Silvia Giacinti

Principal Economist-Statistician BIRD methodology –

DG-S
Mappings
BIRD methodology – Reference / non-reference
Rubric

The multiple codes problem

Several different codification systems exist…

BSI AnaCredit
BS_COUNT_SECTOR INSTTNL_SCTR
BS counterpart sector Institutional sector
2100 - General Government S13 - Central government (excluding social security funds)
1100 - Central Bank (S.121)
S121 - Central banks
00BK - Non-resident banks
122_A - Credit institutions

SHS FINREP
ISSUER_SECTOR CPS
Issuer ESA 2010 sector Counterparty sector
S_13 - General government x1 - General governments
S_121 - The central bank x10 - Central banks
S_122 - Deposit-taking corporations except the central bank
x12 - Credit institutions

… But the input layer should use one single

set of codes!

Mappings 2 www.ecb.europa.eu ©
Rubric

From multiple dictionaries to BIRD: the semantic

integration

FINREP
COREP BIRD
SHS
AE
RES

The integration is made on a micro level, the granularity of each reporting

framework is maintained.

The semantic integration is firstly made analysing variables and members, trying
to find correspondence between the non reference codification systems and the
reference one.

www.ecb.europa.eu ©
Rubric

Some examples of integration

➢ The relevance of the hierarchy for data integration

Area code list SHS

Restricted area code list
SDD
W0 World
MEMBER_CODE DESCRIPTION
KH Cambodia U4 Extra Euro area
KI Kiribati
KH Cambodia
KM Comoros
KN Saint Kitts and Nevis KI Kiribati
KP Korea, Democratic People`s Republic of KM Comoros
KR Korea, Republic of KN Saint Kitts and Nevis
KW Kuwait
Korea, Democratic
KP People`s Republic of
MEMBER_CODE DESCRIPTION

KR Korea, Republic of
W2 Domestic (home or reference area) KW Kuwait
U4 Extra Euro area
U5 Other Euro area member states (all countries except the
W0 World (all entities, including reference area, including IO)
W1 Rest of the World
www.ecb.europa.eu ©
Rubric

Some examples of integration

At domain level :

➢ Split the same non reference domain in more that one reference domain when
the specific domain is describing different categories of information.
SDD
Sector
The RIAD Sector Subtype Destination
12500A Central Clearing Counterparties (CCPs)
RIAD 12500B Financial Vehicle Corporations (FVCs)
ESAST 12500C Securities and derivatives dealers
Source 12500D Financial corporations engaged in lending
S.125.E Central Counterparty Clearing houses (CCPs) 12500E Specialised financial corporations, except CCPs
12500L Financial corporations other than MFIs, non-MMF investment funds,
S.125.A FVCs
88888Z Unknown sector
S.125.B Securities and derivatives dealers 12400Z Non-MMF investment funds
S.125.C Financial corporations engaged in lending 12400Z Non-MMF investment funds
S.125.D Specialised financial corporations, except CCPs 12400Z Non-MMF investment funds
S.125.X other 12400Z Non-MMF investment funds
NN Not applicable / not known 12400Z Non-MMF investment funds
S.124.A bonds
ECB
S.124.B equities
FuInTyp
S.124.C hedge
Destination
S.124.D mixed
01 bonds
S.124.X other
02 equities
03 hedge
04 mixed
05 other
www.ecb.europa.eu ©
BIRD methodology – Reference / non-reference
Rubric

Reference / non-reference

reference
Non-
SHS

S_121

1100 x10

Reference
S121

Mappings 6 www.ecb.europa.eu ©
BIRD methodology – Reference / non-reference
Rubric

Mappings

• Mappings provide the relation between codification systems

• They allow converting one dataset from one codification system to
another

Mappings 7 www.ecb.europa.eu ©
BIRD methodology – Reference / non-reference
Rubric

Mappings: Dealing with complexity

• Mappings are complex because:
– They are n to m (e.g. it is possible to map from 2 variables to 3 variables)
– The same element is sometimes used with different meanings (same element
mapped more than once)

• 4 types of mapping
• Algorithm (for non-enumerated variables)
• Equivalence table (for enumerated variables)
• Deletion (to delete a variable not needed in the output)
• Variable set (for measure dimensions)

Mappings 8 www.ecb.europa.eu ©
BIRD methodology – Reference / non-reference
Rubric

Mappings: A realistic example

Data structures
CUBE ID: ABC
Variable Subdomain Role
FREQ M - Monthly D
AAA_ISIN String D
F1 - Short term debt security
AAA_INST_TYP F2 - Long term debt security O
AAA_MATURITY_DAT
E yyyy_mm_dd O

CUBE ID: REF_ABC

Variable Subdomain Role
ISIN String D
TYP_INSTRMNT 1 - Debt security O
1 - Short term
ORGNL_MTRTY 2 - Long term O
DT_MTRTY dd_mm_yyyy O

FREQ AAA_ISIN AAA_INST_TYP AAA_MATURITY_DATE

M AB123456789 F1 2016_10_05
M CD123456789 F1 2052_01_31 Data instances
M EF123456789 F2 2019_02_29

ISIN TYP_INSTRMNT ORGNL_MTRTY DT_MTRTY

AB123456789 1 1 05/10/2016
CD123456789 1 1 31/01/2052
EF123456789 1 2 29/02/2019
Mappings 9 www.ecb.europa.eu ©
BIRD methodology – Reference / non-reference
Rubric

Mappings: Technical solution

MAPPING_DEFINITION
VARIABLE_EQUIV
MAPPING_TYP MEMBER_EQUIVALE
MAPPING_ID ALENCE_TABLE_ ALGORITHM
E NCE_TABLE_ID
ID
1 Deletion 1
2 Algorithm 2 =
Equivalence
3 3 1
table
4 Algorithm 4 ChangeDateFormat

VARIABLE_MAPPING
MAPPING_ID VARIABLE_ID IS_SOURCE
1 FREQ TRUE
2 AAA_ISIN TRUE MEMBER_MAPPING_ITEM
2 ISIN FALSE EQUIVALENC EQUIVALENCE_TABL IS_SO
E_TABLE_ID E_ROW VARIABLE_ID MEMBER_ID URCE
AAA_INST_TY
3 TRUE 1 1* F1 TRUE
P
TYP_INSTRMN 1 1 TYP_INSTRMNT 1 FALSE
3 FALSE
T 1 1 ORGNL_MTRTY 1 FALSE
ORGNL_MTRT 1 2* F2 TRUE
3 FALSE
Y 1 2 TYP_INSTRMNT 1 FALSE
AAA_MATURIT 1 2 ORGNL_MTRTY 2 FALSE
4 TRUE
Y_DATE
4 DT_MTRTY FALSE

Mappings 10 www.ecb.europa.eu ©