Quantile methods

• an additional euro of support has a stronger stabilising effect on farms at the bottom of the income distribution (lower quantiles) compared to high-income farms (upper quantiles);
• productivity gains from investment subsidies are concentrated among already high-performing farms or are more evenly distributed; and
• direct payments reduce income volatility primarily for smaller or lower-income farms.

The two main tools are:

• QDRF examines how different levels (‘doses’) of support influence different types of farms along the distribution (e.g. low‑income vs. high‑income). The focus is on the variation in the amount of support.

  Example: Imagine farms receive different amounts of investment support: EUR 0, EUR 5 000, EUR 10 000, EUR 20 000. The QDRF tells you, for example:

  • At the 10th percentile (poorest farms): increasing support from EUR 5 000 to EUR 10 000 increases income by X%.
  • At the 90th percentile (wealthiest farms): the same increase in support has almost no effect.

• QCTE  estimates how the treatment effect varies across quantiles, conditional on receiving the treatment. It compares treated vs. untreated farms at each quantile. The focus is on the variation in the effect of being in the treatment group (binary treatment).

  Example: Suppose some farms receive investment support (treatment = 1) and others do not (treatment = 0). The QCTE estimates:

  • At the 25th percentile: receiving support increases income by EUR 2 000.
  • At the 75th percentile: receiving support increases income by EUR 400.

These methods show how CAP support affects farms with very different characteristics or starting conditions, rather than only the ‘typical’ or average farm.

Key assumptions include:

• The relationship between CAP support and outcomes may differ across the distribution (heterogeneous effects).
• The statistical model must be correctly specified for each quantile.
• Sufficient data must be available across all relevant parts of the distribution to produce reliable estimates.

• The policy question is explicitly about distributional impacts (e.g. does support reduce income inequality? who gains most from a measure?).
• Support levels vary across farms, and you suspect that effects differ across farm types or income levels.
• Support is broadly provided to (almost) the entire farm population (e.g. decoupled direct payments), so there is no clear untreated control group, and you want to exploit variation in the level of support rather than a treated/untreated comparison;
• There are large micro‑datasets (e.g. FADN/FSDN) and the assessment wants to go beyond ‘one average effect’.
• They are a good complement to average‑effect methods (e.g. DiD) when equity or targeting issues are central.

Preconditions

• Large cross‑section or panel datasets (many farms, several years) to obtain reliable estimates at different quantiles.
• Continuous or at least multi‑valued treatment (e.g. varying levels of support, not only 0/1).
• Access to strong econometric skills and software, as quantile methods are more complex than standard methods.

When to use quantile methods in the context of CAP Strategic Plan assessments

Quantile methods can be used to analyse how CAP support affects not only the average farm but also the entire distribution of outcomes. 

Here, the ‘distribution of outcomes’ refers to the distribution of the dependent variable (e.g. farm income), but quantile methods can be applied separately or interactively for different groups of farms (e.g. by farm size, type or region) to explore whether distributional effects differ across these categories.

They are suitable for studying how support influences:

• Farm income or entrepreneurial income to see whether lower‑income farms gain proportionally more (or less) than higher‑income ones.
• Agricultural factor income to examine distributional changes across different types of farms.
• Productivity indicators (e.g. total factor productivity) to assess whether support primarily benefits already high‑performing farms or helps lower‑performing farms catch up.
• Environmental or social indicators, provided they are measured at farm level and distributional patterns are relevant, e.g. how support affects low‑ vs. high‑emission farms, or farms with low vs. high employment intensity.

These methods are especially valuable when it is important to understand how outcomes differ across farms, not just the average effect. This is particularly relevant to policy objectives such as fairness (equity), the stability of vulnerable farms (resilience) and targeted support. They help evaluators understand whether CAP measures reduce disparities, reinforce existing gaps or produce heterogeneous effects that may be hidden when looking only at average impacts.

• Quantile methods reveal how CAP support affects different types of farms across the entire distribution – not only the average farm. They help uncover whether low‑income, low‑productivity or otherwise vulnerable farms benefit differently from support compared with higher‑performing farms. This provides a much richer picture than average effects alone. 
• They are especially valuable for evaluating equity, redistribution, and targeting aspects of the CAP. These methods directly answer questions such as ‘Who benefits the most?’, ‘Does support reduce disparities among farms?’ or ‘Are vulnerable groups catching up or falling behind?’ – questions that are central to many CAP objectives.
• They require large datasets and advanced statistical expertise, so they are best used by experienced analysts as a complement to simpler methods. In practice, quantile analysis is most effective when evaluators have access to detailed micro‑data and the capacity to estimate models separately across the distribution. When used alongside more conventional approaches, they add important insights that would otherwise remain hidden.