Two New Exploratory Ideas for Cross National Wealth Analyses

Transcription

1 Two New Exploratory Ideas for Cross National Wealth Analyses A. LWS and LIS Income Variables, Income from Wealth and Income from Labor, an Exploratory Analysis B. Presenting Joint Distribution of Wealth and Income Markus Jäntti, Emilia Niskanen, Eva Sierminska and Timothy Smeeding ECINEC, Berlin, July 2007

2 General Introduction to LWS See home page for LWS 1. Data available November Here are some quick ideas for papers linking LWS and LIS

3 Part A: Income From Wealth and Income From Labor: Niskanen and Smeeding I. LWS INCOME CONCEPTS II. LWS AND INCOME NET WORTH III. NEXT STEPS

4 LWS Income Variables LIS concepts Canberra Group: Wealth and In-kind Income Simplicity Coverage See:

5 LWS NAME WAGE SELF CPRI CPRI1 CPRI2 CPRI3 CPRI4 CPRI5 OCCPEN OCPEN1 OCPENO PUBPEN PUPEN1 PUPEN2 PUPENO SOCIBEN MNSBEN NRCBEN PRIVTRA OTHCINC CONTRIB INCTAX WLTHTAX INTPD INTPDMG INTPDOL PTPD GAIN NCPRI GIFT LUMP LWS Household variables VARIABLE LABEL Wages and Salaries Self Employment Income Cash Property Income Cash property income -interests and dividends Cash property income -rents Cash property income -private savings plans Cash property income -royalties Cash property income -other Occupational and other pensions Occupational pensions Other pensions State old-age and survivors benefits Universal old-age and survivors pensions Employment related old-age and survirors pensions Other state old-age and survivors pensions Social insurance transfers (excl. pubpen) Social assistance cash benefits Near cash benefits Total private regular transfers Other cash income Total mandatory payroll taxes Income tax Wealth/Property taxes Interest paid Interest paid on mortgages Interest paid on other loans Private regular transfers paid Capital Gains Non-cash property income, Imputed Rent Non-regular gifts One-time lump sum income

6 LWS Summary Income Variables EARNINGW Earnings (= WAGE + SELF) FIW MIW TRANS GIW LIS_DPI DPIW DPIT Factor Income (= EARNINGW + CPRI) Market Income (= FIW + OCCPEN) Transfer Income (= PUBPEN + SOCIBEN + MNSBEN + NRCBEN + PRIVTRA) Gross Income (= MIW +TRANS + OTHCINC) Disposable income, LIS standards (= GIW - CONTRIB - INCTAX) LWS Disposable Income (= DPIW - WLTHTAX - INTPD - PTPD + NCPRI) Additional Disposable Income (= DPIW + GAIN)

7 AT 2004 CA 1999 CY 2002 FIN 1998 GE 2002 IT 2002 VARIABLE NAME GROSS NET NO 2002 SE 2002 US 2001 SCF UK 2000 WAGE I YES YES YES YES YES YES YES YES YES YES SELF I YES YES YES YES YES YES YES YES YES YES =EARNINGW I YES YES YES YES n YES YES YES YES YES CPRI = I YES YES YES YES YES YES YES YES YES YES cpri1 + I I YES YES YES YES YES YES YES YES YES cpri2 + I I YES I YES YES I YES YES YES YES cpri3 + I I I I YES YES YES YES YES YES YES cpri4 + I I YES I I I I I YES I YES cpri5 + I I I I I I YES I I I YES =FIW I YES YES YES YES n YES YES YES YES YES OCCPEN = I YES YES I YES YES YES YES YES YES YES ocpen1 + I I I I YES I YES YES I YES I ocpeno + I I I I I YES I YES I I I =MIW I YES YES YES YES n YES YES YES YES YES PUBPEN = I I YES YES YES YES YES YES YES YES YES pupen1 + I I I YES I I I YES I I I pupen2 + I I I YES I YES I YES I YES I pupeno + I I I I I YES I YES I YES I SOCIBEN + I I YES YES YES YES YES YES YES YES YES MNSBEN + I I I YES YES YES YES YES YES YES YES NRCBEN + I I I I YES I YES YES YES YES I PRIVTRA + I YES YES I YES YES YES YES YES YES YES =TRANS I YES YES YES YES YES YES YES YES YES YES OTHCINC + I YES I YES I YES YES I YES YES YES =GIW I YES YES YES YES YES YES YES YES YES YES CONTRIB - I I I I YES* I YES YES YES* YES* YES* INCTAX - I YES I YES YES* I YES YES YES* YES* YES* =LIS_ DPI YES YES I YES YES YES YES YES YES YES YES NCPRI + I I I YES YES YES YES I I I YES WLTHTAX - I I I I I I YES YES YES I YES PTPD - I I YES I YES YES YES YES YES YES YES INTPD = I I I YES I YES YES YES I I I intpdmg - I I I YES I I I I I I I intpdol - I I I I I I I I I I I =DPIW I I I YES YES YES YES YES YES YES YES GAIN + I I YES I I YES YES YES YES I I =DPIT I I I I I YES YES YES YES I I US 2001 PSID

