TRADITIONAL SPANISH FARMING MODEL
Contents of this information file:
1. Model Details
2. Model Usage
3. Model Display
4. Model Rationale

1. MODEL DETAILS
Traditional Spanish Farming Model created with NetLogo

View/download model file: SpanishFarmModel_applet.nlogo

Example Market Scenario Files:
farmCost
marketcostC
marketcostP
marketpricesC
marketpricesP

Please email me if you require more details/info.

Example Model Movies:
Run1

2. MODEL USAGE
This version (V3) of the model now allows the user to decide at setup of a run whether to generate a new "random" landscape (specifying certain parameters such as spatial distribution of land capability, number of farmers etc.) or whether a landscape should be read from file.

Before clicking setup, the user should select either "Random" or "From File".
- "Random" If setup is random, all usual parameters should be specified.
- "From File" If setup is from file, the appropriate files (created using CreateLandscapes.nlogo) should be located in the same directory as the model environment. Using this option in conjunction with the Behaviour Space will allow multiple runs for the same landscape setup to be run. See the Information tab of CreateLandscapes.nlogo for more details.

IMPORTANT NOTE: The display window must be set to the same dimensions as the landscape files being imported from file (Farmers.nldat does not specify a landscape size). The dimensions of the landscape file are denoted on lines 2 (screen-size-x) and 3 (screen-size-y) of the .nldat files. For example, if the landscape file has dimensions of 21 x 21, the view window should be set (click edit) to "Screen Edge X" = 10, "Screen Edge Y" = 10. If the dimensions do not match, an error message will be displayed in the Command Centre and the files will not be loaded.

3. MODEL DISPLAY
Model display options have been changed in this version. Three different types of map may be viewed in the display window using the DisplayMap chooser and the Reload Display button. Select Cover-Owner, LandCap or Roads from the DisplayMap chooser, specify whether patch cover labels should be shown, then click the Reload Display button to view.

The monitors on the right of the display window detail;
- number of farmers in the landscape
- proportion of 'Traditional' farmers
- proportion of 'Commercial' farmers
- proportion of the landscape in use as arable agriculture
- proportion of the landscape in use as pastoral agriculture
- proportion of the landscape in an 'abandoned' state
- proportion of the landscape not used for agriculture
- the size of the largest cluster (in patches) used for agriculture
- mean cluster size (in patches) of all clusters used for agriculture
- number of clusters (counted by cover)
- number of clusters (counted by agent)

4. MODEL RATIONALE
In light of the November Fieldtrip to Madrid I have modfied the structure of the agricultural decision-making module. The primary modification to 'Version 2' of the model is the distinction between two different types of farmer which I will term 'commercial' and 'hobby'. Second, the available land use options are reduced from three (Pastoral, Arable or Dehesa) to two (Pastoral and Arable only). Third, the way fragmentation is calculated has been modified so that the location of the homestead is not required. Finally, new variables are considered when calculating a patch's profit; namely, land capability and distance to the nearest road or track.

Rationale for 'Commercial' and 'Hobby' Farmer Types

It became clear during interviews with local farmers and farming officials that there is a clear distinction between commercial farms that operate in a rationally economic, profit-maximising manner, and those that operate merely to maintain traditional agricultural practices and landscape asthetics.

"Whoever has a vineyard nowadays is like a gardener... they like to keep it, even if they lose money. They maintain vineyards because they have done it all their life and they like it, even having to pay for it. If owners were looking for profitability there would be not a singe vineyard... People here grow wine because of a matter of feeling, love for the land..." [F_NDR]

Thus, for many land owners in the landsape their farm is a hobby kept with no regard for its financial rewards. Clearly the economically rational agent is not appropriate for these 'hobby farmers'. Further, there are many farms across the study area that are run to provide supplementary household income. For example, in one area, Santa Maria, one farmer [LF_SM] stated that of the 80 livestock farmers in the area only 20 (one quarter) made their living solely from farming with the rest taking their primary income from alternate sources (light industry or building nearby) but still keeping some land and livestock active. The farmer termed this supplementary activity 'extensive farming':

"Extensive [pastoral] farming is less time demanding... once a day, 1 or 2 hours can be enough... I see in the future extensive farming will be the most important though - I don’t know if it is more profitable or not - but because it needs less time." [LF_SM]

In San Martin del Valdeglesias, a farming official described a similar situation:

"In general, [pastoral] farming is part-time, with only a few exceptions. The main problem is that it is not economically profitable - it is difficult to sell the product." [FO_SMV]

And again in Villa del Prado, a spokesperson for a Farming Association:

"[arable] agriculture here is part time... there are a few farmers that live on horticulture, and have vineyards as an extra." [APG_VDP]

Finally, in an area traditionally known for its vineyards:

"Part-time workers? Yes, most of them... there are no full-time farmers. Here there are only retired people and their children, who work somewhere else, and help their retired parents with the labouring. There is no way to live on wine production."

