Environmental Engineering Reference
In-Depth Information
Accordingly, the resolution of CAPRI in terms of inputs, outputs, production
activities and regions is driven by data availability, with the EUROSTAT agricultural
domain as the main data source. Regional data aggregate to Member State and
higher level based on clearly defined and fully consistent aggregation rules. To achieve
this database consistency making best use of the information available, CAPRI
employs Bayesian techniques (Britz et al.
2007
: 15ff.).
SEAMLESS-IF does not follow these consistency and full-coverage requirements
for all components. For example, the bio-economic farm models (FSSIM, see
Chapter 5) are incorporated into the system from a bottom-up perspective, sourced
by a bio-physical model (APES, see Chapter 4) for the definition of agricultural
processes and by regional surveys on management practices. These models
are designed to simulate responses of the bio-physical system regarding crop
growth, nutrient fate, the water cycle or soil erosion to a site-specific combination of
farming practices and physical characteristics. The bio-physical models may be
seen as an instrument to construct the production possibility set on a specific
site, by systematically varying different parameters describing the management
practices such as rates and timing of fertilizer application, seeding, and irrigation.
Combined with a behavioural assumption of maximising expected income
penalised by variability to reflect risk aversion and data on costs and revenues
associated with each production possibility set, this information allows building
farm optimisation models able to allocate land and other resources to current
(i.e. observed) and alternative activities. The calibration to crop and livestock
production quantities observed in the base year is performed by using a variant of
Positive Mathematical Programming (PMP). The ultimate link to aggregate market
models is made by extrapolating supply response behaviour in terms of rates of
quantity changes per rate of price change (elasticities) from a sample of FSSIM
farm type models. This is achieved with the EXPAMOD model (Bezlepkina et al.
2007
; Pérez Domínguez et al.
2009)
using data from price-sensitivity experiments
with FSSIM as observations for the estimation of supply response surfaces for
agricultural products depending on variables defining the classification of the
SEAMLESS-IF agri-environmental (e.g. rainfall, solar radiation, soil type, carbon
content) and farm resources (e.g. availability of land, labour and machinery), both
types of variables available at Pan-European level. The regional supply models of
CAPRI are calibrated to the extrapolated price elasticities.
Concepts of Linking Models/Components
From a methodological perspective, we can identify two types of model linking
possibilities here: (i) by means of econometrically estimated response functions,
and (ii) by means of (iteratively) modifying the output of model components
such as to fit the combined output of other models. The first option has been
used in SEAMLESS-IF for linking farm optimisation and aggregated market models
and was already described above. The stand-alone CAPRI model also applies this
