UNITED NATIONS ECONOMIC COMMISSION FOR EUROPE

CONFERENCE OF EUROPEAN STATISTICIANS

Expert Meeting on Statistical Data Confidentiality

26-28 September 2023, Wiesbaden

Remote Access for Scientific Use Files – a New Pathway for German Official

Statistics Microdata Access

Hanna Brenzel ( Research Data Centre of the Federal Statistical Office)

Katharina Cramer (Research Data Centre of the Statistical Offices of the Federal States)

Volker Güttgemanns (Research Data Centre of the Statistical Offices of the Federal States)

Marcel Mathes (Research Data Centre of the Statistical Offices of the Federal States)

[email protected]

Abstract

The fundamental goal of the Research Data Centre of the Federal Statistical Office and the Research Data Centre of the

Statistical Offices of the Federal States (RDC) is not only to provide access to official statistics microdata, but also to

continuously improve and adapt the access to the changing needs of empirical science. In order to meet the broad range of

needs of the empirically working scientific community, the RDC have offered different access paths since their founding, through which differently anonymised data products are made available. Now, the RDC come up with a new remote access

prototype system including a new data product. All access paths differ both in terms of the anonymity degree of the

provided microdata as well as in the access way of data provision. At first, existing and firmly established data access

paths are outlined and their contractual and legal conditions explained. Subsequently, the newly installed remote access

prototype and its features and requirements are presented. Provided that the ongoing evaluation phase turns out positive,

this data access option will define one more way of data access operated regularly in its full version from 2024 onwards.

The analysis potential of the data provided therein will classify between the scientific-use files transmitted to the scientific

institutions and the data provided for on-site analysis at the RDC safe centres. This paper highlights various challenges,

such as data protection requirements and legal framework conditions, which must be considered.

2

1 Introduction

With the establishment of the Research Data Centre (RDC) of the Federal Statistical Office in the fall of 2001

and with the RDC of the statistical offices of the Länder in April 2002, an important cornerstone and a central intersection was created between the scientific community and official statistics as data and information service

provider.

Together, the RDCs offer the empirically working scientific community a coordinated range of data and services for the scientific use of high-quality microdata from official statistics.

Over time, however, expectations of the RDCs have evolved fundamentally, and stakeholders in politics and

scientific communities have been pushing for substantial improvements in data access and data usage capabilities for some time.

Remote access represents an up to date and modern way of accessing data and is accordingly demanded by data

users. The statistical offices of other European countries (e.g., the Netherlands, France or Finland) can be

mentioned as reference benchmarks. They have created the legal and technical prerequisites to make their data available to researchers via remote access some time ago. Last but not least, a remote access system is currently

being set up at European level by Eurostat.

On one hand, the establishment of a remote access system - with the investment in a connectable infrastructure - will advance the continuous development of the RDC. By catering towards the needs of the scientific

community, the status of the RDC as a modern data provider will be consolidated. On the other hand, the

currently complex and inefficient system of data access can be streamlined to a uniform and manageable system without limiting the flexibility of the users.

2 Status Quo

The RDC of the Federal Statistical Office, together with the RDC of the statistical offices of the Länder, offer

access to more than 3,000 different data products for over 90 statistics for scientific use via different ways of

access. They differ both in terms of the anonymity of the accessible data and in the type of data provision. Generally, the existing ways of data access can be divided into two categories, as figure 1 illustrates. In the case

of the so-called "on-site access", the data remains in the secure areas of the statistical offices of the Federation

and the Federal States. Since the RDCs can closely control the access to the data and provide output only after

confidentiality check, the data are only weakly anonymized. With the "off-site access," on the other hand, users can work with the individual data at their own institutes. Since the output are not checked by the data centers,

the individual data has to be more anonymized.

The category “off-site” includes the so-called Public Use Files (PUF), Campus Files (CF) and Scientific Use Files (SUF). “On-site” includes PC workplaces at the RDC, so called “safe centers” and remote execution (see

the homepage of the RDC, https://www.forschungsdatenzentrum.de/en/access).

Safe centers exist in all locations of both RDC. These can be used by researchers to analyse microdata inside

the safe premises of the statistical offices. As the individual data are already protected by the regulation of data

access and the equipment of the PC workstation, formally anonymous microdata can be provided at the safe

centers. Thus, a nationwide infrastructure in Germany is available for these data. The safe centers are equipped with common statistical programs (Stata, R as well as partly SPSS and SAS) and

are completely isolated from the outside. A separate PC workstation with internet connection is available for e-

mail communication and internet searches. In contrast to the safe centers, remote execution does not provide direct access to the microdata. Instead, data

structure files are made available that resemble the original material with regard to structure and variable

values, but do not permit any analyses in terms of content and do not hold any risk of exposing confidential

information. Using these data sets, program codes can be prepared by the users using the statistical programs SPSS, SAS, Stata or R. These program codes are applied by staff of the statistical offices to analyse the original

data. The data users receive the results of those analyses after the relevant confidentiality checks.

SUFs are standardized datasets created by the RDC for popular statistics. SUFs offer lower potential for

analyses than on-site ways of access, but are designed to be suitable for a large proportion of scientific research

projects. Due to the de facto anonymization of microdata, they may be used outside the protected premises of official statistics according to Sect. 16 para. 6 nos. 1 BStatG. Due to legal restrictions, SUF may only be used

3

by researchers who are employed by a research institution that is registered and located in Germany. The use of

SUF may only take place in Germany. Until recently, the SUF were sent by DVD to the respective scientific institution with which user contract was concluded. Since June 2023, recent modernization measures now allow

the SUF to be accessed directly via a download portal to the institution authorized to use the data.

In particular, on-site ways of access entail additional work for both data users and RDC staff. At the same time,

the share of data uses via these access paths steadily increases over time compared to off-site uses. The development of a remote access system therefore pursues the goal of ensuring the technical connectivity to a

modern and demand-oriented data provision for the scientific community. With this technology, the increased

expectations of the research community for an up-to-date and modern data provision can be fulfilled in the long term. In addition, the remote access system holds potential for future innovation by reducing or substituting

existing labor-intensive ways of access (reduction of on-site support, reduction of coordination of appointments

with users, reduction of coordination and support of remote execution, etc.). Consequently, the scarce resources of the RDC could be invested more efficiently, for example in supporting additional data usage or further

developing the data and service offers. At the same time, there is increased potential regarding data parsimony,

as it is expected that this system will reduce the number of intermediate results per project that require

confidentiality checks. Furthermore, the RDC aim to sustainably strengthen their leading role in the group of German RDC.

Figure 1: Ways of data access at the research data centres (RDC) of the statistical offices of the Federation and the Federal States

3 The Remote Access System

3.1 The technical structure

IT and data security play a crucial role in setting up the remote access system. The aim is to ensure that the

remote access system is implemented in compliance with the law while maintaining the required IT security

standards.

A virtual desktop infrastructure based on CITRIX was chosen as the IT-architecture. The system components

set up are located in the so-called IDMZ (Internet Demilitarized Zone), in which procedures are operated that

are to be accessible from the Internet. In the IDMZ, a distinction is made between three areas: Access Area

(Pex), Application Area (Pin1) and Data Area (Pin2). These three areas are separated from each other by

firewalls, which only allow approved communication between the neighboring areas within the application. A

so-called transport encryption secures the communication path between the server and the client.

Two-factor authentication and IP whitelisting are implemented as additional IT security measures for the Citrix

solution. Two-factor authentication means that, in addition to the user-specific work accounts protected by a

personal password, a uniquely generated token must be used for each log-in. IP whitelisting allows only

specific IP addresses to gain access to the remote access system. Prior to each authorized use, the IP address of

the respective facility is allowed or added to the whitelist. This ensures that unauthorized IP addresses do not

initially gain access to the system. This implements geoblocking as a technical measure as well as

strengthening protection against possible (automated) attack attempts.

In addition, app protection is used to, among other things, prevent the user from taking screenshots of the data.

Remote system access is controlled on a per user basis by an access management system, only authorized users

are granted access. Within the system, authorizations are limited to the extent required for data analysis. The

creation of user-specific working accounts, which are managed centrally and secured by the user and access

management, ensures that access is only possible to requested data. Each account is linked to a data folder in

which user-specific official microdata are stored by RDC staff.

In addition to the technical measures, a number of technical and contractual-organizational measures are

introduced to increase data protection. Before the data can be accessed, a user contract has to be concluded

between the scientific institution and the responsible statistical office. It is contractually stipulated that up-to-

4

date software, operating system and virus protection are used on the client side when accessing the virtual

desktop infrastructure. As well as, re-identification of individual cases is illicit. The RDC are legally bound to

check all statistical results for statistical confidentiality that were created within the context of scientific

projects based on provided microdata. This serves the protection of data according to section 16 (6) of the

Federal Statistics Law (BStatG). Should individual cases be part of the output then they have to be blocked

consistently across all results of a project. Data users who plan to re-identify individual cases are liable to

prosecution and are expelled from further data uses.

In order to ensure that the system is tied to a specific location, its use is contractually established and sanctions

are imposed in the event of violations. In addition, it is contractually stipulated that scientific institutions can be

excluded from using the remote access system or from the possibility of carrying out further research projects

via the RDC in the event of serious violations of the terms of use. In the event of a striking breach of contract,

the scientific institutions can also be sanctioned with a penalty payment of up to EUR 20,000.

3.2 Data material in the remote access system

Remote access to formally anonymized data is not feasible within the current legal framework. One possible

way of implementation is to offer remote access for de facto anonymized data with slight modifications, as this

would not require amendment of the law. In this case, the degree of data modification is of utmost relevance: If

the level of anonymization is too high, the data offered will not meet the needs of the scientific community; if

the level of data anonymization is too low, confidentiality can no longer be maintained. The degree of de facto

anonymization therefore largely determines the benefits and coverage of the demand of the scientific

community. In addition, the expected effects on the capacity of the RDC heavily depend on covering as many

of the science community's projects as possible via the remote access system and, in particular, on reducing the

costly uses of remote execution. However, this goal can only be achieved if significantly more data can be

provided via remote access than via the current dissemination path via off-site SUF.

Microdata are described as “de facto anonymous” if it is not possible to completely rule out de-anonymization

but assigning the information to the respective statistical unit “requires unreasonable effort in terms of time,

cost and manpower” (Section 16 (6) of the Federal Statistics Act). According to the Federal Statistics Act,

however, de facto anonymous data may only be used by scientific institutions and only to carry out scientific

projects.

When creating de facto anonymity, the aim is to virtually eliminate the probability of correctly assigning data to

respondents, while preserving the statistical information content as much as possible. Different anonymization

methods can be used for this purpose. Common methods are information reduction (e.g. aggregation, class

formation, censoring) and information modification (e.g. swapping). In order to determine de facto anonymity,

the effort and benefit of deanonymization must be evaluated.

Factual anonymity thus does not completely exclude the possibility of re-identification, but puts its risk in a

cost/benefit ratio. Costs for data users primarily include the consequences for actions in violation of the

contract. Re-identification is strictly prohibited and punishable by fine or imprisonment (Section 203 StGB). In

addition, consequences such as loss of reputation, loss of access to data of official statistics, etc., which threaten

in the event of de-anonymization of the data, must also be considered by scientific users. This is because the

users are obligated to maintain the anonymity of the data both by the formal obligation and the user agreement.

Factual anonymity therefore does not result solely from the remaining information content of the data, but is

composed of a triad: 1) modification of the data material, 2) technical/organizational measures, and 3)

contractual measures. Therefore, it also depends on the access condition, if a microdata set can be described as

Figure 2: Technical infrastructure of the remote access system

5

de facto anonymous. Of crucial importance here is what additional knowledge is available and where the data

access takes place. Depending on whether the microdata is used outside or inside the statistical offices, de facto

anonymity can be achieved with more (off-site SUF) or less (on-site SUF) severe losses of information.

The de facto anonymity of microdata from official statistics is thus not a fixed quantity, but can be mapped

along a continuum. In principle, it can be stated: The higher the technical and contractual measures, the fewer

anonymization measures need to be taken and the higher the analysis potential of the data.

No technical measures are used for the previous off-site SUFs. Factual anonymity must therefore only be

achieved from the two remaining measures: in addition to the contractual commitment and the commitment of

the users, de facto anonymity is achieved by strongly anonymizing the data material itself. For this purpose, a

statistics-specific anonymization concept is developed for each data material.

With the new remote SUF or on-site SUF, de facto anonymity can be achieved by significantly less

modification of the data. This is justified by the high level of technical measures and the associated possibility

to control the data access. In contrast to off-site SUF, the data is not passed on. It is solely possible to view the

data via a virtual desktop (VDI environment). A so-called "transport encryption" secures the communication

path between the server (sender) as well as the client (receiver). An exchange between the technical

infrastructure of the data users and the data on the server of the official statistics or a download of the official

data is thus technically impossible. Thus, unauthorized data linkage is impossible and the RDC has a high level

of use control via log files. With regard to the risk of de-anonymization, data access via remote access therefore

reduces many risks compared to the previous off-site SUFs.

3.3 The use of Remote Access

The remote access system, which is currently under construction, will be set up as a classic remote desktop

version. As in the past, scientific institutions that are entitled to use the system in accordance with Section 16

BStatG have to apply for data access. If the application is approved, the researchers are then able to access the

secure area within their scientific institution by using their own hardware. Within the secure area common

statistical software such as RStudio and Stata is available. The major advantage compared to remote execution

is that researchers can see the microdata and do not have to "blindly" program their syntaxes as before (see

Figure 3). By working directly with and being able to view the data, it should be possible to significantly

reduce the number of intermediate results previously generated via remote execution, thus minimizing a very

labor-intensive process step in the RDC. The goal should be that only final outputs are checked for

confidentiality by the RDC staff and will be released. This also supports the principle of data parsimony.

Figure 3: Remote Access at the RDC

Work on setting up such a system began in November 2021. The system is currently in the evaluation phase.

On one hand, the technical implementation of the system is being tested and its resilience checked using penetration tests. On the other hand, the user-friendliness and the attractiveness of the data material provided is

to be examined thoroughly. In a first step, only absolutely anonymous data material was made available via the

system for a selected group of people. In a second step, off-site SUFs will then be made available to power users who have already completed a valid user application with the RDC. The third step will then be to test the

redesigned on-site/remote SUF material. Since the system requires a redesign of all statistics-specific

anonymization concepts, a gradual integration of the existing data products in the RDC is planned. The start will be made with the most requested data product, the microcensus. In order to be able to evaluate the

operating grade of the system appropriately, DRG statistics will be offered as one of the first data products in

the remote access system in addition to the microcensus. If the evaluation of the system is positive, other data

products that are of high demand will follow.

6

4 BIBLIOGRAPHY

Brenzel, Hanna / Zwick, Markus. An information infrastructure has emerged in Germany – the Research Data

Centre of the Federal Statistical Office. German version published in WISTA | 6 | 2022, p. 54 et seq.

Homepage of the Research Data Centre of the Federal Statistical Office and the Federal States

https://www.forschungsdatenzentrum.de/en

From: Klinger, Prof. Dr. Remo Sent: Monday, July 10, 2023 11:23 AM To: ECE-Aarhus-Compliance <[email protected]> Cc: 'Jürgen Resch extern ; 'Claudia Wesemann | DUH 'Sascha Müller-Kraenner | DUH ; 'Barbara Metz | DUH'

; 'Matthias Walter | DUH' ; 'Dorothee Saar | DUH' Subject: New Communication Dear Sir or Madam, I am enclosing a new complaint that we are filing on behalf of our client, Deutsche Umwelthilfe e.V., directed against the Federal Republic of Germany. We will send the signed original document by mail. Yours sincerely, Professor Dr. Remo Klinger _____________________________ GEULEN & KLINGER Rechtsanwälte Professor Dr. Remo Klinger

Tel.

www.geulenklinger.com

GEULEN & KLINGER

Rechtsanwälte

Commerzbank AG in Berlin

IBAN BIC USt-ID Nr.

*Partner der Sozietät

Communication to the Aarhus Convention‘s

Compliance Committee

I. Information on correspondent submitting the communication

1. This communication is submitted by:

Full name of submitting organisation or person(s):

Permanent address:

Telephone:

Fax:

Email:

Deutsche Umwelthilfe e.V.

Contact person authorised to represent the organisation in connection with the commu-

nication:

Name:

Title:

Jürgen Resch

Executive Director

Represented by:

Name:

Title:

Email:

Telephone:

Fax:

Prof. Dr. Remo Klinger

Lawyer

July 10, 2023

Secretary to the Aarhus Convention Compliance Committee

United Nations Economic Commission for Europe

Environment Division

Palais des Nations

CH-1211 Geneva 10

Switzerland

Dr. Reiner Geulen*

Prof. Dr. Remo Klinger*

Dr. Caroline Douhaire LL.M.

