lisbanner The Luxembourg Wealth Study (LWS) Database – Introduction and Selected Demonstrations Markus Jäntti 1,2 1 Swedish Institute for Social Research, Stockholm University 2 Luxembourg Income Study Seventh Winter School on Inequality and Social Welfare Theory January 10, 2012 Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 1 / 54
45
Embed
The Luxembourg Wealth Study (LWS) Database ...dse.univr.it/it/documents/it7/Jantti-lws-ho_slides.pdfIssues in LWS construction Surveys differ: Purpose: some designed to collect wealth
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
lisbanner
The Luxembourg Wealth Study (LWS) Database –Introduction and Selected Demonstrations
Markus Jäntti1,2
1Swedish Institute for Social Research, Stockholm University 2Luxembourg Income Study
Seventh Winter School on Inequality and Social Welfare TheoryJanuary 10, 2012
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 1 / 54
lisbanner
Introduction
Introduction
other economic resources than income important for well-being(consumption, wealth)little information on wealth on a comparable basis across countriesnational datasets have existed for some while, prominent ones(US Survey of Consumer Finances) have served as examples forother countriesthe ECB is running Household Finance and Consumption Network(HFCN) within the Euro systemno comparable definitions of wealth and debt across countries inpublicly available microdata
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 3 / 54
lisbanner
Introduction
LWS Principles
Built up within the Luxembourg Income Study (LIS) – http://www.lisdatacenter.org/our-data/lis-database/following the same model.Establish a network of experts of micro-data on household networth to share accumulated knowledge and best practicesConstruct a comparable database containing wealth variablesbased on existing datasets to enable cross-country comparisonson household net worth, portfolio composition and wealthdistribution (including liquid assets, debts and other holdings)Produce guidelines for data producers – similar to what has beendone for income distribution statistics through LIS with the finalReport of the Canberra Group→ OECD Working Group currentlyoccupied with producing guidelines
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 4 / 54
Austria Survey of Household Financial Wealth 2004Canada Survey of Financial Security 1999Cyprus Survey of Consumer Finances 2002Finland Household Wealth Survey 1994, 1998Germany Socio-Economic Panel Study 2001, 2006Italy Survey of Household Income and Wealth 2002, 2004Japan Keio Household Panel Survey 2003Luxembourg Socio-economic panel (PSELL) 2007Norway Income and Wealth Survey 2002Sweden Wealth Survey 2002United Kingdom British Household Panel Study 2000United States Panel Study of Income Dynamics 2000
Survey of Consumer Finances 2000-2006
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 5 / 54
lisbanner
Introduction
Issues in LWS construction
Surveys differ:Purpose: some designed to collect wealth data (e.g. CA, IT,US-SCF), some supplemented with special modules (e.g. GE, LX,UK, US-PSID)Source: mostly sample surveys, but supplemented withadministrative data in Nordic countriesSampling frame: some over-sample the richUnit of analysis: generally household, but individual in GE andUK, family in US and CANumber of wealth items: from 7 in UK to 30+ in IT, NW, US-SCFPerfect comparability cannot be achievedDefine basic wealth concept
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 6 / 54
lisbanner
Introduction
Ideal LWS variable structure
DemographicsIncome and consumption aggregatesWealth variables -(household-individual-family level)
I Non-financial assetsI Financial AssetsI Liabilities
Behavioral variablesI Bequest motivationI Inheritance expectationsI Motives for savingsI Inter vivos transfersI Risk attitudeI Income and health uncertainty/riskI Expectations (fertility; income support from the state)
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 7 / 54
lisbanner
Introduction
LWS Wealth Variables: Financial assets
Deposit Accounts: Transaction, Savings and CDs DATotal Bonds: Savings and Other Bonds TBStocks STMutual Funds and other investment funds TMLife insurance LIOther investment./ financial assets(non-pen) OFAPension Assets PA
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 8 / 54
