Building up indices for enablers and outcomes of digital transformation
The synthesis of selected national indicators reflecting enablers and outcomes of digital transformation in Europe was carried out through the development of a series of rank indices; these indices focus on the geographic scope of digital transformation and provide depth to the analysis of enabling conditions for digital transformation at EU-level and for individual Member States. The resulting indices are used to perform an EU-level analysis and monitoring of digital transformation; in addition, this set of indices is also used to support the comparative analysis of individual Member States enabling conditions and outcomes.
A Digital Transformation Enablers' Index (DTEI) and a Digital Technology Integration Index (DTII) were established to identify the performance of Member States in terms of enabling conditions and transformation experienced by means of a positioning score comprised between 0 and 100. These scores allow the analysis of the way countries are distributed in terms of enabling and outcome conditions with respect to each other and with respect to the EU 28 average score.
Methodology used for the construction of the indices
The construction of an index per enabling condition and outcome dimension was based on the analysis of data availability and coverage for each Member State. Many indicators across enabling conditions are available until 2015, with a few exceptions mainly in the dimensions of e-leadership, entrepreneurial culture, and supply and demand of digital skills. Furthermore, the availability of data for outcome indicators varies mostly between 2010 and 2015. The construction of the indices followed rules on data imputation and transformation to overcome availability issues and ensure comparability across countries. In addition, following the application of the imputation procedure, two other steps were carried out in the construction of all rank indices (one per each enabling and output dimension): a data transformation process to limit the effects of extreme or outlying values, and a min-max normalisation process to define the rank index within a common scale for each Member State.
Analysis of data availability and coverage and imputation on missing values
This first step highlighted the fact that several of the national indicators present limitations regarding data coverage. In order to overcome such limitations, a strategy of data imputation was adopted; this strategy allowed the construction of the set of indices based on the following rules:
- Limitations regarding missing information for individual indicators across Member States (where no information was available) were overcome by imputing the EU average of the indicator calculated using the set of available values for the specific indicator.
- Limitations regarding data availability for a single year for a given Member State were overcome by imputing the data of its closest year available, and whenever data was available for both adjacent years the average between the two available years was retained.
Data transformation to limit the influence of extreme values and standardisation to ensure comparability The second step following analysis and imputation involved a process of data transformation to control for outlying values present in the different groups of indicators.
A min-max normalisation procedure was selected and performed in order to allow country comparisons by means of a common range between 0 and 1. The implementation of this procedure requires careful attention to the treatment of extreme or outlying values which can induce a distortion to the resulting rank index.
Within this context, the effects of extreme or outlying values were addressed by transforming the dataset of indicators through the application of a Winsorisation method which limits and reduces the impact of outlying data (OECD, 2008).
Following this transformation, the normalisation process was engaged to ensure comparability of scales in the construction of the indices. As described above, the selected method implemented to normalise the information is a min-max standardisation procedure, which measures the distance from the value of a given Member State to the minimum value in EU28, in the context of a specific indicator.
In essence, the min-max normalisation process provides a ratio of the distance between a given observation and the minimum, with respect to the overall distance between the minimum and maximum observed indicator values in EU28. This ratio is observed as a percentage and ensures comparability providing the rank of the Member States within EU28 in the context of the specific indicator. As a result, rank index per framework condition was computed.