Tuning a consistent information flow between models is a main challenge, since huge variation exists across countries. WP6 has created a new definition and characterization of a typical low input dairy farm, intending to be a better anchor for information flow between different modelling levels.

By Jolien Hamerlinck & Ludwig Lauwers, WP6

The evaluation of the competitiveness of sustainable (S) organic (O) and low-input (LI) dairy (D) farms, or of novel strategies within SOLID farming, requires the integration of information that ranges from micro to macro level. Indeed, the economic and sustainability evaluations happen at various levels. At process level, decision support systems for the animal nutrition and environmental monitoring are developed. At whole-farm level, we question how novel strategies will influence the whole-farm planning, resource use and sustainability outcomes. At sector level, we want to estimate the overall impact of strategic changes on the milk supply and resource demand from SOLID farms. Finally, at policy level, we explore whether and how policies can affect changes in strategy adoption and how it can stimulate conversion to competitive SOLID farms. Models are used at each level to provide information for another level. Each has own data needs, so tuning a consistent information flow between them becomes a main challenge.

Defining typical farms as facilitator for consistent information flow

Each modelling level has its own data needs and assumptions to mimic reality. Therefore, it is extremely difficult to transform outcomes from one level to another. Moreover, detailed data to calculate economic and environmental outcomes at process level are not available at the policy level. The modelling tasks become even more difficult when assessing competitiveness of LI dairying.  Earlier SOLID work on LI definition proves that huge variation across countries exists.

Read more: The economic models in the SOLID project

We try to solve this problem with the concept of typical farms. This method is, for example, also used by the International Farm Comparison Network (IFCN) to compare dairy farms around the world. To account for the CAPRI logic of differentiating dairy farming in two technologies, we will define two typical farms for a country, or group of countries. With the typical-farm concept, we aim at following advantages. First, we make the LI definition more robust and suitable for simulations at various levels. Up to now, we developed a pragmatic approach based on median values for defining LI, which already allowed for some interesting economic and resilience assessments, but suffer from bias from yearly price changes and inconsistency in some key measures. Second, the typical farm facilitates communication between experts, or modelers at various levels. So, the concept is useful to integrate different types of information and data sources.

Characterizing typical LI dairy farms

To construct the LI typical farm, and its counterpart the “non” LI farm, we start from the SOLID indicator for LI:

The SOLID LI definition allows for country–specific differentiation. Besides, the typical farms’ construction is based on six other indicators (see box). We developed an algorithm to derive from the FADN data for the years 2004 to 2009, a group of LI farms that are homogenous for the 7 indicators.

Some first results

At this stage of the project, we started with defining typical farms for Finland, Spain, United Kingdom and Belgium, this to validate the outcomes by our colleagues who use ORGPLAN, DREMFIA, PAFAMO and CAPRI. The results from Belgium are given in Table 1. The LI typical farms are differentiated from the non-LI farms because in CAPRI calculations only can happen with two technology variants.

Evidently, LI farms use considerably less inputs than the other farms. LI farms are more specialized than non-LI farms, these farms have also some cash crops and this group uses the forage more intensive than the LI group. However, both groups have a similar number of dairy cows. LI farms have a lower milk production per cow and use the fodder area less intensive than the non-LI farms.

Such comparisons, as demonstrated above for Belgium, differ from country to country. For example, while the number of dairy cows is approximately the same for the two typical farms in Finland, Belgium and Spain, this is not the case in the UK, where the non-LI typical farm has more dairy cows. This results in higher external input cost per ton milk on the LI typical farms.

Milk production per cow is on the LI typical farm of Finland higher than the milk production on non-LI typical farms in the other countries.

Ongoing research with the typical farms

Above description of typical farms is based on the key figures, used for delimitating the homogenous groups behind the LI typical farm. These descriptions will now be enlarged with other FADN data and reasonable assumptions on physical features (of which data are lacking in FADN) concerning, for example, concentrate use, possible ongoing pasture strategies, biodiversity or N balance. Through this type of descriptions, the typical farms intend to be a better anchor for information flow between the different modelling levels. They provide a template for linking reasonable assumptions to the existing description and this will also facilitate the design of novel strategies to competitive sustainability.

AUTHORS

For further details on country specific results of the low input farms, we advise interested readers to read the summary of deliverable 6.1.

By Jolien Hamerlinck, Jo Bijttebier, Ludwig Lauwers and Simon Moakes

Does the low input European dairy farm exist? Based on a detailed review of existing literature on classification of farming systems and an explorative study of these methodologies on UK and Belgian data, it was concluded that a general definition of the European LI dairy farm was difficult to develop.