8 II. INCOME FROM LABOR VS. INCOME FROM WEALTH

9 1. Incomplete LWS income measures: Simplification of Definitions Table 1A Components of Income and Income From Property/Wealth from LWS Other Components of Country (Year) Gross Income Property Income IDEAL: GIW (GIW includes) (CPRI) +GAIN +NCPRI -INTPD Canada (CA, 1998) GIW (CPRI) Finland (FI, 1998) GIW (CPRI) NCPRI INTPD Germany (GE, 2001) GIW (CPRI) NCPRI Italy (Net) (IT(N), 2002) GIW (CPRI) GAIN NCPRI INTPD Sweden (SE, 2002) GIW (CPRI) GAIN INTPD United States (US, 2000) GIW (CPRI) GAIN IDEAL AGI =GIW -CPRI -GAIN -NCPRI +INTPD Terms: GIW = LWS Gross Cash Income, comparable for all countries (save Italy which is net income) (CPRI) = Cash Property Income (included in all measures of GIW) GAIN = Realized Capital Gains NCPRI = Non-cash Property Income (Imputed Rent) INTPD = Interest Paid AGI = Gross Income Net of Capital Income, (Including Adjustment for Self Employment Income, see text) Source: Niskanen (2006) plus authors concepts

10 2. More complete: Full Income or Income Net Worth A) Haig-Simons Income: Consumption Plus Change Net Worth B) Income types: Labor Income Self-employment Income Income from Capital Taxes and Transfers

11 3. What to do about it? GIW AGI = GIW minus reported capital income (CPRI) FGI1 = AGI plus 2 percent real return FGI2 = AGI plus 4 percent real return

12 Table 1B. Rates of Return for FGI1 and FGI2 Country Income Year CPI CPI plus 2 CPI plus 4 Canada Finland Germany Italy Sweden US SCF Sources: OECD (CPI); Burtless 2007 (rates of return)

13 4. Methods Countries: Canada 1998 Finland 1998 Germany 2001 Italy 2002 (NET) Sweden 2002 The U.S Net Worth 1 vs. New Worth 2 Top codes (not yet)

