# Modelling the implementation of the 2003 CAP reform in Austria

## Modelling agri-environmental policies at sector level

The Positive Agricultural Sector Model Austria (PASMA) is employed to estimate the impact of the 2003 CAP reform on selected economic and environmental indicators. It is designed to model both production linked support measures and decoupled payments. It also includes the program for rural development with 32 agri-environmental measures and the program for Less Favoured Areas.

The model maximises the sum of gross value added plus “other subsidies” according to Economic Accounts of Agriculture (EAA) terminology.2 It is calibrated to historic crop, forestry, livestock, and farm tourism activities by using the method of Positive Mathematical Programming (Howitt, 1995). This method assumes a profit-maximizing equilibrium (e.g. marginal revenue equals marginal cost) in the base run and derives the coefficients of a non-linear objective function from a linear programming model. The calibration method allows deriving marginal costs from observations of average costs and observed levels of production activities. Two major conditions need to be fulfilled: 1) the marginal gross margins of each activity are identical in the base run, and 2) the average Positive Mathematical Programming (PMP) gross margin is identical to the average Linear Programming (LP) gross margin of each activity in the base run. These conditions imply that the PMP and LP objective function values are identical in the base run. This method has been modified and applied in several models (e.g. Lee and Howitt, 1996; Heckelei and Britz, 1999; Arfini and Donati, 2003).

Another important assumption needs to be made: assigning marginal gross margin effects to either marginal cost, marginal revenue, or fractionally to both. In PASMA, the marginal gross margin effect is completely assigned to the marginal cost, and consequently coefficients of linear marginal cost curves are derived. An extension of the PMP method, the multi-variant production approach suggested by Rohm and Dabbert (2003), is also implemented. Their reasoning is that it is easier to switch from management practice A (e.g. standard production with growth regulator) to practice B (without growth regulator) when producing wheat than to switch from wheat to maize. We build on this approach, with one exception. Organic farming is not assumed to be a management variant of the Rohm and Dabbert type but to be a fully separate practice, which can be split into management variants of its own (with and without winter cover crops, etc.).

Therefore, the model differentiates between conventional and organic production systems (crop and livestock) through separate feed and fertilizer balances at regional and structural scales. Transfers between these two production systems are not allowed in the model; however, they compete for the same resources (i.e. land and labour). Consequently, linear marginal cost curves are derived for all activities of both production systems for the base period.

In PASMA, linear approximation techniques are utilized to combine the PMP calibration method with an aggregation method that builds convex combinations of historical crop mixes (Dantzig and Wolfe, 1961; McCarl, 1982; Onal and McCarl, 1991); details of its implementation are provided in Schmid and Sinabell (2005). Other model features such as convex combinations of feed mixes, expansion, reduction and conversion of livestock stands, a transport matrix, and imports of feed and livestock are included to allow reasonable responses in production under various policy scenarios.