Karoline Borwieck

David Krebs

Lukas Rhiel

Telefon

Telefax

E-Mail

www.geulenklinger.com

2

II. Party concerned

2. Federal Republic of Germany.

III. Facts of the communication

1. Reasons that lead to this communication

3. The communication is targeted at the procedure of association participation in the

context of the draft amendment for the Climate Protection Act (Klimaschutzgesetz –

KSG)1 which did not allow for effective public participation and therefore violated ar-

ticle 8 of the Aarhus Convention (AC).

4. The KSG is the main framework legislation mapping out Germany’s route to climate

neutrality. The current German government is planning to amend this Act. In its cur-

rent form, as one of its main operational instruments, the KSG sets yearly emissions

targets for different sectors (so-called sectoral targets).2 If one of the sectors fails to

meet its yearly emissions target, it is obliged to develop an emergency programme

(Sofortprogramm) within three months.3 This is the main tool of the KSG to ensure

that Germany reaches its total greenhouse gas reduction targets as constitutionally

required.4 These emergency programmes are planned to be abolished by the current

German government, which many environmental associations regard as a clear de-

terioration of the German climate protection law.

5. There is no urgent need for this amendment legislation with a functional and consti-

tutional KSG currently in force. The political will to change the law was already in the

coalition agreement concluded almost two years ago.5 There are no urgent reasons

to do so now. Furthermore, the parliamentary procedure for amending the law will

not begin until September and according to current information, the amendment will

not come into force until the beginning of 2024.

6. The draft KSG amendment was sent out on Thursday, June 15, 2023 at 5:28 p.m. by

the responsible Federal Ministry for Economic Affairs and Climate Protection (Bun-

desministerium für Wirtschaft und Klimaschutz – BMWK) to initiate the participation

of associations pursuant to § 47 of the Joint Rules of Procedure of the Federal Min-

istries (Gemeinsame Geschäftsordnung der Bundesministerien – GGO). The com-

municant, Deutsche Umwelthilfe e.V. (DUH), was involved this way. The BMWK set

the response deadline for the associations as Monday, June 19, 2023 at 10:00 a.m.

In Germany, the Working Hours Act (Arbeitszeitgesetz – ArbZG) stipulates a five-day

1 Current KSG: https://www.gesetze-im-internet.de/ksg/. 2 Annex 1 to the KSG. 3 § 8 (1) KSG. 4 Article 20a of the Constitution (Grundgesetz – GG); see: Decision by the German Constitutional Court (Bundesverfassungsgericht – BVerfG) of March 24 2021 – 1 BvR 2656/18. 5 Coalition Agreement “Mehr Fortschritt wagen”, 2021, p. 55.

3

week with a maximum of 8 hours of working time per working day.6 Consequently,

the deadline set by the BMWK was one of one working day (Friday) and a maximum

of two hours (Monday 8:00 a.m. to 10:00 a.m.).

7. As the sole reason for this short comment period, the BMWK stated that this was

“politically imposed.”

8. Factually, the government had put the draft KSG amendment on the agenda for the

cabinet meeting on Wednesday, June 21, 2023.7 As outlined above, there was, how-

ever, no specific reason for the government to do this. Cabinet meetings happen

weekly. Parliament will only put the KSG amendment on their agenda after its sum-

mer break in September.8

9. In terms of content, the draft KSG amendment changes the operational tools of the

KSG comprehensively. The sectoral emergency programmes are drafted to be elim-

inated. Policy adjustments due to missed emissions targets are to be required after

two years and for all sectors collectively – as opposed to after three months and for

each specific sector.

10. The draft KSG amendment is complex and consists of numerous clauses that are

open to interpretation.9

11. Due to this extremely short comment period, DUH was unable to provide compre-

hensive and legal feedback. The feedback DUH was able to provide was one of for-

mulaic and purely general political nature in very short form. An actual legal state-

ment, which also addresses errors, or at least misleading formulations in the draft

law, was not feasible in the short time available. Effective participation was thus im-

possible. Therefore, the German government violated its obligation to ensure effec-

tive public participation by fixing sufficient time-frames under article 8 (a) of the Aar-

hus Convention.

12. The legislative procedure for the Climate Protection Act is just one example of the

significantly too short comment periods set by the German government. The same

applies to a number of other legislative projects. One example is the recent amend-

ment to the environmental provisions of the Road Traffic Act. The documents for the

hearing were sent on June 15, 2023 at 1:00 p.m., and the deadline for comments

was June 16, 2023 at 3:00 p.m., which was only a few hours. Again, this bill did not

contain any regulations with special urgency.

13. The short deadlines vary depending on the ministry responsible. Draft laws from the

ministry originally responsible for the environment sometimes have longer deadlines.

6 §§ 3, 9 ArbZG. 7 https://www.bundesregierung.de/breg-de/bundesregierung/bundeskanzleramt/kabinettssitzun- gen/bundeskabinett-ergebnisse-2197550. 8 https://www.bundestag.de/tagesordnung. 9 Compare the press statement of Minister Habeck: https://www.bmwk.de/Redaktion/DE/Vid- eos/2023/06/230614-pressestatement/video.html.

4

However, the Federal Ministry for the Economy and Climate is responsible for the

climate protection law relevant here. For the Road Traffic Act mentioned as an ex-

ample, the Federal Ministry responsible for transport.

2. Legal and procedural background in Germany

a. Legislative process

14. In Germany, laws are adopted by the parliament. They can be introduced either by

the “midst of parliament,” the Federal Council (Bundesrat) or the government.10 In

the case of governmental introduction, the government firstly submits the draft to the

Bundesrat and after receiving its comments it is submitted to parliament. Ordinarily,

parliament refers the draft to the respective Committee and adopts it after three read-

ings.11

15. The government initiates this legislative process after it has agreed upon its draft

legislation amongst its cabinet. Only after this cabinet meeting, the draft legislation

will be sent to the Bundesrat and parliament.

16. By not acting to put the draft KSG amendment onto the parliamentary agenda before

its summer break, the government underlined that the KSG amendment is not cur-

rently considered a priority and, in particular, that its adoption is not time-critical.

b. Procedural practice

17. With regards to public participation, the Joint Rule of Procedure of the Federal Min-

istries (Gemeinsame Geschäftsordnung der Bundesministerien – GGO) lays out the

ground rules. The GGO stipulates in § 47:

„§ 47 Beteiligung von Ländern, kommunalen Spitzenverbänden, Fachkreisen und

Verbänden

(1) Der Entwurf einer Gesetzesvorlage ist Ländern, kommunalen Spitzenverbänden und

den Vertretungen der Länder beim Bund möglichst frühzeitig zuzuleiten, wenn ihre

Belange berührt sind. Ist in wesentlichen Punkten mit der abweichenden Meinung

eines beteiligten Bundesministeriums zu rechnen, hat die Zuleitung nur im Einverneh-

men mit diesem zu erfolgen. Soll das Vorhaben vertraulich behandelt werden, ist dies

zu vermerken.

(2) Das Bundeskanzleramt ist über die Beteiligung zu unterrichten. Bei Gesetzentwürfen

von besonderer politischer Bedeutung muss seine Zustimmung eingeholt werden.

(3) Für eine rechtzeitige Beteiligung von Zentral- und Gesamtverbänden sowie von Fach-

kreisen, die auf Bundesebene bestehen, gelten die Absätze 1 und 2 entsprechend.

Zeitpunkt, Umfang und Auswahl bleiben, soweit keine Sondervorschriften bestehen,

dem Ermessen des federführenden Bundesministeriums überlassen. Die Beteiligung

10 Article 76 para 1 GG. 11 §§ 75 ff. of the Rules of Procedure of the German Bundestag (Geschäftsordnung des Deutschen Bundestag – GOBT).

5

nach Absatz 1 soll der Beteiligung nach diesem Absatz und der Unterrichtung nach §

48 Absatz 1 vorangehen.

(4) Bei der Beteiligung nach den Absätzen 1 und 3 ist ausdrücklich darauf hinzuweisen,

dass es sich um einen Gesetzentwurf handelt, der von der Bundesregierung noch

nicht beschlossen worden ist. Dem Gesetzentwurf können die Begründung und das

Vorblatt beigefügt werden.

(5) Wird zu einer Gesetzesvorlage eine mündliche Anhörung durchgeführt, sind hierzu

die kommunalen Spitzenverbände einzuladen, wenn ihre Belange berührt sind. Die-

sen soll bei der Anhörung vor den Zentral- und Gesamtverbänden sowie den Fach-

kreisen das Wort gewährt werden.“

English translation (by Deepl.com):

“§ 47 Consultation of Länder, municipal umbrella organisations, expert groups

and associations

(1) Draft legislation shall be submitted to the Länder, the central associations of

the local authorities and the Länder representations to the Federation as

early as possible if their interests are affected. If the opinion of a participating

Federal Ministry is likely to differ on essential points, the legislation shall be

forwarded only in agreement with that Ministry. If the project is to be treated

confidentially, this shall be noted.

(2) The Federal Chancellery shall be informed of the participation. In the case of

draft legislation of particular political importance, its consent must be ob-

tained.

(3) Paragraphs 1 and 2 shall apply correspondingly to the timely participation of

central and general associations as well as of expert groups existing at fed-

eral level. The timing, scope and selection shall be left to the discretion of the

lead Federal Ministry, unless special provisions exist. Participation under par-

agraph 1 shall precede participation under this paragraph and information

under section 48(1).

(4) When participating in accordance with subsections (1) and (3), explicit refer-

ence shall be made to the fact that the bill in question has not yet been

passed by the Federal Government. The bill may be accompanied by the

explanatory memorandum and the preliminary sheet.

(5) If an oral hearing is held on a bill, the municipal umbrella organisations shall

be invited to attend if their interests are affected. They shall be granted the

right to speak at the hearing before the central and general associations as

well as the expert groups.”

18. Hence, § 47 (3) GGO determines that draft legislation is to be submitted to relevant

central and general associations as well as expert groups as early as possible. Ac-

cordingly, this public participation takes place after the participation of the Länder and

their representations and associations but before the cabinet meeting.12

12 See § 51 GGO.

6

19. In response to a question about how the Federal Ministry for the Environment, Nature

Conservation, Nuclear Safety and Consumer Protection (Bundesministerium für Um-

welt, Naturschutz, nukleare Sicherheit und Verbraucherschutz – BMUV) practices

public participation, the BMUV stated in 2022 that, as a rule, a comment period of 4

weeks is provided for public participation regarding draft legislation on environmental

laws. Considering the scope and complexity of the respective draft legislation, this

may be shortened to three or two weeks according to the BMUV. An additional cur-

tailment of the participation period is possible as per the BMUV if there are excep-

tional reasons for faster processing in individual cases. This exemption clause cor-

respondents to article 9 (4) of EU Regulation (EC) No 1367/2006 which implements

the Aarhus Convention.

20. Consequently, although there is no strict timely requirement, both the EU as well as

the German institutions take the stance that effective public participation can only be

ensured if the comment period covers several weeks. Furthermore, in cases of re-

duced comment periods, both institutions maintain that this should only be possible

in individual cases and if exceptional reasons exist.

3. Public participation for the draft KSG amendment

21. As already presented above, the comment period given to associations with regards

to the draft KSG amendment was of one working day plus two hours.

22. The sole reason given by the responsible BMWK was that this short time period was

“politically imposed.” What this means in concrete terms was not explained.

23. The only possible justification that the communicant was able to identify was that the

government had placed this draft on the agenda of the next cabinet meeting. How-

ever, this does not imply that it was politically necessary to accelerate the procedure.

The actual legislative procedure does not begin until the draft legislation is introduced

into the Bundestag. There, however, this draft will not be on the agenda until after

the summer break in September. The period between sending the draft KSG amend-

ment to the associations and the next possible Bundestag session, where this draft

could be on the agenda, is therefore two full months with weekly cabinet meetings.

Furthermore, as detailed above, the government has planned to amend the KSG

since it formed its coalition almost two years ago, there is no particular reason why it

is adopting – and rushing associations in the process – now.

24. Moreover, there is currently a functioning KSG in place which further undermines any

alleged urgency.

4. Infringement of AC article 8

25. The Aarhus Convention maintains that public participation in decision-making is not

a matter of good practice but one of urgency with ever more complex subject matters.

Decision-making requires accurate, comprehensive and up-to-date information

which the public can be a major source of. Effective public participation therefore not

7

only provides for the people to enjoy their rights but also improves the ability of au-

thorities to carry out their responsibilities. For the public to become an effective and

useful part of the decision-making of public authorities, participation requires an

open, regular, and transparent process.13

26. The AC emphasizes that “the public input should be capable of having a tangible

influence on the actual content of the decision. When such influence can be seen in

the final decision, it is evident that the public authority has taken due account of public

input.”14

27. Hence for public participation to be effective, it should be conducted in a way that

allows for the public authorities to be influenced by the public input.

28. Article 8 of the AC standardises the procedure with regards to the preparation of laws

and rules with potential environmental impact.

29. The public participation procedure regarding the draft KSG amendment directly in-

fringes AC article 8 (a) which maintains that time-frames for public participation

should be “sufficient for effective participation.”

a. Applicability of article 8 to the preparation of legislation

30. AC article 8 refers to public participation “during the preparation by public authorities

of executive regulations and/or other generally applicable legally binding rules that

may have a significant effect on the environment.”

31. In 2021, the civic association VIA IURIS brought a communication against Slovakia

before the Aarhus Convention Complaint Committee (from here on: Committee).

There, the Committee found that “preparation of legislation by executive bodies to be

adopted by national parliaments” are included in the provision of Art. 8 AC.15 Accord-

ing to the Committee, “nothing in the title or text of article 8 of the Convention [sug-

gests] that it does not include the preparation of legislation by executive bodies to be

adopted by national parliaments. On the contrary, although the terms “legislation”

and “laws” do not appear in the provision, the wording of article 8 and the ordinary

meaning given to its terms nevertheless support the inclusion of legislation and other

normative instruments of a similar character.”16 The Committee understands the text

of article 8 as a “generic expression intended to cover different kinds of generally

applicable legally binding normative instruments, which may be referred to in different

ways in different jurisdictions.”17 Furthermore, since AC article 8 specifically adds the

term of “generally applicable legally binding normative instruments” and does not limit

13 Implementation Guide, p. 85. 14 Implementation Guide, p. 86. 15 ECE/MP.PP/C.1/2021/19, para. 95 ff. 16 ECE/MP.PP/C.1/2021/19, para. 95. 17 ECE/MP.PP/C.1/2021/19, para. 96.

8

its scope to “executive regulations”, it is, therefore, also applicable to regulations

other than those by the executive branch.18

32. Already in its findings in another case, the Committee had held that article 8 relates

“to any normative acts.”19

33. While AC article 8 is thus applicable to legislation, the Convention is more restrictive

with regards to the term “public authority”. According to AC article 2 (2) this does not

include bodies or institutions acting in a legislative capacity.

34. However, the Committee understands this limitation as strict and precise, meaning

that “it only covers activities by the body or institution with the capacity and power to

adopt the legislation.”20 This strict understanding is also in line with applying article 2

(2) as uniformly as possible to the Parties as their legislative processes likely differ.21

Otherwise, Parties could attempt to exclude the application of the AC by expanding

their preparatory processes with several public authorities involved and with no trans-

parency or public participation.22 Such “comprehensive preparatory procedures are

perfectly in line with the Convention,” however, “they must not be used to exclude

opportunities for members of the public […] to participate.” 23

35. Therefore, public authorities, including governments, “do not act in a legislative ca-

pacity when engaged in preparing laws until the draft or proposal is submitted to the

body or institution that adopts the legislation.” 24

36. The BMWK, a German governmental body, sent out the draft KSG amendment be-

fore it was submitted to the Bundestag (as detailed above). Hence, it acted within its

executive capacity concerning a “generally applicable legally binding rule.” The public

participation in question is, therefore, within the scope of article 8.

b. Violation of article 8 (a)

37. AC article 8 (a) states that Parties should take steps to establish “time-frames suffi-

cient for effective participation.” The Implementation Guide explains that the ele-

ments regulated in article 8 set forth a “basic procedural framework.”25

38. While article 8 (a) does not set specific time-frames, “the Convention states that the

authorities should plan for public participation by fixing their own schedule that is

“sufficient” for effective participation.”26 This is illustrated in the Implementation Guide

18 ECE/MP.PP/C.1/2021/19, para. 97. 19 ECE/MP.PP/C.1/2011/6/Add.1, para. 61. 20 ECE/MP.PP/C.1/2021/19, para. 99. 21 ECE/MP.PP/C.1/2021/19, para. 99. 22 ECE/MP.PP/C.1/2021/19, para. 99. 23 ECE/MP.PP/C.1/2021/19, para. 99. 24 ECE/MP.PP/C.1/2021/19, para. 101. 25 Implementation Guide, p. 185. 26 Implementation Guide, p. 121.

9

with a legislation that it finds to be a typical example of this requirement: The Hun-

garian Act XI of 1987 on Legislation provides that when drafting comment deadlines,

the following four factors are to be taken into account: (1) the person giving the opin-

ion should have the opportunity to form a well-based opinion; (2) the opinion must be

able to be taken into consideration in the drafting; (3) the size of the draft; and (4) the

type of organization giving the opinion.27 This reiterates the need for time-frames that

allow for effective participation and underlines that effective participation is made im-

possible if the deadline does not allow for comprehensive examination.

