Paying farmers to be greener does not necessarily change how they farm
Imre Fertő and Štefan Bojnec
European agricultural policy increasingly relies on financial incentives to encourage farmers to adopt more environmentally sustainable practices. Rather than simply regulating agricultural production, governments pay farmers to undertake voluntary commitments intended to protect biodiversity, improve soil and water quality, and reduce environmental pressures.
The underlying assumption is straightforward: compensating farmers for environmentally beneficial practices should encourage changes in how they manage their land.
But what happens if farmers receive payments without substantially changing their behaviour?
In our recent study published in Environmental Sciences Europe, we investigate this question using Hungarian farm-level data covering 2014–2020. We find little evidence that entering an agri-environment-climate scheme (AECS) leads to broad or persistent reductions in fertiliser use.
The findings highlight a fundamental challenge for evidence-based policymaking: funding environmental commitments and demonstrating additional environmental benefits are not the same thing.
The difference between participation and impact
Agri-environment-climate schemes are an important instrument of the European Union’s Common Agricultural Policy (CAP). They compensate farmers for voluntary commitments that go beyond mandatory environmental requirements, potentially offsetting additional costs and income foregone.
However, participation alone tells us relatively little about whether environmental outcomes have improved.
Farmers may enrol in schemes because their existing practices already satisfy programme requirements. Others may undertake commitments related to grassland management, biodiversity or landscape protection that do not necessarily involve reducing fertiliser applications.
This creates a problem of additionality: would the environmentally desirable activity have occurred without public support?
If participating farms already use fewer fertilisers than non-participants, this difference might reflect the programme’s effectiveness. But it might equally reflect differences in farm size, production specialisation, soil characteristics or existing management practices.
Simply comparing participating and non-participating farms therefore risks attributing pre-existing differences to the policy itself.
The relevant question is not whether participating farms behave differently, but whether participation changes what they would otherwise have done.
What happens when farms enter an environmental scheme?
To investigate this question, we analysed a balanced panel of 957 Hungarian commercial farms observed annually between 2014 and 2020.
We focused on fertiliser application, distinguishing between total fertiliser use and its three main nutrients: nitrogen, phosphorus and potassium. These inputs are essential for agricultural production, but their inappropriate use can contribute to water pollution, nutrient imbalances and greenhouse-gas emissions.
An important feature of our research is that we examine changes in programme participation over time.
Many policy evaluations effectively assume that once a farm joins a programme, it remains a participant. In practice, farmers can enter and leave schemes. Ignoring these transitions may obscure important behavioural responses.
We therefore used a dynamic difference-in-differences approach, comparing changes in fertiliser use among farms entering or leaving the scheme with changes among eligible comparison farms that had not yet changed participation status.
Of the 957 farms, 59 entered after 2014, while 150 were observed leaving during the study period.
The results provide little evidence of a systematic reduction in fertiliser use following entry.
Total fertiliser use declined slightly in the first and second periods after entry, but neither estimate was statistically significant. We also found no statistically significant reductions in nitrogen or phosphorus applications.
There was one exception: potassium use declined significantly in the first period after entry. However, this effect was no longer statistically significant in the second period.
This suggests an isolated short-term adjustment rather than a sustained transformation in nutrient management.
Importantly, the complementary analysis using alternative difference-in-differences methods and different comparison-group restrictions produced similar overall conclusions. The main findings were not driven by a small number of unusual farms or extreme statistical weights.
Why might financial incentives produce limited behavioural change?
At first sight, these findings might appear disappointing. If governments pay farmers to support environmental objectives, should we not expect to observe changes in agricultural inputs?
There are several possible explanations.
First, environmental schemes support different objectives. A farmer receiving payments for maintaining biodiversity-rich grassland may be delivering valuable ecological services without changing recorded mineral fertiliser use. Combining different commitments into a single participation measure can conceal these differences.
Second, farmers may select programmes that fit their existing management practices. If compliance requires little adjustment, enrolment generates limited additional behavioural change even when farmers fulfil their contractual obligations.
