Aligning agri-environmental subsidies and environmental needs: a comparative analysis between the US and EU

We analyzed average US EQIP spending between 2012 and 2014, as these are the most recent years with complete data available (EWG 2020). For the EU (including the UK), we used European Commission data on EAFRD spending for the years 2014 and 2015 reported at NUTS2 level (EC 2018). These years were chosen to reflect the new funding mechanisms in effect since the 2013 CAP reform, but are close enough to the EQIP spending years to warrant a comparison with the US.

EQIP focuses on a set of over 170 practices, ranging from nutrient and crop residue management to irrigation and grazing. To approximate the EAFRD spending allocated for comparable measures, we corrected total EAFRD spending by the proportion of spending for measures M4 ('Investments in physical assets') and M10 ('Agri-environment-climate') using an additional dataset (EC 2020a). These measures were deemed analogous to those of EQIP as they promote sustainable working land and livestock agricultural practices while largely excluding forestry investments. If spending was reported at lower resolution (often NUTS1 level), we used the same proportion for all underlying NUTS2 regions. Since our analysis focuses on spatial targeting rather than temporal analysis of AES impacts, we calculated AES as the average spending per hectare of agricultural land (table 1) by county in the US ($ ha⁻¹) and by NUTS2 region in the EU ($e$ ha⁻¹), modelling 2835 US counties and 211 EU NUTS2 regions.

We selected 14 indicators based on their environmental, socio-economic and policy relevance (table 1). These included: (1) six environmental indicators targeted by AES and capturing environmental need; (2) three socio-economic variables related to how subsidies are allocated and applied for in practice; and (3) four contextual variables describing regional agricultural systems. Table 2 describes their policy relevance within AES frameworks. See appendix D for indicators' maps.

Spatial data were masked to agricultural areas and averaged per region (US county or EU NUTS2). High indicator scores reflect high environmental need across all variables. The direction of soil organic carbon (SOC) and local biodiversity intactness were inverted to achieve this. Socio-economic and contextual indicators were included as drivers of allocation and farmer participation that shape environmental outcomes. Predictors were scaled to zero mean and unit standard deviation.

We conducted two sets of analyses for the continental US (48 states) and most regions of the EU (23 MS) to reflect the two-tier allocation system of the subsidy programs. First, we evaluated AES spending patterns among states ('multi-State models'). Then, we assessed spending patterns within states ('individual-State models') to examine how local spending related to the general patterns observed in the multi-State models. All analyses were conducted in R (R Core Team 2020).

This study covers a limited timespan and historical AES allocation may have influenced the predictors due to changes in agricultural management (Taylor and Morecroft 2009, Meals et al 2010, MacDonald et al 2012). Thus, our findings regarding AES spending are relevant primarily for the years analysed.

3.3.1. Multi-state analysis

3.3.2. Individual-state analysis

In the US, high EQIP spending per hectare was concentrated in the coastal regions, while allocation in the Great Plains region was strikingly lower (figure 1). In the EU, most regions with the highest AES spending per hectare are found in Austria, Netherlands, Germany, Italy, Portugal, the Czech Republic and the UK (figure 1). Total average yearly EAFRD spending was 3.2 times higher than EQIP spending for the years of this study (table 3).

Zoom In Zoom Out Reset image size

For the US, N and P balance, soil erosion, manure runoff, proportion of young farmers and farm income did not remain in the final model (figure 2, appendix E). The total variance explained was relatively high (R² _conditional = 0.65), but only partially related to the fixed effects (R² _marginal = 0.11). SOC and GHG emissions were positively associated with receipt of subsidies, meaning areas with low SOC and high GHG emissions received more subsidies. However, agricultural water stress (AWS) and local biodiversity were negatively associated with spending, suggesting a mismatch with these environmental needs. All socio-economic variables and contextual factors were negatively related to receipt of subsidies. Agricultural land cover ratio showed the largest effect size.

Zoom In Zoom Out Reset image size

In the EU, average AWS, biodiversity, manure runoff, farm size, and proportion of rented land did not remain in the final model (figure 2, appendix E). Together, fixed and random effects explained half of the variance (R² _marginal = 0.25, R² _conditional = 0.54). Of the environmental need indicators, GHG emissions, N balance and SOC were positively associated with receipt of subsidies, indicating a match between subsidies and these environmental needs, while soil erosion and P balance showed a negative association, suggesting a mismatch in allocation. Similarly to the US, the proportion of agricultural land cover had the largest effect size, with a negative relationship to AES payments. Of the socio-economic variables and contextual factors, agricultural income, GDP, and proportion of young farmers were negatively associated with receipt of subsidies. The opposite was true for the proportion of female farmers.

