The impact of farmers' implementation decisions on environmental effectiveness in the Rural Environment Protection Scheme (REPS)

Abstract

The primary objective of this thesis is to investigate the impact of farmers' implementation decisions on the effectiveness of the Rural Environment Protection Scheme (REPS). Three separate empirical papers are used to achieve this objective. The first two papers focus on the impact of farmers' participation decisions on effectiveness, while the final paper addresses whether allowing farmers a role in how scheme measures were implemented into their individual farm management plans impacted on the effectiveness of the scheme. This thesis makes a significant contribution to evaluation literature for voluntary agri-environmental schemes (AESs). In doing so, it provides imperative information for policymakers involved in the design of future AESs, particularly in relation to the Irish context. REPS had broad environmental objectives that covered both pollution abatement and biodiversity conservation. Therefore, it would have been effective if, firstly, the type of farmer who joined had to reduce pollution outputs to participate in the scheme and, secondly, if a wide variety of habitat types were found in the scheme. The first paper in this thesis aims to investigate whether the type of farmer who participated in REPS from 1995 to 2010 met these criteria. There were four separate phases of REPS. Contracts for all the phases were similar, although a few changes were made in attempts to increase participation rates over time. In particular, payment rates increased and restrictions on organic nitrogen production were reduced. Given these changes, the second objective of the first paper in this thesis is to assess whether the type of farmer who joined REPS improved in terms of the scheme's pollution abatement and biodiversity conservation objectives across the four phases. The first objective of the first paper is met by using a random effects logit model applied to National Farm Survey (NFS) data to look at the type of farmer who joined REPS from 1995 to 2010, inclusive. To achieve the second objective, the type of farmer who was most likely to participate in REPS is estimated for four separate years using NFS data. Each chosen year represents a different REPS phase (REPS I, II, III and IV). Results show that the type of farmer who was most likely to participate in REPS over time had, ceteris paribus, lower income levels and chemical usage than non-REPS farmer. However, the type of farmer in each phase differed substantially. Increases in payment rates from earlier to later phases of the scheme did not appear to improve scheme effectiveness with regard to the type of farmer who joined the latter phases of the scheme, whereas the removal of restrictions on organic nitrogen production did. A difficulty associated with the estimation of AES participation decisions using data for actual participants and non-participants is that the final model suffers from sample selection bias. The first objective of the second paper in this thesis is to address the problem of sample selection bias by estimating farmers' participation functions using actual and counterfactual choice data for all Irish farmers from 1995 to 2010. REPS was universally available to every farmer in Ireland from 1994 to 2009. To be deemed effective, REPS should have ideally attracted as many farmers as possible to the scheme by offering them sufficient, but not too generous, compensation rates for the perceived opportunity costs of joining. A phenomenon in Irish agriculture is that many individuals continue to farm despite their farms being commercially non-viable (Hynes and Hennessy, 2012). Given the difference between viable and non-viable farmers, they are expected to view the REPS participation decision differently. The second objective of the second paper in this thesis is to estimate separate REPS participation functions for viable and non-viable Irish farmers to investigate whether they perceived the choice in the same way. These objectives are met using conditional logit models. The REPS participation functions estimated in the second paper are in agreement with economic theory. Results show that viable and non-viable farmers perceived the REPS participation decision differently. This implies that by attempting to appeal to a heterogeneous population using just one contract, the effectiveness of REPS was reduced. Each participant in REPS was given an individual farm management plan, meaning farmers' input into how scheme measures were implemented on their holdings was permitted. The final paper of the thesis investigates how this impacted on the likelihood that farm habitats were assigned to the correct management options, called biodiversity undertakings (BUs), on their farms. To achieve this goal, habitat types are, firstly, assigned to their ideal BUs with the help of ecological experts. Secondly, two multinomial logits are used to estimate how farmers actually assigned the habitats on their farms to BUs. The first model looks at farmers' BU decisions conditional on their joining REPS while the second includes an Inverse Mills Ratio (IMR) to capture the effect of participation on BU choice. The modelling of the relationship between farm habitat and BU assignment under REPS in the third paper is made possible with the use of new data on farmland habitats, which are geolinked to NFS data for 2007. Use was also made of an additional survey taken at this time on the BU choices made by REPS farmers in the NFS. Finally, a comparison between the optimal choices farmers should have made and their actual choices is carried out. The main finding of this study is that farmers did not make optimal choices for the habitats on their farms although, interestingly, the appropriateness of their choices varies across habitats. Consequently, the findings from this paper show that allowing farmers (and their advisors) a role in how scheme measures were implemented on their farms resulted in a decrease in the effectiveness of the scheme.