Content uploaded by Mārtiņš Liberts
Author content
All content in this area was uploaded by Mārtiņš Liberts on Oct 08, 2022
Content may be subject to copyright.
Central Statistical Bureau of Latvia
Method Used to
Produce Population Statistics
Authors: Ieva Aināre,
Mārtiņš Liberts,
Baiba Zukula,
Sigita Purona-Sida,
Jeļena Vaļkovska,
Bruno Opermanis,
Aleksis Jurševskis,
Kristīne Lece,
Ance Ceriņa,
Juris Breidaks,
Jānis Jukāms,
Ruta Beināre
Riga
2022
2
Summary
Within the Population and Housing Census 2011, it was found out that usually resident
population of Latvia accounted for 2 074.6 thousand at the beginning of 2011, which was
7 % fewer people than in the Register of Natural Persons1 supervised by the Office of
Citizenship and Migration Affairs (2 228.0 thousand). The Central Statistical Bureau of
Latvia worked out a method used to estimate population of Latvia more precisely. The
method is based on statistical classification and migration mirror statistics and is aimed at
dividing Latvian population included in the Register of Natural Persons of Latvia into two
classes – persons actually living in Latvia and persons actually living abroad. The statistical
classification model has been developed with the help of logistic regression.
1
Until 1 June 2021– Population Register.
3
Contents
1 Introduction ................................................................................................................................................ 4
2 Definitions .................................................................................................................................................. 5
3 Data sources ................................................................................................................................................ 6
3.1 Office of Citizenship and Migration Affairs ...................................................................................... 6
3.2 Other administrative registers ............................................................................................................. 7
3.3 CSB databases .................................................................................................................................... 9
3.4 Migration mirror statistics ................................................................................................................ 12
3.5 Sample surveys of individuals conducted by the CSB ..................................................................... 13
4 Models and assumptions ........................................................................................................................... 14
4.1 Logistic regression model ................................................................................................................. 14
4.2 Determining registered place of residence ....................................................................................... 16
4.3 International long-term emigration estimates ................................................................................... 18
4.4 Producing statistics ........................................................................................................................... 20
4.5 Initial probabilities and thresholds for probabilities ......................................................................... 22
4.6 Initial estimate of emigration and immigration ................................................................................ 22
4.7 Migration estimates by gender and age group .................................................................................. 23
4.8 Adjustments for probabilities ........................................................................................................... 25
4.9 Producing population, natural increase and migration statistics....................................................... 25
5 Estimating population number and composition precision ...................................................................... 26
5.1 Precision of classification, compared to the data of Census 2011 .................................................... 26
5.2 Precision of classification, compared to SSAIS and survey data ..................................................... 27
6 Conclusions and model improvement ...................................................................................................... 29
References ........................................................................................................................................................ 31
Annex 1 – Tables .............................................................................................................................................. 34
Table 1 Dependent variables of logistic regression model and model coefficient estimates .................. 34
Table 2 Share of people not included in the population by age group; 2012–2020 (%) ......................... 42
4
1 Introduction
Before 2011, the Central Statistical Bureau of Latvia (hereinafter – the CSB) estimated population number and
composition at the beginning of each year by using information from the Register of Natural Persons of the
Office of Citizenship and Migration Affairs (hereinafter – the OCMA). The estimate was based on individual
data about each person. By using Register of Natural Persons data, calculations regarding international long-
term migration were made as well (European Parliament and Council Regulation on Community statistics
regarding migration and international protection (11) stipulates that “emigration” means the action by which a
person, having previously been usually resident in the territory of a Member State, ceases to have his or her
usual residence in that Member State for a period that is, or is expected to be, of at least 12 months;
“immigration” means the action by which a person establishes his or her usual residence in the territory of a
Member State for a period that is, or is expected to be, of at least 12 months, having previously been usually
resident in another Member State or a third country). Natural increase over the year was estimated based on
the information in civil status registers.
Population and Housing Census 2011 (hereinafter – the Census 2011) helped to acquire all the information
necessary for the CSB to specify population and composition thereof in the country. On 1 January 2011,
population of Latvia accounted for 2 074.6 people. Compared to the information published before, population
declined by approximately 155 thousand people or 7 %. It confirmed that part of the population does not fulfil
requirements of the Law on the Register of Natural Persons (16) and that the Register of Natural Persons
information on actual population of the country is incomplete.
In line with Article 4 of the Regulation of the European Parliament and of the Council on demography statistics
(13), Member States may estimate the total population from the legally resident or registered population using
scientifically-based, well-documented, and publicly available statistical estimation methods. Until now, no
common method has been developed in the European Union (hereinafter – the EU) countries for population
estimate. European countries are using differing approaches for population number and composition estimates,
e.g.,:
• based on the information in the Register of Natural Persons (e.g., in Finland (8) etc.);
• by compiling information from local governments (e.g., in Germany (10));
• by using information from several administrative registers (e.g., in Netherlands (6));
• in countries not having identity code, by using data of the Census, sample surveys, and information
compiled by other organisations (e.g., in united Kingdom (5), Ireland (4)).
As of 2016, Statistics Estonia is estimating population number by using a new method that is based on
recording the 'life features' in administrative registers about each person, resulting in the residence index
estimate based on mathematical methods. The index is used to find out place of usual residence of each
person – Estonia or abroad.
5
The number of respondents in the surveys conducted by the CSB is not sufficient to produce high-quality
information in the breakdowns needed for national and international data users. Population statistics shall be
produced in breakdown by gender, age, administrative territory (regions, municipalities, parishes and
municipality towns, neighbourhoods of Riga and Valmiera, densely populated areas), etc.
CSB publishes population statistics based on both administrative territorial division in force until 1 July 2021
and that in force after administrative-territorial reform (data for 2021 and 2022).
CSB population estimate method is based on the data of the Register of Natural Persons as well as other
administrative registers used to estimate the usual place of residence (in Latvia or abroad) of each person
registered in Latvia at the beginning of the year.
CSB uses administrative data not only to calculate population, but also to specify it and for quality control.
Such processing is done for public interest and is provided for in the Statistics Law (18) and respective
Regulations of the Cabinet stipulating measures for protection of sensitive data. Processing of the data used in
the calculations is performed in line with the requirements of Personal Data Processing Law (17) and EU
regulatory enactments (12).
In compliance with the Article 2 (b) of the Commission Implementing Regulation (EU) No 205/2014 laying
down uniformed conditions for the implementation of Regulation (EU) No 1260/2013 of the European
Parliament and the Council on European demographic statistics (13), as regards breakdowns of data, deadlines
and data revisions (14), also methodology used to calculate natural increase was changed. The number of births
includes also children born outside Latvia to mother (permanent resident of Latvia) temporarily (for less than
one year) abroad.
After the transition to an extended use of administrative data, Population and Housing Census 2021 does not
take place in not face-to-face surveys anymore. Statistical data on population, which are supplemented with
population and housing census indicators according to Commission Implementation Regulation (EU)
2017/543 laying down rules for the application of Regulation (EC) No 763/2008 of the European Parliament
and of the Council on population and housing censuses as regards the technical specifications of the topics and
of their breakdowns.
2 Definitions
Usually resident population – resident population of the corresponding administrative territory only includes
persons who have lived in their usual place of residence for at least 12 months as well as persons who have
arrived at their usual place of residence with an intention to stay there for at least one year.
Emigrants – emigrants of the respective administrative territory are the persons discontinuing their usual
residence in the respective territory for a period lasting for or expected to last for at least 12 months.
Immigrants – immigrants of the respective administrative territory are the persons moving to this territory for
a permanent residence from another administrative territory for a period lasting for or expected to last for at
least 12 months.
6
Actual residence – a place of residence defined based on the Population and Housing Census 2011 data and
revised in line with the changes in OCMA Register of Natural Persons showing change of the registered place
of residence and taking place after the critical moment of the Census 2011, as well as based on the information
available in other administrative sources on people in institutional dwellings. Actual residence is published
until 2019 (including).
Registered residence – place of residence in Latvia defined based on the address registered in the OCMA
Register of Natural Persons and information available in other administrative sources on people in institutional
dwellings, as well as by revising registered address of certain population groups (children registered without
parents; fathers registered with children only; persons not having address code). Data by registered residence
are available for the period as of 2020.
Institutional dwellings – various organizations or institutions (hospitals, elderly care facilities, monasteries,
barracks, prisons, etc.) accommodating people and providing them with the necessary shelter and provisions
for a certain period as well as dormitories of educational institutions.
Institutional households – people the need for shelter and provisions whereof is fulfilled by an institution.
Institution is a legal entity providing a group of persons with a short-term accommodation and services.
Institutions usually have facilities and premises that are shared by residents (bathrooms, lounges, dining rooms,
sleeping quarters, etc.) (15).
3 Data sources
3.1 Office of Citizenship and Migration Affairs
Population is estimated by using the data files provided by OCMA from the Register of Natural Persons
(Common Migration Information System sub-system), i.e., information on gender, birth date, birth country,
citizenship, ethnicity, legal marital status, code of administrative territory of registered place of residence in
compliance with the Classification of Territories and Territorial Units (CATTU) (21), type of residence permit,
etc.
Since January 2013, CSB every year receives information on registration of civil status (marriages, births and
deaths) from the OCMA Civil Status Registration Common Information System (hereinafter – the CARIS),
which is a sub-system of OCMA Common Migration Information System (previously information was
received from county civil registry offices in a paper form). Based on this information, the annual data on the
number of births and deaths are produced.
As in compliance with the Section 24 of the Civil Status Registration Law (19), the civil status registry office
(incl. consular representation offices abroad) has to be informed about birth of a child within a month after the
birth, the Register of Natural Persons data received from the OCMA at the beginning of the year are revised
by also including children born in December of the previous year, but registered in January and February of
the current year. In line with the Section 38 of this Law, death should be reported to the civil status registry
office no later than within six working days after the death or after a person was found dead, therefore Register
7
of Natural Persons data received from the OCMA at the beginning of the year are revised by excluding persons
whose death was registered at the beginning of the current year, but who died at the end of the previous year.
In case of stillbirth or child death during delivery, in line with the Section 28 of the Civil Status Registration
Law, the medical institution or specialist is obliged to inform the civil status registry office about this fact
within eight days. With an aim to include also children born outside Latvia to a mother (usually resident of
Latvia) while temporary living abroad (for less than one year), the information of the Register of Natural
Persons about children born outside Latvia and having Latvia as their country of residence is used. Before
including child in the number of births, information on the place of residence of the mother in the previous
year is assessed, and information is compared with the State Health Service data on whether infant has received
state-funded health services (in compliance with the paragraphs 1.1.2–1.1.3 of the Annex 1 to the Regulations
No 55 of the Cabinet “Procedure for Health Care Organisation and Funding” (20) of 2018, until the age of six
months child has to undergo preventive examination once a month, from seven to eleven months – twice during
the mentioned period; if child does not have the examination, nurse or doctor’s assistant makes a home visit).
If the records on a child born abroad and afterwards registered in Latvia are insufficient to define the child as
a usually resident of Latvia, the child is excluded from the birth number as well. Birth number also does not
include children born abroad to a mother residence whereof was not registered in Latvia at all, as well as
children whose both parents are foreigners (do not have identity code).
