And Farm-scale Model Of Arable, Forestry, And Silvoarable Economics

Preparing to load PDF file. please wait...

0 of 0
And Farm-scale Model Of Arable, Forestry, And Silvoarable Economics

Transcript Of And Farm-scale Model Of Arable, Forestry, And Silvoarable Economics

Agroforestry Systems, Volume 81, Number 2, 2011, 93-108.
Pre-print copy of paper: Graves AR, Burgess PJ, Liagre, F., Terreaux, J.-P., Borrel, T., Dupraz, C., Palma J. & Herzog, F. (2011) Farm-SAFE: the process of developing a plotand farm-scale model of arable, forestry, and silvoarable economics. Agroforestry Systems 81: 93-108.
DOI: 10.1007/s10457-010-9363-2
Farm-SAFE: the process of developing a plot- and farm-scale model of arable, forestry, and silvoarable economics
A.R. Graves1, P. J. Burgess1, F. Liagre2, J-P. Terreaux3, T. Borrel4, C. Dupraz4, J. Palma5 and F. Herzog6 1 Cranfield University, Cranfield, Bedfordshire MK43 0AL, UK 2 Assemblée Permanente des Chambres d’Agriculture, 9 Avenue Georges V, 75008 Paris, France 3 Cemagref, 361, Rue J.F. Breton - BP 5095 - 34196 Montpellier Cedex 5, France. 4 Institut National de la Recherche Agronomique, 2 Place Viala, 34060 Montpellier, France 5Technical University of Lisbon, Tapada da Ajuda, 1349-017 Lisboa, Portugal 6Agroscope Reckenholz-Tanikon Research Station ART, Reckenholzstrasse 191, 8046 Zurich, Switzerland
Full address for correspondence: A.R. Graves, Cranfield University, Cranfield, Bedfordshire, MK43 0AL, UK
E-mail address: [email protected]; [email protected]
Key words: Cost-benefit analysis, net present value, economic analysis, economic model, equivalent annual value
Financial feasibility and financial return are two key issues that farmers and land owners consider when deciding between alternative land uses such as arable farming, forestry and agroforestry. Moreover regional variations in yields, prices and government grants mean that the relative revenue and cost of such systems can vary substantially within Europe. To aid our understanding of these variations, the European Commission sponsored a research project called “Silvoarable Agroforestry For Europe” (SAFE). This paper describes the process of developing a new economic model within that project. The initial stages included establishing criteria for the model with end-users and reviewing the literature and existing models. This indicated that the economic model needed to allow comparison of arable farming, forestry and agroforestry systems at a plot- and a farm-scale. The form of comparisons included net margins, net present values, infinite net present values, equivalent annual values, and labour requirements. It was decided that the model would operate in a spreadsheet format, and the effect of phased planting patterns would be included at a farmscale. Following initial development, additional user feedback led to a final choice on a model name, a final method of collating input data, and the inclusion of field-based operations such as varying the cropped area, replacing dead trees, and pruning. In addition options in terms of improved graphical outputs and the ability to undertake sensitivity analysis were developed. Some of the key lessons learnt include the need to establish clear model criteria and the benefits of developing a working prototype at an early stage to gain user-feedback.

Increased population and increased consumption of natural resources per capita are placing increased demands on finite land resources. The ecosystem approach, popularised by the Millennium Ecosystem Assessment (2005), provides one framework for identifying the importance that different stakeholders place on the goods and services that we get from land (Agbenyega et al. 2009). However decisions regarding land use change are still primarily taken by individual land-owners, and profitability is a key consideration (Graves et al. 2009).
During the past 50 years, one of the significant land use changes across the European Union (EU) has been the removal of individual trees from agricultural land, and conversely the reestablishment of trees on agricultural land in woodland blocks. This has been partly a result of the increased mechanisation of agriculture and the availability of EU-related grants for woodland planting. One alternative method for re-establishing trees within an agricultural system is silvoarable agroforestry (Dupraz and Newman 1997; Burgess et al. 2004). It is only recently that the establishment of such systems has been supported by grants associated with the EU Rural Development Regulation 1698/2005 and, the grants are only available in some EU countries. However the decision to establish a silvoarable agroforestry system can be complex because the financial return from the tree component (in the absence of grants) can take many years, and the effect of the trees on crop yields can vary with time. Moreover the likely response will vary with tree spacing and tree species and the grants available can vary substantially between countries.
One method for determining the profitability and feasibility of silvoarable systems, relative to arable and forestry systems, is to use computer-based models (Graves et al. 2005). Although there is some literature describing the results and analyses obtained from using computer models of arable, forestry and silvoarable economics (Wojtkowski et al 1990; Thomas 1991; Willis et al. 1993; Dupraz et al. 1995; Nelson and Cramb 1998), there is less information describing the development of the models. The development of a dynamic computer-based simulation of silvoarable economics is time-consuming, and not aided by the paucity of documentation on existing computer models. This paper aims to describe the process of development of an economic model, called Farm-SAFE, and discusses some of the key lessons learnt.
Within this paper, the development of an economic model of arable, forestry and agroforestry systems is described as a sequential process with significant feedback loops (Figure 1). The first action was to establish the purpose and desired features of the model with the key endusers. This was completed at the same time as a review of existing models. The third step was to develop an initial model, which was then modified in response to additional feedback through model use.


