The Bank of Italy’s experience with microdata dissemination of households and business surveys
description
Transcript of The Bank of Italy’s experience with microdata dissemination of households and business surveys
1
The Bank of Italy’s experience with microdata dissemination ofhouseholds and business surveys
Ivan Faiella
FIRST EUROPEAN DATA ACCESS FORUMFIRST EUROPEAN DATA ACCESS FORUM, 27-28th March 2012, 27-28th March 2012
2
Bank of Italy’s Surveys
To serve economic analysis, the Bank of Italy conducts periodical sample surveys on households, businesses and selected intermediaries. The main features and results are set out in specific issues of the Supplements or other periodical publications. Sample surveys microdata, and the documentation for their use, can be accessed free of charge.
A key feature is the synergy between data users (economists) and data producers (survey statisticians):1.Questionnaire design;2.Quality control 1: continuous interaction between economists and survey statisticians (i.e. hot topics);3.Quality control 2: researchers working at the local branches are also serving as interviewers for some business surveys (trusts, outlier detection and callbacks);4.Quality control 3: data users are randomly assigned in the households sample.
3
Bank of Italy’s Surveys• Survey on Household Income and WealthSurvey on Household Income and Wealth. This biennial survey
gathers information on the income, savings, wealth and other socio-economic indicators of Italian families.
• Survey of Industrial and Service FirmsSurvey of Industrial and Service Firms. This annual survey, carried out in the first half of the year, gathers information on the current and expected activity of Italian industrial and service firms with 20 or more employees, focusing on variables related to employment, investment and sales.
• Business Outlook Survey of Industrial and Service FirmsBusiness Outlook Survey of Industrial and Service Firms. This annual survey, carried out in the second half of the year, gathers qualitative information on the short-term outlook of Italian industrial and service firms with 20 or more employees.
• Survey on inflation and growth expectationsSurvey on inflation and growth expectations. This quarterly survey gathers information related to the expectations held by industrial and service firms with at least 50 employees on inflation rates, economic growth, and own prices and performance.
• Italian housing market surveyItalian housing market survey. The survey, conducted quarterly, covers a sample of real-estate agents and concerns recent developments in the housing market and the short-term outlook.
• Bank Lending Survey (BLS).Bank Lending Survey (BLS). Results for Italy of the bank lending survey conducted in the Euro area by the Eurosystem.
4
Survey on Household Income and Wealth
(SHIW)
Data distributionMicrodata, documentation and main results accessible free of charge (web site)
FrequencyEvery two years
Time dimensionEnd of survey year
Target variablesIncome, Wealth and other socio-economic indicators
Level of observationHouseholds, individuals
Reporting unitHead of household, individuals
Mode of administrationFace to face (CAPI for 75% of the sample)
Sample size8k households (about 20k individuals)
Sample design2 stage stratified “split panel” sample
Sampling FrameMunicipal registers
Target populationNoninstitutionalized households
Year started 1966 (microdata available from 1977)
Collector A private research company
Sponsor The Bank of Italy
5
Survey on Household Income and Wealth
SHIW data are used …SHIW data are used …to appraise the financial behaviour of households and their attitudes towards payment instruments
to support fiscal policy and evaluate income and wealth distribution (microsimulation)
to explore the relations among socio-economic features of households and their economic preferences
for research purposes, inside and outside the Bank of Italy (Italian an foreign universities, private and public research institutes, think-tanks,…): 800 entries in the survey bibliography
SHIW is part of LIS an EA ECB-coordinated survey
6
Survey of Industrial and Service Firms
(Invind)Business Outlook
Survey of Industrial and Service Firms
(Sondtel)
Data distributionDocumentation and main results accessible free of charge (web site)
Anonymized “scrambled” microdata remotely accessible (BIRD)
FrequencyEvery year
Time dimensionsurvey year
Target variablesEmployees, Turnover, Investments, expectations, credit availability
short-term trends of exports, investments, prices, turnover, demand, profits, employment (Sondtel)
Level of observationFirms and Groups
Reporting unitFirms
Mode of administrationWeb + CATI (some PAPI)
Sample size4k firms (about 1k services)
Sample design1 stage stratified panel sample
Sampling FrameASIA (Istat)
Target populationFirms with 20+ empl.
Year started 1970’s (microdata from 1984) 1993 (Sondtel)
Collector BI local branches
Sponsor The Bank of Italy
7
Survey on Household Income and Wealth
Invind and Sondtel data are used to study …Invind and Sondtel data are used to study …
the financial structure of the borrowing of the firms;
Credit rationing, trade credits;
Investments and demand uncertainty;
Time, labour and wage flexibility in the Italian industrial sector
8
Microdata dissemination
• Survey on Household Income and WealthSurvey on Household Income and Wealth. ANONYMISED MICRODATA DOWNLOADABLE FROM THE WEBSITE
• Survey of Industrial and Service FirmsSurvey of Industrial and Service Firms. • Business Outlook Survey of Industrial and Service FirmsBusiness Outlook Survey of Industrial and Service Firms.
