Summary Innovation Index

The SII is calculated using re-scaled values of the indicator data, where the highest value within the group of EU25 countries, Iceland, Norway and Switzerland is set to 1 and the lowest value within the group of EU25 countries to 0. For Bulgaria, Croatia, Romania, Turkey, the US and Japan for those cases where the value of an indicator is above the maximum or below the minimum the re-scaled value is set equal to 1 respectively 0. The SII is then calculated as the average value of all re-scaled values and is by definition between 0 and 1 for the EU25 countries. The 2005 Methodology report provides a more detailed explanation.

The SII values for HR, TR, US and JP are estimated as for these countries available data was insufficient to calculate the SII directly. For the US, data are available for only 15 indicators, Japan for 16 indicators, Turkey for 14 indicators and Croatia for 13 indicators. The SII for these countries was computed as follows:

• Step 1) For all countries an SII is calculated using only data for the 18 non-CIS indicators (“non-CIS SII).
• Step 2) An OLS regression for the EU countries, Iceland, Norway and Switzerland was run with the non-CIS SII from Step 1 as the dependent variable and the 2006 SII as the independent variable. The estimated regression coefficient equals 1.1204 and the estimated constant -0.0054.
• Step 3) The parameter values from Step 2 were then used to compute a 2006 SII estimate for HR, TR, US and JP by substituting the value as computed in Step 2 in the regression equation as follows: SII = (non-CIS SII + 0.0554)/1.1204.

SII growth rate

The growth rate of the SII is calculated differently than in previous years. First, hypothetical composite innovation indices have been calculated for the 4 years prior to the SII. The growth rate of the SII is then calculated as the annual percentage change between the SII and the average over the preceding three years, after a one-year lag. The three-year average is used to reduce year-to-year variability; the one-year lag is used to increase the difference between the average for the three base years and the final year and to minimize the problem of statistical/sampling variability.

In previous scoreboards, changes in the value of the CIS indicators over time were not included in the calculation of the growth rates. This year all CIS indicators are included, but for 5 of these indicators stationary time series have been used for various reasons. For the share of SMEs innovating in-house CIS4 data are not available so, if available, CIS3 data have been imputed for all 5 years. For the share of SMEs having introduced an organisational innovation, we impute CIS4 data for all 5 years when available. If CIS4 data are not available, we impute, if available, CIS3 data from last year’s indicator for the share of SMEs having used non-technological change. For three indicators (the share of innovative SMEs co-operating with others and the sales shares due to both new-to-market and new-to-firm products) for several countries differences between CIS3 and CIS4 results are unexpectedly large and for these indicators stationary CIS data are used, either CIS4 results if available or CIS3 results if CIS4 results are not available. For the share of enterprises receiving public funding for innovation and the turnover share of innovation expenditures, CIS3 and CIS4 results are both included in the calculation of the SII values over the 5-year time period and thus the SII growth rate, except for Portugal for both indicators and for Slovakia for innovation expenditures due to unexpected relatively large changes between CIS3 and CIS4 results.