IAOS 2018 - Using administrative registers to verify information obtained from official surveys , G. Tavares Lameiro da Costa, P.L. do Nascimento Silva
Globally inclusive approaches to measurement_Erhabor Idemudia.pdfStatsCommunications
More Related Content
Similar to IAOS 2018 - Using administrative registers to verify information obtained from official surveys , G. Tavares Lameiro da Costa, P.L. do Nascimento Silva
A Brief Analysis of Medical Device Quality System Enforcement Actions [2015 -...EMMAIntl
Similar to IAOS 2018 - Using administrative registers to verify information obtained from official surveys , G. Tavares Lameiro da Costa, P.L. do Nascimento Silva (20)
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
IAOS 2018 - Using administrative registers to verify information obtained from official surveys , G. Tavares Lameiro da Costa, P.L. do Nascimento Silva
1. Using administrative registers to verify
information obtained from official
surveys
Gustavo Tavares Lameiro da Costa
Pedro Luis do Nascimento Silva
Brazilian Institute for Geography and Statistics
www.ibge.gov.br
2. Problem
Goal
Use Administrative Registers to validate a specific
variable obtained from a Sample Survey.
Brazilian Exports Register
Exports Variable from
Annual Survey of Manufacturing (PIA)
4. Numbers
PIA 2016
Population: 424,098
Take-all stratum: 34,269
Take-some stratum: 389,829
Stratified by: 27 States; ISIC-4; Size of firm
SECEX 2016
Population: 27,702
All Brazilian Exporting Firms
Exporting firms in Manufacturing: 13,600
5. 14 DIGIT CNPJ
Merging databases
8 DIGIT CNPJ
8 DIGIT CNPJ
MIDPOINTS
CLASS EXPORTS
USD:
up to 1 mi
1 - 5 mi
5 - 10 mi
10 - 50 mi
50 - 100 mi
100 mi +
8 DIGIT CNPJ
6. Variables
Total Exports
SECEX
Midpoints from Classes
of Exported Amounts
or
Single Exporter
Municipalities
Total Exports
PIA Gross Industrial Revenue
MULTIPLIED
DIVIDED
USD Currency Exchange Rate
13. CIDAQ
Editing and Imputation Quantitative Data Methodology
SAS Macro with Data Editing Purposes at IBGE
Organization and transformation of data
Robust estimation of the mean vector and
the covariance matrix – Log and Box-Cox Transf.
Data: Incomplete or Suspects
15. Total estimated exports from firms in Groups 2 and 3, and overall
total, for 2007–2016.
CIDAQ’s Suspects
16. CONCLUSIONS
Matching PIA and SECEX data was
efficient
Data limitations were not an obstacle
to identifying disparities between PIA
and SECEX data
Identify and Target
Suspicious Survey Records
Estimate potential bias due to
firms that have exported and
did not declare it to PIA
17. REMARKS AND FUTURE WORK
Stop asking Sales for Export in PIA
Improve PIA’s editing of this
important variable
Thank you!
gustavo.costa@ibge.gov.br
pedro-luis.silva@ibge.gov.br