The US and EU models both had some degree of spatial autocorrelation in the residuals, with Moran's I of 0.16 and 0.15, respectively. However, the indicators remained relatively constant when different spatial modelling methods were used (see appendix H for results).

See appendix F for the distribution of correlation coefficients between spending and predictors.

Average model fit in the US was higher for states with highest (mean R² = 0.57 ± 0.27, see appendix G) than for states with lowest distance from multi-state correlation (mean R² = 0.19 ± 0.07). GHG emissions were significant and positively associated with subsidies allocation in Arkansas and North Carolina, but negatively associated in Montana. Water stress and biodiversity showed negative associations with most selected states, and were significantly negatively associated with allocation in Oregon and Wyoming, and New Mexico and Oklahoma, respectively. Subsidies allocation in Nebraska, Idaho, and Tennessee showed no significant association with the indicator variables.

For the EU MS models, few indicators showed significant associations to AES spending per area (see appendix G). In the UK, only the proportion of agricultural land cover was significantly negatively associated with subsidies, and the positive association with GHG emissions bordered on significance (P = 0.051). In France, average farm income was significantly positively associated with subsidies allocation. Models for Italy and Germany showed no significant associations.

Our analyses showed success of both EQIP and EAFRD in targeting areas of high GHGs, reflecting the programs' goals regarding agricultural emissions reduction and carbon sequestration. This alignment was also reflected in the models of North Carolina and Arkansas, large producers of livestock, rice, soybeans, feed grains, and cotton—all strong drivers of agricultural GHGs (EPA 2015, FTM 2016, NASS 2017). Cropland and livestock emissions reduction are fundamental for meeting Paris Climate Agreement of mitigation targets (Reisinger and Clark 2018, Rogelj et al 2018), and our results suggest that AES may be successfully targeting areas of greatest need. Although both programmes showed a positive association with SOC, this was only significant in the US. As improved SOC sequestration and agricultural GHG emissions mitigation are linked (Frank et al 2017), this may be a potential win-win scenario from the same AES measures. These results are promising and warrant further investigation to understand the drivers of this successful targeting, which could be applied to other areas of environmental need.

The EU model illustrated EAFRD's success in targeting areas of greatest N surplus. The EU has been addressing issues of N imbalance since the 1990s with the nitrates directive (EU 1991), and our results indicate that the spatial distribution of AES funding may help further reduce N surplus. In contrast, this match was not mirrored by P surplus, which was negatively associated with AES spending. This result warrants further exploration, as the EU lacks a common P management strategy (Ronchi et al 2019).

More so than successes, clear mismatches of funding and environmental needs emerged from our results, highlighting potential misdirection of AES and gaps in environmental targeting of EQIP and EAFRD. Although indicators of soil erosion, biodiversity loss, water stress, and nutrient management are explicitly featured in AES programs' goals (table 2), both the US and the EU showed mismatches between policy priorities and funding allocation. Soil erosion was negatively associated with AES in the EU, and did not feature in the US model, while the opposite was true for local biodiversity.

In the EU, the discrepancy between funding allocation and soil erosion comports with previous studies highlighting how the lack of a common binding strategy has inhibited soil conservation efforts (Turpin et al 2017, Helming et al 2018), a shortcoming that the new Healthy Soil Initiative strives to mitigate (EC 2020b). Moreover, the absence of an association with biodiversity conforms with existing literature highlighting insufficient spatial targeting of conservation measures (Pe'er et al 2014, Batáry et al 2015). In the EU, 0.57% of EAFRD funding is reserved for Natura 2000 sites (Measure 12, Dwyer et al 2016); however, the agricultural matrix surrounding protected areas is crucial in supporting biodiversity (Gonthier et al 2014) and biodiversity conservation is among the supporting ecosystem services financed within Measure M10 ('Agri-environment climate').