Using the data files received from the OCMA in January, February and March of the current year as well as
November and December of the previous year, the population of Latvia registered on January 1 of the
respective year, or the Register of Natural Persons frame, is created. The frame serves as a base of the
population estimate. By evaluating person's activities (information about the person in various administrative
registers), part of the people included in the Register of Natural Persons frame are included in the population
estimate, while part of the people are so inactive that it may be concluded that they are abroad.
Register of Natural Persons frame is revised by using information on births and deaths, changed identity codes,
as well as people living in institutional dwellings. Register of Natural Persons frame is supplemented with
people temporary excluded from the OCMA register files, as they are undergoing process of citizenship
change, as well as have registered their place of residence abroad for 1–3 months and will return in February
or March. People born more than 110 years ago, those who have lost their judicial status, citizens of other
countries registered at their workplaces, as well as people imprisoned abroad also are excluded from the
Register of Natural Persons frame.
In 2021 the data on COVID tests and vaccination from the NHS data file were excluded to ensure quality of
the incoming data.
3.2 Other administrative registers
Apart from the OCMA data, population statistics is also produced by using other administrative registers
available to the CSB. The information meeting the needs of the statistical model is that available starting from
8
2010. CSB has access to the data in administrative registers managed by the following institutions:
• State Revenue Service (SRS);
• State Social Insurance Agency (SSIA);
• Ministry of Education and Science (MES);
• State Education Development Agency (SEDA);
• Agricultural Data Centre (ADC);
• Rural Support Service (RSS);
• National Health Service (NHS);
• State Employment Agency (SEA);
• Road Traffic Safety Directorate (RTSD);
• Ministry of Welfare (MW);
• Until 2017 data from higher education institutions (since 2018 – from MES):
o University of Latvia (UL);
o Riga Technical University (RTU);
o Transport and Telecommunications Institute (TSI);
o Riga Teacher Training and Educational Management Academy (RTTEMA)
1
;
o Baltic International Academy (BIA);
o Rezekne Academy of Technologies (RTA)
2
;
o Liepaja University (LiepU);
o Riga Building College (RBC);
o Riga Technical College (RTC);
o Malnava College (MC);
o Riga Medical College of the University of Latvia (RMCUL);
o Daugavpils Medical College (DMC);
o Riga 1st Medical College (R1MC);
o Jāzeps Vītols Latvian Academy of Music (JVLMA);
o Latvian Academy of Culture (LAC);
o Latvian College of Culture at the Latvian Academy of Culture (LAC LCC);
1
Since 2018, RPIVA is a part of University of Latvia and Jāzeps Vītols Latvian Academy of Music.
2
Until 1 January 2016 – Rezekne Higher Education Institution.
9
o State Police College (SPC);
o Red Cross Medical College of Riga Stradiņš University (RSU RCMC);
o Stockholm School of Economics in Riga (SSE Riga);
o RISEBA University of Applied Sciences (RISEBA).
The CSB also has information from the Social Security Administration Information System (hereinafter –the
SSAIS) on persons who have received social benefits from local governments The information is available
from 2012 onwards (in 2012, there are no data on eight municipalities), therefore the SSAIS data are used for
specifying population within the age group 18–30 years and for assessing the quality of the model. As of 2017,
SSAIS data on people in local government long-term social care institutions and people using night shelter
services are available.
The CSB is continuously evaluating additional data sources that might be used for the production of official
statistics, and, as a result, agreements were concluded on acquisition of additional administrative data that are
suitable for the production of population statistics:
• As of 2015, population is specified as well as registered (declared) place of residence is defined by
also using Prison Administration (PA) data on the persons in imprisonment.
• As of 2016, population is specified as well as registered (declared) place of residence is defined by
also using SSIA data on persons who are recipients of benefits/ pensions and are in social care
institutions.
• As of 2018, the CSB uses data of the Ministry of Welfare on people in social care institutions.
• As of 2018, to specify the number of population the CSB has access also to the information on
students from higher education institutions of Latvia (source – MES).
• As of 2019, data of the State Education Quality Service on children not registered with educational
institutions are available.
• As of 2020, Maintenance Guarantee Fund data on persons having submitted a claim for
maintenance , on persons for whose sustenance maintenance is paid, and on persons in lieu of which
maintenance is paid, are available.
3.3 CSB databases
Population estimate also is based on several CSB databases used to select the people as well as ensure that the
data are mutually harmonised.
Depending on the population estimate stage in which the data are used, they may be broken down into:
• Data needed to create specified population datafile (estimate):
✓ birth database;
10
✓ death database;
✓ marriage database;
✓ identity code change database;
✓ institutional dwelling database;
✓ people to be excluded from the Register of Natural Persons frame.
• Logistic regression model needs data which are processed before the use thereof:
✓ data on students of higher education institutions on all years used for the estimate – the data
are updated annually starting from the period on 2011;
✓ SSIA data on benefit recipients in the running year;
✓ internal migration database of the running year.
• Database of young people etc. specific groups:
✓ SSAIS data on benefit recipients aged 18–30 on the running year;
✓ full-time students of all higher education institutions (except for Alberta College) at the
estimate moment;
✓ enrolments of all Latvian pre-school education institutions and schools (except for distance
learning programmes) aged 19 and younger in the running school year;
✓ residents of places of detention, long-term child care institutions (children's homes), medical
institutions and old people’s, care and nursing homes at the estimate moment;
✓ refugees at the estimate moment.
• The data necessary to determine place of residence – State Land Service Address Register (VARIS)
addressing object code (hereinafter –addressing object code) – CATTU transition table.
Birth database additionally to CARIS data includes also:
• children born abroad to mothers residence whereof is registered in Latvia in previous year (at the
beginning or end) and by using population estimate method included in the usually resident
population of Latvia;
• children of foreigners born in Latvia the birth certificate whereof was issued not by OCMA, but by
the consular service of the country of parent nationality, therefore these children are not included
in CARIS data, but, by using population estimate method, are included in the usually resident
population of Latvia.
Death database is based on the CARIS data. In a person browser, information on people not included in
CARIS database, but possible dead, is specified.
11
If the person:
• is citizen or non-citizen of Latvia;
• at the beginning of the previous year was included in the OCMA register;
• at the beginning of the running year is not included in OCMA register anymore, possibility of
whether a person has passed away abroad is verified.
If it is found that the person shall be included in the death database, the date of its death is specified.
Marriage database is used to link correct family status (code) to the people married or divorced at the end
of the year or married or divorced abroad.
Identity code change database is used to link information on the person in registers with the old and new
identity code.
Internal migration database is used as the model testing approved that declaration of the residence in
other municipality increases probability that person is resident of Latvia.
Institutional dwelling database has two parts:
• persons to be included in the estimate definitely – refugees, prisoners, social care clients
(approximately 17 thousand people);
• persons living in institutional dwelling, but depending on their activities might not be included in
the estimate – people living in dormitories of educational institutions, monasteries, social shelters
(approximately 6 thousand people).
Each of the parts have their own addressing object codes that do not overlap.
Population estimate does not include:
• people registered in the blocked addresses (addresses with more than 20 people registered, e.g.,
employees registered at workplace address). Each address having 20 or more people registered and
not classified as institutional dwelling is evaluated separately, verifying citizenship and place of
work as on December 2020 of persons registered there, as well as information on the certain address
available in the Business Register. According to export estimate, citizens of Latvia registered in
such addresses are kept in the population frame, while other people are excluded from the estimate;
• people who have lost their judicial status. OCMA population list has persons on which OCMA
have information that they do not live in Latvia, but there is no documented proof of that, however
the information is available in the OCMA Register of Natural Persons Personal data online browser.
If prior it was found out that person has lost his/her legal status, it is assumed that the person has
the same status also this year. Persons the reasons for changing residence whereof are INL (leaving
Latvia) and SAN (due to death) are also excluded. If person is aged 75 or over, but was not included
in OCMA, SRS or SEA data in the previous year, by using additional information on the person’s
legal status available in the personal data browser, the residence status etc. is evaluated and a
12
decision is made on whether the person shall be included in the Register of Natural Persons frame
or not.
Database of young people etc. specific groups is used to specify emigration data mainly in the youth age
group within the second estimation stage, revising the probabilities used based on the migration. Part of persons
included in this database increases the number of persons included in their age/sex group, but part, regardless
of estimated probability, is included in the usually resident population. The database includes:
• persons aged 18–30 receiving local government and SSAIS benefits;
• enrolments aged 19 and younger of all Latvian pre-school education institutions and schools
(except of vocational and distance learning schools);
• full-time students of all higher education institutions (except for Alberta College as it offers
distance learning);
• residents of prisons, children's homes, medical institutions and old people’s homes;
• refugees.
Refer to the Section 4.8.
Addressing object code – CATTU transition table is used to determine place of registered residence.
The table is updated based on the State Land Service information by including the information on the
addressing object code of each apartment in the housing and house, CATTU code, reason behind the change
(changes in boundaries, correction of an error in the register), and flag for institutional housings.
Databases for specifying addressing object codes
Persons might be granted only such addressing object code of the registered place of residence, which,
accordingly State Land Service information, corresponds to the real place of residence. However, comparing
addressing object codes indicated in OCMA register with codes indicated in Address Register, persons were
found that:
• registered in an apartment, but actually living in a house because it is not divided into apartments;
• registered in a house, but actually living in an apartment, as there are several apartments in this
house;
• village code is indicated as addressing object code of the place of residence.
To adjust these inaccuracies and change addressing object code to the code of the most credible place of
residence of a person, the following is prepared:
• databases of these persons and housings;
• database with areas of housings.
3.4 Migration mirror statistics
International long-term emigration from residence country to another country corresponds with the
international long-term immigration from the residence country to the respective country. This correlation is
called mirror statistics, and it is used to estimate international long-term emigration.
13
When estimating international long-term emigration, the information regarding immigration from Latvia
received from other countries – Denmark, Finland, Sweden, Norway, Spain, the Netherlands, Austria, Iceland,
Germany – is used. Not all countries produce data on immigrants from Latvia, as Article 3 of the Regulation
of the European Parliament and of the Council on Community Statistics on Migration and International
Protection (11) stipulates that countries may provide information on immigrants in breakdown by group of
previous usual place of residence: EU Member States, European Free Trade Association countries, candidate
countries, and other non-member countries.
To estimate emigration of Latvia population to the United Kingdom and Ireland, the information on the number
of UK National Insurance Numbers granted for the first time and the number of Ireland Personal Public Service
Numbers granted for the first time is used. It should be noted that these data are used only to evaluate general
trends, because the respective numbers are given also to the Latvian residents staying in the UK or Ireland for
less than one year, moreover, one person may be registered several times if he/she has reported leaving the
country and has arrived there repeatedly (refer to the Section 4.3). Experts of the UK and Ireland statistical
institutes have pointed out that the number of registered persons in the above mentioned systems cannot be
used as the number of immigrants from Latvia; the numbers may be used only to see some general trends. The
reason behind such a statement is that that the system contains also information on the persons planning to stay
in the country for less than one year; if person is arriving to the country repeatedly, he/she does not have to
register in the system again. Moreover, the Ireland system contains information only the persons aged 15 or
older. Every year Statistics Ireland analyses Personal Public Service Numbers granted, and the tends show that
every year the number of persons granted Personal Public Service Numbers and employed or receiving social
benefits, i.e., residing the country, is growing. Only 37 % of the foreigners who were granted Personal Public
Service Number in 2011 were employed, whereas in 2016 the indicator went up to 54.7 % (1). However in the
case of Ireland, the data on the number of foreigners granted Personal Public Service Number are produced
based on the citizenship and not the former residence country, therefore not only people arrived from Latvia
but also those coming from e.g., United Kingdom and having Latvian citizenship, are included.