14 5. Preliminary Test Results

15 A. Mean Table 2A. Income from Wealth: Effects on Means and Medians Income Measurements All Figures from 2002 (PPP dollars) Net Worth 1 Net Worth 2 GIW 1 AGI 2 FGI1* 3 FGI2* 4 FGI1 5 FGI2 6 FGI2 7 /GIW CA $44,179 $41,923 $44,663 $46,489 $45,380 $47, FI $36,131 $32,977 $35,618 $37,172 NA NA 1.03 GE $41,155 $37,951 $42,166 $44,508 $42,930 $45, IT(N) 6 $28,736 $25,748 $35,141 $38,975 $36,574 $40, SE $36,903 $35,239 $37,872 $39,068 $38,263 $39, US $65,365 $59,688 $71,390 $75,970 $75,495 $81, B. Median CA $35,264 NA 9 $35,909 $37,510 $36,188 $37, FI $28,931 $27,626 $30,057 $31,406 NA NA 1.09 GE $32,154 $30,002 $33,647 $35,265 $33,848 $35, IT(N) 8 $23,132 $21,126 $28,506 $30,943 $28,829 $31, SE $29,623 $28,607 $30,421 $31,235 $30,734 $31, US $39,613 $37,332 $42,923 $45,219 $43,689 $45, Notes: 1. GIW is LWS comparable gross income. 2. AGI is GWI minus all types of property income. 3. FGI1* is AGI plus return on net worth of 2 percent. 4. FGI2* is AGI plus return on net worth of 4 percent. 5. FGI1 is AGI plus return on net worth of 2 percent, business equity included. 6. FGI2 is AGI plus return on net worth of 2 percent, business equity included. 7. Ratio of FGI2 to GIW except for Finland where FGI2* is used 8. Italian values re: net of Tax. 9. We cannot define this term

16 Table 2B. Percentage of Income Components in FGI2 CA FI 1 GE SE IT(N) US Labor Income Capital Income Other Income Notes: 1. FGI2* is used for Finland

17 Table 3. Income from Wealth: The Effect on Inequality: Gini Coefficients for 4 Different Income Aggregates GIW 1 AGI 2 FGI1 3 FGI2 4 FGI2 5 /GIW CA FI GE IT(N) SE US Notes: 1. GIW is LWS comparable gross income. 2. AGI is GWI minus all types of property income. 3. FGI1 is AGI plus return on net worth of 2 percent with business income 4. FGI2 is AGI plus return on net worth of 4 percent with business income 5. Change in Gini: FGI2-GIW. 6. Italian values re: net of Tax 7. FGI1* and FGI2* are used for Finland

18 A. Top End: P95/P50 Table 4. Income from Wealth: The Effect on Inequality by Percentile GIW 1 AGI 2 FGI1 1 FGI2 2 FGI 2 /GIW CA FI GE IT(N) SE US B. Top End: P90/P50 CA FI GE IT(N) SE US Notes: 1. GIW is LWS comparable gross income. 2. AGI is GWI minus all types of property income. 3. FGI1 is AGI plus return on net worth of 2 percent, includes business income 4. FGI2 is AGI plus return on net worth of 4 percent, includes business income 5. Percent change is FGI 2 /GIW 6. Italian values re: net of tax 7. FGI1* and FGI2* are used for Finland

19 Table 4. Income from Wealth: The Effect on Inequality by Percentile (con t) C. Bottom End: P10/P50 GIW 1 AGI 2 FGI1 1 FGI2 2 FGI 2 /GW CA FI GE IT(N) SE US D. Decile Ratio:P90/P10 CA FI GE IT(N) SE US Notes: 1. GIW is LWS comparable gross income. 2. AGI is GWI minus all types of property income. 3. FGI1 is AGI plus return on net worth of 2 percent, includes business income 4. FGI2 is AGI plus return on net worth of 4 percent, includes business income 5. Percent change is FGI 2 /GIW 6. Italian values re: net of tax 7. FGI1* and FGI2* are used for Finland

20 6. Discussion Finland vs. The U.S. Quality of Data Quality of Assumptions Effect on Rank Ordering of Nations, e.g. 90/10 or Gini Ratios, GIW vs FGI2

21 III. NEXT STEPS 1. EXPLORE THE DIFFERENCES FURTHER BY AGE, FAMILY TYPE 2. IMPROVE VALUE OF INCOME FROM WEALTH COMPONENTS 3. TOP CODE WEALTH AT 99 % PERCENTILE

22 IV. COMMENTS AND DISCUSSION, PLEASE? And Second Idea: Joint Distribution of Income and Wealth

23 Part B: Presenting joint distributions of income and wealth Markus Jäntti 1 Eva Sierminska 2 Timothy M Smeeding 3 1 Luxembourg Income Study 2 Luxembourg Wealth Study 3 Syracuse University July 10, 2007 Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

