in efficiency but also other changes not specified explicitly. The indicator comprises a resource variable (numerator) and an economic variable (denominator). There might be changes in one or both variables independent of changes in the resource efficiency (i.e. the relationship between these two variables). For instance, a shift to a service economy or shift of goods production to abroad does not need to imply a improvement in resource efficiency in domestic production of goods. One should keep in mind that the label ‘resource efficiency’ might not cover the load.

Before turning to data and results of the IDA in the next sections, we emphasize that an IDA is not a causal model, but quantifies presumed interactions. Usually an IDA includes factors that are supposed to influence the dependent variable. Some factors have had a strong proven relationship with primary abiotic resource consumption, such as economic activity, while others exert a more complex influence, such as resource efficiency. Our aim is to look into correlations between the change in resource consumption and the change of the various factors. We do not state that these factors ‘cause’ or ‘drive’ the resource consumption exactly with their quantified contributions. Furthermore, there may be hidden mutual relationships between the factors themselves, creating indirect effects and loops. This may artificially resize the respective contributions of drivers in the decomposition.

Instead, we present the decomposition as an integrated macro-economic framework in which various factors play interacting roles in the change in resource use. It is not possible to trace back parts of changes in the resource use to exactly one factor. We pose that the IDA shows whether there are correlations, and that the quantified contributions cautiously indicate the extent of impact of the various factors. Our first aim is to show the relative contribution of the four factors. A second main contribution of our IDA is the longer time perspective, providing insight in structural changes over time. The results give an impression of developments in substitution, recycling, resource efficiency and economic growth, and their potential influence on materials use.

Data comes from regular official statistical sources of CBS Statistics Netherlands on physical material flows (Material Flow Accounts, Waste Statistics) and the National Accounts. This allows statistical offices in other EU member countries (with data available at Eurostat) to use our methodology. Data is available annually for the period 1996-2022.

Material flows

Source: http://opendata.cbs.nl/statline/#/CBS/nl/dataset/83180NED/table?dl=25735. Updated data 2020-2022 (applied in this analysis) remain internal for now, and will be published in 2025.

Data on material flows comes from the economy-wide Material Flow Accounts (MFA). Material consumption is measured as Domestic Material Consumption (DMC), that is, extraction plus import minus export. DMC is a proxy for final (material) consumption (excluding export) of a country. ³⁾ Total DMC equals total consumption of biotic and abiotic materials (Prim). Biomass can be excluded from the DMC leaving only the abiotic materials (Prim_abiotic).

Figure 3.1 shows the development of the primary abiotic resource consumption (in megaton) between 1996 and 2022. One can observe a downward trend in resource consumption over time, notwithstanding short term fluctuations.

Note that DMC estimates apparent consumption in which production of waste and use of energy carriers are also part of consumption. Further, one should know that most biomass is used for food and not for substitution of abiotic materials. However, a distinction between the two types of flows cannot be made in the MFA.

Secondly, note that the MFA data are available only at macroeconomic level. Data on material use at sectoral level can be found in the Material Flow Monitor (MFM), with detailed physical supply and use tables of material flows, with data compatible to the National Accounts. ⁴⁾ However, this Monitor is only available from 2014 onwards (and bi-annual), ⁵⁾ whereas we aim to investigate longer term changes. More importantly, the material use by sector in the MFM is intermediate use by companies, or input of materials (products of other sectors) in the production process to produce new products (often material input for other sectors). This intermediate use cannot be added up over all sectors to a total at the macro-economic level. This would lead to double counting. There is no material use equivalent as for, say, value added in the National Accounts.

Finally, as mentioned above, we measure resource use with domestic material consumption (DMC). Annex A provides an alternative with domestic material input (DMI). DMI does not account for export, and is measured as extraction plus import. DMC is more easily compared between countries as exports are accounted for and there is no double counting of material flows. But the idea behind the application of the DMI indicator is that there is a more direct link with production with the use (input) of materials in this production. Unfortunately, the data in the MFA do not distinguish re-export flows in the import and export data.⁶⁾ Re-exports do not become part of the domestic production process but are goods that enter and leave the Netherlands without hardly any processing. With DMC, the re-exports are cancelled out by definition. Though we do not know the exact size of these re-exports for all years under consideration, from our Material Flow Monitor we know that about one third of total imports in 2022 are re-exports.