39. Therefore, it can be deduced that the public authorities have a duty of care to guar-

antee effective participation.28

40. Truly effective participation is only possible in a deliberative process.29 Deliberation

fosters a collaborative decision-making process which allows “the reconciliation of

strong democracy and demanding environmentalism.”30 This is in line with the pur-

pose of the Convention to guarantee a comprehensive decision-making process re-

garding environmental matters.

41. Finally, the last sentence of article 8 requires the result of the public participation to

be considered as far as possible. In this regard, the Implementation Guide clarifies

that “this provision establishes a relatively high burden of proof for public authorities

to demonstrate that they have taken into account public comments in processes un-

der article 8.”31

42. While it is difficult to gauge the effectiveness of public participation, it is easier to

assess ineffectiveness. It is undoubtedly ineffective if the outcome of the participation

process appears to be a “foregone conclusion.” Therefore, this “closed mind” is “in

principle unlawful under the Aarhus Convention.”32

43. It is thus on the public authorities in question to prove their open mind and demon-

strate their consideration of the results of the public participation procedure.

44. In the case of this communication, the time given for public participation was one day

and two hours. A comprehensive examination of the draft by the public plus the writ-

ing of an extensive statement within one day and two hours is impossible. A state-

ment drafted within this period can be general and broad at best. In addition, the time

the government took to consider these statements was two days. A comprehensive

examination of the submitted statements and an incorporation of these proposals

27 Implementation Guide 121. 28 ECE/MP.PP/C.1/2021/19, para. 103; Eppiney, in: Epiney/Diezig/Pirker/Reitemeyer, Aarhus Konvention, 1. Auflage 2018, Art. 18 AK, marginal no. 4 f. 29 Barritt, The Foundations of the Aarhus Convention: Environmental Democracy, Rights and Stewardship, 2020, p. 153. 30 Barritt, The Foundations of the Aarhus Convention: Environmental Democracy, Rights and Stewardship, 2020, p. 67 f. 31 Implementation Guide, p. 185. 32 Lee, “The Aarhus Convention 1998 and the Environment Act 2021”, in: The Modern Law Review 86 (3), May 2023, p. 756 (782).

10

into the text within two days is also impossible. Hence, a time-period as given in the

process in question indicates a ”closed mind” and does not allow for effective partic-

ipation. This is especially true in the case of the amendment of a critical operational

tool within the main framework legislation for climate protection.

IV. Nature of alleged non-compliance

45. This communication concerns the specific case of non-compliance of the Federal

Government in its involvement of associations regarding the draft of the Climate Pro-

tection Act Amendment. The German government did not make sufficient efforts to

allow the public to effectively participate in the process. It thereby violated article 8

(a) of the Aarhus Convention.

V. Relevant provisions of AC

46. Article 8 (a) of the Aarhus Convention.

VI. Use of domestic remedies

47. Public participation within the German legislative process is regulated by § 47 (3)

GGO (see above). The GGO is an administrative regulation with regularly no external

impact.33 This means, that legal action is not available. To legally challenge govern-

mental action that is based on administrative regulation, the regulation in questions

needs to have developed external impact. This is the case when there is consistent

practice because of the so-called “self-binding of the administration.” Based on the

right to equal treatment of article 3 (1) GG, consistent administrative practice creates

external impact and, thus, a possible legal claim.34

48. Even though the German government claims that it regularly sets statement-dead-

lines of several weeks,35 this is not the case in practice. For years, associations have

complained about deadlines that are not allowing for comprehensive examination of

the respective draft legislation.36 There is no indication that public participation re-

garding environmental legislation are dealt with consistently.

49. Consequently, there is no external impact of § 47 (3) GGO which means that there

is no domestic remedy available to enforce the present violation of article 8 of the

Convention.

33 Epping, BeckOK GG, as of 15.05.2023, Art. 65 GG, marginal no. 19.3. 34 Kluckert, Die Selbstbindung der Verwaltung nach Art. 3 I GG, in: JuS 2019, p. 536 (537). 35 See the letter of the BMUV provided as a supporting document. 36 For example: https://www.handelsblatt.com/politik/deutschland/einwanderungsreform-bun- desregierung-gibt-verbaenden-eine-woche-mehr-zeit-fuer-stellungnahmen/28996260.html (2023), https://www.dggg.de/presse/pressemitteilungen-und-nachrichten/kurze-fristen-bei-ver- baendeanhoerungen (2022), https://www.zfk.de/politik/deutschlan/verbaende-kritisieren- kanzleramt-extrem-kurze-anhoerungs-fristen (2019).

11

VII. International remedies

50. No international remedies were invoked.

VIII. Confidentiality

51. The information of the communicant can be made transparent.

IX. Supporting documents (copies, not originals)

52. Copies of the following documents are supplied in support of the communication:

• Excerpt of the Copy of the GGO (in German and English) (paragraphs 6, 17-18,

47-49) [annex 1],

• Excerpt of the ArbZG (in German and English) (paragraph 6) [annex 2],

• Email of the BMWK to the associations with the draft KSG legislation, specifying

the comment deadline (in German and English) (paragraphs 6-7, 13, 21-23, 36,

44) [annex 3],

• Article 9 EU directive EC/1367/2006 (paragraph 19) [annex 4],,

• Statement submitted by the communicant (in German and English) (paragraphs

11, 44) [annex 5],,

• Reply of the BMUV to an UIG request from June 8, 2022 (in German and English)

(paragraphs 19, 48) [annex 6],,

• Power of attorney for legal representation (in German and English) (paragraph

1) [annex 7].

X. Signature

Prof. Dr. Remo Klinger

Legal representative

Transaction Data for the Price Index of Package Holidays in Germany Amelie Blasius

Federal Statistical Office, Germany

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 2

1) The German HICP

2) Data

3) Projects on German package holidays

4) Methodology: hedonic double imputation

5) Plausibility checks

6) Index results

Agenda

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 3

The German HICP and domestic CPI Update of weights

» HICP: Update of consumption weights every year

» CPI: Update of consumption weights every 5 years

» Methodological changes are integrated with CPI weight updates

096 Package holidays

3.5% expenditure

share in HICP in 2023

Use of transaction data in 2023

Index with 21 destination countriesPublished

indices for 7 destinations in

CPI

destatis.de

» Henn et al. (2019) tested transaction data for German package holidays for 2013-2018

» Calculation methods performed well, none was clearly superior

» Indices using transaction data differ from indices using offer pricesMissing variables: meal type and room category

» Feasibility study in 2021 of Destatis

» Data source: tour operators or Amadeus?

» Transaction data of 2015-2021

» Variable handling, choice of index method, plausibility checks, imputation

Amelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 4

Projects on German package holidays

destatis.de

» Transaction data obtained by the global distribution system (GDS) Amadeus Germany

» Very high market share in Germany

» 21 destinations cover ~90% of bookings through Amadeus

» Bookings are heterogeneous in properties

Amelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 5

Data

Booking date

Travel date Duration Travelers Depart. airport

Dest. airport

Hotel- name

City Country Hotel stars

Meal type Room category

Price

01.03.2023 11.04.2023 10 2 HAM AYT Eftalia Hotel

Alanya - Türkler

Turkey 4 All Inclusive

DOUBLE- ROOM

662.90

06.02.2023 15.10.2023 8 4 CGN FUE allsun Hotel

Playa de Esquinzo

Canary Islands

5 All Inclusive

FAMILY- ROOM

2215.00

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 6

Methodology: hedonic double imputation » Estimation of coefficients in current period t and base period b

1 ln(&#x1d45d;&#x1d45f;&#x1d456;&#x1d450;&#x1d452;&#x1d461;) = &#x1d6fc;&#x1d461; + &#x1d6fd;1,&#x1d461; ∗ ln &#x1d461;&#x1d45f;&#x1d44e;&#x1d463;&#x1d452;&#x1d459;&#x1d459;&#x1d452;&#x1d45f;&#x1d460;&#x1d461; + &#x1d6fd;2,&#x1d461; ∗ ln &#x1d451;&#x1d462;&#x1d45f;&#x1d44e;&#x1d461;&#x1d456;&#x1d45c;&#x1d45b;&#x1d461; + &#x1d6fd;3,&#x1d461; ∗ &#x1d460;&#x1d461;&#x1d44e;&#x1d45f;3&#x1d461;+. . . +&#x1d700;

2 ln(&#x1d45d;&#x1d45f;&#x1d456;&#x1d450;&#x1d452;&#x1d44f;) = &#x1d6fc;&#x1d44f; + &#x1d6fd;1,&#x1d44f; ∗ ln &#x1d461;&#x1d45f;&#x1d44e;&#x1d463;&#x1d452;&#x1d459;&#x1d459;&#x1d452;&#x1d45f;&#x1d460;&#x1d44f; + &#x1d6fd;2,&#x1d44f; ∗ ln(&#x1d451;&#x1d462;&#x1d45f;&#x1d44e;&#x1d461;&#x1d456;&#x1d45c;&#x1d45b;&#x1d44f;) + &#x1d6fd;3,&#x1d44f; ∗ &#x1d460;&#x1d461;&#x1d44e;&#x1d45f;3&#x1d44f;+. . . +&#x1d700;

» Prices are imputed by inserting bookings of both periods into equation (1)

» Price index is calculated via geometric mean

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 7

Methodology: hedonic double imputation

» Reliable indices in months with few bookings

» Takes into account many price-determining characteristics

» Method controls for omitted variable bias

» For each destination, month and over time, importance of variables differs

» (forward) stepwise variable selection

Possible variables • log no of travellers • log duration • booking time

Binary variables • school holidays • online/offline booking • travel service included • number of children • hotel stars (2-5) • meal types (5 categories) • clusters of room category • clusters of room features

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

Deletion of erroneous and implausible bookings

07.06.2023Federal Statistical Office (Destatis) 8

Plausibility checks

» Cancellations, duplicates, bookings with missing values

» Duration, Price, Number of travellers, booking time ≤ 0

» Cruises

» Bookings including a rental car

» Booking time > 366

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

Clearing of outliers by destination country

07.06.2023Federal Statistical Office (Destatis) 9

Plausibility checks

0.5% of bookings are cut from the upper (and lower) end of the spectrum

» price per day and person

» duration

» number of:

» travellers

» adults

» children

Dataset Bookings 2019

Bookings 2020

Bookings 2021

Bookings 2022

Without erroneous bookings 1,656,985 1,029,711 740,073 1,349,654 Without 0.5%- and 99.5%-quantils 1,630,078 1,006,402 720,424 1,312,000 Difference 1.62 % 2.26 % 2.66 % 2.89 %

Number of booking after outlier clearing

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

Influential bookings

07.06.2023Federal Statistical Office (Destatis) 10

Plausibility checks

» Cook‘s Distance: Indicator for the influence of a data point

» Cutoff value for outliers?

» Commonly used rules of thumb (&#x1d437; > 1 or &#x1d437; > 4/&#x1d441; and others) do not apply well

» Little insight through analysis of values of single bookings

» Graphical analysis seems best applicable

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 11

Plausibility checks

0

0,05

0,1

0,15

0,2

0,25

0,3

01/22 03/22 05/22 07/22 09/22 11/22

Cook's Distance for Cuba 2022

CooksD

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

Minimum number of observations

07.06.2023Federal Statistical Office (Destatis) 12

Plausibility checks

Framework derived from own robustness checks:

» More than 140 bookings: Index is robust

» Less than 140 bookings: common rule of thumb

» Insufficient number of bookings: Index is compensated

&#x1d45a;&#x1d456;&#x1d45b; &#x1d441; = &#x1d441; &#x1d452;&#x1d465;&#x1d45d;&#x1d459;&#x1d44e;&#x1d45b;&#x1d44e;&#x1d461;&#x1d45c;&#x1d45f;&#x1d466; &#x1d463;&#x1d44e;&#x1d45f;&#x1d456;&#x1d44e;&#x1d44f;&#x1d459;&#x1d452;&#x1d460; ∗ 10

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

Test of multicollinearity

07.06.2023Federal Statistical Office (Destatis) 13

Plausibility checks

Multicollinearity: two explanatory variables correlate

Ex.: most bookings to 5-star-hotels in Egypt include all-inclusive meals

» Biased index

» Indicator: variance inflation factor (VIF)

» Model improvement by deleting selected variables

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

Further model examination

07.06.2023Federal Statistical Office (Destatis) 14

Plausibility checks

» Graphical analysis of β-coefficients and adjusted R²

» Regular pattern over time

» Some coefficients display a seasonal pattern

» Significant inconsistencies point to a problem in the model

» Appear for dummy variables with little variation

» Bookings might have to be deleted

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 15

Plausibility checks Regression coefficients for Balearic Islands

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 16

Index results

55

75

95

115

135

155

Jan 20

Feb 20

Mrz 20

Apr 20

Mai 20

Jun 20

Jul 20

Aug 20

Sep 20

Okt 20

Nov 20

Dez 20

Jan 21

Feb 21

Mrz 21

Apr 21

Mai 21

Jun 21

Jul 21

Aug 21

Sep 21

Okt 21

Nov 21

Dez 21

Jan 22

Feb 22

Mrz 22

Apr 22

Mai 22

Jun 22

Jul 22

Aug 22

Sep 22

Okt 22

Nov 22

Dez 22

Price indices for 096 package holidays 2020 - 2022

Price index 2015=100 rebased to 2020=100, offer prices Price index 2020=100, transaction data

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 17

Index results

-0,2

-0,15

-0,1

-0,05

0

0,05

0,1

0,15

0,2

-10,0

-5,0

0,0

5,0

10,0

15,0

20,0

Jan 21

Feb 21

Mrz 21

Apr 21

Mai 21

Jun 21

Jul 21

Aug 21

Sep 21

Okt 21

Nov 21

Dez 21

Jan 22

Feb 22

Mrz 22

Apr 22

Mai 22

Jun 22

Jul 22

Aug 22

Sep 22

Okt 22

Nov 22

Dez 22

yo y

in fl

at io

n ra

te

Year on year inflation rates for 096 package holidays, 2021-2022

Annual inflation 2015=100 rebased to 2020=100 Annual inflation 2020=100

Contact Statistisches Bundesamt 65180 Wiesbaden Germany

Contact Person Amelie Blasius [email protected] Phone +49 611 75-4732

www.destatis.de

www.destatis.de/kontakt

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 19

Henn, K., Islam, C.-G., Schwind, P., and Wieland, E. (2019). Measuring Price Dynamics of Package Holidays with Transaction Data. EURONA, 2/2019:95–132.

References

destatis.deAmelie Blasius – Transaction Data for the Price Index of Package Holidays in Germany

07.06.2023Federal Statistical Office (Destatis) 20

Holiday destinations covered by transaction data Published indices for domestic CPI • Balearic islands • Canary islands • Egypt • Greece • Turkey

Further indices • Bulgaria • Cuba • Cypris • Dominican Republic • Italy • Maledives • Mauritius • Mexico • Portugal • Spanish mainland • Thailand • Tunisia • United Arab Emirates

• City trips • Cruises

Measuring the distribution work of couples with household survey data New approaches and findings from the German Microcensus

Thomas Körner, Federal Statistical Office Germany

Group of Experts on Gender Statistics, Geneva, 10-12 May 2023

destatis.de

» Distribution of (paid and unpaid) work in couples – a key indicator for gender equality

» Conceptual considerations

» Employment » Full/part-time job » Family and household

» Indicators and findings

» Average difference of hours usually worked in couples » Employment constellations in couples » Cross-classifying the hours usually worked in (employed) couples

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 2

Overview

destatis.deMeasuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 3

A key indicator for gender equality

Impacts of (paid and unpaid) working time distribution in couples, e.g.

