ASPECTS OF ESTIMATION PROCEDURES AT EUROSTAT
WITH SOME EMPHASIS IN THE OVER SPACE HARMONISATION
Tzavidis
Nikolaos
Supervisor: I. Panaretos
CHAPTER
1
INTRODUCTION
1.1 Introduction
1.2 Micro-Aggregation Techinques
1.3 Backward Calculation Techniques1.4 Sampling Procedures in Eurostat and in Member States
MICRO-AGGREGATION TECHNIQUES
2.1 Introduction
2.2 Confidentiality Techniques
2.3 Micro-Aggregation Methods2.4 Methods Applicable to Quantitative Variables
2.4.1.1 Sorting by a Single Axis
2.4.1.2 Sorting by the First Principal Component
2.4.1.3 Sum of Z-Scores
2.4.2 Classification Methods
2.4.2.1 Adaptation of Hanani's Algorithm
The n x k-Grouping Problem in Rm
The 2 x k-Grouping Problem in Rm
The Hyperplane in the 2 x k-Grouping Problem in Rm
The Algorithm
2.4.2.2 Adaptation of Ward's Method
2.4.3 Individual Methods
2.4.3.1 Individual Ranking Method
Estimation of Variance Loss Due to Micro-Aggregation
by the Individual Ranking Method
2.4.3.2 Weighted Moving Average Method
2.5 Methods Applicable to Qualitative Variables
2.5.1 The Method of the Snake
2.5.2 Calculation of Entropy
2.5.3 Similarity of Distributions
2.6 Evaluation Criteria
2.6.1 Measuring Criteria of Data Confidentiality
2.6.1.1 Value of the Threshold "k"
2.6.1.2 Concetration or Predominance Rule
2.6.1.3 Indicator of Data Perturbation Before and After Micro-Aggregation
2.6.2 Criteria for Evaluating the Maintenance of the Structure
2.6.2.1 Use of Summary Statistics
2.6.2.2 Loss of Information Criteria
2.6.2.3 Further Processing Ability
2.7 Procedures in Eurostat and in European Countries
2.8 Concluding Remarks
BACKWARD CALCULATION TECHNIQUES
3.1 Introduction
3.2 The Introduction of EURO, the 1993 System of National Accounts Regulation and
the 1995 European System of Accounts Regulation. Three Cases Where Backward
Calculation is Required
3.2.1 Backward Calculation Due to the Introduction of EURO
The Problem
Solution
3.2.2 Backward Calculation Due to the Introduction of ESA (European System of Accounts)
1995 Regulation and SNA (System of National Accounts) 1993 Regulation
The SNA Problem
The ESA Problem
3.3 An Overview of Backward Calculation Techniques
3.3.1 Annual Backward Calculation
Full Revision Method
Revision by Superposition of Corrections
Growth Rates Method
Simple Proportional Method
3.3.2 Benchmark Years and Interpolation
The Benchmark Years
Interpolation
Full Benchmark Years
Layer Correction
3.4 The Netherlands Case
3.5 The French Case
3.5.1 The Kalman Filter
The State-Space Represantation of a Dynamic System - Maintained Assumptions
Derivation of the Kalman Filter
Starting the Recursion
Forecasting yt
Updating the Inference About ξ t
Producing a Forecast of ξ t+1
Summary of the Kalman-Filter Process and Remarks
Using the Kalman-Filter to Evaluate the Likelihood Function-Asymptotic Properties
of MLE Estimators
Smoothing
The French Model
Software for the French Retropolation Method
3.6 Cocnluding Remarks
ASPECTS OF SAMPLING PROCEDURES
4.1 Introduction
4.2 Stratified Random Sampling
4.2.1 Description of the Method
4.2.2 Proportionate Sampling of Elements. Properties of the Estimators
- Variance of the Estimators
4.2.3 Disproportionate Sampling of Optimum Allocation
4.2.4 Forming the Strata
4.2.5 Increasing the Efficiency of the Sample
4.3 Systematic Sample
4.3.1 Variance of the Estimated Mean in Systematic Sampling
4.3.2 Stratified Systematic Sampling
4.4 Cluster Sampling
4.4.1 Sub-Sampling With Units of Equal Size
4.4.2 Two-Stage Sampling
4.4.3 Three-Stage Sampling
4.4.4 Stratified Sampling of the Units in the Two-Stage Sampling
4.4.5 Sub-Sampling With Units of Unequal Size
Units Selected With Equal Probabilities
Units Selected With Probability Proportional to a Measure of Size
Units Selected With Probability Proportional to a Measure of Size Ration to Size Estimate
4.4.6 Stratified Sampling of the Units in Two-Stage Cluster Sampling With Unequal Clusters
4.4.7 Application of Multistage Cluster Sampling in European Surveys
4.5 Post Stratification or Stratification After Selection
4.5.1 Post Stratified Non-Response Adjustment
4.6 Raking Ratio Adjustment
4.6.1 Raking Non-Response Adjustment
4.7 Sample Surveys in the European Union
4.7.1 The Labour Force Survey (LFS)
4.7.1.1 The Objectives of the Labour Force Survey
4.7.1.2 The Developmente of the European Union Labour Force Survey - Attempts
for the Harmonisation of the Concepts
4.7.1.3 Technical Features of the European Union Labour Force Survey
4.7.1.4 Sample and Weighting Procedures in Member States and Greece Concerning the
Labour Force Survey
4.7.2 The European Community Household Panel (ECHP)
4.7.2.1 Objectives of the European Community Household Panel
4.7.2.2 Outline of the Design
Multi-Dimensional Coverage
Cross-Sectional Comparability
Longitudinal or Panel Design
4.7.2.3 Sampling Aspects of the First Wave
Sampling Frame Sampling Size and Allocation
Sample Design and Selection
4.7.2.4 Recommendations for Cross-Sectional Weighting
A Step By Step Procedure
Desing Weights
Non - Response Weight
Weights Correcting the Distribution of Households
Weights Correcting the Distribution of Persons
Final Weights
4.7.2.5 Second and Subsequent Waves of the ECHP
4.7.2.6 Longitudinal Weighting
4.7.3 The Household Budget Survey (HBS)
4.7.3.1 The Objectives of the Household Budget Surveys
4.7.3.2 Attempts for the Harmonisation of the Concepts
4.7.3.3 Sampling Aspects of the Household Budget Survey
4.7.3.4 Recommendations Concerning the Weighting Procedures
4.7.3.5 Procedures Followed in Greece Concerning the Household Budget Survey
4.8 Concluding Remarks