Jdemetra CESS Budapest20 Oct 2016 Dario Buono Riccardo Gatto Bogomil Kovachev Gian Luigi Mazzi Rosa Ruggeri Cannata Revisions Timeliness vs accuracy Accuracy vs reliability ID: 556834
Download Presentation The PPT/PDF document "A Revision Analysis Plugin for" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
A Revision Analysis Plugin for Jdemetra+
CESS Budapest,20
Oct
2016
Dario
Buono
, Riccardo
Gatto
,
Bogomil
Kovachev
, Gian Luigi
Mazzi
, Rosa
Ruggeri
CannataSlide2
RevisionsTimeliness vs accuracyAccuracy vs reliability
Types of revisions:
Rout
i
ne – periodic
Within the year
Annual benchmark
…
Occasional – irregular
Changes in estimation procedure
Conceptual changes (new classification etc.)
MixedSlide3
Why revision analysisCheck for: the presence of bias
serial correlation
unit roots
heteros
k
edasticity
…
Apply for:
improvement of the revision process
Harmonization of revision policy
…Slide4
RA plugin – input
Releases
Reference
periodsSlide5
Types of revision analysisHorizontal – successive vintages of the same phenomenonDiagonal – 1
st
release vs 2
nd
release, 2
nd
vs 3
rd
etc.
Vertical – all backward revisions at a particular date Slide6
Horizontal revisionSlide7
Vertical revisionr
=
c2
–
c1Slide8
Diagonal revisionc1 – first release vector
c2 – second release vectorSlide9
Descriptive analysisMean, median, ..Ratio of upward revisionsCorrelation of earlier/later estimate with revision
Skewness
Kurtosis
..Slide10
Parametric analysis – regression basedTheil inequality coefficientRegressing the later vintage on the earlier vintage:
Slope and drift
Regressing the revision on the earlier vintage
Regressing the revision on the previous revision
Check orthogonality
Heteroscedasticity of residuals
NormalitySlide11
Parametric analysis – VAR basedUnit root tests
Dickey-Fuller
Philips-
Perron
Autocorrelation tests
Breusch-Godgrey
Ljung
-Box
Cointegration
test
Vector Error Correction ModelSlide12
News vs NoiseCorrelation between revision and earlier estimate (test if revisions are „noise‟)
Correlation between revision and later estimate (test if revisions are „news‟) Slide13
RA plugin statusOriginally developed for Jdemetra 1.2
A port for version 2.* is available at
https
://github.com/bogosmith/jdemetraplus-revtool