Antoine Bozio. Master APE and PPD Paris October Paris School of Economics (PSE) École des hautes études en sciences sociales (EHESS)

Transcription

1 Top Lecture 7: Labour Paris School of Economics (PSE) École des hautes études en sciences sociales (EHESS) Master APE and PPD Paris October / 141

2 Top Outline of the lecture 6 I. Labour : institutions 1 History 2 Typology 3 Fiscal facts II. Incidence 1 Theory 2 Empirical estimates III. Labour supply responses 1 Theory 2 Traditional estimates 3 Estimates from experiments 4 High-frequency studies 5 Macro vs micro estimates 2 / 141

3 Top Outline of the lecture 7 IV. Elasticity of taxable income 1 Theory V. Optimal tax and transfer system Extensive margin 1 Top 2 Low income redistribution 3 / 141

4 Elasticity of taxable income Top 1 ual framework International top incomes 6 4 / 141

5 Elasticity of taxable income Top Limits of traditional labour supply Quantitative measures of labour supply (hours and employment) are not the only behavioural responses to taxation Deadweight loss of taxation should depend on all behavioural responses Other behavioural margins 1) Effort on the job 2) Career choice 3) Form and timing of compensation 4) Tax avoidance (legal shifting of income) 5) Tax evasion (illegal under-reporting of income) 5 / 141

6 Elasticity of taxable income Top Elasticity of taxable income (ETI) Reported income z captures potentially all margins of behavioural responses to marginal tax rate τ e = 1 τ z z (1 τ) ETI, e, is the % change in reported income when the net-of-tax rate increases by 1% 6 / 141

7 Top Elasticity of taxable income ETI as sufficient statistics (Feldstein, RESTAT 1999) Deadweight loss is the difference between utility loss from taxation W and the revenue from taxation R Marginal deadweight loss ddwl ddwl = dw dr Taxation leads to mechanical effects dm and behavioural revenue effects db dr = dm + DB We have dw = dm (envelope theorem) ddwl = dm (dm + db) = db db depends directly from ETI 7 / 141

8 Top Elasticity of taxable income ETI not as sufficient statistics (Saez, Slemrod and Giertz JEL 2012) Assumption in the basic ETI framework is that reduced z has no other effect on tax revenue Reasonable assumption for real responses (e.g., labour supply responses) Problem if reduced z comes from tax shifting or lead to fiscal externalities Fiscal externalities Shifting between personal/corporate income Shifting over time of income Externalities (e.g., charitable giving) Other parameters needed i) Distinction between real responses vs. shifting ii) How much is shifted income taxed? 8 / 141

9 Elasticity of taxable income Top : high ETI Lindsey (JPubE, 1987) : Reagan, 1981 tax cuts Feldstein (JPE 1995) : Reagan, 1986 tax cuts Very high estimates of ETI More recent studies : smaller ETI More reforms in the U.S. (Goolsbee, 1999) More countries (U.K., Denmark, etc.) Smaller estimates but larger than traditional labour supply estimates Surveys on ETI Saez, Slemrod and Giertz (JEL 2012) Slemrod (NTJ, 1998) 9 / 141

10 Top Feldstein (JPE 1995) Tax reform act (TRA) in 1986 Biggest tax reform in the U.S. since WWII Top MTR down from 50% to 28% Substantial base-broadening : exemptions and preferential tax treatment repealed DiD approach Use panel of tax return data between 1985 and 1988 Construct three income groups 1 Treatment : highest income, with τ 1985 = 49 50% 2 Control 1 : high income, with τ 1985 = 42 45% 3 Control 2 : high income, with τ 1985 = 22 38% DiD approach : exploit differences in MTR e DiD = log(z T ) log(z C ) log(1 τ T ) log(1 τ C ) 10 / 141

11 Table 1: Response of taxable income to changes in MTR Top 1985 data between 1985 and 1988 MTR Income (K$) 1 τ Adjusted taxable income Nb obs , Source : Feldstein (1995), Tab / 141

12 Top Table 2: Response of taxable income to changes in MTR Taxpayer Groups 1 τ Adj. Taxable Ad. Taxable by 1985 MTR Income Income + Loss Percentage Changes, Medium (22-38) High (42-45) Highest (49-50) Differences in differences High vs Med Highest vs High Highest vs Med Implied elasticity estimates High vs Med Highest vs High Highest vs Med Source : Feldstein (1995), Tab / 141

13 Feldstein (JPE 1995) : results Top Very high ETI estimated Estimates between 1.04 and 3.05 Much larger than the usual labour supply elasticities ( ) Implications The U.S. was on the wrong side of the Laffer curve Tax cuts with TRA 86 led to no revenu losses Clinton s 1993 tax increases should lead to no tax revenues The top marginal tax rate should not be much higher than 30% 13 / 141

14 Top Feldstein (JPE 1995) : issues 1) Mean reversion After a negative (positive) income shock, income increases (decreases). This will lead to underestimation. 2) Non-tax related changes in inequality control and treatment groups come from different part of the income distribution. With increasing inequality, the richer will grow richer than the rich or middle income group. It will lead to overestimation. 3) Very small sample 57 tax filers in the treated group Auten and Carroll (RESTAT, 1999) : with larger admin data, they find e=1.10 (compared to Feldstein s 3.05) With additional controls, they preferred estimation is e= / 141

15 Top Feldstein (JPE 1995) : issues 4) Heterogenous elasticity DiD requires homogeneous elasticity If elasticities are increasing in income : then ETI biased upward Example : suppose e T = e and e C = 0 Then : log(z T ) log(z C ) = e log(1 τ T ) and given log(1 τ T ) = 0.5 log(1 τ C ) We obtain : e DiD = 2e 5) TRA86 changed the tax rate and tax base Behavioural effect confounded with definitional effects 6) Short-term vs long-term Some responses could be short term shifting effects 15 / 141

