Green Growth in Agriculture: Long-Term Evidence from European Union Countries

Abstract

In agriculture, the relationship between economic growth and environmental pressures is complex and difficult to measure and compare between countries. This study had two objectives; the first was to build a new green growth accounting framework for agriculture in relation to natural capital and air pollution, and the second was to assess the long-term green growth of agriculture in EU countries. The data for EU27 from 2005 to 2021 were collected and used in the empirical analysis of green growth in agriculture. The findings showed positive real growth in agriculture from both the economic growth and green growth perspectives in most EU countries in the long term. Slow changes in air pollution (expressed in net GHG emissions from agriculture) and in natural capital (expressed in quality-adjusted agricultural land) did not have a significant impact on green growth in agriculture. The empirical analysis also revealed that most EU countries increasingly rely on technological progress to promote agricultural growth, and half of them rely on investments in produced capital. Labour input only made a positive contribution to agricultural growth in Ireland and Malta. This study will significantly contribute to improving the measure of green growth in agriculture, and the results of the empirical analysis will be used by policymakers and economists.

1. Introduction

The relationship between economic growth and environmental pressure in agriculture is complex [1]. Agriculture plays an important role in ensuring food and nutritional security and serves as a major source of rural employment [2]. However, agriculture is also a significant contributor to climate change due to the amount of pollutants emitted from livestock, managed agricultural soils, land use, and land use change. The dilemma between environmental pressure and economic growth obtained widespread attention at the national and global levels. The relevant question is how to maintain economic growth while improving the quality of the environment [3,4]. To achieve this, the concept of green growth has been developed. Green growth is a complex concept that emphasises the quality of development.

Over the last decade, green growth has become one of the most important issues on the political agenda. More and more discussions on the green growth policy are arising [5,6]. At the Rio+20 Conference on Sustainable Development in 2012, green growth emerged as a fundamental theme, and “green economy” and “sustained economic growth” may ensure success in the new field of sustainable development [7]. Green growth in the European Union (EU) is also emerging in various national policy measures. This is especially true for agriculture as a polluting sector [8]. To promote green growth, the EU has created the “European Green Deal” communication that promotes achieving net zero emissions of greenhouse gases (GHGs) by 2050 and decoupling economic growth from resource use [9].

More and more scholars are engaged in the green growth discourse or empirical research [10]. There are several concepts similar to green growth [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15] like environmentally sustainable economic growth [16,17,18,19], ecologically sustainable growth [20,21], inclusive green growth [22,23,24], genuine green growth [25,26], the green economy [15], low-carbon growth [27,28,29], and zero carbon growth [30]. In the academic literature, the above-mentioned terms are used as synonyms, while others are used as separate terms.

Previous studies have focused on the development of the concept of green growth in agriculture [5,11,31,32,33,34] and measuring green growth using such systems of indicators as dashboards and headline indicators [19,23,34,35]. However, according to Kasztelan et al. [36], the agricultural evaluation of green growth is not well understood due to the complexity of the sector and changes in its negative or positive environmental impact. When performing green growth measurement in agriculture, two approaches can be distinguished. These include a framework for measuring green growth by including its key elements, like production, environment, consumption, and its corresponding indicators, and the development of measures in terms of productivity, like environmentally adjusted or green multifactor productivity (EAMFP) [37,38] or environmentally adjusted or green total factor productivity (EATFP) [39,40,41,42,43], as the most commonly used indicators in recent years. Two methods, data envelopment analysis (DEA) and stochastic frontier analysis (SFA), have become the most popular techniques in measuring EATFP [39,41,42,43] and EAMFP [38] in agriculture. This study had two objectives; the first was to build a new green growth accounting framework for agriculture in relation to natural capital and air pollution, and the second was to assess the long-term green growth of agriculture in EU countries.

2. Methodology

2.1. Conceptual Framework of Green Growth Accounting for Agriculture

This paper developed a green growth accounting framework for agriculture using the functional form of the relationship between production factor inputs and production outputs (i.e., the transformation function), taking into consideration labour, produced capital, and natural capital as the inputs, gross value added as the desirable output, and air pollution as the undesirable output. The OECD defines green growth as “fostering economic growth and development while ensuring that the natural assets continue to provide the resources and environmental services on which our well-being relies” [44] (p. 9), which means “aligning economic growth and environmental objectives” [45] (p. 22).

The starting point in our approach to building the green growth accounting framework for agriculture is performing environmentally adjusted multifactor productivity (EAMFP) measurement at the macroeconomic (country) level, as established by Brandt et al. [46,47] and developed by Rodríguez et al. [13,48]. EAMFP extends the traditional MFP measurement framework, which is based on conventional production factors (labour and produced capital) and outputs (gross output or value added) to take account of environmentally related inputs and outputs, such as natural capital and air pollution, respectively. EAMFP can reveal natural contributions and assess the potential for sustainable growth [13]. It is used as the OECD green growth headline indicator [34].

EAMFP is derived from the following transformation function [13]: H (Y, R, L, K, N, t) ≥ 1, where Y represents the desirable output of the economy and R is the undesirable output (air pollution); L, K, and N represent vectors of multiple types of inputs, such as labour, produced capital, and natural capital, respectively; H represents the increase in each input mentioned above and in both desirable and undesirable outputs. The logarithmic derivative of the transformation function H concerning time provides a measure of growth in environmentally adjusted multifactor productivity (EAMFP), which is equal to the sum of the weighted output change minus the sum of the weighted input change over the same time.

