European Commission 2005

1
European Commission Directorate General Economic
and Financial Affairs
Comparison of National versus European
Commission Confidence Indicators
Presentation by Maarten Van der Stadt Joint
EC-OECD Workshop on International Development of
Business and Consumer Tendency Surveys
Brussels, 14-15 November 2005
European Commission 2005
2
INTRODUCTION

• Joint Harmonised EU Programme of Business and
  Consumer Surveys
• Monthly and quarterly surveys
• Carried out at country level
• All EU Member States
• Harmonised questionnaire
• Institutes like IFO (Germany)
• INSEE (France)
• ISAE (Italy

European Commission 2005
3
SECTORS

• Industry
• Services
• Construction
• Retail
• Consumers (households)
• Overall-activity related indicator
• ESI

European Commission 2005
4
SCOPE

• National survey results
• Same basic data
• Sometimes different outcomes
• Factors behind the divergences

European Commission 2005
5
Content of presentation

• EC methodology
• Possible sources
• Effect of methodological difference

European Commission 2005
6
EC Methodology

• Balance statistics
• Three way questions B P N
• Five way questions B PP 0.5P 0.5N - NN
• Seasonal adjustment smoothing
• Dainties
• No smoothing

European Commission 2005
7
EC methodology (cntd)

• Confidence indicators
• Questions used
• Simple average
• ESI
• Same questions
• Standardised to mean 0, stadev 1
• Fixed weights

European Commission 2005
8
Possible sources of differences

• Rounding
• Quantification
• Seasonal Adjustment
• Smoothing
• Selection of questions
• Weighting scheme
• Standardisation
• Sector coverage

European Commission 2005
9
Analysis

• Graphical comparison level series
• Correlation coefficient level series
• Graphical comparison month on month change
• Correlation coefficient m-o-m change
• Concordance of direction of change

European Commission 2005
10
Examples

• Finish CCI s.a. vs non-s.a. indicator
• Danish ICI Different method of s.a.
• Italian ICI Different method of s.a.
• Hungarian CCI Different questions
• French ICI Combination of differences
• German ESI vs IFO BCI
• German ICI vs IFO BCI

European Commission 2005
11
Finish CCI (seasonal adjustment)
European Commission 2005
12
Finish CCI (cntd)

• Good fit at levels (correl 0.94)
• M-o-m changes
• Less strong correspondence (cc 0.83)
• 16 of cases different direction of change
• S.A. has noticeable effect on m-o-m changes

European Commission 2005
13
Danish ICI (seasaonal adjustment)
European Commission 2005
14
Danish ICI (cntd)

• High correspondence at level (cc 0.96)
• M-o-m change
• Less correcpondance (cc 0.83)
• 9 of cases different direction of change
• Surprisingly high, but smaller than non-s.a.

European Commission 2005
15
Italian ICI (seasonal adjustment)
European Commission 2005
16
Italian ICI (ctnd)

• Levels nearly identical (cc 0.99)
• M-o-m changes
• Still fairly high correspondence (cc 0.90)
• 14 of cases difference direction of change
• Surprisingly high number in light of high
  correspondence

European Commission 2005
17
Hungarian ICI (different questions)
European Commission 2005
18
Hungarian CCI (cntd)

• Level series nearly identical (cc 0.99)
• M-o-m changes
• Correlation declines to 0.86
• 8 of cases different direction of change

European Commission 2005
19
French ICI (several method. Diff.)
European Commission 2005
20
French ICI (cntd)

• High correspondence at levels (cc 0.97)
• M-o-m change
• Much lower correspondence (cc 0.72)
• Nearly 1 in 4 cases different direction of change

European Commission 2005
21
Ifo indicator EC ESI for Germany
European Commission 2005
22
Ifo indicator EC ESI (cntd)

• Level
• Cc of 0.88 surprisingly high
• EC clearly lags the IFO
• M-o-m- changes
• Very low correspondence (cc 0.46)
• 1/3 of observations do not coincide in d.o.c

European Commission 2005
23
Ifo indicator EC ICI for Germany
European Commission 2005
24
Ifo Indicator EC ICI (cntd)

• Level visually fits better, but cc still 0.88
• M-o-m change
• cc somewhat higher than for ESI (0.58)
• Smaller share (0.25) point in different direction
• ICI fits better then ECI
• Considerable differences

European Commission 2005
25
CONCLUSIONS

• At level series effect of methodological
  differences is relatively small
• Effect on m-o-m changes can be considerable
• Seasonal adjustment
• S.A. has noticeable effect on m-o-m change
• Effect of method of s.a. is smaller, but by far
  not negligable
• Selected questions
• Large overlap no strong effect
• Small overlap more substantial deviations
• Different sector coverage in OAI can lead to
  substantial difference
• Number of dimensions
• The more dimensions differ, the more
  discrepancies.

European Commission 2005