User Manual for Software BV4.1, Version 2.

1
(Program Version in English)

June 2015

Federal Statistical Office

1
 Imprint

Published by: Statistisches Bundesamt (Federal Statistical Office), Wiesbaden

Distribution and Subject-related information General information

on the software BV4.1: on the data supply:

Statistisches Bundesamt Information service,

Gruppe Mathematisch-Statistische Methoden tel: +49 (0) 611 / 75 24 05
Referat C 101 fax: +49 (0) 611 / 75 33 30
65180 Wiesbaden info@destatis.de
Germany www.destatis.de
bv4.1@destatis.de
Release calendar
of the Press Office:
http://www.destatis.de/press

Published in June 2015

Statistisches Bundesamt, Wiesbaden 2015

All rights reserved.

2
Contents
1 Introduction .................................................................................................... 1
1.1 Codes and definitions ..................................................................................... 1
1.2 Brief outline of the mathematical strategies of the BV4.1 procedure............... 1
1.2.1 Calendar regressors ........................................................................................ 2
1.2.1.1 Trading day component ................................................................................... 3
1.2.1.2 Working day component (not counting Saturdays) .......................................... 3
1.2.1.3 Working day component (counting Saturdays) ................................................ 4
1.2.2 Level shifts ...................................................................................................... 4
1.3 Special qualities of the BV4.1 procedure......................................................... 4
1.4 Short form for characterising BV4.1 analyses.................................................. 4
1.5 References ...................................................................................................... 5
2 The BV4.1 Software ......................................................................................... 6
2.1 Capabilities and program limits ...................................................................... 6
2.2 Technical requirements and instructions for installation ................................. 7
2.3 Structure of the program folder ....................................................................... 7
2.4 Adaptation of the calendar adjustment ........................................................... 8
3 Input Files ....................................................................................................... 11
3.1 Requirements to be met by the input files containing
the time series to be analysed......................................................................... 11
3.1.1 EXCEL files ....................................................................................................... 11
3.1.2 CSV files .......................................................................................................... 13
3.1.3 ACCESS files and SQL Server files ................................................................... 13
3.2 Requirements to be met by the input files containing
up to 50 user-defined regressor variables ....................................................... 13
3.2.1 EXCEL files ....................................................................................................... 14
3.2.2 CSV files .......................................................................................................... 14
3.2.3 ACCESS files and SQL Server files ................................................................... 14
4 Output Files ..................................................................................................... 15
4.1 Txt output files with BV4.1 analysis results ..................................................... 15
4.2 Txt output files with spectrum analysis results................................................ 16
4.3 Output option "integrated, complete" ............................................................. 16
4.4 Output option "integrated, reduced" ............................................................... 17
4.5 Output option "separated (reduced)" .............................................................. 17
4.6 Output files with diagrams .............................................................................. 18
4.7 Naming conventions for output files................................................................ 18
5 Running BV4.1 ................................................................................................ 20
5.1 Front screen page ............................................................................................ 20
5.2 Main menu screen page .................................................................................. 21
5.3 Specification of the decimal separator, the column separator for CSV files
and the data path for the input and output files / Selection
of the input file containing the time series to be analysed .............................. 23
5.4 Selection of the time series to be analysed and
specification of the analysis periods ............................................................... 25
5.5 Manual specification of the analysis options .................................................. 26
5.5.1 Presetting of level shifts.................................................................................. 28
5.5.2 Selection of user-defined regressors ............................................................... 29

3
5.6 Successive analyses ....................................................................................... 30
5.7 Result screen page .......................................................................................... 31
5.8 Generating diagrams ....................................................................................... 32
5.8.1 Standard diagrams .......................................................................................... 33
5.8.2 Yearly curves of a component .......................................................................... 34
5.8.3 Diagram of the results of successive analyses ................................................ 35
5.8.4 Diagram of power spectra ............................................................................... 36

4
1 Introduction
In Germany, the decomposition and the seasonal adjustment of economic time series with the BV
procedure have a long tradition. The mathematical bases were developed in the late sixties at the
Berlin Technical University and the German Institute for Economic Research (DIW) (Nullau, Heiler
et al. (1969)). Shortly after (1972), the Federal Statistical Office (DESTATIS) established a first
practicable version of the procedure to provide the general public with information on trends and
seasonally adjusted data of major business-cycle indicators. From 1983 the BV4 version of the
procedure was used, a further development by DESTATIS (Nourney, M. (1983), (1984)). BV4
turned out to be very suitable for German official statistics.

In the course of 2004, BV4 was replaced by the new version BV4.1 with methodological
improvements concerning the estimation of outliers and calendar effects. In addition, the user
can now specify explanatory variables, which are to be considered with the analysis (Speth, H.-
Th. (2004)).

1.1 Codes and definitions

O = original series,
T = trend-cycle component (without level shift component),
S = seasonal component,
CA = calendar component,
U + = user-defined component (including level shifts),
U = user-defined component (without level shifts),
LS = level shift component,
EX = outlier component,
R = residual component,
TC =T + LS =trend-cycle component,
SADJ = O S CA U = seasonally adjusted series,
CADJ = O CA U = calendar adjusted series,
I = EX + R = irregular component.

1.2 Brief outline of the mathematical strategies of the BV4.1 procedure

With BV4.1, it is possible to analyse monthly and quarterly time series.

The first part of the BV4.1 procedure is the identification of (potential) additive outliers. It is
based initially on the assumption that within sufficiently short moving time intervals with fixed
length M - the so-called basic spans - the time series is the realisation of a Normal stationary
process. That way the conditional distributions of the observations (data) from the time series
directly left and/or right from outside of the particular basic spans (backward and forward
identification of outliers) are determined. If the difference between an observation and the thus
determined (conditional) expected value is more than a certain fixed multiple TAU (sigma limit) of
its standard deviation, then the observation is regarded as an outlier.

In the second part of the procedure the integrated estimation of outliers, calendar effects and of
the effects of series specific user-defined variables is accomplished. Based on the general
additive model for time series decomposition

1
 O =T + S + CA + U + + EX + R ,

this is done using the following linear regression model

h k l m n
O= iTi + i Si + iCAi + iU i+ + i EX i +
=i 1 =i 1 =i 1 =i 1 =i 1

( Ti = trend-cycle regressors, Si = seasonal regressors, CAi = calendar regressors, U i+ = series

specific user-defined regressors, EX i = series specific outlier dummy regressors according to
the outliers identified at step 1, = error term) but "filtered" by the linear BV4.1 filter procedure
F for trend-cycle and seasonal adjustment (cf. below). This leads to the model

h k l m n
F (O) = F iTi + F i Si + i F (CAi ) + i F (U i+ ) + i F ( EX i ) + * ,
= i1 = i 1= i1 =i 1 =i 1

where * = F ( ) is the new error term. Because

h k
F iTi 0 and F i Si 0 ,
i =1 i =1

the model used for the estimation of the model parameters i , i , i (or the components CA ,
U and EX ) is
l m n
F (O) =
=i 1 =i 1 =i 1
i F (CAi ) + i F (U i+ ) + i F ( EX i ) + ** .
The method of estimating the parameters is that of ordinary least squares.

