COMPSTAT 2010 19 TH INTERNATIONAL CONFERENCE ON COMPUTATIONAL - - PowerPoint PPT Presentation
COMPSTAT 2010 19 TH INTERNATIONAL CONFERENCE ON COMPUTATIONAL - - PowerPoint PPT Presentation
COMPSTAT 2010 19 TH INTERNATIONAL CONFERENCE ON COMPUTATIONAL STATISTICS Paris France August 22-27 INFLUENCE OF THE CALIBRATION WEIGHTS ON RESULTS OBTAINED FROM CZECH SILC DATA Jitk tka a Bartoov and Vladisla islav v Bna University
OBJECTIVE OF THE CONTRIBUTION
To reveal the connection between values of calibration weights and chosen statistical characteristics
- f the Czech households.
2
Basic ic statist istica ical charact cter erist istics ics of calibrati tion
- n weights
hts. Influe uence nce of calibrati tion
- n weights
ts on the income e distribution ibution in Czech h Republic ic. Influe uence nce of the calibrati tion
- n weights
ts on the measurem emen ent t of monet etar ary y pover erty ty in the Czech h Republic ic.
CONTENT
National variant of european survey EU EU-SIL ILC C as a continuation of former MICROCEN OCENSUS SUS survey. Const nstruc ruction tion of
- f calibration
bration weights hts for sample survey in Czech Republic. Dependence ndence of
- f calibration
bration weights ts on chosen variables. Inf nflue uence nce of
- f the calibration
bration weights ts to the results
- f survey.
3
Table
le 1: Rate e of successfu ssfull lly surveyed ed households holds according ing to the regio ion n of the Czech ch Republic blic
REASONS FOR USE OF CALIBRATION WEIGHTS
4
Source ce: Mikrocensus 2002, EU - SILC 2005 and 2008
Succ Succes essfu fully lly s surv rveyed d flats flats (%)
Region egion 2002 2002 2005 2005 2008 2008 Region egion 2002 2002 2005 2005 2008 2008
Ca Capit ital al Pra Pragu gue 61,9% 51,1% 69,5% Hradec Králové 65,9% 62,9% 81,3% Central Bohemian 67,8% 63,7% 84,4% Pardubice 80,7% 68,1% 85,0% South Bohemian 76,2% 62,9% 87,0% Vysočina 78,7% 73,5% 90,0% Plzeň 77,0% 73,3% 82,3% South Moravian 69,8% 60,0% 83,6% Kar arlov
- vy Var
ary 81,3% 61,1% 83,6% Olo lomouc uc 77,5% 74,4% 84,0% Ústí nad Labem 84,0% 64,6% 84,1% Zlín 78,6% 67,3% 88,1% Liberec 68,8% 64,0% 83,3% Mor
- rav
avia ian–Sil Silesia esian 73,8% 73,9% 86,9%
CONSTRUCTION OF CALIBRATION WEIGHTS
number of permanently occupied flats number of inhabitants per flat number of retirees (both working and not working) number of unemployed number of self employed age of the leading person size groups of municipalities
5
BASIC STATISTICAL CHARACTERISTICS OF CALIBRATION WEIGHTS
Table
ble 2: Basic statistical
tistical charact cteristics eristics of calibration bration weights ts
6
Sour urce ce: EU - SILC 2007
Minimum mum 100.0 Me Mean 417.9 Maxim imum 3475. 5.0 3rd qua quartile ile 493.6 Media ian 369.8 1st qua quartile 294.6
- St. deviati
tion
- n
205.5 Weigh ght sum 404334 3341
DISTRIBUTION OF CALIBRATION WEIGHTS IN DEPENDENCE ON INCOMES
7
Source ce: EU – SILC 2005 – 2007
Fig. . 1: : Kernel el density ity estimat imates es of calibration ration weights hts distri tribu buti tion
- n
DEPENDENCE OF CALIBRATION WEIGHTS ON INCOME OF HOUSEHOLDS
8
Source ce: EU - SILC 2007
- Fig. 2:
: Dependence
nce of calibrat ratio ion n wei weights ts on income of households
DEPENDENCE OF CALIBRATION WEIGHTS ON NET OF INCOME
9
- Fig. 3: Calibration weights of the households with different
number of members
Source: ce: EU - SILC 2007
DEPENDENCE OF CALIBRATION WEIGHTS ON THE HOUSEHOLD SIZE
Fig.
