The directory /res0/resdata/bankitalia
contains the files from the 'year/endyear'
directories described below.
A list of files and their sizes is available.
The directory /res0/resdata/bankitalia/panel
contains the 'panel/endpanel' files described below.
For information on using fmrisc, please contact baum
****************************************************************
* *
* README FILE *
* *
****************************************************************
IN THE DIRECTORY BANKITA YOU FIND VERSION 0.0 OF THE STATA FILES,
ROUTINES AND EXPLANATORY REPORT PROCESSED USING THE SURVEY OF
BANCA D`ITALIA CALLED "INDAGINE SUI BILANCI DELLE FAMIGLIE ITALIANE"
FROM 1977 TO 1995.
THE BANK OF ITALY IS IN NO WAY RESPONSIBLE FOR THE TRANSFORMATION
OF THE ORIGINAL ASCII FILES INTO STATA ONES. THIS TRANSFORMATION HAS
BEEN REALIZED BY ANTONIO FILIPPIN AT IGIER, THANKS TO CNR FUNDING.
The source ASCII files have been made available by the Bank of
Italy under the following constraints that apply also to anyone
is going to use the Stata files:
1) DISTRIBUTION MUST BE FREE OF CHARGE, EXCLUDING ALSO ANY
SPONSORSHIP;
2) THE AUTHOR IS REQUESTED TO CITE THE SOURCE;
3) DATA CANNOT BE GIVEN OR USED FOR PURPOSES DIFFERENT FROM
SCIENTIFIC RESEARCH;
4) DATA PROCESSING TAKES PLACE UNDER THE USER'S OWN WHOLE
RESPONSIBILITY. THE USER CAN IN NO WAY INVOLVE THE BANK OF
ITALY, WHICH MUST BE CITED ONLY IN THE NOTE:
"Processing of Banca d'Italia data, Indagine sui bilanci delle
famiglie italiane".
IN ORDER TO ALLOW THE COLLECTION OF REFERENCES CONCERNING THE
SURVEY, PLEASE SEND COPY OF YOUR WORK BASED ON THESE DATA TO THE
BANK OF ITALY AT THE ADDRESS SPECIFIED BELOW.
FOR INFORMATION ON ORIGINAL ASCII FILES PLEASE CONTACT:
*****************************************************************
* *
* Banca d'Italia *
* Servizio Studi, Divisione A.I.S. *
* Via Nazionale 91 *
* 00184 Roma *
* *
* Tel ++39.6.47924159 *
* Tel ++39.6.47924158 *
* *
* e-mail: dal.4158@interbusiness.it *
* *
*****************************************************************
FOR INFORMATION ON STATA FILES PLEASE CONTACT:
*****************************************************************
* *
* Antonio Filippin *
* c/o IGIER BOCCONI *
* via Salasco, 5 *
* 20136 Milano *
* *
* fax: ++39.2.58363302 *
* e-mail: antonio.filippin@uni-bocconi.it *
* *
*****************************************************************
****************************************************************
* *
* INTRODUCTION *
* *
****************************************************************
THE ORIGINAL DATA PROVIDED BY THE BANK OF ITALY ARE COLLECTED
BY QUESTIONNAIRES GIVEN TO SELECTED SAMPLES OF HOUSEHOLDS. DATA
ARE STORED IN DIFFERENT FILES YEAR BY YEAR, WITH THE EXCEPTION
OF THE PERIOD FROM 1977 TO 1983, IN WHICH THEY HAVE BEEN
TREATED TOGETHER, BECAUSE THE TYPE OF QUESTIONNAIRE AND THE
CRITERIA USED IN COLLECTING DATA DID NOT CHANGE OVER THOSE YEARS.
THE AIM OF THIS WORK IS TO MAKE AVAILABLE THIS DATA SET IN STATA
FORMAT WITH DOCUMENTATION.
TO REACH THIS GOAL:
1) WE READ THE ORIGINAL ASCII FILES IN STATA FORMAT (.dta),
2) WE VERIFIED THE LOGICAL CONSISTENCY OF THE DATA SET,
3) WE WROTE A REPORT ON THE PROBLEMS ENCOUNTERED IN THE STEP 1)
AND 2) PLUS DOCUMENTATION.
