1、Taxes and Labour Supply Theory, Evidence and PolicyDavid Phillips (IFS),Outline,The Basics: Income, Substitution and Deadweight Loss Enriching the Model: Fixed costs and more 3. Measurement Issues: Econometric Problems 4. A Review of the Emprical Evidence Alternate Measures: Taxable Income Elasticit
2、ies Policies: Mini Jobs for Lone ParentsConclusions,The Basics,Labour Supply is the supply of effort and time by individuals for monetary compensation. Working out revenue implications of tax reforms Working out the optimal tax rate structure.Theoretical and empirical work has traditionally focused
3、upon hours of work but more recently a recognition of: The participation decision (extensive margin) Effort and Compensation FormBut what basic lessons can we use from Econ 101?,Income and Substitution Effects,H,l,NI,A,A,A,Higher tax rate reduces net income. If leisure is a normal good this reduces
4、demand for leisure and increases Labour Supply.,Higher tax rate reduces the marginal wage and hence the price of an extra hour of leisure. This acts to decrease Labour supply.,Tax has an ambiguous effect on Labour Supply,Abolition of the 10p Starting Rate And Cutting Basic Rate to 20p,H,l,NI,Old Tax
5、 Schedule,New Tax Schedule,What about Labour Supply?,Abolition of the 10p Starting Rate And Cutting Basic Rate to 20p,Those not working or earning less than the personal allowance are unaffected. Those previously paying starting rate have lower income (work more) but face a higher marginal rate (wor
6、k less) Ambiguous. Those paying basic rate and earning upto about 19,000 have lower income (work more) and a lower marginal rate (work more) Work MORE. Those paying basic rate and earning more than 19,000 have higher income (work less) but face a lower marginal rate (work more) - Ambiguous,Elasticit
7、y of Labour Supply,There are two types of elasticity we are interested in: The uncompensated elasticity. The compensated elasticity.Linked together by the Slutsky Equation:h = m - (w*h)/(w*T+y)* i h = Compensated Elasticity w = Wage m= Uncompensated Elasticity h = hours of work i = Income Elasticity
8、 T = time endowment y = Unearned Income,Tax Rates and Tax Revenues,The uncompensated elasticity determines revenue:Change in Revenue (1- m)*(%(1-t)w)*w*h*t,H,l,NI,A,A,Here a tax rise increases revenue but if labour supply is elastic would see a fall in revenue.,Tax Rates and Deadweight Loss,When sub
9、stitution occurs there is always deadweight loss. It is determined by the compensated elasticity.,The Hours Distribution,Now imagine a tax cut. The model predicts some not working will enter work and do 1 or 2 hours per week.,But we dont observe this. Why?,Adding Fixed Costs of Work,When you start w
10、ork there are certain fixed costs:Transport costs Work clothing costsChildcare costs And others,H,A,NI,A,Fixed costs mean not working better than working few hours -provided leisure and income are both goods,l,The Welfare System: Non Convexity,The welfare system (benefits and tax credits) mean that
11、people on low incomes can face very high effective marginal tax rates. This non convexity causes similar discontinuous labour supply responses.,Introducing non-convexities (e.g. WFTC) can encourage jumps in labour supply (e.g. From 10 to 30 hours),A,A,H,l,NI,The Family: Harmony or Bargaining?,Unitar
12、y Model: Family is a single unit with a single utility function and behaviour of all members chosen to maximise this. Uf(c, li, lj) This implies: Full income poolingSymmetry of response to changes in other partners wage. Neither Hold,Collective Model: Family as a group of individuals who engage in (
13、pareto-efficient) bargaining. This group of model allows one to extend models from just labour supply to looking at intra-household allocations of welfare.,Putting it together,hi = f(wi,wj, t, b, yi, yj, fi, fj, c, r)t = tax system b = benefit system f = fixed costs r = real interest rate Analysis o
14、f Labour supply cant be done by a simple OLS regression of hours on wages because of the complexity of the budget constraint and substitution over time and within the household.,Econometric Issues (1),You have data a cross section of hours, wages and demographic characteristics. But theres a problem
15、.People have different preferences for work/leisure.Those who like work are likely to work longer hours.They are also likely to work harder, engage in training more, have tried harder at school and are probably more able. Spurious positive correlation between hours and wagesIf taxes exhibit increasi
