Decomposing the Europe 2020 Index

See, for example, Freudenberg (2003) for extensive discussion.

The normalised indicators for each country j at time t for objective i, denoted by \( NX _{j,i}^t\), are defined as follows: \( NX _{j,i}^t\) is bounded from below by zero and from above by one. If \( NX _{j,i}^t=1\), it means that country j at time t has the best performance on objective i. Lower values indicate worse performances.

In particular, Tertiary education is measured as the percentage of people, for the age group 30–34, that have successfully completed university or university-like education (availability: 2003–2015); R&D is measured as the ratio of the expenditure of R&D over GDP (availability: 2003–2015); Greenhouse gas emissions is given as an index taking 1990 as the base year (availability: 1990–2015); Renewable energy is measured as the share of renewable energy in gross final energy consumption (availability: 2004–2015); Energy efficiency is measured as the gross inland consumption of energy divided by GDP (availability: 2000–2015); Employment is measured as the rate of employment in the age group 20–64 (availability: 2001–2015); Early school leavers are measured as the percentage of the population aged 18–24 with, at most, lower secondary education and not in further education or training (availability: 2000–2015); and Poverty is measured by the people at risk of poverty or social exclusion (availability 2004–2015). Also, for some years and countries, there could have been a break in the estimation, low reliability issue, or difference in the definition. Hopefully, these issues are rather marginal, overall making the data good quality. When these cases occur, we recompute the composite indexes and their decomposition using two different strategies: linear regression and simple average. This has no important impact on the results, and on our main conclusions. As such, the results displayed in Sect. 3.2 are those using the data taken from Eurostat.

At this point, it should be clear that the Europe 2020 index is also obtained as a geometric average of the sub-indexes. The same strategy is used in Pasimeni (2012, 2013), and Pasimeni and Pasimeni (2016).