In the last part the procedure proceeds with the estimation of the trend-cycle and the seasonal
component of series O , based on the time series adjusted for outliers, calendar effects and the
effects of the user-defined variables O* =O CA U + EX
( = T + S + R* ). This is done using
fixed linear filters (fixed filter approach) derived from different component-specific local (moving)
regression models, where locally the trend-cycle and the seasonal component are approximated
by polynomials and trigonometric functions. The model parameters are estimated by the method
of weighted least squares. For details cf. Nourney, M. (1983), (1984).

1.2.1 Calendar regressors

With BV4.1, the following three regression models for calendar adjustment are directly available
to the user. In addition, it is possible to use other regressors choosing suitable user-defined
variables (cf. 3.2).

2
1.2.1.1 Trading day component

Starting point of the trading day adjustment is a modelling approach for the trading day
component considering the differing day-of-the-week compositions of different months or
quarters:

8
Ct = i dt (i ) ,
i =1

where
t= (individual) period (month or quarter),
dt (i ) = numbers of Mondays ( i = 1 ), ..., Saturdays ( i = 6 ) in period t which are not public
holidays,
number of Sundays ( i = 7 ),
number of public holidays, which are not Sundays ( i = 8 ),
i = regression coefficients ( i = 1,...,8 ).

Let dt (i ) denote the mean number of "day" i in the periods of the same name as t and d (i )
denote the mean number of "day" i in the period (month or quarter). Then

( ) ( )
8 8 8
Ct = i d (i ) + i dt (i ) d (i ) + i dt (i ) dt (i ) .
i =1 i =1 i =1

C1 C2 C3

Because C1 = const. , this calendar effect is assigned to the trend-cycle component. C 2 , the
length-of-period effect, is assigned to the seasonal component. So with BV4.1 the trading day
component is modelled by

( d (i) d (i) ) .
l 8
=CAt
= i CAti
i =1 i =1
i t t

1.2.1.2 Working day component (not counting Saturdays)

The model of the working day component (not counting Saturdays) is a variation of the model of
the trading day component. Now, all the weekdays from Monday to Friday which are not public
holidays have identical effects, and Saturdays, Sundays and public holidays have identical
effects:

2
Ct = i dt (i ) ,
i =1
where
t= (individual) period (month or quarter),
dt (i ) = number of the weekdays Monday to Friday in period t which are not public holidays
( i = 1 ),
number of Saturdays, Sundays and public holidays which are not Saturdays or Sundays
( i = 2 ),

3
i = regression coefficients ( i = 1 or 2 ).

In addition, it is assumed that 2 = 0 . Then similar considerations as in chapter 1.2.1.1 lead to

( )
CAt = 1 dt (1) dt (1) .

1.2.1.3 Working day component (counting Saturdays)

The model of this working day component essentially corresponds to the model of the working
day component not counting Saturdays. It only differs in the definition of dt (i ) :

dt (i ) = number of the weekdays Monday to Saturday in period t which are not public holidays
( i = 1 ),
number of Sundays and public holidays which are not Sundays ( i = 2 ).

1.2.2 Level shifts

Users often know the periods (months or quarters) where level shifts occur in the time series. To
improve the analyses it is useful to provide the procedure with this information. In these cases,
the first two parts of the BV4.1 procedure as described above (cf. 1.2), based on the original
series, are only used to produce provisional estimates of the level shifts. After that they are
repeated, but then the identification of outliers is done based on the provisionally level shift
adjusted series.

Regarding output files and diagrams, the BV4.1 procedure assigns level shifts to the trend-cycle
component.

1.3 Special qualities of the BV4.1 procedure

Efficient modelling of changing seasonal time series structures.

Trend-cycles are depicted plausibly in terms of economic points of view.
Low cost-benefit ratio. To produce high-quality results, the user of BV4.1 requires
neither a special training or even expert knowledge nor long-term experiences in dealing
with the procedure.
Due to the fixed filter approach, BV4.1 results on principle do not depend on the
respective user.
Unless outliers occur, components of partial series add up to the corresponding
component of the aggregate series (i.e. there is no difference between indirect and
direct analysis of aggregate series.)

1.4 Short form for characterising BV4.1 analyses

Regarding tables and diagrams of the results, the user is informed about the particular BV4.1
options by the following short form:

BV4 (P) E(E1/E2) C(C1) S(S1) R(R1),

4
where
BV4 = name of the procedure,
P = code letter for the variant of BV4.1,
M: variant to analyse monthly series,
Q: variant to analyse quarterly series,
E = indicator of the parameters of the outlier identification procedure,
E1 = length M of the basic span,
E2 = sigma limit TAU,
C = indicator of the option of the calendar adjustment according to 1.2.1,
C1 = code letter of the variant of the calendar component,
k or K: trading day component, German public holiday situation, where All Saints
Day respectively is or is not treated as a public holiday (cf. 2.4),
a or A: working day component (not counting Saturdays), German public holiday
situation, where All Saints Day respectively is or is not treated as a public
holiday (cf. 2.4),
w or W: working day component (counting Saturdays), German public holiday
situation, where All Saints Day respectively is or is not treated as a public
holiday (cf. 2.4),
C: trading day component, user-defined calendar input,
D: working day component (not counting Saturdays), user-defined calendar input,
E: working day component (counting Saturdays), user-defined calendar input,
S = indicator of the number of level shifts,
S1 = number of level shifts,
R = indicator of the number of user-defined regressors,
R1 = number of user-defined regressors.

1.5 References

Nourney, M. (1983). Umstellung der Zeitreihenanalyse. Wirtschaft und Statistik 11 (ed. Federal
Statistical Office), pp. 841-852.
Nourney, M. (1984). Seasonal adjustment by frequency determined filter procedures. Statistical
Journal of the United Nations ECE 2, pp. 161-168.
Nullau, B., Heiler, S. et al. (1969). Das "Berliner Verfahren" - Ein Beitrag zur Zeitreihenanalyse.
DIW-Beitrge zur Strukturforschung 7.
Speth, H.-Th. (2004). Komponentenzerlegung und Saisonbereinigung konomischer Zeitreihen
mit dem Verfahren BV4.1, Methodenberichte 3 (ed. Federal Statistical Office).