3.5 Sample surveys of individuals conducted by the CSB
The accuracy of the population estimate is evaluated by using data from various CSB surveys that contain
identity codes of respondents:
• Labour Force Survey (LFS) (starting from 2011);
• “EU Statistics on Income and Living Conditions” (EU-SILC) survey (from 2011);
• European Health and Social Integration Survey (EHSIS) (on 1 September 2012);
• European Health Interview Survey (EHIS) (at the end of 2014 – beginning of 2015 and at the end
of 2019 – beginning of 2020);
• Community Survey on ICT Usage in Households and by Individuals (ICT) in 2017-2020, 2021;
• Adult Education Survey (AES) in 2016;
14
• Mobility Survey of Latvia Population (MOBA) in 2017;
• External Migration Survey (ĀMA) in 2017 and 2018.
• “Careers of Doctorate Holders” (DH) survey in 2019.
While getting prepared for the Census 2021, in 2015 the CSB carried out a Population Microcensus. The
Microcensus data were used to estimate the overall international immigration in Latvia during 2015 and will
be used to evaluate accuracy of the population statistics produced.
External Migration Survey was conducted in 20 thousand households by surveying respondents twice – at the
end of 2017 and 2018 (aiming to find out information about 2016, 2017 and 2018). Each survey wave allowed
to acquire information from more than 35 thousand households, 7 % of which filled in electronic questionnaires
on-line.
4 Models and assumptions
4.1 Logistic regression model
Logistic regression model (hereinafter – the model) is based on the assumption that probability that a person
registered in the Register of Natural Persons of Latvia with an index ( varies from to ) is usually resident
of Latvia may be expressed as:
where the dependent variable (indicated with s distributed binomially.
In its turn, the binary auxiliary variables of the person , indexed with an index (from to ), which take
the values:
For example, if -th binary variable is information on whether the person receives old-age pension, then
for all persons i receiving old-age pension, and 0 for all other persons.
With the help of administrative register data, 206 binary variables were developed on each person registered
in the Register of Natural Persons of Latvia, e.g., gender, age groups, indicators on the fact that person has
received wage or salary, social benefits, has acquired education, etc. (refer to the Table 5 in the Annex 2).
To produce estimates of the model coefficients , the data on usual place of residence acquired within the
Census 2011 as well as information in administrative registers on 2010, 1 January 2011, or 1 March 2011 was
used.
During the model development phase, several possible model versions that differed in the choice and
construction of explanatory or independent variables were studied. Also, different variables derived from the
15
administrative registers available to the CSB were created and included in the model, e.g., based on the age of
a person the binary variables characterising belonging of the person to one of the ten age groups (0–9, 10–19,
…, 90–99, 100+) or one of the five age groups (0–4, 5–9, …, 95–99, 100+).
The explanatory variables were standardised as follows:
where –value of the k-th variable of the i-th person,
– the average value of the k-th variable,
– standard deviation of the k-th variable.
Model versions were compared by their predictive power. The model predictive power was measured with the
help of pseudo Nagelkerke determination coefficient that is calculated as follows:
where is a model without independent variables and is a model with selected independent
variables, and is a likelihood function. The maximum value of Nagelkerke determination coefficient is one.
Model coefficient estimates were analysed by their statistical significance ().
The model was initially formed by using solely OCMA data. best model based on OCMA data explained only
13 % of the model dependent variable dispersion (by pseudo Nagelkerke determination coefficient), therefore
model was supplemented with additional independent variables that were created by using the data from
administrative registers available to the CSB. In order to determine how additional administrative register data
influence the prediction power of the model, the data were included in the model sequentially (refer to the
Table 1).
Table 1 Comparison of logistic regression models
No
Administrative register data included in logistic regression model
Nagelkerke
1
OCMA
1 047 445
0.13
2
OCMA; SRS
823 391
0.35
3
OCMA; SRS; RSS; ADC; SEA
763 518
0.41
4
OCMA; SRS; RSS; ADC; SEA; SSIA
674 877
0.49
5
OCMA; SRS; RSS; ADC; SEA; SSIA; MES; UL; RTU; information about parents3
594 059
0.56
6
OCMA; SRS; RSS; ADC; SEA; SSIA; MES; UL; RTU; information about parents;
SHS
527 458
0.61
3
Information about parents is nine model variables containing information about parents of a person. The variables are
calculated only on persons aged 0–25. Value of the variable used for other persons equals 0. The data sources used
for these nine variables are described in the Table 5 of the Annex 2 “Dependent variables of logistic regression model
and model coefficient estimates”. The description of the variables includes a phrase “only for persons aged 0–25”.
16
No
Administrative register data included in logistic regression model
Nagelkerke
7
OCMA; SRS; RSS; ADC; SEA; SSIA; MES; UL; RTU; information about parents;
SHS; CSDD; TSI / RTTEMA / BIA / RTA / LiepU
508 462
0.61
8
OCMA; SRS; RSS; ADC; SEA; SSIA; MES; UL; RTU; information about parents;
SHS; CSDD; TSI / RTTEMA / BIA / RTA / LiepU / RBC / RTC / MC / RMCUL /
DMC / RMC1 / JVLMA / LAC / LAC LCC / SPC / RCMC / SSE Riga
506 414
0.61
Within the step 8, the model was able to explain 61 % of dependant variable dispersion (by pseudo Nagelkerke
determination coefficient).
Model compliance to data was assessed additionally by calculating negative double-log likelihood (-2 log
likelihood () as follows:
,
In statistics is a badness-of-fit indicator. The greater the value, the worse is the fit of the model
to the data.
Table 1 contains pseudo Nagelkerke determination coefficient and value changes depending on the
volume of additional information used in the model. There were no cases where both statistics give
contradictory model comparison (if model has a higher Nagelkerke determination coefficient, as compared
to model , then in all cases the value for model was lower than the value for model ). All of
the developed models were assessed additionally by performing demographic analysis of the model results.
Initially model was created with the help of binary logistic regression algorithm integrated in the statistical
calculation software IBM SPSS Statistics (2). As the model was developed further, the software code was re-
written in the statistical programming language R (7). Independent variables of the developed model and
estimates of the coefficients thereof are presented in the Table 5 of the Annex 2.
For each person registered in the Register of Natural Persons of Latvia the estimated probability with
which this person may be regarded as usually resident of Latvia may be expressed as follows:
4.2 Determining registered place of residence
Until 2020 (including) for all persons included in the population estimate, both actual and registered place of
residence was determined, but as of 2020 only registered place of residence is determined by publishing both
versions of the summary tables on the website databases from 2014-2019.
Actual place of residence is determined based on the place of residence specified in the Census 2011 and
changes made in the OCMA Register of Natural Persons. However, as years pass by, it becomes more and
more difficult to determine place of actual residence, as people in Register of Natural Persons change their
registered place of residence not only when moving to another residence, but also due to other reasons, e.g., to
avoid paying higher real estate tax, a person not actually living in the address is registered there, or to have a
place in a suitable kindergarten child is registered within the administrative territory of that kindergarten. Often
17
when emigrating abroad, registered place of residence is not changed or it is done with a delay of several years.
Therefore, in the Census 2021 that was conducted solely based on administrative registers (23), only registered
place of residence was used and a decision was taken not to determine the actual place of residence anymore.
Data on the registered place of residence are published in the database also for 2020.
Registered place of residence is determined on several stages:
• finding initial addressing object code;
• allocating people into families based on family structure algorithm;
• adjustment of addressing object codes by changing addressing object code of a female living
separately to that of a dwelling in which her husband lives together with common children aged
0–15. Such an approach is used because, when evaluating family structure recorded in the
Census 2011, it was concluded that such an adjustment reflects actual situation more precisely. The
initial addressing object code is defined by using addressing object codes which are added from the
Register of Natural Persons January datafile. For persons that cannot be given an addressing object
codes in such a way, the codes together with a date are added from the Register of Natural Persons
datafile of December and November of the previous year and Register of Natural Persons monthly
datafile of February and March of the running year, verifying whether addressing object code is not
one of an institutional dwelling. Such codes are not added to ensure that dates match and
institutional dwellings include only persons residing them on 1 January.
For persons not having addressing object code, the addressing object codes of relatives are added in the
following order – spouse, mother, father, child (the youngest child having addressing object code), i.e., if
person is not married, addressing object code of a mother is added, etc. The codes of relatives are added
regardless of the initial CATTU code of a person and fact whether the relative is included in the population
estimate of the respective year, except for a separate cases when address of a child is changed to that of a
mother or father (refer to the explanation further in the text).
None of the personal addressing object codes is changed to institutional dwelling code, i.e., an addressing
object code of a relative in institutional dwelling is not added.
For children aged 15 and younger (except for those living in institutional dwellings), a fact whether also any
of the parents is registered in the dwelling is verified. If child is registered alone (without any of the parents),
his/her addressing object code is changed to that of a mother (if mother does not have one, to that of a father).
Address of a child is changed to the dwelling address of parents also if child is registered with another adult,
e.g., grandmother, as in line with the UN Population and Housing Census recommendations (9) family (family
nucleus) is formed by family members of two consecutive generations (children and parents). Addressing
object code of a child is not changed to that of a parent if population estimate method allowing to conclude
that any of the parents lives abroad. As this approach does not fully allow to eradicate the number of cases
when child is the only occupant of the dwelling, since 2021 situations when none of 0-15 years old child’s
18
parents lives in Latvia, but a child does not live in institutional housing, it is assumed that most probably he/she
lives with some of grandparents, and his/her addressing code is changed to some of grandparents’ addressing
codes.
Within the second stage of determining registered place of residence, with the help of family nucleus analysis
algorithm the people living in private dwellings are broken down into families. After dividing persons into
families, residences of separate people are adjusted by changing addressing object code of a female living
separately to that of a dwelling in which her husband lives together with common children aged 0–15.
This adjusted database is supplemented with persons the address codes whereof do not meet the population
group “family” defined in the Population and Housing Census methodology, i.e., people living in institutional
dwellings, people living in houses not separated into apartments, as well as single people (not having spouse,
parents and children) not having addressing object code and residence place whereof is known only at
municipality level. In line with the updated addressing object code-CATTU transition table, registered
residence of each person is defined at the level of municipality rural territory, city and municipality – based on
both administrative territorial division in force until 1 July 2021 and that in force after administrative-territorial
reform.
Addressing object codes of place of residence are specified for persons whose codes, accordingly State Land
Service information, do not correspond to the actual place of residence, by changing them to a housing code
of relative or other suitable housing code.
4.3 International long-term emigration estimates
International long-term emigration is estimated based on the data sources listed in the Section 2.4.
