The study analysed existing databases for agricultural market data on errors and discrepancies and to elaborate the possibilities to harmonise datasets for policy modelling. The study supports DG AGRI in improving quality and timely availability of data for market modelling and ensuring that data from different sources are consistent. This study aims to provide a structure for a consolidated database for policy modelling which does not alter existing databases. Within this report, existing databases are analysed to derive key insights for setting-up a harmonised metabase. As available databases comprise statistical databases as well as scientific model databases, both groups are studied. For the purpose of this study, statistical databases are defined as providers of the information that international institutes receive from their reporters, while the reporters are required to provide harmonised, complete, consistent, and where possible, timely data series for establishing models or other quantitative methods. Nevertheless, a statistical database can also serve as a model database, such as e.g. PS&D. Statistical databases from international institutions (FAO, USDA, Eurostat), as well as model databases (AGLINK/COSIMO, AGMEMOD, CAPRI/CAPSIM, ESIM, FAPRI, GTAP, FARM, IMPACT), were studied to find ways of consolidating data and providing insights that allow for a better comparison of model results. For this reason, various classification schemes used in agricultural statistics were reviewed (country, product, balance item, year, unit), as was the manner in which the different modelling groups have dealt with these classifications in their databases. Besides a common classification, a harmonised database for market modelling purposes will require further efforts to be applied to a consolidation effort for the original data. Such a procedure must be supplemented by methods dealing with completion and balancing.