This part-time nature must be acknowledged in the behaviour of agents of the model. In general, we might say that these hobby and part-time farmers are less concerned with the economic state of the market, and their activities will not be sensitive to changes in it. In contrast, those farmers for whom their farm is their sole source of income treat their land as a commercial enterprise:

"There are some young farmers, 5 or 6 of whom are less than 25 years old, that are making important investments. If someone wants to live from livestock farming, they need to have an entrepreneurial vision, a business mentality, like in any company." [LF_SM]

These farmers will adopt land use and farming practices to maximise their income. Different behaviours [model rules] are therefore appropriate for each type of agent; 'commercial farmer' or 'hobby farmer'.

Model Structure
Figure X illustrates the step taken during one 'time-step' of a model run. Each time-step represents one season of the year, for seasons per year. This model description will briefly describe setup procedures before moving through each of the 12 steps illustrated in the flow chart (Figure X).

Setup Procedures
At setup the defined number of agents (farmers) are created with a random wealth, age (between 25 and 60 years) perspective (i.e. 'commercial' or 'hobby'). 'Clusters' of patches are then randomly generated across the landscape, the size of which is determined by the cluster_probability variable (see Saura & Martinez-Millan 2000 for method). Clusters are then randomly assigned a land use (Crops, Pasture, Abandoned) and then randomly assigned to an agent. One of each agents patches is then assigned as the 'homestead' (this assignment is now redundant as homestead is NOT used to calculate a patch's fragmentation value).

Model Execution
Step 1. Start or iterate season loop (add one to 'season').

Step 2. If this time-step is the fourth season of the year, increase farmers' ages by one year.

Step 3. If this time-step is the fourth season of the year, check if commercial farmers have reached retirement age (currently 65 years).

Step 4. If a commercial farmer has reached the retirement age a check is made to establish if the farmer has a son. The code for this check is;

( rand_no < (propM + 0.25) )

Thus, if a randomly generated number (between 0 and 1) is less than the proportion of farmers that are 'commercial' plus 0.25 then the farmer does have a son to inherit the farm as a 'commercial farmer'. The likelihood that a commercial farmer has a son is based on the proportion of commercial farmers in the neighbourhood not because this is the mechanism that dictates whether a farmer has a son (clearly that is not right!) but because this is likely to be an important factor in determining whether a son WANTS to continue his father's business. 0.25 is added to the likelihood to ensure that when there are no commercial farmers in the landscape, there is still a chance that a son will want to continue the business - this accounts simply for personal choice (of the son) and the father's individual influence over his son's attitudes (which are likely to be just as, if not more, important than the proportion of the local community that has the 'commercial perspective'). The son's age is randomly set to a value between 20 and 40, ensuring that the value is less than the dying farmer's age minus 20 (this assumes that farmers do not have children before the age of 20).

If the farmer does not have a son, the farmer becomes a hobby farmer. This transition is made because it is assumed that having farmed his land all his life the farmer is unlikely to want to simply give up his land for nothing (a sentiment that interviews suggested to be strong).

Step 5. Check if farmer (both perspectives) dies.

A random is generated between 0 and 1. If this number is less than the probability for the farmer's age (provided by the 'death probabilities' file that is read in during set-up; data source: http://www.lifetable.de/data/MPIDR/ESP_1998-1999.txt) the farmer is deemed to have died during this year.

Step 6. Check if the farmer has a son.

If the dying farmer is commercial and older than 40 years the method described in Step 4. is used; if a son is present he inherits the farm as a commercial farmer. If the farmer is younger than 40 it is assumed that either he does not have a son, or if he does the son is not old enough to take over the running of the farm. If a son is not present the farm 'dies'; ownership of all patches is released to 'nobody' and are assumed to be abandoned.

If the dying farmer is a hobby farmer and older than 40, checks are made to establish if there is a son to inherit the farm - as either type of farmer. The first check establishes if the son inherits as a commercial farmer;

(mean_tot_patch_profit + propM - ( 3 * age / 100 )) > 0

If this code is true a son inherits the farm as a commercial farmer. Thus, if the mean patch profit of all patches currently in use by commercial farmers (i.e. NOT abandoned; this value will generally lie between 0 - 2) plus the proportion of neighbouring farms that are commercial minus 3 times the age of the farmer divided by 100, is greater than zero, the son becomes a commercial farmer. This assumes that the son will be willing to become a commercial farmer when the profit in the landscape is generally high, there are other commercial farmers in the landscape (i.e. he sees that others are finding it possble to make a living from their land) and his age is low (and therefore he assumed to be more willing to take a risk and 'give it a go').

