Improving the Representation of Public Procurement in the GTAP Framework

Overview In 2012, €352 billion of the €420 billion EU public procurement budget was open to member countries of the WTO Agreement on Government Procurement (GPA). This represents 85% openness for the EU, which contrasts with the value of other GPA members such as the USA, whose level of public procurement offered to foreign bidders was of approximately €178 billion of €556.25 billion (32%) and for Japan where the open procurement figures were about €27 billion of the €96.4 billion (28%). Given the large, and growing, size of these public expenditures, access by potential foreign suppliers has become increasingly important. Yet the analytical tools used to quantify the impact of restrictive public procurement decisions remain woefully underdeveloped. This paper reports on a project, being undertaken by the Center for Global Trade Analysis and collaborators, aimed at improving the representation of public procurement in the GTAP data base and modelling framework. The Nature of the Challenge Due to the fact that any public procurement agreements will involve multiple countries, purchasing a wide range of goods and services, supplied by many different regions, this question is best addressed in a multi-region, general equilibrium framework. Yet the scope for global CGE modelling of any new issue, such as public procurement, is inevitably limited by data. If current purchases cannot be separately identified in the data, then there is little hope of estimating the likely impact of liberalization. Currently most global CGE models rely, directly or indirectly on the GTAP Data Base (Narayanan, Aguiar and McDougall, 2012). Unfortunately, the current structure of GTAP poses serious limitations when it comes to analysing public procurement. There are several key limitations which deserve special mention. First of all, it is likely that the current data base does not accurately capture the mix of domestic and imported goods which are purchased by governments. T...