On the Predictive Content of Production Surveys:

1
On the Predictive Content of Production Surveys

• A Pan-European Study

2
The Research Question (1)
Production Expectations

• 68,000 companies and 27,000 consumers
• Ex-ante judgments about the future industrial
  production levels, i.e. Production Accounts
  (OECD)

Predictive Content
3
(No Transcript)
4
The Research Question (2)

• ? Predictive Content of
• able to predict better than simple
  extrapolation

5
Literature Review

• Bergström (1995), Lindström (2000), Öller and
  Tallbom (1996), Teräsvirta (1986)
• Significant predictive content in Finland and
  Sweden
• Hanssens and Vanden Abeele (1987)
• No systematic predictive content in Belgium,
  France, Germany, The Netherlands and Italy

Ongoing Debate
6
Our Contribution

• Previous Research
• Limited set of countries and data range
• Univariate, within-country analysis
• No correlation between countries
• Bivariate tests
• Country by country predictive content

• Our Contribution
• 12 EU countries, 216 observations until 2002
• Multi Cross-country analysis
• Allow for correlations
• Bivariate Multivariate tests
• European (cross-country) predictive content

7
(No Transcript)
8
Bivariate Granger-causality tests
9
Haugh Test Results
significant at the 5 probability level
10
Haugh Test Results
significant at the 5 probability level
11
Haugh Test Results
significant at the 5 probability level
12
Our Contribution

• Previous Research
• Limited set of countries and data range
• Univariate, within-country analysis
• No correlation between countries

• Our Contribution
• 12 EU countries, 216 observations until 2002
• Multi Cross-country analysis
• Allow for correlations

13
Our Contribution

• Previous Research
• Limited set of countries and data range
• Univariate, within-country analysis
• No correlation between countries
• Bivariate tests
• Country by country predictive content

• Our Contribution
• 12 EU countries, 216 observations until 2002
• Multi Cross-country analysis
• Allow for correlations
• Bivariate Multivariate tests
• European (cross-country) predictive content

14
Multivariate Granger-causality tests
Clout
European
Receptivity
15
Methodology

• Bivariate Analysis
• ARMA filtering
• Haugh (1976) test
• Hanssens (1980),
• Leeflang and Wittink (1992)

• Multivariate Analysis
• VAR filtering
• El Himdi Roy (1976) test
• This paper

• Pre-filtering
• Granger-causality test
• Marketing applications

16
Pre-Filtering of the Data
17
El Himdi Roy Test (1997)
18
El Himdi Roy Test (1997) (2)
A is the asymptotic covariance matrix of

19
El Himdi Roy Test Results

• P-value 0.000
• When considered all together,
• significant predictive content of the 12
  European Production Expectations.

European
20
Clout Receptivity

• Adaptation of the El Himdi and Roy test
• Other filtering
• Part of the cross-correlations considered

21
Clout Receptivity Results

• High clout for France and Germany (p0.00)
• High receptivity of Austria (p0.00), Denmark
  (p0.03) and Finland (p0.03)
• Isolated position of UK and Ireland

22
General Conclusion

• About the question of the predictive content of
  the production expectations.
• High European predictive content
• Variability in clout and receptivity across
  Europe
• El Himdi and Roy in marketing
• ? other applications