If this check is false a check is made to establish if the farmer has a son to inherit the farm as a hobby farmer in a similar manner to that described in Step 4. The code for this check is;

( rand_no < (propT + 0.25) )

Thus, if a randomly generated number (between 0 and 1) is less than the proportion of farmers that are 'hobby farmers' plus 0.25 then the farmer does have a son to inherit the farm as a hobby farmer. As above, the likelihood that a hobby farmer has a son is based on the proportion of hobby farmers in the neighbourhood not because this is the mechanism that dictates whether a farmer has a son (clearly that is not right!) but because this is likely to be an important factor in determining whether a son WANTS to continue in the steps of his father. 0.25 is added to the likelihood to ensure that when there are no hobby farmers in the landscape, there is still a chance that a son will want to continue the business - this accounts simply for personal choice (of the son) and the father's individual influence over his son's attitudes (which are likely to be just as, if not more, important than the proportion of the local community that has the 'hobby perspective').

Again, the son's age is randomly set to a value between 20 and 40, ensuring that the value is less than the dying farmer's age minus 20 (this assumes that farmers do not have children before the age of 20). If both checks are false the farm 'dies'; ownership of all patches is released to 'nobody' and are assumed to be abandoned.

Step 7. Commercial farmers calculate their profit for this season.

Profit is calculated for the three potential land uses (for each patch) using the following code:

Crops profit = ( valueC * lcap ) - ( calc_frag * 2 * costC ) - (road_dist / screen-size-x) )
Pasture profit = ( valueP * lcap ) - costP - (road_dist / screen-size-x) )
Abandoned profit = -0.1

Crops:
The current value of crops is multiplied by the land capability value (a value scaled between 0 and 2 indicating the suitability of the patch for agriculture based on slope, soil and climate). From this, twice the fragmentation value of the farm multiplied by the current cost to produce crops is subtracted, followed by the subtraction of the distance to the nearest road divided by the maximum possible distance (giving a range for this value of 0 - 1). Thus, the highest profit is earnt by patches with a high land capability, low fragmentation value, and low distance to the nearest road when the value for crops is high and the cost of production is low.

Fragmentation value has changed from previous versions of the model. These previous versions assumed the location of the farmhouse ('homestead') was known for each farm; this data is not available for SPA56 so the calculation has been duly changed. Fragmentation is now calculated using the following code;

1 - ( prop_farm / max_dist )

where prop_farm is the proportion of the total farm area composed by the cluster in which the patch under consideration lies, and max_dist is the maximum distance between the patch under consideration and another patch owned by the same farmer. Thus, when prop_farm is large and max_dist is small, the fragmentation value of the patch is low. This index penalises patches in small clusters that are of larger distances from other patches owned by the farmer. This index has a theoretical range of 0 (when max_dist is infinite) to 1.

Pasture:
Pasture profit is calculated in a very similar manner to crops, the difference being that the fragmentation value of the patch is not considered. The rationale behind this is that land for grazing does not afford much advantage by being clustered in large patches. Small areas of land may be used for grazing just as easily as large. However, the distance to the nearest road or trail is deemed important as this will ease movement of livestock between areas of pasture and to the market.

Abandoned:
There is no immediate value afforded by owning land in an abandoned state. However, the costs of doing so are also minimal. As the land may be useful in the future, and long term planning or forecasting of the state of the market is not represented in the model the profit per abandoned patch per season is therefore set at -0.1. This value is minmial compared to the proces and costs of the other land uses (which should range between 0 and 2).

Once profit for all individual patches has been calculated, the sum of these values is calculated for the entire farm. If the total farmed area of the farm exceeds a 'maximum single farmer area', for each patch exceeding this area a further cost is subtracted from the total farm profit. This cost is designed to reflect the infrastructure and labour required to farm an area greater than that possible by a single farmer with no hired labour. Thus, while currently abritrary, in the future the given area will be set to the maximum area that could be maintained by a single farmer not employing hired labour.

Step 8. Commercial farmers check their annual profit at each year's end. If their wealth is negative, go to Step 11.

Step 9. If annual profit is equal to, or less than, -5.0% of their wealth, go to step 10. Otherwise, go to Step 12.

