Lecture 4: Taxation and income distribution Public Economics 336/337 University of Toronto Public Economics 336/337 (Toronto) Lecture 4: Income distribution 1 / 33
Introduction In recent years we have been increasingly aware that changes in income distribution are concentrated at the top of the distribution the one percent. This raises a number of questions what is happening, and what is causing it? inequality trends who are the 1%? 0.1%? etc. structural economic changes should taxes on the rich rise? requires a theory of optimal taxation and comparative statics of optimum understand potential benefits and costs of tax increases at the top Public Economics 336/337 (Toronto) Lecture 4: Income distribution 2 / 33
Income inequality in Canada: Trends Some observations (based on Lemieux, Riddell and St-Hilaire, 2016): Market income inequality is high and rising average person in the top 10 per cent earns 20 times that of average in bottom 10 per cent Tax/transfer policies reduce inequality a lot but not the trend towards greater inequality Economic growth has favoured the rich have earnings grown for the median family? depends on your baseline... earnings fell in bottom decile 1979-1999 income growth concentrated in top 1% and above Public Economics 336/337 (Toronto) Lecture 4: Income distribution 3 / 33
Trends in market income inequality 40 Share of total income received by top and bottom fractiles Canada, 1982-2014 Source: Cansim 204-0002 35 30 25 20 15 10 5 0 Top 1 % Top 10 % Bottom 50 % Public Economics 336/337 (Toronto) Lecture 4: Income distribution 4 / 33
Effects of the tax-transfer system 65000 Median incomes, 1976-2015 Economic families, 2015 real dollars Source: Cansim 206-0011 60000 55000 50000 45000 40000 1976 1981 1986 1991 1996 2001 2006 2011 Market income Total income After-tax income Public Economics 336/337 (Toronto) Lecture 4: Income distribution 5 / 33
Trends in inequality 0.5 Income Inequality, Canada, 1976-2011 Gini coefficient, Before and after taxes and transfers, Economic families Source: Cansim 202-0709 0.45 0.4 0.35 0.3 0.25 0.2 1976 1981 1986 1991 1996 2001 2006 2011 Market income Total income After-tax income Public Economics 336/337 (Toronto) Lecture 4: Income distribution 6 / 33
Income trends for the very rich The survey data show an increase in incomes concentrated among the top decile. But who in this group, and why? Survey data cannot tell us about the very richest in society there are not enough of them to be reliably sampled. Instead, use data from income tax returns. Public Economics 336/337 (Toronto) Lecture 4: Income distribution 7 / 33
Worldwide trends in top income inequality Source: Atkinson et al. (2013) Public Economics 336/337 (Toronto) Lecture 4: Income distribution 8 / 33
Worldwide trends in top income inequality Source: Atkinson et al. (2013) Public Economics 336/337 (Toronto) Lecture 4: Income distribution 9 / 33
How rich is rich? What does it take to get into the top 1% top 0.1%? Table: Top income thresholds by country, 2014 Fractile Australia Canada (2010) US p50 30,203 p90 82,538 85,993 100,774 p95 110,453 111,884 141,438 p99 204,649 217,560 349,807 p99.9 586,253 742,373 1,603,811 p99.99 2,750,217 8,587,335 Note: Total income for tax purposes, 2014 CAD at PPP exchange rates. Source: http://wid.world Public Economics 336/337 (Toronto) Lecture 4: Income distribution 10 / 33
Trends at the very top Source: Milligan and Smart (2015) Public Economics 336/337 (Toronto) Lecture 4: Income distribution 11 / 33
Income growth hockey stick in US, Canada Source: Lemieux and Riddell (2015) Public Economics 336/337 (Toronto) Lecture 4: Income distribution 12 / 33
Top incomes and economic policy The top 1% are very rich but by definition, it s not a lot of people! Should we care about the incomes of this very small group of very rich people? 1 understanding broader economic changes In last 30 years, 28% of aggregate income growth went to top 5%, just 16% of growth went to bottom half! 2 politics of redistribution economic and political consequences of the top surge? 3 revenue considerations the rich are important source of revenue Top 1% pay 21% of PIT revenues (top 0.1%: 7.5%) the rich are hard to tax their response to taxes appears much higher better accountants? Public Economics 336/337 (Toronto) Lecture 4: Income distribution 13 / 33