****************************************************************
* *
* DON'T FORGET *
* *
****************************************************************
a) WE ASSUME YOU HAVE A BASIC KNOWLEDGE OF STATA SOFTWARE;
WE USED "STATA VERSION 5.0", THAT AUTOMATICALLY REPARTITIONS
MEMORY AS NECESSARY. IF YOU USE A VERSION UP TO 4.O YOU NEED
TO REPARTITION MEMORY "MANUALLY", RUNNING THE COMMANDS:
set maxvar ### OR set maxobs ###.
YOU CAN DO IT IN TWO WAYS:
1) ENABLE THESE COMMANDS IN THE *.DO FILES (REMOVING THE "*")
2) GIVE THEM DIRECTLY TO STATA (IN THE MEMSIZE.TXT FILE YOU FIND
THE NR. OF VARIABLES AND OBSERVATIONS FOR EACH FILE.
b) "*" STANDS FOR JOLLY LETTERS; FOR EXAMPLE WHEN YOU FIND
"*F84.*" IT STANDS FOR RF84.LOG, AF84.LOG, RF84.DO AND
AF84.DO AS WELL.
c) WHEN year IS INCLUDED IN A FILE NAME, IT STANDS FOR 1977-83,
1984, 1986, 1987, 1989, 1991, 1993 AND 1995 ALTERNATIVELY.
****************************************************************
* *
* STRUCTURE OF THE WORK *
* *
****************************************************************
THE STRUCTURE OF OUR WORK IS THE SAME THROUGHOUT THE DIFFERENT YEARS,
EVEN IF THE STRUCTURE OF DATA SETS PROVIDED BY BANCA D'ITALIA FROM 1989
TO 1993 IS DIFFERENT WITH RESPECT TO THE PREVIOUS PERIOD (1977-1986).
FOR EACH YEAR (OR GROUP OF YEAR AS FOR 1977-83) WE CREATE TWO
DIRECTORIES: origyear and endyear.
THERE IS ALSO A DIRECTORY pub/panel CONCERNING THE PANEL DATA SET
FROM 1989 TO 1995.
In bankita/year/origyear you will find:
(Note: these files are not available on fmrisc.)
1) THE ORIGINAL FILES CONTAINING THE DATA IN ASCII FORMAT
PROVIDED BY THE BANK OF ITALY.
A) PERIOD 1977-1986.
orig7783 contains:
f7783.asc (The initial f standing for "family"),
p7783.asc (The initial p standing for "percettore", that is
income receiver),
l7783.asc (The initial l standing for "labour"), that contains
info about employees, provided by the Bank of Italy
separately from the original tapes.
(Similarly for 1984, 1986 and 1987).
B) PERIOD 1989-1993.
orig1989 contains:
f89tsqf.asc information about hh composition
comp89.asc information about hh members
(Original name for this file was "carcomp")
rper89.asc information about income recipients
rfam89.asc information about hh incomes
ricfam89.asc information about net real wealth (hh)
confam89.asc information about consumption
f89tsb1.asc information about personal expectation on income
and inflation
f89tsb2.asc information about working activity
f89tsb3.asc information about dependent working activity
f89tsb4.asc information about assistants
f89tsb5.asc information about self employed workers (also in
companies employing up to 19 workers)
f89tsb6.asc information about partners in companies employing
20 or more workers
f89tsb7.asc information about pensions
f89tsb8.asc information about other sources of income
f89tsd1.asc information about owned real estates
f89tsd2.asc information about sold/given real estates
Orig1991 contains:
f91tsqf.asc information about hh composition
comp91.asc information about hh members
(Original name for this file was "carcomp")
rper91.asc information about income recipients
rfam91.asc information about hh incomes
ricfam91.asc information about net real wealth (hh)
confam91.asc information about consumption
f91tsb1.asc information about personal expectation on income
and inflation
f91tsb2.asc information about working activity
f91tsb3.asc information about dependent working activity
f91tsb4.asc information about self employed workers
f91tsb5.asc information about family firms
f91tsb6.asc information about partners in companies
f91tsb7.asc information about pensions
f91tsb8.asc information about other sources of income
f91tsb9.asc information about job change and job search
f91tsd1.asc information about owned real estates
f91tsd2.asc information about sold/given real estates
f91tsf1.asc information about intergenerational transfers
f91tsg1.asc information about (dependent worker) head of hh
f91tsg2.asc information about (self employed) head of hh
orig1993 contains:
f93tsqf.asc information about hh composition