DATA ACCESSIBLE THROUGHT A REMOTE PROCESSING SYSTEM (BIRD)
9
Households microdata
SHIW data distribution SHIW data distribution (BI website)(BI website)……
Two versions: historical and annual archives
Extensive documentation
Complete questionnaire
Data anonymised: all information collected is released with very few exceptions (nut1+ codes, health status, name of financial institutions used by HHs)
JRR weights provided to have a correct estimate of design variance without providing geog. details (Faiella, 2008)
10
Firms microdata
Invind and Sondtel data distribution …Invind and Sondtel data distribution …
In 1974 the interviews for a survey of manufacturing firms were conducted for the first time by the Bank’s branches. Since then, survey microdata have been used by economists at the Research Department for policy use as well as for economic research and the target population of the business surveys has been extended over the years.
Business survey data had never been made available outside the Bank until 2008…because
1. Anonymisation can be unsafe if there are many outliers in the dataset. This tends to happen with business surveys that collect data on firms with very differing sizes.
2. Firms are confident that data are collected for the purpose of economic analysis only and that confidentiality is guaranteed. A mutual climate of trust has been built over the years.
11
Firms microdata
trade-off between security and accessibility…trade-off between security and accessibility…
Find a balance between the confidentiality commitment towards the sample firms and the commitment to transparency and accountability in the domain of applied economic research.
Protecting confidentiality in microdata collected in surveys has two motivations: it is both required and sanctioned by the law and is expected by survey participants….
…. But access to microdata is increasingly asked for research purposes by the scholars’ community.
Possible solutions: 1. data lab, where the researcher has to show up in person at the
place where data are stored: here she can login to the desired dataset while her processing is carefully scrutinised;
2. Remote Access Data Lab (Trewin, 2003): let researchers access the lab remotely via some secure device (“Remote Execution” in EDAF terminology).
12
Bank of Italy’s Remote access to micro Data
BIRD: the Bank of Italy Remote Access Data Lab …BIRD: the Bank of Italy Remote Access Data Lab …
With BIRD (Bank of Italy’s Remote access to micro Data) users carry out their statistical and econometric analyses without having direct access to the micro data; they send an e-mail containing a program written in one of the prescribed languages and the system sends back an e-mail with the results of the calculations.
13
BIRD: additional safeguards …BIRD: additional safeguards …
The measures taken to safeguard data confidentiality of the remotely available datasets are first of all the usual anonymisation measures adopted in the Public Use Files (as in the SHIW).
Preventive treatment of all the quantitative variables: a cut-off upper value is defined and, for top 5 big companies (99.5th percentile), the corresponding values are set equal to it, plus a disturbance term preserving data variability.
Besides that, the Bank of Italy follows four principles to let external users access its business survey data:
1. the researcher’s eligibility is checked (2-3 working days);2. automatic legitimacy checks are performed on the commands
used to access data; 3. automatic and manual checks are performed on the log, the
output and the logic of submissions;4. checks increase as long as disclosure risk increases (e.g. if the
researcher needs to use the database with the original - non-perturbated - values)
Bank of Italy’s Remote access to micro Data
14
BIRD documentation (BI website)…BIRD documentation (BI website)…
Extensive documentation
Complete methodology
Description of the archives
Program “samples”
Original questionnaires
Bank of Italy’s Remote access to micro Data
15
BIRD usage (March 2011)BIRD usage (March 2011)
Bruno, D’Aurizio and Tartaglia (2011)
Bank of Italy’s Remote access to micro Data
16
BIRD usage (March 2011)BIRD usage (March 2011)
Bruno, D’Aurizio and Tartaglia (2011)
Bank of Italy’s Remote access to micro Data
17
What next?
BIRD “evolution”BIRD “evolution”
NOWProviding customised datasets to the users (available now)
NEAR FUTUREUse the platform to access other BI survey data
NEXT FUTURECommon platform (“Data archiveData archive”) among BI, Istat, Bruno Kessler foundation and MIUR. The project aims to create a single access that allows:
1. flexible queries of microdata (research topics, keywords or variables, data wikies);
2. overview of the data available and how they can be accessed;
3. harmonized documentation.
18
The Bank of Italy’s experience with microdata dissemination of
households and business surveysIvan Faiella
ThanksThanks!!