1.Model developer establishes criteria for model with end-users
2. Model developer reviews existing models and literature

3. Model developer creates new working version of model

4. Model developer uses model with endusers

Figure 1. Schematic diagram showing how the criteria for the model, the development of the model, and its use is affected by feedback from the end-users of the model.
Establishing the criteria for the model The primary purpose of the model was to address a research objective of an EU-sponsored project called “Silvoarable Agroforestry for Europe” which started in August 2001 and was completed in January 2005. The specific aims of the project included reducing the uncertainties concerning the viability of silvoarable systems and the extrapolation of plotscale results to individual farms (Dupraz et al. 2005). The criteria for the model were agreed at a workshop meeting including researchers and end-users in September 2002. The principal end-users were researchers and extension advisors in a range of countries including France, the Netherlands, Spain and Switzerland. The agreed criteria for the model are categorised in Table 1 using the model characteristics described by Graves et al. (2005). The criteria are categorised under the headings of: 1) model background, 2) systems modelled, 3) objective of the economic analysis, 4) viewpoint of the analysis, 5) spatial scale, 6) temporal scale, 7) generation and use of biophysical data, 8) model platform and interface, and 9) input requirements and outputs generated.


Table 1. Criteria established for the economic model in September 2002, categorised using the framework described by Graves et al. (2005).

Characteristic 1. Background 2. Systems modelled
3. Objectives of economic analysis
4. Viewpoint of analysis 5. Spatial scale
6. Temporal scale 7. Generation and use of
biophysical data 8. Platform and
interface 9. Inputs and outputs

Criteria for the economic model. The model should be able: 1.1 To operate in English 1.2 To be initially designed and used as a research tool 1.3 To operate as a “closed” format model 2.1 To model silvoarable, arable, and forestry systems 2.2 To model coincident and spatially-zoned silvoarable systems 2.3 To model crop rotations 2.4 To model multi-planting schemes 3.1 To use a common conceptual framework of farm economics including net
margins 3.2 To account for the effect of time on the value of money by discounting 3.3 To compare the profitability of the systems. Discounted future benefits and
costs of each system should be aggregated and a net present value, infinite net present value, and equivalent annual value calculated. 3.4 To determine the feasibility of the systems. Discounted future benefits and costs of all farm systems should be aggregated and a net present value, infinite net present value, and equivalent annual value calculated. 3.5 To examine the sensitivity of each system to changes in input values 4.1 To simulate the view-point at a micro-economic scale, from the perspective of a single farmer 5.1 To operate at a one-hectare scale 5.2 To operate at a farm scale. Variation in land heterogeneity and enterprise diversity should be accounted for using four land units, each capable of simulating one or more of an arable, forestry, and silvoarable system. 5.3 To “establish” different areas of forestry and silvoarable systems in different years 6.1 To use a yearly time-step 6.2 To use a maximum rotation of 60 years 7.1 To initially be a stand-alone model capable of using annual crop and tree yield data from an external source. 8.1 To be a spreadsheet „workbook” model, using an available and inexpensive modelling platform 8.2 To use a direct interface to make it easily transferable between different language versions of the software 9.1 To reduce input requirements by storing key parameters 9.2. To use databases to store key physical and financial data 9.3 To produce both tabular and graphical output