5
2 The BV4.1 Software
2.1 Capabilities and program limits

The BV4.1 software offers:

BV4.1 analyses of monthly and quarterly time series. The maximum length of the
analysis period is 360 periods (data). The minimum length is 60 periods for monthly
and 17 periods for quarterly series.
BV4.1 analyses with up to 15 user-defined explanatory variables.
3 options for modelling calendar effects (cf. 1.2.1 and 2.4).
Estimation of power spectra of time series, components and seasonal and calendar
adjusted series. Thereby the Parzen window is used. The truncation point is set at 2/3 of
the length of the particular analysis period.
User-friendly graphical user interfaces (GUI).
Support of the file formats
- CSV,
- EXCEL,
- ACCESS,
- SQL Server.
Possibility of mass production of time series decompositions and seasonal
adjustments. If there is a collinearity problem between the regressors, the software will
modify the choice of the regressors by itself to guarantee that the program runs
smoothly and inform the user about the modifications.
Possibility to execute so-called successive analyses (i.e. analyses where the analysis
periods are extended gradually by one additional period). This option is useful
especially when such revisions of results are to be examined which are due to the BV4.1
procedure itself.
Various possibilities of graphic evaluations of the BV4.1 analysis results and the
corresponding power spectra.
2 variants to run the analyses:
- using the input file to provide BV4.1 with information about the time series and the
analyses to be performed,
- using the input file to provide BV4.1 with information about the time series and using
the GUI to provide BV4.1 with the information on the analyses to be performed.
For each time series analysed the software produces two txt output files - one (cf. 4.1)
containing detailed information on
- the BV4.1 analysis parameters,
- the identified outliers,
- the regression coefficients,
- the complete results of the time series decomposition and the seasonal adjustment
and
- selected percentage changes,
and the other containing the results of the corresponding spectrum analysis (cf. 4.2).
It is to be taken into account that the given information concerning the number of
degrees of freedom of the test statistics for the significance of the coefficients and
components estimated according to chapter 1.2 could only be approximations.
Options for the production of output files intended for the processing of the results of
the time series analyses:
- series specific files with the complete results of the time series decomposition and
the seasonal adjustment plus selected percentage changes (cf. 4.3),

6
 - series specific files with selected components, adjusted series and percentage
changes (cf. 4.4),
- component specific files with the original series, the trend-cycle components, the
seasonally adjusted series and the calendar adjusted series of all the time series of
the respective run, including the respective percentage change rates (cf. 4.5).

2.2 Technical requirements and instructions for installation

Technical requirements:
Windows 7+,
Java Runtime Environment (JRE) 1.7.0+.

Processing file formats:

CSV,
EXCEL 2002+,
ACCESS 2002+ or
SQL Server 7.0+.

Installation:
For the installation of the BV4.1 software, the user has to run the program "install.exe". Now the
installation program starts and will guide the user through the installation.

The BV4.1 software is a Java program. So it is necessary to install a suitable Java version. If the
user wants, he can do this by the installation program, too. If the user doesn't want this, the
installation program will ask him to choose one of the JRE (Java Runtime Engine) the user has
already installed on his PC. It is important to note that BV4.1 requires at least JRE 1.6.0.

If the user wants to use the ODBC format "SQL Server" he has to install a DSN (Data Source
Name) named "BV41_SQL". The data base indicated has to contain all input files required for the
analysis (time series files and regressor files (cf. Ch. 3)).

2.3 Structure of the program folder

By the installation program, with the exception of the sub-folders "Test" and "Remarks" which are
generated during a program run, the following structure of the BV4.1 program folder (here e.g.
named BV4.1) is generated.

Except SQL Server input files, for which a specific SQL Server data base has to be used, files
containing the time series which are to be analysed have to be stored in the folder "Series", and
files containing the user-defined regressor variables have to be stored in the folder "Regressors".
Results of the BV4.1 analyses will be stored by the program in sub-folders of the folder "Results"
(e.g. folder "Test" below). Only SQL Server output files, intended for processing of the results (cf.
4.3 to 4.5), will be stored in the specific SQL Server data base (also containing the SQL Server
input files).

7
2.4 Adaptation of the calendar adjustment

The BV4.1 software particularly supports calendar adjustments according to the approaches
described in chapter 1.2.1. But without further actions, only the German public holiday situation
can be considered (public holidays: New Years Day, Good Friday, Easter Monday, May Day,
Ascension of Christ, Whit Monday, Corpus Christi, national holiday, All Saints Day (optional),
Penance Day (until 1994), Christmas Day, Boxing Day, New Years Eve), the calendar adjustment
is restricted to analysis periods between 1949 and 2050 and the calculations of the means
dt (i ) and d (i ) are based on the complete cycle of the dates of Easter (5700000 years) and the
related other public holidays.

For adaptation to other situations, the user has to provide the corresponding information for
monthly series in the files "BV4M_Weekdays_N_X.txt", "BV4M_Weekdays_MNM_X.txt" and
"BV4M_Weekdays_MNEM_X.txt", stored in the two folders named "Monate", and for quarterly
series in the files "BV4Q_Weekdays_N_X.txt", "BV4Q_Weekdays_MNQ_X.txt" and
"BV4Q_Weekdays_MNEQ_X.txt", stored in the two folders "Quartale" (cf. 2.3).

In the files "BV4M_Weekdays_N_X.txt" and "BV4Q_Weekdays_N_X.txt" have to be stored the

numbers dt (i ) for monthly and quarterly time series, respectively (cf. 1.2.1.1). In the files
"BV4M_Weekdays_MNEM_X.txt" and "BV4Q_Weekdays_MNEQ_X.txt" have to be stored the
numbers dt (i ) for monthly and quarterly time series, respectively, and in the files
"BV4M_Weekdays_MNM_X.txt" and "BV4Q_Weekdays_MNQ_X.txt" the numbers d (i ) for
monthly and quarterly time series, respectively (cf. 1.2.1.1). For program technical reasons the
files "BV4M_Weekdays_N_X.txt" and "BV4Q_Weekdays_N_X.txt" may cover only periods up to
200 years.

8
Structure of the data file "BV4M_Weekdays_ N_X.txt"
Record Position Contents
14 Arbitrary.
5 ff. 17 Date formatted as MM YYYY (e.g. 03 2001).
8 Blank.
9 ff. Corresponding numbers dt (i ) , i = 1, ..., 8 (separated by blanks).