In the case of the agricultural industry, the logarithmic derivative of the transformation function H with respect to time t can be expressed as follows:

∂lnEAMFP/∂t = ∂lnGVA/∂t − ε_YGHGnet × (∂lnGHG_net/∂t) − ε_YL × (∂lnL/∂t) − ε_YK × (∂lnK/∂t) − ε_YN × (∂lnN/∂t);

(1)

where ∂lnEAMFP/∂t denotes the real growth of the environmentally adjusted multifactor productivity, which can be interpreted as technological progress; ∂lnGVA represents the real growth of gross value added as a desirable output in agriculture; ∂lnGHGnet represents the change in net GHG emissions in agriculture as an undesirable output; ∂lnL denotes the change in the agricultural labour force in terms of annual work units; ∂lnK denotes the real change in produced capital in terms of agricultural capital stock; and ∂lnN denotes the change in natural capital in terms of quality-adjusted agricultural land (QAAL). Outputs and inputs are weighted with their elasticity concerning the transformation function: εYGHGnet denotes the elasticity of undesirable output, and εYL, εYK, and εYN denote elasticities of labour, produced capital, and natural capital, respectively. The methodology to obtain elasticities is described in detail in Section 2.4 and Section 2.5 below.

Rearranging Equation (1) provides a green growth accounting equation that decomposes pollution-adjusted GVA growth into growth in factor inputs (such as labour, produced capital and natural capital) and EAMFP growth, as follows:

∂lnGVA/∂t − ε_YGHGnet × (∂lnGHG_net/∂t) = ε_YL × (∂lnL/∂t) + ε_YK × (∂lnK/∂t) + ε_YN × (∂lnN/∂t) + ∂lnEAMFP/∂t

(2)

The left side of Equation (2), i.e., ∂lnGVA/∂t − ε_YGHGnet × (∂lnGHG_net/∂t), refers to pollution-adjusted GVA growth, which can measure the green growth in agriculture and could be useful in the analysis of sustainable development of the agricultural industry. This indicator provides valuable information on the economic growth of the agricultural industry, taking into consideration the contribution of natural capital and the negative impact of pollutants.

2.2. Output Variables and Data

Joint production in agriculture produces both desirable and undesirable outputs, also known as good and bad outputs. Desirable or good output is defined as the preferred or planned product of the production process, usually expressed as gross domestic product [13,49,50]. In the case of the agricultural industry, the desirable output can be measured by gross agricultural output [51,52,53], value added [54], or specific output in a narrow analysis like crop output [2]. In this study, gross value added (GVA) was used as a metric of desirable agricultural output. The GVA indicator in the Economic Accounts for Agriculture (EAA) is measured as the difference between the value of output and the value of intermediate consumption. It reflects the value generated by producing goods and services in the agricultural industry. The EAA is the official harmonised source of agricultural industry information in the EU [55]. The data on GVA at real prices in the agricultural industry by EU countries were obtained from the Eurostat database under the following codes: online data code [aact_eaa04] and indicator code [20000].

According to Rodríguez et al. [48], undesirable or bad output is defined as negative by-products of the production process, such as air pollution and water waste. In the case of agriculture, the undesirable output is environmental pollution from agricultural production, such as the greenhouse gas (GHG) emissions from livestock breeding and managed agricultural soils, or it comes from sources that do not have a single point of origin [13,37,52,56,57].

According to Lewis [58], non-point source pollution is a diffuse source of water quality degradation that is difficult to measure. Non-point source pollution includes urban runoff, storm sewers, drainage from waste disposal sites and landfills, and airborne pollutants that settle in the water. In the case of agriculture, non-point source pollution refers to emissions into water, like total nitrogen and total phosphorus emissions, nutrient runoff from agricultural production, and rainwater carrying agricultural sediments like soil particles from agricultural fields into nearby lakes or streams [52,59]. The coverage of non-point source pollution is constrained by data limitations. The agricultural pollution data used in this analysis contained clearly identified variables (Figure 1).

In terms of GHGs, net agricultural pollution is affected by both GHG emissions and removal [56,60,61,62,63,64]. We considered net GHG emissions in agriculture (GHG_net) as the balance between the GHG emissions from production (cultivation of crops and livestock) and the carbon sequestration in grassland and pastures. GHG_net can be described as follows [65]:

GHG_net = GHG + (±CO₂)

(3)

where GHG denotes greenhouse gas emissions in CO₂ equivalent and ±CO₂ denotes positive or negative carbon dioxide per annum in terms of emissions/removal from agriculture.

In this study, environmental pollution from agriculture is expressed in terms of net GHG emissions, taking into consideration both the GHG emissions in CO₂ equivalent from agricultural activities and the CO₂ uptake by grassland and pastures. European Environment Agency (EEA) data on agricultural GHG emissions (including carbon sequestration) in CO₂ equivalent were used for the empirical analysis. These data are available in the Eurostat database under the following codes: online data code [env_air_gge] and codes of the source sectors for GHG emissions such as Agriculture [CRF3], Cropland [CRF4B], Grassland [CRF4C], and Wetlands [CRF4D].

2.3. Input Variables and Data

Produced (or physical) capital, labour, and land are fundamental production factors in agriculture (Figure 2).