Patterns in top income inequality: Causes and correlates What factors might be causing the rise in top income inequality? 1 Quantity factors: 2 Price factors: Public Economics 336/337 (Toronto) Lecture 4: Income distribution 14 / 33
Patterns in top income inequality: Causes and correlates What factors might be causing the rise in top income inequality? 1 Quantity factors: capital ownership and rise in r/w human capital 2 Price factors: stars, globalization pressures, and winner take all markets CEO compensation and bargaining for rents tax changes and tax avoidance responses Public Economics 336/337 (Toronto) Lecture 4: Income distribution 15 / 33
Who are the top one per cent? Population characteristics, Canada, 2011 Top 1% All Earners (percent) Men 79.7 49.1 Graduate degree 35.6 7.8 Industry: Finance insurance and real estate 14.9 4.2 Business services 19.2 6.8 Health and social service 13.3 8.3 Occupation: Management 36.3 7.1 Science 16.0 5.0 Health 13.8 4.5 Art, culture and sport 1.2 2.5 Source: Lemieux and Riddell (2016) Public Economics 336/337 (Toronto) Lecture 4: Income distribution 16 / 33
Being rich: Not just for capitalists anymore 70 Share of top incomes from wages and salaries Share of total income including capital gains, Canada, 1982-2014 Source: Cansim 204-0002 60 50 40 30 20 10 0 1982 1987 1992 1997 2002 2007 2012 Top 0.01 percent income group Top 1 percent income group Public Economics 336/337 (Toronto) Lecture 4: Income distribution 17 / 33
The very top in Canada and US: Evidence for brain drain? Source: Veall (2012) Public Economics 336/337 (Toronto) Lecture 4: Income distribution 18 / 33
Types of tax equity When we evaluate fairness of tax system, we typically look to two principles. 1 vertical equity: individuals with higher welfare (income) should pay higher tax can be justified by ethical notions of fairness, diminishing marginal utility of income, etc. 2 horizontal equity: individuals who are equally well off before taxes should pay the same tax taxation should not be arbitrary or capricious some preferences and choices should not be considered e.g. chocolate vs. vanilla ice cream lovers defining this is hard... Public Economics 336/337 (Toronto) Lecture 4: Income distribution 19 / 33
Measuring progressivity How do we measure vertical equity/progressivity of tax system? How is it related to work disincentives? For any tax system, let T (Y ) be tax paid at pre-tax income Y. Define: average tax rate: A(Y ) = T (Y )/Y marginal tax rate: M(Y ) = dt (Y )/dy Definition: A tax system is: progressive proportional regressive if A(Y ) is increasing in Y independent of Y decreasing in Y Graded exercise A flat-rate income tax pays a benefit B and has a constant marginal tax rate t. Write down T (Y ). Under what conditions is this tax progressive? Public Economics 336/337 (Toronto) Lecture 4: Income distribution 20 / 33
Tax rates on top earners: The Diamond-Saez formula The rich are not like you and me: They are a very good source of tax revenue. Are current top tax rates at appropriate levels? How should tax policy change with the surge at the top? Following Diamond and Saez (2011), we derive an operational formula for the optimal top marginal tax rate. Key consideration: avoidance of personal income tax, e.g. through changes to work and savings behaviour reclassification of income to lightly taxed forms (stock options, capital gains, business income) shifting income to low-tax jurisdictions Public Economics 336/337 (Toronto) Lecture 4: Income distribution 21 / 33
The rate increase dτ causes a mechanical transfer of (z z )dτ from the average high-income taxpayer to government revenues. If the social value of consumption by top earners is g < 1, then the average net gain to the transfer is (1 g)(z z )dτ Public Economics 336/337 (Toronto) Lecture 4: Income distribution 22 / 33
