The final version of the GEM-E3-MONROE model is a large scale applied CGE model that covers the whole world aggregated in 33 countries/regions and 50 economic activities. All countries and sectors are linked with endogenous bilateral trade transactions. The model has a detailed representation of the manufacturing sectors of clean energy technologies and knowledge spillover matrices linked to R&D expenditures for these technologies. The simulation time period of the model is up to 2050 in five-year time steps.  Firms decision on R&D that increases sectoral productivity and on the training of its employees are endogenous. Households decide endogenously on their consumption, savings, labour supply and education level.  The overall spending of public R&D is exogenous but the decision amongst different energy technologies is endogenous. The labour market feature unemployment for five skills/education categories and the labour supply is based on empirical estimations of the wage curve for each EU member state and key trade partners (USA, China, India). An age cohort satellite model is linked with the core GEM-E3-MONROE model ensuring that labour productivity applies to the educated cohort that enters into the labour market. The model has an explicit link between human capital and the absorptive capacity of knowledge spillovers.

GEM-E3 is calibrated to the latest IO statistics to the GTAP v10 database with base year 20114 and has a representation of taxes and subsidies and in particular R&D tax incentives.

Model improvements in R&D and human capital

GEM-E3 has been further developed to incorporate specific characteristics of R&I policies and the links to human capital and can be used to assess the effects of R&I programs into several dimensions including economic, competitiveness, and employment implications. GEM-E3 modelling improvements focus on the representation of specific features of private and public R&D expenditure, knowledge spillover effects, R&I instruments, the representation of human capital, and its linkages to absorption of knowledge spillovers. The GEM-E3 model has been extended to include: five labour skills, endogenous skills supply, endogenous knowledge absorption and stochastic public R&I decisions.


To capture the structural effects of policies that can cause a potential mismatch between supply and demand for specific skills, the GEM-E3 model has been expanded with a representation of five distinct labour skills, .namely: unskilled workers (level 1), service and shop workers (level 2), technicians (level 3), clerks (level 4) and managers (level 5).  This is combined with the endogenisation of households’ decision for education that influences the level of its future wage and income. The optimal labour demand by skill is derived from the firms’ cost minimization problem. The nesting of the production function depends on the substitution possibilities that characterize the production technology of each production sector. Each type of labour skill is characterized by a specific labour productivity (technical progress of labour), with low skills typically having lower productivity. Based on extensive econometric estimations and using data from the EUROSTAT database, the labour productivity across EU member states is econometrically estimated, by explicitly linking technical progress of labour with the skill levels of employees (identifying low, medium and high-skilled).

This improved model version of GEM-E3 includes an explicit representation of the link between human capital stock, labour productivity and the firms’ capacity to absorb knowledge produced elsewhere (spillovers). The inclusion of human capital aims to improve the representation of:

  • Labour productivities differentials across countries and sectors
  • The link between human capital and the creation of knowledge through R&D
  • The knowledge spillovers through modelling of absorptive capacity
  • Potential for capacity growth of new high value added economic activities through increased skills availability (i.e. higher share of workforce attending tertiary education)

The new labour force is an endogenous cumulative variable that accumulates the working age population. The households’ choice on education affects the working age population that in the period t can be added to the labour force based on its’ decision for education.

For each skill category the demand-supply mismatch results into a skill specific unemployment rate. The model assumes full labour mobility across sectors for each skill type. The supply of labour for each skill is determined via an empirically determined wage curve, linking wages with unemployment rate. The labour supply function is calibrated to a wage elasticity of -0.1.

The “innovation and spillovers module” of the new GEM-E3 model assumes that total factor productivity is composed from an endogenous and an exogenous part (in conventional CGE modelling total factor productivity is set exogenously). The endogenous part represents the concept of innovation-induced endogenous growth and is composed of: i) the learning by doing effect, ii) the learning by research effect, iii) the impact of knowledge spillovers, iv) the human capital stock measure and v) the productivity of the previous period. The availability of Human Capital is essential to enable productivity growth induced by R&D and knowledge spillovers. The index of human capital stock is constructed based on the shares of each skill type to the total labour force; these indicate that the respective skills embodied at each skill type are more productive relative to those in the lowest skill type.

The R&D capacity of firms is linked to human capital availability. In particular the total factor productivity growth induced by R&D depends on the absorptive capacity of firms, as the elasticity of the cumulative R&D stock with respect to productivity is a function of the human capital stock. The cumulative R&D expenditures are expressed by the following formula:

In the GEM-E3 model, each sector optimizes allocation of resources in R&D simultaneously with decisions about acquiring capital, labour, energy, material. Each sector decides on its R&D expenditures based on its production function and the share of R&D expenditures in intermediate demand. R&D expenditures is a demand for R&D services addressed to the R&D supply sector, which is represented as a separate production sector in the GEM-E3 model. The R&D expenditures do not increase production capacities, as investment does, but they improve the quality and reduce the costs of the produced goods and services. As resources are limited, R&D expenditures may exert a crowding out effect on investment, but only temporarily because the improvement, thanks to R&D, enlarges the market prospects and it may induce higher investment in the long term. Higher quality and lower costs imply better possibilities to use the economy resources both in the short and in the long term and thus R&D expenditures may induce positive economic growth.

The capacity of firms to absorb knowledge spillovers is linked to human capital availability through a specific elasticity of total factor productivity with respect to cumulative spilllovers, which is a function of the human capital stock. Spillovers are proxied by applying the bilateral imports shares to the R&D expenditures by country.

All parameters related to the specification of endogenous total factor productivity growth (i.e. learning by doing, learning by research, knowledge spillovers, human capital) are estimated using advanced econometric techniques with cross country data for the EU member states for the period 2005-2016 and data for R&D expenditures for China, USA, Korea, Japan and Russia.

Public R&D expenditures are assumed to increase the global stock of knowledge by sector.