In 2012 after the Census 2011, CSB recalculated emigration indicators on the period from 2000 to 2010. The
recalculation was carried out by using information from the Census, which included questions on emigration
from Latvia, as well as data from the Register of Natural Persons on emigration registered during this period.
The data obtained were compared with other country statistics on immigration from Latvia. As a result, it was
concluded that migration calculation corresponds to the statistics on immigrants from Latvia produced in other
countries (mirror statistics).
International emigration change coefficient is calculated starting from 2011. E.g., to compare 2010 with
2011, a coefficient was calculated, to compare 2011 with 2012 – , etc. To estimate the total
emigration in 2011, the previously calculated coefficient was attributed to the volume of emigration
estimated for 2010, whereas to get the total volume of emigration in 2012, the coefficient was attributed
to the volume of emigration estimated 2011. In such a way, the total annual emigration is calculated:
e.g.:
As a result, the estimate of international long-term emigration from 2011 onwards was acquired.
In order to break down emigration into country groups, initially the emigration estimate by country is
developed.
19
Based on the information in the OCMA Register of Natural Persons, the data (country of residence) are
developed on each emigrant; the persons that do not have record on other country of residence (in line with
the Register, their country of residence is Latvia) are allocated to the same country as similar emigrants (based
on demographic characteristics).
• In line with the other country statistics on immigration from Latvia (Denmark, Finland, Sweden,
Norway, the Netherlands, Austria, Germany, Spain, Iceland) the number of persons emigrated to
these countries is specified. Emigration from Latvia to the United Kingdom is estimated by using
information on changes in the number of National Insurance Numbers granted for the first time to
the Latvia residents arriving to the UK. Changes in emigration to the UK are found out by
calculating coefficient of changes in the number of first-time-granted national insurance numbers
in the UK (1), from 2011 onwards, attributing the amount of numbers granted during a year against
the previous year. In order to estimate emigration to the UK in 2011, the calculated coefficient
l2011 was attributed to the emigration of the Latvia residents to the UK calculated by the CSB in
2010 (in accordance with the international long-term emigration re-calculation carried out based
on the Census 2011 results) and, correspondingly, the coefficient l of each of the following years
is attributed to the calculated emigration to the UK in the previous year.
• Emigration from Latvia to Ireland is estimated by using coefficient of change in the number of
Personal Public Service Numbers granted in Ireland for the first time to the persons arriving from
Latvia (4). Similarly as it was in the case of the UK, the coefficient of change was used and
emigration from Latvia to Ireland in 2010 was calculated in line with the re-calculation of
international long-term emigration carried out based on the results of the Census 2011.
Immigration by country group is estimated by using OCMA Register of Natural Persons data. The persons that
do not have record on the previous country of residence are allocated to the same country as similar immigrants
(based on demographic characteristics).
While analysing international long-term emigration and immigration registered in the OCMA Register of
Natural Persons, the data at individual level were compared with the population number acquired.
Conclusions:
1. Only part of the persons who have received resident permits due to capital investment live in Latvia
on a permanent basis (in line with the Section 23, Paragraph 28 of the Immigration Law (22) – a
foreigner shall have the right to apply for a temporary residence permit for a period of time
not exceeding five years if he or she has made investment in the equity capital of a capital company)
or due to purchase of real estate (in accordance with the Section 23, Paragraph 29 of the Immigration
Law – a foreigner shall have the right to apply for a temporary residence permit for a period of time
not exceeding five years if he or she has acquired in the Republic of Latvia and he or she owns one
functionally related built-up real estate). E.g., out of almost 4 000 persons who received Latvia resident
permit in 2013 due to the purchase of real estate or capital investment, only approximately 10 % were
included in the CSB population estimate (previously already pointed out by the OCMA).
20
2. Emigrants do not immediately provide information to the Register of Natural Persons. For example,
according to the CSB estimate, two thirds of the emigrants registered in 2013 have actually emigrated
already in the previous years and are included in non-registered emigration. It was also confirmed by
the Census 2011 data: the registered year of emigration not always corresponds with the actual
situation, since many people register their place of residence abroad only when it is required by some
specific reasons.
3. Not all children (aged 0–1) born abroad and having place of usual residence registered in Latvia
(Register of Natural Persons) are included in the total population number. In accordance with the
population estimate, only a part of them live in Latvia permanently.
4.4 Producing statistics
Coding
Migration statistics is produced by defining variables for each person (refer to the Table 3).
Table 3 Variables used for population estimate
Variable
Name
Value
Description
Classification status at the
beginning and at the end of the
year
11
Person is registered as a usually resident both at the beginning and
at the end of the year
10
Person is registered as a usually resident only at the beginning of
the year
01
Person is registered as a usually resident only at the end of the year
00
Person is not registered as a usually resident both at the beginning
and at the end of the year
Classification status at the end
of the year
1
Person is classified as a usually resident at the end of the year
0
Person is not classified as a usually resident at the end of the year
NULL
Person is not registered as a usually resident at the end of the year
Classification status at the
beginning of the year
1
Person is classified as a usually resident at the beginning of the year
0
Person is not classified as a usually resident at the beginning of the
year
NULL
Person is not registered as a usually resident at the beginning of the
year
Classification status at the
beginning and at the end of the
year
11
Person is classified as a usually resident both at the beginning and
at the end of the year
10
Person is classified as a usually resident only at the beginning of the
year
01
Person is registered as a usually resident only at the end of the year
00
Person is not classified as a usually resident both at the beginning
and at the end of the year
Death characteristic
1
Person has died during the corresponding year
0
Person has not died during the corresponding year
Birth characteristic
1
Person was born during the corresponding year
0
Person was not born during the corresponding year
Binary variable characterising
affiliation of the person to the
domain
1
Person belongs to domain
0
Person does not belong to domain
21
Number of registered emigrants (total and in domain ) is calculated as follows:
(1)
(2)
Number of registered immigrants (total and in domain ) is calculated as follows:
(3)
(4)
Number of non-registered emigrants (total and in domain ) is calculated as follows:
(5)
(6)
Number of non-registered immigrants (total and in domain ) is calculated as follows:
(7)
(8)
Image 4 schematically shows the classification of persons depending on the values of pre-defined variables.
Image 4 Population classification scheme
22
4.5 Initial probabilities and thresholds for probabilities
From 2012 onwards, the initial probability to be classified as a usually resident of Latvia is estimated for each
person who is registered as a usually resident of Latvia at the beginning of a year. The initial probability is
estimated using personal variables (refer to the Section 4.1) that are calculated using administrative data (refer
to the Section 4. 1 and 3.2) and the logistic regression model (refer to the Section 4.1).
The initial population number estimate is calculated for groups (by totally acquiring 202 groups) in the
following breakdowns:
• males aged 0, 1 , 2, …, 99, 100 and older;
• females aged 0, 1 , 2, …, 99, 100 and older.
In each group , the estimate is acquired as the sum of initial probabilities of the respective group:
Where is the initial probability to be a usually resident of Latvia for person from group and is the
person index set for the group .
Residents in each group are sorted in descending order by probability and numbered according to the order:
, where is the number of registered persons in the group , . Within each group , a
person with sequence number
is found; the initial probability of this person is denoted
with . For the group , the probability threshold is defined:
4.6 Initial estimate of emigration and immigration
Net migration is the difference between the number of immigrants and emigrants during a year:
,
(13)
where is net migration during year;
is the number of immigrants during year;
is the number of emigrants during year.
Population at the beginning and at the end of a year is characterised by the following equation:
(14)
where is population at the beginning of the year;
is population at the end of the year;
is the number of births during the year;
is the number of deaths during the year.
23
The net migration may be expressed from the equation (14) as follows:
and estimated as:
.
(15)
The estimate
of the total emigration during a year is made with the method described in Section 4.3
(except for 2015). The total immigration
may be estimated with an equations (13) and (15) as follows:
.
The migration statistics of 2015 was estimated based on data of the Population Microcensus. By using the
respective data, the total immigration volume
was estimated, and by using equation (13) the total
emigration volume in
was estimated as follows:
.
Emigration and immigration estimates on the period from 2011 to 2019 are shown in the Table 4.
Table 4 Total emigration and immigration estimates
Year
Emigration (thousands)
Immigration (thousands)
2011
30
10
2012
25
13
2013
23
8
2014
19
10
2015
20
9
2016
20
8
2017
18
10
2018
15
11
2019
15
11
2020
12
9
2021
13
13
4.7 Migration estimates by gender and age group
At the beginning of each year, the values of the variables , , and are calculated for each person (refer
to the Table 2). At first, registration status at the beginning and at the end of a year is computed (. Persons
with residence permit in Latvia up to one year are counted separately (they are not included in the number of
usually residents registered in Latvia). The values of birth (DZi) and death (MIi) variables also are assigned for
each person in compliance with the civil status records.
Each person who was registered as usually resident of Latvia at the beginning of the year gets classification
status at the beginning of a year , and it is the same as the classification status at the end of the previous year.
Classification status at the beginning of a year was calculated for the first time for the beginning of 2011 by
using the data of Census 2011 as well as Register of Natural Persons information on January and February of
2011. For the following years, it was assumed that is equal to the of the previous year.
24
For each person who is registered as usually resident of Latvia at the end of a year, the initial classification
status at the end of the year is calculated:
There are cases when is adjusted based on additional administrative data not included in the model. Persons
which regardless of the estimated probability are included in the usually resident population:
• people aged 18–30 receiving local government and SSAIS benefits;
• enrolments of all Latvian pre-school education institutions and schools aged 19 and younger;
• full-time students of all higher education institutions (except for Alberta College) at the estimate
moment;
• residents of prisons, children's homes, medical institutions, religious organisations and old people’s
homes;
• refugees.
If person shall be included in the set of usually resident population, the values and of the persons are
defined as equal to .
If during the estimation process it is found out that the set of usually resident population includes person not
meeting inclusion criteria (e.g., person has died in the previous year, but the death fact was registered later),
the values and of the person are defined as equal to 0.
The initial classification status at the beginning and at the end of the year is calculated for each person as
follows:
where is the classification status of person at the beginning of a year.
With the help of , , , and , the , , , and for each group as well as ,
, , and totally in Latvia is calculated using formulas (1)–(8). Migration is estimated in each of
the 202 groups:
where
is the total emigration (refer to the Section
4.3), and
is the total emigration (refer to the Section
4.3).
25
4.8 Adjustments for probabilities
In each age and gender group , the initial probabilities are adjusted. It is done to ensure that migration statistics
acquired by summing adjusted probabilities is harmonised with the migration estimates
described in the Section 5.4. The probabilities are adjusted in three steps:
1) For each group , a constant is calculated in a way that by adding to all probabilities of the
group and summing them the emigration
estimated in the step 5.4 is acquired.
2) For each group , a constant is calculated in a way that by adding to all probabilities of the
group and summing them the immigration
estimated in the step 5.4 is acquired.
3) The adjusted probabilities are calculated in line with (16).
(16)
After the adjustment of probabilities, the adjusted classification status of each person () is computed:
By using adjusted classification status , registered and non-registered migration in each group is recalculated:
,
,
and
. Adjustment constant and in each group are defined in a way to
ensure that
is as close to
and
as possible and as close to
as possible.