We found that subsidy allocation in the US concentrated in areas of reduced water stress need, while the variable was excluded from the EU model. Agriculture is a major cause of freshwater ecosystems degradation (Allan 2004, Poole et al 2013) and water efficiency is becoming increasingly important in light of climate change and groundwater depletion (Marshall et al 2015, Cotterman et al 2018). Thus, this mismatch warrants further policy attention. While it is possible that high water stress and at-risk biodiversity areas are more likely to be enrolled in the CRP program in the US due to its focus on environmentally sensitive lands, or in complementary initiatives like the Conservation Reserve Enhancement Program, this result also highlights the lack of geographical targeting by EQIP since its 2002 reform (Shortle et al 2012, Drevno 2016, Hellerstein 2017). At the state level, water stress emerged as a strong negative predictor in Oregon, indicating a mismatch between EQIP spending and environmental need. Oregon's water supply is threatened by droughts and irrigation demands, with repercussions for agriculture and public health (OEC 2012, Schimpf and Cude 2020). In the EU, the post-2020 CAP reform revised the water exploitation index to relate water stress to availability of renewable water resources, possibly enabling more efficient AES allocation in the future (EEA, 2020).

Nutrient surplus and runoff, especially when tied with water stress and flooding, have serious socio-environmental repercussions that can be ameliorated with sustainable practices (Jones et al 2017, Blanco-Canqui 2018, Roy et al 2021); however, they did not remain in the US model. As the US accounts for large proportions of global livestock production and cropland N surplus (West et al 2014, FAO 2019b), nutrients are known threats to the US water supply (Grant et al 2002, Howarth et al 2002). EQIP subsidizes nutrient management plans and infrastructure development to improve surface water quality, and mandates 50% of total spending to livestock producers (NRCS 2017). Notably, it provides waste management assistance to concentrated animal feeding operations, leading sources of nonpoint water pollution due to chemical inputs and manure runoff from animal feed crops (Burkholder et al 2007, Martin et al 2018). While conservation practices have been shown to reduce nonpoint source loadings locally (Poudel 2016, Liu et al 2018, Sneeringer et al 2018), significant amounts of agricultural land are over-fertilised and need improved nutrient management (Jackson et al 2000, NRCS 2011, 2013, 2014a, Long et al 2018). Moreover, these 'pay-the-polluter' initiatives have been criticised due to insufficient resources, as well as their full reliance on voluntary compliance (Collins 2012, Shortle et al 2012, Shortle and Uetake 2015, Drevno 2016). It should be noted that demand for EQIP funding consistently exceeds allocation (Stubbs 2010), thus, funding limitations may constrain AES' ability to optimize spatial targeting; our results highlight a potential gap in AES targeting which should be further investigated, particularly due to the severe impacts of nutrient surplus and manure runoff associated with US high livestock densities.

In some cases, social context may explain the mismatch between funding and environmental need. In the EU, AES were allocated to areas with higher proportions of female producers, suggesting that EAFRD program goals regarding gender equality and rural development successfully influenced funding allocation, despite, or perhaps in competition with, environmental need. In the US, instead, the proportion of female producers was negatively associated with county spending. This could be due to gender inequality not being a main focus of EQIP per se, but only of overarching USDA policies (NRCS 2019). There may be opportunities for EQIP to learn from EAFRD's focus on social inclusion, especially considering the propensity of women farmers to farm in sustainable, conservation-oriented ways in line with AES program goals (Paul and Fremstad 2016).

Neither program was successful at targeting younger farmers, although only the EU includes additional benefits for young farmers in policy targets (table 2). This supports the findings from an EU audit that despite policy priorities, young farmer participation is declining due to poorly defined interventions and unsatisfactory monitoring systems (ECA 2017). Rented land did not remain in the EU model, and our findings suggest a negative association between tenancy and AES in the US. While Reimer et al (2013) argued that EQIP funding may be preferred over CRP in states with larger proportions of rented land, other studies indicated land ownership as a positive predictor of EQIP participation at local scales (Parker et al 2007, Nyaupane et al 2012, Zhong et al 2016). Rented land is not an explicit target of either AES program; however, previous studies have evidenced lower uptake of sustainable practices by tenant farmers, particularly in conventional systems (Sklenicka et al 2015, Walmsley and Sklenička 2017, Ranjan et al 2019). Local EQIP initiatives and selected EAFRD measures could thus be tailored towards tenants to increase widespread adoption of sustainable practices in conventional landscapes.