Review existing models and literature At the same time as establishing the criteria for the model, we reviewed existing computer models of silvoarable economics. The principal models examined were the Agroforestry Calculator (Agriculture Western Australia and Campbell White and Associates Pty Ltd, 2000), the Agroforestry Estate Model (Knowles and Middlemiss, 1999), POPMOD (Thomas, 1991), ARBUSTRA (Liagre, 1997) and the Water Nutrients and Light Capture in Agroforestry Systems model (WaNuLCAS) (Van Noordwijk and Lusiana, 1999, 2000, 2003). Using the criteria described in Table 1, it was possible to characterise the available models and these results have been described by Graves et al. (2005).
Initial development Based on existing plot- and farm-scale models Although it would have been possible to start from scratch, our philosophy was to build on existing models. Of the available models, it was eventually decided to use POPMOD (Thomas, 1991) and ARBUSTRA (Liagre, 1997) as a basis for new economic model.

POPMOD provides an empirical model of tree and crop yields to inform the economics of arable, silvoarable and poplar forestry systems at a one-hectare scale. The ARBUSTRA model, whilst lacking an empirical model of tree and crop yields, allows analysis of different combinations of agriculture/agroforestry/forestry systems within a “farm-level analysis” (Figure 2). These farm-scale features allow the analysis of the effect of different planting patterns and an assessment of the feasibility of introducing new systems in terms of capital and labour requirements (Table 2). In addition the project team had free access to and experience of using both these models and there were no copyright issues. Even so, integrating the two models involved substantial translation issues as the POPMOD model was developed in English and the ARBUSTRA model was developed in French.

Figure 2. A schematic representation of different spatial scales of modelling, the one-hectare, unit and farm scale. A one-hectare scale analysis may be used for unit-scale analysis which in turn may be used for farm scale analysis.

Table 2. Some differences between one-hectare and farm-scale modelling.

One-hectare scale modelling  Useful for comparing farm enterprises
 Comparison on a per unit area basis  Usually a single planting and clear-felling date for
forestry and agroforestry  Spatial heterogeneity not represented  Analysis is based on partial budgets of competing

Farm scale modelling  Useful for comparing farm profitability and labour
use with and without a specified enterprise  Comparison over a user-defined area of land  Several forestry and agroforestry planting and clear-
felling dates may be defined  Spatial heterogeneity represented  Analysis can include farm fixed costs


Choice of modelling platform In theory it was possible to develop the model within a spreadsheet, a database, a programming language, or a graphical development environment, such as Stella™ (Systems Thinking Software™, 2005) or ModelMaker™ (ModelKinetix®, 2005). Because the model was to be used by research and extension organisations in different countries, the platform needed to be readily available and/or inexpensive. It was also important that users could operate and modify the model themselves. Hence, it seemed optimal to use a spreadsheet platform, and specific software chosen was Microsoft ® Excel. This choice was also coloured by the fact that the chosen POPMOD and ARBUSTRA model were also spreadsheet based. The decision to use a spreadsheet platform focussed on spreadsheet cell functions enabled the use of the model in different language versions (e.g. English, French, German, Italian, and Spanish). Another advantage of using a widely-used spreadsheet programme was the availability of add-on applications such as Crystal Ball® Risk Analysis Software and Solutions (Decisioneering® Incorporated, 2005), and Insight.xla 2.0, developed by AnalyCorp® (Savage, 2003). These could be used to help in the optimization and uncertainty analysis.

Form of field-scale economic analysis
Various conceptual models of farm economics have been used depending on the
circumstances and objectives of the analysis. Within the model, economic analyses were
initially undertaken at a one-hectare scale (Figure 2). The financial value of each enterprise was calculated in terms of a net margin (units: € ha-1) determined as the revenue (R; units: € ha-1) minus variable costs (V; units: € ha-1) such as seed, fertilisers and sprays, and the „assignable fixed costs‟ of labour and machinery associated with the enterprise (A; units: € ha1) (Equation 1). A similar approach when comparing arable and forestry systems has
previously been used by Willis et al. (1993) and Burgess et al. (1999).

