Sectoral specialisation in the EU a macroeconomic perspective



4 ANNEXES


Table 15 Business cycle component- sensitivity analysis (cont’)

I

Public services

1980-2001

1980-1991

1991-2001

Belgium

0,66

[0,58; 1,07]

0,69

[0,69; 1,36]

0,71

[0,51; 0,97]

Denmark

0,67

[0,67; 0,95]

0,53

[0,49; 0,92]

0,95

[0,92; 1,36]

Germany

0,28

[0,19; 0,31]

0,33

[0,15; 0,33]

0,35

[0,25; 0,38]

Greece

na

na

na

Spain

0,51

[0,47; 0,60]

0,29

[0,21; 0,57]

0,58

[0,56; 0,61]

France

0,41

[0,38; 0,46]

0,70

[0,60; 0,77]

0,28

[0,25; 0,34]

Ireland

na

na

na

Italy

0,32

[0,30; 0,34]

0,31

[0,24; 0,42]

0,33

[0,30; 0,38]

Luxembourg

na

na

na

Netherlands

0,41

[0,41; 3,66]

na

0,41

[0,41; 3,46]

Austria

1,33

[0,36; 1,35]

na

1,31

[0,38; 1,33]

Portugal

na

na

na

Finland

0,28

[0,23; 0,28]

0,08

[0,05; 0,08]

0,35

[0,31; 0,39]

Sweden

0,32

[0,28; 2,25]

0,24

[0,12; 2,16]

0,34

[0,31; 2,34]

United Kingdom

1,21

[1,17; 1,21]

1,02

[0,91; 1,02]

1,71

[1,71; 1,76]

Sources: Eurostat, NCBs, ECB calculations.

industries in total employment). The second
term (shift-effect) reflects the ability of a
country to move resources from low to high-
productivity sectors.

4.2.2 INDICATORS FOR BUSINESS CYCLE
ANALYSIS

4.2.2.1 EXTRACTING BUSINESS CYCLE
COMPONENTS

There are several competing approaches on how
to measure the business cycle.56 While potential
output growth measures would in principle be
available, band-pass filters provide a more
flexible treatment of short-term fluctuations and
have been extensively used in the literature. In
this report, the business cycle component was
calculated using a band-pass filter with the
Burns-Mitchell parameter set at 6, 32, 12, i.e.
the business cycle component contains all
fluctuations with a period of between 1.5 and 8
years.57 As band-pass filters require the loss of
data at the beginning and the end of the series,
the quarterly gross value added series were
extended by three years backward and forward
in time using an AR(5) process. The business
cycle component for the monthly data presented
in 3.2.2 was constructed using a similar
methodology by applying a band-pass filter
with parameters 18, 96, and 36 on the monthly
real industrial production values, including all
fluctuations with a period between 1.5 and 8
years in the monthly business cycle component.

In order to evaluate the sensitivity of the results
with respect to the chosen window, the series
were filtered using different parameter settings,
ranging from 1 year to 11 years, thereby
covering all standard business cycle lengths
(i.e. Kitchin-cycles with a periodicity of
approximately 3 years and Juglar-cycles with a
periodicity of 7-11 years) but leaving out the
seasonal fluctuations. The results of the
minimum and maximum values for relative
sectoral output volatility that the selection of
different periods yield are presented in Table
15. The sensitivity analysis shows that, while
the use of different parameters for the band-
pass filter will have an impact on the volatility
measure, the sensitivity of the measured relative
volatility remains limited with respect to the

56 For an overview of the different approaches used to determine
potential output, see C. Giorno, P. Richardson, D. Roseveare and
P. van den Noord (1995), “Estimating potential output, output
gaps and structural budget balances”, OECD Economics
Department Working Paper, 152.

57 Compared with other available filters, the Baxter-King filter,
which is now widely used in the field, has simultaneous
advantages in that it removes the unit root, respects the phase
and isolates cycle frequencies without re-weighting past
frequencies (see the discussion in A.-M. Agresti and B. Mojon
(2001), “Some stylised facts on the euro area business cycle”,
ECB Working Paper no 95, Appendix 1).

ECB

Occasional Paper No. 19
July 2004



More intriguing information

1. Unilateral Actions the Case of International Environmental Problems
2. Wirkt eine Preisregulierung nur auf den Preis?: Anmerkungen zu den Wirkungen einer Preisregulierung auf das Werbevolumen
3. The name is absent
4. The name is absent
5. Biological Control of Giant Reed (Arundo donax): Economic Aspects
6. Experience, Innovation and Productivity - Empirical Evidence from Italy's Slowdown
7. Biologically inspired distributed machine cognition: a new formal approach to hyperparallel computation
8. Automatic Dream Sentiment Analysis
9. Quality practices, priorities and performance: an international study
10. THE ANDEAN PRICE BAND SYSTEM: EFFECTS ON PRICES, PROTECTION AND PRODUCER WELFARE