2.3.1. Labour

According to Fuglie [66], farm labour input is the total number of adults (salaried and non-salaried) who are economically active in agriculture. In this study, the labour input was measured in terms of annual work units (AWUs). In the EU, the EAA provides detailed information on the AWUs in the agricultural industry. The number of AWUs is estimated by dividing the number of hours actually worked per year by the annual number of hours corresponding to full-time jobs in the country [55]. For persons working less than full-time on farms, the agricultural labour input in AWU terms is calculated as the quotient of the number of hours actually worked per week or year and the number of hours typically worked in a full-time job during the same work period [67]. Both the salaried and non-salaried labour force inputs were used as agricultural labour input in this study. The data on total labour input in the agricultural industry by EU countries were obtained from the Eurostat database under EAA’s agricultural labour input statistics (online dataset code [aact_ali01] and indicator code [40000]).

2.3.2. Produced Capital

Produced (physical) capital is a fundamental factor in the agricultural production process and is used as the input variable in terms of capital stock in EAMFP analysis. Despite the recent growing interest in the estimation of agricultural capital stock, agricultural capital stock datasets at the national level comparable between countries are still scarce [68]. This is a major obstacle to empirical research on the contribution of capital to the growth of the agricultural industry, as well as economic development more broadly [69]. This is also true for EU countries, as capital stock data for the agricultural industry are not included in the EAA dataset.

Currently, FAOSTAT’s data on capital stocks aggregated at the level of the wider primary production industry of biological raw materials, including agriculture, forestry, and fisheries [70], help to partially overcome this data limitation. However, agriculture’s contribution to the primary production industry varies greatly across EU countries and changes over time (as shown by structural data on labour input and GVA output at the beginning and end of the considered study period in Figure A1 and Figure A2 in Appendix A). For instance, in 2021, employment in agriculture ranged from 55% of all primary production jobs in Sweden to 97% in the Netherlands and Poland, while the total contribution of agriculture to the total GVA of the primary production industry ranged from 31% in Finland to 99.9% in Luxembourg. Considering these findings, the estimates of agricultural capital stocks for the EU countries in this study were calculated using a variation of the perpetual inventory method (PIM) applied in the FAO methodological and computational concepts of the FAO global database on agriculture investment and capital stock [69].

According to Vander Donckt and Chan [69], the capital stock equation is expressed as follows:

K_i,t = (1 − δ_i) × K_i,t−1 + (1 − δ_i) × GFCF_i,t

(4)

where K_i,t denotes agricultural net capital stock for country i at the end of period t, K_i,t−1 denotes initial agricultural capital stock at the end of the previous period t − 1, δ_i is a country-specific depreciation rate, and GFCF_i,t denotes gross fixed capital formation for country i at period t. The estimates of the initial net capital stock in agriculture for each EU country in period t − 1 (i.e., before the start of the time series of gross fixed capital formation data in 2005–2021) were obtained by multiplying the FAOAST aggregated data on net capital stocks in agriculture, forestry and fisheries in 2004 by the coefficient of agricultural employment share in the same year, as shown in Figure A2 of Appendix A. The data on gross fixed capital formation at real prices in agriculture by EU countries were obtained from the Eurostat EAA’s database under the following codes: online data code [aact_eaa04] and indicator code [34000].

Given the differences in the composition of agricultural fixed assets by type across countries [68] and its variability over time, this study used a depreciation rate that varies by country and over time to construct a longitudinal series of estimates of agricultural capital stocks for EU countries. For each EU country, the implicit depreciation rate was calculated as the ratio of the depreciation amount to the value of the total fixed assets, using standard FADN indicators such as Total fixed assets at closing valuation [SE437] for the previous year t−1 and Depreciation [SE360] for the current year t, and annual average farms data series from the FADN Public Database of the European Commission. The depreciation rate is expressed as a 5-year moving average to eliminate random annual fluctuations in the [SE437] and [SE360] data.

2.3.3. Natural Capital

According to Ascui and Cojoianu [71] and Barbier [72], natural capital is defined as natural resources used in the economy, such as land, forest, and fossil fuels. Natural capital also consists of ecosystems that provide flows of environmental goods and services, which underpin the global economy and human well-being.

In this study, natural capital in the form of quality-adjusted agricultural land [66] was applied as the input (N) variable in the EAMFP model (Equations (1) and (2)) to measure green growth in agriculture, taking into consideration the differences in the productive capacity of land among its types (i.e., irrigated and rainfed cropland and permanent pastures) and across countries or regions. Agricultural land is a heterogeneous input, with some cropland capable of multiple harvests per year, while some permanent pastures yield very low amounts overall [73]. In the quality-adjusted agricultural land approach, various land types are converted into “rainfed cropland equivalents” according to their relative productivity using different quality weights; these are then aggregated into a quality-adjusted agricultural land indicator with quality weights varying by land type and region [66,73].

The amount of quality-adjusted agricultural land (QAAL) is calculated as follows [73]:

QAAL_i,t = α_i × Crain_i,t + ρ_i × Cirrg_i,t + β_i × P_i,t

(5)

where Crain denotes a rainfed cropland area in time t for country i, Cirrg denotes irrigated cropland area, P denotes permanent pasture area, and α, β, and ρ represent quality weights by land type for country i, as listed in

Table 1. Land quality weights by land type used to measure quality-adjusted agricultural land in EU countries.