Recycling

Source: https://opendata.cbs.nl/statline/#/CBS/nl/dataset/83555NED/table?dl=2573B

Data on recycling is available from waste statistics. In the case that a country compiles waste accounts it is recommended to use these figures instead of the waste statistics reported to Eurostat, because waste accounts represent waste recycled within a country. Data on recycling in the Eurostat database comprises treatment of domestically generated waste. This means domestic generated waste that is treated abroad is also taken into account and, on the other hand, imported waste recycled domestically is excluded.

Biomass waste that is used for purposes of feed for animals (and not substitution of abiotic materials) should not be included. Recycled animal and vegetable waste can be omitted by assuming that most of it is used for animal feed. Due to a revision of the Dutch waste accounts we could not leave this waste category out and, therefore, it is included in our analysis. In the model, Total is Prim plus the amount of recycled (secondary) materials. Note that the original recycling source data are bi-annual data. To estimate the recycling amounts for the years without data points, we interpolated them by drawing straight lines between the adjacent years with data points. Figure 3.2 shows the amount of recycled materials (in megaton). We see a very slow upward trend. The largest increase occurred before 2000, thereafter the amounts fluctuate.

Economic activity

Source: https://opendata.cbs.nl/statline/#/CBS/nl/dataset/85865NED/table?dl=A9518

Data on economic activity come from the National Accounts. We apply GDP in market prizes estimated as chain linked volumes. Chain-linked level series are obtained by successively applying previous year's price's growth rates to the current price figure of a specific reference year, here 2021. Measured in constant prices of 2021, Dutch GDP increased from 553 billion euro in 1996 to 936 billion euro in 2022, an increase of 70 percent in 26 years.

Changes in the factors

Table 3.1 presents the values of the four factors at time points with 6 or 7 years interval in the period 1996-2022. This shows some of the dynamics in the data, though it does not show the fluctuations between the time points.

We see the share of primary abiotic resource consumption in total primary consumption decreases somewhat from 73 percent in 1996 to 67 percent in 2022, indicating substitution of abiotic materials for biotic materials. This occurs mainly after 2006. The recycling rate of the Netherlands is already very high, so there is rather little dynamism to be seen. The share of primary materials in total of primary and secondary (recycled) materials only slowly decreases over time, from 82 percent in 1996 to 77 percent in 2022.

Further, we can observe a clear trend in resource efficiency. The amount of materials used per unit value added decreased steadily over time from 0,42 kg per euro in 1996. to 0.25 kg per euro in 2022. The ratio however, is difficult to interpret, as it combines two elements which are different in nature. The total of resources consumed (numerator) remained relatively constant throughout time, while the economic activity (GDP, the denominator), increased strongly.

Estimation method

We applied an IDA estimation method developed by Delahaye et al. (2020). This calculates an average of the contribution of each factor in a year from a system of equations comprising variants which resemble each other very closely. In turn, this mathematical system is based on a decomposition approach by Dietzenbacher and Los (1998).⁷⁾ A main disadvantage of this method is that the number of equations increases quickly with the number of factors in the IDA. With four drivers we have 4! (faculteit) = 24 (non-unique) equations per driver to calculate the average contribution of the driver under consideration. We implemented the estimates in the programming language R. We refer to Delahaye et al. (2020, Annex) for a basis R script and more detail on their method. Our update of this basis R script and additional explanation is available on request.

Figures 4.1 to 4.4 show the changes of primary abiotic resource consumption and its drivers between 1996 and 2022. The Figures tell the same tale but in different visualisation formats.

• Figure 4.1 (line graph) shows the relative changes since 1996 (in percentages of the 1996 level). Figures 4.2 to 4.4 (bar diagrams) add all annual changes to a net balance of total change (in megaton) in the respective periods. ⁸⁾
• Figure 4.1 shows all short term changes, whereas Figures 4.2 to 4.4 aim to highlight structural (longer term) changes.
• Figure 4.3 and 4.4 divide the period 1996-2022 into two subperiods of 13 years, with 2009 as the split year. This year was also a watershed year in some respects. In 2009, the first SDE subsidies were committed to applicants. ⁹⁾ SDE has had a large influence on the use of biomass for energy. Furthermore, the global financial crisis of 2007-2008 started to have large impact on the Dutch economy.
• The orange line or bar ('Abiotic resource consumption') shows the change in primary abiotic resource consumption. The other lines or bars represent the contribution of drivers behind this change: 1) substitution of biotic for abiotic materials, 2) changes in recycling of secondary materials, 3) changes in resource efficiency and 4) economic growth. The Figures should be read as follows. For instance, the dark green line or bar ('Economic growth') shows the resource consumption which would have been taken place if only economic growth determined this consumption and other factors did not have had a dampening effect. If one adds up all four contributions, the result is the size of the change in resource consumption in the orange line or bar ('Abiotic resource consumption').