» Access to managerial positions » Pay gap » Pension gap Issues particularly striking in couples with children (40-50% earnings „child penalty“ for women; Bönke et al. 2020)

Distribution of paid work as a proxy for the distribution of unpaid work

Distribution of (paid and unpaid) work in couples

“The unequal sharing of unpaid work, with women bearing the

brunt of housework and childcare, is one of the main drivers of gender inequality”

(Gimenez-Nadal/Molina 2022)

destatis.de

Labour Force Concept of the International Labour Organization (ILO 2013) » Extensive defnition of employment, one hour criterion » Persons absent from their job are considered employed provided they have a job

attachment » Persons on annual leave, sick leave, partental leave, maternity leave often continue to be

counted as employed

Risk of over-estimating the employment of mothers (in particular of young children)

Concept of “realised employment” developed by the FSO Germany (Hochgürtel 2018) » Persons employed according to the Labour Force Concept of the ILO » Provided that they were not absent form their job due to maternity leave or parental leave

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 4

Conceptual issues – employment (1)

destatis.de

Employment vs. realised employment of mothers and fathers (Germany, 2021)

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 5

Conceptual issues – employment (2)

63% 58%

65%

74% 80%

82% 83%

14%

47%

62%

73%

79% 82% 83%

91% 91% 91% 91% 91% 91% 91% 86%

90% 91% 91% 91% 91% 91%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

below 1 year 1 to below 2 years 2 to below 3 years 3 to below 6 years 6 to below 10 years 10 to below 15 years

15 to below 18 years

Age of the youngest child

Employment - mothers

Realised employment - mothers

Employment - fathers

Realised emplyoment - fathers

destatis.de

Definition of the ILO (2008) » Part-time job: Hours worked less than those of

comparable full-time job » Depending on national labour markets part-time jobs

are a highly heterogeneous group » Part-time jobs held by women may sytematically differ

from those held by men » Further differences in operationalisation

Relying solely on the full-time/part-time distinction risks to over-estimate female employment participation

Hours usually worked should be analysed in addition

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 6

Conceptual issues – part-time jobs

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

below 10 hours 11 to 20 hours 21 to 30 hours 31 to 40 hours 41 to 50 hours

Fathers Mothers

Hours usually worked in realised part-time employment (Germany, 2022)

destatis.de

Housekeeping vs. household dwelling concept both limit household members to the dwelling » In most surveys, also families are restricted to the persons living in one dwelling » Not in line with family constellations beyond the traditional core family

» „Single“ parents » Blended families including biological as well as foster children

» Frequently incomplete data regarding the actual split of care responsibilities

Age boundary for children » Children living in a household with their parents might be over 18 years old » Breakdown by the age of the youngest child is highly useful for analysis

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 7

Conceptual issues – households and families

destatis.de

Indicator definition » Difference weekly hours usually worked » Population: couples, whose youngest child

is less than three years old » Partners not in realised employment count

for „0“ hours Pros & cons » Simplicty, reduction of complexity » Covers all couples, wheter employed or

not » Suitable as a performance indicator, but

not for differentiated analysis

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 8

Proposed indicators – Working time difference Difference of hours usually worked in couples (Germany)

0

5

10

15

20

25

30

35

2008 2010 2012 2014 2016 2018 2020 2022

Ho ur

s

Year

destatis.de

Indicator definition » Cross-tabulation of (realised) employment

status of both partners of a couple » Population: couples, whose youngest child is

less than three years old » Possible extension to full-time/part-time work Pros & cons » Covers all couples, wheter employed or not » Included in UNECE Gender Statistics

questionnaiore » Differentiation of employment situation of the

mother and the father

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 9

Proposed indicators – Employment constellations Employment constellations of couples (Germany, 2021)

Both employed 37%

Man employed / woman maternity or parental leave

21%

Woman employed / man parental leave

1%

Man employed / woman not employed

30%

Woman employed / man not employed

2%

None employed 9%

destatis.deMeasuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 10

Proposed indicators – Employment constellations (2) Employment constellations of couples (Germany, 2021)

Both employed 37%

Man employed / woman maternity or parental leave

21%

Woman employed / man parental leave

1%

Man employed / woman not employed

30%

Woman employed / man not employed

2%

None employed 9%

FT/PT-Employment constellations of couples (Germany, 2021)

Both full time 25%

Man full time / women part

time 66%

Woman full time / man part time

3%

Both part time 6%

destatis.de

Indicator definition » Cross-tabulation of hours usually worked of

couples in (realised) employment » Population: employed couples, whose youngest

child is less than three years old Pros & cons » Focus on employed couples (37% of the

population) » Needs to be combined with employment

constellations » Very differentiated picture » Definiton of hour bands may affect the results

Measuring the distribution work of couples

Federal Statistical Office (Destatis) 11

Proposed indicators – Hours worked (1) Working time constellations of couples

0

1

2

3

4

5

6

0 1 2 3 4 5 6below 10 11 to 20 21 to 30 31 to 40 41 to 50 51 or more hours usually worked of the father

Ho ur

s us

ua lly

w or

ke d

of th

e m

ot he

r

be lo

w 1

0

11 to

2 0

2

1 to

3 0

3

1 to

4 0

41 to

5 0

5

1 or

m or

e

destatis.deMeasuring the distribution work of couples

Federal Statistical Office (Destatis) 12

Proposed indicators – Hours worked (2) Couples with children below 3 years

0

1

2

3

4

5

6

0 1 2 3 4 5 6below 10 11 to 20 21 to 30 31 to 40 41 to 50 51 or more Hours usually worked of the father

Ho ur

s us

ua lly

w or

ke d

of th

e m

ot he

r

be lo

w 1

0

11 to

2 0

2

1 to

3 0

3

1 to

4 0

41 to

5 0

5

1 or

m or

e

Couples without children

0

1

2

3

4

5

6

0 1 2 3 4 5 6below 10 11 to 20 21 to 30 31 to 40 41 to 50 51 or more Hours usually worked of the father

Ho ur

s us

ua lly

w or

ke d

of th

e m

ot he

r

be lo

w 1

0

11 to

2 0

2

1 to

3 0

3

1 to

4 0

41 to

5 0

5

1 or

m or

e

destatis.de

Indicators on paid work of couples can be used as a proxy for the distribution of unpaid work » No perfect correlation of paid / unpaid work: time use survey data still need » Labour Force Surveys based on household samples provide a rich and frequently available

data source Conceptual decisions need to be taken with care » Employment vs. realised employment » Pros & cons of working with breakdowns by full-time or part-time work » Issues of analyses of „single“ parents and blended families Different indicators complement one another » No indicator alone shows the full picture » Selection of indicator(s) depends on the specific reasearch context

Measuring the distribution work of couples

25.01.2021Federal Statistical Office (Destatis) 13

Conclusions

Contact Statistisches Bundesamt 65180 Wiesbaden Germany

Thomas Körner [email protected] +49 611 75-4413

www.destatis.de

www.destatis.de/kontakt

GDP Flash Estimate and GDP Nowcast: An R-Shiny App for GDP Estimation

Arne Ackermann

Meeting of the Group of Experts on National Accounts

Session 3: Real time indicators & nowcasting

25-27 April 2023

destatis.dedestatis.de

1. Background

2. Econometric approach

3. R-Shiny App

4. Outlook

An R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 2

Outline

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 3

1. Background: Early GDP estimates in Germany

Nowcast t+10

Flash Estimate t+30

1st calculation t+55

» Purely econometric approach

» For internal use only

» Expert and econometric approach

» Publication of GDP flash

» Expert approach

» Publication of detailed results

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 4

2. Econometric approach

» Bottom-up approach: GDP as sum of its aggregates

» Production side (gross value added of 15 sectors), expenditure side (9 aggregates)

» Two-step approach to estimate aggregates (bridge equation)

» Estimation method: seasonal ARIMA models with external regressors

GDP aggregat

Gross value added manufacturing

Indicators

Industrial production

Predictors

Business situation

Aggregation on quarterly frequency

Estimation of GDP aggregate

Estimation of missing months

GDP aggregateIndicatorsPredictors

destatis.dedestatis.de

» Economic plausibility checks of estimation results

» Testing and inclusion of new data sources, e.g. new digital data

» Dealing with crises such as the corona pandemic and the Ukraine war

An R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 5

2. Econometric approach: Challenges

destatis.dedestatis.de

» Development of an R-Shiny environment for econometric estimation of GDP and its main aggregates

» Advantages:

» Compact graphical representation of the estimated GDP aggregates including models, indicators and predictors

» Easy inclusion of new (digital) data in the existing data set and estimation models

» Flexible adaption of models and evaluation thanks to mapping of model parameters

» Clear and user-friendly user interface

An R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 6

3. R-Shiny-App

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 7

3. R-Shiny App: Data input

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 8

3. R-Shiny App: example private consumption

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 9

3. R-Shiny App: example private consumption

GDP aggregate Private consumption

Indicators

Turnover retail trade

Turnover trade of motor vehicles

Turnover accommodation and food service activities

Import of services (NA concept)

Predictors

New passenger car registrations of private owners

Advanced VAT returns in accommodation and food services activities

Aggregation on quarterly frequency

Estimation of GDP aggregate

Estimation of missing months

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 10

3. R-Shiny App: example private consumption

TO retail trade TO motor vehicles trade TO acc. & food services Import of services

Passeng. car registration Adv. VAT ret. acc. & food

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 11

3. R-Shiny-App: example private consumption TO retail trade TO motor vehicles trade TO acc. & food services Import of services

Passeng. car registration Adv. VAT ret. acc. & food

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 12

3. R-Shiny App: example private consumption

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 13

3. 3. R-Shiny App: example private consumption

TO retail trade

TO motor vehicles trade

TO acc. & food services

Import of services

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 14

3. R-Shiny-App: aggregated GDP (expenditure side)

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 15

3. R-Shiny-App: example GVA manufacturing

Industrial production

Prod. expectations

Truck toll mileage ind.

Electricity production

destatis.dedestatis.deAn R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 16

3. R-Shiny-App: aggregated GDP (production side)

destatis.dedestatis.de

» Increase depth of estimation (divide large or important areas)

» Inclusion of an (pseudo) out-of-sample analysis inside the application:

» Estimation and illustration of estimation errors

» Additional criterion for model selection

An R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 17

4. Outlook

realisiation out-of-sample estimate

destatis.dedestatis.de

Special thanks to Dr. Claudia Fries, Dr. Joao Claudio and Xaver Dickopf.

An R-Shiny App for GDP Estimation

19.04.2023Federal Statistical Office (Destatis) 18

Questions?

Contact Statistisches Bundesamt 65180 Wiesbaden Germany

Contact Person Arne Ackermann [email protected] Phone +49 611 75-4923

www.destatis.de

www.destatis.de/kontakt

Identifying economic ownership of Intellectual Property Products: The German experience using the GMPG Decision Tree

Meeting of the Group of Experts on National Accounts

Geneva, Switzerland, 25-27 April 2023

destatis.deIdentifying economic ownership of Intellectual Property Products

Definition

„the result of research and development, investigation or innovation leading to knowledge, use of which is restricted by law or other means of protection” (ESA 3.132)

➢ challenging issue: identification of economic ownership of IPPs

➢ Guide on Measuring Global Production (GMGP) offers a decision tree

Intellectual Property Products (IPPs)

25.04.2023Meeting of the Group of Experts on National Accounts, Geneva, Switzerland, 25-27 April 2023 2

destatis.deIdentifying economic ownership of Intellectual Property Products

Share of GFCF in IPPs of total GFCF in Germany for the years 2000-2020 , percentage value

IPPs in German National Accounts

25.04.2023Meeting of the Group of Experts on National Accounts, Geneva, Switzerland, 25-27 April 2023 3

10%

12%

14%

16%

18%

20% 1

9 9

2

1 9 9

3

1 9 9

4

1 9 9

5

1 9 9

6

1 9 9

7

1 9 9

8

1 9 9

9

2 0 0

0

2 0 0

1

2 0 0

2

2 0 0

3

2 0 0

4

2 0 0

5

2 0 0

6

2 0 0

7

2 0 0

8

2 0 0

9

2 0 1

0

2 0 1

1

2 0 1

2

2 0 1

3

2 0 1

4

2 0 1

5

2 0 1

6

2 0 1

7

2 0 1

8

2 0 1

9

2 0 2

0

Share of GFCF in IPPs of total GFCF

destatis.deIdentifying economic ownership of Intellectual Property Products

GMGP decision tree

25.04.2023 4

Control/ ownership of unit

The unit is part of a MNE

The unit produced the IPP

Production of the IPP

Type of producer

The unit is a main producer of other goods and services and is expected to use the IPP in its production process

The unit is a main IPP producer.

Decision about the economic ownership

The unit may, or may not, receive funding from the parent as compensation for IPP development costs but this aspect is not decisive.

The unit does not receive income from royalties or licenses to use, but either receives compensation for IPP development from the parent or sells the IPP originals to the parent.

The unit receives income from royalties or licenses to use, or does not receive any compensation for IPP development from the parent, so it can be assumed that it is expected to obtain income from royalties and licences to use in the near future.

Income and expenditure related to IPP

destatis.deIdentifying economic ownership of Intellectual Property Products

GMGP decision tree

25.04.2023 5

The unit did not produce the IPP

Production of the IPP

Type of producer

The unit is a main producer of other (non IPP) goods and services and may use the IPP in production

The unit is not a producer of other (non IPP) goods and services. Its main output is IPP related.

Decision about the economic ownership

The unit pays royalties or licenses to use

No IPP related payments are being observed. IPP use may be indirectly observed based on the nature of the production process (with usually high IPP requirements) and above average returns to capital.

Purchase of the IPP from the parent and income from royalties and licenses to use may, or may not, be observed.

Income and expenditure related to IPP

The unit purchased the IPP original for use in production

destatis.deIdentifying economic ownership of Intellectual Property Products

Using the GMGP decision tree

• Data basis: Multinational enterprises (MNEs) of the GNI reservation on globalisation

• Key advantages of the test sample:

• Data exchange between different institutions in Germany

• Further shared data of EU member states

• Additional information from direct MNE contacts

German Approach

25.04.2023Meeting of the Group of Experts on National Accounts, Geneva, Switzerland, 25-27 April 2023 6

destatis.deIdentifying economic ownership of Intellectual Property Products

Relevant data sources

Relevant data sources

25.04.2023Meeting of the Group of Experts on National Accounts, Geneva, Switzerland, 25-27 April 2023 7

Node Control/ownership of the unit Is the unit part of an MNE? Data sources

1 Production of the IPP Is the unit an IPPs producer? SBS, public sources, other

2 Type of producer What is the main kind of activity of the unit? NACE REV. 2 - ISIC REV. 4

3 Use of IPPs Is the unit expected to use the IPPs in the production process? Public sources

4.1 Income from royalties Does the unit receive income from royalties or licenses to use IPPs? Balance of payments

4.2 Expenses for royalties Does the unit pay royalties or for licenses to use IPPs? Balance of payments

5 Compensation for R&D Does the unit receive compensation for IPPs development? Balance of payments

6.1 Income from selling IPPs Does the unit receive income from selling IPPs? Balance of payments

6.2 Expenses for buying IPPs Has the unit expenses for buying IPPs? Balance of payments

destatis.deIdentifying economic ownership of Intellectual Property Products

Depending on the MNE case:

• GMGP decision tree is useful

• for small units with a distinct business activity

• GMGP decision tree is not feasible

• for large and complex units e.g. parent companies

• IPP flows are not available on the level of the individual transaction

General problem of data availability

Results

25.04.2023Meeting of the Group of Experts on National Accounts, Geneva, Switzerland, 25-27 April 2023 8

destatis.deIdentifying economic ownership of Intellectual Property Products

• Data availability

• Insufficient data basis in the German/European statistical system

• Limited information of financial reporting of MNEs

• Direct MNE contact can be useful

• Utilization of the derived results?

• Adequate coordination involving all relevant NSIs

• Possibility to share microdata

Open issues when applying the GMPG decision tree

25.04.2023Meeting of the Group of Experts on National Accounts, Geneva, Switzerland, 25-27 April 2023 9

destatis.deIdentifying economic ownership of Intellectual Property Products

➢ German experience led to differing results

➢ Useful instrument for a limited range of units

➢ Sufficiently large data basis is required to apply the GMPG decision tree

➢ Current data basis in ESS needed to be extended

➢ Remaining obstacles regarding the utilisation of the results are not solved, e.g. microdata exchange

Conclusions

25.04.2023Meeting of the Group of Experts on National Accounts, Geneva, Switzerland, 25-27 April 2023 10

Thank you for your attention!

Questions?

Contact Statistisches Bundesamt 65180 Wiesbaden Germany www.destatis.de

Accounting for CO2 emission trade certificates in the Capital account Meeting of the Group of Experts on National Accounts, Geneva Susanne Goldhammer, Deutsche Bundesbank

Accounting for CO2 emission trade certificates in the Capital account

Outline:

▪ European Emission Trading System/ Cap and trade principle

▪ Economic asset classification

▪ Data sources used in Germany

▪ Insights: trade volume, seasonal trend, trading partners, information about respondents

▪ Emissions trade certificates in the Capital account

▪ Lessons learned

▪ Outlook and discussion

Page 2 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Cap and Trade principle

Source: Umweltbundesamt / Deutsche Emissionshandelsstelle (2022) https://www.umweltbundesamt.de/publikationen/der-europaeische-emissionshandel

Page 3 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Emission Trading System

Sale of

certificates

Purchase of

certificates

Higher CO2

emissions

Lower CO2

emissions - - - Cap

Installation A Installation B

CO2

emissions

Installation A CO2

emissions

Installation B

Emission trade certificates – key issues

− EU Emissions trading allowances entitle an installation or an aircraft operator to emit 1 tonne

of carbon dioxide equivalent (1 metric tonne of CO2 or an amount of another greenhouse gas

with an equivalent global-warming potential) during a specified period.

− At a certain moment, installations or aircraft operators must present allowances to cover each

unit of emissions.