16 Gruber and Saez (JPubE 2002) Top Data Panel data from Exploit all tax changes rather single reform Methodology IV regression analysis ln(z it ) = e ln(1 τ it ) + X it + ν it Endogeneity issue : τ it linked to z it Use predicted change in ln(1 τ it ) assuming income stays constant Isolate changes in tax law as the only source of variation 16 / 141

17 Gruber and Saez (JPubE 2002) Top Source : Gruber and Saez (2002), Tab. 4. Figure 1: Basic elasticity results 17 / 141

18 Gruber and Saez (JPubE 2002) Figure 2: Elasticity results by heterogeneous groups Top Source : Gruber and Saez (2002), Tab / 141

19 Gruber and Saez (JPubE 2002) Top Results ETI estimates of 0.4, and elasticity of broad income of 0.12 Higher ETI for top incomes ( ) Smaller ETI for non-top incomes ( ) Issues : results are fragile 1) Sensitive to exclusion of low income 2) Sensitive to controls for mean reversion (Kopczuk, 2005) 3) Bundles together small tax change and big tax changes 19 / 141

20 Kleven and Schultz (AEJ-EP 2014) Top Danish data Full-population admin data over 25 years Sample is 37 million obs. Danish tax reforms Stable income distribution throughout the period Clear and large tax variations Separate variations for labour and capital income Method Compelling graphical DiD evidence Define treatment/control pre-reform and follow the same group before and after the reforms 20 / 141

21 Kleven and Schultz (AEJ-EP 2014) Top Figure 3: Top income shares in Denmark Source : Kleven and Schultz (AEJ-EP 2014), Fig / 141

22 Kleven and Schultz (AEJ-EP 2014) Figure 4: Danish tax reforms Top Source : Kleven and Schultz (AEJ-EP 2014), Fig / 141

23 Kleven and Schultz (AEJ-EP 2014) Top Figure 5: Danish 1987 reform : labour income Source : Kleven and Schultz (AEJ-EP 2014), Fig. 4.A 23 / 141

24 Kleven and Schultz (AEJ-EP 2014) Top Figure 6: Danish 1987 reform : labour income, large vs small cuts Source : Kleven and Schultz (AEJ-EP 2014), Fig. 4.B. 24 / 141

25 Kleven and Schultz (AEJ-EP 2014) Top Figure 7: Danish 1987 reform : positive capital income Source : Kleven and Schultz (AEJ-EP 2014), Fig. 4.C. 25 / 141

26 Kleven and Schultz (AEJ-EP 2014) Top Figure 8: Labour income elasticities : heterogeneity Source : Kleven and Schultz (AEJ-EP 2014), Tab / 141

27 Kleven and Schultz (AEJ-EP 2014) Top Results Small labour income elasticities ( ) Larger capital income elasticities ( ) Larger elasticities when estimated from larger reforms (frictions, cf. Chetty 2012) Larger labour income elasticities for self-employed, women with kids Limited income shifting between labour and capital income Implications Broad base and strong enforcement leads to modest behavioural responses even under high marginal tax rates 27 / 141

28 at kinks Top Saez (AEJ-EP, 2010) Key prediction of standard labor supply model : individuals should bunch at (convex) kink points of the budget set Amount of bunching at kinks provides non-parametric estimates of intensive elasticity Formula for elasticity ε c = dz/z dt/(1 t) = excess mass at kink % change in net of tax rate 28 / 141

29 Figure 9: Indifference curves and bunching Top Sources : Saez (2010), Fig. 1.A. 29 / 141

30 Figure 10: Density distributions and bunching Top Sources : Saez (2010), Fig. 1.B. 30 / 141

31 Figure 11: Estimating excess bunching using empirical densities Top Sources : Saez (2010), Fig / 141

32 Saez (AEJ-EP, 2010) Top Some evidence of bunching in the U.S. Evidence of bunching at first tax bracket in 1960s implied elasticity of 0.2 No evidence of bunching at higher tax brackets Evidence of bunching at first kink point of EITC implied elasticity of 0.25 Mechanisms for bunching Itemization for income tax Self-employment income for EITC 32 / 141

33 Figure 12: Earnings density distribution and EITC Top Sources : Saez (2010), Fig. 3.A. 33 / 141

34 Top Figure 13: Earnings density distribution and EITC Sources : Saez (2010), Fig. 3.B. 34 / 141

35 Figure 14: Earnings density distribution : wage earners vs self-employed Top Sources : Saez (2010), Fig. 4.A. 35 / 141

36 Figure 15: Earnings density distribution : wage earners vs self-employed Top Sources : Saez (2010), Fig. 4.B. 36 / 141

37 Top Figure 16: around first bracket U.S. income tax ( ) Note : Married tax filers. Sources : Saez (2010), Fig. 6.A. 37 / 141

38 Top Figure 17: Itemizing effects on bunching (first bracket, ) Note : Married tax filers. Sources : Saez (2010), Fig. 7.A. 38 / 141

39 Why not more bunching at kinks? Top 1 True elasticity is small Figure 18: Simulation of bunching according to elasticity Sources : Saez (1999), Fig 5.1 and / 141

40 Why not more bunching at kinks? Top 1 True elasticity is small 2 Randomness in income generation process Saez (2009) shows that year-to-year income variation too small to erase bunching if elasticity is large 3 Salience and information Liebman and Zeckhauser (2004) : Schmeduling Chetty, Friedman and Saez (AER, 2013) : information on EITC 4 Frictions Adjustment costs and institutional constraints Chetty, Friedman, Olsen, and Pistaferri (QJE, 2011) Kleven and Waseem (QJE, 2013) : exploiting notches 40 / 141