comp93.asc information about hh members
(Original name for this file was "carcomp")
rper93.asc information about income recipients
rfam93.asc information about hh incomes
ricfam93.asc information about net real wealth (hh)
confam93.asc information about consumption
f93tsb1.asc information about working activity
f93tsb2.asc information about dependent working activity
f93tsb3.asc information about self employed workers
f93tsb4.asc information about family firms
f93tsb5.asc information about partners/managers in companies
f93tsb6.asc information about pensions
f93tsb7.asc information about other sources of income
f93tsd1.asc information about owned real estates
f93tsd2.asc information about sold/given real estates
orig1995 contains:
q95a.asc information about intergenerational comparison
q95b1.asc information about working activity
q95b2.asc information about income sources
q95c1.asc information about means of payment
q95c2.asc information about savings
q95d.asc information about real estates
q95e.asc information about consumption goods
q95f.asc information about insurances
(previous eight files correspond more or less
to the f**tsqf.asc files from 89 to 93)
rper95.asc information about income recipients
rfam95.asc information about hh incomes
ricfam95.asc information about net real wealth (hh)
confam95.asc information about consumption
comp95.asc information about hh members
(Original name for this file was "carcom95")
allb1.asc information about dependent working activity
allb2.asc information about self employed workers
allb3.asc information about family firms
allb4.asc information about partners/managers in companies
allb5.asc information about pensions
allb6.asc information about other sources of income
alld1.asc information about owned real estates
alld2.asc information about sold/given real estates
usciti.asc information about members that left the hh
rotaz1.asc information about employed people
(Only if with head of hh's even year of birth)
rotaz2.asc info about empl., unempl., seeking 1st job
(Only if with head of hh's odd year of birth)
2) THE LOGICAL CHAIN OF FILES WE USED TO REVISE THE DATA SET.
(Note: these files are not available on fmrisc.)
A) PERIOD 1977-1986.
For example, orig7783 contains:
*7783.dct That is the dictionary used to read the data in
ASCII format (pay attention that * stands for "f", "p" or
"l" alternatively).
r*7783.do The initial "r" standing for "read-file". This
file contains the program used to insert the ASCII
data in Stata using the dictionary mentioned above.
The final output of the process is r*7783.dta (i.e.
the original data in Stata format separately for "f",
"p" and "l" files).
We don't make r*7783.log and r*7783.dta available
for two reasons:
first because it takes a lot of memory;
secondarily because you can obtain it very easily
giving to Stata the command: "do r*7783".
a*7783.do The initial "a" standing for "analysis-file".
This file contains the program used to analyse the
inner logical consistency of the data set, to find
existing problems (that will be pointed with P#, #
being the ordinal number in the file) and to solve
them (S#). In this file we also create the value labels.
The final output of the process is a*7783.dta (i.e.
the final data in Stata format separately for "f",
"p" and "l" files).
We don't make a*7783.log and a*7783.dta available
for two reasons:
first because it takes a lot of memory;
secondarily because you can obtain it very easily
giving to Stata the command: "do a*7783".
fpl7783.do The initial "fpl" standing for family, percettore
and labour, respectively. This file contains the
program use to merge af7783.dta, al7783.dta and
ap7783.dta together.
The final output of the process is fpl7783.dta (i.e.
the final data in Stata format with "f", "p" and "l"
files together, both for each year separately and for
all the period long). We copied these files in another
directory (end7783), in which you will find also a file
called "book7783.txt" (see below).
NOTE: This chain is identically repeated for two of the three
kinds of data set ("f" and "p") for 1984 and 1986.