24 Outline 1 Introduction 2 Data 3 Descriptive results 4 The multivariate distribution of wealth 5 Regression results 6 Concluding comments Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

25 Introduction Outline 1 Introduction 2 Data 3 Descriptive results 4 The multivariate distribution of wealth 5 Regression results 6 Concluding comments Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

26 Introduction Introduction Examine the joint distribution of income and wealth in selected countries. Exploratory analysis to compare nature of association across countries. Use common definitions (limits number of countries) and comparative units. Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

27 Introduction Why study the joint distribution? Informs us of the nature of the data. Wealth and income clearly related, but possibly in quite different ways. May reveal interesting differences that could be related to institutional and sectoral differences across countries. Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

28 Data Outline 1 Introduction 2 Data 3 Descriptive results 4 The multivariate distribution of wealth 5 Regression results 6 Concluding comments Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

29 Data LWS datasets analysed Canada (1999) Germany (GSOEP 2002) Italy (SHIW 2002) Sweden (2002) United States (SCF 2001) Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

30 Data Income and wealth variables Variable Symbol LWS definition Disposable income dispincome lis dpi = grossincome taxes Gross income grossincome giw Taxes taxes inctax + contrib Net worth networth nw2 = wealth debt Wealth wealth tfa1 + tnf2 Debt debt td Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

31 Data Issues in analysis Choose comparable but inclusive concept of net worth (nw2). Focus on non-outlier observations: retain the observations where both net worth and disposable income is within the inner 98 percent of the values. Express all money values in terms of international US dollars in 2002 prices use domestic deflator and PPP for actual individual consumption. Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

32 Data Sample sizes and outliers Canada Germany Italy Sweden United States Pre-shaving Post-shaving Difference Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

33 Data Pre-shave percentiles Percentiles Country Income Canada Germany Italy Sweden United States Wealth Canada Germany Italy Sweden United States Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

34 Descriptive results Outline 1 Introduction 2 Data 3 Descriptive results 4 The multivariate distribution of wealth 5 Regression results 6 Concluding comments Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

35 Descriptive results Descriptive results Show marginal distributions. Study the bivariate distribution using: Quartile-group cross-tabulations. Bands of income related to median (< 50%, 50% 100%, 100% 150%, > 150% of median). Bivariate density estimates. Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

36 Descriptive results Density estimates: disposable income Canada Germany Italy Sweden United States 5e 05 4e 05 Density 3e 05 2e 05 1e 05 0e Disposable income Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

37 Descriptive results Density estimates: net worth 0e+00 6e+05 0e+00 6e+05 Canada Germany Italy Sweden United States 1.5e e 05 Density 5.0e e+00 0e+00 6e+05 0e+00 6e+05 Net worth 0e+00 6e+05 Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

38 The multivariate distribution of wealth Outline 1 Introduction 2 Data 3 Descriptive results 4 The multivariate distribution of wealth 5 Regression results 6 Concluding comments Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

39 The multivariate distribution of wealth Income-wealth quartile groups Wealth across the top / Income down the side Country/ Income down the side United States Sweden Italy Germany Canada United States Sweden Italy Germany Canada United States Sweden Italy Germany Canada United States Sweden Italy Germany Canada IncomeQG : 1 IncomeQG : 1 IncomeQG : 1 IncomeQG : 1 WealthQG : 1 WealthQG : 2 WealthQG : 3 WealthQG : 4 : IncomeQG : 2 IncomeQG : 2 IncomeQG : 2 IncomeQG 2 WealthQG : 1 WealthQG : 2 WealthQG : 3 WealthQG : 4 IncomeQG : 3 IncomeQG : 3 IncomeQG : 3 IncomeQG : 3 WealthQG : 1 WealthQG : 2 WealthQG : 3 WealthQG : 4 IncomeQG : 4 IncomeQG : 4 IncomeQG : 4 IncomeQG : 4 WealthQG : 1 WealthQG : 2 WealthQG : 3 WealthQG : Proportion / Wealth across the top Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