After finding optimum and values, is adjusted also in line with additional administrative data not
included in the model. Irrespective of the value found, the set of usually resident population includes
residents of prisons, children's homes, medical institutions, old people’s homes as well as refugees. If
necessary, birth, and death databases are also adjusted to ensure that there are no logical contradictions. After
this adjustment, the final classification status is set, and it is not changed any more.
4.9 Producing population, natural increase and migration statistics
By summing the persons with , the total population of Latvia is calculated as well as in breakdown by
sex, age and municipality and any other random population domain. Thus statistics in all demography tables
is harmonised.
The data on natural increase are produced by using information from civil status registers managed by the Civil
Registry Offices (refer to the Section 3.3). To produce data on long-term international migration:
• persons with are included in immigration;
• persons with are included in emigration;
• persons with and died during the year in Latvia are included in immigration;
• persons with and born during the year in Latvia are included in emigration.
26
5 Estimating population number and composition precision
5.1 Precision of classification, compared to the data of Census 2011
One of the ways used to estimate precision of the classification is calculation of classification values for the
data of 1 March 2011 and comparison of the results at individual level with the data of the Census 2011. As a
result, it is possible to calculate confusion matrix characterising number of cases when the values defined by
the classification meet or do not meet the data acquired in the Census 2011. The confusion matrix may be
calculation for the whole population as well as random sub-set of the population, e.g., males, females, or people
at a certain age. When analysing confusion matrix the following characteristics shall be taken into account:
• the calculation is based on the classification values determined from the initial values (without
migration adjustment);
• all data sources have errors (also data of the Census 2011);
• confusion matrix characterises precision at a micro level (individual level). Precision at micro-level
(error of each individual) is not directly related to an error at macro-level (total population of a
country or population sub-set).
Table 5 Confusion matrix of total population and gender breakdown
Population
domain
Indicator
Census 2011 result
Classification
includes in the
population set
Classification does
not include in the
population set
Total
All population
Number
Census 2011 includes in the population set
1 912 867
154 848
2 067 715
Census 2011 does not include in the
population set
155 070
275
155 345
Total
2 067 937
155 123
2 223 060
Proportion
Census 2011 includes in the population set
0.8605
0.0697
0.9301
Census 2011 does not include in the
population set
0.0698
0.0001
0.0699
Total
0.9302
0.0698
1.0000
Males
Number
Census 2011 includes in the population set
862 811
81 956
944 767
Census 2011 does not include in the
population set
82 071
174
82 245
Total
944 882
82 130
1 027 012
Proportion
Census 2011 includes in the population set
0.8401
0.0798
0.9199
Classification does not include in the
population set
0.0799
0.0002
0.0801
Total
0.9200
0.0800
1.0000
Females
Number
Census 2011 includes in the population set
1 050 056
72 892
1 122 948
Census 2011 does not include in the
population set
72999
101
73100
Total
1 123 055
72 993
1 196 048
Proportion
Census 2011 includes in the population set
0.8779
0.0609
0.9389
Census 2011 does not include in the
population set
0.0610
0.0001
0.0611
Total
0.9390
0.0610
1.0000
27
5.2 Precision of classification, compared to SSAIS and survey data
Each person has an estimate acquired by using the model and it may not meet the actual residential status
of the person. By using the data of surveys and administrative registers described in the Section 3, it is possible
to estimate the share of people classified as living abroad (, but actually being residents of Latvia.
Target population of the surveys covers usually resident population of Latvia living in private households,
therefore only such persons are respondents of the survey. However, it should be taken into account that
population with the model is estimated at the beginning of the year, while reference periods of surveys usually
do not meet beginning of the year. Only persons living in private households are surveyed, thus the data cannot
be used to make conclusions about persons living in institutional dwellings.
Table 6 Share of people not to be included in population estimate (%)
Data source
population estimate year 01.01.
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
SSAIS (2012)
1,05
0,92
SSAIS (2013)
1,14
0,98
SSAIS (2014)
1,25
1,09
SSAIS (2015)
1,47
1,33
SSAIS (2016)
1,71
1,59
SSAIS (2017)
1,89
1,50
SSAIS (2018)
1,81
1,60
SSAIS (2019)
1,96
1,77
SSAIS (2020)
2,05
1,87
SSAIS (2021)
2,32
1,98
EHSIS (2012)
0,49
0,93
LFS (2011)
1,15
LFS (2012)
0,98
1,29
LFS (2013)
1,36
1,42
LFS (2014)
1,71
1,84
LFS (2015)
1,99
2,10
LFS (2016)
1,92
2,08
LFS (2017)
2,13
2,22
LFS (2018)
2,35
2,43
LFS (2019)
2,37
2,40
LFS (2020)
2,46
2,50
LFS (2021)
3,05
2,88
SILC (2011)
1,19
SILC (2012)
1,01
1,23
SILC (2013)
1,31
1,45
SILC (2014)
1,38
1,49
SILC (2015)
1,49
1,66
SILC (2016)
1,80
1,99
SILC (2017)
2,11
2,17
SILC (2018)
2,09
2,17
SILC (2019)
2,37
2,41
SILC (2020)
2,26
2,29
SILC (2021)
2,41
2,35
EHIS (2014)
1,43
1,54
EHIS (2019)
0,85
1,03
ICT (2017)
0,89
1,05
ICT (2018)
0,98
1,11
28
Data source
population estimate year 01.01.
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
ICT (2019)
0,75
0,79
ICT (2020)
0,41
0,53
ICT (2021)
0,38
0,43
ICT (2022)
0,74
AES (2017)
1,47
MOBA (2017)
0,67
0,77
ĀMA (2017)
3,56
3,60
ĀMA (2018)
3,66
3,73
DH (2019)
0,30
0,30
EU-GBV (2021)
1,42
1,56
Differences in the data and model classification results may be partly explained by a mismatch between the
data and population statistics reference periods, e.g., there are cases when person is indicated as an usually
resident of Latvia in a survey at the beginning of a year, while at the end of the year the same person has moved
to another country or died. If the changes are registered in the Register of Natural Persons, the data on those
persons are not used for the accuracy analysis. However, there are cases when a person who is indicated as an
usually resident in a survey changes his/her place of residence, but does not register it. It is possible that part
of the persons who are indicated as usually residents in the survey, but have not been included in the population
number, have been living in Latvia.
Approximately 9 % of Latvia’s residents receive local government benefits. The information on the persons
receiving benefits in 2012 was used to assess model quality, while starting from 2013 it is used to specify
population estimates. The SSAIS information cannot be used in the logistic regression model, as the CSB does
not have the local government data on 2010 and 2011.
It is more difficult to estimate another type of error – persons classified as usually residents (, but
actually are living abroad. The Labour Force Survey allows to acquire information on such persons, however,
their number is too small to make informative conclusions. Moreover, it is very likely that in several cases
persons that have arrived to or departed from the country during the year are classified wrongly, because such
persons may be very active in Latvia within a certain period of time and being registered in various databases,
while spending most of their time abroad, thus they should be classified as persons living abroad (.
In 2016, more detailed analysis of the declared places of residence of Latvia population was performed by geo-
referencing and visualising the respective data cartographically. The analysis allowed detecting places with a
great number of persons having declared their place of residence at the workplace thereof, therefore the method
used to estimate usually resident population was improved, and, in line with the internationally adopted
definition, the number of persons having declared their place of usual residence at their workplace (enterprise
registered in Latvia) but being foreigners and not living in Latvia actually (e.g., lorry drivers, who are guest
workers from other countries), was excluded from the number of usually resident population. The changes
were introduced starting from the data on 2017.
29
Due to the mentioned reason the population change, e.g., in former Mārupe municipality constitutes an increase
of 5 % in 2015, whereas in 2016 a decline of 0.5 %. The actual population increase in 2015 and 2016 constitutes
2–3 %. Similar situation may be observed in Riga and former Carnikava municipality, however population
change in these areas is smaller.
The largest share of the persons classified as living abroad ( but actually being residents of Latvia, is
registered in External Migration Survey (4.09 %), especially in the age group 28–36.
6 Conclusions and model improvement
The further work of the CSB on the methodology used to estimate population number will take place in two
directions – improvement of the existing logistic regression model for population number and migration flow
estimation as well as work on the development of new population statistics methodology (model). The CSB
aims at developing a model that is based on administrative data, but directly uses data of the Census 2011,
which are becoming increasingly outdated every year.
Improvement of the existing model envisages incorporation of additional explanatory variables that will be
developed from the data already available to the CSB or other supplementary data that may be accessed by the
CSB in the future. Verification and improvement of the existing model is also needed in cases when there are
changes in administrative data or their quality, structure or form of receipt. For example, as information in
respect to data of SEA and SRS self-employed persons changed, the pattern was adjusted for assessment of
2020.
Model improvement requires annual (with periodicity of at least one year) individual data on the time period
starting from 2010 (one year before the Census 2011). Availability of such a data is very limited, as the CSB
already has access to the individual data with the respective time series available in the largest Latvian
administrative sources. Also, the data on the several recent years or latest situation are very valuable to assess
quality of the estimated population number. To evaluate precision of the model, the CSB uses the following
additional individual data:
• individual data from higher education institutions of Latvia about students provided by MES;
• individual data of the Riga municipal limited liability company “Rīgas satiksme” about
personalised e-ticket;
• individual data from the SEDA regarding persons who have received a study/student loan for
studying abroad;
• data of the Ministry of Education and Science about children at school age who have not been
registered with any of the educational institutions.
With an aim to be able to estimate population in specific groups, also in future it has been planned to seek for
administrative registers that might have individual data, especially on young people, long-term emigrants and
homeless people.
30
While getting prepared for the Census 2021, in 2015 the CSB conducted a Population Microcensus with the
aim to acquire annual population, composition thereof, and long-term migration estimates independent from
the methodology used for producing population statistics described.
Analysis of the total population allows concluding that the difference between the population statistics and
results of the Microcensus constitutes 37 thousand (1.9 %), which is statistically significant difference, as the
margin of error is 29 thousand (relative margin of error is 1.5 %). The results of the Microcensus show that
the total population is overestimated.
There is a problem in the age group 19–24, in which in five cases verification of the hypothesis shows
statistically significant difference between the population statistics and results of the Microcensus. Population
statistics estimates exceed the Microcesnsus estimates. In this case, there is a reference to actual differences,
since the fact that there are differences in several age groups in a row shows that sampling error is not the
reason behind the difference.
Comparison of the Microcesnsus results with the population statistics approved that calculation of precise
migration flows would need panel survey conducted for at least two years in a row. In 2017, the first stage of
the survey was initiated. The sample includes 20 000 dwellings and it is aimed at asking about usual residents
of the dwelling on 1 December 2016 and 1 December 2017. A separate sample was formed for Valka
municipality and Valka town with an aim to estimate the number of employed people working in Estonia. The
survey includes questions about mother tongue, language mainly spoken at home and foreign language
knowledge.
The second stage of the survey took place from 1 October 2018 to 15 December 2018. Within it, in addition
to the questions about usual residents of a dwelling and questions about employment country asked to residents
of Valka municipality and town residents also questions about highest level of education completed and
country where education was acquired were asked. The results of the survey were published in the 2nd quarter
of 2019.
