Chapter 15 Income Taxation Reading • Essential reading – Hindriks, J and G.D. Myles Intermediate Public Economics. (Cambridge: MIT Press, 2005) Chapter 15. • Further reading – Blundell, R. (1992) ‘Labour supply and taxation: a survey’, Fiscal Studies, 13, 15—40. – Feldstein, M. (1995) ‘The effect of marginal tax rates on taxable income: a panel study of the 1986 tax reform act’, Journal of Political Economy, 103, 551—572. – Hindriks, J. (2001) ‘Is there a demand for income tax progressivity?’, Economics Letters, 73, 43—50. – Kanbur, S.M.R. and M. Tuomala (1994) ‘Inherent inequality and the optimal graduation of marginal tax rates’, Scandinavian Journal of Economics, 96, 275—282. Reading – Myles, G.D. (2000) ‘On the optimal marginal rate of income tax’, Economics Letters, 66, 113—119. – Romer, T. (1975) Individual welfare, majority voting and the properties of a linear income tax, Journal of Public Economics, 7, 163—168. – Roberts, K. (1977) ‘Voting over income tax schedules’, Journal of Public Economics, 8, 329—340. – Tuomala, M. Optimal Income Tax and Redistribution. (Oxford: Clarendon Press, 1990) [ISBN 0198286058 hbk]. • Challenging reading – Diamond, P.A. (1998) ‘Optimal income taxation: an example with a U-shaped pattern of optimal marginal tax rates’, American Economic Review, 88, 83—95. – Mirrlees, J.A. (1971) ‘An exploration in the theory of optimum income tax’, Review of Economic Studies, 38, 175—208. Reading • Seade, J.K. (1977) ‘On the shape of optimal tax schedules’, Journal of Public Economics, 7, 203—235. • Saez, E. (2001) ‘Using elasticities to derive optimal tax rates’, Review of Economic Studies, 68, 205—229. • Weymark, J.A. (1986) ‘A reduced-form optimal income tax problem’, Journal of Public Economics, 30, 199—217. Income Taxation • Income taxation is a major source of government revenue • It is also a major source of contention – The income tax is a disincentive to effort and enterprise so the rate of tax should be kept as low as possible – Income taxation is well-suited to the task of redistribution which requires that high earners pay proportionately more tax on their incomes than low earners • The determination of the optimal income tax involves the resolution of these contrasting views Taxation and Labor Supply • The effect of income taxation on labor supply can be investigated using the standard model of consumer choice • This highlights the importance of competing income and substitution effects • Assume – The consumer has a given set of preferences over allocations of consumption and leisure – The consumer has a fixed stock of time to divide between labour supply and leisure • The choice is made to maximize utility Taxation and Labor Supply • Preferences are represented by U = U(x, L - ℓ) = U(x, ℓ) • L is the stock of time, ℓ is labor supply, and x is consumption – Leisure time is L - ℓ • • • • Labour is assumed unpleasant so ∂U/∂ℓ < 0 Each hour of labour earns wage w Income before taxation is wℓ With tax rate t the budget constraint is px = (1 – t)wℓ Taxation and Labor Supply • Alternatively the preferences of the consumer can be be written in terms of income • Let z ≡ wℓ denote income before tax • Utility in terms of income is U = U(x, z/w) • The budget constraint becomes px = (1 - t)z • The consumer’s indifference curves depend upon the wage rate Taxation and Labor Supply Consumption • Fig. 15.1a depicts the choice between leisure and consumption • The budget constraint depends on the wage • Fig. 15.1b depicts the choice between before tax income and consumption • The indifference curves depend on the wage • In both cases the budget constraint depends on the tax rate 1 t wL p x* Consumption L * a. Leisure L Leisure px 1 t z x* Before tax income z* b. Before tax income Figure 15.1: Labor supply decision Taxation and Labor Supply Consumption • The initial choice is at a • In Fig. 15.2a an increase in w shifts the budget constraint • In Fig. 15.2b an increase in w shifts the indifference curve • The choice moves to c – a to b is the substitution effect – b to c is the income effect • The total effect can raise or lower labor supply but increases income c b a Leisure a. Leisure Consumption c a Before tax income b. Before tax income Figure 15.2: Effect of a wage increase Taxation and Labor Supply Consumption • Income z* in Figs. 15.3a and b is a threshold level of income below which income is untaxed c b – The budget constraint is kinked at b • Points a and c are interior solutions • Point b is a corner solution • A consumer at a corner may be unaffected by a tax change a L z* w a. Leisure Consumption Leisure c b a Before tax income z* b. Before tax income Figure 15.3: A tax threshold Taxation and Labor Supply Consumption • For many tax systems the marginal rate of tax has several discrete increases • Figs 15.4a and b display the case of four marginal rates • The marginal rates increase with income • The budget