The "l" files are slightly different from the "f" and
"p" ones, because Bank of Italy provided the
related data set in the same files with 1986 and 1987
ones. For this reason we ran the logical chain of the
"l" files for 1984, 1986 and 1987 together, separating
the "l" files for each year only at the end of al8487
file. For the same reason you will find also the *8487.*
files instead of only the *84.* and *86.* ones.
(Similarly for 1984, 1986 and 1987)
B) PERIOD 1989-1995.
As far as 1989-95 period is concerned, the logical chain
*.dct (where "*" stands for every file)
r*.do (where "r" stands for read-file)
a*.do (where "a" stands for analysis-file)
is the same; also in this case we don't make available
r*.log, r*.dta, a*.log and a*.dta because they take a lot of
memory and they can be obtained very easily.
As for this period it is not possible to merge
all files, you will find
merge*.do
containing the program used to merge only acomp*.dta,
arper*.dta, arfam*.dta, aconfa*.dta and aricfa*.dta.
You find the final output (merge*.dta) in the "endyear"
directory, toghether with all the other *.dta files.
In bankita/year/endyear you will find:
(Note: these files are on fmrisc in /res0/resdata/bankitalia)
1) THE FINAL DATA FOR EACH YEAR (AND TOGETHER FOR THE PERIOD
FROM 1977 TO 1983).
A) PERIOD 1977-1987.
In the end7783 directory you find:
fpl7783.dta
fpl77.dta
fpl78.dta
fpl79.dta
fpl80.dta
fpl81.dta
fpl82.dta
fpl83.dta
(Similarly for 1984, 1986 and 1987)
B) PERIOD 1989-1995.
In the end1989 directory you find:
f89tsqf.dta
merge89.dta (that is the merge of acomp89.dta, arfam89.dta,
arper89.dta, aricfa89.dta and aconfa89.dta)
f89tsb1.dta
f89tsb2.dta
f89tsb3.dta
f89tsb4.dta
f89tsb5.dta
f89tsb6.dta
f89tsb7.dta
f89tsb8.dta
f89tsd1.dta
f89tsd2.dta
(Similarly for 1991, 1993 and 1995)
2) AN EXPLANATORY TEXT FILE CALLED "bookyear.txt" IN WHICH
THERE IS A LIST OF ALL THE VARIABLES AND RELATED CODES.
****************************************************************
* *
* PANEL DATA SET *
* *
****************************************************************
In bankita/panel/origpan you find:
(Note: these files are not available on fmrisc.)
nquest.do program used to get the household interviewed
at least twice (nque8993.dta).
the programs used to obtain the panel data sets.
-------------+-------------------------------------------------+--------+
file name | information | period |
-------------+-------------------------------------------------+--------+
pahhques.do | Households' composition | 89-95 |
-------------+-------------------------------------------------+--------+
pamerge.do | Households' income, wealth | 89-95 |
-------------+-------------------------------------------------+--------+
paexpect.do | Expectations about income and inflation | 89-91 |
-------------+-------------------------------------------------+--------+
padepwk.do | Dependent working activity | 89-95 |
-------------+-------------------------------------------------+--------+
paselfwk.do | Self employed workers | 89-95 |
-------------+-------------------------------------------------+--------+
paworkin.do | Working activity | 89-93 |
-------------+-------------------------------------------------+--------+
papens.do | Pensions | 89-95 |
-------------+-------------------------------------------------+--------+
paownres.do | Owned real estates | 89-95 |
-------------+-------------------------------------------------+--------+
paothinc.do | Other income sources | 89-95 |
-------------+-------------------------------------------------+--------+
NOTE: we don't make available pa*.log and the input pa*.dta files,
because they take a lot of memory and because they can be
obtained very easily running the pa*.do files and the
correspondent a*.do files, respectively.
For example, if you want to obtain pa8991qf.dta file
(that is the panel 89-91 of info about hh composition),
you need pane89qf.dta and pane91qf.dta: you can get
them running af89tsqf.do and af91tsqf.do files
respectively.