Given EAFRD's re-distributive goals, it is somewhat surprising that farm income was not significantly associated with AES allocation in the EU, although a negative association did exist. This could be due to our focus on measures M4 ('Investments in physical assets') and M10 ('Agri-environment-climate'), while excluding other EAFRD funds. Furthermore, there are myriads of drivers which could motivate farmer AES applications that we do not capture here, including personal beliefs, previous experiences and crop prices (McCracken et al 2015, Pavlis et al 2016, Holland et al 2020). Without detailed information on the applications received and approved, we cannot isolate drivers of farmer adoption from those of EQIP and EAFRD's selection criteria when interpreting socio-economic patterns of AES allocation.

Contextual factors related to agricultural production systems may also explain mismatches between AES spending and environmental need, as previous studies have suggested that production system type and structure are a larger driver of subsidy allocation than environmental conditions (Reimer et al 2013, Reimer and Prokopy 2014, Zasada et al 2018). Our findings suggest that the potential for AES to remediate environmental issues may be curtailed within policy frameworks by limited participation from farmers engaged in highly intensive and expansive operations. This may be due to higher utilisation costs of certain conservation practices in intensive farming, or a greater predisposition for these practices in lower intensity areas (see Früh-Müller et al 2019). This was evidenced by significant negative associations with the proportion of agricultural land cover in the US and EU, and with farm size in the US. In the US, AES spending per area was strikingly lower in central regions typically dominated by large farms, while coastal areas had higher spending, despite being less dominated by agriculture. Similarly, in the EU, predominantly agricultural regions such as central Germany and France received lower payments per area.

Our results align with Zasada et al (2018), who also found that smaller farms were more likely to receive higher AES in the EU. However, previous research on the influence of farm size on US conservation practice adoption reports contrasting results, with some finding larger farms more (Baradi 2009) and other less (McLean-Meyinsse et al 1994) likely to adopt conservation practices. Others assert that the influence of farm size varies depending on the management practice and conservation program in question (Soule et al 2000, Lambert et al 2007, Reimer 2015). Future research should further explore whether these farms are less likely to engage with AES—particularly since they are drivers of sustainability issues and experience environmental risks that AES may help mitigate.

Allocation is administered by the USDA's Natural Resource Conservation Service (NRCS), distributed from the federal government to each state first, and then distributed within each state to its counties. For the national allocation to states, NRCS determines each state's EQIP funds using a formula that reflects national priorities and available natural resources, including: the significance of environmental and natural resource concerns and the opportunity for environmental improvement, the ways the program can best assist producers in complying with Federal, State, local, and Tribal environmental laws, and the amount of agricultural land in different land use categories. For state-level fund distribution, the State Conservationist develops an allocation formula, considering State and local level resource concerns, science-based information on environmental status, and relevant local programs and specialized farming operations (e.g. specialty crops, livestock, organic, small-scale), among other factors (NRCS 2019). The incentives provided include technical assistance and cost-shares of up to 75% of implementation costs.

Funds are provided in the form of less favoured area payments, agri-environment schemes, and investment support towards rural development in MS. The framework for fund allocation is based on a two-tier process: at the central stage (EU level), the overall framework is established, financial modalities are outlined and eligibility criteria are defined. Additionally, the EU outlines a set of six priority areas, including fostering knowledge transfer and innovation; enhancing viability and competitiveness of agriculture; promoting food chain organization, animal welfare and risk management; promoting resource efficiency, and low-carbon and climate resilient agriculture; preserving and enhancing ecosystems; and promoting social inclusion, poverty reduction, and economic development in rural areas. The distribution of the overall amount for rural development between MS is based on objective criteria and past performance. At the second level, MS develop national strategic plans and set quantitative objectives for priority areas. At least four of these national level priorities must address the EU level priorities. A minimum value of 30% of rural development funds must be set aside for environmental management measures, falling primarily within the measures considered in this study: M4 'Investments in physical assets', which receives the highest share of EAFRD budget (24%), and M10 ('Agri-Environment Climate'), which receives the third highest share (20%, ENRD CP 2015). There are several AESs across different EU MS, and the allocation mechanisms for specific subsidies tend to vary depending on the focus of the scheme (EC 2013).