constraint is kinked at each point of increase Consumption a. Leisure b. Before tax income Leisure Before tax income Figure 15.4: Several thresholds Taxation and Labor Supply Consumption • It may not be possible to continuously vary hours of work • A minimum working week gives a choice between 0 hours and the minimum ℓm • This causes a discontinuity in the budget constraint • Figs. 15.5a and b show a discontinuity in labor supply as the tax rate changes Consumption T m T a. Leisure Leisure w m Before tax income b. Before tax income Figure 15.5: Taxation and the participation decision Empirical Evidence • The theoretical analysis of labor supply makes three major points – The effect of a wage or tax change depends on income and substitution effects – Kinks in the budget constraint can make behaviour insensitive to taxes – The participation decision can be sensitive to taxation • The theory does not predict the size of these effects • Empirical evidence is required to provide quantification Empirical Evidence • Evidence on the effect of income taxes can be found in – The results of taxpayer surveys – Econometric estimates of labor supply functions • Two points are important in choosing s survey sample – Labor supply is insensitive to taxation if working hours are determined by the firm or by union/firm agreement – The effect of taxation can only be judged when workers who have the freedom to vary hours of labor Empirical Evidence • Surveys usually conclude that changes in the tax rate have little effect on the labor supply decision • If correct the labor supply function is approximately vertical – This results from the income effect almost entirely offsetting the substitution effect – This predicts taxation will have little labor supply effect • Different groups in the population may have different reactions to changes in the tax system • This is now considered by reviewing some econometric analysis Empirical Evidence • Tab. 15.1 reports estimates of labor supply elasticities for three groups • The substitution effect (compensated wage) is positive but the income effect is always negative • The elasticity for married men is the lowest • The elasticity for unmarried women is the largest – Participation effect Married Women Married Men Lone Mothers US UK US UK US UK Uncompensated wage 0.45 0.43 0.03 -0.23 0.53 0.76 Compensated wage 0.90 0.65 0.95 0.13 0.65 1.28 Income -0.45 -0.22 -0.98 -0.36 -0.18 -0.52 Table 15.1: Labor-supply elasticities Source: Blundell (1992) Optimal Income Taxation • The optimal income tax balances efficiency and equity to maximise welfare • A interesting model must have the following attributes: – An unequal distribution of income so there is an equity motivation for taxation – The income tax must affect labor so that it has efficiency effects – There must be no restrictions placed on the optimal tax function • The Mirrlees model of income taxation is the simplest that has these attributes Optimal Income Taxation • All consumers have identical preferences but differ in their level of skill • The level of skill determines the hourly wage • Income is the product of skill and hours worked • The level of skill is private information and cannot be observed by the government – This makes it impossible to tax directly. – A tax levied on skill would be the first-best policy but this not feasible • The government employs an income tax as a second-best policy Optimal Income Taxation • The government is subject to two constraints when it chooses the tax function – The income tax must meet the government’s revenue requirement – The tax function must be incentive compatible • View the government as assigning to each consumer an allocation of labor and consumption • Incentive compatibility requires that each consumer must find it utility maximizing to choose the allocation intended for them – No alternative allocation should be preferred Optimal Income Taxation • If a consumer of skill level s supplies ℓ hours of labour they earn income of sℓ before tax • Denote the income of a consumer with skill s by z(s) • For a consumer with income z the income tax paid is given by T(z) – T(z) is the tax function the analysis aims to determine • A consumer who earns income z(s) can consume x(s) = c(z(s)) = z(s) – T(z(s)) Optimal Income Taxation • Fig. 15.6 illustrates the budget constraint • Without taxation the budget constraint is the 45o line • T(z) < 0 when the consumption function is above the 45o line • T(z) > 0 when the consumption function is below the line • The gradient of the consumption function is 1 – T′ x T zˆ xˆ c z T 45o zˆ Figure 15.6: Taxation and the Consumption function z Optimal Income Taxation • Preferences are assumed to satisfy the agent monotonicity condition • At any point (z, x) the indifference curve of a consumer of skill s1 is steeper than the curve of a consumer of skill s2 if s2 > s1 • This is shown in Fig. 15.7 • Consumers of lower skill are less willing to supply labor x High-skill Low-skill xˆ zˆ Figure 15.7: Agent monotonicity z Optimal Income Taxation • Fig. 15.8 shows the consequence of agent monotonicity • The low-skill consumer chooses a • The indifference curve of the high-skill is flatter and cannot be at a tangency • The choice for the highskill must be further to the right • Income is increasing with skill x Low-skill High-skill a z Figure 15.8: Income and skill Optimal Income Taxation • Consider the consumption function in Fig. 15.9 • No consumer will locate on the downward-sloping section • This part of the consumption function can be replaced by the flat dashed section • This shows c′(z) > 0 so 1 – T′(z) > 0 – The marginal tax rate is less than 100 percent x z Figure 15.9: Upper limit on tax rate Optimal Income Taxation • Fig. 15.10 shows the marginal tax rate must be positive • Start with c1 with c1′ > 1 and move to c2 with c2′ = 1 – c2 chosen so tax revenue is unchanged • High-skill moves from h1 to h2, low-skill from l1 to l2 – Consumption is transferred from high skill to low skill so welfare rises • c1 could not be optimal x c1 h1 h2 c2 l2 l1 z Figure 15.10: Lower limit on tax rate Optimal Income Taxation • The highest-skill consumer should face a zero marginal rate of tax • In Fig. 15.11 ABC does not have this property • Replace with ABD where BD has gradient of 1 – Highest-skill consumer moves to b – Utility rises but tax payment is unchanged – No-one is worse-off • ABC cannot be optimal x D b C B A 45 o z Figure 15.11: Zero marginal rate of tax Optimal Income Taxation • A tax system is progressive if the marginal rate of tax increases with income – A zero rate at the top shows progressivity cannot be optimal – Most tax systems are progressive • This result is valid only for the highest-skill consumer – The implications for lower skills are limited – Observed systems may only be ‘wrong’ at the very top • Result questions preconceptions about the structure of taxes Two Specializations • There are two specializations of the general model that provide additional insight • The quasi-linear model restricts the form of the individual utility function • The individual utility function becomes U = u(x) – z/s • Rawlsian taxation adopts a specific social welfare function • Social welfare is evaluated by W = min{U} Two Specializations • Assume there are just two consumers – The high-skill is sh and the low-skill sl • The optimal tax problem is equivalent to choosing the allocations ah and al for these consumers • Incentive compatibility requires that the consumer of skill i prefers allocation i • The low-skill will never mimic the high-skill so only one incentive compatibility constraint is binding u(xh) – zh/sh = u(xl) – zl/sh Two Specializations • Fig. 15.12 illustrates the role of allocations • The allocations al and ah are incentive compatible • These cannot be optimal since xh can be reduced and xl raised without violating incentive compatibility x High-skill Low-skill al ah – The change raises welfare • The high-skill must be indifferent between al and ah z Figure 15.12: Allocations and the consumption function Two Specializations • The resource constraint xl + xh = zl + zh and the incentive compatibility condition can be solved to give zl = (1/2)[xl + xh – sh[u(xh) – u(xl)]] zh = (1/2)[xl + xh + sh[u(xh) – u(xl)]] • Using these the optimal allocation of consumption for utilitarian social welfare solves max blu(xl) + bhu(xh) – [(sl+sh)/2slsh][xl + xh] • Where bl = (3sl – sh)/2sl and bh = (sl+sh)/2sl Two Specializations • The welfare weights bl and bh incorporate incentive compatibility and resource implications • For the high-skill the solution to the optimization is u′(xh) = 1/sh so that MRSh = 1 – This captures the zero marginal rate for the highestskilled • For the low skill u′(xl) = (sl+sh)/sh(3sl – sh) so MRSl = sh (3sl – sh)/sl(sl+sh) < 1 – The low-skill faces a positive marginal rate of tax Two Specializations • Rawlsian taxation aims to maximize the utility of the worst-off • Assume all tax revenue is redistributed as a lump-sum grant • It can then be assumed that the optimal Rawlsian tax maximizes the grant • A consumer of skill s earns income z(s) so z-1(s) is the skill level associated to each income • If F(s) is the cumulative distribution of skill then G(z) = F(z-1(s)) is the cumulative distribution for income Two Specializations • Since revenue is maximized any small change in the tax function must have no effect on revenue • Consider a increase in the marginal rate of DT′ at income z • Tax payments increase from all those with income above z • Holding labor supply constant the total increase is [1 – G(z)]zDT′ • The tax increase reduces labor supply and leads to a revenue loss g(z)T′zesDT′/(1 – T′) where es is the elasticity of labor supply Two Specializations • At the optimum the gain must equal the loss [1 – G(z)]zDT′ = g(z)T′zesDT′/(1 – T′) • Solving this equation T′/(1 – T′) = [1 – G(z)]/esg(z) • This implies the marginal tax rate (T′) will be high at income z when – The labor supply elasticity is low – There are few taxpayers with income z • Even for Rawlsian taxation there will not be progressivity unless [1 – G(z)]/esg(z) increases in z Numerical Results • The theory describes some characteristics of the optimal income tax function • A numerical analysis is required to generate more precise results • Numerical results employ the social welfare function 1 eU W 0 e s ds, e 0 e 0 U s ds , e 0 • The social welfare function is utilitarian if e = 0 • Higher values of e give more concern for equity Numerical Results • The density function for the skill distribution is given by f(s) • This is assumed to be log-normal with a standard deviation of s = 0.39 – This value is similar to that for observed income distributions – But skill and income may not have the same distribution • The individual utility function is Cobb-Douglas U = log(x) + log(1 – ℓ) Numerical Results • Tab.15.2 presents the optimal tax rates for a utilitarian welfare function • The average rate of tax is negative for the lowskilled but increases with skill – The negative tax is an income supplement • The marginal tax rate first rises with skill and then falls. – The maximum rate is around the median of the skill distribution Income Consumption Average tax (%) Marginal tax (%) 0 0.03 - 23 0.055 0.07 -34 26 0.10 0.10 -5 24 0.20 0.18 9 21 0.30 0.26 13 19 0.40 0.34 14 18 0.50 0.43 15 16 Table 15.2: Utilitarian case (e = 0) Numerical Results • The results in Tab. 15.3 involve a greater concern for equity • The average tax rate starts lower but rises higher • The marginal tax rate is higher for all income levels • The marginal rate is highest at a low income level Income Consumption Average tax (%) Marginal tax (%) 0 0.05 - 30 0.05 0.08 -66 34 0.10 0.12 -34 32 0.20 0.19 7 28 0.30 0.26 13 25 0.40 0.34 16 22 0.50 0.41 17 20 Table 15.3: Some equity considerations (e = 1) Numerical Results • The outcome is a negative income tax with the government supplementing income • The maximum average rate of tax is low • The marginal tax rate first rises with income and then falls. – The system is not marginal rate progressive • The marginal rate of tax is close to constant – The consumption function is almost a straight line • The zero tax for the highest-skill consumer is reflected in the fall of the marginal rate at high incomes Tax Mix: Separation Principle • Governments use both income and consumption taxes • Chap. 14 showed that efficient commodity taxes should be inversely related to the elasticity of demand – This implies a system of differential commodity taxation • The question to address now is the role of differential commodity taxation when there is an optimal nonlinear income tax • The answer is dependent on the relation between commodity demand and labor supply Tax Mix: Separation Principle • Recall that the success of the income tax is limited by incentive compatibility – The high-skill will mimic the low-skill • Differential commodity taxes are justified if they relax the incentive compatibility constraint – This can be done by making the consumption bundle of the low-skill less attractive to the high-skill • If the utility function is separable between consumption and labor incentive compatibility cannot be relaxed – Separable utility has the form U = U(u(x), ℓ) Tax Mix: Separation Principle • Fig. 15.13 displays nonseparable preferences • Changing prices from p to p′ makes the consumption plan of the low-skill less attractive to the high-skill • The utility of the low-skill is not affected • Incentive compatibility is relaxed x2 I h2 I h1 I p p' x1 Figure 15.13: Differetial taxation and nonseparability Voting over a Flat Tax • The political process determine the tax system through voting • Assume skills are distributed with cumulative distribution F(s), mean s and median sm • A vote is taken over a linear tax with lump-sum benefit b and constant marginal tax rate t • Consumer preferences are represented by the quasi-linear utility function U = x – (1/2)(z/s)2 Voting over a Flat Tax • Given the budget constraint x = [1 – t]z + b the chosen income of a consumer with skill s is z(s) = [1 – t]s2 • The government budget constraint is b = tE(z(s)) = t[1 – t]E(s2) • Substituting for b and z in the utility function and maximizing over t gives the optimal tax of the median voter tm = (E(s2) – sm2)/(2E(s2) – sm2) Voting over a Flat Tax • Using the choice of income the tax can be written tm = (E(z) – zm)/(2E(z) – zm) • The model predicts the political tax rate is determined by the position of the median voter in the income distribution • As income inequality rises (E(z) – zm increases) the tax rate rises • In practice median income is below mean income so voters will vote for redistribution