Results on migration from 1 December 2017 until 1 October 2018 (exact dates of the survey) are compiled,
which are compared to migration indicators of 2018 calculated by the CSB. The data collected (concerning
12 months) differ from the annual migration assessment – volume of emigration is larger and exceeds 20 %,
but volume of immigration is smaller (by 38 %). It is characteristic that at the end of the year number of
immigration cases is increasing, as people take decision about the future place of residence, but the survey did
not include last months of 2018. Another factor, also reflected by age structure of immigrants data, might be
language barrier. It is observed that until the age of 40 significantly less immigrants (mainly English-speaking
immigrants) were surveyed, opposite trend was observed in larger age groups. Breakdown of immigrants by
regions shows differences between the registered and actual place of residence. Main differences are observed
in immigration indicators in Riga, Kurzeme and Latgale.
External Migration Survey has provided valuable information on the quality of migration assessment of the
CSB, information is acquired that attention needs to be paid to migration volumes, however, it is positive that
31
most significant demography indicators have similar structure, but differences are explained with nature of
both population estimate methodology and survey of the CSB.
Within the framework of grant project G-19.10 "Urban and territorial statistics in 2019", in 2019 elaboration
of assessment of the population estimation method and new or improved method was started. For the purpose
of this test individual data were received from SSIA, CSDD and SEA on all cases, when the person has
appeared in some of administrative registers maintained by these institutions.
In 2020, along with the closing of the project G-19.10, testing of use of the Sol-Logit model for population
estimates was started. Considering that logic regression and Sol-Logit model give differing results, it was
decided to carry out coverage survey to evaluate precision of the methods and test some other methods that
may be suitable for population estimates based on administrative data. In 2020 a substantiation was assessed
and sample size was evaluated to organise coverage survey as well as CSB applied for a grant project to
develop and test other methods as well as organise a pilot survey in 2023–2024 and thus get ready for the
coverage survey.
References
1) Central Statistics Office (18 December 2017) Foreign Nationals: PPSN Allocations, Employment and
Social Welfare Activity. Access at:
https://www.cso.ie/en/releasesandpublications/er/fnaes/foreignnationalsppsnallocationsemploymenta
ndsocialwelfareactivity2016/
2) IBM Corporation (1989, 2011) IBM SPSS Statistics 20 Algorithms. Access at: http://www-
01.ibm.com/support/docview.wss?uid=swg27021213
3) Population estimate method used in Estonia in 2017. Access at: https://www.stat.ee/news-release-
2018-007
4) Population estimates in Ireland. Access at:
https://www.cso.ie/en/statistics/population/populationandmigrationestimates/
5) Population estimates in United Kingdom. Access at:
https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationestimat
es
6) Population statistics in the Netherlands. Access at:
https://opendata.cbs.nl/statline/#/CBS/en/dataset/37943eng/table?ts=1529480855777
7) R Core Team (2018). R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vīne, Austrija. Access at: https://www.R-project.org/.
8) Population statistics in Finland. Access at: http://tilastokeskus.fi/til/vrm_en.html
32
9) The United Nations Statistics Division (2017) Principles and Recommendations for Population and
Housing Censuses, Rev.3. Access at:
https://unstats.un.org/unsd/demographic/sources/census/census3.htm
10) Population statistics in Germany. Access at:
https://www.destatis.de/EN/FactsFigures/SocietyState/Population/CurrentPopulation/CurrentPopulat
ion.html
European Union Regulations
Document includes legal norms resulting from:
11) Regulation (EC) No 862/2007 of the European Parliament and of the Council of 11 July 2007 on
Community statistics on migration and international protection. Access at: https://eur-
lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32007R0862
12) Regulation (EC) No 223/2009 of the European Parliament and of the Council of 11 March 2009 on
European statistics and repealing Regulation (EC, Euratom) No 1101/2008 of the European Parliament
and of the Council on the transmission of data subject to statistical confidentiality to the Statistical
Office of the European Communities, Council Regulation (EC) No 322/97 on Community Statistics,
and Council Decision 89/382/EEC, Euratom establishing a Committee on the Statistical Programmes
of the European Communities. Access at: https://eur-lex.europa.eu/legal-
content/EN/ALL/?uri=CELEX%3A32009R0223
13) Regulation (EU) No 1260/2013 of the European Parliament and of the Council of 20 November 2013
on European demographic statistics.
Access https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32013R1260
14) Commission Implementing Regulation (EU) No 205/2014 laying down uniformed conditions for the
implementation of Regulation (EU) No 1260/2013 of the European Parliament and the Council on
European demographic statistics, as regards breakdowns of data, deadlines and data revisions. Access
at: https://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2014:065:0010:0026:EN:PDF
15) Commission Implementing Regulation (EU) 2017/543 of 22 March 2017 laying down rules for the
application of Regulation (EC) No 763/2008 of the European Parliament and of the Council on
population and housing censuses as regards the technical specifications of the topics and of their
breakdowns (Text with EEA relevance). Access at: https://op.europa.eu/en/publication-detail/-
/publication/3dcf9f49-0f95-11e7-8a35-01aa75ed71a1/language-en
33
Regulations of the Republic of Latvia
16) 20.06.2002. Law “Declaration of Place of Residence Law” (“Latvijas Vēstnesis”, 104 (2679),
10.07.2002, „Ziņotājs”, 16, 22.08.2002) [in force since 07.07.2003] with amendments. Access at:
https://likumi.lv/ta/en/en/id/64328-declaration-of-place-of-residence-law
17) 23.03.2000. Law “Personal Data Protection Law” (“Latvijas Vēstnesis”, 132, (6218) 04.07.2018) [in
force since 05.07.2018]. Access at: https://likumi.lv/ta/en/en/id/300099-personal-data-processing-law
18) 23.03.2015. Law “Statistics Law” (“Latvijas Vēstnesis”, 118 (5436), 18.06.2015) [in force since
01.01.2016] with amendments. Access at: https://likumi.lv/ta/en/en/id/274749-statistics-law
19) 29.11.2012. Law “Law on Civil Status Registration” (“Latvijas Vēstnesis”, 197, 14.12.2012) [in force
since 01.01.2013] with amendments. Access at: https://likumi.lv/ta/en/en/id/253442-law-on-
registration-of-civil-status-documents
20) 28.08.2018. Regulations of the Cabinet No 555 “Procedure for Health Care Organisation and Funding”
(“Latvijas Vēstnesis”, 176 (6262), 05.09.2018) [in force since 06.09.2018] with amendments. Access
at (in Latvian): https://likumi.lv/ta/id/301399
21) 21.03.2017. Regulations of the Cabinet of Ministers No 152 “Regulations for Classification of
Administrative Territories and Territorial Units” (“Latvijas Vēstnesis”, 61, 23.03.2017) [in force since
24.03.2017]. Access at (in Latvian): https://likumi.lv/ta/id/289582-administrativo-teritoriju-un-
teritorialo-vienibu-klasifikatora-noteikumi
22) 31.10.2002. Law “Immigration Law” (“Latvijas Vēstnesis”, 169 (2744), 20.11.2002) [in force since
01.01.2003] with amendments. Access at: https://likumi.lv/ta/en/en/id/68522-immigration-law
23) 02.06.2015. Regulations of the Cabinet of Ministers No 280 “On Plan of Activities for Preparation
and Organisation of Population and Housing Census 2021” (“Latvijas Vēstnesis”, 108 (5426),
04.06.2015) with amendments. Access at (in Latvian): https://likumi.lv/ta/id/274448-par-pasakumu-
planu-2021-gada-tautas-skaitisanas-sagatavosanai-un-organizesanai
34
Annex 1 – Tables
Table 1 Dependent variables of logistic regression model and model coefficient estimates
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
OCMA
Changed placed of residence during the year
0.081
0.005
0.000
1.085
OCMA
male
0.203
0.004
0.000
1.225
OCMA
single
-0.017
0.013
0.194
0.983
OCMA
married
-0.039
0.012
0.002
0.962
OCMA
divorced
-0.171
0.009
0.000
0.843
OCMA
born in Russia, Ukraine or Belarus
-0.023
0.005
0.000
0.977
OCMA
born in EU (except LV)
-0.086
0.003
0.000
0.918
OCMA
Estonian
-0.003
0.003
0.436
0.997
OCMA
German
-0.038
0.003
0.000
0.963
OCMA
Russian
-0.049
0.004
0.000
0.953
OCMA
Ukrainian
-0.021
0.004
0.000
0.979
OCMA
Pole
-0.028
0.004
0.000
0.972
OCMA
Jew
-0.108
0.003
0.000
0.898
OCMA
Roma
-0.016
0.002
0.000
0.984
OCMA
unspecified/ unknown ethnicity
-0.024
0.004
0.000
0.977
OCMA
lives in Kurzeme region
0.016
0.007
0.023
1.013
OCMA
lives in Pierīga region
0.013
0.006
0.017
1.008
OCMA
lives in Riga region
0.008
0.006
0.181
0.976
OCMA
lives in Vidzeme region
-0.024
0.006
0.000
0.978
OCMA
lives in city of Daugavpils
-0.022
0.004
0.000
0.978
OCMA
lives in Jelgava city
-0.022
0.004
0.000
0.978
OCMA
lives in Liepāja city
-0.045
0.005
0.000
0.956
OCMA
lives in Rēzekne city
-0.042
0.004
0.000
0.959
OCMA
lives in Valmiera city
-0.031
0.004
0.000
0.970
OCMA
lives in Ventspils city
-0.026
0.004
0.000
0.975
OCMA
lives in Tukums county
-0.022
0.004
0.000
0.978
OCMA
lives in Talsi county
-0.017
0.004
0.000
0.983
OCMA
lives in Strenči county
-0.015
0.004
0.000
0.985
OCMA
lives in Ropaži county
0.014
0.004
0.000
1.014
OCMA
lives in Ogre county
-0.014
0.004
0.000
0.986
OCMA
lives in Nereta county
0.027
0.004
0.000
1.028
OCMA
lives in Grobiņa county
-0.011
0.004
0.003
0.989
OCMA
lives in Cēsis county
-0.035
0.004
0.000
0.966
35
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
OCMA
lives in Burtnieki county
0.035
0.004
0.000
1.036
OCMA
lives in Valka county
-0.017
0.004
0.000
0.983
OCMA
lives in Rugāji county
0.016
0.004
0.000
1.016
OCMA
lives in Saldus county
-0.016
0.004
0.000
0.984
OCMA
lives in Mārupe county
0.025
0.004
0.000
1.025
OCMA
lives in Ērgļi county
-0.014