− Relevant for National Accounts: issuance of emissions trading allowances by government and

surrender of allowances

− Relevant for Balance of Payments: trade in emissions trading allowances between residents

and non-residents (changes in economic ownership)

Page 4 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Classification of the Emissions trading allowances

▪ Allowance entitles the holder to do a certain activity: to "emit one tonne of carbon dioxide

equivalent during a specified period“

▪ Allowance does not entitle property rights

▪ Allowance is registered and transferable

▪ Allowance is tradeable

▪ Allowance does not have a corresponding liability

Non-produced non-financial asset

License

Page 5 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Source: BPM6, Table 5.1

Emission trade certificates in Germany: data source

− Direct reporting system

− We collect emission trade certificates with the same „form“ as for securities and financial

derivatives: we want to be informed about acquisition (outgoing payments) and disposal

(incoming payments), and we differentiate between foreign emission trade certificates and

domestic emission trade certificates

− We give explanatory information in our explanatory notes as well as in a separate leaflet on

energy trade („Notice on Energy Trading“)

Page 6 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Emission trade certificates in Germany: development of trade volume

Page 7 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Emission trade certificates in Germany: insight in seasonal trend of monthly source data

Page 8 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Emission trade certificates in Germany: main trading partners 2022

Page 9 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Emission trade certificates in Germany: main respondents 2022

Page 10 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

NACE Rev.2 Division Number

of reports

Reported volume

Financial and insurance activities 64.19 404 53,25 bn EUR

64.90 66 5,84 bn EUR

66 878 57,75 bn EUR

Electricity, gas, steam and air-conditioning 35 758 49,13 bn EUR

Manufacture of chemicals and chemical products 20 85 2,96 bn EUR

Manufacture of coke, and refined petroleum products 19 55 2,36 bn EUR

Manufacture of glass and glass products 23 115 0,87 bn EUR

Manufacture of basic metals 24 53 0,71 bn EUR

Emission trade certificates (net) in the German Capital Account

Page 11 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Emission trade certificates in Germany – lessons learned

− In our direct reporting system, we are not able to separate between primary market and secondary market, but we are able to separate derivatives

− Separation between foreign and domestic certificates is difficult

− If respondents are not able to identify foreign and domestic certificates separately, they are allowed to report the total under domestic certificates

− Banks report on behalf of their clients (in this case, we do not know the branch of the client)

− Clearing by stock exchange causes geographic bias: Clearing house becomes central counterparty, change in economic ownership happens. At the moment, we „look through“ the clearing house.

− Domestic transactions (e.g. domestic airline buys emission trade certificates from domestic stock exchange) are not reported.

Page 12 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Outlook and discussion

− Proposed next generation of EU own resources: European Commission plans to use 25 %

from revenue of EU emissions trading flows as own resources

− BPM6 Update Process:

Guidance Note B.6 Sustainable finance

(„Making purchases and trade in CO2 emission permits visible”)

Guidance Note WS.7 Treatment of Emission Trading Schemes

Page 13 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Outlook and discussion – Guidance Note WS.7

− Proposed option 4 (emission permits as financial assets) is problematic from a conceptual perspective for both, the BPM and the SNA:

− Emission permits do not comply with the definition of financial claims in the SNA:

SNA 2008 para. 3.35: "A financial claim is the payment or series of payments due to the creditor by the debtor under the terms of a liability.".

Yet emission permits do not include a payment from the debtor (the government) to the creditor (the holder of the permit) in any case.

The fact that emission permits are traded on markets does not make them financial assets as pointed out in BPM6 para. 5.8.

− GN WS.7 recommends to record every transaction of EPs at market prices and all positions at the auction price. This would be extremely data demanding since the auction price of every single EP worldwide has to be known at every point in time. The recording of positions at auction price breaks with the principle that positions are to be recorded at market prices.

− Data from a centralized body that coordinates the sales and purchases of emission permits (e.g. Union Registry) is not in accordance with change in economic ownership principle

− With emission permits as financial assets, they would of course no longer appear in the Capital account

Page 14 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Accounting for CO2 emission trade certificates in the Capital account

Thank you for your attention!

Page 15 17 April 2023

Susanne Goldhammer, Deutsche Bundesbank

Looking forward to your questions

and experiences during the

discussion!

*Prepared by Thomas Körner. I am grateful to Matthias Keller, who provided the tabulations of the results from the Microcensus. The views expressed in this paper are those of the author and do not necessarily coincide with the views of the Federal Statistical Office. NOTE: The designations employed in this document do not imply the expression of any opinion whatsoever on the part of the Secretariat of the United Nations concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries.

Economic Commission for Europe Conference of European Statisticians Group of Experts on Gender Statistics Geneva, Switzerland, 10–12 May 2023 Item H of the provisional agenda New approaches to measuring unpaid work and work-life balance

Measuring the distribution work of couples using household survey data. New approaches and findings from the German Microcensus

Note by the Federal Statistical Office Germany*

Abstract

While indicators like the employment participation rate or the hours usually worked of men and women are widely used as standard indicators in the field of gender statistics, the division of labour of couples less frequently taken into account. This is surprising as – especially when children are born – the division of labour of mothers and fathers are renegotiated at family level, often with the effect that women reduce their engagement in paid work to focus on unpaid care work. Against this background, informing policies to promote gender equality often particularly requires indicators on the division of labour of couples. Based on household surveys, such data are easily accessible. Still, the analysis requires a careful application of the internationally agreed concepts such as the employment status and the measurement of working time in order to avoid misinterpretations. Based on recent findings from Germany, the contribution presents approaches to develop suitable indicators and discusses the conceptual pitfalls.

Working paper 27

Distr.: General 25 April 2023 English

Working paper 27

2

I. Introduction

1. Distribution of paid and unpaid work in couples is a key indicator regarding gender equality. The inequalities of the time spent on paid and unpaid work by men and women are one of the main reasons, why women reach lower earnings (Allmendinger 2022) as well as for gender gaps in various social fields. As pointed out by Schrenker and Zucco (2020), unequal distribution of paid and unpaid work between men and women significantly contributes to the Gender Pay Gap.

2. Earnings of men and women, as shown by the regular publications on the Gender Pay Gap, still differ in almost all countries and sectors. It has been frequently noted that the Gender Pay Gap becomes particularly visible when couples have their first child: At this crucial step in life, the distribution of paid and unpaid between the two partners of a couple gets renegotiated. The result is often that the female partner interrupts paid employment to take care of the child, while the man continues paid work (and even increases it). When care obligations decrease as the children grow older, women usually return to the labour market, yet typically in part-time jobs. As a result, when estimating the average earnings over the lifetime, the gap between men and women is around 40 to 45 percent in Germany, with even larger gaps for women with children, which are due to a “child penalty” (Bönke et al. 2020; Kleven et al. 2019).

3. While the distribution of earnings as well as the hours spent on paid employment are readily available from a range of data sources, especially Labour Force Surveys, estimates regarding the distribution of unpaid work are provided less frequently and with more limited precision: Given the more complex data collection process such data are collected by time use surveys in larger intervals and with smaller sample sizes (compared, e.g., to Labour Force Surveys), which consequently have only limited possibilities to provide detailed breakdowns.

4. We argue that indicators on the distribution of paid work in couples can be used as a proxy for the distribution of both paid and unpaid work in couples with and without children. It can be assumed that the partner who is spending less work in paid employment is spending more time on unpaid care and household work. It should however be kept in mind that this relationship is not perfect: As shown by time use surveys, also in couples in which both partners work full-time, the woman on average indicates to spend more time on unpaid household work. In the case of Germany, comparing mothers and fathers in full-time employment, the mothers still provided twice as much time of care work (28 versus 15 minutes per day; Hobler/Pfahl 2017). At the same time, a reduction of the inequality in the hours worked in paid employment is often presented as a key element to close the gender care gap, i.e. the average difference in the number of hours spent on unpaid care work by women and men (Schäper et al. 2023).

5. Figures regarding the distribution of paid work in couples is equally available from many Labour Force Surveys, provided that these are conducted on the basis of a household survey. In Germany, in the Microcensus, one percent of the population is interviewed each year, including detailed information regarding the person’s household situation and detailed information on working time and paid employment.

6. Although indicators on the distribution of paid employment in couples are sometimes used, including in international data bases, their operationalisation is not straightforward. Important conceptual elements of possible indicators require further attention. This concerns especially the definition of employment according to the labour force concept of the International Labour Organization (ILO), the definition of full-time and part-time employment, but also the household and family concepts used.

Working paper 27

3

7. This paper, in section II discusses some of the methodological and conceptual challenges. In section III, a range of different indicators, based on results from the German microcensus for the year 2021 are presented. In the concluding section IV, we summarise the findings and provide suggestions for further developments.

II. Conceptual and methodological challenges

A. Defining employment of couples (with children)

8. The labour force concept of the International Labour Organization (ILO 2013) is the international standard guiding the conceptualisation of work and employment in official statistics. The labour force exhaustively subdivides the entire population in three mutually exclusive groups, i.e. employed persons, unemployed persons and persons outside the labour force. In this concept, employment is defined in an extensive way, i.e. also persons engaged in small jobs of one hour or more are considered as employed (Körner 2012).

9. In the labour force concept, persons on temporary absence during the reference week are considered as employed, provided that they maintain a job attachment during their absence. For this reason, persons with a job but not at work are usually included as employed if they were, e.g., on annual leave, sick leave, parental leave, educational leave, leave due care for others, or absent due to strikes or lockouts.

10. The labour force concept therefore may lead to biased conclusions, when comparing the employment situation of mothers and fathers (Hochgürtel 2018; Kahle/Keller 2018): Mothers not at work due to parental leave (which may be longer than three months “where the return to employment in the same economic unit is guaranteed”; ILO 2013: 6) may be counted as employed, thereby systematically over-estimating the employment of mothers. As men and women are typically affected differently by the different reasons for being absent from the job, analyses on the employment situation of mothers and fathers, as well as of couples may be biased if the Labour Force Concept is applied in a naïve way.

11. Against this background, for the area of analyses regarding the employment of men and women, the Federal Statistical Office (FSO) Germany developed the concept of “realised employment”, building on the available criteria of the labour force concept (Hochgürtel 2018). The group of peopled in realised employment comprises all persons that are employment according to the labour force concept provided that they were not absent from their job due maternity leave or parental leave. As shown in figure 1, referring to the persons in realised employment instead of all employed leads to significant differences in the results, in particular regarding mothers with young children. In 2021 the realised employment rate for mothers with children below the age of one year was 14% while overall employment rate for mothers with children below the age of one year amounted to 63%.

Working paper 27

4

Figure 1 Employment and realised employment for mothers and fathers, by the age of the youngest child (Germany, 2021)

B. Defining full-time vs. part-time employment

12. A frequent indicator to capture a person’s volume of work is distinguishing whether the main job is a full-time or a part-time job. This is sometimes less straightforward as it may seem at first sight, since the hours usually worked within the group part-time jobs (as well as full-time jobs) show considerable variation, which is different for women and for men (figure 2). In the context of work arrangements in couples this can lead to biased results, if the working time of men working part-time is different from the working time of women in part-time employment.

13. According to the ILO convention 175, a part-time worker is “an employed person whose normal hours of work are less than those of comparable full-time workers” (ILO 1994). The ILO resolution on the measurement of working time reaffirms this definition, but refers “contractual hours or hours usually worked” (ILO 2008) (and not normal hours), which is of limited importance in the present context. Assuming that most full-time workers would work around 40 hours per week results in a situation in which full-time workers are a relatively homogeneous group (regarding their working time), while part-time workers can usually work between 1 and 39 hours per week (Kahle/Keller 2018). This is a very heterogeneous group, meaning that working part-time can have diverse implications not only on the related earnings, but also regarding the distribution of work in couples.

14. Looking at the question how part-time employment is usually made operational in surveys further complicates the matter. While the in the Labour Force Surveys of EU member states it is recommended to base the measurement of full-time and part-time jobs on the self- assessment given by the respondent, it is common practice in many other surveys to define a fixed threshold (e.g. 30 hours) distinguishing full-time and part-time jobs. Further rules may apply in data editing to correct the data for implausible replies. The German Microcensus, e.g., until 2019 considered all persons as working full-time if their hours usually worked are

63% 58%

65% 74%

80% 82% 83%

14%

47%

62% 73%

79% 82% 83% 91% 91% 91% 91% 91% 91% 91%86% 90% 91% 91% 91% 91% 91%

below 1 year 1 to below 2 years

2 to below 3 years

3 to below 6 years

6 to below 10 years

10 to below 15 years

15 to below 18 years

Employment - mothers Realised employment - mothers

Employment - fathers Realised emplyoment - fathers

Working paper 27

5

36 or above (regardless of the self-assessment provided in the survey). Persons usually working less than 25 hours are considered part-time workers, while only for those whose hours usually worked in the range from 25 to 36 were categorised according to their self- assessment in the survey. Conclusions regarding to the distribution of work in couples can diverge at least to some extent depend on the definition applied in a given survey.

Figure 2 Distribution of hours usually worked by mothers and fathers in in realised part-time employment (Germany, 2022)

15. A further aspect to be noted is that survey estimates on full- and part-time work usually refer the main job. Secondary jobs are usually not considered, even though their working time may sum up to a full-time job.

C. Defining the household situation of couples

16. According to the UNECE recommendations on the 2020 census round household is defined as “persons who combine to occupy the whole or part of a housing unit and to provide themselves with food and possibly other essentials for living” (UNECE 2015: 162). The two main household concepts in use – the housekeeping concept as well as the household dwelling concept (often used in censuses) – have in common that the basic criterion delimiting the household is that the household members share the same dwelling.

17. Linking the concept of “household” or in particular “family” to the criterion of occupying the same dwelling can be useful for many analytical purposes and may be a straightforward decision to make both concepts operational in surveys. However, it may deviate from the everyday life’s perception of households or families. This may apply, e.g., if not all members of a family share the same dwelling, as persons living in another household by definition cannot be members in further households.

18. Since several decades this delimitation is no longer in line with some household and family constellations, in particular as regards families: Families are no longer limited to the traditional core families living together in a stable way in the same dwelling. A considerable

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

below 10 hours 11 to 20 hours 21 to 30 hours 31 to 40 hours 41 to 50 hours

Fathers Mothers

Working paper 27

6

share of families are recomposed as patchwork families after a separation of the parents. The children living together in a households are no longer necessarily the biological descendants of the adults they live with. In complex or blended families both members of the couple have at least one pre-existing child. Furthermore, children may live in several households at a time, shifting between the households of different biological and step parents, which is often not reflected by survey concepts.

19. Surveys often have to make a choice what they can cover, and analyses of the division of labour of couples is typically limited by such decisions. In household surveys of official statistics (such as the German Microcensus), the focus is often on constellations of living together within the boundaries of one household. Social ties or family relationships to persons living in other households cannot currently be displayed. This means that only those persons are considered as mother or father who live together with their children in the same household. This might be problematic in some family types, such as “single parents” who share the care responsibilities after a separation.

20. To date, there is no solution in the Microcensus regarding families that extend beyond the boundaries of the household. Therefore, the following analyses refer to couples that live in the same household, knowing that this does not fully reflect all current forms of family life.

21. Also the definition of a child deserves some attention: In surveys like the Microcensus, there is principally no age boundary for children, e.g. in many standard tabulations, also full age children living in the household might be included. For analyses of the division of labour of parents, it is advisable to focus on underage children only. Apart from that, it is advisable to apply breakdowns by the age of the youngest child in the household in order to be able to make meaningful comparisons. Persons living together with their full-age children are usually included as households without children.

III. Indicators and findings

22. As shown in the preceding section, analysing the division of paid and unpaid work in couples requires some basic conceptual decisions. For the reasons already mentioned, we have chosen to focus on persons in realised employment, i.e. we do not include employed persons who were temporarily absent from their job due to maternity leave or parental leave.

23. As already shown in figure 1, the realised employment rate of mothers and fathers shows large differences: While 90% of the fathers were in realised employment in 2021, this applied to only 69% of the mothers. The gap gets much bigger, the younger the children living in the household are. Note that the gap entirely disappears if we compare men and women without children. If we focus on an age range comparable to the group of parents, e.g. 21 to 55 years, the realised employment rate is 83% for men and 82% for women.

24. If we focus on couples living together, the results become much more focussed on the specific distribution of work negotiated by the mother and father living together with their child or children.

25. A simple, but very rough indicator boiling down the division of paid work of couples, is the average difference of the weekly hours usually worked in parental couples. In Germany, the indicator refers to mothers and fathers from 15 to 64 years whose youngest child is less than three years old. In includes couples with at least one partner in realised employment (partners not in realised employment counting for 0 hours). Such an indicators presents in

Working paper 27

7

one figure the inequality in the distribution of paid employment and unpaid care work between women and men. According to data from the German Microcensus, the average difference of hours usually worked in couples was 20.9 in 2021 (BMFSFJ 2023). Compared to the year 2008, the difference decreased by 30% from 29.9 hours. Despite this reduction the indicator shows a persistent inequality in the division of paid and unpaid work between men and women, in particular for couples with young children.