EU Countries	Permanent Pasture (β)	Rainfed Cropland (α)	Irrigated Cropland (ρ)
Bulgaria, Czechia, Hungary, Poland, Romania, Slovakia	0.094	1.000	1.570
Estonia, Latvia, Lithuania, Finland, Sweden, Ireland, Denmark	0.094	1.000	1.001
Slovenia, Croatia, Greece, Italy, Malta, Portugal, Spain, Cyprus	0.094	1.000	1.972
Luxemburg, Belgium, France, Austria, Germany, Netherlands	0.094	1.000	1.279

Source: authors’ own elaboration for EU countries based on USDA information [73] for four European regions and Cheba and Bąk [74].

. The rainfed cropland area is the difference between the total cropland area and the irrigated cropland area (Crain = Ctotal − Crrig). Compared to the usual metric of agricultural land in physical hectares, the aggregate QAAL indicator better reflects how changes in agricultural land use (i.e., changes in natural capital, as in this study N = QAAL) contribute to output growth. It also indirectly reflects other components of natural capital, such as water used for crop irrigation and multiple ecosystem functions of permanent pasture areas. Permanent meadows and pastures provide habitat for a high diversity of plant and animal species, prevent soil erosion, and improve soil structure and fertility and carbon capture and storage.

Agricultural land data for EU countries collected from the FAOSTAT Land, Inputs, and Sustainability database were used for empirical analysis. The equivalent area of quality-adjusted agricultural land by type was calculated using data obtained from the FAOSTAT Land Use statistics based on the following aggregated items: cropland, cropland area actually irrigated, and permanent meadows and pastures.

2.4. Elasticities with Respect to Outputs

The EAMFP measurement (Equation (1)) requires information on the elasticity with respect to undesirable outputs and their price [46]. According to Rodríguez et al. [13], the elasticity with respect to air pollution as the undesirable output is defined as the change in output associated with an increase in pollution when the input use remains constant. Net greenhouse gas emissions from agriculture as the air pollutant output are included in the green growth accounting framework for agriculture.

The elasticities with respect to inputs, which are traded on markets and have clear prices, can be determined using the profit maximisation method used in the green growth accounting framework for agriculture. The calculation of the monetary value of net GHG emissions is based on carbon emission prices obtained from the EU Emissions Trading System (EU ETS). Elasticity with respect to net GHG emission can be expressed as follows [13]:

ε_YGHGnet = GHG_net/(GVA + GHG_net)

(6)

where GHG_net denotes net GHG emissions in agriculture (in monetary value), and GVA denotes agricultural gross value added.

2.5. Elasticities with Respect to Inputs

EAMFP growth accounting requires information on the elasticity with respect to inputs of labour, produced capital, and natural capital expressed in quality-adjusted agricultural land area, as given in Equation (1). The elasticities with respect to market-traded input can be computed using a profit-maximisation approach usually used in traditional multifactor productivity analyses. Under this approach, the elasticities of the transformation function are equal to the shares of labour, produced capital, and natural capital in this input mix [13,48]. The main specificity of the profit-maximisation approach in the case of the EAMFP accounting framework for agriculture is that the calculation of elasticities must rely not only on the implicit prices of purchased labour, produced capital, and agricultural land but also on the opportunity costs of the same factors of production owned by the farm. Opportunity costs (otherwise known as implicit costs) of own production factors refer to the income or other benefits that the farm could obtain from the next best use of these factors instead of using them in agricultural production on the farm.

The elasticity measurement using both explicit and implicit costs of labour, produced capital, and natural capital (expressed in terms of the quality-adjusted agricultural land) can be calculated as follows:

ε_YL = L/γ, ε_YK = (D + C + I)/γ, and ε_YN = N/γ

(7)

where L denotes the costs of paid and unpaid labour input, D denotes deprecation of fixed capital, C denotes intermediate consumption of working capital, I denotes the costs of rented and owned capital, N denotes the costs of leased and owned land, and γ denotes the total inputs costs (

Table 2. Indicators and data sources used to calculate elasticities with respect to inputs.

Indicators	Code	Description of Definition and Calculation	Data Sources: Indicators [Code] in the FADN Public Database of the European Commission
Costs of paid and unpaid labour input	L = Lpl + wLupl	The cost of paid labour input Lpl is the amount of wages and social security charges (and insurance) of wage earners in the accounting year.	Wages paid (EUR) [SE370]
		The implicit cost of unpaid labour input wLupl is the amount obtained by multiplying the hours worked by unpaid labour input in the accounting year by the wage per hour worked w by paid labour input. (wLupl = SE016 × SE370/SE021)	Unpaid labour input (h) [SE016] Paid labour input (h) [SE021]
Deprecation of fixed capital	D	The cost of consumption of fixed capital in the accounting year.	Depreciation (EUR) [SE360]
Intermediate consumption of working capital	C	The intermediate consumption is the total cost of circulating capital in the accounting year.	Total intermediate consumption (EUR) [SE275]
Costs of rented and owned capital	I = Irc + rIoc	The cost of rented capital Irc is the total amount of interest and financial charges paid on loans obtained for the purchase of fixed and circulating capital, interest, etc.	Interest paid (EUR) [SE380]
		The implicit cost of owned capital Ioc is the amount obtained by multiplying the total assets in ownership by long-term interest rates r. (rIoc = r × (SE436 + SE437)/2)	Total assets, closing valuation (EUR) [SE436] Total assets, opening valuation (EUR) [SE437] Long-term interest rates—Maastricht criterion interest rates [irt_lt_mcby] (Eurostat online data code: irt_lt_mcby_a)
Costs of leased land and owned land	N = Nrl + Nol	The cost of leased land Nrl is the rent paid for farmland in the accounting year.	Rent paid (EUR) [SE375]
		The implicit cost of owned land Nol is the amount obtained by multiplying the rent per hectare of UAA rented by the holder r under a tenancy agreement by the area of owned UAA. (Nol = SE375/SE030 × (SE025 − SE030))	Rented UAA (ha) [SE030] Total utilised agricultural area (ha) [SE025]
Costs of total inputs	γ	The costs of total inputs linked to the agricultural activity of the holding and related to the output of the accounting year. (γ = L + D + C + I + N)

Source: authors’ own elaboration based on definitions of variables used in FADN standard results [75].