40 The multivariate distribution of wealth Income-wealth quartile groups Canada Germany Proportion Italy Sweden United States Income wealth cell Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

41 The multivariate distribution of wealth Income-wealth median-based groups Wealth across the top / Income down the side Country / Income down the side meb : 0.5 memeb : 0.5 memeb : 0.5 memeb : 0.5 me althb : 0.5 med thb : me thb : me althb : 1.5 media United States Sweden Italy Germany Canada meb : meb : meb : meb : m althb : 0.5 med thb : me thb : me althb : 1.5 media United States Sweden Italy Germany Canada meb : meb : meb : meb : m althb : 0.5 med thb : me thb : me althb : 1.5 media United States Sweden Italy Germany Canada meb : 1.5 medi meb : 1.5 medi meb : 1.5 medi meb : 1.5 medi althb : 0.5 med thb : me thb : me althb : 1.5 media United States Sweden Italy Germany Canada Proportion / Wealth across the top Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

42 The multivariate distribution of wealth Joint distribution of income and net worth Non-parametric density estimates: Canada Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

43 The multivariate distribution of wealth Joint distribution of income and net worth Non-parametric density estimates: Germany Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

44 The multivariate distribution of wealth Joint distribution of income and wealth Canada and Germany relative to the US Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

45 The multivariate distribution of wealth Joint distribution of income and wealth Italy and Sweden relative to the US Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

46 Regression results Outline 1 Introduction 2 Data 3 Descriptive results 4 The multivariate distribution of wealth 5 Regression results 6 Concluding comments Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

47 Regression results Bivariate regressions of income and wealth Simple bivariate regressions relate disposable income and net worth to selected covariates: Report: dispincome = f dpi (age, education, fam. struct) + ɛ dpi networth = f nw (age, education, fam. struct) + ɛ nw [ ] ([ ] [ ]) (1) ɛdpi 0 σ 2 N, dpi ρσ dpi σ nw 0 ɛ nw Share of variation accounted for by f ( ) Regression coefficients Standard deviations σ dpi, σ nw Correlation ρ σ 2 nw Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

48 Regression results Regression results: share of variance explained DisposableIncome NetWorth United States Sweden Italy Germany Canada R 2 in regression Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

49 Regression results Regressions results: disposable income Coefficient estimates and confidence intervals Canada Germany Italy SwedenUnited State Intercept Fam: sng, no kids Fam: sng parent Fam: other Covariate Fam: cpl, no kids Educ: Unknown Educ: Medium Educ: High Age: 70 Age: Age: Coefficient estimates Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

50 Regression results Regressions results: net worth Canada 1e+05 3e+05 Germany Italy 1e+05 3e+05 SwedenUnited State Intercept Fam: sng, no kids Fam: sng parent Fam: other Covariate Fam: cpl, no kids Educ: Unknown Educ: Medium Educ: High Age: 70 Age: Age: e+05 3e+05 1e+05 3e+05 Coefficient estimates 1e+05 3e+05 Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

51 Regression results Residual standard deviation of disposable income Regression residuals United States Sweden Country Italy Germany Canada Residual standard deviation of disposable income Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

52 Regression results Residual standard deviation of net worth Regression residuals United States Sweden Country Italy Germany Canada Residual standard deviation of net worth Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

53 Regression results Residual correlation of disposable income and net worth Regression residuals United States Sweden Country Italy Germany Canada Residual correlation of disposable income and net worth Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

54 Concluding comments Outline 1 Introduction 2 Data 3 Descriptive results 4 The multivariate distribution of wealth 5 Regression results 6 Concluding comments Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33

55 Concluding comments Concluding comments Substantial differences in the range of variation, even after shaving the data. US has much greater variation that the rest. The association between income and wealth also greatest in the United States. Jäntti, Sierminska, Smeeding (LIS, LWS) Income and wealth July 10, / 33