Following these records, it is possible to add some notes. The text has to
start with the capital letter "N" (for "Note") at the first position of the record.

Structure of the data file "BV4M_Weekdays_MNEM_X.txt"

Record Contents
1-4 Arbitrary.
5 d (1) , t = January, ..., December (separated by blanks).
t
6 dt (2) , t = January, ..., December (separated by blanks).



12 dt (8) , t = January, ..., December (separated by blanks).
13 ff. Arbitrary.

Structure of the data file "BV4M_Weekdays_MNM_X.txt"

Record Contents
1-4 Arbitrary.
5 d (i ) , i = 1, ..., 8 , regarding months (separated by blanks).
6 ff. Arbitrary.

9
Structure of the data file "BV4Q_Weekdays_ N_X.txt"
Record Position Contents
14 Arbitrary.
5 ff. 17 Date formatted as QQ YYYY (e.g. 03 2001).
8 Blank.
9 ff. Corresponding numbers dt (i ) , i = 1, ..., 8 (separated by blanks).

Following these records, it is possible to add some notes. The text has to
start with the capital letter "N" (for "Note") at the first position of the record.

Structure of the data file "BV4Q_Weekdays_MNEQ_X.txt"

Record Contents
1-4 Arbitrary.
5 d (1) , t = 1st quarter, ..., 4th quarter (separated by blanks).
t
6 dt (2) , t = 1st quarter, ..., 4th quarter (separated by blanks).



12 dt (8) , t = 1st quarter, ..., 4th quarter (separated by blanks).
13 ff. Arbitrary.

Structure of the data file "BV4Q_Weekdays_MNQ_X.txt"

Record Contents
1-4 Arbitrary.
5 d (i ) , i = 1, ..., 8 , regarding quarters (separated by blanks).
6 ff. Arbitrary.

10
3 Input Files
3.1 Requirements to be met by the input files containing the time series to be analysed

After successful program installation, a number of example files are available at the sub-folder
"Series" of the folder "Data" (cf. 2.3). The following sections describe the structure and the
contents of the input files expected by the program.

3.1.1 EXCEL files

Column A:
Row 1: Headline "Record" (or "Satz").
Rows 2, 3, etc.: running number (identification number to be used by the program for
error referencing).

Column B:
Row 1: Headline "Period" (or "Zeitkennung"):
Rows 2 to 49: arbitrary.
Rows 50 and following: period in the EXCEL format "date". For monthly series the format
01.MM.YYYY has to be used, whereas periods for quarterly series have to be stored
according to the following rule: 01.02.YYYY for the first quarter, 01.05.YYYY for the
second quarter, 01.08.YYYY for the third quarter and 01.11.YYYY for the fourth quarter of
a year.

Column C:
Row 1: Headline "Description" (or "Beschreibung").
Row 2 and following: arbitrary (it is recommended to use this column for descriptions of
the contents of the rows of the following columns (cf. below)).

The contents of the following columns (each containing the data of a time series as well as the
particular specification of the options) are listed in the following table:

Row Content
1 Arbitrary (but empty entries are not allowed).
2-11 Descriptive information (e.g. title) on the time series (to be used in the output files).For
each row up to 72 characters are permissible.
12 Short name for the time series (from 1 to 7 characters, the first has to be a letter).

This short name serves for identification of the time series during the program run as well
as at the output files.
13 Integral part of the name of the txt output files (from 1 to 8 characters, the first has to be
a letter).
14 Start period of the analysis period (serial number of the period within a year).
15 Start year of the analysis period (four digits).
16 End period of the analysis period (serial number of the period within a year).
17 End year of the analysis period (four digits).

Remark: The analysis period has to cover 17 to 360 quarters or 60 to 360 months.

11
Row Content
18 Number of the decimals places of the figures shown in the output files (max. 6).

Remark:
Percent change rates, however; are tabulated to at least one decimal place.
19 Number of available observations (data of the time series) given below (row 50 and
following).
20 Sigma limit TAU for the outlier identification (0 < TAU < 9.99).

Usual specification: "3.00".

21 Control parameter and length of the basic span M for the outlier identification.

Remarks: No outlier adjustment: "0", or 3 < M for quarterly and 11 < M for monthly series.

Usual specification: "8" for quarterly and "24" for monthly series.
22 Control parameter for the calendar adjustment according to the approaches described in
the chapters 1.2.1 and 2.4.

No calendar adjustment: "0".

Trading day adjustment based on the German public holiday situation without
consideration of All Saints Day: "1".
Working day adjustment (not counting Saturdays) based on the German public holiday
situation without consideration of All Saints Day: "2".
Working day adjustment (counting Saturdays) based on the German public holiday
situation without consideration of All Saints Day: "3".
Trading day adjustment based on the German public holiday situation with consideration
of All Saints Day: "4".
Working day adjustment (not counting Saturdays) based on the German public holiday
situation with consideration of All Saints Day: "5".
Working day adjustment (counting Saturdays) based on the German public holiday
situation with consideration of All Saints Day: "6".
Trading day adjustment based on the user-defined calendar input: "7".
Working day adjustment (not counting Saturdays) based on the user-defined calendar
input: "8".
Working day adjustment (counting Saturdays) based on the user-defined calendar input:
"9".
23 Number of level shifts (max. 10) given in the following 10 rows.
24 Date of the 1st level shift, formatted as MMYYYY or QQYYYY (e.g. 032001), or "0", if no
level shift is given.
25 Date of the 2nd level shift, formatted as MMYYYY or QQYYYY, or "0", if no 2nd level shift is
given.



33 Date of the 10th level shift, formatted as MMYYYY or QQYYYY, or "0", if no 10th level shift
is given.
34 Name (1 to 8 characters, the first has to be a letter) of that file, whose columns contain
the user-defined regressor variables.

Remark: There has to be a name assignment even if the usage of user-defined regressors
is not intended.

12
Row Content
35 Column number of the 1st user-defined regressor variable or "0", if no regressor variable
is to be used.
36 Column number of the 2nd user-defined regressor variable or "0", if no 2nd regressor
variable is to be used.



49 Column number of the 15th user-defined regressor variable or "0", if no 15th regressor
variable is to be used.
50 Observation (date) of the time series, belonging to the period in column B, or blank.

Remark:
Empty entries (in-between given observations) are not allowed.

One EXCEL input file may consist of several EXCEL sheets built according to the above rules (e.g.
to read simultaneously monthly and quarterly series).