16、ng marginal rates, those working longer hours will face higher marginal rates and have lower net wages. Spurious negative correlation between hours and wages,Econometric Issues (2),Second Generation models attempted to account for the downward bias induced by increasing marginal rates of tax. very c
17、arefully modelled the tax system.assume utility maximising and rational behaviour.People bunch at the kinks as marginal rates change. Desired labour supply below kink at post-kink tax rate, and above kink at pre-kink tax rate. When you combine this with a linear labour supply modelRestricts so that
18、must estimate a positive uncompensated wage elasticity. H = + W + Y,Econometric Issues (3),Solutions: Use time series changes in wages (potentially a panel data series). Preferably have exogenous reforms e.g. Reduction of top rates of income tax, introduction of E.I.T.C.Remember to control for aggre
19、gate preference changes (e.g. Social acceptability of women working).E.G Blundell, Duncan & Meghir (1998)Avoid modelling the complex tax system by using only women paying basic rate (nearly linear budget constraint).Avoid preference problem by making use of changes in income distribution and tax ref
20、orm in the 1980s.Makes use in differential changes in post-tax wages for different cohorts and educational groups.,Econometric Issues (4),Solutions:Treat labour supply as discrete (e.g. full time, part time, not work) and estimate income at these points using full tax and benefit system.E.G Meghir &
21、 Phillips (2008)Men have a 1-0 choice of working or not, with income in work estimated as a weighted average of income at various hours points. Use predicted wages for both non-workers and workers to account for (spurious) correlation of hours of work and pre-tax wages due to preferences.Use the dif
22、ferential trends across regions in wages and housing benefits (out of work income) as exogenous changes in work incentives.,Elasticity Estimates (Female),Elasticity Estimates (Male),Other aspects of Labour Supply,So far we have focused on hours but what about other aspects of labour supply?intensity
23、 of effort (payment by results)effort to seek promotion investment in skills, education and complimentary capitalIf substitution is easier for effort/investment than for work-hours hours elasticities underestimate welfare cost of taxation. Problem: we cannot measure effort.,Taxable Income Elasticity
24、 (1),. but we can measure income.Income reflects both hours of work and effort per hour. It is a proxy for total effort. Two measures:Broad Income reflects changes in effort.Taxable Income reflects changes in effort but also shifts in income from taxable to non-taxable forms (tax avoidance) and tax
25、evasion. Reallocation of income also entails welfare costs, and of course, revenue implications.,Taxable Income Elasticity (2),Main method of calculating taxable income elasticities is difference-in-differences. Two groups one affected by a change in tax rates (H), the other not (L).Elasticity = dif
26、ference in % change in income . Elasticity = lnEH lnEL . ln(1-tH) - ln(1-tL),Taxable Income Elasticity (3),Problems with this methodology: Mean ReversionE = + t + Xi +Uit Some people in group H likely to have transitory high income, and vice versa. Revert to mean: expect high income to fall, low inc
27、ome to rise. Estimates of elasticity will be downwardly biased.General Equilibrium Labour of different prices is probably labour of different types to some extent. Increased supply of those benefiting from tax cut (e.g. High skilled) will depress price of this type, reducing income. Estimates of ela
28、sticity will be downwardly biased.,Taxable Income Elasticity (4),Problems with this methodology: Secular Trends in Skills prices The price of various types of Labour may be changing for reasons other than tax reforms e.g. trade or technological change. Varies, but in 1980s/1990s, elasticity will be
29、upwardly biased.Timing If a tax rise (or cut) is announced, people will bring forward (or put back) income if they have such flexibility e.g. Bonuses, dividends, capital gains. Estimates of elasticity upwardly biased unless account for these timing effects.,Elasticity Estimates,Reforming Support for
