EU-EMS (European Economic Modelling system) is a flexible modelling system build by PBL Netherlands Environmental Assessment Agency  that has modular structure that allows to choose model form that fits the best policy question at hand. The model system database includes the representation of 62 countries of the world and one Rest of the world region. It has detailed regional dimensionality for EU28 countries and includes them as consisting of 276 NUTS2 regions. Sectoral and geographical dimensions of the model are flexible and can be adjusted to the needs of specific policy or research question. Mathematical forms chosen for modelling of demand, supply and production are also flexible and can be modified according to the needs of the modeler.

Long-term growth SCGE model developed in MONROE project combines the modelling of sectoral structural change via sector-specific technological progress and dynamic changes in consumer preferences with representation of changes in regional labour supply by education type. Labour supply in the model is modelled as the combination of regional demographic projections of Eurostat with modelling of catch-up in education levels and participation rates between different EU regions. Endogenous growth part of the model includes three parts: (1) knowledge production with modelling of the number of patents as a function of private and public R&D, number of researchers, specific capital stocks (ICT etc.) and the stock of accumulated regional and rest of the world knowledge. (2) knowledge adoption that is modeled as sector and region specific catch-up process that depends upon the stock of human capital and R&D. (3) knowledge spillover effects between regions that transfer knowledge via trade, migration and FDI flows.  

Goods and services are consumed by households, government and firms, and are produced in markets that can be perfectly or imperfectly competitive. Spatial interactions between regions are captured through trade of goods and services (which is subject to trade costs), factor mobility and knowledge spill-overs. The model includes New Economic Geography (NEG) features such as monopolistic competition, increasing returns to scale and migration.

Endogenous growth in the model results in technological changes via product and process innovations. Product innovations in the model are captured via endogenous number of the varieties of various final consumption goods purchased by households. Process innovations are represented by the changes in productivity and share parameters of the sectoral production functions. Technological change in the model can be both capital and labour biased and results in either reduction or increase in labor demand. The model includes representation of sectoral employment by three levels of education and  regional voluntary unemployment that is captured by the Beveridge curve.  

Model improvements in R&D and human capital

Following the paper of Griffith, Redding and Van Reenen (2001) our model assumes that R&D has two roles in the development of TFP. The first role is in knowledge creation or stimulation of innovation that has received a lot of attention in both theoretical and empirical literature. The second role is in facilitation of adoption or imitation of knowledge that has been created in other countries or sectors. Griffith, Redding and Van Reenen (2001) use a panel of OECD countries and find a strong empirical evidence for the second role of R&D in adoption of knowledge. Griffith, Redding and Van Reenen (2001) present a general equilibrium model of endogenous growth through increasing productivity that incorporates both roles of R&D investments. They augment the conventional quality ladder model to allow the size of innovations to be a function of the distance behind the technological frontier, deriving a equation for TFP growth as a function of the knowledge adoption that is captured by the growth of the technological frontier and an interaction between technological gap, and R&D per unit of sectoral output, as well as the interaction between technological gap and human capital. The levels of human capital and R&D capture the absorptive capacity of the particular sector. The growth of TFP is also linked to knowledge creation that is explained by R&D and human capital stocks.

To model R&D intensity we simply assume that R&D decision follow an AR(1) process with a constant term and a vector of regressors, representing the following six types of variables:  a) barriers to trade and investments, b) barriers to entrepreneurship, c) state control, namely governmental distortion in the market such as price ceilings d) government-financed R&D expenditures, e) public expenditures on R&D, and finally f) public expenditures on education and social programs. To limit possible reverse causality and multi-collinearity problems, the policy vector is lagged one period, where each period presents 5 years, while in the regressions we include each variable one by one. Because of the dynamic specification of our Research intensity equation, the equation is estimated using the Arellano and Bond methodology. 

In order to integrate the econometrically estimated sector-specific equations for R&I intensity and productivity into the regional-economic model, we need to have data at the regional level about the initial productivity levels of the sectors in different regions as well as the regional level of human capital and R&I expenditures by private and public sector. This type of data can be constructed using regional accounts data and regional science and technology data from Eurostat.  Several of the policy indicators used in the regressions including barriers to trade and investments, barriers to entrepreneurship and state control are national level indicators that represent national policy and hence cannot be calculated at the regional level. Other types of the policy indicators such as government-financed R&D expenditures of private sector, public expenditures on R&D as a share of overall R&D expenditures and public expenditures on education and social programs can be derived at the regional level using Eurostat data.

Once the indicators for various explanatory variables of the econometric equations have been constructed at the regional level they can be substituted to the econometrically estimated sector-specific regressions for R&I intensity and productivity. This way we have constructed regional sector specific versions of the equations that can be used further for policy analysis and linked to the rest of the regional-economic model via (1) changes in sectoral TFP that enters into the production function and (2) increase in R&I expenditures that is linked to the demand of the firms for intermediate inputs.