3.1.2 CSV files

Time series files in CSV format correspond in structure and content exactly to the EXCEL input
files described above (cf. 3.1.1). Note: Empty entries, which in practice may occur when no
observations are available for some time periods, have to be considered by putting a column
separator character.

3.1.3 ACCESS files and SQL Server files

The structure and the contents of time series files in ACCESS format and in SQL Server format
correspond in most parts to the EXCEL files described above (cf. 3.1.1). The difference is that the
first row of an EXCEL sheet, particularly containing the column headlines, is missing in the
corresponding ACCESS and SQL Server files. The headlines "Record" (or "Satz"), "Period" (or
"Zeitkennung") and "Description" (or "Beschreibung") have now to be placed as a column
description.

One ACCESS input file may consist of several ACCESS tables built according to the above
descriptions (e.g. to read in both, monthly and quarterly time series).

3.2 Requirements to be met by the input files containing up to 50 user-defined regressor

variables

After successful program installation, example files are available at the sub-folder "Regressors"
of the folder "Data" (cf. 2.3). These files contain the data of the regressor variables for trading day
adjustment (cf. 1.2.1.1) of quarterly or monthly time series for the period from 1991 to 2010
(according to the German public holidays situation without consideration of All Saints Day, cf.
2.4).

13
The names of the regressor files may be up to 8 characters long (not counting the file name
extension) and have to start with a letter.

3.2.1 EXCEL files

Column A:
Row 1: Headline "Record" (or "Satz").
Rows 2, 3, etc.: running number.

Column B:
Row 1: Headline "Period" (or "Zeitkennung").
Rows 2 to 49: arbitrary.
Rows 50 and following: period in the EXCEL format "date". For monthly series the format
01.MM.YYYY has to be used whereas periods for quarterly series have to be stored
according to the following rule: 01.02.YYYY for the first quarter, 01.05.YYYY for the
second quarter, 01.08.YYYY for the third quarter and 01.11.YYYY for the fourth quarter of
a year.

The following up to 50 columns are intended for containing the data of the regressor series:
Row 1: Identifier of the regressor variable (up to 10 characters long, starting with a letter).
The identifier is necessary for the selection of regressor variables at the GUI (cf. 5.5.2),
and it is used for labelling the corresponding regression tables within the txt output files
generated by the program.
Rows 2 and following: Data of the regressor series corresponding to the periods noted in
column B. For missing values, a "*" has to be put. Empty entries are not allowed, and the
maximum length of the numerical values of the regressor variables is 10 characters
(inclusive sign).

3.2.2 CSV files

Regressor input files in CSV format correspond in structure and contents exactly to the EXCEL
regressor files described above (cf. 3.2.1).

3.2.3 ACCESS files and SQL Server files

Structure and contents of regressor input files in ACCESS format and in SQL Server format
correspond in most parts to the EXCEL regressor files described above (cf. 3.2.1). The difference
is that the first row of an EXCEL sheet, particularly containing the column headlines, is missing in
the corresponding ACCESS and SQL Server files. The headlines "Record" (or "Satz") and "Period"
(or "Zeitkennung") and the regressor identifiers have now to be placed as a column description.

When working with SQL Server format the user has to store them together with the time series
files in one data base.

14
4 Output Files
For each program run a separate sub-folder is generated within the folder "Results", containing
the txt output files with detailed information about the BV4.1 analyses and the spectrum
analyses undertaken. The name of this sub-folder (such as "Test" in 2.3) is defined by the user at
the GUI (cf. 5.2). If during the program run important additional information concerning single
time series analyses are outputted by BV4.1, which the user should be aware of, an additional
sub-folder "Remarks" is also generated (cf. 2.3). For the convenience of the user, the txt output
files of all abnormal time series analyses are separately listed there once more.

The generated sub-folder also contains the output files which were chosen by the user by the
option "Output variants" (cf. 5.2), provided that they were exported in the formats EXCEL, CSV or
ACCESS (cf. 5.7). The ACCESS output files are always stored in the file Output.mdb. Output files
chosen by the user by the option "Output variants", but exported in SQL Server format, are always
stored in the data base, which was accessed by the system source BV41_SQL.

4.1 Txt output files with BV4.1 analysis results

For each BV4.1 time series analysis performed during one program run, the software generates a
txt output file, containing the following information:

Descriptive information on the time series, taken from the input file.
Where necessary: important additional information on abnormal analysis runs (cf.
above).
Information on the specification of the BV4.1 options used for the analysis.
Depending on the chosen options:
- Information on the outlier identification process.
(Note: If TAU is chosen too small, it is possible that a change of sign between a
potential outlier according to the identification process and the corresponding
regression estimate will occur. In this case, the observation will not be considered as
an outlier any more and the note "no replacement" is placed into the table named
"Information on the replacement of outliers".)
- Information on regression coefficients and corresponding test statistics.
(Note: The given numerical values for test statistics and degrees of freedom are
approximate values only. For consideration of the BV4.1 filter procedure F for
preliminary trend-cycle and seasonal adjustment, 15 degrees of freedom for monthly
series and 7 degrees of freedom for quarterly series are estimated.)
- Table containing results for the calendar component CA , on the length-of-the-period
effect ( L ) and on the total CA + L .
Table, containing the results of the time series decomposition and the seasonal
adjustment. In detail, these information cover the
- original series O ,
- annual percentage change rate O _ R of O ,
- trend-cycle component TC (= T + LS ),
- percentage change rate TC _ R of TC on the previous period,
- seasonally adjusted series SADJ (= O S CA U ),
- percentage change rate SADJ _ R of SADJ on the previous period,
- calendar adjusted series CADJ (= O CA U ),

15
 - annual percentage change rate CADJ _ R of CADJ ,
- seasonal component S ,
- calendar component CA ,
- user-defined component U ,
- level-shift component LS ,
- irregular component I (= EX + R ),
- outlier component EX ,
- residual component R and the
- effects of each user-defined regressor variable.

Periods for which outliers were identified or for which the user preset level shifts are
marked by "*" and "+", respectively.

For technical reasons, in the txt output file the given name of the input file is always "input.txt"
and the given format extension of the regressor file is always "txt".

4.2 Txt output files with spectrum analysis results

For each BV4.1 time series analysis performed during one program run, the software generates a
further txt output file, containing the following information:

Descriptive information on the time series, taken from the input file.
Information on the specification of the BV4.1 options used for the analysis.
Table, containing the estimated power spectra of the
- original series O ,
- trend-cycle component TC (= T + LS ),
- seasonally adjusted series SADJ (= O S CA U ),
- calendar adjusted series CADJ (= O CA U ),
- seasonal component S ,
- calendar component CA ,
- user-defined component U ,
- irregular component I (= EX + R ),
- residual component R .