26. While it is helpful for a quick assessment to concentrate a complex phenomenon like the division of labour in couples to one single number, this simplification at the same time prevents from a differentiated analysis. The average mixes couples in which one partner works full-time and the other is not in paid employment, with full-time/part-time constellations and couples in which both partners work part-time. A considerable variation in reality is thus strongly reduced in complexity.

27. Another approach is to focus less on the number of hours usually worked, but on the general types of division of labour. Looking at couples with children who are less than three years old, broad categories are mother and father are both in (realised) paid employment, only one of the partners is in paid employment, and none of the partners. As shown based on the Microcensus for the year 2021, in 9% of the couples with children under the age of three years neither the man nor the woman was in (realised) paid employment and 37% of the couples were both employed. In 51% of the couples in this group, the father was employed, while the mother was either not employed or in parental or maternity leave. Only in 3% of the couples, the mothers was in paid employment, while the father was not employed or in parental leave. This picture shows that in more than half of the couples a traditional division of labour prevails, in which the mother fully interrupts her employment, while the man continues to be employment (and even increases the working hours, as further analyses show). Cases in which the father interrupts his employment are still very rare.

Figure 3 Employment constellations of couples, whose youngest child is less than three years old (Germany, 2021)

Both employed 37%

Man employed / woman maternity or parental leave 21%

Woman employed / man parental leave 1%

Man employed / woman not employed

30%

Woman employed / man not employed

2%

None employed 9%

Working paper 27

8

28. Following this analytical approach, we can distinguish couples in which the woman entirely stopped working with a highly traditional division of labour. It is analytically fruitful to distinguish this group before taking a closer look into the couples in which both the mother and the father are in paid employment. It should however be noted that such an analysis only covers 37 % of the couples whose youngest child is less than three years old. Out of these 37 % of couples, in which both partners are employed, in two thirds the father is in full-time paid employment, while the mother is employed part time (see figure 4). In 25% of the couples, both partners are working full time, and in 6% both are employed part time. In only 3 percent of the couples, the mother holds a full-time job, while the father is working part time. Note that a very similar distribution can also be observed for employed couples with children aged less than 18 years old (both full-time 27%; both part-time 5%; man full- time/woman part-time 3%; woman part-time/man full-time 66%), the main difference being that in this group the share of couples, in which both partners are in paid employment amounts to 66% (instead of 37% for the couples with a child aged less than 3 years).

Figure 4 Employment constellations of couples in realised employment, whose youngest child is less than three years old (Germany, 2021)

29. The figures presented in figure 4 still not provides a fully accurate picture. As already noted, the range of hours usually worked by part-time (but also full-time) workers is subject to considerable variation. This is particularly relevant, since mothers in part-time employment tend to work less hours than fathers in part-time employment (see figure 2). Similarly, fathers in full-time employment on average work more hours than mothers in full-time employment.

30. A possible solution is to analyse the hours usually worked by couples, broken down by the hours worked of the father and the mother. Figure 5 shows the hours worked by the father on the x-axis and the hours worked by the mother on the y-axis. The figure shows, e.g. that the most frequent employment constellation of couples with children below the age of three is that both the mother usually works 21 to 30 hours and the father 31 to 40 hours (which

Both full time 25%

Man full time / women part time

66%

Woman full time / man part time

3%

Both part time 6%

Working paper 27

9

applies to 20 % of the couples).1 The bubbles on the diagonal represent cases, in which the mother’s and the father’s usual working hours fall into the same category. The (few) bubbles above the diagonal represent couples in which the mother usually works more hours than the father and the bubbles below the diagonal represent cases in which the father usually works longer. Overall, the hours usually worked are equal in 24.1 % of the couples, whose youngest child is less than three years old. In 73.1 % of the couples, the father falls into a higher category of hours usually worked and only in 2.8 % the mother is working longer hours. Compared to the results shown in figure 4, one can conclude that the comparison based on the full-time/part-time distinction underestimates the differences in working time of mothers and fathers.

Figure 5 Employment constellations of couples in realised employment, whose youngest child is less than three years old (Germany, 2021)

1 Note that in figure 5 data cannot be presented for several data points due to insufficient cell size.

0

1

2

3

4

5

6

0 1 2 3 4 5 6below 10 11 to 20 21 to 30 31 to 40 41 to 50 51 or more

hours usually worked of the father

ho ur

s u su

al ly

w or

ke d

of th

e m

ot he

r be

lo w

1 0

11 to

2 0

2

1 to

3 0

31

to 4

0

41

to 5

0

5 1

or m

or e

Working paper 27

10

IV. Conclusions

31. Information on the distribution of paid work in couples can provide important hints also regarding the distribution of unpaid work in couples. Without measuring unpaid work directly, information on paid work can be used as a proxy for arrangements regarding unpaid work, and at least describes the potential of an equal distribution of unpaid work.

32. The indicators presented in this paper can be a highly useful complement to results on the division of unpaid work from Time Use Surveys, typically collected in large intervals and in relatively small samples. The indicators presented in this paper can be produced on the basis of Labour Force Surveys based on a household sample. As Labour Force Surveys are usually run more frequently and based on larger sample sizes, more differentiated analyses are possible, e.g. focussing on the situation of couples with young children by their socio- economic status.

33. As we have shown, several conceptual decisions need to be taken that have implications on the interpretation of the indicators. This includes the definition of employment (for which we suggest the concept of realised employment), which is facing different requirements when applied to the division of labour of couples, compared, e.g. to economic analysis. We have also shown that the common distinction between full-time and part-time employment is going along with some uncertainty, since the group if part-time employed tends to be very heterogeneous and often a harmonised operationalisation is lacking. Using the information regarding the hours usually worked is analytically richer, however more difficult to communicate.

34. All the indicators presented in section III have their right, and are appropriate measures in different contexts. We have shown that, nevertheless, relying on some of the indicators proposed only can lead to biased results, as, e.g., the inequality of the division of labour of couples might be under-estimated.

V. References

Allmendinger, Jutta, 2022: Es geht nur gemeinsam! Wie wir Geschlechtergerechtigkeit erreichen. Berlin.

BMFSFJ (Federal Ministry for Family Affairs, Senior Citizens, Women and Youth), 2023: Equality Atlas. At https://www.bmfsfj.de/bmfsfj/meta/en/equality/equalityatlas?view=

Bönke, Timm et al., 2020: Wer gewinnt? Wer verliert? Die Entwicklung und Prognose von Lebenserwerbseinkommen in Deutschland. Gütersloh: Bertelsmann Stiftung. At: https://www.bertelsmann-stiftung.de/de/publikationen/publikation/did/wer-gewinnt-wer-verliert- 2020

Eurostat, 2021: EU Labour Force Survey Explanatory Notes (to be applied from 2021Q1 onwards). European Commission: Luxembourg. At https://ec.europa.eu/eurostat/documents/1978984/6037342/EU- LFS+Explanatory+notes+from+Q1+2021+onwards.pdf

Hobler, Dietmar and Svenja Pfahl, 2017: Zeitaufwand für Hausarbeit 2012/2013. In: WSI GenderDatenPortal. At https://www.boeckler.de/data/wsi_gdp_zeitverwendung_20170601_01.pdf

Working paper 27

11

Hochgürtel, Tim, 2018: Realisierte Erwerbstätigkeit zur Messung des Vereinbarkeitsarrangements von Familie und Beruf. In: Wirtschaft und Statistik 1/2018.

ILO, 1994: C175 - Part-Time Work Convention, 1994. At https://www.ilo.org/dyn/normlex/en/f?p=NORMLEXPUB:12100:0::NO::P12100_ILO_CODE: C175

ILO, 2008: Resolution concerning the measurement of working time. Adopted by the Eighteenth International Conference of Labour Statisticians (November/December 2008) At https://www.ilo.org/wcmsp5/groups/public/---dgreports/--- stat/documents/normativeinstrument/wcms_112455.pdf

ILO, 2013: Resolution concerning statistics of work, employment and labour underutilization. Adopted by the Nineteenth International Conference of Labour Statisticians (October 2013) At https://www.ilo.org/global/statistics-and-databases/standards-and-guidelines/resolutions- adopted-by-international-conferences-of-labour-statisticians/WCMS_230304/lang--en/index.htm

Kahle, Irene und Matthias Keller, 2018: Realisierte Erwerbstätigkeit von Müttern und Vätern zur Vereinbarkeit von Familie und Beruf. In: Wirtschaft und Statistik 3/2018.

Kleven, Henrik et al. 2019a: Child Penalties Across Countries: Evidence and Explanations. AEA Papers and Proceedings, 109, 122–26. At https://www.henrikkleven.com/uploads/3/7/3/1/37310663/klevenetal_aea-pp_2019.pdf

Körner, Thomas, 2012: Measuring the Labour Status in Official Statistics: The Labour Force Concept of the International Labour Organisation and its Implementation in the Labour Force Survey. In: Hoffmeyer-Zlotnik, Jürgen/Warner, Uwe (editors). Demographic Standards for Surveys and Polls in Germany and Poland: National and European Dimension. GESIS: Cologne 2012, pp. 123–138.

Schäper, Clara et al., 2023: Gender Care Gap and Gender Pay Gap Increase Substantially until Middle Age. DIW weekly report 9/2023, pp. 84-88. At https://www.diw.de/documents/publikationen/73/diw_01.c.867440.de/dwr-23-9-1.pdf

Schrenker, Annekatrin and Aline Zucco, 2020: The gender pay gap begins to increase sharply at age of 30. DIW weekly report 10/2020, pp. 75-82. At https://www.econstor.eu/bitstream/10419/220007/1/1692667602.pdf

UNECE, 2015: Conference of European Statisticians Recommendations for the 2020 Censuses on Population and Housing. United Nations: New York and Geneva.

*Подготовил Томас Кёрнер. Я благодарен Маттиасу Келлеру, который предоставил таблицы результатов

микропереписи. Мнения, выраженные в этой статье, являются мнениями автора и не обязательно совпадают с

мнением Федерального статистического управления.

ПРИМЕЧАНИЕ: Обозначения в настоящем документе не подразумевают выражения какого-либо мнения

Секретариата Организации Объединенных Наций в отношении юридического положения любой страны,

территории, города или края или их властей или в отношении делимитации ее границ.

Европейская экономическая комиссия

Конференция европейских статистиков

Группа экспертов по гендерной статистике Женева, Швейцария, 10-12 мая 2023 года

Пункт H предварительной повестки дня

Новые подходы к измерению неоплачиваемого труда и баланса между работой и личной

жизнью

Измерение распределения труда в парах на основе данных обследования домохозяйств. Новые подходы и результаты микропереписи в Германии

Записка Федерального статистического управления Германии

Аннотация

В качестве стандартных показателей в гендерной статистике часто

используются такие показатели, как доля участия в занятости или обычное

рабочее время мужчин и женщин, при этом разделение труда в парах

принимается во внимание реже. Это удивительно, поскольку, особенно когда

рождаются дети, разделение труда матерей и отцов пересматривается на

семейном уровне, из-за чего зачастую женщины сокращают свою занятость на

оплачиваемой работе, чтобы сосредоточиться на неоплачиваемой работе по

уходу за детьми. В этом контексте для выработки политики по продвижению

гендерного равенства часто особенно востребованы показатели, касающиеся

разделения труда между партнерами в паре. Такие данные легко доступны,

если опираться на обследования домохозяйств. Тем не менее, во избежание

неверного толкования при анализе необходимо аккуратно применять

согласованные на международном уровне концепции, такие как статус

занятости и измерение продолжительности рабочего времени. На основе

недавно полученных в Германии результатов в докладе представлены подходы

Рабочий документ 27

Distr.: General

23 апреля 2023 г. 6:51:00

English

Рабочий документ 27

2

к разработке подходящих показателей и рассматриваются типовые ошибки,

связанные с неправильным использованием концепций.

I. Введение

1. Распределение оплачиваемой и неоплачиваемой работы между партнерами в паре

является ключевым показателем гендерного равенства. Неравное время,

затрачиваемое мужчинами и женщинами на оплачиваемую и неоплачиваемую работу,

является одной из основных причин более низких заработков женщин (Allmendinger

2022), а также гендерного разрыва в различных социальных сферах. Как отмечают

Schrenker and Zucco (2020), неравное распределение оплачиваемой и неоплачиваемой

работы между мужчинами и женщинами в значительной степени способствует

гендерному разрыву в оплате труда.

2. Доходы мужчин и женщин, как показывают регулярные публикации о гендерном

разрыве в оплате труда, по-прежнему различаются практически во всех странах и

секторах. Часто отмечается, что гендерный разрыв в оплате труда становится

особенно заметным, когда у пары рождается первый ребенок: На этом решающем

этапе жизни пересматривается распределение оплаченного и неоплачиваемого труда

между партнерами. В результате женщина часто прерывает оплачиваемую работу,

чтобы заботиться о ребенке, в то время как мужчина продолжает оплачиваемую

работу (и даже работает больше). Когда по мере взросления детей обязанности по

уходу сокращаются, женщины обычно возвращаются на рынок труда, но, как правило,

на неполный рабочий день. В результате при оценке среднего заработка в течение

жизни разрыв между мужчинами и женщинами в Германии составляет около 40-45

процентов, причем разрыв еще серьезнее для женщин с детьми, что связано с

«детским наказанием» (Bönke et al. 2020; Kleven et al. 2019).

3. В то время как распределение заработков, а также количество часов, затрачиваемых

на оплачиваемую работу, легко получить из ряда источников данных, особенно из

обследований рабочей силы, оценки распределения неоплачиваемого труда

предоставляются реже и с меньшей точностью. Учитывая более сложный процесс

сбора данных, такие данные собираются в ходе обследований затрат времени, и это

происходит реже и с меньшими размерами выборки (по сравнению, например, с

обследованиями рабочей силы). В итоге возможности для подробной разбивки данных

ограничены.

4. Мы показываем, что индикаторы распределения оплачиваемой работы в парах можно

использовать в качестве косвенного показателя распределения как оплачиваемой, так

и неоплачиваемой работы в парах с детьми и без детей. Можно предположить, что

партнер, который тратит меньше времени на оплачиваемую работу, больше времени

тратит на неоплачиваемую работу по уходу и работу по дому. Однако следует иметь в

виду, что это соотношение не совершенно: как показывают обследования затрат

Рабочий документ 27

3

времени, в парах, в которых оба партнера работают полный рабочий день, женщина в

среднем указывает, что тратит больше времени на неоплачиваемую работу по дому. В

случае Германии при сравнении матерей и отцов, работающих полный рабочий день,

матери по-прежнему вдвое больше времени уделяют работе по уходу (28 минут по

сравнению с 15 минутами в день; Hobler/Pfahl 2017). В то же время сокращение

неравенства в продолжительности рабочего времени на оплачиваемой работе часто

представляется как ключевой элемент, позволяющий устранить гендерный разрыв в

затратах времени на работу по уходу, то есть среднюю разницу в количестве часов,

затрачиваемых женщинами и мужчинами на неоплачиваемую работу по уходу

(Schäper et al. 2023).

5. Данные о распределении оплачиваемой работы в парах можно также получить из

многочисленных обследований рабочей силы при условии, что они проводятся на

основе обследования домохозяйств. В Германии в ходе микропереписи каждый год

опрашивается один процент населения. Им задаются вопросы для получения

подробной информации о домохозяйстве, в котором живет человек, а также

подробной информации о рабочем времени и оплачиваемой занятости.

6. Хотя показатели распределения оплачиваемой занятости в парах иногда

используются, в том числе в международных базах данных, их операционализация

является непростой задачей. Необходимо уделить больше внимания важным

понятийным элементам потенциальных показателей. Это касается, в частности,

определения занятости в соответствии с концепцией рабочей силы Международной

организации труда (МОТ), определения полной и неполной занятости, а также

используемых концепций домохозяйства и семьи.

7. В данном документе в разделе II обсуждаются некоторые методологические и

концептуальные проблемы. В разделе III представлен ряд различных показателей,

полученных по итогам микропереписи населения Германии за 2021 год. В

заключительном разделе IV мы подводим итоги и даем предложения относительно

дальнейшей работы.

II. Концептуальные и методологические проблемы

A. Определение занятости пар (с детьми)

8. Концепция рабочей силы Международной организации труда (МОТ, 2013 г.) является

международным стандартом, регулирующим разработку концепции работы и

занятости в официальной статистике. Понятие рабочей силы исчерпывающим образом

разделяет все население на три взаимоисключающие группы: занятых, безработных и

лиц, не входящих в состав рабочей силы. В этой концепции занятость определяется

экстенсивно, то есть занятыми также считаются лица, занятые небольшими работами

продолжительностью один час и более (Körner 2012).

9. В концепции рабочей силы лица, временно отсутствующие в течение учетной недели,

считаются занятыми при условии, что они сохраняют привязанность к работе во время

своего отсутствия. По этой причине лица, имеющие работу, но не работающие,

обычно включаются в число занятых, если они находились, например, в ежегодном

отпуске, отпуске по болезни, отпуске по уходу за ребенком, учебном отпуске, отпуске

по уходу за другими людьми или отсутствовали из-за забастовки или локаута.

