Construction and updating of a Ugandan CGE database

This paper documents (1) the structure of a CGE database; (2) the data manipulation steps in creating such a database from published data; (3) updating a SAM; and (4) describe features of the updated SAM. The database is constructed for a Ugandan CGE model. The building blocks for creating a database for a CGE model are official data from an Input/output (IO) table, or from a Supply Use Table (SUT), or from a SAM. Often the structure of the published data is not in the required format of a CGE database, and so a major task is to transform the official data into a form required by a CGE database. The first step in this task is typically a review of the primary source of data. We then proceed by identifying any implausible, unusual and negative values. We adjust these elements and rebalance the database to ensure that the balancing conditions hold. We then proceed to create the matrices required by the CGE model. Typically we create (1) a source dimension for all user-specific matrices, (2) user and source-specific margin matrices, (3) user and source-specific tax matrices and (4) industry-specific land rentals. It is likely that as we adjust data and create the required matrices, we violate the balancing conditions. Therefore in each step in the database construction stage, we check the balancing conditions and when appropriate we rebalance the database to ensure that the balancing conditions hold. Having constructed the 2002 database that conforms to the CGE structure, we update the database to 2009. We then proceed to create an additional sector namely, RawOil sector. In terms of the database, we create an additional industry and an additional commodity. Our final task is to create, based on the 2009 database, an updated SAM. The CGE database does not provide information on transfers between economic agents. We therefore adjust the transfer elements based on shares estimated for 2002.