One of the main causes was the variety of farming systems in Europe. To fully elaborate competitiveness issues within and between the farming systems across Europe, LI farms were defined for each country as those farms with the lowest 25% expenditure on inputs for that country. The inputs taken into account to identify LI farming systems, were the
costs for fertilizers, crop protection, purchased feed for the ruminants and energy, expressed as € per grazing livestock unit. A first study of the FADN database (2007-2008), revealed that LI farms were smaller, with fewer animals, a slightly higher family
labor percentage and lower milk yields. Besides these structural differences, LI farms were less profitable than other holdings, but also received lower support payments (see summary of deliverable 6.1 for more information).

Country specific differences in performance of high input (HI) and LI dairy farms
With the pragmatic country-specific definition of low input dairy farms as a tool, we can differentiate farms in each country into high versus low input farms. Figure 1 represents the average economic profit per annual working unit (AWU) for dairy farms across Europe for the years 2007-2008 (with direct input costs along the x-axis). The relative economic
profit of the farms in the different countries is compared to the economic profit of the EU 27 HI farm, e.g. the profit of a dairy farm in Denmark is more than 7 times greater than that of the EU average. In the figure, each country is represented by two dots, interconnected with a line. The left dot represents the average profit of the LI farms of that country while the right dot illustrates the average profit of the HI farm.

The length of the line indicates the relative difference in input expenditure between HI and LI; the slope indicates the difference in profit: A downward slope indicating that LI holdings perform better than HI, (which is strongly pronounced in Finland, but is also the case in Spain and Ireland) i.e. additional inputs have resulted in lower profitability. In several other countries however the line slopes upward; HI farms have higher economic
profits compared to the LI farms. The position in the figure demonstrates very clearly the variety in farm size of the different farming systems within Europe. The immediate expenditures for a LI farm in Denmark, for example, are 10 times higher than those for a LI farm in Italy, reflecting the variation in dairy systems across the EU. These data reveal further insights on the real behavior of LI farms: some belong indeed to another farming system, while other LI farms, like in the Netherlands and Denmark, may still belong to a similar production system but are more efficient than the HI farms in their country.

Are LI farms more resistant to future volatility of milk and feed prices compared to HI farms? 

Due to current and future tendency of high volatility of milk, feed and energy prices, the resilience of LI and HI dairy farms to this volatility is of interest.

Figure 2 illustrates this volatility for milk prices in Europe and Belgium. Two sensitivity analyses have been undertaken based on two observations in the milk price evolution.

First, the average milk price during a longer period (2007-2012) is lower compared to that of the period 2007-2008, though the recent trend is again upwards. Secondly, there are significant changes, illustrated by the high milk price in 2008 and the pronounced decline
in the following year 2009.

For 2007 – 2012, milk prices declined by approximately 5.5% and yet feed prices increased by about 3% in comparison with the period 2007-2008. In 2009, milk prices
were very low and declined by 30% while feed prices only declined by 13 %. Based on these figures, we designed a trend and a shock scenario, (based upon FADN data results from 2007-2008) to simulate the effect of both scenarios. Table 1 summarizes the average economic performances of LI and HI dairy farms in EU27 in 2007-2008 compared to the scenario simulations of reduced longer term prices and a price drop shock scenario.

The results show that LI farms are more resilient towards price fluctuations than HI farms. Where HI farms had a higher economic profit in 2007-2008, they have a lower
income when assuming trend conditions and were more affected by extremely low prices as those observed in 2009. These results are confirmed by the country specific data (Figure 3). When prices decline, either in the long or short term, the economic advantage of HI farms decreases in these countries where HI farms perform better and in countries where LI farms perform better; this comparative advantage increases when prices decline.

Conclusions
Earlier analysis of LI farms across Europe revealed lower profitability of the LI farms compared to the high input ones. However, although this tendency can be extended to several European countries, the opposite is true for some countries; LI farms perform better than HI farms. Although their lower use of inputs produces less output, lower inputs may result in increased efficiency in the use of fertilizer, crop protection, feed, and energy on these farms. Moreover, in all European countries, LI farms seem to be more resistant to price fluctuations, which become more and more important in the post quota era, and may be of particular relevance to family farms where reduced income fluctuation is as important as absolute profit.

MACSUR gathers the excellence of existing research in livestock, crop, and trade science
to describe how climate variability and change will affect regional farming systems and food
production in Europe in the near and the far future and the associated risks and opportunities for European food security. A knowledge hub is an innovative, tailor-made instrument developed by FACCE-JPI, associating 3 complementary dimensions: networking, research and capacity building.

The knowledge hub consists of 73 partners from 17 countries all over Europe and Israel
including Aberystwyth University and Aarhus University. The overarching challenge is to
develop a pan-European capability in the development, use and interpretation of models
to perform risk assessments of the impacts of climate change on European agriculture.
The project focusses on the technical and informational integration of suitable existing
models and their application in regional case studies that reflect the European diversity in
soil, climate, socio-economy and agricultural systems.

To address this the following challenges must be met:

1. Identify and address a range of issues between models in different themes to enable
their closer integration including issues of scale and data processing.

2. Train a new generation of scientists to work across models which contribute to greater
integration of models. This challenge can be described as focusing on the creation of
integrated modellers as opposed to integrated models.

