Transcript FAME - OECD

FAME
2nd World Forum on Statistics, Knowledge and Policy
Istanbul 2007
Measuring and Fostering the Progress of Society
[email protected]
www.sungard.com
FAME for the Public Sector
A database designed for economic time series analysis
SunGard’s Forecasting Modelling Analytical Environment (FAME) has
long been viewed as a market leader in providing solutions to the
Public Sector for time series storage and manipulation.
FAME provides unparalleled database management facilities for
storing time series data, The analytical tools give users the power
as well as the flexibility required to improve the handling of many
of the tasks of Central Banks and Statistical agencies.
Six of the world’s 10 largest central banks and eleven of the world’s
20 largest commercial banks rely on FAME for the management of
mission-critical time series data.
2
SunGard
FAME
for Data
the Public
Management
Sector Solutions
A Database designed for economic time series analysis
which provides a number of unique features that are
essential to Public Sector users:









Efficient storage and retrieval of time series for unparalleled speed
Data types specifically designed for storing time series
Objects are stored separately from each other and accessed through their
unique names, enabling models based on object name but with no relational
overhead
Naming conventions that accommodate the GESMES structure
Only the raw data needs to be stored; Time Scale conversion is done adhoc
User-defined attributes allows auxiliary data to be stored with the time
series itself
Client/server architecture can be combined with local databases
Flexible graphing and reporting package
Dynamically evaluated formulas e.g. building a national accounts model that
automatically aggregates different economic levels
3
SunGard
FAME
for Data
the Public
Management
Sector Solutions
The FAME analytical engine offers a wide range of features
including:




Powerful manipulation and analysis of time series data means there are no
constraints on the size of your models or the depth of your history
800 pre-defined analytical functions minimizes the amount of coding needed
which means quick deployment and easy ad-hoc analysis
Unique flexibility in time scale conversion without any coding
Analytical routines can be saved and shared internally as well as with users
in other public sector organization
FAME is used primarily in 5 different areas of Public Sector
institutions:





Economic Research - Fast and easy ad-hoc econometric analysis
Statistics Agencies – Dynamic models for calculating and validating
aggregated measures of economic activity based on low-level raw data
Monetary Policy/Monetary Analysis – Scenario analysis
Banking Supervision – Simple dynamic aggregation of data
Ministry of Finance – Fast ad-hoc analysis of data at different frequencies
4
SunGard
FAME
andData
the OECD
Management Solutions
FAME is heavily used by two directorates at the OECD:
 Economics Directorate
 Statistics Directorate
5
SunGard
FAME
andData
the OECD
Management
– Economics
Solutions
Directorate
 Analytical Database (ADB) Management System
The ADB Management System is the Economic Department’s ETL platform for
managing macro economic data. The application, build using VBA and the
FAME/OLE technology, uses Excel as a front-end for statisticians to define the data
sources and management for time series from Member Countries. These userfriendly instructions are translated into FAME 4GL and extensive metadata are
generated on the fly.
 Forecast Entry System (FE)
Forecast Entry is a multi-country, multi-frequency accounting framework based on
Excel and OLE technology. Country experts enter forecasts of input macro economic
variables into Excel spreadsheets. Multiple, simultaneous write access to FAME
FRDB databases ensures the computation in real time of all accounting identities and
aggregates.
6
SunGard
FAME
andData
the OECD
Management
– Economics
Solutions
Directorate
 FAME Graphics
The FAME 4GL is used to produce graphics for the “Economic Studies” and
“Economic Outlook” publications.
 FAME/Populator
The FAME/Populator is used widely in ECO, in order to bring FAME data into Excel
for calculations or graphing. For example, the Populator is used to produce the
tables in the Statistical Tables Annex to the “Economic Outlook” publication.
 OECD FAME Wizard
The “OECD FAME Wizard” is an Excel-based GUI that uses the FAME/OLE
technology. It allows users to browse and select data from FAME databases.
 Econometric work: regression, modelling
 Reports for internal documents
7
SunGard
FAME
andData
the OECD
Management
– Statistics
Solutions
Directorate
FAME is used in the maintenance of three principal SQL databases;
MEI (Main Economic Indicators), QNA (Quarterly National
Accounts), SNA (Annual National Accounts)

Calculations:
Calculations include:
 Seasonal adjustment
 Aggregations for composite indicators
 Chain linking
The FAME-SQL communication is ensured by an interface written using the
FAME C API

Data Collection
A FAME 4GL program is used to read GESMES/EDI files containing national
accounts data submitted by Member Countries
8
FAME: Object Oriented / Vector Based Database





Highly tuned and optimized
database structure
Faster performance than
relational databases
Unique “Time-Intelligent”
Architecture
Cross sectional screening
attributes
Industrial strength, enterprisewide data access

FAME’s Core Technology consists of 2 components:
1. Database Engine
2. Analytical Engine

FAME 4GL is the native querying language

A full set of APIs (ODBC/JDBC, Java, C/C++, OLE server, web queries) gives
any application fast and efficient read/write access to the data stored in FAME
databases, as well as complete access to the FAME analytical functionality.
This allows for easy and efficient integration between the FAME databases and
statistical packages e.g. TROLL, TRAMO SEATS, EViews and SAS.
9
Tight, concise financial engineering language

FAME Percent Change
convert average
freq business
date 03jan05 to 28feb05
which not missing(yell.close)
graph YELL.CLOSE, YELL.CLOSE[T-1], pct(YELL.close)

SQL Percent Change
SELECT DataPointDate, Close,
(SELECT MAX(DataPointDate)
FROM Yell R1
WHERE R1.InstrumentID = R.InstrumentID
AND R1.DataPointDate < R.DataPointDate)
AS PrevDate,
(SELECT Close FROM Yell R1
WHERE R1.InstrumentID = R.InstrumentID
AND R1.DataPointDate =
(SELECT MAX(DataPointDate)
FROM Yell R2
WHERE R1.InstrumentID = R1.InstrumentID
AND R2.DataPointDate < R.DataPointDate))
AS PrevClose,
(SELECT (((R.Close - R1.Close) / R1.Close) * 100)
FROM Yell R1
WHERE R1.InstrumentID = R.InstrumentID
AND R1.DataPointDate =
(SELECT MAX(DataPointDate)
FROM Yell R2
WHERE R1.InstrumentID = R1.InstrumentID
AND R2.DataPointDate < R.DataPointDate))
AS CalcRead
FROM YELL R
WHERE DataPointDate > (
SELECT MIN(DataPointDate)
FROM Yell R1
WHERE R1.InstrumentID = R.InstrumentID)
AND DataPointDate BETWEEN DATE('2005-01-03')
AND DATE('2005-02-28')
ORDER BY InstrumentID, DataPointDate;
10
FAME Desktop
11