30、 Lone Parents,Fewer than 60% of lone parents employed versus about 70% of mothers in couples:Similar number of lone parents working 16+ hours.But far fewer lone parents working 1 15 hours. Why?No incentive to do so:Income Support is Tapered away at 100% above earnings of 20 a week.Housing Benefit an
31、d Council tax have tapers of 65% and 20%. When combined with income tax, tapers of over 95%. No working tax credit (WTC) unless you work 16 hours per week.,The Gains of Labour., are less for lone parents for 8 16 hours., and they face high effective tax rates,IFS and One Parent Families were asked b
32、y the Joseph Rowntree Foundation to look at ways of improve work incentives.Giving WTC to lone parents working less than 16 hours per week. Increasing the earnings disregard for income support and other means-tested benefits.Reducing the taper rate on income support and other means-tested benefits.-
33、 does more to encourage jobs with higher rates of pay.- but administratively more burdensome as still need to know about small changes in income.,Making mini jobs Pay.,Model used is:estimated using repeated cross sections from the Family Resources Survey (FRS) roughly 17,000 lone parents.uses cross-
34、sectional and time-series (policy reforms) variation in work incentives. discrete choice: not work, 1 15 hours, 16 23, 24 29, 30 37, 38+. income estimated at each of these choices.they take up all benefits and tax credits entitled to.use of childcare determined by family circumstances and hours not
35、price or subsidies. repeatedly draw a prediction for each lone parent to get weights for each hours choice, before and after reform.,Modelling Lone Parents Labour Supply.,Results,WTC reduces average hours, offsetting extra tax credits. Increased disregards also encourage 16 30 hours work, increasing
36、 average hours and earnings.,Conclusions (1),Income and substitution effects can give some important insights (e.g. the 10p tax).- But non-convexities and fixed costs means can see jumps in Labour supply in response to reforms.Compensated versus uncompensated elasticities- Tax rises have an ambiguou
37、s effect on labour supply and revenue but always entail deadweight loss.Estimation of model made difficult due to complex budget constraint and unobserved preferences for work.- Use exogenous variation in incentives due to secular wage changes or tax reforms as basis of estimation.,Conclusions (2),E
38、stimated responsiveness varies by group- Low for men- Higher for women, particularly lone parents- Participation decision particularly sensitiveTaxable Income Elasticities- Picks up effort and reallocation of income- But problems if simple diff-in-difference techniques used.Example: Lone Parents and
39、 Mini-jobs.- Lone parents face poor incentives to enter work at low hours.- Reforms to benefits and tax credits can have a big impact on labour supply for this sensitive group.,Main Readings,C. Meghir & D. Phillips (2008) Labour Supply and Taxes http:/www.ifs.org.uk/wps/wp0804.pdfK. Bell, M. Brewer
40、& D. Phillips (2007) Lone Parents and Mini-jobs http:/www.jrf.org.uk/bookshop/eBooks/2110-lone-parents-minijobs.pdf,Additional References,Aaberge, Colombino, Strom (1999) “Labour Supply in Italy: An Empirical Analysis of Joint Household Decisions with Taxes and Quantity Constraints“, Journal of Appl
41、ied Econometrics, Vol 14. No 4. Ashenfelter & Heckman (1974) “The Estimation of Income and Substitution Effects in a model of Family Labor supply“, Econometrica, Vol 42, No 1Blomquist & Newey (2002), “Nonparametric Estimation with Nonlinear Budget Constraints“, Ecometrica, Vol. 70, No. 6Blundell, Du
42、ncan, Meghir, (1992), “Taxation in Empricial Labour Supply Models: Lone Mothers in the UK“, The Economic Journal, Vol 102, No 411Blundell, Duncan, Meghir (1998) “Estimating Labor Supply Responses using Tax Reforms, Econometrica, Vol 66, No 4Bourguignon & Magnac (1990) “Labor Supply and Taxation in F
43、rance“, The Journal of Human Resources, Vol 25, No 3Cogan (1981), “Fixed Costs and Labor Supply“, Econometrica, Vol 49, No 4Feldstein (1995), “The Effects of Marginal Tax Rates on Taxable Income: a Panel study of the 1986 Tax Reform Act“, Journal of Political Economy, Vol 103, No 3Goolsbee (1999) “E
44、vidence on the High-Income Laffer Curve from Six Decades of Tax Reform“, Brookings Papers on Economic Activity, Vol 1999, No 2Gruber & Saez (2000) “The Elasticity of Taxable Income: Evidence and Implications“, NBER Working Paper SeriesHausman (1981), Labour Supply: How Taxes affect Economic Behaviour“, Tax and the Economy, Brookings Institute,