In bankita /panel/endpanel you find:
(Note: these files are on fmrisc in /res0/resdata/bankitalia/panel)
----------+------------------------+-------------------------------------+
file name | information | longitudinal data availabile |
----------+------------------------+-------------------------------------+
phhq*.dta | hh's composition | 89-95 89-93 91-95 89-91 91-93 93-95 |
----------+------------------------+-------------------------------------+
pmer*.dta | hh's income, wealth | 89-95 89-93 91-95 89-91 91-93 93-95 |
----------+------------------------+----------------+--------------------+
pexp*.dta | Expectations about income and inflation | 89-91 |
----------+-----------------------------------------+--------------------+
pdwk*.dta | Dependent working activity | 89-95 89-93 91-95 |(*)
----------+-----------------------------------------+--------------------+
pswk*.dta | Self employed workers | 89-95 89-93 91-95 |(*)
----------+-----------------------------------------+--------------------+
pwor*.dta | Working activity | 89-93 |(*)
----------+-----------------------------------------+--------------------+
ppen*.dta | Pensions | 89-95 89-93 91-95 |(*)
----------+-----------------------------------------+--------------------+
powr*.dta | Owned real estates | 89-95 89-93 91-95 |(*)
----------+-----------------------------------------+--------------------+
poin*.dta | Other income sources | 89-95 89-93 91-95 |(*)
-------------------------------------------------------------------------+
(*) data for only two years ar not made available directly. To get them
is sufficient to enable the correspondent commands in the respective
pa*.do file.
****************************************************************
* *
* OPTIONS TO ACCESS AND TRANSFER THE DATA *
* *
****************************************************************
SEVERAL OPTIONS ARE AVAILABLE. CHOOSE THE ONE YOU PREFER
BETWEEN THE FOLLOWING THREE:
1) TO FTP-GET ONLY THE ORIGINAL FILES (.ASC) IN THE DIRECTORIES
origyear;
2) TO COPY ALL THE FILES, IF YOU THINK THAT THE FINAL DATA ARE
USEFUL FOR YOUR WORK, BUT YOU WANT TO VERIFY THE PROCEDURE
WE USED, OR TO CHECK THE CHANGES WE MADE IN TRANSFORMING THE
ORIGINAL DATA SETS.
WE SUGGEST TO COPY ALL THE FILES EVEN IF YOU THINK THAT THE
CHANGES WE MADE ARE NOT (OR ONLY PARTIALLY) NECESSARY FOR
YOUR WORK, OR YOU NEED TO MAKE DIFFERENT ONES, BECAUSE DO-FILES
ARE VERY EASY TO MODIFY;
3) TO COPY ONLY THE FINAL DATA IN THE DIRECTORIES endyear;
*****************************************************************
* *
* NOTES *
* *
*****************************************************************
HERE IS A LIST OF NOTES, EXPLAINING IN SOME DETAILS THE
PROBLEMS WE FOUND IN THE "ANALYSIS-FILES". EACH NOTE
CONTAINS AN IDENTIFICATION NUMBER AND A LIST OF ALL THE
PROGRAMS (*.DO FILES) INVOLVED IN EACH SPECIFIED PROBLEM.
WE ADVICE YOU TO READ THE NOTES AND THE RELATED PROGRAMS
TOGETHER, IN ORDER TO MAKE EASIER THE COMPREHENSION OF
THE PROBLEMS.
(1a)
ALPHANUMERIC MISSING VALUES
In the ASCII file "Y", "N" or "&" stand for alphanumeric
missing values, which cannot be read by Stata unless
previously specifying "str# varname %#s" in the dictionary.
If we specify this, however, Stata could not read those
variables like numbers, so we prefer not to use this
specification and let "Y", "N" or "&" be read as a dot.
