Green Fiscal Instruments and the Circular Economy Transition in the EU-27: Does the Structure of Environmental Taxation Matter?

1. Introduction

The transition to a circular economy is one of the most ambitious economic objectives of the European Union. The core idea is that materials already introduced into the economy should be used for as long as possible through recycling, repair and reuse, rather than extracting new raw materials and increasing the volume of waste. The significance of this transition is not solely environmental. The European Union imports a large proportion of the raw materials on which its industry depends, which is why every tonne of secondary material simultaneously reduces both the pressure on nature and external resource dependence. The Circular Economy Action Plan sets the target of doubling the share of secondary materials in overall consumption by 2030. Real progress, however, remains limited: in 2024 this share was only 12.2% and has barely changed over the past decade. Behind the EU average lie considerable differences between member states as well – from over 30% in the Netherlands to below 2% in some countries.

This raises the question of which instruments policy can use to accelerate the circular transition. One of the possible instruments is taxation. The logic of environmental taxation is well known: when pollution and resource use become more expensive, firms and households have a stronger incentive to save materials and seek more efficient solutions. In practice, however, the green tax systems in the EU are constructed differently. Around three-quarters of environmental tax revenues come from the taxation of energy, which is directed primarily at fuels rather than at material flows. Pollution and resource taxes, which are most directly connected to the idea of a circular economy, account for only 4–5% of revenues, and in some member states are almost absent. Moreover, the overall environmental tax burden in the EU is not increasing but declining – from around 2.7% of GDP in 2010 to 2.2% in 2024. This gives rise to an important paradox: policies increasingly emphasise the circular economy, yet the fiscal systems remain dominated by energy taxation.

The economic literature examines in detail the relationship between environmental taxes, emissions, energy efficiency and innovation. Considerably less is known, however, about whether and how the tax structure influences circular outcomes themselves – the share of secondary materials, recycling, value added and employment in the circular sectors. Even less studied is the territorial and sectoral dimension of this question. Agriculture and rural areas occupy a specific place in the circular agenda. A substantial part of circular practices there, such as the valorisation of biomass, composting and on-farm reuse, takes place outside the formal sectors that statistics measure. At the same time, the waste infrastructure in these areas is sparser and more expensive to maintain. The question of whether identical fiscal instruments lead to identical circular outcomes in agricultural and in urbanised economies therefore has direct relevance for policy, but has so far rarely been posed empirically.

The present article examines these questions through a panel of the 27 EU member states over the period 2010–2024, and 2005–2023 respectively for the economic indicators of the circular sectors. The database is constructed entirely from official Eurostat data. Circular outcomes are measured through five indicators from the EU's official monitoring framework, while the fiscal side is measured through environmental tax revenues and their structure. The study tests four hypotheses: whether the overall environmental tax burden is associated with circular outcomes; whether it is not only the size but the structure of taxation that matters; whether the fiscal effects manifest with a lag; and whether the effect depends on the agricultural profile of the economy. The methodology combines panel models with two-way fixed effects and clustered standard errors, lag specifications, dynamic system GMM, and models with interactions and conditional marginal effects.

The contribution of the article is that it offers a systematic assessment of the relationship between the structure of environmental taxation and the official circular economy indicators in the EU; it introduces the agricultural dimension into this analysis; and it draws practical conclusions for the differentiation of green fiscal policies.

2. Literature Review and Hypotheses

2.1. Theoretical Framework

The theoretical foundation of the study rests on three related strands: the theory of environmental taxation and the double dividend, the Porter hypothesis, and the theory of policy instruments for the circular economy (Pigou, 1920; Baumol & Oates, 1988; Pearce & Turner, 1990; Fullerton et al., 2010; Ekins & Speck, 2011). The common logic among them is that markets often fail to reflect the full environmental cost of production, consumption and resource use. When pollution, waste or the extraction of primary raw materials is not correctly priced, firms and households have a weaker incentive to reduce material consumption, to recycle, or to use secondary resources. In this sense, environmental taxes can be regarded as an instrument for correcting these externalities.

The idea of the double dividend of environmental taxation suggests that environmental taxes can have two simultaneous effects. The first is environmental: making environmentally harmful activities more expensive should reduce pollution and the inefficient use of resources. The second is economic: the revenues from these taxes can be used to reduce other distortionary taxes, for example on labour or capital, or to finance environmental investments (Pearce & Turner, 1990; Bovenberg & de Mooij, 1994; Goulder, 1995; Ekins & Speck, 2011). Later reviews of the literature show that the double dividend is not an automatic outcome but depends on the design of the tax, on the use of the revenues, and on the structure of the economy (Schöb, 2003; Freire-González, 2018).

The relationship between environmental regulation and economic outcomes is also considered through the Porter hypothesis. According to it, well-designed environmental policies can stimulate innovation that partially or fully offsets the initial costs of compliance with the regulations (Porter, 1991; Porter & van der Linde, 1995; Jaffe & Palmer, 1997; Ambec & Barla, 2002; Lanoie et al., 2011; Ambec et al., 2013). In the context of the circular economy, this means that fiscal incentives can encourage firms to adopt more efficient technologies, to reduce material losses, to develop repair and recycling activities, and to use secondary raw materials. This effect, however, is likely only when the instrument is sufficiently clearly targeted at the corresponding environmental problem.

The theory of policy instruments for the circular economy complements this framework. The circular transition is not exhausted by reducing emissions or by lower energy consumption. It requires slowing, narrowing and closing material flows through product design, repair, reuse, recycling and new business models (Ghisellini et al., 2016; Bocken et al., 2016; Geissdoerfer et al., 2017; Kirchherr et al., 2017; Reike et al., 2018). Policy instruments must therefore be directed not only at energy, but also at materials, waste and resource efficiency (OECD, 2025; European Environment Agency, 2025). In this sense, the overall environmental tax burden may be too broad an indicator if it is dominated by energy taxes, which do not act directly on material flows.

From this follows the main theoretical logic of the present study. If environmental taxation acts as a genuine incentive for the circular transition, a higher tax burden should be associated with better circular outcomes. But if the structure of taxation is dominated by taxes directed at energy and transport rather than at pollution and resources, this effect may be weak or unclear (Fullerton et al., 2010; OECD, 2025; European Environment Agency, 2025). It is therefore necessary to examine not only the overall size of environmental taxes, but also their internal structure.

2.2. Empirical Studies on Environmental Taxes and Environmental Outcomes

The empirical literature on environmental taxes is relatively rich, but its main focus is on emissions, energy consumption and energy efficiency. Studies on EU and OECD countries typically test whether environmental taxes lead to a reduction in pollution or to a change in energy behaviour (Bosquet, 2000; Miller & Vela, 2013; Aydin & Esen, 2018; Wolde-Rufael & Mulat-weldemeskel, 2023). Morley (2012), for example, uses a panel of EU countries and Norway and finds a negative relationship between environmental taxes and pollution, but does not find a clear relationship with energy consumption. This shows that environmental taxes can have an effect on some environmental outcomes, but this effect is not the same for all indicators.

Similar conclusions are found in more recent studies as well. Some of them confirm that environmental taxation can contribute to reducing emissions, especially when it is linked to carbon pricing or to taxes on energy products (Andersson, 2019; Best et al., 2020; Metcalf & Stock, 2023; Kohlscheen et al., 2025; Döbbeling-Hildebrandt et al., 2024). Others, however, emphasise that the effect depends on the tax structure, on the level of economic development, on the energy mix, and on the institutional environment (OECD, 2017; European Environment Agency, 2016, 2022; Aydin & Esen, 2018; Miller & Vela, 2013). This means that environmental taxes cannot be regarded as a homogeneous instrument. The same overall size of the tax burden may contain very different incentives depending on which activities are taxed (Blagoeva & Georgieva, 2023).

For the present study, the distinction between energy, transport, pollution and resource taxes is particularly important. In the European context, environmental tax revenues are heavily concentrated in energy taxes, while pollution and resource taxes account for a much smaller share (Eurostat, 2024; OECD, 2017). This creates a risk that the overall indicator of the environmental tax burden may conceal different mechanisms. Energy taxes can support climate objectives, but do not necessarily directly encourage recycling, reuse, or the replacement of primary with secondary materials. Conversely, pollution and resource taxes are closer to the logic of the circular economy, because they are connected to waste and material flows (Mayer et al., 2019; Haupt et al., 2017).