0.004
0.000
0.986
OCMA
lives in Brocēni county
-0.015
0.004
0.000
0.985
OCMA
lives in Līvāni county
-0.013
0.004
0.000
0.987
OCMA
lives in Auce county
-0.023
0.004
0.000
0.978
OCMA
lives in Krāslava county
-0.019
0.004
0.000
0.981
OCMA
lives in Jēkabpils county
0.025
0.004
0.000
1.026
OCMA
lives in Dobele county
-0.017
0.004
0.000
0.983
OCMA
lives in Balvi county
-0.016
0.004
0.000
0.984
OCMA
lives in Alūksne county
-0.021
0.004
0.000
0.979
OCMA
citizen of Latvia
-0.015
0.006
0.012
0.985
OCMA
aged 1
-0.068
0.009
0.000
0.860
OCMA
aged 2
-0.150
0.008
0.000
0.826
OCMA
aged 3
-0.191
0.008
0.000
0.793
OCMA
aged 4
-0.232
0.008
0.000
0.769
OCMA
aged 5
-0.262
0.008
0.000
0.772
OCMA
aged 6
-0.259
0.008
0.000
0.753
OCMA
aged 7
-0.284
0.008
0.000
0.738
OCMA
aged 8
-0.303
0.008
0.000
0.748
OCMA
aged 9
-0.291
0.008
0.000
0.741
OCMA
aged 10
-0.300
0.008
0.000
0.741
OCMA
aged 11
-0.299
0.008
0.000
0.741
OCMA
aged 12
-0.289
0.008
0.000
0.749
OCMA
aged 13
-0.276
0.008
0.000
0.759
OCMA
aged 14
-0.293
0.008
0.000
0.746
OCMA
aged 15
-0.299
0.009
0.000
0.741
OCMA
aged 16
-0.297
0.009
0.000
0.743
OCMA
aged 17
-0.311
0.009
0.000
0.632
OCMA
aged 18
-0.402
0.009
0.000
0.669
OCMA
aged 19
-0.322
0.009
0.000
0.717
OCMA
aged 20
-0.320
0.009
0.000
0.726
OCMA
aged 21
-0.349
0.010
0.000
0.705
36
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
OCMA
aged 22
-0.370
0.010
0.000
0.691
OCMA
aged 23
-0.400
0.010
0.000
0.670
OCMA
aged 24
-0.426
0.010
0.000
0.653
OCMA
aged 25
-0.426
0.010
0.000
0.653
OCMA
aged 26
-0.282
0.010
0.000
0.755
OCMA
aged 27
-0.290
0.010
0.000
0.748
OCMA
aged 28
-0.279
0.010
0.000
0.756
OCMA
aged 29
-0.273
0.009
0.000
0.761
OCMA
aged 30
-0.263
0.009
0.000
0.768
OCMA
aged 31
-0.260
0.009
0.000
0.771
OCMA
aged 32
-0.250
0.009
0.000
0.779
OCMA
aged 33
-0.237
0.009
0.000
0.789
OCMA
aged 34
-0.240
0.009
0.000
0.786
OCMA
aged 35
-0.233
0.009
0.000
0.792
OCMA
aged 36
-0.221
0.009
0.000
0.802
OCMA
aged 37
-0.211
0.009
0.000
0.810
OCMA
aged 38
-0.210
0.009
0.000
0.811
OCMA
aged 39
-0.206
0.010
0.000
0.814
OCMA
aged 40
-0.202
0.010
0.000
0.817
OCMA
aged 41
-0.184
0.009
0.000
0.832
OCMA
aged 42
-0.173
0.009
0.000
0.841
OCMA
aged 43
-0.175
0.009
0.000
0.839
OCMA
aged 44
-0.169
0.009
0.000
0.844
OCMA
aged 45
-0.163
0.009
0.000
0.850
OCMA
aged 46
-0.158
0.010
0.000
0.854
OCMA
aged 47
-0.159
0.010
0.000
0.853
OCMA
aged 48
-0.157
0.010
0.000
0.855
OCMA
aged 49
-0.147
0.010
0.000
0.863
OCMA
aged 50
-0.141
0.010
0.000
0.869
OCMA
aged 51
-0.138
0.010
0.000
0.871
OCMA
aged 52
-0.131
0.010
0.000
0.877
OCMA
aged 53
-0.117
0.010
0.000
0.890
OCMA
aged 54
-0.114
0.010
0.000
0.892
OCMA
aged 55
-0.107
0.010
0.000
0.898
OCMA
aged 56
-0.101
0.009
0.000
0.904
OCMA
aged 57
-0.093
0.009
0.000
0.912
37
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
OCMA
aged 58
-0.088
0.009
0.000
0.916
OCMA
aged 59
-0.074
0.009
0.000
0.928
OCMA
aged 60
-0.100
0.009
0.000
0.905
OCMA
aged 61–64
-0.175
0.018
0.000
0.839
OCMA
aged 65
-0.045
0.012
0.000
0.956
OCMA
aged 66
-0.028
0.013
0.026
0.972
OCMA
aged 67
-0.021
0.013
0.115
0.979
OCMA
aged 68
-0.026
0.013
0.050
0.974
OCMA
aged 69
-0.032
0.013
0.018
0.969
OCMA
aged 70
-0.019
0.013
0.157
0.981
OCMA
aged 71
0.001
0.013
0.958
1.001
OCMA
aged 72
0.012
0.014
0.373
1.012
OCMA
aged 73
0.012
0.013
0.377
1.012
OCMA
aged 74
0.002
0.013
0.890
1.002
OCMA
aged 75
-0.014
0.012
0.233
0.986
OCMA
aged 76
-0.015
0.012
0.203
0.985
OCMA
aged 77
-0.008
0.012
0.488
0.992
OCMA
aged 78
-0.024
0.011
0.033
0.977
OCMA
aged 79
0.011
0.012
0.374
1.011
OCMA
aged 80
-0.013
0.011
0.219
0.987
OCMA
aged 81
-0.007
0.010
0.477
0.993
OCMA
aged 82
-0.020
0.010
0.052
0.981
OCMA
aged 83
0.004
0.010
0.709
1.004
OCMA
aged 84
-0.013
0.009
0.150
0.987
OCMA
aged 85
-0.013
0.009
0.136
0.987
OCMA
aged 86
-0.019
0.008
0.014
0.981
OCMA
aged 87
-0.001
0.008
0.143
0.989
OCMA
aged 88
-0.004
0.007
0.542
0.996
OCMA
aged 89
-0.017
0.006
0.007
0.983
OCMA
aged 90
-0.015
0.006
0.008
0.985
OCMA
aged 91
-0.015
0.005
0.004
0.986
OCMA
aged 92
-0.016
0.005
0.001
0.984
OCMA
aged 93
-0.013
0.005
0.005
0.987
OCMA
aged 94
-0.013
0.004
0.001
0.987
OCMA
aged 95
-0.015
0.004
0.000
0.985
OCMA
aged 96
-0.016
0.004
0.000
0.984
38
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
OCMA
aged 97
-0.012
0.003
0.000
0.988
OCMA
aged 98
-0.014
0.003
0.000
0.986
OCMA
aged 99
-0.013
0.003
0.000
0.988
OCMA
aged 100+
-0.022
0.003
0.000
0.978
OCMA
has reached the statutory retirement age
-0.563
0.025
0.000
0.570
MES
is attending pre-school education
establishment or secondary/special/
vocational education institution
1.063
0.009
0.000
2.896
SSIA
is receiving service pension
0.127
0.004
0.000
1.135
SSIA
is receiving old-age pension
2.095
0.014
0.000
8.129
SSIA
is receiving disability pension etc. benefits/
compensations for disabled persons
0.673
0.009
0.000
1.960
SRS
information about employee – employee has
received unpaid leave
-0.053
0.008
0.000
0.948
SRS
information about employee – acquisition of
the status of employee that is employed
during custodial sentencing
0.081
0.016
0.000
1.084
SRS
information about employee – acquisition of
the status of employee or micro enterprise
employee that is to be insured in compliance
with all types of state social insurance
0.480
0.008
0.000
1.617
SRS
information about employee – loss of
employee or micro enterprise employee status
-0.354
0.011
0.000
0.702
SRS
employee (worker) for 1 month
0.036
0.004
0.000
1.037
SRS
employee (worker) for 2 months
0.237
0.005
0.000
1.268
SRS
employee (worker) for 3 months
0.264
0.006
0.000
1.303
SRS
employee (worker) for 4 months
0.267
0.006
0.000
1.306
SRS
employee (worker) for 5 months
0.263
0.006
0.000
1.301
SRS
employee (worker) for 6 months
0.274
0.006
0.000
1.315
SRS
employee (worker) for 7 months
0.274
0.007
0.000
1.315
SRS
employee (worker) for 8 months
0.283
0.007
0.000
1.327
SRS
employee (worker) for 9 months
0.286
0.007
0.000
1.331
SRS
employee (worker) for 10 months
0.367
0.008
0.000
1.443
SRS
employee (worker) for 11 months
0.590
0.011
0.000
1.805
SRS
employee (worker) for 12 months
1.998
0.019
0.000
7.374
SRS
self-employed person
0.261
0.011
0.000
1.329
UL/ RTU
studies at UL or RTU
0.261
0.006
0.000
1.298
39
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
TSI/ RTTEMA/
BIA/ RTA/ LiepU/
RBC/ RTC/ MC/
RMCUL/ DMC/
R1MC/ JVLMA/
LAC/ LAC LCC/
SPC/ RSU RCMC/
SSE Riga
studies at TSI / RTTEMA / BIA / RTA /
LiepU / RBC / RTC / MC / RMCUL / DMC /
R1MC / JVLMA / LAC/ LAC LCC/ SPC/
RSU RCMC or SSE Riga
0.181
0.005
0.000
1.198
RSS
may be found in Rural Support Service data
0.160
0.009
0.000
1.173
ADC
herd owner
0.181
0.009
0.000
1.198
Census/ OCMA
lives in institutional dwelling
0.022
0.003
0.000
1.022
SRS
state social insurance contributions calculated
from EUR 44.02 to 89.46
0.039
0.011
0.000
1.040
SRS
state social insurance contributions calculated
from EUR 89.46 to 123.54
0.067
0.013
0.000
1.069
SRS
state social insurance contributions calculated
from EUR 123.54 to 187.44
0.163
0.016
0.000
1.178
SRS
state social insurance contributions calculated
from EUR 187.44 to 293.94
0.262
0.019
0.000
1.300
SRS
state social insurance contributions calculated
from > EUR 293.94
0.306
0.020
0.000
1.358
SRS
registered income < EUR 2059
0.157
0.004
0.000
1.171
SRS
registered income from EUR 2059 to 3821.22
0.130
0.006
0.000
1.139
SRS
registered income from EUR 3821.22 to
5576.34
0.120
0.006
0.000
1.127
SRS
registered income from EUR 5576.34 to
8660.58
0.093
0.007
0.000
1.097
SRS
registered income from EUR 8660.58 to
13691.64
0.082
0.009
0.000
1.085
SRS
registered income < EUR 13691.64
0.103
0.010
0.000
1.109
MES/ UL/ RTU/
TSI/ RTTEMA/
BIA/ RTA/ LiepU/
RBC/ RTC/ MC/
RMCUL/ DMC/
R1MC/ JVLMA/
LAC/ LAC LCC/
SPC/ RSU RCMC/
SSE Riga
parents – only for persons aged 0–25
0.036
0.006
0.000
1.037
RSS/ADC
parents (RSS or ADC) – only for persons
aged 0–25
0.066
0.004
0.000
1.068
SHS
parents (state compensated health care
service) – only for persons aged 0–25
0.180
0.007
0.000
1.197
40
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
SSIA
parents (SSIA, childbirth allowance, childcare
benefit, state family benefit, maternity
benefit, etc. benefits/ allowances,
compensations for the performance of
guardian’s and foster family’s duties ) – only
for persons aged 0–25
0.214
0.006
0.000
1.239
SSIA
parents (disability allowances/benefits or
pension) – only for persons aged 0–25
0.041
0.004
0.000
1.042
SEA
parents (SEA) – only for persons aged 0–25
0.078
0.005
0.000
1.081
SRS
parents (SRS) – only for persons aged 0–25
0.130
0.010
0.000
1.139
SRS
monthly income of mother or father <EUR
1278 – only for persons aged 0–25
0.239
0.009
0.000
1.270
SRS
monthly income of mother or father >EUR
1278 – only for persons aged 0–25
0.128
0.005
0.000
1.136
CSDD
during the last year, has changed driver’s
licence abroad
-0.041
0.002
0.000
0.960
CSDD
during the last year, has obtained a driver’s
licence
0.043
0.004
0.000
1.044
CSDD4
during the last year, has obtained or changed
boat driver licence
0.019
0.004
0.000
1.020
CSDD
during the last year, has obtained a bicycle
driver licence
0.074
0.011
0.000
1.077
CSDD
during the last year, had to pay penalty for
violations of road traffic rules
0.147
0.004
0.000
1.158
OCMA
born in Latvia and is not Latvian, and is not
citizen of Latvia
0.113
0.005
0.000
1.119
SSIA/SRS
is receiving childbirth allowance/ childcare
benefit, family allowance, maternity benefit
and other benefits for the adoption and care of
a child, or the person went on or returned
from a childcare leave
0.555
0.005
0.000
1.742
SSIA
is receiving pension for the loss of a provider,
other special benefits/ allowances or funeral
benefit
0.171
0.006
0.000
1.186
SHS/SSIA
is receiving compensation for harm to the
participants of the Chernobyl Nuclear Power
Plant accident consequence elimination,
benefits for losing one’s ability to work (if
harmed while at work) or sickness benefits,
and receiving state-funded health care
services
0.266
0.013
0.000
1.304
5 From the year 2021 changes regarding the variable have been implemented since it was not possible directly to detect
the type of changed driver's license (boat or car) in the Road Traffic Safety Directorate (RTSD) data received in 2021.