Рабочий документ 27

4

10. Таким образом, концепция рабочей силы может привести к необъективным выводам

при сравнении статуса занятости матерей и отцов (Hochgürtel 2018; Kahle/Keller 2018):

Матери, не работающие из-за отпуска по уходу за ребенком (который может длиться

более трех месяцев, «когда гарантировано возвращение на работу в ту же

экономическую единицу»; МОТ 2013: 6), могут считаться занятыми, тем самым

систематически завышается занятость матерей. Поскольку мужчины и женщины, как

правило, по-разному реагируют на разные причины отсутствия на работе, анализ

ситуации с занятостью матерей и отцов, а также пар может быть необъективным, если

концепция рабочей силы применяется примитивно.

11. На этом фоне для анализа занятости мужчин и женщин Федеральное статистическое

управление (ФСУ) Германии разработало концепцию «фактической занятости»,

основываясь на имеющихся критериях концепции рабочей силы (Hochgürtel 2018). В

группу людей фактической занятости входят все лица, являющиеся занятыми в

соответствии с концепцией рабочей силы, при условии, что они не отсутствовали на

работе в связи с отпуском по беременности и родам или отпуском по уходу за

ребенком. Как показано на Рисунке 1, учет лиц, имеющих фактическую занятость, а

не всех занятых, приводит к значительным изменениям в результатах, в частности, что

касается матерей с маленькими детьми. В 2021 году уровень фактической занятости

матерей с детьми в возрасте до одного года составлял 14%, а общий уровень занятости

матерей с детьми в возрасте до одного года составлял 63%.

Рисунок 1

Занятость и фактическая занятость матерей и отцов в зависимости от возраста

самого младшего ребенка (Германия, 2021 г.)

B. Определение полной и неполной занятости

12. Часто в качестве показателя для определения объема выполняемой работы

используется понятие основной работы на условиях полной или частичной занятости.

Иногда это не так просто, как может показаться на первый взгляд, поскольку обычное

63% 58%

65%

74% 80% 82% 83%

14%

47%

62%

73% 79% 82% 83%

91% 91% 91% 91% 91% 91% 91% 86% 90% 91% 91% 91% 91% 91%

below 1 year 1 to below 2 years

2 to below 3 years

3 to below 6 years

6 to below 10 years

10 to below 15 years

15 to below 18 years

Employment - mothers Realised employment - mothers

Employment - fathers Realised emplyoment - fathers

Рабочий документ 27

5

рабочее время при групповой работе на условиях неполного рабочего дня (а также на

условиях полной занятости) имеет значительные колебания, которые различаются для

женщин и мужчин (рисунок 2). В контексте организации работы в парах это может

привести к необъективным результатам, если рабочее время мужчин, работающих

неполный рабочий день, отличается от рабочего времени женщин, занятых неполный

рабочий день.

13. Согласно Конвенции МОТ 175 работник, занятый неполное рабочее время, означает

работающее по найму лицо, нормальная продолжительность рабочего времени

которого меньше нормальной продолжительности рабочего времени работников,

занятых полное рабочее время и находящихся в сравнимой ситуации» (МОТ 1994).

Резолюция МОТ об измерении рабочего времени подтверждает это определение, но

здесь речь идет о «контрактном времени или обычном рабочем времени» (МОТ, 2008

г.) (а не о нормальном времени), что имеет ограниченное значение в данном

контексте. Если предположить, что большинство работников, занятых полный

рабочий день, будет работать около 40 часов в неделю, получится ситуация, в которой

работники, занятые полный рабочий день, представляют собой относительно

однородную группу (в отношении их рабочего времени), в то время как работники,

занятые неполный рабочий день, обычно могут работать от 1 до 39 часов в неделю

(Kahle/Keller 2018). Это очень неоднородная группа, а это означает, что работа

неполный рабочий день может иметь различные последствия не только для

соответствующих заработков, но и для распределения работы в парах.

14. Изучение вопроса о том, как неполный рабочий день обычно используется в

обследованиях, еще больше усложняет ситуацию. В то время как в обследованиях

рабочей силы в государствах-членах ЕС рекомендуется основывать измерение

занятости полный и неполный день на самооценке, данной респондентом, во многих

других обследованиях обычной практикой является определение фиксированного

порогового значения (например 30 часов) для разграничения работы на условиях

полной и неполной занятости. Дополнительные правила могут применяться при

редактировании данных для исправления данных в случае неправдоподобных ответов.

Например, для целей микропереписи в Германии до 2019 года все лица с

продолжительностью обычного рабочего времени 36 или более часов считались

работниками на полной занятости (независимо от самооценки, полученной в ходе

обследования). Лица, обычно работающие менее 25 часов, считались работающими

неполный рабочий день, при этом только тех, кто обычно работал от 25 до 36 часов,

относили к той или иной категории в соответствии с их самооценкой по данным

обследования. Выводы о распределении работы в парах могут быть хотя бы в какой-то

степени неоднозначными в зависимости от определения, применяемого в том или

ином обследовании.

Рисунок 2

Распределение обычного рабочего времени матерей и отцов при фактической

неполной занятости (Германия, 2022 г.)

Рабочий документ 27

6

15. Еще один аспект, который следует отметить, заключается в том, что оценки

обследований относительно полной и неполной занятости обычно касаются основного

места работы. Дополнительные места работы обычно не рассматриваются, даже если

отработанное там рабочее время может в сумме составлять полный рабочий день.

C. Определение статуса домохозяйства для пар

16. Согласно рекомендациям ЕЭК ООН по проведению переписи 2020 года

домохозяйство определяется как «лица, которые совместно занимают всю или часть

жилой единицы и обеспечивают себя продуктами питания и, возможно, другими

предметами первой необходимости для жизни» (ЕЭК ООН, 2015 г.: 162. Две основные

используемые концепции домохозяйства – концепция ведения домашнего хозяйства, а

также концепция жилища домохозяйства (часто используемая в переписях) –

объединяет то, что основным критерием разграничения домохозяйства является то,

что члены домохозяйства проживают в одном и том же жилище.

17. Увязка понятия «домохозяйство» или, в частности, «семья» с критерием проживания в

одном и том же жилище может быть полезна для многих аналитических целей и

простым решением для использования обоих понятий в обследованиях. Однако оно

может отличаться от повседневного восприятия домохозяйств или семей. Такое может

быть, например, если не все члены семьи проживают в одном и том же жилище,

поскольку лица, проживающие в другом домохозяйстве, по определению не могут

быть членами дальнейших домохозяйств.

18. По прошествии нескольких десятилетий это разграничение уже не соответствует

некоторым структурам домохозяйства и семьи, в частности это касается семей. Семьи

теперь – это не только традиционные семьи, постоянно проживающие вместе в одном

жилище. Значительная часть семей превращается в сводные семьи после того как

родители начинают жить по отдельности. Дети, живущие вместе в домашнем

хозяйстве, больше не обязательно являются биологическими потомками взрослых, с

которыми они живут. В сложных или смешанных семьях у обоих членов пары есть, по

крайней мере, один ранее существовавший ребенок. Кроме того, дети могут

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

below 10 hours 11 to 20 hours 21 to 30 hours 31 to 40 hours 41 to 50 hours

Fathers Mothers

Рабочий документ 27

7

проживать одновременно в нескольких домохозяйствах, перемещаясь между

домохозяйствами разных биологических и приемных родителей, что часто не

отражается в концепции обследования.

19. При проведении обследований часто приходится выбирать их охват, и такие решения

обычно ограничивают анализ разделения труда в паре. В обследованиях домохозяйств

официальной статистики (таких как микроперепись в Германии) основное внимание

часто уделяется структурам совместного проживания в пределах границ одного

домохозяйства. Социальные связи или семейные отношения с лицами,

проживающими в других домохозяйствах, в настоящее время невозможно отобразить.

Это означает, что матерью или отцом считаются только те лица, которые живут

вместе со своими детьми в одном домохозяйстве. Это может быть проблематично в

некоторых типах семей, таких как «одинокие родители», которые делят между собой

обязанности по уходу после развода.

20. На сегодняшний день в микропереписи нет решения относительно семей, выходящих

за пределы домохозяйства. Поэтому нижеследующий анализ относится к парам,

проживающим в одном домохозяйстве, при этом мы знаем, что это не в полной мере

отражает все современные формы семейной жизни.

21. Кроме того, заслуживает внимания определение ребенка. В обследованиях, подобных

микропереписи, для детей принципиально не устанавливаются возрастные границы,

то есть во многие стандартные таблицы также могут быть включены

совершеннолетние дети, проживающие в домохозяйстве. Для анализа разделения

труда родителей целесообразно ориентироваться только на несовершеннолетних

детей. Кроме того, рекомендуется применять разбивку по возрасту самого младшего

ребенка в домохозяйстве, чтобы можно было проводить осмысленные сравнения.

Лица, проживающие вместе со своими совершеннолетними детьми, обычно

включаются в состав домохозяйств без детей.

III. Индикаторы и выводы

22. Как показано в предыдущем разделе, для анализа распределения оплачиваемой и

неоплачиваемой работы в парах необходимо принять несколько основных

концептуальных решений. По уже упомянутым причинам мы решили сосредоточить

внимание на лицах, имеющих фактическую занятость, то есть мы не включаем

занятых лиц, которые временно отсутствовали на работе из-за отпуска по

беременности и родам или отпуска по уходу за ребенком.

23. Как уже показано на рисунке 1, уровень фактической занятости матерей и отцов

сильно различается: В то время как 90% отцов имели фактическую занятость в 2021

году, то же самое верно только для 69% матерей. Разрыв становится тем больше, чем

младше дети, живущие в домохозяйстве. Обратите внимание, что разрыв полностью

исчезает, если сравнивать мужчин и женщин без детей. Если мы сосредоточимся на

возрастном диапазоне, сравнимом с группой родителей, то есть от 21 до 55 лет, то

реальный уровень занятости составляет 83% для мужчин и 82% для женщин.

24. Если мы рассмотрим пары, живущие вместе, результаты будут гораздо больше

сосредоточены на конкретном распределении труда, о котором договорились мать и

отец, живущие вместе со своим ребенком или детьми.

Рабочий документ 27

8

25. Простым, но очень грубым показателем распределения оплачиваемой работы внутри

пары является средняя разница между обычным рабочим временем партнеров. В

Германии этот показатель относится к матерям и отцам в возрасте от 15 до 64 лет, у

которых младшему ребенку меньше трех лет. Он включает пары, где хотя бы один из

партнеров имеет фактическую занятость (партнеры, не имеющие фактической

занятости, обозначаются как работающие 0 часов). Такой показатель отражает в одной

цифре неравенство в распределении оплачиваемой занятости и неоплачиваемой

работы по уходу между женщинами и мужчинами. Согласно данным микропереписи в

Германии средняя разница между обычным рабочим временем в парах в 2021 году

составляла 20,9 часа (BMFSFJ 2023). По сравнению с 2008 годом разница

уменьшилась на 30% с 29,9 часов. Несмотря на это сокращение, индикатор

демонстрирует сохраняющееся неравенство в распределении оплачиваемой и

неоплачиваемой работы между мужчинами и женщинами, особенно для пар с

маленькими детьми.

26. Хотя для экспресс-оценки удобно представить сложное явление, такое как разделение

труда внутри пары, одним числом, тем не менее такое упрощение препятствует

дифференцированному анализу. Под средним значением скрываются пары, в которых

один партнер работает полный рабочий день, а другой не работает по найму, пары, в

которых партнеры работают полный/неполный рабочий день, и пары, в которых оба

партнера работают неполный рабочий день. Таким образом, значительные реальные

различия сильно упрощаются.

27. Другой подход заключается в том, чтобы сосредоточиться не на продолжительности

обычного рабочего времени, а на общих типах разделения труда. Изучая пары с

детьми в возрасте до трех лет, можно выделить такие широкие категории как: и мать,

и отец имеют (фактическую) оплачиваемую работу, только один из партнеров имеет

оплачиваемую работу, и ни один из партнеров не имеет оплачиваемой работы. Как

показали данные микропереписи населения за 2021 год, в 9% пар с детьми в возрасте

до трех лет ни мужчина, ни женщина не имели (фактической) оплачиваемой работы, а

в 37% пар работали оба партнера. У 51% пар этой группы отец работал, а мать либо не

работала, либо находилась в отпуске по уходу за ребенком или в отпуске по

беременности и родам. Лишь в 3% пар матери работали по найму, а отец не работал

или находился в отпуске по уходу за ребенком. Это показывает, что более чем у

половины пар преобладает традиционное разделение труда, при котором мать

полностью прерывает свою занятость, а мужчина продолжает работать (и даже

работает более продолжительное время, как показывает дальнейший анализ). Случаи,

когда отец прерывает работу, все еще очень редки.

Рабочий документ 27

9

Рисунок 3

Структура занятости в парах, где младший ребенок моложе трех лет (Германия,

2021 г.)

28. Следуя этому аналитическому подходу, мы можем выделить пары, в которых

женщина полностью перестала работать и произошло весьма традиционное

разделение труда. С точки зрения анализа разумно выделить эту группу, прежде чем

рассматривать пары, в которых и мать, и отец имеют оплачиваемую работу. Однако

следует отметить, что такой анализ охватывает только 37% пар, в которых младшему

ребенку меньше трех лет. Из этих 37% пар, в которых работают оба партнера, в двух

третях случаев отец работает на оплачиваемой работе полный рабочий день, а мать

работает неполный рабочий день (см. Рис. 4). В 25% пар оба партнера работают

полный рабочий день, а в 6% – оба работают неполный рабочий день. Только в 3%

пар мать работает полный рабочий день, а отец работает неполный рабочий день.

Обратите внимание, что очень похожее распределение можно наблюдать и для

работающих пар с детьми в возрасте до 18 лет (оба родителя работают полный

рабочий день - 27%; оба работают неполный рабочий день - 5%; мужчина работает

полный рабочий день/женщина работает неполный рабочий день - 3%; женщина

работает неполный рабочий день/мужчина работает полный рабочий день - 66%). При

этом основное отличие состоит в том, что в этой группе доля пар, в которых оба

партнера находятся на оплачиваемой работе, составляет 66% (вместо 37% для пар с

ребенком в возрасте до 3 лет).

Both employed 37%

Man employed / woman maternity or parental leave 21%

Woman employed / man parental leave 1%

Man employed / woman not employed

30%

Woman employed / man not employed

2%

None employed 9%

Рабочий документ 27

10

Рисунок 4

Структура занятости в парах, имеющих фактическую занятость, где младший

ребенок моложе трех лет (Германия, 2021 г.)

29. Цифры, представленные на Рис.4, все еще не дают полностью точной картины. Как

уже отмечалось, диапазон продолжительности обычного рабочего времени при

неполной занятости (но также и при работе полный день), подвержен значительным

колебаниям. Это особенно актуально, поскольку матери, занятые неполный рабочий

день, как правило, работают меньше часов, чем отцы, занятые неполный рабочий день

(см. Рис. 2). Точно так же отцы, занятые полный рабочий день, в среднем работают

больше часов, чем матери, занятые полный рабочий день.

30. Возможным решением является анализ продолжительности обычного рабочего

времени пары с разбивкой по времени, отработанному отцом и матерью. На Рис. 5

рабочее время отца отложено по оси абсцисс, а рабочее время матери — по оси у. На

рисунке показано, например, что наиболее частой структурой занятости пар с детьми

в возрасте до трех лет является вариант, при котором мать обычно работает от 21 до

30 часов, а отец от 31 до 40 часов (что применимо к 20 % пар).1 Кружки на диагонали

представляют собой случаи, когда обычное рабочее время матери и отца попадает в

одну и ту же категорию. (Несколько) кружков над диагональю представляют пары, в

которых мать обычно работает больше часов, чем отец, а кружки под диагональю

обозначают случаи, в которых отец обычно работает дольше матери. В целом

продолжительность рабочего времени, как правило, одинакова у 24,1 % пар, где

младший ребенок моложе трех лет. В 73,1% пар продолжительность обычного

рабочего времени отца больше и только в 2,8% пар мать работает дольше отца. По

сравнению с результатами, показанными на Рис. 4, можно сделать вывод, что

1 Обратите внимание, что на Рисунке 5 данные не могут быть представлены для нескольких точек данных из-

за недостаточного размера ячейки.

Both full time 25%

Man full time / women part time

66%

Woman full time / man part time

3%

Both part time 6%

Рабочий документ 27

11

сравнение, основанное на разнице между полной и неполной занятостью,

недооценивает разницу в продолжительности рабочего времени матерей и отцов.

Рисунок 5

Структура занятости в парах, имеющих фактическую занятость, где младший ребенок

моложе трех лет (Германия, 2021 г.)

IV. Выводы

31. Информация о распределении оплачиваемой работы между партнерами в парах может

также дать важные косвенные указания относительно распределения неоплачиваемой

работы в парах. Без прямого измерения неоплачиваемого труда информация об

оплачиваемой работе может использоваться в качестве косвенного указания на

договоренности относительно неоплачиваемого труда и, по меньшей мере, описывает

потенциал равного распределения неоплачиваемого труда.