Nevertheless, empirical studies rarely link environmental taxes to the official circular economy indicators. Most studies remain within the bounds of emissions, energy or overall environmental efficiency (Bosquet, 2000; Morley, 2012; Miller & Vela, 2013; Döbbeling-Hildebrandt et al., 2024). This creates a research gap. It is still not sufficiently clear whether environmental taxes are associated with a higher degree of circular material use, with better recycling, or with the development of the circular sectors in terms of value added, investment and employment (Saidani et al., 2019; Mayer et al., 2019; Pauliuk, 2018; Afolabi & Islam, 2025; Georgieva, 2026). The present study addresses precisely this gap, by testing the relationship between the tax structure and several different circular outcomes in the EU-27.

2.3. Measuring the Circular Economy

Measuring the circular economy is a methodological challenge in its own right. The circular economy is a multidimensional process and cannot be exhausted by a single indicator. It encompasses the use of secondary materials, the reduction of waste, recycling, repair, reuse, innovation, investment and changes in employment. For this reason, various approaches to measuring it exist in the literature – from individual indicators to composite indices (Saidani et al., 2019; Claudio-Quiroga & Poza, 2024; Krasteva-Hristova & Moneva, 2026).

The European Union's official framework for monitoring the circular economy offers a systematic set of indicators grouped into several strands. These include production and consumption, waste management, secondary raw materials, competitiveness and innovation, as well as global sustainability and the sustainability of supply (European Commission, 2020; Eurostat, 2024). This framework is particularly suitable for comparative analysis between member states, since it uses harmonised Eurostat data.

The lead indicator in this framework is the circular material use rate, known as CMUR. It measures the share of secondary materials in the total use of materials in the economy. A high value of CMUR means that a larger proportion of materials is returned to the economic cycle, rather than being replaced by new primary raw materials. This indicator is suitable for assessing material circularity, but it also has limitations. It does not capture all forms of reuse, repair or internal circularity within firms and farms. It is therefore necessary to use it together with other indicators.

The recycling rate of municipal waste measures a narrower dimension of the circular economy. It shows the efficiency of systems for the collection, sorting and processing of waste. This indicator is important, because recycling is one of the most visible elements of the circular transition. At the same time, it does not cover all material flows in the economy and may underestimate circular practices that take place outside municipal waste systems.

The economic dimension of the circular economy is captured through the indicators of gross value added, private investment and employment in the circular sectors. These are important, because they show whether the circular transition is translating into economic activity, jobs and an investment process. The literature increasingly emphasises that the circular economy should not be regarded solely as an environmental strategy, but also as an economic transformation linked to new business models and a change in production systems (Bocken et al., 2016; Geissdoerfer et al., 2017; Kirchherr et al., 2017).

Alongside the official indicators, composite indices for the circular economy are also being developed. They attempt to combine different dimensions of the transition into a single aggregate indicator. Claudio-Quiroga and Poza (2024), for example, construct a multidimensional index of the circular economy in Europe, using the European Commission's approach and including production and consumption, waste management, secondary raw materials, competitiveness and innovation. Such indices are useful for ranking and comparison between countries, but may conceal the differences between the individual dimensions of the circular transition. For this reason, the present study uses several separate dependent variables, rather than relying on a single composite index alone.

2.4. The Circular Economy in Rural Areas and Agriculture

The circular economy has a specific significance for rural areas and agriculture. In agricultural systems, material flows are often linked to biomass, organic residues, soil resources, water and local chains of production and consumption. Circular practices in agriculture can therefore include composting, the valorisation of plant and animal residues, the reuse of resources on farms, the reduction of losses, and the development of the bioeconomy (Toop et al., 2017; Donner et al., 2020; Hamam et al., 2021; Muscat et al., 2021). These practices are close to the logic of the circular economy, but are not always clearly reflected in the official statistical indicators.

The literature on the circular economy in agriculture emphasises that the agricultural sector has considerable potential for applying circular principles, but also has specific constraints. Velasco-Muñoz et al. (2021) point out that the theoretical framework of the circular economy is not yet fully adapted to the particularities of agriculture. The reason is that agricultural systems operate with biological cycles, seasonality, spatial dispersion and a strong dependence on natural conditions. This distinguishes them from the industrial systems on which much of the circular economy literature was originally built.

Rural areas also face infrastructural constraints. Systems for the collection, sorting and processing of waste are typically sparser, more expensive and more difficult to maintain in territories with a low population density. This can limit the effect of fiscal incentives. Even when taxes create the correct price signal, households and firms may not have a real opportunity to respond through recycling, reuse or participation in formal circular chains. The territorial context is therefore important for assessing the effectiveness of green fiscal instruments (OECD, 2020a; OECD, 2020b; Girard, 2022).

The relationship between the circular economy, the bioeconomy and rural areas is particularly important for the present study. The bioeconomy allows agricultural residues, biomass and organic flows to be regarded as a resource rather than as waste (European Commission, 2018; Stegmann et al., 2020; Muscat et al., 2021; Barañano et al., 2021). In this sense, circular practices in rural areas can have considerable environmental and economic potential. But if they take place outside the formal sectors, the official indicators of value added, investment and employment in the circular economy may underestimate them. This creates the possibility that the effect of environmental taxes may appear weaker in agricultural economies, not only because the actual response is weaker, but also because part of it remains statistically invisible.

This particularity motivates the inclusion of the agricultural profile as a moderating factor. If economies with a higher agricultural specialisation have a different structure of material flows, a weaker waste infrastructure and a larger share of informal or on-farm circular practices, then the effect of environmental taxation on the officially measured circular indicators may be weaker or different (Girard, 2022; Bianchi et al., 2024; Khanna et al., 2024). This is particularly important for policy, because identical fiscal instruments may not yield identical results across all member states.

On the basis of the theoretical and empirical literature reviewed, the present study formulates the following hypotheses:

H1: A higher environmental tax burden, measured as a percentage of GDP, is positively associated with circular outcomes, including the circular material use rate and the recycling of municipal waste.

H2: Resource and pollution taxes have a stronger positive effect on circular outcomes compared with energy and transport taxes.

H3: The effect of environmental taxation on employment and value added in the circular sectors is positive, but manifests with a lag.

H4: In countries with a higher share of agriculture, the effect of environmental taxation on circular outcomes is weaker or different; that is, the agricultural profile moderates the relationship between fiscal instruments and the circular transition.

3. Materials and Methods

3.1. Data and Sample

The study uses panel data for the 27 member states of the European Union. The database is constructed entirely from official Eurostat sources. The full set of collected information covers the period 2000–2024, but the actual estimation periods are determined by the availability of the individual dependent variables.

For the two main indicators of the circular transition – the circular material use rate and the recycling of municipal waste – the analysis covers the period 2010–2024. For these indicators, 405 observations are available for the EU-27, respectively. For the three economic indicators of the circular sectors – gross value added, private investment and employment – the period is 2005–2023, with 513 observations used for each of them. The difference in the periods is due to the different coverage of the series published by Eurostat.

Table 1 presents the variables used, their definitions, units of measurement and sources. All sources are Eurostat datasets, extracted in July 2026. Some of the variables are used directly from Eurostat, while others are calculated by the authors on the basis of the published data. In the table, these variables are designated as derived.

Circular outcomes are measured through five indicators from the EU's official monitoring framework for the circular economy. The main indicator is the circular material use rate, denoted as CMUR. It shows what proportion of the materials used in the economy comes from secondary materials, that is, from materials that have been returned to the economic cycle. For this reason, CMUR is used as the lead indicator of progress towards a circular economy.

It should be borne in mind that the underlying waste statistics are collected at two-year intervals. For this reason, some of the intermediate annual values are estimated by Eurostat. This does not invalidate the use of the indicator, but requires the results to be interpreted with care, especially when short-term changes are considered.

The second indicator is the recycling rate of municipal waste. It reflects a narrower dimension of the circular economy, related specifically to waste management. While CMUR measures the return of materials to the economy more generally, the recycling of municipal waste shows how effectively the systems for the collection and processing of waste function.

The remaining three dependent variables describe the economic dimension of the circular transition. These are the gross value added of the circular sectors, private investment in these sectors, and employment in them. The circular sectors include activities such as recycling, repair and reuse. Gross value added and private investment are expressed as a percentage of GDP, while employment is measured in full-time equivalent per 1,000 inhabitants.