41 Why not more bunching at kinks? Top Adjustment costs Search cost, cost of acquiring information about taxes Institutional constraints (e.g., 35 hours week) These frictions reduce elasticities Chetty, Friedman, Olsen, and Pistaferri (QJE 2011) Data on full Danish population Sample restriction : wage-earners million people per year Exploit kinks in Danish tax system 41 / 141

42 Figure 19: Marginal Tax Rates in Denmark in 2000 Top Sources : Chetty, et al (2009). 42 / 141

43 Figure 20: Income Distribution for Wage Earners Around Top Tax Cutoff Top Sources : Chetty, et al (2009). 43 / 141

44 Figure 21: Income Distribution for Wage Earners Around Top Tax Cutoff Top Sources : Chetty, et al (2009). 44 / 141

45 Figure 22: Married Women vs. Single Men Top Sources : Chetty, et al (2009). 45 / 141

46 Figure 23: Teachers vs. Military Top Sources : Chetty, et al (2009). 46 / 141

47 Top Figure 24: Taxable income distributions in 1994 Sources : Chetty, et al (2009). 47 / 141

48 Figure 25: Taxable income distributions in 1995 Top Sources : Chetty, et al (2009). 48 / 141

49 Figure 26: Taxable income distributions in 1996 Top Sources : Chetty, et al (2009). 49 / 141

50 Figure 27: Taxable income distributions in 1997 Top Sources : Chetty, et al (2009). 50 / 141

51 Figure 28: Taxable income distributions in 1998 Top Sources : Chetty, et al (2009). 51 / 141

52 Figure 29: Taxable income distributions in 1999 Top Sources : Chetty, et al (2009). 52 / 141

53 Figure 30: Taxable income distributions in 2000 Top Sources : Chetty, et al (2009). 53 / 141

54 Top Figure 31: Observed Elasticity vs. Size of Tax Change Sources : Chetty, et al (2009). 54 / 141

55 at kinks Top Results Search costs attenuate observed behavioural responses : find larger elasticities around large kink points Groups with more flexibility respond more (secondary earners, self-employed) Overall elasticities estimated from bunching are small in magnitude (perhaps because frictions prevent full response) 55 / 141

56 Top Evidence top incomes Piketty and Saez (QJE, 2003) Use application of Piketty Les hauts revenus en France (2001) Use tax returns to produce estimates of top income share for the U.S. (top 1%, top 0.1%, top 0.01%) Description of inequality over Suggest interpretation that tax policy or social norms behind the recent increase World top income database Large effort of data collection (A. Atkinson, T. Piketty, F. Alvaredo, E. Saez) Using tax returns computation of income shares across the world and over time Project based at PSE : Wealth and Income Database (WID.world) 56 / 141

57 Top income shares times series Top Saez, Slemrod and Giertz (JEL, 2012) Use top income share data for the U.S. Relate change in s t to change MTR, to get ETI Two empirical strategies 1 Tax reform episode e = 2 Full time series log(s 1 ) log(s 0 ) log(1 τ 1 ) log(1 τ 0 ) log(s t ) = α + elog(1 τ t ) + ν t 57 / 141

58 Top Table 3: Elasticity estimates using top income share time series Top 1 percent Next 9 percent (1) Panel A. Tax reform episodes 1981 vs (ERTA 1981) vs (TRA 1986) vs (OBRA 1993) vs (OBRA 1993) Panel B. Full time series No time trends Linear time trend Linear and square time trends Linear, square, and cube time trends Source : Saez, Slemrod and Giertz (2012), Tab / 141

59 Income share times series Top Figure 32: Top 1 percent income share and marginal tax rate Source : Saez, Slemrod and Giertz (2012), Fig. 1.A. 59 / 141

60 Income share times series Top Figure 33: Next 9% income share and marginal tax rate Source : Saez, Slemrod and Giertz (2012), Fig. 1.B. 60 / 141

61 Top Income share times series Long-run evidence in the U.S. 1) 1% share started to increase in 1981, precisely when top MTR was reduced 2) Sharp jump in 1% share in 1986 with TRA86 3) 1% share increased further in the 1990s despite increase in the top MTR 4) No correlation for incomes in 90-99th percentile 5) Top income shares sometimes do not respond to large rate cuts e.g., Kennedy tax cuts in the 1960s Anatomy of behavioural responses Income shifting between corporate and personal tax base e.g., shift from C-corporations to S-corporations Shift in top income composition from dividends to wage and S-corp. 61 / 141

62 Income share times series Top Figure 34: The Top 0.01 Percent U.S. Income Share, Composition, and Marginal Tax Rate, Source : Saez, Slemrod and Giertz (2012), Fig / 141

63 International top incomes Top Piketty, Saez and Stancheva (AEJ-EP 2014) Exploit pre-tax top 1% income share from 18 OECD countries since 1960 Relate changes to MTR to infer elasticity Results Very small elasticity in : Large elasticity in : Test impact on GDP per capita : no significant effect 63 / 141

64 Top International top incomes Figure 35: Top 1 percent share and top marginal tax rate in Source : Piketty, Saez and Stancheva (AEJ-EP 2014), Fig. 2.A. 64 / 141

65 Top International top incomes Figure 36: Top 1 percent share and top marginal tax rate in Source : Piketty, Saez and Stancheva (AEJ-EP 2014), Fig. 2.B. 65 / 141

66 Top International top incomes Figure 37: Changes in Top Income Shares and Top Marginal Tax Rates Source : Piketty, Saez and Stancheva (AEJ-EP 2014), Fig / 141