32. Индикаторы, представленные в этом документе, могут быть очень полезным

дополнением к результатам о разделении неоплачиваемого труда по данным

0

1

2

3

4

5

6

0 1 2 3 4 5 6до 10 от 11 до 20 от 21 до 30 от 31 до 40 от 41 до 50 51 и более

продолжительность обычного рабочего времени отца, часов

п р

о д

о л ж

и те

л ьн

о ст

ь о

б ы

ч н

о го

р аб

о ч

ег о

в р

ем ен

и м

ат ер

и ,

ч ас

о в

д о

1 0

о т

1 1

д о

2 0

о т

2 1

д о

3 0

о т

3 1

д о

4 0

о т

4 1

д о

5 0

5 1 и

б о

л ее

Рабочий документ 27

12

обследований затрат времени, которые обычно проводятся через большие интервалы

времени и на относительно небольших выборках. Индикаторы, представленные в

настоящем документе, могут быть получены на основе обследований рабочей силы на

выборке домохозяйств. Поскольку обследования рабочей силы обычно проводятся

чаще и имеют больший размер выборки, возможен более дифференцированный

анализ, например, можно сконцентрироваться на парах с маленькими детьми и

рассмотреть их социально-экономическое положение.

33. Как мы показали, необходимо принять несколько концептуальных решений,

влияющих на интерпретацию показателей. Это в том числе определение занятости

(для которого мы предлагаем концепцию фактической занятости), к которому

предъявляются различные требования применительно к разделению труда между

партнерами в паре, по сравнению, например, с экономическим анализом. Мы также

показали, что общее различие между занятостью полный и неполный рабочий день

сопровождается некоторой неопределенностью, поскольку группа, занятая неполный

рабочий день, склонна к большой неоднородности, и зачастую отсутствует

унифицированная практическая реализация. Использование информации об обычном

рабочем времени предоставляет больше возможностей для анализа, однако передача

информации затруднена.

34. Все индикаторы, представленные в разделе III, имеют право на существование и

являются подходящими показателями в различных контекстах. Тем не менее, мы

показали, что опора лишь на некоторые из предложенных показателей может

привести к необъективным результатам, так как, например, неравенство в разделении

труда между партнерами в парах может быть недооценено.

V. Литература

Allmendinger, Jutta, 2022: Es geht nur gemeinsam! Wie wir Geschlechtergerechtigkeit

erreichen. Berlin.

BMFSFJ (Federal Ministry for Family Affairs, Senior Citizens, Women and Youth), 2023:

Equality Atlas. At https://www.bmfsfj.de/bmfsfj/meta/en/equality/equalityatlas?view=

Bönke, Timm et al., 2020: Wer gewinnt? Wer verliert? Die Entwicklung und Prognose von

Lebenserwerbseinkommen in Deutschland. Gütersloh: Bertelsmann Stiftung. At:

https://www.bertelsmann-stiftung.de/de/publikationen/publikation/did/wer-gewinnt-wer-verliert-

2020

Eurostat, 2021: EU Labour Force Survey Explanatory Notes (to be applied from 2021Q1

onwards). European Commission: Luxembourg. At

https://ec.europa.eu/eurostat/documents/1978984/6037342/EU-

LFS+Explanatory+notes+from+Q1+2021+onwards.pdf

Hobler, Dietmar and Svenja Pfahl, 2017: Zeitaufwand für Hausarbeit 2012/2013. In: WSI

GenderDatenPortal. At

https://www.boeckler.de/data/wsi_gdp_zeitverwendung_20170601_01.pdf

Hochgürtel, Tim, 2018: Realisierte Erwerbstätigkeit zur Messung des

Vereinbarkeitsarrangements von Familie und Beruf. In: Wirtschaft und Statistik 1/2018.

Рабочий документ 27

13

ILO, 1994: C175 - Part-Time Work Convention, 1994. At

https://www.ilo.org/dyn/normlex/en/f?p=NORMLEXPUB:12100:0::NO::P12100_ILO_CODE:

C175

ILO, 2008: Resolution concerning the measurement of working time. Adopted by the Eighteenth

International Conference of Labour Statisticians (November/December 2008) At

https://www.ilo.org/wcmsp5/groups/public/---dgreports/---

stat/documents/normativeinstrument/wcms_112455.pdf

ILO, 2013: Resolution concerning statistics of work, employment and labour underutilization.

Adopted by the Nineteenth International Conference of Labour Statisticians (October 2013) At

https://www.ilo.org/global/statistics-and-databases/standards-and-guidelines/resolutions-

adopted-by-international-conferences-of-labour-statisticians/WCMS_230304/lang--en/index.htm

Kahle, Irene und Matthias Keller, 2018: Realisierte Erwerbstätigkeit von Müttern und Vätern zur

Vereinbarkeit von Familie und Beruf. In: Wirtschaft und Statistik 3/2018.

Kleven, Henrik et al. 2019a: Child Penalties Across Countries: Evidence and Explanations. AEA

Papers and Proceedings, 109, 122–26. At

https://www.henrikkleven.com/uploads/3/7/3/1/37310663/klevenetal_aea-pp_2019.pdf

Körner, Thomas, 2012: Measuring the Labour Status in Official Statistics: The Labour Force

Concept of the International Labour Organisation and its Implementation in the Labour Force

Survey. In: Hoffmeyer-Zlotnik, Jürgen/Warner, Uwe (editors). Demographic Standards for

Surveys and Polls in Germany and Poland: National and European Dimension. GESIS: Cologne

2012, pp. 123–138.

Schäper, Clara et al., 2023: Gender Care Gap and Gender Pay Gap Increase Substantially until

Middle Age. DIW weekly report 9/2023, pp. 84-88. At

https://www.diw.de/documents/publikationen/73/diw_01.c.867440.de/dwr-23-9-1.pdf

Schrenker, Annekatrin and Aline Zucco, 2020: The gender pay gap begins to increase sharply at

age of 30. DIW weekly report 10/2020, pp. 75-82. At

https://www.econstor.eu/bitstream/10419/220007/1/1692667602.pdf

UNECE, 2015: Conference of European Statisticians Recommendations for the 2020 Censuses

on Population and Housing. United Nations: New York and Geneva.

Economic Commission for Europe Conference of European Statisticians Group of Experts on National Accounts Twenty-second session Geneva, 25-27 April 2023 Item 7 of the provisional agenda Real-time indicators and nowcasting

GDP Flash Estimate and GDP Nowcast: An R-Shiny App for GDP Estimation

Prepared by the German Federal Statistics Office1

Summary

In recent years DESTATIS has been publishing a GDP flash estimate 30 days after the end of the quarter and a purely model-based GDP nowcast 10 days after the end of the quarter for internal use. Since spring 2020, the GDP flash estimate and nowcast have been facing new challenges related to the ongoing corona pandemic, supply and material bottlenecks, price increases, and the war in Ukraine. GDP Flash estimate and GDP nowcast: An R-Shiny app for GDP estimation by DESTATIS presents a new tool developed to cope with these additional uncertainties. The application allows to carry out the model estimations more flexible, faster and less error-prone manner than before. It permits loading a comprehensive set of indicators that can be expanded if new indicators become available. Furthermore, it is possible to test various estimation scenarios based on previously defined models of the GDP sub-aggregates. The application also includes a variety of graphical evaluation options to analyse the estimated GDP, production- and use-side components and the underlying monthly economic indicators

1 Prepared by Arne Ackermann and Claudia Fries, Destatis.

United Nations ECE/CES/GE.20/2023/9

Economic and Social Council Distr.: General 22 March 2023 English only

ECE/CES/GE.20/2023/9

2

I. Scope of the R-Shiny App

1. Since the second quarter of 2020, the Federal Statistical Office has been publishing gross domestic product (GDP) as a GDP flash estimate 30 days after the end of the quarter. For this publication, expert estimates for production and expenditure side are supplemented by econometric estimates2. In addition, a purely model-based GDP nowcast is estimated 10 days after the end of the quarter for internal use (see Dickopf, Janz and Mucha, 2019). After an initial feasibility study in 2019, the econometric model as well as the estimation tool for the nowcast has been continuously further developed.

2. Since spring 2020, the GDP estimates (flash estimate and nowcast) have been facing new challenges. The corona pandemic, supply and material bottlenecks, price increases and the Russian war of aggression in Ukraine have led to increasing estimation uncertainties for the time series econometric models of GDP flash estimate and GDP nowcast (see Ackermann, Dickopf and Mucha, 2021). To cope with these additional uncertainties, we started to develop an R-Shiny app as a graphical user interface for the estimation process. With this new tool, it is possible to carry out the model estimations more flexibly, faster and less error-prone than before. The app allows loading a comprehensive set of indicators that can be expanded if new indicators become available. Furthermore, it is possible to test various estimation scenarios based on previously defined models of the GDP sub-aggregates. The app also includes a variety of graphical evaluation options to analyse the estimated results of GDP, its production and use side-aggregates and the underlying monthly economic indicators. Other features as e.g. an integrated revision analysis could be added in the future.

3. This contribution will first explain the underlying econometric approach. Then, we present the main functionalities and some application examples of the R-Shiny app that is used internally by the Federal Statistical Office for the econometric part of the GDP flash estimate and for the GDP nowcast.

II. Econometric Approach

4. We estimate the quarterly gross domestic product following the bottom-up structure of the detailed GDP calculation. Therefore, estimated values for aggregates of GDP are calculated first and then aggregated to form one overall GDP result for the production side, and one for the expenditure side. Thus, the results of the production and expenditure approach are determined independently. The basis of the calculations on the production side is the gross value added in currently 15 aggregated economic sectors, plus taxes on goods and less subsidies on goods (see Table 1, first column). In order to increase the estimation precision in trade (section G) and to better assess the impulses of each of the three subsections, trade has been divided into its three subsections G45, G46 and G47. Thus, trade (section G) is estimated indirectly as the sum of the estimates of the three subsections. This procedure is also planned for other aggregates on both the production and the expenditure side (e.g. private consumption expenditure). On the expenditure side, we estimate 9 aggregates (see Table 1, second column) and manually add an assumed value for changes in inventories and acquisitions.

2 Following the Delphi method (see appendix Figure 7)

ECE/CES/GE.20/2023/9

3

Table 1 Estimated components of GDP

Production side Expenditure side

NACE Section

Economic Area

A Agriculture, forestry and fishing Private consumption expenditure B Mining and quarrying General government expenditure C Manufacturing Gross fixed capital formation, Machinery and

equipment D Energy supply Gross fixed capital formation, Buildings E Water supply, sewerage, waste

management and remediation activities Gross fixed capital formation, Other products

F Construction Export of Goods G G45 G46 G47

Wholesale and retail trade, sale and repair of motor vehicles and motorcycles Wholesale and retail trade and repair of motor vehicles and motorcycles Wholesale trade, except of motor vehicles and motorcycles Retail trade, except of motor vehicles and motorcycles

Export of Services

H Transportation and storage Import of Goods I Accommodation and food services Import of Services J Information and communication K Financial and insurance activities L Real estate activities M, N Business services O, P, Q Public services, education, health R, S, T Other services Taxes on products Subsidies on products

5. The individual aggregates and subsections are estimated using bridge equations. If suitable indicators are available for an aggregate, these are included in the corresponding model as external regressors. Currently, up to six indicators are included for each aggregate. Most indicators are available on a monthly basis, while GDP aggregates are available on a quarterly basis. Therefore, the monthly indicators are aggregated to quarterly frequency for estimation. At the time of the estimation, normally only one to two monthly values of the underlying indicators for the current quarter are available. The missing monthly values have to be estimated before aggregating to quarterly frequency (Eurostat, 2016). The two steps of the procedure are summarized for the example of private consumption in Figure 1. The estimation of missing monthly values might include up to three predictors as external regressors, if available.

6. Methodologically, ARIMA models with external regressors are used for the individual estimates in both steps. Seasonality in the time series is considered within the modelling, and indicators and predictors are adjusted for seasonal effects.

ECE/CES/GE.20/2023/9

4

Figure 1 Estimation steps, example private consumption

III. Features: Application

A. Menu and Data Input

7. The menu bar on the left allows to switch between different parts of the estimation and to choose several options which will be explained in the following sections.

8. On the first page “Dateneingang”, four data files need to be selected for upload as input for the estimations:

• Steuerungsdatei: In this file the estimation models with respect to external regressors for the aggregates and all indicators are described.

• Datensatz Indikatoren: This file contains the time series of all possible indicators and predictors in monthly frequency.

• Datensatz Entstehung: This file contains the time series of all production side aggregates.

• Datensatz Verwendung: This file contains the time series of all expenditure side aggregates.

9. The menu bar allows to set the start date of the model estimation (“Start Modellkalibrierung”), to set the quarter to be estimated (“Aktuelles Quartal”), to manually set the value for inventories and acquisitions and to choose the time interval to be displayed in the graphical outputs.

ECE/CES/GE.20/2023/9

5

Figure 2 Menu bar and data upload

B. Selection of approach and aggregate: example private consumption expenditure

10. The list in the menu under “BIP-Seite” allows to switch between production or expenditure approach. In the example in Figure 3, the expenditure approach (“Verwendung”) is chosen.

11. If the second menu point “Aggregate” is selected, output appears on the right panel. In the header of this panel, a drop-down menu allows to choose between the aggregates of the selected GDP approach. The example in Figure 3 shows the output for private consumption expenditure.

12. The graph displays the chain index for unadjusted (blue line) and seasonally adjusted (red line) data. The nowcasts for both series are displayed as dots at the current edge of the time line. On the left hand of the graph the estimated model is printed.

13. Below, drop-down menus show the indicators for private consumption and their predictors as given by the input file “Steuerungsdatei” and described in Figure 1. The drop- down menus allow to change the indicators and their predictors in a flexible way by choosing from the time series set given in the file “Datensatz Indikatoren”. To facilitate the analysis and choice of indicators and predictors, the time series are displayed graphically below the menus as shown in Figure 4. Additionally, the estimated models can be displayed (see Figure 5).

ECE/CES/GE.20/2023/9

6

Figure 3 Estimation of private consumption expenditure

ECE/CES/GE.20/2023/9

7

Figure 4 Estimation of indicators for private consumption expenditure

ECE/CES/GE.20/2023/9

8

Figure 5 Estimation of indicators for private consumption expenditure, including display of estimated model

ECE/CES/GE.20/2023/9

9

C. Calculation of GDP

14. If the estimation models of all aggregates of one GDP side, here the expenditure approach, are chosen, the calculation of aggregated GDP for the expenditure side can be displayed be selecting the third point “BIP” in the menu bar and clicking the button “BIP rechnen” at the top of the right panel. Then, the right panel is showing a graph with unadjusted (blue) and seasonally adjusted (red) GDP (see Figure 6). In addition, a table under the graph summarizes the estimates for GDP and all aggregates, giving level estimates as well as growth rates and confidence intervals for the aggregates. Final estimation models for each aggregate can be displayed below.

15. The same procedure is to be applied to obtain estimates for the production approach of GDP.

16. The last button in the menu bar allows to print the estimates for each approach to an Excel-file.

IV. Advantages and further development

17. For our regular quarterly nowcast and flash estimation, the R shiny app has the following advantages:

• It provides a compact graphical representation of the estimated GDP aggregates and the indicators used for them.

• The inclusion of new (digital) data in the existing data set and testing of this data indicators and predictors is easy and quick. It allows to adjust estimation models in a flexible way, which is especially helpful in times of crises or other unusual economic developments

• The user interface does not only allow fast and easy adoption of models, but also helps to understand and interpret economic interdependencies thanks to the visualization of the time series employed in the estimation.

18. Our plans for further development of the app in the future comprise two areas. First, we envisage to include a (pseudo) out-of-sample analysis within the application. The estimation errors of a given estimation model could be calculated and displayed to provide the user with an additional criterion to select the best possible model inside the application. Second, we plan to increase the depth of calculation for more than one aggregate.

ECE/CES/GE.20/2023/9

10

Figure 6 Estimation of GDP for expenditure side

ECE/CES/GE.20/2023/9

11

Literature

Ackermann, Dickopf and Mucha (2021). Flash und Nowcast: Schnellschätzungen des Bruttoinlandsprodukts in der Corona-Pandemie. In WISTA – Wirtschaft und Statistik, Ausgabe 4/2021.

Dickopf, Janz and Mucha (2019. Vom BIP-Flash zum BIP-Nowcast: Erste Ergebnisse einer Machbarkeitsstudie zur weiteren Beschleunigung der BIP-Schnellschätzung. In: WISTA Wirtschaft und Statistik. Ausgabe 6/2019.

Eurostat (2016). Overview of GDP flash estimation methods. In: Eurostat statistical working papers. p. 15 ff.

ECE/CES/GE.20/2023/9

12

Appendix

Figure 7 Delphi-method-style three-pillar-model to obtain a reconciled quarterly GDP at t+30