The fiscal variables are based on Eurostat statistics on environmental taxes. The main explanatory variable is the overall environmental tax burden, measured as environmental tax revenues as a percentage of GDP. In addition to it, the three main categories of environmental taxes according to Eurostat's standard classification are also used: energy taxes, transport taxes, and pollution and resource taxes.

The overall environmental tax burden and the three main categories are taken directly from Eurostat's env_ac_tax dataset. However, since pollution and resource taxes are published there as a single category, for the purposes of the study they are separated into two distinct variables: pollution taxes and resource taxes. This separation is carried out using the data by economic activity from the env_ac_taxind2 dataset. The values in millions of euros for all economic activities and households are used, after which they are related to GDP at current prices. The derived indicators for pollution taxes and for resource taxes as a percentage of GDP are obtained in this way.

Control variables are also included in the analysis. The first is GDP per capita, measured in purchasing power standard and used in logarithmic form. This indicator controls for the differences in the level of economic development between countries.

The second control variable is the intensity of research and development activity. It is expressed as R&D expenditure as a percentage of GDP. This indicator is important, because the circular transition often depends on technological solutions, innovation, new production processes and more efficient management of materials.

The third control variable is the agricultural specialisation of the economies. It is measured through a location quotient, denoted as LQ. This coefficient shows whether a given country has a stronger or weaker agricultural profile compared with the average level for the EU-27. It is calculated as the ratio between the country's share in the agricultural gross value added of the EU-27 and its share in the total gross value added of the EU-27. A value above one means that the country has an agricultural specialisation above the average for the European Union.

Since LQ is calculated as a ratio of shares within the same year, the overall EU level across individual years is accounted for through the year fixed effects. This allows the indicator to be used for comparison between countries.

3.2. Multicollinearity

Before estimating the models, a check for multicollinearity between the explanatory variables was carried out. This is necessary, because a strong dependence between the regressors can hinder the reliable estimation of the individual effects.

The check was performed through the variance inflation factors, known as VIF. The values obtained do not indicate a problem with multicollinearity. In the baseline specification, the maximum value of VIF is 2.43, and in the specification by categories of environmental taxes it is 2.46. Both values are well below the conservative threshold of 5, which is usually used as a guide to a potential problem.

The strongest pairwise correlation between the explanatory variables is between the logarithm of GDP per capita and agricultural specialisation. It is negative and amounts to -0.72. This is to be expected, because the more strongly agricultural economies in the EU typically have lower levels of income per capita. This relationship, however, does not create a serious problem for the estimates, since the models include fixed effects for country and year.

The descriptive statistics of all variables are presented in Section 4.1, in Table 2.

3.3. Econometric Strategy

The empirical strategy is constructed so as to correspond to the four hypotheses of the study. Four main types of specification are used: a baseline fixed-effects model, a model by categories of environmental taxes, a dynamic specification, and a model with an interaction according to the agricultural profile of the economy.

First, to test H1, a linear panel model with fixed effects for country and year is used. It has the following general form:

Y_it = β·ETAX_it + γ′X_it + μ_i + λ_t + ε_it

(1)

In this equation, Y_it is the corresponding circular indicator for country i in year t. ETAX_it is the overall environmental tax burden. X_it is the vector of control variables, which includes the logarithm of GDP per capita, R&D intensity and agricultural specialisation. μi denotes the country fixed effects, λt the year fixed effects, and εit is the error term of the model.

The country fixed effects make it possible to account for all persistent national characteristics that do not change substantially over time. These include, for example, the institutional environment, geographical features, sectoral structure and long-term differences in infrastructure. The year fixed effects account for changes common to all countries, including European policies, legislative changes, economic shocks and other factors that affect the entire EU in a given year.

The standard errors are clustered at the country level. This makes it possible to account for the possible dependence between observations within the same country over time. There are 27 clusters in the analysis, corresponding to the 27 member states.

To test H2, the baseline model is re-estimated, with the overall environmental tax burden replaced by the individual categories of environmental taxes. First, the three main categories are used: energy taxes, transport taxes, and pollution and resource taxes. The last category is then further divided into pollution taxes and resource taxes. This tests whether the structure of environmental taxation matters more than its overall size.

To test H3, two dynamic extensions are used. The first includes the current value of the environmental tax burden and two of its lags. The model has the following form:

Y_it = Σ_k=0..2 β_k·ETAX_i,t−k + γ′X_it + μ_i + λ_t + ε_it

(2)

This specification makes it possible to test whether the fiscal effect manifests in the same year, after one year or after two years. The long-term effect is calculated as the sum of the three coefficients – the current one, the first lag and the second lag.

The second dynamic extension takes into account the fact that circular indicators usually change slowly and depend on their own values in previous years. For this purpose, a dynamic model of the following form is used:

Y_it = ρ·Y_i,t−1 + β·ETAX_it + γ′X_it + μ_i + λ_t + ε_it

(3)

This model is estimated through a two-step system GMM approach of Blundell–Bond with the Windmeijer correction of the standard errors. The method is suitable for dynamic panel data, when the dependent variable is strongly inertial and when endogeneity between the variables is possible.

In order to avoid the problem of instrument proliferation, the instrument matrices are collapsed, and the depth of the instruments is limited. Lags of the second and third year are used for the dependent variable, and lags of the first and second year for the tax variable, which is treated as predetermined. The control variables are instrumented by themselves. In this way, the number of instruments remains between 23 and 27 and does not exceed the number of groups.

The validity of the dynamic models is checked through the Arellano–Bond tests for first- and second-order serial correlation, as well as through the Hansen test for the overidentifying restrictions. These tests are necessary in order to assess whether the instruments used are appropriate and whether the models do not suffer from serious diagnostic problems.

To test H4, a model with an interaction term between the overall environmental tax burden and agricultural specialisation is used. The model has the following form:

Y_it = β₁·ETAX_it + β₂·(ETAX_it×LQ_it*) + β₃·LQ_it* + γ′X_it + μ_i + λ_t + ε_it

(4)

In this equation, LQ_it* is agricultural specialisation, centred around a reference value of 1.415, which corresponds to the average for the full panel over the period 2000–2024. This value is used as a common reference point for all models, since the estimation samples for the individual dependent variables cover different periods and have different means (1.33 for CMUR and recycling; 1.38 for the indicators of the circular sectors). This means that the coefficient β₁ shows the effect of the environmental tax burden at this reference level of the agricultural profile. The coefficient β₂ shows whether this effect changes when agricultural specialisation increases or decreases. The choice of centring constant does not affect β2, nor the conditional marginal effects calculated at specific values of LQ.

The conditional marginal effects are calculated as:

∂Y/∂ETAX = β₁ + β₂·LQ*

(5)

They show what the effect of the environmental tax burden is at different levels of agricultural specialisation. The standard errors of these effects are calculated by the delta method. The effects are presented at three representative levels of LQ, corresponding to the group means used in the descriptive analysis.

4. Results

4.1. Descriptive Analysis

Table 2 presents the main characteristics of the variables used for the 27 member states of the European Union over the period 2010–2024.

The mean value of CMUR over the period is 9.05%. This value, however, conceals considerable differences between individual countries. The lowest recorded value is 1.2% for Romania, and the highest is 32.7% for the Netherlands in 2024. In other words, the leading country has a level of circular material use more than 25 times higher than that of the country with the lowest result.

Over time, a gradual improvement is observed. The unweighted mean for the countries increases from 7.9% in 2010 to 10.9% in 2024. The median also rises – from 5.9% to 9.4%. This shows that the transition towards a more circular economy is progressing, albeit slowly. Nevertheless, the levels attained are still far from the European Union's target of doubling circularity by 2030.

The differences between countries remain clearly pronounced. The greatest improvement over the period is observed in Malta, Estonia, Italy, the Czech Republic and Belgium. In these countries, the circular material use rate increases by 13.3, 11.5, 10.3, 9.5 and 9.2 percentage points respectively. At the same time, in five countries the indicator declines. The deterioration is most marked in Luxembourg, where the fall is 12.7 percentage points, and in Finland, where the decrease is 8.5 percentage points. This means that the overall progress in the EU is not distributed evenly between countries.

