THE USE OF OPTIONAL DIMENSIONS IN SDMX DATA

THE USE OF OPTIONAL DIMENSIONS IN SDMX DATA STRUCTURE DEFINITIONS. APPLICATION IN BANCO DE ESPAÑA DATABASES Carlos S. Morillo Gálvez Specialist in Statistical Information Management and Dissemination Systems 6 TH SDMX GLOBAL CONFERENCE Addis Ababa October 2 – 5, 2017 DEPARTMENT OF STATISTICS

SUMMARY - Introduction - Business case - Conclusions DEPARTMENT OF STATISTICS 2

INTRODUCTION Overview - FAME database system - Based on SDMX information model since 1998 - Five types of databases: - Public data (national and international) - Primary data - Derived data - Confidential data - Exchange of information DEPARTMENT OF STATISTICS 3

INTRODUCTION Overview - Banco de España collects and disseminates internally national (Bd. E, INE) and international (IMF, OECD) data - BIEST DEPARTMENT OF STATISTICS 4

INTRODUCTION Overview - Each time-series follows a DSD - Databases contain time-series from different DSDs - The DSDs are stored in one database as case series - And so are the codelists… DEPARTMENT OF STATISTICS 5

INTRODUCTION Overview The ID of any time-series is formed by code of the DSD, followed by the value codes of each dimension. - Monthly ECB spot exchange rate of Euro against US Dollar: DTC. CBCE. USD. EUR. M • • • - DTC: code of the DSD CBCE: ECB spot exchange rate USD: quoted currency EUR: base currency M: monthly Annual IMF spot exchange rate of US Dollar against Moroccan Dirham: DTC. CFMI. MAD. USD. A - Monthly market exchange rate of Euro against Uruguayan Peso: DTC. CMME. UYU. EUR. M DEPARTMENT OF STATISTICS 6

BUSINESS CASE - Unemployment rate Employment rate Activity rate Employed persons Inactive persons Employed persons Unemployment rate Employed persons of women of men in Madrid of men in Andalucia over 65 years old between 25 and 65 years old over 65 years old in agriculture of foreigners in Madrid of nationals in military forces with a permanent job with a temporary job with a permanent job in Madrid Sex Reference area Age Economic activity Nationality Type of contract DEPARTMENT OF STATISTICS 7

BUSINESS CASE First approach: one unique DSD (for instance, EAPS) - Unemployment rate of women EAPS. UER. F. ZZ. ZZ. ZZ. A - Employment rate of men in Madrid EAPS. EMR. M. ZZ. ZZ. MAD. A - Inactive persons between 25 and 65 years old EAPS. INA. ZZ. A 25_65. ZZ. ZZ. A - Unemployment rate of foreigners in Madrid EAPS. UER. ZZ. ZZ. EXT. ZZ. MAD. A - Employed persons with a temporary job EAPS. EMP. ZZ. ZZ. TEMP. ZZ. A DEPARTMENT OF STATISTICS 8

BUSINESS CASE First approach: one unique DSD - PROS: - Only one DSD to work with, referring a single topic/statistic - CONS: - Abuse of ZZ value (not considered) - Too long series code (not 100% meaningful) - Series difficult to be found by means of dimensions DEPARTMENT OF STATISTICS 9

BUSINESS CASE Second approach: one DSD per crossing (total: 8 DSDs) - Unemployment rate of women: DSD 1. UER. F. A - Employment rate of men: DSD 1. EMR. M. A - Employment rate of men in Madrid: DSD 2. EMR. M. MAD. A - Activity rate of men in Andalucia: DSD 2. ACR. M. AND. A - Employed persons over 65 years old: DSD 3. EMP. AGT 65. A - Inactive persons between 25 and 65 years old: DSD 3. INA. A 25_65. A DEPARTMENT OF STATISTICS 10

BUSINESS CASE Second approach: one DSD per crossing - PROS: - Series codes are fully understandable - CONS: - Too many DSDs to work with, although all of them are related - When adding/removing a crossing dimension, the DSD changes - Need for identifying the corresponding DSD DEPARTMENT OF STATISTICS 11

BUSINESS CASE Bd. E approach: one DSD with optional dimensions - Main magnitude and frequency are compulsory - Rest of dimensions are optional. - _{letter} introduces the code of an optional dimension DEPARTMENT OF STATISTICS 12

BUSINESS CASE Bd. E approach: one DSD with optional dimensions - Unemployment rate of women DEP. UER. _SF. A - Employment rate of men in Madrid DEP. EMR. _SM. _AMAD. A - Inactive persons between 25 and 65 years old DEP. INA. _EA 25_65. A - Unemployment rate of foreigners in Madrid DEP. UER. _PEXT. _AMAD. A - Employed persons with a temporary job DEP. EMP. _CTEMP. A DEPARTMENT OF STATISTICS 13

BUSINESS CASE FYI, the real DSD is: With 20 optional dimensions. DEPARTMENT OF STATISTICS 14

BUSINESS CASE DEPARTMENT OF STATISTICS 15

BUSINESS CASE DEPARTMENT OF STATISTICS 16

BUSINESS CASE Bd. E approach: one DSD with optional dimensions - PROS: - Only one DSD to work with. - Series code easy to read and interpret. - CONS: - There must be one at least. DEPARTMENT OF STATISTICS 17

CONCLUSIONS - Given a DSD, optional dimensions help analysts in searching of time series more easily. - Only for storage purposes and internal dissemination (for the time being). - To exchange data, SDMX 2. 1/2. 0 compliant DSDs are used. DEPARTMENT OF STATISTICS 18

Thank you very much for your attention carlos. morillo@bde. es DEPARTMENT OF STATISTICS