Private Pension Savings and the Taxability of High Income Earners Christopher Heady School of Economics University of Kent International Conference on Taxation Analysis and Research London, 1st – 2nd December 2011
Introduction • The share of top income recipients in gross incomes has increased strongly. • Should governments respond by increasing top rates of personal income tax? • Problem of incentive effects, represented by the elasticity of taxable income. • This elasticity is affected by tax planning, including use of tax-preferred pensions. • How important is this effect?
How should tax policy respond? • Optimal tax theory says that tax rates should be higher with more inequality • But we might not have started at the optimum • Need to worry about incentive effects, as summarised by the elasticity of taxable income, capturing: • Incentives to earn more money • Incentives to avoid and evade taxes
Increase top income tax rates? • Raising these taxes is clearly redistributive • But the marginal tax elasticities of taxable income suggest that raising the rates will have limited revenue-raising effect. • However, this may be over-stated: • The estimates do not take account of increases in other tax bases or in the future • Elasticity can be reduced by limiting opportunities for tax planning and evasion
Private pensions as an example of tax avoidance • The use of tax-preferred savings reduces tax liabilities during working life but increases them during retirement. • This is a pattern that is common to many other tax avoidance strategies (e.g. keeping income in corporations) • The elasticity of taxable income only captures the reduction in taxes. • How important are these effects?
Modelling private pension behaviour • Need to model choice between pension and non-pension savings • Empirically important • To have more than inter-temporal substitution • This requires pensions to have a disadvantage, to set against tax gains: • Lock-in of savings, so the funds cannot be used to meet short-term contingencies • Modelled by a random adverse shock
A very simple model • Taxpayer lives for two periods: • Work and retirement • Taxpayer has fixed income in period 1 but chooses pension and non-pension savings • Sequence of events: • Taxpayer chooses pension savings • Random shock is either realised or not • Taxpayer chooses non-pension savings • Choices result in consumption in periods 1 and 2
Some more very simple detail • There is no return on savings and no pure time preference • Utility is sum of expected value of concave functions of consumption in each period • The shock is a fixed amount, ε, which occurs with probability (1-π) • There are no bequests • Tax is levied on earnings minus pension savings, and (at a lower rate) on pension income
Analytical results • If the shock does not occur, consumption is the same in each period because marginal savings are not tax-preferred • If the shock does occur, consumption is different in the two periods because: • Marginal savings are tax-preferred (there is no non-pension savings in this case) • It is not rational to save enough to fully compensate for the (uncertain) adverse shock
Numerical examples • The analytical results do not show how important these effects are • Parameter values: • Income in period 1 = 150 • Shock (ε) = 50 • Top marginal tax rate, t1 = 40% or 50% • Income at which top marginal rate starts = 90 • Tax paid on income below top tax bracket = 30 • Flat rate tax on pension income = 20% • Utility function is constant elasticity of marginal utility, α = 1 or 2
Pension savings results • Pension savings increase with tax rate • This could account for a large part of observed difference between elasticities of broad and taxable income • Effect is larger for smaller αbecause of implied larger elasticity of substitution
Tax revenue results • Tax revenues increase with tax rate in both periods although rate only increased in period 1 • Increase relatively more in period 2 with higher elasticity of substitution (low α)
Implications for optimal rates • Based on US data, the revenue-maximising top marginal tax rate is: • 53.4% using standard (Gruber and Saez, 2002) estimate of elasticity of taxable income • Adjusting elasticity to account for extra tax revenue during retirement (assuming α = 1) increases rate to 55.2% • Preventing additional tax-preferred pension contributions of top tax rate payers increases rate to 66.0%
Conclusions • Tax-preferred pensions can be a major form of tax avoidance • Effects on optimal tax rates of recognising future revenue increase is small. • But effect of limiting tax–preferred pension contributions could be substantial • Need for more detailed modelling • More time periods could increase tax avoidance • But, closing down one method of tax avoidance might increase others