The regional comparison also shows substantial differences. The mean value of CMUR in the western and northern economies is 12.4%. In the southern countries it is 7.6%, and in the countries of Central and Eastern Europe it is 6.8%. Higher levels of circularity are therefore concentrated mainly in the western and northern parts of the EU. This confirms that the circular transition depends not only on common European policies, but also on the economic, institutional and technological capacities of the individual countries.

A similar, but more clearly pronounced, positive trend is observed in the recycling of municipal waste. This indicator is a narrower measure, because it reflects specifically waste management rather than the overall use of materials in the economy. Its mean value rises from 26.8% in 2010 to 41.0% in 2023. This shows progress in waste management, but does not automatically imply the same level of progress in the overall circularity of materials.

For the fiscal indicators, the picture is different. Total environmental tax revenues decline from an average of 2.70% of GDP in 2010 to 2.19% in 2024. At the end of the period, the lowest value is recorded in Ireland – 0.98% of GDP, and the highest in Greece, where environmental taxes reach 3.76% of GDP. Consequently, while the circularity indicators gradually improve, the overall burden of environmental taxation as a share of GDP is rather declining.

More important, however, is the question of the structure of these taxes. The data show that environmental taxation in the EU is heavily concentrated on energy taxes. They account, on average, for between 76% and 78% of all environmental tax revenues. Transport taxes have a considerably smaller share – between 17% and 20%. Pollution and resource taxes, which are most directly connected to material flows and resource efficiency, account for only 4–5% of total revenues.

This result is particularly important for the present study. Pollution and resource taxes ought to be closest to the logic of the circular economy, because they are directed at limiting pollution, the use of resources and the pressure on the environment. In reality, however, their fiscal share is very small. The additionally calculated indicators by economic activity show that pollution taxes amount, on average, to only 0.08% of GDP, and resource taxes to 0.04% of GDP. In some countries the values are zero. This shows that the instruments that can most directly influence resource use still have a limited significance in the structure of environmental taxation. It is precisely this result that underlies the second hypothesis of the study.

Pollution taxes and resource taxes, considered separately, are calculated as derived indicators. They are obtained from Eurostat's data by economic activity and related to GDP. Owing to data availability, these indicators cover the period up to 2023. The same applies to the indicators related to the circular sectors.

The economic dimensions of the circular transition remain relatively stable as a share of GDP. The gross value added of the sectors connected to the circular economy is, on average, 1.78% of GDP. Private investment in these sectors is, on average, 0.67% of GDP. A clearer positive change is observed in employment. It increases from 10.1 to 12.8 full-time equivalents per 1,000 inhabitants between 2010 and 2023. This shows that the circular sectors are gradually expanding their role in the labour market, even though their economic share in GDP remains comparatively limited.

The agricultural profile of the economies is also included in the analysis. It is measured through a location quotient relative to the average level for the EU-27. This coefficient shows the extent to which a given country is more strongly or more weakly specialised in agriculture compared with the EU average. The values range from 0.75 in the western and northern countries to 1.88 in the countries of Central and Eastern Europe. This difference is substantial, because it makes it possible to test whether agricultural specialisation matters for the relationship between environmental taxation and circularity. This is connected to the fourth hypothesis of the study.

The initial correlations give only an indicative impression of the relationships between the variables. In the pooled sample, the relationship between the overall environmental tax burden and circularity is very weak and positive (r = 0.03). A clearer positive relationship is observed for pollution and resource taxes (r = 0.17). These results, however, are not sufficient for definitive conclusions. They do not account for the persistent differences between countries, nor for the common changes over time that may affect all economies simultaneously. For this reason, in the next section the analysis moves to panel models with fixed effects.

4.2. Baseline Results: Overall Environmental Tax Burden and Circular Indicators

This section presents the baseline results from the panel models with fixed effects for country and year. The aim is to test whether the overall environmental tax burden is associated with the level of development of the circular economy in the EU countries.

Table 3 includes five separate models. In each of them, the dependent variable is a different indicator of the circular economy: the circular material use rate (CMUR), the recycling rate of municipal waste, the gross value added of the circular sectors, private investment in these sectors, and employment in the circular economy. The main explanatory variable is the total value of environmental taxes as a percentage of GDP.

The models are estimated by ordinary least squares with fixed effects for country and year. The country fixed effects make it possible to account for the persistent differences between individual economies that do not change substantially over time. These include, for example, the institutional environment, geographical features, sectoral structure and the existing waste management infrastructure. The year fixed effects account for changes and shocks common to all countries, such as the 2022 energy crisis or changes in European legislation. The standard errors are clustered by country, in order to account for the possible dependence between observations within the same country.

The main result from Table 3 is that no statistically significant relationship is found between the overall environmental tax burden and most of the circular economy indicators. For CMUR, the coefficient is negative (-0.868), but it is not statistically significant (p = 0.224). This means that it cannot be concluded that an increase or decrease in the overall environmental tax burden leads to a substantial change in the circular material use rate.

The picture is similar for the remaining dependent variables. For the recycling of municipal waste, the gross value added of the circular sectors and employment in the circular economy, the coefficients are likewise not statistically significant. This shows that, within the period under consideration, the overall environmental tax burden is not systematically associated with better circular outcomes.

The only partial exception is observed for private investment in the circular sectors. There the coefficient is negative (-0.076) and is statistically significant only at the 10% significance level (p = 0.077). This means that an increase in total environmental taxes of one percentage point of GDP is associated with a fall in private investment in the circular sectors of around 0.08 percentage points of GDP. This result, however, should be interpreted with caution, because the significance is weak.

Taken together, the results do not confirm the first hypothesis in its aggregate formulation. In other words, when environmental taxes are considered in aggregate, without distinguishing between their individual types, no convincing evidence is found that a higher environmental tax burden leads to higher circularity.

This result is important, because it shows that the size of environmental taxation alone is not sufficient to explain the development of the circular economy. It is also consistent with the descriptive analysis from the previous section. As noted, the overall environmental tax indicator is heavily dominated by energy taxes. These are connected mainly to the consumption of energy and fossil fuels, whereas circular material use relates to the reuse, recycling and return of materials to the economy. The relationship between energy taxes and material flows is therefore not direct.

Moreover, the decline in the overall environmental tax burden over the period may reflect not so much a weakening of environmental policy as a change in the tax base itself. For example, electrification, the increase in energy efficiency and the decline in the consumption of certain fuels can reduce the revenues from energy taxes. Thus the overall indicator mixes two different effects: on the one hand, the political price signal of the taxes, and on the other, the mechanical reduction of the tax base.

The way in which the revenues from environmental taxes are used is also of additional significance. If these revenues are not directed towards circular economy measures, for example towards recycling, reuse, repair or resource efficiency, then the fiscal signal may remain weak. This explains why the overall size of environmental taxation is not necessarily a sufficient factor for improving the circular indicators.

The results therefore point to a more important conclusion: what matters is not only how large the environmental tax burden is, but what its structure is. If environmental taxation is concentrated mainly on energy and transport, it may have a limited influence on material flows. For this reason, in the next section the overall environmental tax burden is broken down by categories of tax, in order to test whether the individual types of tax have different effects.

Among the control variables, the clearest result is observed for the intensity of research and development activity. An increase in R&D expenditure of one percentage point of GDP is associated with a rise in CMUR of 3.64 percentage points, and the result is statistically significant (p = 0.021). This is logical, because the transition to a circular economy depends to a large extent on technology, innovation, new production solutions and more efficient management of materials.

Agricultural specialisation has a weakly negative effect on employment in the circular sectors. The coefficient is -1.240 and is significant only at the 10% significance level (p = 0.082). This result shows that in the more strongly agricultural economies the formal circular sectors, such as recycling, repair and reuse, may be relatively less developed. The question of whether the agricultural profile also alters the fiscal effect itself is examined further in Section 4.5.

Income per capita is not statistically significant in any of the models. This can be explained by the use of fixed effects for country and year. In this type of model, it is primarily the changes within individual countries over time that are analysed, rather than the constant differences between richer and poorer economies. Part of the income effect is therefore captured by the fixed effects themselves.

The R² (within) indicator in the table shows what proportion of the internal variation over time is explained by the models after the removal of the fixed effects. Its values are relatively low in some of the specifications, which is to be expected for complex processes such as the circular transition. The explanatory power is highest for the model of the gross value added of the circular sectors, where R² (within) is 0.152. For CMUR the value is 0.082, and for recycling it is 0.037. This shows that a considerable part of the dynamics of the circular indicators is probably determined also by other factors that are not included in the baseline specification.