In such a way Stata reads even the alphanumeric characters
as numeric, mixing the true missing values like "Y", "N"
or "&" (i.e. in a household there is a 2nd non income
receiver component, but for a reason there aren't the
concerning data) and the logical ones (i.e. in a household
there isn't a 2nd non income receiver component): in both
cases we will have a dot.
rp7783 rf7783
rf86 rp86
rf87 rp87
rf89tsqf
rf91tsqf rf91tsb4 rf91tsb5 rf91tsb6 rf91tsb9 rf91tsd1 rf91tsd2
rf91tsf1 rf91tsg1 rf91tsg2 rcomp91
(1b)
TOO MANY MISSING VALUES
We specify "str# varname %#s", even if we are simply
handling with alphanumeric missing values. This would
not be strictly necessary, but it allows us to avoid
a too long list of messages:
' ' cannot be read as a number for varname[observation number]
We will transform these variables in numeric ones in
the a*.* files.
ap87
af91tsqf af91tsb2 af91tsb7 af91tsd1 af91tsg1 af91tsg2
(1c)
NUMERIC MISSING VALUES
We replace numeric missing values with dots.
af87
(2)
STRINGS
The specification "str# varname %#s" is necessary, because
Banca d'Italia uses alphabethic codes, which are neither
true nor logical missing values.
If we don't use this specification, we would lose all the
observations related to these codes, because Stata would
read them as missing values.
For this reason in a first time (r*.* files) we keep these
codes by specifying "str# varname %#s"; then (a*.* files)
we recode it as a number.
af7783 ap7783;
af84 ap84;
af86 ap86;
af87
af89tsqf af89tsb1 af89tsb2 af89tsb3 af89tsb8 af89tsd1 af89tsd2
af91tsqf acomp91
af93tsqf acomp93
(3)
MISMATCHING VARIABLES
Banca d'Italia gives different names and codes to the
same variables in different files. We try to make them
as uniform as possible, because it is useful when we
merge files.
al7783
al8487
af87 ap87
(4)
VARIABLE GENERATION
When useful (and possible, of course) we generate those
variable generally used in other years. Sometimes we
generate variables necessary in order to make easier tha
analysis of data sets.
aconfa89 af89tsb7
(5)
VARIABLE ELIMINATION
We cancel out empty or non useful variables
af89tsqf af89tsb1 af89tsb2 af89tsb3 af89tsb4 af89tsb5 af89tsb6
af89tsb7 af89tsb8 af89tsd1 af89tsd2
af91tsqf
aq95b1 aq95c2 aq95f
(6)
MISMATCHING CODES
The same value label is represented by different codes in
different files. For example in *p7783.* files "seeking
first occupation" is represented by "2", while in the
*f7783.* ones it is represented by "9".
We try to get them as uniform as possible.
af7783 ap7783
af84 ap84
af86 ap86
af87
af89tsqf acomp89 af89tsb2
af91tsqf af91tsb1 af91tsb2 af91tsd1 af91tsd2 af91tsg1
af93tsqf af93tsb1
(7)
REGION CODES
The number of codes for this variable is different
according to the years:
* from 1977 to 1980 there are 10 categories.
* from 1981 to 1984 there are 17 categories.
We create a new variable "reg17" in order to separate
the new codes from the old ones.
* in 1986 there are 17 categories.
In Banca d'Italia records there is "reg" variable, with
17 codes, slightly different with respect to "reg17" one
used in 1981-84.
* in 1987 there isn't "region" variable.
* from 1989 to 1993 there are 20 categories.
Everytime, when possible, we "translate" the new codes
in old ones in the "reg10" variable.
In such a way it is possible to have a variable (reg10)
that allow us to compare classified by region data
after and before 1980 at the same time. As viceversa
is not possible, that is we cannot translate old codes
in new ones, we let the new codes before 1981 as a dot.
af7783
af84
af86
af89tsqf acomp89
af91tsqf acomp91
af93tsqf acomp93
aq95a acomp95
(8)
AGE CODES
Information about age of the people is different according
to the years (and the files):
* from 1977 to 1983 Banca d'Italia provides information only
about the class of age of people. In the l7783 file there
was a variable called "eta" (age: nr. of years), but it
was empty so we canceled it out.
Furthermore, codes for classes of age change from file to
file. In order to merge data sets we renamed variables eta
and eta7 "etacl" giving them the same codes used for "eta1"
in f7783 file.