Information about both obtained boat driver’s licences from the period 2010-2020 and changed licences during the
year 2020 was used to derive the variable accordingly.
41
Administrative
register
Description of variable
(municipalities defined based on the
administrative territorial division in force
until 1 July 2021)
B
S.E
Sig
ExpB
SHS/ SSIA
has received state-funded health care services
at least once during the year, but does not
receive any sickness benefits from SSIA
0.931
0.004
0.000
2.536
SHS/ SSIA
is receiving compensation for harm to the
participants of the Chernobyl Nuclear Power
Plant accident consequence elimination paid
to persons who have lost 10–25 % of their
ability to work, benefits for losing one’s
ability to work (if harmed while at work) or
sickness benefits, but the respective person
has not received any sickness benefits from
SSIA during the year
0.030
0.009
0.001
1.030
SEA/ SSIA
is registered with the SEA as an unemployed
person or a person seeking employment and
receives unemployment benefit
0.314
0.005
0.000
1.368
SEA/ SSIA
is registered with the SEA as an unemployed
person or a person seeking employment, but
not receiving unemployment benefit
0.616
0.004
0.000
1.852
model constant
5.326
0.012
0.000
205.639
42
Table 2 Share of people not included in the population by age group; 2012–2020 (%)
Age
Population on 01.01.2012
Population on 01.01.2013
SSAIS
EHSIS
DA (2011)
SSAIS
EHSIS
SILC (2012)
SSAIS
EHSIS
DA (2012)
SSAIS
EHSIS
SILC (2013)
0
0.83
0.00
0.00
2.38
0.93
0.35
0.00
0.74
0.00
0.00
1
0.82
0.00
1.27
0.00
0.76
0.56
0.00
0.00
0.00
0.00
2
0.74
0.52
0.96
0.00
0.00
1.17
0.00
0.63
0.77
1.56
3
0.77
0.49
1.90
1.22
0.63
0.56
1.44
1.35
0.00
0.74
4
0.75
2.59
1.46
1.12
2.45
1.20
0.95
0.44
1.25
0.61
5
0.72
1.55
0.00
0.65
0.00
0.75
1.46
0.93
2.45
2.05
6
0.42
0.50
0.00
0.71
0.76
0.63
0.00
0.00
0.00
0.70
7
0.50
2.59
0.59
0.65
0.69
0.51
0.53
1.13
0.76
1.36
8
0.18
0.90
1.42
0.00
0.00
0.57
1.18
0.00
0.69
0.00
9
0.34
0.57
0.58
0.68
0.75
0.34
0.94
0.00
0.00
0.00
10
0.10
0.52
0.00
0.00
0.00
0.43
1.16
0.00
1.50
0.84
11
0.12
0.51
0.51
0.00
0.00
0.19
0.49
0.00
0.00
0.00
12
0.16
0.00
0.00
0.00
0.00
0.37
1.01
0.95
0.00
0.00
13
0.10
0.58
0.00
0.00
0.00
0.19
0.56
0.00
0.00
0.00
14
0.18
0.00
0.00
0.49
0.67
0.00
0.17
0.00
0.58
0.00
0.00
15
0.03
0.00
0.00
0.00
0.00
0.00
0.29
0.86
0.49
0.00
0.00
0.00
16
0.30
0.54
0.00
0.00
0.63
0.00
0.31
0.65
0.00
0.52
0.00
0.74
17
0.32
0.00
0.79
0.43
0.56
0.54
0.37
0.54
0.46
0.00
0.00
0.00
18
0.57
0.84
0.72
0.39
1.56
1.18
0.53
0.41
1.74
0.93
0.54
1.25
19
1.50
0.66
1.06
0.38
1.67
2.03
0.67
1.26
0.79
0.87
1.78
1.37
20
2.98
0.29
2.06
2.00
4.47
2.80
1.71
1.32
3.09
3.67
1.52
0.56
21
2.85
0.83
3.69
2.75
3.27
3.55
2.84
0.87
2.40
4.76
4.21
3.68
22
3.73
1.16
3.28
2.45
0.97
1.76
2.86
0.83
3.53
4.13
4.57
4.52
23
3.42
0.56
3.96
3.38
3.43
3.35
3.56
2.33
2.45
6.88
2.35
3.80
24
3.54
0.00
3.97
2.29
4.21
3.01
3.36
1.69
4.22
2.82
3.91
4.52
25
2.79
0.00
4.67
3.92
2.59
1.78
3.23
2.78
3.21
2.75
3.03
3.03
26
3.69
0.00
2.44
2.94
3.43
2.94
2.93
0.00
2.94
2.40
2.96
3.87
27
2.64
0.00
5.47
4.23
3.09
4.61
3.02
0.00
3.43
2.96
2.94
2.50
28
2.74
0.00
2.51
2.88
3.93
2.60
2.07
0.00
4.23
4.64
3.95
5.22
29
2.50
0.00
0.51
2.37
2.89
2.65
2.11
3.28
3.37
1.07
1.31
1.88
30
2.69
3.77
2.42
2.37
1.34
0.78
2.21
1.89
2.96
3.50
2.65
2.07
31
1.52
1.79
3.38
2.69
1.90
1.90
1.88
3.77
2.84
3.11
0.78
0.71
32
2.03
0.00
1.28
2.35
0.00
0.73
1.89
3.57
3.23
2.73
3.82
2.27
33
1.59
2.13
0.98
0.99
4.44
2.96
2.33
0.00
2.36
3.92
0.73
1.61
34
1.20
2.13
0.89
2.20
2.75
1.20
1.37
6.38
1.98
1.62
3.55
4.23
35
1.59
0.00
2.53
0.87
0.00
2.26
1.15
2.13
2.21
2.99
2.41
3.39
43
Age
Population on 01.01.2012
Population on 01.01.2013
SSAIS
EHSIS
DA (2011)
SSAIS
EHSIS
SILC (2012)
SSAIS
EHSIS
DA (2012)
SSAIS
EHSIS
SILC (2013)
36
1.13
3.28
0.00
0.00
1.49
1.03
1.44
3.57
1.73
2.67
2.26
2.89
37
1.45
3.85
1.91
1.15
1.80
2.45
0.76
4.92
0.75
2.12
1.03
1.11
38
1.40
0.00
2.90
1.69
1.81
1.14
1.12
2.56
1.15
2.49
1.84
2.86
39
1.65
0.00
0.38
1.19
1.65
1.05
1.17
0.00
2.95
1.66
1.70
1.18
40
1.31
0.00
1.83
1.14
1.27
0.92
1.29
2.60
3.56
1.64
2.11
2.62
41
1.26
1.10
2.02
0.82
2.40
0.96
1.04
0.00
1.15
2.09
0.92
0.93
42
1.77
0.00
2.71
1.63
0.93
0.49
0.64
1.10
0.82
1.14
2.88
3.00
43
1.61
0.00
1.93
2.16
0.54
0.00
1.17
1.09
2.04
1.27
1.48
1.41
44
1.78
1.02
1.92
0.80
1.38
1.79
1.29
0.00
2.16
1.23
0.55
1.08
45
1.43
0.00
1.10
1.08
0.51
0.56
1.28
1.03
1.20
2.34
2.70
2.24
46
1.09
0.00
1.45
1.63
0.46
0.00
0.95
0.82
0.72
1.39
0.56
0.54
47
1.74
0.96
0.60
0.96
0.00
0.43
0.61
0.00
1.65
1.55
0.00
0.47
48
1.81
1.71
0.95
0.96
1.37
0.49
0.97
0.96
0.64
2.11
0.44
0.83
49
1.67
0.00
0.91
0.66
1.21
1.29
1.13
1.71
1.61
1.00
0.98
0.51
50
1.52
0.00
1.22
1.90
1.23
1.72
1.34
0.00
1.33
1.68
1.29
0.45
51
1.65
0.83
1.16
1.55
2.99
2.46
0.68
0.75
1.93
1.66
2.16
4.35
52
2.01
0.64
1.18
1.16
1.98
1.33
1.26
0.83
2.17
1.92
2.07
1.37
53
1.75
0.00
1.57
0.63
2.58
1.69
1.28
0.64
1.73
0.91
1.34
1.87
54
1.31
0.78
1.70
1.31
1.52
2.07
1.20
0.00
1.27
1.28
2.12
1.28
55
1.98
1.54
2.01
2.08
1.72
1.36
0.94
0.00
2.30
0.63
2.50
2.04
56
0.99
0.00
0.32
0.67
0.87
0.00
1.09
1.54
3.14
2.74
1.37
1.46
57
1.69
0.70
1.44
1.46
0.93
0.53
0.75
0.00
1.01
1.69
0.44
1.34
58
1.79
0.00
1.13
1.15
1.53
1.49
1.04
0.70
1.47
1.06
1.08
1.55
59
1.47
0.65
1.16
0.40
0.49
0.90
0.95
0.00
1.53
1.61
2.02