Plots are assigned to agents on a auction where individual farmers bid according to their shadow price.
Farmers calculate the profit they would have without that specific plot and compare it with the profit they would achieve with the bidding plot.
The difference in profit is the shadow price, that become the bid offer once adjusted by a couple of parameters to take into account fixed and variable negotiation costs and overheads.
As plots are fully heterogeneous we can't use previously made bids by a farmer as an information on what it will be the bid for a certain plot (a technique used in AgriPoliS).
At the same times, for computational reasons, we can't ask on each free plot to ALL agents of the model to bid for it. When stopping to collect the bids?
We ended up asking to bid only to the farmer with the higher likelihood to offer the highest bid 1).
Just before the simulation start the model performs a calibration step where, for each soil type, a certain number of bids are compared with the characteristics of the farms making such bids, like number of plots already available on that soil type and distance.
Once we have estimated this coefficients we can go to the effective renting procedure. Here, for every free plot, we order the farms basing on what it is the estimated bid and only at the firsts of them we ask to make the real bid.
Obviously more farms we take for the initial calibration stage and more farms we gonna asking for the real bidding, more chances we get to collect the region-wide highest bid, but also slower the simulation is.
We gonna ask the bid to the closer farms. We made some test where we saw the highest bid (with the current level of transport costs) is always made by the closest 10-15 farms, so we gonna ask to bid to the closest 20 farms and we are pretty sure to get the highest bid..
Obviously more farms we gonna asking for the bidding, more chances we get to collect the region-wide highest bid, but also slower the simulation is.