Net margin = R  V  A

Equation 1

When comparing arable and forestry systems, whereas the costs and revenue from arable systems take place within a 12-month period, the timber revenue from trees can occur many years after the costs of establishment. Since most people have a preference for immediate income, there is an opportunity cost to immobilizing capital in long-term projects. Within the model, future benefits and costs were therefore reduced or “discounted” using the approach developed by Faustmann (1849) to value forestry investments. The net present value (NPV; units: € ha-1) was therefore calculated using Equation 2 where the revenue (Rt), variable costs (Vt), and assignable fixed costs (At) are specified for each year (t) over a time horizon of T (years), and i is the discount rate (Equation 2):

 NPV  tT (Rt  Vt  At ) t0 (1  i)t

Equation 2

In addition in order to compare systems including tree species with different rotations, the model was also developed to calculate an infinite NPV (NPVInfinite; units: € ha-1). This is the
NPV defined over an infinite time horizon, in which each replication has a rotation of n years.
It is defined (Equation 3) as:

(1  i)n

NPVInfinite  NPV


(1  i) 1

Equation 3

The NPV was also expressed as an annuity, termed the “equivalent annual value” (EAV; units: € ha-1 a-1) (Equation 4):

EAV = NPVInfinite  i

Equation 4

Structure of the model The philosophy in building the model was to develop distinct worksheets to contain the key components of the economic analysis (Figure 3; Table 3). The primary worksheet was called “Option and results” and this functioned as the control worksheet for selecting the appropriate inputs and presenting the results. This structure also simplified loading and simulation of different scenarios because each scenario could be saved, rather than needing manual input each time it was used. The input worksheets comprised three physical yield templates labelled “Arablesystem”, “Forestrysystem”, and “Agroforestrysystem”. There were also four financial templates labelled “Arablefinance”, “Treevalue”, “Treegrant”, and “Treecost”. The inputs required in these worksheets are provided in the appendix to this paper.

Options and r esults

Sys tem selection

Planting calander

Economic options

Sensiti vity anal ysis

One-hectare results

Unit-scale results

Farm-scale results

Arablesystem Arablefinance Agroforestrysystem Crop optimisation
Treevalue Treegrant Treeco st

Plot 1-4 Calculation of arable economics
Calculation of agroforestr y economics
Calculation of forestr y

Unit 1-4
Calculation of arable
Calculation of agroforestr y economics
Calculation of forestr y

Calculation of arable
Calculation of agroforestr y economics
Calculation of forestr y


Production and LER

Plot scal e produc tion

Land equi val ent ratio

Graphic resu lts Plot scal e results

Unit scale results

Farm-scale results

Figure 3. Schematic representation of the SAFE economic model. Each box represents a separate worksheet within the Microsoft Excel workbook.


Table 3. Worksheets within the SAFE economic model.

Worksheet name

Worksheet function

Data manipulation

“Options and results”

Allows selection of data stored in “Arablesystem”, “Arablefinance”,

“Agroforestrysystem”, “Treesystem”, “Treecost”, “Treevalue” and “Treegrant”.

Allows selection of analytical criteria (e.g. discount rate and rotation length)

Data storage


Stores production data for arable systems


Stores production data for agroforestry systems


Stores production data for forestry systems


Stores data on the prices, grants and costs associated with arable systems and

the crop component of agroforestry systems


Stores data on the prices of tree products


Stores data on the grant systems associated with trees


Stores data on the costs associated with forestry systems and the tree component

of agroforestry systems

Data modelling

“Crop optimisation”

For plots 1 – 4, calculates the optimal rotation of the crop component of the

silvoarable system

“Plot 1”, “Plot 2”, “Plot 3”, For four plots 1 – 4, models one-hectare-scale economics and labour

“Plot 4”

requirements of arable, forestry and silvoarable systems

“Unit 1”, “Unit 2” “Unit 3”, For four land units 1 – 4, models unit-scale economics, labour, and land use

“Unit 4”

requirements of arable, forestry and silvoarable systems.


Models farm-scale economics of arable, forestry and silvoarable systems at the

farm scale

Data manipulation and results

“Options and results”

Stores production and economic one-hectare-, unit- and farm-scale results, in

numerical form as tabular data, for the final year of the rotation

“Production and LER”

Stores one-hectare-scale production and land equivalent ratios in graphical form

for the duration of the rotation

“Graphic results”