* 1984, 1986, 1987: Banca d`Italia provides detailed
information about "age" of the sample, in fact number of
years is specified. In order to make possible a
confrontation among the data from 1984 to 1987 and the
other years, we generate:
1) "etacl" variable (age in 9 classes) for income recipients
and employees, with the same codes as in 1977-83 (and we
call "etacl" variable "eta7" in l8487.asc file.
Furthermore, codes for age (nr. of years) change from file
to file.
In order to merge data sets we call it "etaan" in every
file (e.g. "v10" in p84.asc file).
2) "eta#" in 1984-1986, "etacl#" in 1987 for non income
recipients.
* from 1989 to 1993 Banca d`Italia provides detailed
information about "age" of the sample, in fact year of
birth (in f**tsqf.asc) or nr. of years (comp**.asc) are
specified. In order to make possible a confrontation among
the data from 1989 to 1993 and the other years, we generate
"etacl#" variable (age in 9 classes) with the same codes as
in 1977-83 in both files.
Summary of age codes in final (*.dta) data, where
n.a. = not available
lower case= variable provided by Banca d'Italia (also with
another name)
UPPER CASE= variable we generated:
| income recipient | non-income rec. | dependent worker |
--------+---------+---------+---------+---------+----------+---------+
| class | #year | class | #year | class | #year |
--------+---------+---------+---------+---------+----------+---------+
1977-83 | etacl | n.a. | eta# | n.a. | etacl | n.a. |
1984 | ETACL | etaan | ETA# | etaan# | ETACL | etaan |
1986 | ETACL | etaan | ETA# | etaan# | ETACL | etaan |
1987 | ETACL | etaan | ETACL# | eta# | ETACL | etaan |
| f**tsqf | comp** |
--------+-----------------------+----------------------------|
| class | #year | class | #year |
--------+-----------+-----------+---------------+------------|
1989 | ETACL# | anasc# | cleta ETACL | eta |
1991 | ETACL# | anasc# | cleta ETACL | eta |
1993 | ETACL# | anasc# | cleta ETACL | eta anasc |
1995 | | cleta ETACL | eta anasc |
ap7783 al7783
af84 al8487
af86 ap86
af87
af89tsqf acomp89
af91tsqf acomp91
af93tsqf acomp93
acomp95
(9)
NOT ALLOWABLE OBSERVATIONS
We drop those observations that are incorrect and not
replaceable with exact ones.
It can happen for two reasons:
a) the code does not exist (i.e. there is a code "8" for
"apset", while this variable can assume only values from
1 to 7);
b) the code exists, but it is inconsistent with the
particular situation (i.e. there is a "non income
receiver" coded as "head of household", while the head
of household is defined as the member of household which
provide the major income).
af7783 ap7783
ap86
af87
af89tsqf af89tsb1 af89tsb2 af89tsd1 af91tsg1
af91tsb4
af93tsb5
aallb4 aq95c1 aq95d
(10)
INCONSISTENT OBSERVATIONS
It concerns observations that we prefer not to classify
as not allowable, because we can substitute them with
consistent ones, without dropping them.
This is the case, for example, when the sum of the different
"income from employment" of a worker in the *l7783.* files
doesn't match with his "total income from employment" in
the *p7783.* files.
Verifying the whole situation of this worker (through the
different files) we manage to know that the right value is
the *p7783.* one.
In such a way we can substitute an inconsistent observation,
instead of dropping it. For this reason we prefer not to
classify this situation as "not allowable observations".
af7783 fpl7783
af84 ap84 al8487
af89tsb4
(11)
INCOMPLETE OBSERVATIONS
Data set looses some information, because there are more
actual situations than coded variables. For example data
set provides information only about five "non income
receivers", even if there are households with more than
five.
af7783
af84
af86
af87
(12)
CONSISTENCY CHECK FAILURE
This note contains explanations for mismatching in
aggregation of variables (income, wealth, etc.).
Most of the times it is only a problem of approximation
(+/- 1000 lire). When it is not a problem of approximation
we explain the reason for the mismatching.