The F-statistics show that the models are, on the whole, statistically significant. The strongest overall significance is observed for the model of the gross value added of the circular sectors (F = 20.75, significant at 1%). The model for CMUR is also significant at the 1% level (F = 8.03). The remaining models are significant at the 5% or 1% level, depending on the specification.

In the table, ***, ** and * denote the levels of statistical significance at 1%, 5% and 10% respectively. The dependent variables are measured as follows: CMUR and recycling are in percentages; the gross value added and private investment in the circular sectors are expressed as a percentage of GDP; employment in the circular sectors is measured as full-time equivalent per 1,000 inhabitants. For CMUR and recycling, the period is 2010–2024, with 405 and 395 observations respectively. For the gross value added, private investment and employment in the circular sectors, the period is 2005–2023, with 513 observations used for each of these models.

4.3. The Structure of Environmental Taxation

Since no clear relationship was found in the previous section between the overall environmental tax burden and the circular indicators, in this part the analysis turns to the structure of environmental taxation. The aim is to test whether the individual types of environmental taxes have a different significance for the development of the circular economy.

Table 4 presents fixed-effects models for country and year, in which the overall environmental tax burden is divided into the three main categories according to Eurostat's classification: energy taxes, transport taxes, and pollution and resource taxes. The dependent variables are the same as in Table 3: the circular material use rate, the recycling of municipal waste, the gross value added of the circular sectors, private investment in the circular sectors, and employment in the circular economy. The models also include the same control variables: GDP per capita, measured in purchasing power standard, R&D intensity and agricultural specialisation. The standard errors are clustered by country.

Table 5 extends the analysis by dividing the last category into two separate components: pollution taxes and resource taxes. These two variables are calculated as derived indicators from Eurostat's data by economic activity. Since the available data for them reach up to 2023, the estimation period for these models is shortened to 2010–2023 for CMUR and recycling. For the indicators related to the circular sectors, the period is 2005–2023.

The results show that the absence of a significant effect for the overall environmental tax burden does not mean that all environmental taxes are irrelevant. On the contrary, when the taxes are considered separately, it becomes apparent that the different categories have a different direction and strength of impact. This means that the aggregate indicator conceals important differences in the structure of taxation.

The first important result concerns energy taxes. They account for the largest part of environmental tax revenues, but do not display a positive relationship with the circular indicators. On the contrary, for private investment in the circular sectors the effect is negative and statistically significant. In Table 4 the coefficient is -0.120 at p = 0.016, and in Table 5 it is -0.104 at p = 0.045. This means that higher energy taxes are associated with lower private investment in the circular sectors.

This result can be explained by the nature of energy taxes. They are directed mainly at the consumption of energy and fossil fuels, rather than directly at material flows. At the same time, some of the activities connected to the circular economy, such as waste processing, repair, reuse and recycling, can be energy-intensive. Higher energy taxes may therefore act as an additional cost burden for these activities, rather than as a direct incentive to close material cycles.

In the extended specification in Table 5, a weakly negative relationship is also observed between energy taxes and the recycling of municipal waste. The coefficient is -3.036 and is statistically significant only at the 10% significance level. This means that the result should be interpreted with caution, but it is in the same direction as the effect on investment.

The second important result concerns transport taxes. They display a strong positive relationship with the recycling of municipal waste. In Table 4 the coefficient is 20.256 at p = 0.002, and in Table 5 it is 21.033 at p = 0.003. This means that higher transport taxation is associated with higher levels of recycling.

To assess the practical magnitude of this effect, one can use the standard deviation of transport taxes, which is 0.33 percentage points of GDP. An increase of one such magnitude corresponds to approximately 6.7 percentage points of a higher recycling rate. Nevertheless, this result should not be interpreted as a direct causal effect. It is more likely that transport taxes reflect a broader environmental ambition of policy. Countries that strengthen transport taxation often also apply other measures for waste management, recycling and resource efficiency.

The third important result concerns pollution taxes. They are the most directly oriented towards limiting waste and polluting flows. In Table 5 they are positively associated with the gross value added of the circular sectors. The coefficient is 1.627 and is statistically significant at the 10% significance level. This means that higher pollution taxes are associated with a higher economic contribution of the circular sectors. Although the significance is weak, the direction of the effect is in line with the expectations of the second hypothesis.

Resource taxes do not display a statistically significant effect in any of the specifications. At first sight, this may appear to be a lack of support for the hypothesis. Such a conclusion, however, would be premature. Resource taxes in the EU have a very small fiscal share – on average around 0.04% of GDP. Moreover, in some countries the values are zero, and their variation over time is limited. With such weak variation, the panel model can hardly establish a reliable effect.

The more accurate interpretation is that resource taxation in the EU is not yet applied on a sufficient scale for its effect on the circular economy to be clearly assessed. This in itself is an important result. It shows that, despite the strategic significance of resource efficiency, resource taxes remain an underused instrument in the green fiscal systems of the member states.

Overall, the results provide partial support for the second hypothesis. The different categories of environmental taxes do indeed have different effects on the circular indicators. Energy taxes are negatively associated with private investment in the circular sectors. Transport taxes display a positive relationship with recycling, but probably reflect a broader environmental policy rather than an independent causal mechanism. Pollution taxes have the expected positive sign with respect to the gross value added of the circular sectors. Resource taxes remain statistically insignificant, but this is due more to their limited application than to a demonstrated absence of an effect.

Consequently, the aggregate environmental tax burden is not a sufficiently good indicator of the fiscal support for the circular transition. It is dominated by energy taxes, which are not directly targeted at material flows. For assessing the relationship between environmental taxation and the circular economy, it is more important to consider not only the size of the taxes, but their structure. The robustness of these results is tested in the next section through a dynamic specification.

In the tables, ***, ** and * denote the levels of statistical significance at 1%, 5% and 10% respectively. The R² (within) indicator shows what proportion of the variation within countries over time is explained by the respective model after accounting for the fixed effects. For the models in Table 4, the R² (within) values range from 0.052 for private investment to 0.201 for the gross value added of the circular sectors. In Table 5, they range from 0.038 for private investment to 0.220 for gross value added. This shows that the models best explain the dynamics of the economic contribution of the circular sectors, but capture the variation in investment more weakly.

4.4. Dynamics: lagged effects and system GMM

This section tests whether the effect of environmental taxation on the circular indicators manifests immediately or with a lag over time. This is important, because the transition to a circular economy usually does not occur immediately after a change in fiscal policy. Decisions on investment, changes in production processes, the development of recycling infrastructure or the creation of jobs all require time.

The analysis is carried out in two steps. First, Table 6 presents fixed-effects models for country and year, in which the overall environmental tax burden is included not only with its current value, but also with two lags. This means that it is tested whether environmental taxes influence the circular indicators in the same year, after one year or after two years. The standard errors are clustered by country, and the dependent variables are the same as in the previous tables: CMUR, the recycling of municipal waste, the gross value added of the circular sectors, private investment and employment in the circular economy.

Owing to the inclusion of two lags, the period for some of the models is shorter. For CMUR and recycling, the analysis covers the period 2010–2024. For the gross value added, private investment and employment in the circular sectors, the period begins in 2007 and ends in 2023. The long-term effect in Table 6 is calculated as the sum of the three coefficients – the current effect, the effect after one year and the effect after two years.

The results from the lag specification provide partial support for the third hypothesis. This is seen most clearly for employment in the circular sectors. It does not respond in a statistically significant way either to the current value of environmental taxes or to the first lag. At the second lag, however, the effect becomes positive and weakly significant. The coefficient is 0.834 at p = 0.061. This means that an increase in the environmental tax burden is associated with higher employment in the circular sectors approximately two years later.

This result is logical from an economic point of view. The creation of new jobs in activities such as recycling, repair, reuse or waste processing cannot happen immediately. Investment decisions, organisational changes and the adaptation of firms are required. It is therefore possible that the fiscal signal manifests on employment with a lag.

Similar logic is observed for the investment channel. In the previous section, a negative relationship was found between energy taxes and private investment in the circular sectors. In the dynamic specification, this effect manifests precisely at the second lag, where the coefficient is -0.074 and is statistically significant at p = 0.049. This means that higher energy taxes can be associated with a contraction of circular investment after around two years, rather than immediately.

