Work Package 4

Work Package 4 - Time series and multivariate analysis

Lead participant: SOTON


Start month: May 2018

End month: October 2019


This work package has the objective to improve the data dimension by means of time series analysis as well as discontinuity models. At the same time traditional multivariate analysis will be used to explore maps of European countries and regions depending on the traditional database available MIP, extended MIP, Europe 2020, SDGs and possible combinations and extensions as emerged by WP1.
Multivariate time series modelling will be fruitful especially if auxiliary series derived from big data sources in WP2 that are generally observed at higher frequency are combined with series obtained from repeated surveys. In addition it can be used as a form of small area estimation by borrowing strength over time and space. This can be utilized to improve the timeliness of indicators by making more precise first releases, sometimes called nowcasting. Moreover multivariate structural time series models will be considered to combine series of SDGs and welfare with auxiliary series at different frequency.
The second goal of this WP is the identification of the maps of European countries and regions depending on the different frameworks applied (MIP, SDGs). By means of multivariate analysis we shall provide the relation between the classification of countries and the indicators used.
This work package has also the objective to explore the feasibility of a micro database based on the integration of the actual survey data stemming from different sources and countries.
For some time now National Statistical Institutes in Europe have been reviewing approaches to streamlining survey design and data collection processes across multiple surveys that on many instances collect information on similar topics. Such a streamlining process is likely to create breaks in the series of survey estimates known as survey discontinuities.
It is important to demonstrate to the users that the design and methodology for the new survey is appropriate and good quality, and to produce estimates of the change in the series (discontinuity) caused by the change from one design to another. In addition, there is need for estimating survey discontinuities at disaggregated geographical areas (domains). A best practice will be derived according to the proposed database.

Description of work:

The main specific tasks within WP4 aimed at achieving the objectives are the followings: 

Task T4.1: Multivariate time series models for sustainability and welfare indicators.
Starting from the database developed in WP2, this task develops multivariate time series models for Ssustainability Development Goal indicators (SDG’s) and welfare indicators. Principal goals of the tasks will be nowcasting, detecting common components in the indicators and mixed frequency models.

Task T4.2: This task provides a multivariate framework able to identify maps of European countries and regions depending on the different frameworks applied: traditional MIP, extended MIP, Europe 2020, SDGs and possible combinations and extensions as emerged by WP1. This analysis will permit an assessment on the impact of the well-being and SDG indicators of the maps.

Task T4.3: Study alternative estimators of discontinuities and analysis of availability of data to solve the problems in the data concatenation surge in the database. By means of the techniques developed in this task will be able to expand our dataset dimension. 


D4.1: Report on nowcasting and mixed frequency model for the integrated analysis of well-being and SDGs.

D4.2: Report on multivariate analysis on MIP and well-being and SDGs indicators.

D4.3: Report on alternative estimators of discontinuity.


Events calendar