The main line of our results is that between 1996 and 2022, that substitution, recycling and resource efficiency apparently go hand in hand with dematerialisation over time. These factors were mitigating the upward pressure of economic activity. Only in more recent years, this apparently led to a substantial decrease of primary abiotic resource consumption.

In the time period 1996-2022, an overall decline of 15 percent in abiotic resource consumption (solid orange line in Figure 4.1, 'Abiotic resource consumption') can be observed. The orange bar ('Abiotic resource consumption') in Figure 4.2 also shows that resource consumption decreased on balance. However, if we look at the two subperiods 1996-2009 and 2009-2002 (Figures 4.3 and 4.4), it is clear that nearly all of the decrease took place after 2009. Then apparently more efforts were made to reduce materials use.

Turning to the potential drivers behind the reduction on material consumption, we see in Figures 4.1 and 4.2 that the strongly increased economic activity is the main driver behind increases in resource consumption. That is, without the mitigating contributions by the other three drivers, primary abiotic resource consumption would have increased substantially. But in the period before 2009 the upward pressure of economic growth on material consumption was larger than in the period thereafter (Figures 4.3 and 4.4).

As mentioned above, the three other drivers play their expected role. They all exert a downward pressure on primary abiotic material consumption, dampening the impact of economic growth. Substitution, recycling and resource efficiency apparently go hand in hand with dematerialisation over time.

But we see that substitution and recycling seem to have a small impact on balance (Figure 4.1 and 4.2). Increased substitution had a somewhat larger contribution to the decline of resource consumption than increased recycling. The recycling rate is already around 80 percent in the Netherlands. Therefore, in the near future we cannot expect recycling to have much additional impact on resource use reduction. The contribution of recycling was larger before 2009 than after 2009 (Figure 4.3 and 4.4). The contribution of substitution of abiotic by biotic materials is probably overrated because food and feed are also included. However, this driver can have more impact in the future, because currently the biobased economy is still small but might take flight. In the period after 2009, the downward pressure of substitution was clearly present, while it is apparently nearly absent before 2009.

The driver that counteracts the effect of economic growth most is resource efficiency. If the increase in resource efficiency had not taken place, consumption of resources had grown significantly. However, this driver is not well defined. Resource efficiency is defined as total material consumption divided by GDP. These two variables are different in nature, being physically or economic. A direct link between the two is difficult to establish. The reasons behind increase of efficiency can be manifold. These may lie in specific developments of the numerator (total materials consumption), in the denominator (GDP), and/or in the interplay between the two elements of the ratio. A reduction in material consumption while holding GDP constant leads to a decrease of the ratio, hence an improvement of resource efficiency. On the other hand, an increase in GDP while holding material consumption constant, also leads to a decrease of the ratio. Hence what is actually happening in the nominator or denominator of the ratio is important to know.

For instance, the ratio decreases (and resource efficiency improves) when resource intensive industry moves abroad and is replaced by a more service driven sector. This structural change of the economy affects the use of primary abiotic resources as the production of goods is more material intensive than the production of services. Such a structural change took place in the Netherlands after the Second World War. However, if we look at the share of various sectors in gross value added in more recent years, we see rather a shift out of mining as a consequence of the downward re-evaluation and closing of the Groningen natural gas field, than a shift out of manufacturing. The share of manufacturing remains relatively stable. We observe only a small further shift to services. Another possible factor is a change in importance of certain economic activities like repair, refurbishment or re-use. Without further sectoral analysis with data of higher quality, we cannot state that the Dutch economy actually produces with a higher material efficiency (less materials per unit of value added).

For government policy that aims for a transition towards a circular economy it would be relevant to know what exactly is behind the observed ‘resource efficiency’ improvement. Further investigation is needed to divide the resource efficiency driver into relevant components, including the shift of producing goods to producing services. Tentative experiments with a sectoral shift model stumbled on lack of long run data on materials use by sector. Whether there are real improvements in resource efficiency probably has to be assessed at a lower aggregation level, in particular at the sectoral or company level, or even at the product level. In this, we need more detailed data of high quality on materials use at a lower aggregation level.

In this paper, we presented an index decomposition analysis (IDA) of changes in primary abiotic material consumption in the Netherlands. Our main result shows that between 1996 and 2022, substitution, recycling and resource efficiency apparently go hand in hand with dematerialisation over time. These factors were mitigating the upward pressure of economic activity. Only in more recent years, this apparently led to a substantial decrease of primary abiotic resource consumption.