The increase in MTR causes a behavioural response and decline in revenue on average equal to τdz = τezdτ/(1 τ) where is the elasticity of taxable income. e = dz/z dτ/(1 τ) Public Economics 336/337 (Toronto) Lecture 4: Income distribution 23 / 33
At the optimum, the net gain is zero: (1 g)(z z τ )dτ ez 1 τ dτ = 0 Simplifying, τ 1 τ = 1 g z z e z Let 1/a denote the second term, which is roughly constant for empirical distributions at large z. (a is the Pareto parameter.) Then τ = 1 g 1 g +ae We have a formula for optimal tax as a function of inequality a, elasticity e, and social welfare judgement g. Exercise: When g = 0, what is this a formula for? Public Economics 336/337 (Toronto) Lecture 4: Income distribution 24 / 33
Tax rates and tax avoidance Faced with higher tax rates, the rich may reduce taxable incomes through real responses: e.g. work less, invest less, migrate out of jurisdiction avoidance responses: e.g. reclassify income, evasion Incorporating these responses, taxable income is a function Y (1 τ). Feldstein (1999): ETI is relevant to inefficiency of taxes, not just revenues When tax base shrinks, deadweight costs of tax avoidance rise by same amount as tax paid drops So the behavioural effect of taxes on revenues also measures the marginal excess burden of taxation: the deadweight economic costs of tax avoidance It doesn t matter if responses are real or avoidance : deadweight costs are the same. Public Economics 336/337 (Toronto) Lecture 4: Income distribution 25 / 33
Avoidance responses: An example Public Economics 336/337 (Toronto) Lecture 4: Income distribution 26 / 33
Tax rates and tax avoidance: International evidence How large are tax responses? Piketty, Saez, Stantcheva (2014) compare tax rates and top shares in 18 OECD countries plot in 1960-64 period, 2005-09 period and change within countries over time evidence for a causal effect of taxes? surge in top share where tax rates have declined most some evidence for increasing tax response over time Public Economics 336/337 (Toronto) Lecture 4: Income distribution 27 / 33
Estimating the ETI How can we estimate the ETI: Look at response to tax reforms 1 Time series approach: regress total taxable income of top 1% on tax rates What s the problem with this? logy Ht = a +e log(1 τ Ht ) +u t 2 Difference in difference approach: We know that earners outside the top 1% are less responsive to taxes - good control group logy Ht logy Lt = a +e log(1 τ Ht ) +u t (Note log(y Ht /Y Lt ) log(y Ht /(Y Ht +Y Lt )) = logs t approx. the log share of income accruing to the top one percent so this is known as the share approach to estimating ETI.) What s the problem with this? Public Economics 336/337 (Toronto) Lecture 4: Income distribution 28 / 33
Public Economics 336/337 (Toronto) Lecture 4: Income distribution 29 / 33
Public Economics 336/337 (Toronto) Lecture 4: Income distribution 30 / 33
Public Economics 336/337 (Toronto) Lecture 4: Income distribution 31 / 33
Implications for tax policy Recall our formulas for the effects of a tax increase: mechanical effect: M = (z z )dτ behavioural effect: B = ez τ 1 τ dτ Given data on inequality in each province we can simulate the effect of a tax increase, and the revenue maximizing rate. How strong do you think is the case for a tax increase? Graded exercise Fill out the missing entries in the following table. Public Economics 336/337 (Toronto) Lecture 4: Income distribution 32 / 33
Figure: Revenue simulations for Canadian provinces (e = 0.51) Top 1% Incomes Pareto Top Effect of a 1% increase in MTR: Rev Max Province Average Threshold a MTR M B Total MTR ($000) ($000) (%) ($ per taxfiler) (%) Newfoundland 332.2 206.5 2.6 43.0 1257.0-1279.7-22.7 42.6 PEI 229.2 156.4 3.1 46.9 Nova Scotia 299.3 176.0 2.4 49.4 New Brunswick 263.3 162.2 2.6 42.8 Quebec 357.2 184.6 2.1 47.7 Ontario 501.3 233.3 1.9 46.3 Manitoba 333.0 181.2 2.2 46.0 Saskatchewan 356.4 209.5 2.4 43.5 Alberta 739.9 325.3 1.8 39.0 British Columbia 419.8 211.2 2.0 43.1 Public Economics 336/337 (Toronto) Lecture 4: Income distribution 33 / 33