For the recycling of municipal waste, a more complex pattern emerges. The effect at the first lag is negative and statistically significant (-4.327, p < 0.001), whereas at the second lag it is positive and stronger (+8.324, p = 0.007). The cumulative long-term effect is positive and amounts to 2.686. This can be interpreted as a dynamic resembling a J-curve: initially, costs and adjustment difficulties appear, but subsequently a positive effect follows, connected to a change in behaviour, better organisation and the development of infrastructure.

For CMUR and the gross value added of the circular sectors, the lagged effects are not statistically significant. This means that for these two indicators no convincing evidence is found that the overall environmental tax burden exerts an influence either immediately or with a lag of one or two years.

The second part of the dynamic analysis is presented in Table 7. There, dynamic panel models are used, estimated through a two-step system GMM approach of Blundell–Bond with the Windmeijer correction of the standard errors. This method is suitable when the dependent variable depends strongly on its own previous values. In this case this is important, because the circular indicators usually change slowly and have strong inertia.

The models include the dependent variable with a lag of one year. This accounts for the fact that the current level of a given circular indicator depends on its level in the previous year. The tax variable is treated as predetermined, that is, it is allowed to be influenced by previous processes, but not by the current random deviations in the model. Year dummies are also included, as are the control variables.

In order to avoid the problem of instrument proliferation, the instrument matrices are collapsed, and the depth of the instruments is limited. GMM instruments are used for the lagged dependent variable with lags of the second and third year, as well as for the tax variable with lags of the first and second year. The control variables are instrumented by themselves. The number of instruments is between 23 and 27 and does not exceed the number of groups, which are 27 countries. This is an important condition for a more reliable estimation of the dynamic models.

The results from the system GMM show very strong inertia in the circular indicators. The coefficient of the lagged dependent variable is 0.944 for the gross value added of the circular sectors and 0.972 for employment. For recycling, the value is almost unity – 0.999. This means that the current values of these indicators are very strongly connected to their values in the previous year.

For CMUR, the coefficient even slightly exceeds unity and reaches 1.089. This indicates an extremely high persistence of the indicator over time and requires the results for this variable to be interpreted with particular caution. A further reason for caution is the AR(2) test, which for CMUR is borderline, with p = 0.062.

Once this strong inertia is accounted for, the current effect of the overall environmental tax burden disappears in most cases. The tax variable is not statistically significant for CMUR, gross value added, private investment and employment. This confirms the conclusion from Table 6 that the fiscal effects do not manifest immediately, but, if they exist, are more likely to unfold with a lag.

The only exception is the recycling of municipal waste. There the current coefficient of the overall environmental tax burden is negative and statistically significant (-3.344, p = 0.024). This result, however, should not be interpreted as evidence of a long-term negative effect. Given the almost complete inertia of recycling, it rather indicates a short-term difficulty in adjustment. In other words, initially the higher tax burden may create costs or organisational difficulties before any positive effects manifest.

The diagnostic tests in Table 7 broadly support the validity of the models used. The AR(1) and AR(2) tests are Arellano–Bond tests for serial correlation in the first differences. In dynamic panel models, the presence of AR(1) is expected, but not of AR(2). In most specifications, the AR(2) test does not indicate a problem of second-order serial correlation. For recycling, for example, the p-value of AR(2) is 0.716, and for gross value added it is 0.111. The Hansen test checks the validity of the overidentifying restrictions, that is, the extent to which the instruments used are appropriate. The p-values obtained do not indicate a rejection of the validity of the instruments, which supports the reliability of the specifications.

In addition, the positive relationship between R&D intensity and the gross value added of the circular sectors is retained in the dynamic specification as well. The coefficient is 0.028 at p = 0.020. This reinforces the conclusion from the baseline models that technological and innovation capacity matters for the economic development of the circular sectors.

In summary, the dynamic analysis does not change the main results from the static models, but refines them. The circular indicators are strongly inertial and change slowly. The overall environmental tax burden does not display a positive current effect on them. Where a fiscal effect is observed, it manifests above all with a lag of around two years. This is seen for employment in the circular sectors and for the investment channel.

Consequently, the third hypothesis receives partial support. The results show that it is important to take account of the temporal dimension of the fiscal effects. Environmental taxes do not automatically lead to immediate changes in the circular economy. The possible effects manifest gradually, after a period of adjustment. These findings prepare the next step of the analysis, which examines whether the agricultural profile of the economies alters the relationship between environmental taxation and the circular indicators.

In the tables, ***, ** and * denote the levels of statistical significance at 1%, 5% and 10% respectively. In Table 6, the R² (within) indicator reflects what proportion of the internal variation within countries is explained by the models after accounting for the fixed effects. Table 7 also presents the diagnostic tests for the dynamic models: AR(1), AR(2), the Hansen test, the number of instruments, the number of observations and the number of groups.

4.5. Heterogeneity According to the Agricultural Profile of the Economies

This section tests the fourth hypothesis of the study, namely whether the relationship between the environmental tax burden and the circular indicators depends on the agricultural profile of the individual economies. The logic of this test is that countries with a more strongly pronounced agricultural specialisation may respond differently to green fiscal instruments compared with the more urbanised and industrialised economies.

To this end, in Table 8 an interaction term between the overall environmental tax burden and the agricultural specialisation coefficient is added to the baseline model. Agricultural specialisation is measured through a location quotient relative to the average level for the EU-27. This coefficient is centred around the reference value of 1.415, defined in Section 3.3. This means that the main effect of the environmental tax burden in the table is interpreted at this reference level of the agricultural profile.

The models in Table 8 are estimated with fixed effects for country and year, with the standard errors clustered by country. The dependent variables are the same as in the previous tables: the circular material use rate, the recycling of municipal waste, the gross value added of the circular sectors, private investment in these sectors, and employment in the circular economy.

Table 9 presents the conditional marginal effects of the overall environmental tax burden at three different levels of agricultural specialisation. The first level is LQ = 0.75 and corresponds approximately to the average profile of the western and northern countries. The second is LQ = 1.42 and corresponds to the reference value used for centring. The third is LQ = 1.88 and corresponds to the average profile of the countries of Central and Eastern Europe. The marginal effects are calculated by the delta method on the basis of the estimates from Table 8.

The results show that the agricultural profile matters for some of the relationships between environmental taxation and the circular indicators. The clearest support for the fourth hypothesis is observed for the recycling of municipal waste and for employment in the circular sectors.

For recycling, the interaction between the environmental tax burden and agricultural specialisation is negative and highly statistically significant. The coefficient is -5.705 at p < 0.001. This means that as agricultural specialisation increases, the positive effect of the environmental tax burden on recycling weakens.

This result is seen more clearly in Table 9. At a weakly agricultural profile, characteristic of the western and northern countries, the marginal effect of the tax burden on recycling is positive and statistically significant. At LQ = 0.75 it is 4.440 at p = 0.024. This means that in the more weakly agricultural economies a higher environmental tax burden is associated with a higher recycling rate of municipal waste.

At the reference level of the agricultural profile, the effect is no longer statistically significant. At LQ = 1.42 the coefficient is 0.647, but does not differ substantially from zero. At a higher agricultural profile, characteristic of the countries of Central and Eastern Europe, the sign of the effect even becomes negative. At LQ = 1.88 the coefficient is -2.007, but is likewise not statistically significant. The positive relationship between environmental taxation and recycling is therefore observed mainly in the more weakly agricultural economies.

A similar dependence is found for employment in the circular sectors. The interaction between the tax burden and agricultural specialisation is negative and statistically significant. The coefficient is -0.896 at p = 0.017. This means that the effect of environmental taxation on circular employment also weakens at a more strongly pronounced agricultural profile.

At LQ = 0.75, the marginal effect on employment is positive and weakly significant. The coefficient is 1.067 at p = 0.051. This shows that in the western and northern economies a higher environmental tax burden may be associated with higher employment in the circular sectors. At the reference level, the effect decreases and is no longer significant, and at a high agricultural profile it practically disappears. At LQ = 1.88 the coefficient is 0.054 and is not statistically significant.

For private investment in the circular sectors, the interaction in itself is not statistically significant. Nevertheless, the conditional marginal effects reveal an important feature. The negative relationship between the environmental tax burden and circular investment is statistically significant precisely at a high agricultural profile. At LQ = 1.88 the coefficient is -0.085 at p = 0.038. This means that in the more agricultural economies a higher environmental tax burden may be associated with a contraction of private investment in the circular sectors.