. 3: Ca Calibrat
ratio ion n wei weights hts of the households with different rent number of members
10
Source ce: EU - SILC 2007
DEPENDENCE OF THE CALIBRATION WEIGHTS ON THE SOCIAL GROUP AND MUNICIPALITY
1 1 Low
- wer
er employees ees 2 2 Self employed ed 3 3 Highe her r employees ees 4 4 Retired ired with EA members 5 5 Retired ired without
- ut
EA EA members 6 6 Unem employed ed 7 7 Other ers 8 8 All ll househ ehold
- lds
11
Table le 3: Samp mple le sizes s and means ns of calibra rati tion
- n weight
ghts of differen erent t social l group ups
Sour urce: ce: EU - SILC 2007
So Social g ial gro roup of
- f t
the he head head of
- f h
hou
- usehold
ehold
1 2 3 4 5 5 6
7 8
sa samp mple le size size
2385 2385 802 2297 418 3423 3423 258 258 110 9675 9675
mean mean of
- f w
weig eights hts
420.4 630.1 433.6 429.2 332.3 731.1 380.4 417.9
Table le 4: Samp mple le sizes s and means ns of calibra rati tion
- n weight
ghts of differen erent t mun unici cipali paliti ties es
Type e of
- f munic
icip ipalit ality
cap apital tow ital town cou
- unty
ty s seat eat urban rban villag villages es villag villages
sa samp mple le size size
864 864 1423 3952 3952 3436 3436
mean mean of
- f w
weig eights hts
617 617.3 .3 446 446.4 .4 395.6 38 381.7
DEPENDENCE OF THE CALIBRATION WEIGHTS ON THE SOCIAL GROUP
Fig.
. 4: : Calib
ibration ation wei weights ts of the households lds from different rent social groups ps
12
Source: ce: EU - SILC 2007
DEPENDENCE OF THE CALIBRATION WEIGHTS ON THE TYPE OF MUNICIPALITY
Fig.
. 5: Calib
ibration ation wei weights ts of the households lds from different rent types
- f municipal
icipalities ities
13
Source: ce: EU - SILC 2007
INFLUENCE OF CALIBRATION WEIGHTS ON THE ESTIMATES OF INCOME CHARACTERISTICS
Table
le 4: Income
- me charact
cteris eristics tics of the househ useholds
- lds from
m different erent social group ups
Table
le 5: Income
- me charact
cteris eristics tics of the househ useholds
- lds from
m different erent mun unicip ipaliti alities es
14
Source: ce: EU - SILC 2007
Differe Difference betw e betwee een weig eighted an hted and u d unweig eighted hted characteris haracteristic tics
So Social g ial gro roup of
- f t
the he head head of
- f h
hou
- usehold
ehold
1 2 3 4 5 5 6 7 8 mea ean (C (CZK ZK)
10628 17486 17486 10878 14424 14424 2374 3243 1181 1181 22131 22131
med median n (CZK) (CZK)
8727 14198 14198 8658 14237 14237 10128 9260 -1288 1288 20582 20582
standard dev eviatio iation n (CZK) (CZK)
5294 19872 19872 3135 9512 9512
- 143
143 2289 -4574 21380 21380
Differe Difference betw e betwee een weig eighted an hted and u d unweig eighted hted characteris haracteristic tics
Type e of
- f munic
icip ipalit ality
cap apital tow ital town cou
- unty
ty s seat eat urban rban villag villages es villag villages
mea ean (C (CZK ZK)
17387 17387 20973 20973 18278 21102 21102
med median n (CZK) (CZK)
26499 26499 21500 21500 16152 16152 18862
standard dev eviatio iation n (CZK) (CZK)
13035 9799 9799 23437 23437 20544 20544
1 1 – low
- wer
r employ
- yees,
2 2 – se self em employed, , 3 – higher r employ
- yees,
, 4 – ret retired red wit with ec econom nomical cally act active ve members rs, 5 5 – ret retired red wit without
- ut ec
econo
- nomical
cally a act ctive m ve members rs, , 6 6 – un unemploy
- yed,
, 7 – ot
- thers
rs, 8 – al all house
- useholds
DIFFERENCE BETWEEN WEIGHTED AND UNWEIGHTED CHARACTERISTICS
15
- 5000
5000 10000 15000 20000 25000
Difference of means (CZK) Difference of medians (CZK) Difference of standard deviations (CZK)
DEFINITIONS OF THE CONSUMING UNIT
H – total income per household SJ SJ – equivale alent nt scale
- f OECD
D EJ EJ – equivalent alent scale
- f EU
R – income per representative
16
- ch
ch – num umbe ber r of childre ren n betw etwee een n 0 and 13 13
- op
- p – num
umber ber of ot
- ther
her childre ren n and membe mbers s (except ept „he head“ of househ usehold
- ld)