).

The elasticities with respect to each input were calculated using aggregated performance indicators of farms included in the FADN system. The data obtained from the FADN Public Database of the European Commission and Eurostat Database were used to calculate elasticities with respect to each input (for more detail, see Table 2).

3. Results

3.1. The Growth of GVA and Pollution-Adjusted GVA in Agriculture

Over the long term (2005–2021), most EU countries (twenty-two) achieved positive growth in agriculture from both the economic growth and green growth perspectives, i.e., in terms of GVA growth and pollution-adjusted GVA growth, respectively, as indicated in Figure 3a. The remaining five countries of the EU recorded a negative trend of both indicators in the long term. However, the values of both indicators were close to zero in Greece and Cyprus, which means that GVA and net GHG emissions in agriculture have remained stable over time. The gap in agricultural growth between EU countries was narrower when growth was measured in terms of pollution-adjusted growth rather than real economic growth.

According to Rodríguez et al. [76], the adjustment of production output growth for pollution abatement measures the extent to which economic growth is influenced by emission reduction efforts. A positive abatement adjustment indicates a decrease in pollution over the period, while a negative one indicates an increase. In this study, the adjustment of GVA growth for pollution abatement measured the extent to which economic growth in the agricultural industry is affected by efforts to reduce net GHG emissions from agricultural activities, including efforts to increase CO₂ absorption in grassland and pastures (Figure 3b). The agricultural GVA growth adjustment for pollution abatement was measured using the elasticity of GVA with respect to net GHG emissions from agriculture and the change in net GHG emissions. The elasticity values calculated for each country are presented in Appendix A.

In fourteen EU countries that reduced their net agricultural GHG emissions over time, the adjustment in GVA growth (i.e., the adjustment is positive) shows that efforts to reduce net GHG emissions reduced the growth of agricultural gross value added, i.e., slowed down economic growth in this industry. For instance, in the Slovakian, Romanian, and Portuguese agriculture industries, GVA grew at an average annual rate of 0.88%, 0.35%, and 0.09%, respectively, over the period analysed, while the pollution-adjusted GVA grew by 1.09%, 0.58%, and 0.19%, respectively, over the same period (Figure 3b), considering the efforts made by these countries in reducing their net GHG emissions. These adjustments were influenced by technological changes in agricultural production (e.g., a shift to soil-friendly tillage methods and automated fertigation systems), by structural changes in farming (e.g., a shift to less GHG-intensive farming), and by the implementation of CAP agri-environmental measures in EU countries. In the other eleven EU countries, where efforts to reduce agricultural GHG emissions were relatively weak, the GVA adjustment was close to zero.

In contrast, in thirteen EU countries that increased their GHG emissions from agriculture over the long term, the negative adjustment of GVA growth provides insights into the extent to which national income is generated at the expense of environmental quality. For instance, Bulgaria, Estonia, Slovenia, and Latvia’s GHG emissions from agriculture increased the most over the analysed period, reducing pollution-adjusted GVA growth (i.e., green growth) by 0.21%, 0.40%, 0.12%, and 0.10% points, respectively (Figure 3b). In countries where net GHG emissions from agriculture grew slowly over time, the GVA adjustment was close to zero, which indicates that economic growth in agriculture is decoupled from GHG emissions. An economic system that invests heavily in reducing pollution naturally faces the prospect of relatively less pollution-intensive long-term growth. Overall, as Rodríguez et al. [76] noted, economies that invest heavily in the reduction of pollution naturally face the prospect of relatively less pollution-intensive long-term growth.

3.2. Contribution of Production Factors and EAMFP to Agricultural Green Growth

Table 3. Contribution of inputs and EAMFP to pollution-adjusted GDP growth in agriculture of EU countries (2005–2021).