lisbanner
Introduction
LWS Wealth Variables: Non-Financial assets
Principal residence PRInvestment real estate IRBusiness Equity BEVehicles VHDurables/Collectibles DR/ CLOther non-financial assets ONF
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 9 / 54
lisbanner
Introduction
LWS Wealth Variables: Liabilities
Total Liabilities TDTotal Home secured debt HSD
Principal residence mortgage MGOther property mortgage OMGOther home secured debt OHSD
Vehicle loans VLTotal Installment debt (inc. credit card bal) ILEducational loans ELOther loans from financial institutions OLInformal debt ID
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 10 / 54
lisbanner
Introduction
Wealth Summary Variables
Risky assets RA =TB+ST+TMTotal assets TA=sum of all assetsHome secured debt HSD=MG+OMG+OHSDNon-housing debt NHD=TD-HSDTotal financial assets TFA1=DA+ST+TB+TMTotal non-financial assets TNF1=PR+IR
introducing Japanese data on wealth and on income (2003)compare with
I Canada 1999I Germany 2001I Sweden 2002I United Kingdom (UK) 2000I United States (US) 2000
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 15 / 54
lisbanner
The distribution of wealth and income
Data definitions
focus now onI gross wealth (not net worth = gross wealth − debt)I gross income (not on disposable income = gross income − direct
taxes)
household is the sharing unitindividuals are the unit of analysisboth wealth and income have been equivalizedoutlying observations can be very influential: also study “shaved”data (inner 98 percent of the marginal distributions of both incomeand wealth)
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 16 / 54
lisbanner
The distribution of wealth and income
Analysis
aspects of the distribution of wealth and its componentsdistributional comparisons (cumulative distribution functions)Lorenz, Generalized Lorenz and Absolute Lorenz orderings ofcountriesdescriptive quantile regressions (10th, 50th [median] and 90thpercentiles as functions of age, education of household head andhousehold structure)examine the (conditional) joint distribution of income and wealth
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 17 / 54
lisbanner
The distribution of wealth and income
Sample sizes and outliers
Canada Germany Japan Sweden UK USPre-shaving 13520 10909 5139 14355 7102 6680Post-shaving 12477 8977 3475 11621 5012 5972Difference 1043 1932 1664 2734 2090 708
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 18 / 54
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 25 / 54
lisbanner
The distribution of wealth and income The distribution and composition of wealth
Boxplot: LWS data
Wealth
Canada
Germany
Japan
Sweden
UK
US
0e+00 2e+05 4e+05 6e+05 8e+05
●
●
●
●
●
●
No
0e+00 2e+05 4e+05 6e+05 8e+05
●
●
●
●
●
●
Yes
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 26 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Distributions compared for wealth: Canada vs US
0e+00 2e+05 4e+05 6e+05 8e+05 1e+06
0.0
0.2
0.4
0.6
0.8
1.0
wealth
Prop
ortio
n <=
x
n:26155245 m:0n:220223.923 m:0
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 27 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Distributions compared for income: Canada vs US
0e+00 2e+05 4e+05 6e+05 8e+05 1e+06
0.0
0.2
0.4
0.6
0.8
1.0
wealth
Prop
ortio
n <=
x
n:26155245 m:0n:220223.923 m:0
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 28 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Distributions compared for wealth: Japan vs US
0e+00 2e+05 4e+05 6e+05 8e+05 1e+06
0.0
0.2
0.4
0.6
0.8
1.0
wealth
Prop
ortio
n <=
x
n:13229 m:0n:220223.923 m:0
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 29 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Distributions compared for income: Japan vs US
0e+00 2e+05 4e+05 6e+05 8e+05 1e+06
0.0
0.2
0.4
0.6
0.8
1.0
wealth
Prop
ortio
n <=
x
n:13229 m:0n:220223.923 m:0
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 30 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Distributions compared for wealth: Germany vs Japan
0e+00 2e+05 4e+05 6e+05 8e+05 1e+06
0.0
0.2
0.4
0.6
0.8
1.0
wealth
Prop
ortio
n <=
x
n:12045115.474 m:0n:13229 m:0
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 31 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Distributions compared for income: Germany vs Japan