For technical reasons, in the txt output file the given name of the input file is always "input.txt"
and the given format extension of the regressor file is always "txt".

4.3 Output option "integrated, complete"

If the output option "integrated, complete" was chosen (cf. 5.2) and the user clicks on one of the
export buttons after the analyses were accomplished, for each of the analysed time series an
output file is generated, which contains the complete set of the time series components as well
as the important adjusted series. In detail:

indicator of level shifts ("X"),

indicator of outliers ("X"),
original series O ,
annual percentage change rate O _ R of O ,
trend-cycle component TC (= T + LS ),

16
 percentage change rate TC _ R of TC on the previous period,
seasonally adjusted series SADJ (= O S CA U ),
percentage change rate SADJ _ R of SADJ on the previous period,
calendar adjusted series CADJ (= O CA U ),
annual percentage change rate CADJ _ R of CADJ ,
seasonal component S ,
calendar component CA ,
user-defined component U ,
level shift component LS ,
irregular component I (= EX + R ),
outlier components EX and
residual component R .

In addition, the file contains the short form of the BV4.1 analysis (cf. 1.4) and the descriptive
information on the time series (given in the input file (cf. 3.1)).

4.4 Output option "integrated, reduced"

If the output option "integrated, reduced" was chosen (cf. 5.2) and the user clicks on one of the
export buttons after the analyses were accomplished, for each of the analysed time series an
output file is generated, which contains a reduced set of components and adjusted series. In
detail:

indicator of level shifts ("X"),

indicator of outliers ("X"),
original series O ,
annual percentage change rate O _ R of O ,
trend-cycle component TC (= T + LS ),
percentage change rate TC _ R of TC on the previous period,
seasonally adjusted series SADJ (= O S CA U ),
percentage change rate SADJ _ R of SADJ on the previous period,
calendar adjusted series CADJ (= O CA U ),
annual percentage change rate CADJ _ R of CADJ .

In addition, the file contains the short form of the BV4.1 analysis (cf. 1.4) and the descriptive
information on the time series (given in the input file (cf. 3.1)).

4.5 Output option "separated (reduced)"

If the output option "separated (reduced)" was chosen (cf. 5.2) and the user clicks on one of the
export buttons after the analyses were accomplished, 4 output files are generated, where all the
time series analysed during the run, the corresponding trend-cycle components TC (= T + LS ),
the seasonally adjusted series SADJ (= O S CA U ) and the calendar adjusted series
CADJ (= O CA U ) are each stored separately (including the respective percentage change
rate). Again the files contain for each series the short form of the BV4.1 analysis (cf. 1.4), the
descriptive information on the time series (given in the input file (cf. 3.1)) and the information on
outliers identified and level shifts preset (indicated by an "X").

17
4.6 Output files with diagrams

Diagrams which where created by the user during the program run can be stored as png files.

4.7 Naming conventions for output files

Let "NAME" denote the code name to identify the output files of a program run (cf. point 4 in
5.2), let "NAME1" denote the integral part of the name of the txt output file defined in the input
file (cf. 3.1) and let "NAME2" denote the short name of the analysed time series (cf. 3.1). Then
the output files are named as follows:

NAME.NAME1 -txt file of the detailed results of the BV4.1 time series analysis,
NAME.s"i"NAME1 -txt file of the detailed results of the ith BV4.1 time series analysis of a
program run for generating successive analyses,
NAME.NAME1.spec - txt file with the power spectra for the results of the BV4.1 analysis,
NAME.s"i"NAME1.spec - txt file with the power spectra for the results of the ith BV4.1
analysis of a program run for generating successive analyses,
NAME.NAME2.IL or NAME_NAME2_IL - file or table of the output variant "integrated,
complete" (cf. 4.2), generated using the file formats EXCEL and CSV or
ACCESS and SQL Server, respectively,
NAME.NAME2.IS or NAME_NAME2_IS - file or table of the output variant "integrated,
reduced" (cf. 4.3), generated using the file formats EXCEL and CSV or
ACCESS and SQL Server, respectively,
NAME.OS or NAME_OS - file or table of the output variant "separated (reduced)" (cf. 4.4) con-
taining the original series ( O ), generated using the file formats EXCEL
and CSV or ACCESS and SQL Server, respectively,
NAME.TC or NAME_TC - file or table of the output variant "separated (reduced)" (cf. 4.4) con-
taining the trend-cycle components TC (= T + LS ), generated using
the file formats EXCEL and CSV or ACCESS and SQL Server,
respectively,
NAME.SA or NAME_SA - file or table of the output variant "separated (reduced)" (cf. 4.4) con-
taining the seasonally adjusted series SADJ (= O S CA U ),
generated using the file formats EXCEL and CSV or ACCESS and SQL
Server, respectively,
NAME.CA or NAME_CA - file or table of the output variant "separated (reduced)" (cf. 4.4) con-
taining the calendar adjusted series CADJ (= O CA U ),
generated using the file formats EXCEL and CSV or ACCESS and SQL
Server, respectively,
NAME.NAME2.OS.S or NAME_NAME2_OS_S - file or table of the output variant "separated
(reduced)" (cf. 4.4) containing all original series ( O ) of a program run
for generating successive analyses, generated using the file formats
EXCEL and CSV or ACCESS and SQL Server, respectively,
NAME.NAME2.TC.S or NAME_NAME2_TC_S - file or table of the output variant "separated
(reduced)" (cf. 4.4) containing all trend-cycle components TC
(= T + LS ) of a program run for generating successive analyses,
generated using the file formats EXCEL and CSV or ACCESS and SQL
Server, respectively,
NAME.NAME2.SA.S or NAME_NAME2_SA_S - file or table of the output variant "separated
(reduced)" (cf. 4.4) containing all seasonally adjusted series SADJ
(= O S CA U ) of a program run for generating successive

18
 analyses, generated using the file formats EXCEL and CSV or ACCESS
and SQL Server, respectively,
NAME.NAME2.CA.S or NAME_NAME2_CA_S - file or table of the output variant "separated
(reduced)" (cf. 4.4) containing all calendar adjusted series CADJ
(= O CA U ) of a program run for generating successive analyses,
generated using the file formats EXCEL and CSV or ACCESS and SQL
Server, respectively.