Third, agricultural decisions are shaped by complex economic and agronomic considerations. Fertiliser applications depend on expected yields, input prices, weather conditions, soil nutrient stocks and crop rotations. Financial incentives may be too weak or indirect to produce immediate changes against these pressures.
Finally, behavioural adjustments may take time. Farmers may need technical advice, new equipment or confidence in the continuity of policy support before changing established production systems.
Our data cannot distinguish conclusively between these explanations. Nevertheless, they illustrate why the relationship between public payments and environmental performance should not be taken for granted.
Environmental improvements cannot be reduced to one indicator
A further question is whether changes in fertiliser use are accompanied by changes in agricultural productivity.
Reducing fertiliser applications while maintaining output could indicate improved partial input productivity. Conversely, reduced input use accompanied by falling output might suggest an economic trade-off.
We therefore examined land productivity, measured as agricultural output per hectare.
We found no statistically significant changes following programme entry.
However, this should not be interpreted as evidence that farm efficiency remained unchanged. Output per hectare is influenced by prices, production choices and other inputs, and does not measure overall technical or environmental efficiency.
More fundamentally, the absence of measurable changes in fertiliser use is not proof that an environmental scheme has failed.
Biodiversity improvements, soil carbon accumulation, erosion prevention and water retention may develop over longer periods and may not be visible in farm accounting records.
Similarly, reducing fertiliser quantities does not automatically guarantee lower nutrient losses. Environmental outcomes depend on when, where and how nutrients are applied, alongside soil conditions and crop requirements.
Our results therefore reveal a limitation in the evidence available to policymakers, not a comprehensive verdict on the environmental effectiveness of the schemes.
From paying for participation to evaluating environmental outcomes
What do these findings mean for the future of European agricultural policy?
The first implication is that environmental policy evaluation should move beyond measuring participation and expenditure.
Numbers of enrolled farms, hectares under contract and payments distributed are important administrative indicators. But they cannot establish whether a policy has generated additional environmental benefits.
Where reducing nutrient pressure is an explicit objective, evaluations need information on the commitments farmers undertake, their previous management practices and the changes that follow participation.
Where schemes target biodiversity, soil protection or climate resilience, those outcomes should be measured directly rather than inferred from general input indicators.
Second, programme design needs to recognise the diversity of agricultural systems.
Small family farms and large commercial holdings differ in administrative capacity, access to advice and opportunities to invest in new technologies. Uniform incentives may therefore produce different responses across farms.
Third, environmental monitoring and advisory support deserve greater attention. Better monitoring could help establish whether payments correspond to meaningful management changes, while technical advice could reduce the costs and uncertainty associated with more sustainable practices.
These are plausible policy improvements rather than solutions directly demonstrated by our estimates. Their effectiveness should itself be evaluated.
Finally, agricultural income support and environmental payments need to remain conceptually distinct. Supporting farm incomes and rural livelihoods can be legitimate policy objectives. But when payments are justified by environmental benefits, their effectiveness should be assessed against those objectives.
Better data are essential for better environmental policy
Our analysis also points to a wider research challenge.
Farm-level accounting data make it possible to examine changes in economic behaviour, but they provide only a partial picture of environmental sustainability.
The transition from the EU’s Farm Accountancy Data Network to the broader Farm Sustainability Data Network offers an opportunity to improve this evidence base.
Yet collecting more indicators will not be sufficient on its own.
Credible evaluation requires linking information on programme participation and contractual commitments with management practices, environmental outcomes and sufficiently long farm histories. It also requires appropriate comparison groups to distinguish policy-induced changes from developments that would have occurred anyway.
This matters beyond agriculture. Across many areas of environmental and social policy, governments increasingly use financial incentives to encourage desirable behaviour. However, the number of beneficiaries or the amount of money spent is often much easier to observe than the changes those incentives actually produce.
The experience of Hungarian farms illustrates why this distinction matters.
Our findings do not imply that agri-environmental payments should be abandoned. They suggest that policymakers need to be more precise about what these payments are intended to achieve, how those objectives translate into farm-level behaviour and what evidence would demonstrate success.
Ultimately, the effectiveness of environmental policy should be judged not by how many actors receive incentives, but by what changes because those incentives exist.
Environmental Sciences Europe.