0e+00 2e+05 4e+05 6e+05 8e+05 1e+06
0.0
0.2
0.4
0.6
0.8
1.0
wealth
Prop
ortio
n <=
x
n:12045115.474 m:0n:13229 m:0
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 32 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Lorenz curves for wealth: within country
Population share
Lore
nz cu
rve
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.4 0.8
Canada
0.0 0.4 0.8
Germany
0.0 0.4 0.8
Japan
0.0 0.4 0.8
Sweden
0.0 0.4 0.8
UK
0.0 0.4 0.8
US
NoYes
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 33 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Lorenz curves for wealth: country ordering
Population share
Lore
nz cu
rve
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
No
0.0 0.2 0.4 0.6 0.8 1.0
Yes
CanadaGermanyJapanSwedenUKUS
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 34 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Lorenz curves for income: country ordering
Population share
Lore
nz cu
rve
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
No
0.0 0.2 0.4 0.6 0.8 1.0
Yes
CanadaGermanyJapanSwedenUKUS
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 35 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Generalized Lorenz curves for wealth: countryordering
Population share
Lore
nz cu
rve
0
50000
100000
150000
0.0 0.2 0.4 0.6 0.8 1.0
No
0.0 0.2 0.4 0.6 0.8 1.0
Yes
CanadaGermanyJapanSwedenUKUS
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 36 / 54
lisbanner
The distribution of wealth and income Distributional comparisons
Generalized Lorenz curves for income: countryordering
Population share
Lore
nz cu
rve
0
10000
20000
30000
40000
0.0 0.2 0.4 0.6 0.8 1.0
No
0.0 0.2 0.4 0.6 0.8 1.0
Yes
CanadaGermanyJapanSwedenUKUS
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 37 / 54
lisbanner
The distribution of wealth and income Quantile regressions for wealth and gross income
Median regressions: Gross incomeCoefficient estimates and 95% CI
Coefficient estimates and 95% CI
Cova
riate
Age: 30−50
Age: 50−
Educ: High
Educ: Medium
Fam: cpl with kids
Fam: other
Fam: sng, no kids
Fam: sng parent
(Intercept)
−20000
●
●
●
●
●
●
●
●
●
Canada
−20000
●
●
●
●
●
●
●
●
●
Germany
−20000
●
●
●
●
●
●
●
●
●
Japan
−20000
●
●
●
●
●
●
●
●
Sweden
−20000
●
●
●
●
●
●
●
●
●
UK
−20000
●
●
●
●
●
●
●
●
US
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 38 / 54
lisbanner
The distribution of wealth and income Quantile regressions for wealth and gross income
Median regressions: wealthCoefficient estimates and 95% CI
Coefficient estimates and 95% CI
Cova
riate
Age: 30−50
Age: 50−
Educ: High
Educ: Medium
Fam: cpl with kids
Fam: other
Fam: sng, no kids
Fam: sng parent
(Intercept)
−100000 150000
●
●
●
●
●
●
●
●
●
Canada
−100000 150000
●
●
●
●
●
●
●
●
●
Germany
−100000 150000
●
●
●
●
●
●
●
●
●
Japan
−100000 150000
●
●
●
●
●
●
●
●
Sweden
−100000 150000
●
●
●
●
●
●
●
●
●
UK
−100000 150000
●
●
●
●
●
●
●
●
US
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 39 / 54
lisbanner
The joint distribution of wealth and gross income
Why study the joint distribution?
Informs us of the nature of the data.Wealth and income clearly related, but possibly in quite differentways.May reveal interesting differences that could be related toinstitutional and sectoral differences across countries.
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 41 / 54
lisbanner
The joint distribution of wealth and gross income
Income-wealth quartile groups
Proportion / Wealth across the top
Coun
try/ I
ncom
e do
wn th
e sid
e
CanadaGermany
JapanSweden
UKUS
●
●
●
●
●
●
: WealthQG 1 : IncomeQG 1
0.05 0.10 0.15
●
●
●
●
●
●
: WealthQG 2 : IncomeQG 1
●
●
●
●
●
●
: WealthQG 3 : IncomeQG 1
0.05 0.10 0.15
●
●
●
●
●
●
: WealthQG 4 : IncomeQG 1
CanadaGermany
JapanSweden
UKUS
●
●
●
●
●
●
: WealthQG 1 : IncomeQG 2
●
●
●
●
●
●
: WealthQG 2 : IncomeQG 2
●
●
●
●
●
●
: WealthQG 3 : IncomeQG 2
●
●
●
●
●
●
: WealthQG 4 : IncomeQG 2
CanadaGermany
JapanSweden
UKUS
●
●
●
●
●
●
: WealthQG 1 : IncomeQG 3
●
●
●
●
●
●
: WealthQG 2 : IncomeQG 3
●
●
●
●
●
●
: WealthQG 3 : IncomeQG 3
●
●
●
●
●
●
: WealthQG 4 : IncomeQG 3
CanadaGermany
JapanSweden
UKUS
0.05 0.10 0.15
●
●
●
●
●
●
: WealthQG 1 : IncomeQG 4
●
●
●
●
●
●
: WealthQG 2 : IncomeQG 4
0.05 0.10 0.15
●
●
●
●
●
●
: WealthQG 3 : IncomeQG 4
●
●
●
●
●
●
: WealthQG 4 : IncomeQG 4
Markus Jäntti (SOFI/LIS) LIS/Wealth and income January 10, 2012 42 / 54