Countries	Output Growth			Input Growth			Residual Growth
	Pollution- Adjusted GVA Growth	GVA Growth	Adjustment for Pollution Abatement	Contribution of Labour	Contribution of Produced Capital	Contribution of Natural Capital	Growth of EAMFP
Czechia	1.108	1.103	0.005	−0.106	0.111	−0.003	1.105
Latvia	1.097	1.199	−0.102	−0.240	0.265	0.005	1.068
Slovakia	1.087	0.881	0.206	−0.290	−0.690	−0.002	2.068
Sweden	0.989	0.964	0.025	−0.087	0.158	−0.001	0.918
Lithuania	0.852	0.874	−0.022	−0.122	0.228	0.006	0.740
Luxembourg	0.782	0.709	0.073	−0.094	0.559	0.003	0.314
Ireland	0.668	0.678	−0.010	0.014	0.282	0.002	0.370
Austria	0.582	0.586	−0.004	−0.057	0.132	−0.003	0.510
Romania	0.579	0.347	0.231	−0.242	0.311	−0.002	0.511
Croatia	0.530	0.474	0.056	−0.103	−0.185	0.000	0.818
Hungary	0.523	0.555	−0.033	−0.091	0.297	−0.005	0.322
Denmark	0.436	0.390	0.046	−0.057	−0.166	0.001	0.657
Estonia	0.433	0.837	−0.405	−0.314	0.641	0.004	0.101
Bulgaria	0.415	0.625	−0.210	−0.285	0.339	0.007	0.354
Poland	0.381	0.384	−0.003	−0.102	0.093	−0.002	0.392
Germany	0.348	0.336	0.013	−0.041	0.133	−0.001	0.258
France	0.218	0.214	0.005	−0.048	0.163	−0.001	0.104
Portugal	0.186	0.094	0.092	−0.185	0.197	−0.002	0.176
Italy	0.185	0.188	−0.003	−0.055	−0.050	−0.003	0.292
Netherlands	0.154	0.148	0.005	−0.009	0.222	−0.002	−0.057
Belgium	0.147	0.146	0.001	−0.075	0.425	0.001	−0.204
Spain	0.107	0.101	0.006	−0.032	0.122	−0.002	0.019
Cyprus	−0.005	−0.018	0.013	−0.184	−0.676	−0.013	0.868
Greece	−0.012	−0.041	0.029	−0.166	0.262	−0.010	−0.098
Slovenia	−0.209	−0.090	−0.119	−0.074	0.122	0.004	−0.260
Finland	−0.211	−0.195	−0.016	−0.109	0.061	0.000	−0.162
Malta	−0.331	−0.324	−0.007	0.359	0.483	0.001	−1.175

Source: authors’ own calculations.

presents data from an empirical analysis of green growth in agriculture over the long term (2005–2021), where pollution-adjusted GVA growth is decomposed into the contribution of individual production factors (labour, produced capital, and natural capital) and pollution-adjusted multifactor productivity. The results show significant differences between EU countries.

During the analysed period, the contribution of EAMFP was particularly sizeable in two-thirds of the EU countries. Many of them relied more on technological progress than on traditional production factors (labour and produced capital) to promote green growth in agriculture. For instance, when analysing the contribution of EAMFP to agricultural growth compared to the other growth components (Table 3), it was found that productivity growth more than compensated for the declining contribution of all three production factors (labour force, produced capital, and natural capital) in EU countries such as Slovakia, Italy, Denmark, Croatia, and Cyprus. Additionally, in almost a third of EU countries (like Latvia, Czech Republic, Sweden, Lithuania, Romania, Austria, Poland, and Germany), EAMFP accounted for more than 50% of pollution-adjusted GVA growth on average.

Meanwhile, in the rest of the EU countries, produced capital was an important source of gross value-added growth in the agricultural industry, playing a bigger role than labour or natural capital (Table 3). During the analysed period, fixed capital formation was the main source of agricultural growth and more than compensated for the declining contribution of other factors of production (labour and natural capital) to pollution-adjusted GVA growth in Finland, the Netherlands, Belgium, and Slovenia. Produced capital fuelled a significant share of pollution-adjusted GDP growth in some other EU countries. For instance, more than 50% of agricultural green growth can be directly attributed to capital investment in Portugal, France, Malta, the Netherlands, Greece, and Spain, while in Estonia and Luxembourg, it represents more than 80%.

In most EU countries, excluding Ireland and Malta, the gradually decreased labour input over the long period under consideration had a negative contribution to economic growth in agriculture and, therefore, needed to turn to other factors to fuel their economic growth (Table 3). In the countries (e.g., Bulgaria, Slovakia, Romania, Estonia, and Latvia) where the agricultural labour force shrunk the most over time (on average by −4.5% to −8.0% p.a.), the decline in labour input had a significant negative impact on pollution-adjusted GDP growth (between −0.25% and −0.32%). To offset the negative impact on agricultural growth due to reduced labour input, Slovakia and Latvia relied on EAMFP growth and technological progress, Bulgaria and Romania relied on both EAMFP growth and produced capital investment, and Estonia relied mainly on increasing investment in produced capital.

In contrast, Ireland and Malta increased their labour use in agriculture over time. The slow growth in labour input did not significantly contribute to green agricultural growth in Ireland (i.e., increasing pollution-adjusted GVA growth by 0.01%). Meanwhile, efforts to increase the use of the labour force in agriculture played a significant role in stimulating economic growth in Maltese agriculture (i.e., increasing pollution-adjusted GDP growth by 0.36%).

Natural capital, expressed in terms of quality-adjusted agricultural land, played a low-significance role in the growth of agriculture in all EU countries. The contribution of natural capital to pollution-adjusted GVA growth was low compared to the contribution of produced capital and labour (Table 3). This can be explained by the fact that EU countries have reached a stage where it is no longer possible to substantially increase the amount of land used for agricultural production due to the natural constraints of land area. Land is being taken out of agricultural production due to the expansion of infrastructure and cities, afforestation, etc. The results of this study show that in most EU countries, the average annual change in natural capital over the long term was less than 1%, either downwards or upwards, as countries increased or reduced their agricultural area. In Cyprus and Greece, the annual decrease in agricultural area was 1.9% and 1.3%, respectively, while in Latvia, the average increase was 1.3%. As a result, the contribution of natural capital to pollution-adjusted GVA growth in the agricultural industry was close to zero in all EU countries.