For CMUR and the gross value added of the circular sectors, no significant moderation by the agricultural profile is found. This means that for these two indicators there is no convincing evidence that the effect of environmental taxation differs substantially between more agricultural and more weakly agricultural economies.

The robustness of the results was additionally checked for the two more important channels established in the previous analysis. First, the negative effect of energy taxes on private investment in the circular sectors does not change substantially according to the agricultural profile. The interaction is not statistically significant, with a p-value of 0.817. This means that the negative relationship between energy taxes and circular investment appears to be structural rather than specific only to the agricultural economies.

Second, the positive relationship between transport taxes and recycling is concentrated mainly in the more weakly agricultural economies. At LQ = 0.75 the effect is positive and statistically significant, with a coefficient of 17.61 at p = 0.004. At LQ = 1.88 the effect decreases to 7.68 and is no longer statistically significant. The interaction is borderline significant, with p = 0.104. This result is consistent with the more general conclusion that green fiscal instruments act more weakly in economies with a more pronounced agricultural profile.

The economic interpretation of these results can be explained in two directions. On the one hand, in the more agricultural economies some of the circular practices do not pass through the formal sectors that official statistics measure. For example, the valorisation of biomass, composting, on-farm reuse and the internal use of residual resources often take place at the level of the farm or the local community. For this reason, they are not always clearly reflected in the indicators for the formal circular sectors or in the statistics on municipal waste.

On the other hand, in rural and more weakly urbanised areas there is often more limited infrastructure for the collection, sorting, processing and recycling of waste. This reduces the ability of households and firms to respond to the price incentives created by environmental taxes. Even when the fiscal signal exists, it may not lead to the expected result if there are no practical opportunities to change behaviour.

The results thus show that identical green fiscal instruments can have a different effect in different types of economy. In the more urbanised and more weakly agricultural countries, they are more readily associated with higher recycling and employment in the circular sectors. In the more agricultural economies, however, this effect weakens or disappears, and for investment it may even become negative.

This conclusion has important significance for policy. If environmental taxes are applied without accompanying investment in infrastructure, they may deepen the disparities between the more urbanised and the more agricultural countries. In order genuinely to support the circular transition, green fiscal instruments must be combined with investment in circular infrastructure in rural areas, including systems for the collection and processing of waste, composting, the valorisation of biomass, and local reuse chains.

Consequently, the fourth hypothesis receives partial but substantial support. The agricultural profile of the economy alters the effect of environmental taxation above all for recycling and employment in the circular sectors. This shows that the circular transition cannot be regarded solely as a pan-European process. It also has a clearly pronounced territorial dimension, connected to the differences between agricultural, industrial and urbanised economies.

This heterogeneity is directly connected to the broader question of the circular economy in rural areas. It shows that circular transition policies must be adapted to the structure of the economy. In the more agricultural countries, the fiscal instruments need to be complemented by targeted measures for the development of a circular bioeconomy, agri-environmental practices, local infrastructure and better statistical recording of circular practices outside the formal sectors.

In the tables, ***, ** and * denote the levels of statistical significance at 1%, 5% and 10% respectively. The R² (within) indicator shows what proportion of the internal variation within countries is explained by the models after accounting for the fixed effects. In Table 8, the R² (within) values range from 0.031 for private investment to 0.206 for the gross value added of the circular sectors. This shows that the models best explain the dynamics of the economic contribution of the circular sectors, but capture the variation in investment more weakly.

5. Discussion

The results obtained show that the relationship between environmental taxation and the circular economy cannot be explained solely through the overall size of environmental taxes. What is more important is how the tax system is structured, at which activities it is directed, when its effect manifests, and in what type of economy it operates. In this sense, the main conclusion of the study is that what matters for the circular transition is not simply how high the green fiscal burden is, but precisely what is taxed, with what lag the effect manifests, and whether the economy has a more agricultural or a more urbanised profile. Such a conclusion is in keeping with the broader circular economy literature, which regards the transition not as a single measure but as a systemic change in the way materials are used, retained and returned to the economy (Geissdoerfer et al., 2017; Kirchherr et al., 2017; Bocken et al., 2016).

This overall picture explains why the first hypothesis is not confirmed in its aggregate formulation. The overall environmental tax burden in itself does not display a statistically significant relationship with circular outcomes. This does not mean, however, that environmental taxes are irrelevant. Rather, it means that the overall indicator is too broad and conceals the differences between the individual types of tax. It is precisely when the tax burden is decomposed by category that it becomes apparent that the effects are different and, in places, opposing. This provides partial support for the second hypothesis. The dynamic models additionally show that some of the effects manifest with a lag, which supports the third hypothesis, above all for employment and the investment channel. Finally, the interaction models confirm that the agricultural profile of the economy matters for the strength and direction of the effect, which provides support for the fourth hypothesis.

The null result for the overall environmental tax burden is important, because it directs attention to the difference between fiscal revenue and a genuine incentive for change. In the literature on environmental taxation, this distinction is well known through the debate on the double dividend – whether environmental taxes serve primarily to raise revenue, or whether they can simultaneously improve environmental outcomes and economic efficiency (Pearce & Turner, 1990; Bovenberg & de Mooij, 1994; Goulder, 1995; Freire-González, 2018). In the case of the EU, a large proportion of environmental taxes is linked to energy and fuels. These matter for climate policy, but are not directed directly at material flows, reuse or recycling. It is therefore logical that the overall tax burden does not display a clear relationship with the circular economy indicators.

A further explanation is that the overall share of environmental taxes in GDP may be declining not because environmental policy is weakening, but because the tax base itself is changing. Electrification, higher energy efficiency and changes in fuel consumption can reduce the revenues from energy taxes. Thus one and the same indicator mixes two different processes: on the one hand, the political price signal, and on the other, the mechanical contraction of the tax base. For this reason, the overall environmental tax burden is not a sufficiently good indicator of whether fiscal policy supports the circular transition. For future research and for policy monitoring, it is more appropriate to work with the individual categories of environmental taxes, rather than only with their overall share in GDP (European Environment Agency, 2016, 2022; Eurostat, 2024).

The clearest evidence of the significance of the tax structure is the negative effect of energy taxes on private investment in the circular sectors. This result points to a possible tension between climate and circular policy. Energy taxes are an important instrument for restricting the consumption of fossil fuels and for reducing emissions. At the same time, some of the activities connected to the circular economy – for example recycling, the processing of secondary raw materials, and repair – also require energy. When energy becomes more expensive through taxation, the costs of these activities can increase. Thus an instrument that is directed at climate objectives may indirectly hinder investment in circular sectors. This result complements the understanding that the circular economy is not reduced solely to recycling, but involves a comprehensive restructuring of production, repair, resource and logistics processes (Geissdoerfer et al., 2017; McCarthy et al., 2018).

This conclusion is not an argument against energy taxation. Rather, it shows that green fiscal policy must be designed more precisely. If energy taxes are increased without there being compensatory mechanisms for the circular industries, an undesirable side effect may result. Such mechanisms could be investment incentives, the targeted use of part of the revenues, support for energy efficiency in recycling firms, or other instruments to prevent the contraction of circular investment. The fact that the negative effect manifests with a lag suggests that firms do not respond immediately, but revise their investment decisions once the higher energy costs have already become part of their cost plans. This is also consistent with the more general logic of the Porter hypothesis, according to which regulations and fiscal incentives can encourage innovation, but only when they are sufficiently well targeted and predictable for business (Porter & van der Linde, 1995).

Unlike energy taxes, pollution taxes display a positive relationship with the gross value added of the circular sectors. This result is logical, because these taxes are more directly connected to waste and polluting flows. When pollution itself is taxed, the incentive to limit waste, to process and to reuse is clearer. The positive relationship with the value added of the circular sectors therefore supports the idea that fiscal instruments work better when the tax base is close to the environmental problem that the policy seeks to influence. This is consistent with the understanding of environmental taxes as a market instrument that can correct externalities when it is correctly directed at the source of the environmental pressure (Pearce & Turner, 1990; OECD, 2017; European Environment Agency, 2016).

