Assessment of Simulation Behavior of Different Mathematical Programming Approaches

This paper investigates the simulation behaviour of different PMP methods being developed in the past. About 800 of identical farms for eight years from the German FADN 1 were used to aggregate 45 farm groups. The aggregated farms were calibrated for 1996/97 and the observed prices, direct payments and yields for 2002/03 were applied. The ex-post simulation results underline that the simulation behaviour is mainly controlled by the considered PMP methodology. Even the Maximum Entropy approach proposed by Paris and Howitt (1998) for one observation on base year allocation did not improve the findings. In addition first experiences with an alternative model to PMP where the Q matrix is recovered using multiple observation proposed by Heckelei and Wolff (2003) are illustrated for one particular farm.