67 Top Important public debate Concern that top skilled individuals move to low tax countries Bigger concern than supply-side story within country Little research on tax induced migration U.S. State variations (Moretti and Wilson, 2015 ; Young et al. 2015) Special schemes for football players (Kleven, Landais and Saez AER 2013) Danish tax scheme (Kleven, Landais, Saez and Schultz, QJE 2013) 67 / 141

68 Top European football market Bosman ruling (1995) : elimination of the rule for maximum 3 foreign players Beckham law in Spain (2004) : top MTR reduced from 45% to 24% for foreign workers Kleven, Landais and Saez (AER 2013) Exploit Bosman ruling and country specific schemes to assess impact of MTR on football players migration Significant migration responses : high elasticities ( ) 68 / 141

69 Top Figure 38: Cross-Country Correlation between Tax Rates and Migration Source : Kleven, Landais and Saez (AER 2013), Fig. 1.A. 69 / 141

70 Top Figure 39: Cross-Country Correlation between Tax Rates and Performance Source : Kleven, Landais and Saez (AER 2013), Fig. 1.C. 70 / 141

71 Figure 40: Effects of the 2004 Beckham Law in Spain Top 71 / 141

72 Top 1991 Danish tax scheme Higher earners (above 100K euros) taxed at flat rate 25% for three years (instead of regular top rate of 59%) Kleven, Landais, Saez and Schultz (QJE 2013) Exploit Danish admin data DiD strategy (below/above threshold) Results : scheme double number of highly paid foreigners in Denmark Very high elasticities (above 1) Tax competition across countries will reduce ability to tax 72 / 141

73 Figure 41: Number of foreigners by income groups Top Source : Kleven, Landais, Saez and Schultz (2013), Fig / 141

74 Top 1 to optimal taxation 2 models 3 Mirrlees (1971) model 4 labour 1) Optimal linear : Laffer curve 2) Optimal top income tax rate 3) Optimal non linear schedule 5 Optimal transfer programmes 74 / 141

75 Top The purpose of the optimum tax literature (...) does not assume that policy is formed by some benevolent dictator who read the Journal of Public Economics in order to find out what to do. The purpose of the analysis is rather to illuminate the structure of the arguments, explaining the relationship between instruments, constraints, and objectives. Anthony B. Atkinson (1995, p. 17) 75 / 141

76 Top General idea Normative approach about tax design Describe objective of redistribution through social welfare function Model the trade-off between equity and efficiency Social welfare function (cf. lecture 1) Welfarism : social welfare based solely on individual utilities SWF defines the way to model aggregate welfare SWF = µ i u i with µ i the social weight on ind. i Utilitarian or Benthamite SWF : µ i = 1 i 76 / 141

77 Top Simple optimal taxation model Notations Utility u(c) identical for all, depends on after-tax income c Pre-tax income z is fixed (i.e., no behavioural responses), with density h(z) T (z) is tax schedule Government welfare maximisation With utilitarian SWF 0 u ( z T (z) ) h(z)dz Subject to budget constraint T (z)h(z)dz E(λ), with E the revenue requirement for the gov. 77 / 141

78 Top Simple optimal taxation model Solving the model Lagrangian FOC L = [u(z T (z)) + λt (z)]h(z) L T (z) = 0 = [ u (z T (z)) + λ]h(z) u (z T (z)) = λ z T (z) is constant for all z, i.e., after-tax income the same for all Utilitarianism and egalitarianism Utilitarianism with decreasing marginal utility and no behavioural responses leads to perfect egalitarianism (Edgeworth, EJ 1897) 78 / 141

79 Top Simple model : issues 1 No behavioural responses Obvious problem : 100% MTR would destroy incentives to work and thus the assumption that pre-tax income is exogenous is unrealistic Optimal needs to incorporate behavioural responses (Mirrlees, 1971 : efficiency-equity trade-off 2 Utilitarianism Even absent behavioural responses, many people would object to 100% redistribution Citizens view on fairness impose bounds on redistribution Alternatives to utilitarianism have been discussed in the literature (cf. lecture 1) 79 / 141

80 Top George Akerlof, American economist, Nobel Prize in Famous for his article on the market for lemons. Akerlof (AER, 1978) Government can observe characteristics X and define the tax system as T (X, z) If X is correlated with endowments or ability and immutable, then redistribution can be efficient Logic : tagging on immutable characteristics leads to no deadweight loss 80 / 141

81 Top Potential characteristics for tagging Used in current tax/benefit systems Age : e.g., minim income for pensioners Children : e.g., child benefits Disability : e.g., disability benefits Not used, but correlated with earnings Gender, height, beauty, DNA, etc. Disadvantages of tagging Perverse incentives by stigmatisation of tagged individuals Horizontal equity issue : characteristics used reflect direct needs or direct ability to earn Administrative costs e.g., medical test for disability 81 / 141

82 Top Alesina, Ichino and Karabarbounis (AEJ-EP 2011) Gender-based taxation Higher labour supply elasticity of women + lower average income Lower taxation of women Mankiw and Weinzierl (AEJ-EP 2010) Height is correlated to income (+5cm = +4%) model : tall person should pay $4500 more than short one at same level of income Contradict horizontal equity If result non acceptable, then is welfarism (and optimal taxation) flawed? 82 / 141

83 Top Figure 42: Wage distribution by height for adult male in the U.S. Source : Mankinw and Weinzierl (2013), Fig / 141

84 Top The James Mirrlees, British (Scottish) economist, Nobel Prize in Mirrlees (REStud 1971) Technical paper Huge impact on information economics (e.g., models with asymmetric information in contract theory) Until late 1990s not connected with empirical evidence Assumptions 1 Workers have fixed productivities (pre-tax wage are fixed) 2 Workers choose their effort/hours of work (intensive margin only) 84 / 141