19
5 Running BV4.1
5.1 Front screen page

This is the front screen page of the BV4.1 application. The user can choose between an English
and a German language version of the application by clicking on the button named "English" or
"Deutsch", respectively. Clicking the "Quit" button closes the application.

20
5.2 Main menu screen page

3
4

5
7

Clicking one of the language buttons on the front screen page switches the GUI to the screen
page of the main menu. The above screen shot shows the main menu with the defaults.

1 Possible formats of the file containing the time series to be analysed ( 5.3).

2 Field for the listing of the imported time series with the possibility of changing the list
and of specifying series specific analysis periods ( 5.4).

3 Options regarding the way to specify the options of the BV4.1 analyses:

- Using the option "Use of default options", all analyses are carried out with outlier
adjustment (TAU=3.00 and M=24 for monthly time series, and M=8 for quarterly time
series) as well as with calendar adjustment according to the trading day model (cf.
1.2.1.1) based on the German public holiday situation without consideration of All
Saints Day (cf. 2.4). Level shifts and user-defined regressor variables wont be taken
into account. The results are outputted with one decimal place.
- Using the option "Use of input file options", all analyses are carried out with the
specification of the options given in the input file (cf. Ch. 3).
- Using the option "Manual specification of the options", the user has the possibility to
specify the options by using the GUI ( 5.5).

Default: "Use of default options".

21
4 Input field to specify the name of the sub-folder of the folder "Results" (maximum four
digits) in which the output files of the program run will be stored (for time series in SQL
Server format only the txt output files will be stored there) and label/name component
of the output files (cf. 4.7)

5 Option to carry out successive analyses ( 5.6).

Successive analyses are practicable only if one single time series is to be examined and
only in combination with the output variant "separated (reduced)" ( 5.7).

Default: "no".

6 Option to generate diagrams of the results (cf. 5.8).

To generate diagrams of results of past program runs it is sufficient to click on the

"Continue" button (e.g. without reading in any time series). Then the GUI switches
directly to the screen page "Generating diagrams" ( 5.8).

Default: "yes".

7 After filling in the main menu, the analysis run is started by clicking on the "Continue"
button ( 5.7).

By clicking on the "New" button, the GUI will switch to the default version of the screen
page of the main menu.
Clicking the "Language" button switches the GUI back to the front screen page, where it
is possible to change the selected program language.

22
5.3 Specification of the decimal separator, the column separator for CSV files and the data
path for the input and output files / Selection of the input file containing the time
series to be analysed

1 Clicking on "Configuration" will open the "Options" window, where some default options
may be changed.

For the decimal separator character, one can select between decimal point and decimal
comma. For the column separator character for CSV files, one can select between
semicolon and comma. The changes are valid only for the current BV4.1 run. If one
restarts BV4.1, the defaults will be adopted again. The defaults for the German version
are decimal comma and semicolon. For the English version, the defaults are decimal
point and comma.

23
 In the field "Data path" a new path for the folder "Data" (containing the input and output
files (cf. 2.3)) can be determined. It is possible, too, to change the name of the folder
"Data". But it is essential that the folder contains the sub-folders "Series" (for the input
files containing the time series), "Regressors" (for the input files containing the user-
defined regressor variables) and "Results" (for the output files).

Default: Path of the folder "Data" set up during the installation of BV4.1 (cf. 2.3).

2 To import time series first of all one has to specify the format of the file containing the
desired series. A click on the corresponding format button will open the "Load" window
3 (for SQL Server: a window named "Selection"). All available files for the chosen
format are shown in this window (for SQL Server: only those files are shown which
satisfy the requirements described in chapter 3). By selecting the desired file and by
clicking on the "Open" button, all existing time series in the file will be read in. If there is
more than one sheet/table in an EXCEL- or ACCESS- file another window will open to
select the desired sheet/table. It is possible to read in time series from different
sheets/tables by marking them. This way it is possible, too, to read in simultaneously
quarterly and monthly time series. (Confer chapter 3 for the requirements to be met by
the input files.)

24
5.4 Selection of the time series to be analysed and specification of the analysis periods

After selecting the input file containing the time series, all series from that file are listed in the
field 1 . Time series, which are to be analysed, have to be marked by a small hook in the column
"Active". As default, all imported time series are marked. For deactivating a time series, the user
has to click on the corresponding box in column "Active". By clicking on the buttons "activate" or
"deactivate", several time series can be simultaneously activated or deactivated. Before doing
this, the desired time series have to be marked.

For the imported time series, the analysis periods are adopted from the input file. To change
analysis periods the user has to click on the corresponding fields in the column "Analysis
period". Then the dialog window 2 named "Analysis period" will be opened where the changes
can be made by using scroll down lists.

25
5.5 Manual specification of the analysis options

4
6
5

Choosing the option "Manual specification of the options" on the main menu screen page
switches the GUI to the above screen page. It is possible now, to specify the analysis options
independently of the specifications in the input file. Except for the presetting of the level shifts
and the selection of the user-defined regressor variables, the specifications made will be
adopted uniformly for all time series.

1 Options regarding the outlier adjustment (cf. 1.2): If outlier adjustments are desired, the
length M of the basic span for outlier identification and the sigma limit TAU have to be
chosen from the scroll down menus. If monthly and quarterly series will be analysed at
the same run, M can be specified separately for each type of series.

Defaults: "Yes" for outlier adjustment, M=24 for monthly series, M=8 for quarterly
series, and TAU=3.00.

2 Options regarding the calendar adjustment approaches described in the chapters 1.2.1
and 2.4.

Default: Trading day component based on the German public holiday situation (without
considering All Saints Day) and the other settlements described in chapter 2.4.

26
3 Input fields for the presetting of level shifts by using scroll down menus (cf. 5.5.1).

Defaults: Specifications from the input file.

4 Option to incorporate user-defined regressor variables in the time series analyses (cf.
5.5.2).

Default: No use of user-defined regressors.

5 Field to specify the number of decimal places of the figures in the output files. Valid
options are 0, 1, ..., 6.

Default: 1.

6 Buttons:

- "OK": The given specifications are adopted and the GUI switches to the screen page
of the main menu for specifying further options or for starting the BV4.1 analyses.
- "Cancel": The given specifications are cancelled and the GUI switches to the screen
page of the main menu.
- "New": The GUI switches to the default version of the screen page of the main menu.

27
5.5.1 Presetting of level shifts

The presetting of level shifts or the cancellation of already defined level shifts is handled by
clicking on the corresponding boxes in the field 1 . Whenever one clicks on such a box the
dialog window 2 named "Level shift" is opened, where the level shifts can be chosen by using
a scroll down list. For each individual time series, up to 10 level shifts can be preset.

