WP1: Database construction. In order to calibrate, initialize and test the model a comprehensive historical database will be built containing standard aggregate economic indicators such as GDP and unemployment as well as fine grained microeconomic data about EU households, firms, and financial institutions. This will integrate together data that the various principal investigators of the project have already collected. Substantial new data will also be collected, which will require integrating data from multiple EC member state, academic, and private sources in a consistent way. There will also be mixed levels of data availability and access by variable and geography, requiring statistical adjustments and estimations. Such a consistent, integrated microeconomic database will in itself be a valuable deliverable of the CRISIS project.

 

There are significant challenges in creating an integrated data set to underlie the model due to differences in methodologies and definitions from the numerous data sources that will be required. Creating such a data set for the EU will present further challenges due to differences in data availability and methodologies across individual member states; moreover, EU data protection laws limit access to survey microdata for households and firms. Finally, the project will have to face all of the challenges normally related to designing, programming, and testing such a large database.

 

The consortium members have significant experience in integrating multi-source data sets and using techniques for adjusting methodological and definitional differences. Statistical estimation techniques, and where necessary the use of parameters from comparable countries/regions where data is available can also compensate for data availability issues. The workplan includes phases for needs assessment, source inventory, and data architecture design to ensure that these issues are considered up-front, and that a realistic detailed scope is developed.

 

WP 1 will have to meet the following four objectives:

• To parameterize key relationships in constructing the agent-based models developed in WP2 and WP3 and for the general ABM to be built in WP8.

• To initialise the models in a state approximating the current state of the EU economy, or in a historical state (e.g. 2006 just before the crisis).

• To test the performance of the models versus historical data.

• To enable the models to respond realistically when played in game-model