3.3. Comparison of EAMFP and MFP in Agriculture

EAMFP estimates a country’s ability to generate production output from a given set of inputs, including environment-related inputs and pollution as undesirable environment-related outputs [48]. Unlike the MFP measurement framework, where the production output (produced goods and services) is compared to combined production inputs (labour and produced capital), the EAMFP measurement framework clarifies the contribution of environmental pollution and natural capital to production output, i.e., adjustment from pollution abatement and natural capital use, respectively, and also clarifies the decreased contribution of traditional inputs (labour and produced capital) to production output [48].

In this study, Figure 4 presents MFP and EAMFP indicators in agriculture for each country in the EU. The value of the average annual growth rate of traditionally calculated MFP over a long period was higher than the value of the same EAMFP indicator in two-fifths of the EU countries. Meanwhile, an inverse difference between both indicators was found in one-quarter of the EU countries. The same values of both indicators, MFP and EAMFP, were found only in Ireland. The biggest positive value gap (0.09% and above) between both the MFP and EAMFP indicators was found in Estonia, Latvia, Bulgaria, and Slovenia, while the biggest negative value gap (−0.09% and above) between MFP and EAMFP was found in Romania.

Depending on the contribution of natural capital and the adjustment for air pollution emission reduction, traditional MFP could be over- or undervalued concerning EAMFP [48], i.e., when MFP > EAMFP, the MFP growth was probably overestimated, and conversely, when MFP < EAMFP, the MFP growth was perhaps underestimated.

In this study, the traditionally calculated MFP indicator was likely to be overestimated in Estonian, Latvian, Bulgarian, and Slovenian agriculture due to unconsidered net GHG emissions. Conversely, the MFP indicator was likely to be underestimated in Romania, Slovakia, Luxembourg, and Portugal due to GHG emissions from agricultural reduction efforts, including efforts to increase CO₂ absorption in grassland and pastures.

4. Discussion

Green growth is a complex issue and difficult to measure and compare [36]. In this study, a pollution-adjusted GVA accounting framework was developed to empirically measure and compare agricultural green growth across EU countries. The results of the empirical analysis revealed that most EU countries have achieved positive green growth in the agricultural industry from both the economic growth and green growth perspectives in the 2005–2021 period. Only three countries, namely, Slovenia, Finland, and Malta, recorded a negative trend in both GVA growth and pollution-adjusted GVA growth, while the values of both indicators were close to zero in Greece and Cyprus. The gap in agricultural growth between EU countries was narrower when growth was measured in terms of pollution-adjusted growth rather than real economic growth. Kasztelan et al. [36] conducted an empirical analysis of green growth in agriculture in twenty-five EU countries five years ago and found significant differences between them. Unlike our study, the authors used an aggregate index calculated from a set of selected agricultural indicators in their agricultural green growth analysis.

The empirical results of this study showed that slow changes in environmental pollution in the long term, expressed in terms of net GHG emissions, did not have a significant impact on pollution-adjusted GVA growth in the agricultural industry of any EU country. The European Union’s Common Agricultural Policy (CAP) aims to reduce environmental pollution from agricultural production to levels that are considered to be safe for human health and natural ecosystems [77]. Due to environmental standards in agriculture set by the EU, the EU plays an important role in mitigating the impact of agriculture on climate change, and since 2013, the fight against climate change has become one of the main objectives of the CAP. Although the European Commission allocated over EUR 100 billion of CAP funds to the 2014–2020 period to tackle climate change [ibidem], as the results of this study have shown, the net GHG emissions in almost half of EU countries did not decrease over this period. One of the possible solutions to this problem is the involvement of the agricultural sector in the carbon trading system. Yu et al. [78] stated that carbon trading has been widely recognised around the world as a core environmental management tool to reduce the negative effects of various economies on the climate. The authors took the stance that the implementation of a carbon dioxide trading system in the agricultural sector can significantly promote the improvement of total green factor productivity in China’s agriculture. In the near future, the EU will establish a Union certification framework for carbon removal, carbon farming, and permanent carbon storage in products, aiming to facilitate and accelerate carbon removal activity and reduce GHG emissions from the soil in agriculture [79]. Once operational, this certification framework should encourage farmers to carbon capture in soil and reduce net GHG emissions from agriculture, thus contributing to the green growth of agriculture.

This study showed that the pollution-adjusted GVA growth in agriculture was based on technological progress in most of the EU countries. This conclusion is supported by other studies that have found that increasing technological innovation capacity has a statistically significant positive effect on changes in green growth in agriculture, with both direct and indirect positive effects [80]. Moreover, green growth in agriculture is improved by green technology innovation, i.e., digital transformation of agriculture [81,82]. Huang et al. [83] stated that energy conservation and emission reduction are the primary drivers of EAMFP, accounting for more than half of China’s total agricultural green factor. The empirical results of this study showed that agricultural EAMFP growth more than compensated for the declining contribution of the labour force, produced capital and natural capital in Slovakia, Italy, Denmark, Croatia, and Cyprus, and accounted for more than 50% of pollution-adjusted GVA growth on average in almost a third of EU countries. This study revealed that the contribution of natural capital (expressed as quality-adjusted agricultural land in use) to agricultural green growth was close to zero in most EU countries due to the amount of utilised agricultural area that remained stable for a long time (bearing in mind minimal annual change rates, either downwards or upwards). Hamilton et al. [84] held the view that the natural capital contribution to productivity growth is particularly significant in resource-dependent countries.