28
5.5.2 Selection of user-defined regressors

By clicking on "yes" in field 1 , the "Load" window 2 or, if the time series are stored in SQL
Server files, a window named "Selection" will open and all available files of the folder
"Regressors" (cf. 2.3) or within the SQL Server data base are listed, respectively. For time series
files in EXCEL, ACCESS and CSV format only files of the same format are available. For time series
files in SQL Server format, only those files of the data base are listed, which meet the
requirements stated in chapter 3.2.

After the selection of the desired regressor file, the following screen page is displayed:

29
Now it is possible to select up to 15 variables from the (up to 50) regressor variables of the
chosen regressor file for each time series (in the example shown, the regressor variables were
named "Mo", "Tu", "We" etc.). The desired regressor variables are selected by clicking on the
corresponding boxes. Selected variables are marked by a small hook. By clicking on the "OK"
button, the selections will be adopted.

5.6 Successive analyses

If the option "successive analysis" from the main menu is chosen, the GUI switches to the above
screen page. There, the beginning A of that period can be defined, for which successive analyses
will be carried out. (The end of the period for which successive analyses will be carried out is
always the end of the analysis period as determined in the main menu. The analysis periods of
the single successive analyses always begin with the starting period of the analysis period as
determined in the main menu.)

To specify A, one has to click on the corresponding field 1 in column A and then to select the
desired date from the list shown at the dialog window 2 named "Starting date A".

By clicking on the "OK" button, the specification will be adopted and the screen page of the main
menu is displayed again. There, further options can be specified or the analysis run can be
started.

30
5.7 Result screen page

2 3

Clicking the "Continue" button on the main menu screen page opens the screen page shown
above and the analyses will be carried out.

1 After the analyses are finished, the txt output file (cf. 4.1) for the first time series
selected in the main menu is displayed.

The BV4.1 results for the other time series analysed during the analysis run and the
results of the spectrum analyses can be displayed by clicking on the scroll down arrow
in the box named "Time series". A scroll down list is presented showing the short names
of the selected time series (as specified in the input file (cf. 3.1)). If successive analyses
were carried out, this is indicated by a "(S i)" placed at the end of the series name,
where i means that this result belongs to the ith time series analysis of that analysis
run. Only the name of the txt file corresponding to the analysis based on the longest
analysis period has no additional notification. The corresponding results of the
spectrum analyses can be identified by the name ending "spec".

The individual analysis results may be printed out by clicking on the "Print" button.

2 For further processing of the analysis results, there are three alternatives to export them
(cf. 4.3 - 4.5). The export is carried out by clicking on the button of the desired file
format.

31
 The exported files are stored within the sub-folder "Results" or within the BV4.1 data
base for SQL Server format. They can be identified by the code name defined in the main
menu.

3 Buttons:

- "Continue": The screen page for generating diagrams is displayed ( 5.8).

- "Back": The GUI switches to the screen page of the main menu.
- "New": For carrying out new analyses. The default version of the screen page of the
main menu is displayed again.

5.8 Generating diagrams

1
2
3
4

1 The box named "Code name of the desired analysis results" has to contain the code
name (cf. point 4 in 5.2) of that results, which are to be plotted. By clicking on the
scroll down arrow, a list of the currently held analysis results will be displayed.

If, during the program run, analyses were carried out, that box will contain the current
code name.

2 This option allows for the creation of diagrams of up to 4 original time series, estimated
time series components, adjusted series and percentage change rates (so called
standard diagrams ( 5.8.1)). If this option is activated, the user has to select the
series to be displayed from the scroll down lists 5 .

32
 3 This option allows for the creation of diagrams of up to 4 power spectra or log-spectra of
original time series, estimated time series components and adjusted series ( 5.8.4). If
this option is activated, the user has to select the series whose spectra are to be
displayed from the scroll down lists 5 .

4 This option allows for the creation of diagrams of yearly curves of an original time series,
an estimated time series component, a seasonally adjusted series or a percentage
change rate. Here, up to ten yearly curves of a series can be superimposed ( 5.8.2). If
this option is activated, the user has to select the series to be displayed from the scroll
down lists 5 .
6 If successive analyses were carried out during the program run chosen in box 1 , this
option is activated, too. It is now possible to plot diagrams, showing the results of
successive BV4.1 analyses for the trend-cycle component TC , the seasonally adjusted
series SADJ , the calendar adjusted series CADJ , the percentage change rate of TC
on the previous period, the percentage change rate of SADJ on the previous period and
the annual percentage change rate of CADJ ( 5.8.3).

By clicking on the "Back" button, the result screen page will be displayed again (cf. 5.7). By
clicking on the "Continue" button, the desired diagram will be displayed. Clicking the "New"
button switches the GUI to the default version of the screen page of the main menu.

5.8.1 Standard diagrams

33
 1 In this field of the screen page, the ranges of the axes of the diagram shown can be
changed. To change the value axis the desired starting and end values have to be
entered in the corresponding boxes. To change the time axis the beginning and the end
of the time interval have to be selected from the corresponding scroll down menus.
Changes will be adopted by clicking on the "Show" button.

By clicking on the "Further Options" button, it is possible to change the distances

between the grid lines and to specify axis breaks.

The diagram can be printed out by clicking on the "Print" button. It can be saved by
clicking on the "Save Print" button (cf. 4.6).

By clicking on the "Back" button, the starting screen page for generating diagrams will
open again. Clicking the "New" button switches the GUI to the default version of the
screen page of the main menu.

2 The legend for a component of the diagram in each case is composed of the short form
of the underlying time series, the denomination of the plotted component (cf. also 1.1)
and the short form of the BV4.1 analysis undertaken (cf. 1.4).

5.8.2 Yearly curves of a component

34
 1 By clicking on the "Further Options" button, it is possible to change the distances
between the grid lines and the selection of the years to be displayed. Otherwise, cf.
5.8.1.

2 Cf. 5.8.1.

5.8.3 Diagram of the results of successive analyses

1 By clicking on the "Further Options" button, it is possible to add the original time series.
Otherwise, cf. 5.8.1.

2 Cf. 5.8.1.

3 For the components resulting from different successive analyses, only the last two
numerical values of each component are displayed (connected by a line).

35
5.8.4 Diagram of power spectra

1 By clicking on the "Spectra" button, it is possible to switch from the presentation of log-
spectra to spectra. Vice versa it is possible to switch from the presentation of spectra to
log-spectra. Otherwise, cf. 5.8.1.

2 Cf. 5.8.1.

36