3. Determine the contribution that can be made to reducing uncertainty over the impacts
of climate change on European food security by adopting integrated models of crop
production, animal production and trade.

MACSUR plans close co-operation with other international research networks like AgMIP
and the involvement of political stakeholders. Expected outcomes of the project is a procedure for integrating models, assessment of food security in European case studies, and an assessment of how uncertainty in food security could be reduced further.

Read more here.

By Simon Moakes, Institute of Biological, Environmental and Rural Sciences (IBERS), Aberystwyth University, Wales

To compare the economic and physical structure of organic, low input and conventional dairy farms we are using European farm economic data (FADN). This will allow us to report the size, profitability and other characteristics of the sector, and to identify the best performing holdings. Within the data, organic and conventional farms are clearly identified, but before any analysis could be undertaken it was necessary to develop a classification to define low input dairy farms. 

A more specific classification was needed
We required a method to allow the comparison of low-input dairy systems with organic and conventional farms that could be used within the data constraints of variables available within the FADN database. An initial review of previous work identified that previous projects such as IRENA, SEAMLESS and CEAS had tried to classify farms systems according to their intensity or type of dairy system. After testing these methodologies with national data in UK (ABER) and Belgium (ILVO) and following feedback from project partners and stakeholders, it was clear that although useful that
a more specific classification was needed for use within SOLID.

The cut-off value should be higher
The IRENA project used costs for fertiliser, crop protection and purchased concentrate feed per hectare as an indication of farming intensity (and potentially therefore
environmental impact), but the cut-off value for defining low input (<€80/ha) were very low compared to the values for EU dairying. Therefore it was felt that the cut-off value should be higher and that additional inputs, such as energy costs and other purchased feed should also be included. Following further testing and consultation it was agreed that the low input definition would include costs for fertiliser, crop protection, purchased feed for ruminants and energy costs (e.g. fuel and electricity).

Inputs per grazing livestock unit
These input costs would need dividing by the number of cows, sheep/goats, livestock units (LU) or by area e.g. hectares. Due to a lack of area data for some countries due to common or shared grazing, it was decided that inputs per grazing livestock unit (GLU) would be the preferable denominator. The cut-off values specified by IRENA were very low compared
to the values of most dairy holdings, therefore it was also decided that holdings in the bottom 25% of input expenditure would be classified as LI. The 25% cut-off value for all EU dairy holdings was calculated at €419/GLU for dairy cow holdings and at €359/GLU for sheep/goat holdings. Values were also calculated per country as there was large variability in input costs, as shown in Figure A below.

Dairy holdings were split into three groups for the analysis; organic, low input and conventional systems. Initial results indicating the structure of dairy farming in the EU is shown in Figure B below, and highlights that low input classified holdings were smaller, with less livestock, a slightly higher family labour percentage and lower milk yields.

LI classified holdings had lower output
Figure C indicates that LI classified holdings had lower output and despite lower costs, their profitability was almost half that of than organic and conventional farms.

Comparison of different classes
Following further statistical analysis the different classes can be compared for efficiency of input use for both physical yield e.g. milk yield per head or financial performance e.g. profit per cow or worker unit. Figure D below shows increasing net margin (NM) per kg of
milk against increasing inputs (fertiliser, crop protection, feed and energy) for German dairy cow holdings (LI holdings – green, conventional – blue and organic – cream). The figure indicates that organic systems achieved the highest NM and that despite lower
overall farm profitability LI systems resulted in high NM per milk kg. It also indicates the large variation in performance with some holdings in each class losing money for every kg of milk produced.

Conclusion and next step
In conclusion, it was possible to identify low input conventional dairy holdings
through their expenditure on a range of inputs, to allow us to compare their
performance against other farm groups. Initial analysis at EU level indicates LI
system profitability is below that of organic and conventional systems, but that efficiency may be higher.

The next step is to identify the characteristics of more profitable holdings and efficiency analysis will highlight optimum input output values that achieve maximum financial
returns. Economic analysis will then be discussed with environmental results to provide an integrated policy output on optimising the sustainability of EU dairy systems.

The Finnish DREMFIA model describes the agricultural supply-demand conditions at the regio­nal or national level. This sector model brings the essential farm-level responses to the market level.

At farm level, we can use ORGPLAN, a farm management tool designed to assess the effects of farm conversion to an organic system. The software is adapted for evaluating the transition from a medium or high input dairy system to a low input or organic, forage-based system.

PAFAMO assesses impact of strategic choice variables in a programming mo­del where resources (capital, land, labor) are scarce. It wants to be a tool to support farmers in assessing the impact of their strategic choices.

The more detailed process models (such as those in WP3 and WP4) must serve as input for PAFAMO and ORGPLAN,  which then will provide exogenous information to sector and policy models.

To guarantee a fluent up-scaling of results from process to macro levels, the concept of typical farms is proposed.