Stores production and economic one-hectare-, unit- and farm-scale results, in

graphical form for the duration of the rotation

Form of farm-scale economic analysis In order to allow an analysis of the effects of introducing agroforestry or forestry systems at a farm-level, it was assumed that most farms can be described in terms of up to four land units, which each unit representing a given level of productivity. The user was required to specify the area of each land unit (alu; units: ha) and this was assumed to remain constant. For example a farm may comprise one land unit of 50 ha of sandy soil and a second land unit of 100 ha of a clay soil.
In a simple comparison of forestry or silvoarable enterprises a single planting year can be assumed. However if all the planting on a large farm took place in a single year, this could cause serious disruptions to farm cash-flow and the demand for labour. Hence the economic model was designed to allow the analysis of phased planting schemes where the user could specify that a certain area or proportion of land was planted to forestry and/or silvoarable agroforestry in each year. In any particularly year (t), new areas of forestry (anewfor: units: ha) and silvoarable agroforestry (anewsil; units: ha) could be planted assuming that the total did not exceed the total area of the land unit (alu). As the rotation proceeds, forestry (afellfor: units: ha) and silvoarable (afellsil; units: ha) plots may also be “clear-felled” in each year. The area of forestry (afor: units: ha) (Equation 5) and silvoarable agroforestry (asil; units: ha) (Equation 6) plots in year t is therefore obtained by adding the area of new planting and subtracting the areas of clear-felled systems.

Equation 5
Equation 6
Lastly the revenue and costs of up to four units were aggregated in a worksheet labelled “Farm” which also included the fixed costs of the farm (F; units: € farm-1). Thus, the NPV of the farm (NPVfarm; units: € farm-1) (Equation 7) can be expressed as:
Equation 7 Where: l is one of four possible land units, Nar, Nfor, and Nsil is the net margin (€ ha-1) of the arable, forestry and silvoarable enterprises respectively in each land unit l in year t. The other inputs include , , and as the area (ha) of the arable, forestry, and silvoarable systems respectively in each land unit l in year t, Ft is the farm fixed cost in year t (€ farm-1), and T is the time horizon (years). A farm infinite NPV (€ farm-1) and a farm EAV (€ farm-1 a1) were also calculated with Equations 3 and 4 respectively.
The results for the one-hectare-, unit- and farm-scale calculations of timber and crop production, undiscounted and discounted cash flows, land and labour requirements were tabulated as single numerical totals for the final year of the rotation in the “Options and results” worksheet along with other criterion such as the NPV, infinite NPV, and EAV (Table 3).
Feedback from using the model An initial version of the plot-scale economic model was developed within the first twelve months of the project, and a farm-scale model was developed soon after. However the model continued to be developed through the project by an iterative process of use and refinement. This was greatly aided by the development of a project website, which was used to store the project outputs and provided a forum where discussion of all aspects of the project could take place. For example, the initial version of the model with a description of the model and sample exercises was placed on the project website, so that project members could use the model and provide feedback. During the project, some of the key issues included the naming of the model, the generation and/or collation of the physical and economic data for the model, the inclusion of specific field operations, and improved ways of presenting the outputs (Table 4).

Table 4. Additional feedback provided which led to additional features within the model.

Naming Input
Field-based operations
Output analysis

Problem How do you distinguish between multiple models? How do you minimise data entry requirements?
National differences in the subsidy regime Can you stop the crop rotation when no longer profitable? Can you include the effect of establishing a grass sward? Can you vary the cropped area during a rotation? Can you include the effect of poor tree establishment? Can you include the effects of pruning? Can you illustrate the results? Can you determine the sensitivity of different inputs? Can you model one-off changes in prices in a future year?
Can you allow for incremental changes in prices and costs from a given future year?

Solution Provide discrete model names
Use “identifiers” for default datasets
Collect “default” data for specific systems Include a range of grant options
Include a feedback loop to stop cropping when unprofitable Include effect of creating a grass sward
Include the proportion of cropped land
Include the impact of replacing dead trees
Include pruning and pruning labour model Include graphs of key outputs Include spreadsheet routines which allow changes in key inputs Include spreadsheet routines to specify year and degree of one-off change in prices and costs Include spreadsheet routines to specify year and degree of incremental change in prices and costs

Naming the model and identifying its role in a family of models A key activity within any modelling project is identifying a suitable name for the model(s). A number of models were developed within the SAFE project and it was decided that the model name should make reference to the overall project. It was finally agreed that the plot- and farm-scale economic model would be called Farm-SAFE (Figure 4).

Biophysical analysis Hi-SAFE Yield-SAFE

Plot-scale Economic analysis

Farm-scale Economic analysis

Figure 4. Schematic diagram showing the relationship between the two biophysical models, one bioeconomic model and an economic model.