THRESHOLD OF MONETARY POVERTY FOR DIFFERENT TYPES OF CONSUMING UNITS
Table 6: Influence
luence of calibra rati tion
- n weight
ghts s on the thresh shold
- ld of mone
neta tary y pover erty ty for differe erent nt types es of consuming ming un units
Sour urce ce: Mikrocensus 2002, EU-SILC 2005 – 2007
Ye Year ar
Type of the e of the cons
- nsum
uming ing un unit it
Threshold hreshold of m
- f monet
- netary p
pov
- verty (CZ
(CZK)
weighted eighted esti timat ate un unweighted eighted esti timat ate Dif ifferen erence betwe tween weighted eighted and and un unweighted eighted
2002 2002 household
116909 114554 2355
representative
52000 53522
- 1522
2005 2005 household
132549 123246 9303
repr epres esentativ entative
58200 58230
- 30
30
def
- ef. E
. EU
78786 78786 76500 76500 2286 2286
def
- ef. OE
. OECD
68223 68223 67199 67199 1024 1024
2006 2006 hou
- useh
ehold
- ld
139743 128088 11655 11655
representative
60912 60384 528
def
- ef. E
. EU
83052 83052 79568 79568 3484 3484
def
- ef. OE
. OECD
72000 72000 69926 69926 2074 2074
2007 2007 hou
- useh
ehold
- ld
152069 139718 12351 12351
representative
65850 65246 604
def
- ef. E
. EU
89611 89611 86129 86129 3482 3482
def
- ef. OE
. OECD
89611 89611 86129 86129 3482 3482
THRESHOLD OF MONETARY POVERTY FOR DIFFERENT TYPES OF CONSUMING UNITS
18
20000 40000 60000 80000 100000 120000 140000 160000 household representative household representative
- def. EU
- def. OECD
household representative
- def. EU
- def. OECD
household representative
- def. EU
- def. OECD
2002 2005 2006 2007
weighted unweighted
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD
Income per consuming unit (EU) Lower employees
19
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD Income per consuming unit (EU) Self-employed
20
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD
Income per consuming unit (EU) Higher employees
21
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD
Income per consuming unit (EU) Retired with EA members
22
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD
Income per consuming unit (EU) Retired without EA members
23
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD
Income per consuming unit (EU) Unemployed
24
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD
Income per consuming unit (EU) Other households
25
INFLUENCE OF CALIBRATION WEIGHTS ON THE INCOME DISTRIBUTION AND POVERTY THRESHOLD
Income per consuming unit (EU) All households
26
RATE OF HOUSEHOLDS UNDER THE THRESHOLD FOR DIFFERENT TYPES OF CONSUMING UNITS
Table 7: Influen
ence ce of calibra rati tion
- n weight
ghts s on the rate of househ useholds
- lds
un under der the thres esho hold ld of mone netar tary y pover erty ty
Sour urce: ce: Mikrocensus 2002, EU-SILC 2005 – 2007
Ye Year ar
Type of the e of the cons
- nsum
uming ing un unit it
- Freq. u
- eq. under
der risk isk-of
- f-pov
- ver
erty th thresh eshold
- ld
Pear earson son in in. . test test
weighted es eighted esti timat ate un unweighted e eighted esti timat ate
statistics istics χ2 p-value lue
ab absol
- lut
ute rel elat ative ab absol
- lut
ute rel elat ative
2002 2002
household 1833 22.99 % 1782 22.35 % 0.894304 0.344314 representative 672 8.43 % 757 9.49 % 5.423770 0.019864
2005 2005
household 1095 25.17 % 1012 23.26 % 4.210827 0.040167 representative 439 10.09 % 439 10.09 % 0.001267 0.971608 def.
- ef. EU
331 331 7.61 % 291 291 6.69 % 2.633580 0.104626 def.
- ef. OECD
CD 176 176 4.05 % 167 167 3.84 % 0.194245 0.6594065
2006 2006
household 1878 25.10 % 1691 22.60 % 12.729007 0.000360 representative 753 10.06 % 733 9.80 % 0.269714 0.603523 def.
- ef. EU
570 570 7.62 % 469 469 6.27 % 10.342669 0.001300 def.
- ef. OECD
CD 297 297 3.97 % 253 253 3.38 % 3.490078 0.061738
2007 2007
household 2409 24.90 % 2193 22.67 % 13.178869 0.000283 representative 858 8.87 % 832 8.60 % 0.4052132 0.524409 def.
- ef. EU
697 697 7.20 % 566 566 5.85 % 14.315212 0.000155 def.