Step 10. The 'negative profit' year counter is increased by one. If this counter is now equal to five (i.e. the last five consecutive years have been loss-making) go to Step 11. Otherwise, go to Step 12. (N.B. this counter is re-set to zero if annual profit is not less than -5% of wealth.)

Step 11. A check is made to see if the commercial farmer changes to become a hobby farmer. This check is made by the follwoing code:

rand_no < (propT + 0.25) or age-of self >= 50

If either statement in this block is true, the commercial farmer becomes a hobby farmer. The first half of the block (before 'or') is the same as found in Step 6 when the check to establish if the farmer has a son to inherit the farm as a hobby farmer is made, and is based on the same premises and assumptions. The second statement in the block checks the age of th farmer. If the farmer is 50 or older he becomes a hobby farmer. This is based on the assumption that a younger farmer will want to move onto another job elsewhere because they still have 'time on their side' to start a different career. If older than 50, it is assumed that the farmer will be less inclined (or skilled) to endeavour to find a new full-time career and will therefore maintain the farm as a supplementary income.

However, if both statements are false, the farm 'dies'; ownership of all patches is released to 'nobody' and are assumed to be abandoned.

Step 12. Hobby farmers check the size of their farm. If the farm is larger than the 'maximum single farmer area' (see Step 7) the farmer reduces farm size by abandoning the patch furthest from a road/trail until the specified area is reached. Distance to road is used as it is assumed that the hobby farmer will reduce maintainence costs rather than maximising productivity. This step is only needed for the case in Step 11 where a commercial farmer becomes a hobby farmer.

Step 13. Commercial farmers estimate their profit for the next season with the land they currenly own. Each possible configuration of their land (i.e. with each patch they own in each of the three possible land use states) is checked and patches change land use accordinly to maximise profit for the next season.

Farmers' estimates for their profit in the next season is based on their estimates of the values and costs for crops and pasture for that next season. These estimates are based on the values and costs of the current and previous seasons, and the accuracy of the farmers' estimate in the previous season. For example, the code for a farmer to estimate the value of crops for the next season is;

valueC + actual_value_diffcC + random-float 0.5 * ( valueC - prev_est_valueC )

where actual_value_diffcC is the difference between the previous value of crops and the current value of crops, and prev_est_valueC is the previous estimated value of crops. Thus, the current estimate is given by the error in the last estimate multiplied by a random number between 0 and 0.5 added to the current value of crops and actual_value_diffcC. This method ensure farmers can estimate future prices reasonably well when values and costs change slowly, but perform worse when changes are rapid.

If the land use configuration of the land currently owned can be modified to imrove profit land use conversions are made. A 'ladder' of land uses is considered to restrict some land use conversions. In this ladder Cropland is above Pasture land which in turn is above Abandoned land. An unlimited number of patches conversions 'down the ladder' may be made in any one season. Only one conversion 'up the ladder' may be made in any one season. Conversion up the ladder (e.g. from abandoned to cropland) will require both resources of both time and money and so representation of this process should be slowed within the model. Conversions down the ladder require vastly less resources and are simply achieved by reducing maintainance levels.

If conversions to currently owned land are made and the program continue to the next iteration (Step 1.) Otherwise, if land currently owned is in the optimal state to maximise profit, Step 14 is executed.

Step 14. Commercial farmers bid to buy abandoned patches that neighbour their own land. All neighbouring abandoned patches are examined to assess if their ownership and conversion, to either cropland or pasture, will increase the farmer's profit in the season. The neighbouring abandoned patch which will increase profit by the largest margin is then bid for. The bid a farmer will offer is given by the estimated increased profit it will afford multiplied by (100 - age of farmer). The age of the farmer is considered as this gives a rough guide to the number of years the patch (if bought) will provide to the farmer.

If the bid being made is larger than the asking price of the current owner, ownership passes to the bidding farmer and land use change to the most profitable state. The asking price of a farmer is set as the current wealth of the farmer divided by the total number of patches owned by that farmer. If the patch is abandoned but unowned (i.e. 'dead' as it once belonged to a farm that was 'killed') the asking prices is set to the 'current market price' for patches. This price is set to;

40 * mean_tot_patch_profit

where mean_tot_patch_profit is the mean patch profit for the season across all patches owned by commercial farmers in the landscape. Multiplying by 40 gives an estimate of potential profit to be made by that patch (in the current market state) over the next decade.

If a bid is not as large as the asking price, ownership stays with the current owner and the patch remains in the 'abandoned' state. Whatever the result of a bid, the once all farmers bids have been considered the program continues to the next iteration (Step 1.)