85 Top The Standard labor supply model Individual maximizes utility u(c, l) with c is consumption, l labor supply Budget constraint is c = wl T (wl) with w wage rate, T (.) income tax Heterogeneity Individuals differ in ability w with density f (w) Gov. maximises a SWF SWF = G(u(c, l))f (w)dw Budget constraint : T (wl)f (w)dw E Individual FOC : w(1 T )u c + u l = 0 where G(.) is increasing and concave 85 / 141

86 The : results Top An integrated tax/benefit system T (.) < 0 at the bottom (transfers) T (.) > 0 further up General results 1) Non-negative MTR : T (.) > 0 (i.e., rules out EITC/working tax credit) 2) MTR should be 0 at the top if the skill distribution is bounded 86 / 141

87 Top Connecting optimal models to data Atkinson (1995), Diamond (AER 1998), Piketty (RFE 1997) and Saez (REStud 2001) General idea : deriving the optimal tax schedule as functions of elasticities and income distribution Surveys : Diamond and Saez (JEP 2011), Piketty and Saez (HPE 2013) Three sets of approaches 1) Optimal linear (Laffer curve) 2) Optimal top income tax rate 3) Optimal non linear schedule 87 / 141

88 Top Optimal linear The popular Laffer curve Raising tax rates leads to behavioural effects Revenue increases will be less than mechanical effects Revenue curve inverted U-shape Arthur Laffer s sketch Dinner with Dick Cheney and Donald Rumsfeld in 1974 in Washington D.C. Notion known for a long time Dupuit (1844) : revenue curve (cf. excess burden) 88 / 141

89 Top Optimal linear Optimal linear tax rate Linear tax rate τ with lump-sum grant T (0) Individuals earn z and consume c = (1 τ)z + T (0) Maximize u(c, z) to get z(1 τ, R) (labour supply choice) At the aggregate, total tax revenues R(τ) = τz(1 τ, R) Revenu maximizing linear tax rate τ R (τ) = Z τ dz d(1 τ) = 0 τ = e with e = 1 τ dz Z d(1 τ) Top of the Laffer curve depends on e the elasticity of aggregate earnings to the net-of-tax rate 89 / 141

90 Top Optimal linear Gov. maximises generalized SWF General SWF G(u i ) with Pareto weights ω i ḡ average normalised social marginal welfare weight weighted by pre-tax income Optimal linear rate (see Piketty and Saez, 2013) τ = 1 ḡ 1 ḡ + e Consequences on optimal tax rates 1 τ decreases with aggregate elasticity e 2 τ decreases with redistribution taste ḡ e.g., no taste for redistribution (ḡ = 1), τ = 0 e.g., Rawlsian SWF (ḡ = 0), τ = 1 1+e 3 τ increases with inequality (higher inequality leads to lower ḡ 90 / 141

91 Top Top marginal tax rate Notations Denote z the threshold for the top marginal tax rate τ z is the average income of the N taxpayers above z g is the social value of additional income of high earners Elasticity of earnings z with the net of tax rate 1 τ e = 1 τ z z (1 τ) Small increase in the top marginal tax rate dτ 1 Mechanical effect dm = N(z z)dτ 2 Behavioural response τ db = N e z 1 τ dτ 91 / 141

92 Top Top marginal tax rate Net effect of this small increase dτ τ dm+db+dw = N(z z)dτ Nez 1 τ dτ gn(z z)dτ [ z τ ] dm + db + dw = N(z z)dτ 1 g e z z 1 τ At the optimum, this expression must be zero Note a = z z z, the thinness of the income distribution τ = 1 g 1 g + ae The top marginal tax rate is : Decreasing with the social welfare weight g Decreasing with elasticity of reported earnings Decreasing with the thinness of the income distribution 92 / 141

93 Top Optimal non-linear schedule Notations Denote T (z) the tax schedule at earnings level z T (z) is the marginal tax schedule H(z) is the cumulative distribution of taxpayers h(z) is the density distribution of taxpayers G(z) average social value of 1 for earners above z Small increase in the marginal tax schedule dτ in the income range (z, z + dz) 1 Mechanical effect 2 Behavioural response 3 Welfare effect dm = (1 H(z))dτdz T (z) db = e z dτ 1 T (z) h(z)dz dw = dmg(z) 93 / 141

94 Optimal non-linear schedule Top Optimal marginal tax schedules T (z) 1 T (z) = 1 e 1 H(z) (1 G(z)) zh(z) Determinants of the optimal tax schedule 1 Elasticity of reported earnings e 2 Thinness of the income distribution 3 Social value of consumption for individuals with given earnings level Implications Negative marginal tax rates are never optimal 94 / 141

95 Optimal non-linear schedule Top Saez (REStud 2001) Estimate optimal tax schedule on U.S. data Estimate the shape of earnings distribution Estimate earnings elasticity Choose social welfare function Estimate optimal tax schedule Brewer, Saez and Shephard (MR 2010) U.K. data for Mirrlees Review Earnings elasticities estimated using 1980s tax changes Large standard errors around estimates e.g. top marginal tax rate in main scenario between 50.4% and 64.5% 95 / 141

96 Optimal non-linear schedule Figure 43: Optimal tax simulations (U.S.) Top 96 / 141

97 Optimal non-linear schedule Figure 44: Optimal tax sensitivity, labour elasticity (U.K.) Top Source : Brewer et al. (2010), Fig. 2.4A. 97 / 141

98 Top Optimal non-linear schedule Figure 45: Optimal tax sensitivity, redistribution preference (U.K.) Source : Brewer et al. (2010), Fig 2.4B. 98 / 141