We emphasize that our results must be interpreted carefully. The decomposition analysis does not present a causal link but can uncover correlations. The driver ‘resource efficiency‘ is still subject of research. We aimed to contribute by decomposition analysis of primary abiotic resource consumption in two ways:

• A framework of factors underlying changes in primary abiotic resource consumption:
  There are four factors: substitution of abiotic for biotic materials, recycling, resource efficiency and economic growth. Our aim was to investigate correlations between the change in the resource consumption and the change of the various factors. Quantified contributions cautiously indicate the extent of direct impact of the various factors.
• A long time perspective:
  Our IDA revealed, among other things, that substitution, recycling and resource efficiency apparently go hand in hand with dematerialisation over time. The results show the potential of long time series data for an analysis providing insight in structural changes in substitution, recycling, efficiency, and economy over time.
  
  There are various potential research lines for the future, which are often intertwined:
• Measurement of resource use:
  In Annex A, we already investigated an alternative measurement of resource use, namely input (DMI) instead of consumption (DMC). We might scrutinize further on data and measurement of resource use. Among other things it is important to distinguish re-exports from domestic import and export, and to distinguish food and feed from other biomass, in order to be able to interpret results based on DMC and DMI
• Defining resource efficiency:
  Further, we need to discuss the definition and measurement of ‘resource efficiency’, the ratio ‘total material consumption to GDP’. Do domestic producers indeed use less materials in production, or does the ratio change mainly due to other factors such as economic structural change or a shift to industries abroad?
• Introducing sectoral shift:
  An experimental estimation with an IDA model with a factor measuring sectoral economic shift stumbled upon lack of high quality sectoral data on material use.
• Additional sectoral analysis:
  With high-quality data would be also a valuable addition to insight in material flows and use and the factors underlying these flows and use. Think of, for instance, input-output data on materials used in production chains and footprints.

However, such data improvements are very extensive and bring their own data and measurement issues, whereas we currently just aimed for a first impression of potential relationships in an integral long run framework.

As an alternative to DMC, we measure the materials use with DMI (Domestic Materials Input). DMC is equal to domestic extraction plus imports minus exports, whereas DMI is equal to extraction plus imports. The idea behind measuring with DMI is that there is a more direct link with production with the use (input) of materials in this production. Also because of this direct link, it might be more appropriate to measure economic activity by (gross) production value rather than GDP. In the main text, we utilize the DMC indicator and GDP in the IDA.

However, data in the MFA do not distinguish re-export flows in the import and export data. By estimating DMC, the re-exports are cancelled out, though we do not know the exact size of these re-exports in the total of import or export. DMC is also more easily compared between countries as it accounts for exports and there is no double counting of material flows. By applying DMI, we actually take into account amounts of imports which will be re-exported and not used domestically. Re-exports do not become part of the domestic production process but are goods that enter and leave the Netherlands without hardly any processing. This gives a bias in interpreting the results based on DMI for the Netherlands. For instance, transport and storage of resources by refineries in the port of Rotterdam are partially re-exports.

With these caveats in mind, we observe the following. The DMI data show that primary abiotic resource input actually increased between 1996 and 2022 (see Figure A.1). This is in contrast to the decrease when resource use is measured with DMC (Figure 3.1 in the main text). This might partially be a data issue as DMI includes imports that will be re-exported abroad, and not used in domestic production. DMC omits such re-exports. Whether the increase also partially represents a real development, e.g. increased imports due to shift of manufacturing industries to abroad, is not clear beforehand.

Recycling data remain the same in this alternative application. Finally, we apply gross production value from the National Accounts. Gross production increased from 987 billion euro in 1996 (in constant prices of 2021) to 1823 billion euro in 2022, increasing with 85 percent in 26 years.

The IDA results with DMI and production are displayed in Figures A.2 to A.5, similar to Figures 4.1 to 4.4 in the main text. They seem to tell a similar story, though there are differences. Primary abiotic resource input (DMI) increases, whereas the resource consumption (DMC) decreased. Particularly before 2009 the DMI increased strongly (Figures A.4 and A.5).

The relative roles of the drivers are more pronounced with DMI than in the DMC variant. With DMI and production, we see the relative contributions of production growth and changes in resource efficiency are larger than in the results based on DMC and GDP, and the role of substitution and recycling smaller. Particularly the relative contribution of recycling is nearly absent (or underrated) in the whole period, following from the fact that the recycling amounts are the same as in the DMC analysis but material amounts are much larger in the DMI measure.