Farm Accountancy Data Network (FADN) is a primary data source of current EU farm models. FADN data derives from stratified sampling and thus attaches a weighting factor to farms denoting how many farms are represented by a single observation. A modeler can upscale the model results to the whole population by multiplying individual farm output with the weighting factor. This works adequately for deriving aggregated results due to the statistical consistency of the sampling methodology.
In the case of farm ABMs the above process is differentiated. The farm population must first be recreated, the simulation is run and results are directly derived (no upscaling is needed). This brings the question of how to recreate the population from the limited in number FADN farms. One common and easy method is cloning. Each farm contained in the FADN sample is cloned as many times as its weighting factor.
Cloning may introduce distortions in the simulation due to the existence of interactions between farms and also due to its multi-year span. In this paper we contact a computational experiment to test whether the cloning method is adequate for use in farm ABMs.