99 Optimal transfer programmes Top Two approaches 1 Intensive margin : Mirrlees (1971) 2 Extensive margin : Diamond (JPubE 1980), Saez (QJE 2002), Laroque (ECMA 2005) : negative income tax Lump-sum grant T (0) for those with no earnings High MTR at the bottom : a) target transfers (low cost) b) intensive response does not generate large losses (earnings low at the bottom) 99 / 141

100 Optimal transfer programmes Top Diamond and Saez (JEP 2011) g 0 social marginal weight on zero earners e 0 elasticity of fraction non-working to the bottom net-of-tax rate Optimal bottom marginal tax rate with intensive margin only τ 1 = g 0 1 g e 0 Implications of the formula If society values redistribution towards zero earners, τ 1 will be high e.g., with g 0 = 3, e 0 = 0.5, then τ 1 = 80% 100 / 141

101 Optimal transfer programmes Top Extensive margin responses With fixed cost of work, extensive margin might be more responsive Empirical literature finds bigger labour supply elasticities at the extensive margin Participation labour supply (Saez, QJE 2002) Income when working c i = w i T i Income when not working c 0 Person works if c i θ > c 0, with θ fixed cost of work 101 / 141

102 Optimal transfer programmes Figure 46: Introducing in-work credit Top Source : Piketty and Saez (2013), Fig / 141

103 Optimal transfer programmes Top Results (extensive margin only) Negative MTR are optimal (i.e., in-work credit are optimal) NIT is not optimal Implications In practice, both intensive and extensive margin exist Trade-off between negative MTR in phase-in of in-work credit (good for extensive margin) against high MTR in phase-out (bad for intensive margin) 103 / 141

104 Top 1 At the top of the distribution Do tax cuts pay for themselves? What should be top marginal tax rates? 2 At the bottom of the distribution Traditional welfare Basic income and negative income tax Welfare to work 104 / 141

105 Debate among economists Top Mankiw, Weinzierl and Yagan (JEP 2009) 1 Optimal MTR schedule could decline at high incomes 2 Flat tax and universal lump-sum transfer is close to optimal 3 Capital income should not be taxed Diamond and Saez (JEP 2013) 1 Very high earnings should be subject to rising MTR and higher rates than current U.S. policy for top earners 2 Tax/transfer policy toward low earners should include subsidization of earnings and should phase out the subsidization at a relatively high rate 3 Capital income should be taxed 105 / 141

106 Debate among economists Top Mankiw (JEP 2013) : defending the one percent Technology and great innovators lead to wealth Equality of opportunity is better than ex post equality Utilitarianism is flawed as philosophical guideline Piketty (2013) Confiscatory rates for top incomes are necessary Optimal top marginal tax rate should be around 80% Prevent rent-seeking, no objective to raise revenue 106 / 141

107 Income tax reforms Top Large cuts in top marginal tax rates US, 1972 : top marginal tax rate from 72% to 60% US, Tax Reform Act 1986 : top marginal tax rate from 50% to 28% UK, 1979 : top marginal tax rate from 83% to 60% UK, 1988 : top marginal tax rate from 60% to 40% Sweden, 1991 : top marginal tax rate from 80% to 50% Modest reversal? Bush then Clinton tax increase in early 1990s : top marginal rate from 28% to 40% Brown in latest budget, increase from 40% to 50%. 107 / 141

108 Top The side of the Laffer curve Do tax cuts pay for themselves? With Reagan, idea that tax cut would pay for themselves Large deficits have followed the Tax Reform Act 1986 Tax increases from Bush senior and Clinton : large budget surplus Popular view that tax cuts visibly don t pay for themselves On the wrong side of the Laffer curve? Use the formula of linear taxation τ = 1 ḡ 1 ḡ + e Choose different plausible elasticity e Different redistribution taste ḡ 108 / 141

109 The side of the Laffer curve Top Table 4: Optimal linear tax rate formula Elasticity e =0.25 Elasticity e =0.5 Elasticity e =1 ḡ τ ḡ τ ḡ τ A. Optimal linear rate τ Rawlsian SWF 0 80% 0 67% 0 50% Utilitarian SWF % % % B. Revealed preference for redistribution ḡ US tax level (35%) % % % EU tax level (50%) % % % Sources : Piketty and Saez (2013), Tab / 141

110 Sources of top income inequality Top 1 Technology Technology favours skilled workers IT favours entrepreneurs who can reach globally 2 Supply side story Top income now higher because top marginal rates are lower People at the top work more 3 Tax avoidance story Top earners avoid less when top tax rate decreases 4 Rent-seeking Top earners extract more pay when top rates are low 110 / 141

111 Top income tax Top Three main positions (1) Lower top marginal tax rates (supply side) Broaden the tax base and international coordination (to reduce avoidance) (3) Increase top marginal tax rates (to lower rent-seeking) Piketty, Saez and Stancheva (AEJ-EP 2014) Discuss the optimal policy in terms of three elasticities a) labour supply e 1 b) tax avoidance e 2 c) compensation bargaining e / 141

112 Top Top income tax Real changes vs. tax avoidance According to the tax avoidance story, increase in top income shares is misleading U.S. top incomes shares were as high in the 1960s as today but reported income was smaller Evidence against this scenario 1) Correlation with MTR similar when using narrow tax-base measure (e.g., excluding capital gains) 2) Charitable giving (tax deductible) has grown along with top incomes Evidence in favor of this explanation Causal impact of change MTR have a hard time finding large real effects 112 / 141

113 Top income tax Top Supply-side vs. rent seeking? According to supply side story, lower MTR should have led to higher growth According to rent-seeking story : higher top income share is at the expense of bottom 99% Evidence 1) No correlation between MTR and growth 2) CEO pay across countries negatively correlated with top MTR 113 / 141