The results for resource taxes are different. They do not display a statistically significant effect in any of the specifications. This should not, however, be interpreted as evidence that resource taxation is ineffective. The more likely explanation is that resource taxes in the EU are used far too little. At an average level of around 0.04% of GDP, with zero values in some member states and limited change over time, the panel data do not contain sufficient variation to estimate their effect reliably. In this sense, the result is more of a diagnosis of the absence of resource taxation on a significant scale than a refutation of its potential. Such an interpretation is consistent with the European analyses, according to which environmental taxation remains heavily concentrated on energy, while pollution and resource taxes have a limited fiscal share (European Environment Agency, 2022; Eurostat, 2024).

This has important policy significance. If the EU genuinely wishes to encourage a more efficient use of materials, resource taxes ought to occupy a more visible place in the green fiscal system. The reasons why this is not happening are understandable. The costs of such taxes are concentrated on specific industries, such as extraction, construction and resource-intensive manufacturing, whereas the benefits are more widely distributed. In addition, individual member states may be concerned about a loss of competitiveness or about tax competition. For this reason, a coordinated European approach appears more appropriate than separate national solutions. In this context, the discussion on the own resources of the EU budget and on gradually shifting the tax burden from labour to resources offers a natural framework for the future development of this instrument (European Environment Agency, 2016, 2022; OECD, 2017).

The dynamic results complement this picture. They show that the circular indicators are strongly inertial and change slowly. This is to be expected, because the circular transition requires investment, changes in infrastructure, new organisational practices and the adaptation of firms and households. The absence of an immediate effect from environmental taxes therefore does not necessarily mean the absence of an effect altogether. Where a response is observed, it manifests with a lag of around two years. This is particularly important for employment and investment, because hiring and investment decisions are usually taken after a period of adjustment. Similar logic is found in studies on the employment and macroeconomic effects of the circular transition, where it is emphasised that changes in the labour market and the sectoral structure unfold gradually (Chateau & Mavroeidi, 2020; McCarthy et al., 2018).

The agricultural profile of the economies adds another important dimension to the analysis. The results show that the effects of environmental taxation are not identical across all countries. In the more weakly agricultural and more urbanised economies, the positive relationship between the tax burden, recycling and circular employment is more visible. In the more agricultural economies, this effect weakens or disappears, and for investment a negative relationship is observed. This means that one and the same fiscal instruments can yield different results depending on the structure of the economy. This conclusion is significant, because the circular economy in rural areas and in agriculture often has a different logic from that in industrial and urban systems (OECD, 2020b; Velasco-Muñoz et al., 2021).

This agricultural gradient can be explained in two ways. The first explanation is related to measurement. In agricultural economies, some of the actual circular practices – the valorisation of biomass, composting, the reuse of resources on farms – take place outside the formal circular sectors and outside the municipal waste flows that official statistics measure. As a result, part of the actual circularity remains poorly visible in the indicators used. The second explanation is structural. In rural areas, the infrastructure for the collection, sorting, processing and recycling of waste is usually more limited and more expensive to maintain. This reduces the possibility for households and firms to respond to the price incentives, even when these are correctly set. Such constraints are also considered in the literature on the circular economy in rural areas, where the significance of local infrastructure, the bioeconomy and territorial differences is emphasised (OECD, 2020b; Donner et al., 2020; Velasco-Muñoz et al., 2021).

In both cases, the policy conclusion is the same: uniform green fiscal instruments are not sufficient. If they are applied without accompanying investment and without account being taken of the territorial differences, they may deepen the disparities between the more urbanised and the more agricultural economies. Environmental revenues must therefore be linked more clearly to investment in circular infrastructure in rural areas – for example systems for the collection and processing of waste, composting, the valorisation of biomass, and local reuse chains. Better recognition and measurement of informal agricultural circularity is also needed, including through bioeconomy indicators. The link with rural tourism may also be important, because the demand for local, short and more circular chains creates a market premium that a fiscal incentive alone can hardly provide (OECD, 2020b; Velasco-Muñoz et al., 2021).

More broadly, the results show that the circular transition cannot rely on a single type of instrument alone. Environmental taxes matter, but their effectiveness depends on their structure, on the way the revenues are used, and on the available infrastructure. A similar conclusion stands out in the literature on market instruments, including green bonds. These can assist the financing of environmental projects, but in themselves are not sufficient for a comprehensive transformation of the economy (Flammer, 2021; Fatica & Panzica, 2021; Tang & Zhang, 2020). The present study leads to a similar conclusion regarding the fiscal side: in their current structure, environmental taxes act slowly, unevenly and sometimes in an undesirable direction.

Consequently, the more convincing approach is a policy mix. The tax structure, the targeted use of revenues and infrastructural investment must be designed together. Moreover, this mix must be territorially sensitive. The more urbanised economies can respond more quickly to price incentives, because they already possess more developed infrastructure and formal circular sectors. The more agricultural economies need not only a tax signal, but also investment, institutional and statistical support. Only with such a combination can green fiscal instruments become a genuine factor in accelerating the circular transition, rather than remaining a general fiscal indicator with a limited connection to material circularity.

6. Conclusions

The article investigates the relationship between the structure of environmental taxation and the circular transition in the 27 member states of the European Union over the period 2010–2024. Circular outcomes are measured through five indicators from the EU's official monitoring framework. The results obtained make it possible to formulate four main conclusions. First, the overall environmental tax burden is not a sufficiently informative indicator of the relationship between fiscal policy and the circular economy. It does not display a systematic relationship with circular outcomes, since it is dominated by energy taxes and by the changing tax base. Second, the structure of taxation matters more than its overall size. Energy taxes are associated with lower private investment in the circular sectors, whereas pollution taxes display a positive relationship with their value added. Resource taxes remain too limited as a fiscal share for their effect to be reliably estimated. Third, the fiscal effects do not manifest immediately. Where they are observed, they unfold with a lag of around two years, and the circular indicators themselves are characterised by strong inertia. Fourth, the effects are territorially uneven. The positive relationships are more visible in the weakly agricultural economies and weaken as the agricultural profile increases, while the negative investment effect is significant precisely in the more agricultural member states.

Several policy recommendations follow from these conclusions. In the first place, a change in the structure of green taxation is needed. Greater emphasis should be placed on pollution and resource taxes, because they are more directly connected to material flows and to the logic of the circular economy. This reorientation would be more effective if it were coordinated at EU level, since individual member states may refrain from more substantial resource taxation owing to concerns about tax competition. In the second place, the design of environmental taxes must be more precise. If energy taxes increase the costs of energy-intensive circular activities, such as recycling, processing and repair, compensatory mechanisms need to be provided for. These could be investment incentives, differentiated rates, or the targeted use of part of the revenues. In the third place, green fiscal policy must take account of territorial differences. Part of the environmental revenues should be directed towards circular infrastructure in rural areas, and informal agricultural circular practices must be better recognised and measured. For rural areas, the combination of fiscal incentives with the demand created by rural tourism and short supply chains appears particularly promising.

The results must be interpreted within several limitations. The study uses aggregated national data, which do not make it possible to capture the differences between rural and urban territories within individual countries. And it is precisely there that the effects of green fiscal instruments probably differ most strongly. In addition, the estimates show within-country associations after fixed effects have been accounted for, but do not represent full causal identification. Tax changes are not random and can often be part of broader policy packages, which is particularly important in interpreting the results for transport taxes. Another limitation is that the official circular economy indicators measure mainly the formal sectors and municipal waste flows. For this reason, part of the informal circularity, especially in agriculture, remains outside the scope of the analysis. The agricultural specialisation coefficient is also an indirect measure of the rural character of the economy. Finally, the high persistence of CMUR requires caution in interpreting the dynamic estimates, and the limited variation in resource taxes does not allow a reliable estimation of their effect.

These limitations also outline possible directions for future research. One natural next step is analysis at the regional NUTS 2 level, where the differences in circular infrastructure, the agricultural profile and tourist pressure can be traced more precisely. Another direction is the use of microdata from agricultural holdings, which would make it possible to investigate how fiscal incentives reach the individual producer and whether they actually change their behaviour. The inclusion of bioeconomy indicators would also bring the analysis closer to the actual circularity in rural areas. The interaction between green fiscal instruments and market mechanisms for green financing also deserves additional attention. The results obtained show that the circular transition can hardly be accelerated through a single instrument. It is more likely that progress will depend on a well-composed policy mix that combines tax structure, the targeted use of revenues, infrastructural investment and territorial differentiation.