Original data for 1989 (rfam89.asc and rper89.asc)
contain a mistake: "ycf" and "jycf" variables (financial
income) don't take into account negative interest,
therefore ycr+ycf>yc (and jycr+jycf>jyc). We generate
correct values for "ycf" ("jycf") in arper89.do
(arfam89.do), subtracting yc-ycr (jyc-jycr).
af7783
af84
af87 ap87
arper89 arfam89
arper91 arfam91 aricfa91
arper93 arfam93 aricfa93 aconfa93
arper95 arfam95 aricfa95 aconfa95
(13)
ANNUAL FILES
Data set fpl7783.dta and al8487.dta contain more than
one year: we create one file per year.
fpl7783
al8487
(14a)
MERGING PROBLEMS-MORE THAN ONE RECORD PER WORKER (HOUSEHOLD)
In the rl7783.* files we have data in "long" format,
i.e. there can be more than one record per worker (for
example if he has two activities), while in the *p7783.*
files it cannot be, i.e. the data are in "wide" format.
We reshape the data set from "long" to "wide", in such
a way that here will be only one record per worker even
if he has more than one activity, both activities will
appear in the same record. This may cause the lost of
some variables. In al7783 file we loose variables "oretot"
and "orestra", which didn't contain any observation, while
variable "attivp"=0 is implied by the new variables
"reddito0" and "mesilav0" and "attivp"=1 is implied by
the new variables "reddito1" and "mesilav1".
From 1989 there are more files per year and it is
impossible to do the same procedure; for this reason
we only signal that there is this problem, without
merging the files.
In 1995 we have to reshape aq95b1 and aq95b2 files in order
to merge the files concerning hh compostition (aq95a...aq95f).
This is necessary otherwise it isn't possible to get
the correspondent panel, because Stata would have
"no room to add more observations".
al7783
al8487
af89tsb3 af89tsb4 af89tsb5 af89tsd1 af89tsd2
af91tsd1 af91tsd2 af91tsf1
af93tsb2 af93tsb3 af93tsb5 af93tsb6 af93tsd1 af93tsd2
aallb1 aallb2 aallb4 aallb5 aalld1 aalld2
aq95b1 aq95b2
(14b)
MERGING PROBLEMS: "NORD" VARIABLE
l7783.asc and l8487.asc files contain the "nord" variable,
while p7783.asc, p84.asc, p86.asc and p87.asc files don't.
This is a problem when we decide to merge the two types of
data set, because it makes quite hard to connect the same
worker as "employee" and "income earner" within his/her
household, in particular if there are other "income
earners" with similar characteristics.
First of all we try to identify people without using "nord",
that is we use all the variables common to both data sets in
order to get the most precise "identikit" (in according with
this purpose we create a new variable "yl" in the *l7783.*
files).
Secondarily we count how many people have the same identikit
in both data sets, and then we give them an artificial
number of order (control1), so we will not have twin
observations.
fpl7783
fpl84
fpl86
fpl87
(14c)
MERGING PROBLEMS: OVERLAPPING VARIABLES
Banca d'Italia gives the same name to different variables
in different files (ex: yl in rper89 and in rfam89 files).
this would be a problem when we will merge files, because
one of the two variables would be "covered". For this
reason we change the name of these variables in one of the
two files.
arfam89 aconfa89
arfam91 aconfa91
arfam93 aconfa93
arfam95 aconfa95
(15a)
STRANGE OBSERVATIONS (OPTIONAL COMMANDS)
If we are dealing with very strange observations, but we
have not sure reasons to drop them (for example a "wealth
in valuable goods" equal to 99999), we let you to choose
if keep them or not. If you want to drop them run the
optional commands.
af7783 al7783
af84
af86
af87
af89tsb3 af89tsb4
af91tsqf af91tsf1 af91tsg1
af93tsqf
(15b)
DOUBLE VARIABLES (OPTIONAL COMMANDS)
Banca d`Italia uses the same name for slightly different
variables, for example "ym" for incomes of different
sources. This is a problem if you want to merge the
involved files. If you want to avoid this overlapping,
you can run the optional commands.
af89tsb4 af89tsb5 af89tsd2
af91tsb4 af91tsb5 af91tsb6 af91tsd2 af91tsg2
af93tsb3 af93tsb4 af93tsb5 af93tsd2
aallb2 aallb3 aallb4 aalld2