114 Top income tax Figure 47: Growth and change in top marginal tax rate Top Source : Piketty, Saez and Stancheva (AEJ-EP 2014), Fig. 4.A. 114 / 141

115 Top income tax Top Figure 48: Average CEO compensation Source : Piketty, Saez and Stancheva (AEJ-EP 2014), Fig. 5.A. 115 / 141

116 Table 5: Synthesis of various scenarios Top Scenario 1 : Scenario 3 : Standard Scenario 2 : Compensationsupply-side Tax-avoidance effects bargaining tax effects effects (a) Current narrow tax base (b) After base broadening Panel A. Total elasticities e = e 1 + e 2 + e 3 = 0.5 e 1 = 0.5 e 1 = 0.2 e 1 = 0.2 e 1 = 0.2 e 2 = 0.0 e 2 = 0.3 e 2 = 0.1 e 2 = 0.0 e 3 = 0.0 e 3 = 0.0 e 3 = 0.0 e 3 = 0.3 Panel B. Optimal top tax rate τ = (1 + tae 2 + ae 3 )/(1 + ae) Pareto coefficient a = 1.5 Pareto coefficient t = 20% τ = 57% τ = 62% τ = 71% τ = 83% Source : Piketty, Saez and Stancheva (AEJ-EP 2014), Tab / 141

117 Top 1 Traditional welfare 2 Basic income and NIT 3 In-work credit 117 / 141

118 Traditional welfare Top Targeted benefits : tagging Benefits on grounds of disability Income support for the old Income support for lone mothers Traditional welfare Disregards for incentive effects Means-testing with 100% tax rate (100% taper rate or 100% benefit withdrawal) Creation of poverty trap : once on welfare, no incentives (i.e., no rewards) to go back to work 118 / 141

119 Traditional welfare Top 119 / 141

120 Negative income tax Top Milton Friedman in Capitalism and Freedom (1962) proposed the replacement of all welfare benefits by a negative income tax (NIT) : Guaranteed income paid by the government Each additional dollar of income taxed at a marginal rate (below 100%) Randomized experience in the U.S. in the 1970s (cf. lecture 6) Small elasticities for male Larger for women 120 / 141

121 Negative income tax Top 121 / 141

122 Iron triangle of redistribution Top Labour supply effects of NIT Lower marginal tax rates for low incomes : positive effects for the individuals not working Higher marginal tax rates higher in the income distribution : negative effects on labour supply The iron triangle of redistribution 1 Redistribution to the poor (high replacement income) 2 Incentives to work (low marginal tax rates) 3 Low cost to the government 122 / 141

123 Negative income tax Top 123 / 141

124 Negative income tax Top 124 / 141

125 Welfare to work Top Welfare reforms in the 1990s Welfare to work or workfare Removing high marginal tax rates on low incomes Politically attractive to condition welfare on work Spread of these reforms In the US, Earned Income Tax Credit (EITC) In the UK, Working Families Tax Credit (WFTC) and then Working Tax Credit (WTC) In France, Prime pour l emploi (PPE) and Revenu de solidarité active (RSA) In Singapore, Workfare Income Supplement (WIS) 125 / 141

126 Tax credit Top 126 / 141

127 Mix of policies Top 127 / 141

128 The EITC in the US Top The Earned Income Tax Credit (EITC) Large increase under Clinton administration Now the largest cash antipoverty programme in the US ($34.6 billion in 2006) EITC amounts depend on the number of children (higher for families) EITC is computed based on family income Three components 1 An increasing subsidy part (40% per dollar of wage top-up) 2 A constant amount (no tax) 3 Then a taper rate of 21% as benefits are withdrawn with increasing income 128 / 141

129 The EITC in the US Top 129 / 141

130 Top Studies of the impact of the EITC DiD estimation (Eissa and Liebman, 1996) Table 6: Estimation of the 1986 EITC increase on employment Before After Diff. Women with kids Women without kids Diff Results Positive impact on participation of lone mothers ( ppts) No negative effects on married men s labour supply Modest reduction in married women s labour supply Relatively successful redistribution programme Flaws of the programme Low amount to the childless No increase with more than two children 130 / 141

131 Top Tax credit programme in the UK In the UK, the term tax credit programme describe a flurry of schemes, which have been modified, abolished, created... In 2003 two main schemes replaced the older ones 1 Child Tax Credit (CTC) : tax credit for families with children. Family element at 545 p.a., taped away above 50,000 of family income at a rate of 6.7% Child element at 2085 p.a. per child, taped away above 15,575 of family income at a rate of 39% 2 Working Tax Credit (WTC) : in-work tax credit for low income. Minimum of 16 hours work for families, and minimum of 30 hours work for childless adults. Taper rate is 39% and entitlement is exhausted at 15,575 and the basic amount is hours week bonus of / 141

132 The tax credit expansion in the UK Top Sources : Adam, S. and Brewer, M., Take-up of Family Credit and Working Families Tax Credit (2005) 132 / 141

133 The UK tax credits Top Figure 49: WTC : family with one child Source : Brewer, M., The New Tax Credit (2003) 133 / 141

134 The UK tax credits Top Figure 50: WTC : couple without children Source : Brewer, M., The New Tax Credit (2003) 134 / 141

135 The impact of the WFTC in the UK Top Impact evaluation DiD estimation (Brewer et al. 2005) Structural modeling (Brewer et al. 2006) Results Increase in participation by lone mothers (+5 ppts) Slight decrease in participation by mothers in couple Slight increase in participation by fathers in couple Puzzle : EITC vs WFTC WFTC twice as large as EITC but half the impact Interactions with other benefits (housing benefit etc.) 135 / 141

136 Interactions of UK benefits Top Source : Blundell (2006) 136 / 141

137 NIT vs EITC Top results With intensive margin only : NIT is superior With extensive margin : EITC is superior Other questions Incidence of EITC? Role of minimum wages? Issue of non take-up Role of active policies towards poor households? 137 / 141