Since the quantity of agricultural land is more or less constant in most EU countries, the contribution of natural capital to pollution-adjusted GVA growth would be increased by way of extended areas equipped for irrigation or investments in new, more efficient irrigation technologies. This conclusion is supported by other studies [66,73], which argue that an increase in irrigated land area has a positive effect on output growth. Irrigation helps to protect crops from irregular rainfall and increase their vitality, yield, and quality [77] and is or can be an instrument to reduce drought risk in agriculture production in various EU regions (especially in the arid and semi-arid areas of southern Europe, e.g., Spain, Portugal, Italy, Greece and southern France, in which irrigation allows crop production where water would otherwise be a limiting factor, and in more humid and temperate areas, e.g., Denmark, Belgium, the Netherlands, Luxembourg, north and central France, Germany, southern Sweden, and eastern Austria, in which irrigation provides a way of regulating the seasonal availability of water to match agricultural needs) [85].

The contribution of labour to pollution-adjusted GVA growth suggests that all EU countries (except Malta and Ireland) had a labour input decrease during the period 2005–2021 and, therefore, needed to turn to other factors to fuel their economic growth. For instance, some EU countries relied on EAMFP growth and technological progress (e.g., Slovakia and Latvia) to offset the negative labour input impact on agricultural growth, while other countries relied on EAMFP growth and produced capital investment (e.g., Bulgaria and Romania), and the rest of the countries relied on increasing investment in produced capital (e.g., more than 80% in Estonia). These findings are supported by other studies [86,87,88], which highlight that the decline in labour force growth over time has been accompanied by the ageing of farmers and worker migration, alongside the growth of productivity.

5. Conclusions and Research Limitations

5.1. Conclusions

For the empirical analysis of agricultural green growth in each individual EU country, this study developed a pollution-adjusted GVA growth accounting framework for agriculture based on the environmentally adjusted multifactor productivity (EAMFP) measure that is used as the OECD green growth headline indicator at the macroeconomic (i.e., country) level. The study sample comprised EU27 countries from 2005 to 2021. The framework allows accounting for natural capital (quality-adjusted agricultural land) and greenhouse gas emissions (positive or negative carbon dioxide per year in terms of emissions/removal from agriculture). The findings indicated that long-run pollution-adjusted GVA growth rates differ between EU27 countries. The key difference in the overall growth performance was due to the extent to which they relied on increased productivity gains or better use of factor inputs.

We found that more than two-thirds of EU countries increasingly relied on technological progress (i.e., growth in EAMFP) rather than on traditional production factors (i.e., growth in labour and produced capital) with the aim of increasing GVA in agriculture. Meanwhile, in the rest of the EU countries, produced capital was an important source of gross value-added growth in the agricultural industry and played a bigger role than labour or natural capital. Furthermore, agricultural labour only contributed to increased GVA growth in agriculture in Ireland and Malta. On the contrary, in the remaining countries, the contribution of labour reduced GVA growth in agriculture. The natural capital, expressed in terms of quality-adjusted agricultural land, played a low-significance role in agricultural growth in all EU countries.

5.2. Research Limitations and Future Directions

The present study has some limitations, mainly related to the narrowed expression of natural capital. Some research [13,46,47,48,84,89] limits the scope of natural capital to traded natural commodities or land because of the lack of necessary information on all-natural capital assets (such as stock, flow, and price information). Rodríguez et al.’s [76] recent study involved the use of renewable natural capital, non-cultivated biological resources, and ecosystem services at the macroeconomic level. In this study, quality-adjusted agricultural land served as a natural capital input variable in the EAMFP model, considering the variations in the productive capacity across three types of agricultural land (such as irrigated and rainfed cropland and permanent pastures). In forthcoming research, this variable should be supplemented with a fourth type of agricultural land, i.e., cropland under organic management. It has been estimated that long-term ecological land management leads to lower yields than conventional management [90]. However, these differences are highly context-dependent, taking into consideration the characteristics of both management systems and the local context [91]. Therefore, as a first step, large-scale detailed studies on yield differences between organic and conventional crops are needed to establish valid quality weights for organic croplands. In addition, in forthcoming research, the quality-adjusted labour input (i.e., human capital attributes such as academic qualifications, trade qualifications, and experience [92]) could be applied in an analogous way to the quality-adjusted agricultural land variable in the agricultural green growth accounting framework. Wang et al. [93] stated that an increase in labour quality leads to an increase in quality-adjusted labour input. It has been estimated that quality-adjusted labour input can adjust labour input and contribute to a greater extent to real MFP growth [93,94]. However, it is difficult to obtain data on labour quality at the level of the agricultural industry.

Second, the elasticities with respect to the inputs were based on the FADN data of commercial farms with an economic size of at least EUR 4000 of standard output. Thirdly, the elasticities with respect to undesirable outputs were estimated by their carbon emission prices obtained from the EU ETS. These prices reached EUR 26 per CO₂ equivalent in 2020 and increased more than twice thereafter. Therefore, environmental pollution might have a greater impact on agricultural green growth analysis in the future.

Despite the limitations mentioned above, this study will significantly contribute to improving the green growth accounting system in agriculture, and the results of this empirical analysis will be used by policymakers and economists, bearing in mind the crucial role of green growth in the policy space [95]. In addition, based on the results of the empirical analysis, decision-makers will be able to identify areas for improvement so that all EU countries can fully embark on the path of green agricultural growth.