- ef. OECD
CD 363 363 3.75 % 324 324 3.35 % 2.179265 0.139881
RATE OF HOUSEHOLDS UNDER THE THRESHOLD FOR DIFFERENT TYPES OF CONSUMING UNITS
28
0% 5% 10% 15% 20% 25% 30% household representative household representative
- def. EU
- def. OECD
household representative
- def. EU
- def. OECD
household representative
- def. EU
- def. OECD
2002 2005 2006 2007
weighted (%) unweighted (%)
CONCLUSIONS
Sam Sample le su survey Mi Mikr krocensu
- census and
and Czec Czech EU EU-SILC sur survey pr provid vides es an an inf nfor
- rmat
mation
- n
ab about
- ut
incom ncomes es and and
- th
ther so social cial and and dem emog
- grap
raphi hic charact characteri eristi tics cs of
- f Czec
Czech ho house useho holds
- lds. Th
The data ata files les cont
- ntain cali
alibrat bration
- n
wei eigh ghts th that at can can signi significan cantl tly influence fluence the the resul esults of
- f real
realized zed an analyses
- alyses. It
appears that the role of calibration changes with number of household members, grows with the growing incomes, etc. The paper focuses on the strength of influence of the calibration weight on the risk of monetary poverty in the Czech Republic. We
We had had sh shown wn that hat the the bias bias of
- f
results results occurred
- ccurred in
in all all cases cases (u (usuall ually highe higher valu alues) es) and and in in more
- re th
than an hal half
- f
- f cases
cases thi his cha hange nge was as statisti tatisticall ally signi significan cant (o (on the he 5% le level el). Thus hus, the he un unwei eight ghted ed resu esult lts ar are sli light htly dist storted ed; on
- nly
ly in in hal half of
- f cases
cases the the bias bias is is stati atisti tical cally ly signi nifi ficant cant.
In order to create a complex insight on the problem of biasing the results of measuring the relative poverty by calibration weights, our analyses were based
- n the study of different definitions of consuming unit which handles the
monetary poverty from different perspectives. We
We sh shown wn th that the the choi choice ce of
- f
scale ale can can suppres press or
- r emp
mphasize hasize the the influen luence ce of
- f calibrat
bration
- n weigh
ghts. An An impor
- rta
tant nt out
- utcome
come is is the he inf nflu luence ence of
- f consu
- nsuming
ming un unit defi efini nition on
- n the
the ri risk sk of
- f poverty
ty of
- f Czec
Czech ho househ usehol
- lds. And
And th there refor
- re the
the suitab suitable le definition tion of
- f
consuming nsuming unit plays ys the the key role in in identi ntifyi ying ng of
- f relat
ative pover erty ty in in soci
- ciety
ty.
REFERENCES
BARTOŠOV OVÁ, , J. (2009): ): Analysis ysis and Modell llin ing of Financ ncial ial Power of Czech Households eholds. Aplimat, Journal of Appl. Math. 2(3), STU Bratislava, 31-36. BARTOŠOV OVÁ, , J. and BÍNA, , V. (2009): ): Modell lling ng of income
- me distribution
ribution of Czech househ useholds
- lds in years 1996 - 2005
- 2005. Acta Oeconomica Pragensia 17(4), Oeconomica,
Prague, 3-18. BARTOŠOV OVÁ, , J. and BÍNA, , V. (2009): ): Financi ncial al Power of the Czech Househol eholds
- ds. In:
EURISBIS09: Book of Abstracts. TILAPIA, Cagliari, 201-202. BÍNA, , V. (2009): ): The role of calibrat ration
- n weights
s in SILC data. In: Finanční potenciál domácností '09 (Proceedings of workshop of GAČR 402/09/0515). University of Economics in Prague, CDROM. NICODEM DEMO, O, C., LONGFOR ORD, D, N.,T. (2009): ): A sensitivity vity analysi ysis s of poverty y defini nitions
- ns
used d with EU-SILC. In: Finanční potenciál domácností '09 (Proceedings of workshop of GACR 402/09/0515). Univ. of Economics in Prague, CDROM. STANKOVIČOVÁ, I., BARTOŠOVÁ, J. (2009): Príspevok k analýze subjektívnej chudob
- by
y v SR a ČR. Forum Statisticum Slovacum 5(3), SŠDS, Bratislava, CDROM.
ACKNOWLEDGEMENT
The paper was supported by Czech ch Science ence Foundation ndation: Project GAČR 402/09/0515 „Anal nalysis ysis and Modelling elling of Financi ancial al Power er
- f Czech
ch and Slovak ak Ho Household seholds“
Thank you for your attention and wish a nice day!
Happy to answer your questions in writing at: bartoso sov@fm. fm.vs vse.cz e.cz
31