138 Top Akerlof, G. (1978), The Economics of as Applied to the Optimal Income Tax, American Economic Review, Vol. 68, pp Alesina, A., Ichino, A. and Karabarbounis, L. (2011), Gender-Based Taxation and the Division of Family Chores, American Economic Journal : Economic Policy, Vol. 3, No. 2, pp Atkinson, A. (1995), Public Economics in Action. Basic Income/Flat Tax Proposal, Lindahl Lectures Series, Oxford University Press. Auten, G. and Carroll, R. (1999), The Effect of Income Taxes on Household Income, Review of Economics and Statistics, Vol. 81, pp Blundell, R. (2006), Earned income tax credit policies : Impact and Optimality, The 2005 Adam Smith Lecture, Labour Economics, Vol. 13, pp Brewer, M., Saez, E. and Shephard, A. (2010), Means-Testing and Tax Rates on Earnings, in The Mirrlees Review : Dimensions of Tax Design, pp , Oxford University Press. Diamond, P. (1980), Income Taxation with Fixed Hours of Work, Journal of Public Economics, Vol. 13, pp Diamond, P. (1988), Optimal Income Taxation : An Example with a U-shaped pattern for the optimal marginal rates, American Economic Review, Vol. 88, No. 1, pp Diamond, P. and Saez, E. (2011) The Case for a Progressive Tax : From Basic Research to Policy Recommendations, The Journal of Economic Perspectives, Vol. 25, No. 4, pp Dupuit, J. (1844), De la mesure de l utilité des travaux publics [reproduced in Revue française d économie, 10, 1995, pp ] Edgeworth, F. (1897), The Pure Theory of Taxation III, The Economic Journal, 7 (28), pp Feldstein, M. (1995), The Effects of Marginal Tax Rates on Taxable Income, Journal of Political Economy, Vol. 103, No. 3, pp Feldstein, M. (1999), Tax Avoidance and the Deadweight Loss of the Income Tax, Review of Economics and Statistics, Vol. 81, No. 4, pp Goolsbee, A. (1999), Evidence on the High-Income Laffer Curve from Six Decades of Tax Reform, Brookings Papers on Economic Activity, Vol. 1999, No. 2, pp Gruber, J. and Saez, E. (2002), The Elasticity of Taxable Income : Evidence and Implications. Journal of Public Economics 84 (1), pp / 141

139 Top Heady, C. (1993), Optimal Taxation as a Guide to Tax Policy : A Survey, Fiscal Studies, Vol. 14, No. 1, pp Kleven, H. and Schultz, E. (2014), Estimating Taxable Income Responses using Danish Tax Reforms, American Economic Journal : Economic Policy, Vol 6, No. 4, pp Kleven, H., C. Landais, E. Saez, and Schultz, E. (2014), Migration and Wage Effects of Taxing Top Earners : Evidence from the Foreigners Tax Scheme in Denmark, The Quarterly Journal of Economics, 129 (1), pp Kleven, H., C. Kreiner, and Saez, E. (2009), The Optimal Income Taxation of Couples, Econometrica 77, pp Kleven, H., C. Landais and Saez, E. (2013), Taxation and International Migration of Superstars : Evidence from the European Football Market, American Economic Review 103 (5), pp Lindsey, L. (1987), Individual Taxpayer Response to Tax Cuts : , Journal of Public Economics 33, pp Mankiw, G. and Weinzierl, M. (2010) The Optimal Taxation of Height : A Case Study of Utilitarian Income Redistribution, American Economic Journal : Economic Policy 2 (1), pp Mirrlees J. (1971), An Exploration in the Theory of Optimum Income Taxation, Review of Economic Studies, Vol. 38, No. 2, pp Piketty T. (1997), La Redistribution fiscale contre le chômage, Revue française d économie, Vol. 12, No. 1, pp Piketty, T. (2013), Le Capital Au XXIe Siècle, Seuil. Piketty, T. and Saez, E. (2003), Income Inequality in the United States, , Quarterly Journal of Economics, Vol. 116, pp Piketty, T., E. Saez and Stantcheva, S. (2014), Optimal Taxation of Top Labor Incomes : A Tale of Three Elasticities, American Economic Journal : Economic Policy 6(1), pp Saez, E. (2001), Using Elasticities to Derive Optimal Income Tax Rates, Review of Economic Studies, Vol. 68, pp / 141

140 Top Saez, E. (2002), Optimal Income Transfer Programs : Intensive Versus Extensive Labor Supply Responses, Quarterly Journal of Economics, Vol. 117, No. 3, pp Saez, E, Slemrod, J. and Giertz, S. (2009), The Elasticity of Taxable Income with Respect to Marginal Tax Rates : A Critical Review, Journal of Economic Literature, 50(1), pp Slemrod, J. (1998), Methodological Issues in Measuring and Interpreting Taxable Income Elasticities, National Tax Journal, 51(4), pp Moretti, E. and Wilson, D. (2015), The Effect of State Taxes on the Geographical Location of Top Earners : Evidence from Star Scientists, NBER Working Paper No Slemrod, J. (1995), Income Creation or Income Shifting? Behavioral Responses to the Tax Reform Act of 1986, American Economic Review, Vol. 85, pp Young, C., Varner, C., Lurie, I. and Prisinzano, R. (2016), Millionaire Migration and the Taxation of the Elite : Evidence from Administrative Data, American Sociological Review, Vol. 81, No. 3, pp / 141

141 Top Lecture 7: Labour Paris School of Economics (PSE) École des hautes études en sciences sociales (EHESS) Master APE and PPD Paris October / 141