Trade-offs in sustainable dairy farming systems
View/ Open
Soteriades2016.docx (1.787Mb)
Date
28/11/2016Author
Soteriades, Andreas Diomedes
Metadata
Abstract
A key challenge facing dairy farming is to meet the increasing demand for dairy
products from a growing and more affluent global population in a period of
unprecedented socio-economic and environmental change. In order to address this
challenge, policies are currently placing emphasis on ‘sustainable intensification’
(SI), i.e. producing ‘more’ outputs and services with ‘less’ resources and
environmental impacts. Determining whether or not SI can deliver greater yet
sustainable dairy production requires understanding of the relationships between
sustainability pillars (environmental; economic; and social) and farm aspects (e.g.
on-farm management; and animal productivity) under particular farming systems and
circumstances (e.g. regional bio-physical conditions). Trade-offs between pillars and
aspects is inevitable within a farming system. Many widely-used assessment
methods that aim to measure, scale and weight these pillars and aspects are unable to
fully capture trade-offs between them. The objectives of this thesis are: 1) to identify
key trade-offs in dairy farming systems to inform greater yet sustainable food
production; and 2) to introduce models and methodologies aiming at a more holistic
measurement and better understanding of dairy farm sustainability.
This thesis assesses the sustainability of French and UK dairy farming systems via a
farm efficiency benchmarking modelling framework coupled with statistical
analyses. It explores the relationships between pillars, aspects and technical,
economic and environmental performance; and identifies important
drivers/differentials in dairy farm efficiency. Importantly, it also suggests ways in
which farm inputs and outputs can be adjusted so that improvements in
environmental, technical and economic performance become feasible.
Efficiency benchmarking was performed with the multiple-input – multiple-output
productive efficiency method Data Envelopment Analysis (DEA). DEA calculates
single aggregated efficiency indices per farm by accounting for several farm inputs
and outputs which the DEA model endogenously scales and weights. In this work,
the notion of farm inputs and outputs was extended to also include ‘undesirable’
outputs (greenhouse gas emissions) and environmental impacts (e.g. eutrophication,
acidification etc.) of dairy farming. The DEA models employed belong to the family
of ‘additive’ models, which have several advantages over ‘traditional’ DEA models.
These include their ability (i) to simultaneously increase outputs and reduce inputs,
undesirable outputs and environmental impacts; (ii) to identify specific sources of
inefficiency. These ‘sources’ represent a farm’s shortfalls in output production and
its excesses in input use and/or in undesirable outputs and environmental impacts,
relatively to the other farms; (iii) to position undesirable outputs in the output set
rather than consider them as inputs or ‘inverse’ outputs; and (iv) to rank farms by
efficiency performance. Importantly, this thesis also proposes a new additive model
with a ranking property and high discriminatory power. In a second stage, DEA was
coupled with partial least squares structural equation modelling (SEM) so as to
develop and relate latent variables for environmental performance, animal
productivity and on-farm management practices.
The results suggested that the efficacy of SI may be compromised by several on-farm
trade-offs between pillars, aspects and farm inputs and outputs. Moreover, trade-offs
depended on particular farming systems and circumstances. Increasing animal
productivity did not always improve farm environmental performance at whole farm-level.
Intensifying production at animal and farm-levels, coupled with high reliance
on external inputs, reduced farm environmental performance in the French case, i.e. a
significant negative relationship was found between intensification and
environmental performance (SEM path coefficients ranged between -0.31 and -0.57,
p < 0.05). Conversely, in the UK case, systems representing animal-level
intensification (via genetic selection) for increased milk fat plus protein production
performed better, on average, than controls of UK average genetic merit for milk fat
plus protein production in terms of technical efficiency (DEA scores between 0.91–
0.92 versus 0.78–0.79) and environmental efficiency (scores between 0.92–0.93
versus 0.80), regardless of whether on a low-forage or high-forage diet. The levels of
inefficiency in (undesirable) outputs, inputs and environmental impacts varied
among farming systems and depended on the regional and managerial characteristics
of each system. For instance, in France, West farms had higher eutrophication
inefficiencies than East farms (average normalized eutrophication inefficiencies
were, respectively 0.141 and 0.107), perhaps because of their more intensive
production practices. However, West farms were more DEA-efficient than East
farms as the former benefited from bio-physical conditions more favourable to dairy
farming (mean DEA score ranks were 97 for West and 83 for East). Such findings
can guide policy incentives for SI in different regions or dairy systems.
The proposed modelling framework significantly contributes to current knowledge
and the search for the best pathways to SI, improves widely-used modelling
approaches, and challenges earlier findings based on less holistic exercises.