Econ 551 Government Finance: Revenues Fall 2019 Given by Kevin - - PowerPoint PPT Presentation

ECON 551: Lecture 8b 1 of 44

Econ 551 Government Finance: Revenues Fall 2019

Given by Kevin Milligan Vancouver School of Economics University of British Columbia Lecture 8b: Taxing Lower Income Workers

ECON 551: Lecture 8b 2 of 44

Agenda

• 1. Overview
• 2. Basic models
• 3. Saez 2002
• 4. Evidence on in-work tax credits

ECON 551: Lecture 8b 3 of 44

Overview

Let’s recall the Mirrlees optimal income tax formula we derived a few weeks ago: 𝑈′ (1 − 𝑈′) = (1 + 1 𝜁) (1 − 𝐺(𝑥0)) 𝑥0𝑔(𝑥0) For low-ability workers:  Low wage (𝑥0), so productivity gain from their work not great.  Lots of workers above them in the distribution, so big revenue gain (1 − 𝐺(𝑥0))  Both these things push to higher marginal tax rates at low ranges. Also:  We want to redistribute toward them.  In presence of fixed transfer, may choose corner solution of no work.  Both these things push to lower marginal tax rates at low ranges. So…it’s complicated.

ECON 551: Lecture 8b 4 of 44

How could we tax low earners?

In most income taxes, there is basic amount which effectively leaves low earners untaxed or only very lightly taxed.  This suggests the income tax bracket/rate schedule not going to be particularly important. Instead, other features of our tax/transfer system are going to be much more important  Provincial social assistance schemes—how do they handle earnings?  Employment Insurance—how does it handle earnings?  Transfers that reduce with income—e.g. GST tax credit, or full negative income tax.  Earned income supplements—credits that provide multiplicative subsidy for earnings.

ECON 551: Lecture 8b 5 of 44

Example of an earned-income supplement: WITB (now CWB)

 Working Income Tax Benefit introduced in 2007.

• Now changed to Canada Workers Benefit

 Supplements earnings over a threshold at 21% rate.  Phased out once income over another threshold at 17%  In 2013 maxed out at $1230 (singles) $1952 (others). Many other countries have similar programs.  Earned Income Tax Credit (EITC) in US.  Working Tax Credit in UK. In Canada, many provinces have their own similar programs as well.

ECON 551: Lecture 8b 6 of 44

How WITB is calculated:

 Line 8 from Step 1 is a measure of your earned income.

 Line 15 from Step 1 is based on couple line 236 net income.  Note starting point of $4,750; pivot at $12,301/$16,579.  Note phase in and phase out rates.

ECON 551: Lecture 8b 7 of 44

WITB by earnings:

ECON 551: Lecture 8b 8 of 44

WITB vs. EITC:

ECON 551: Lecture 8b 9 of 44

Canada Child Benefit: Since 2016

 Basic benefit of $6,400/yr (<age 6) and $5,400/yr (>=age 6)  Phased out starting at $30,000.  Everyone under 150,000 (P90) better off.

ECON 551: Lecture 8b 10 of 44

All the refundable tax credits:

This adds up to $1,181 per month!

ECON 551: Lecture 8b 11 of 44

Questions we will try to address:

What is sensitivity of work decision to these incentives? Is intensive (hours) or extensive (in/out) decision more important?

ECON 551: Lecture 8b 12 of 44

Agenda

• 1. Overview
• 2. Basic models
• 3. Saez 2002
• 4. Evidence on in-work tax credits

ECON 551: Lecture 8b 13 of 44

Review: The static labour supply model

 Transfer of A.  Wage rate of w0.

leisure After- tax income l=24 A U0 Slope w0 Z

ECON 551: Lecture 8b 14 of 44

Add in a WITB-style credit

Benefit phase-in rate: s Benefit phase-out rate: t

leisure After- tax income l=24 A Slope w0 B C D E Phase-in range. Slope w0 (1+s) Receiving max benefit; not yet phasing out. Slope w0 Phase-out range. Slope w0 (1-t)

ECON 551: Lecture 8b 15 of 44

What happens to a corner-solution guy?

Moves from corner solution at A to point C with introduction of EITC.  With any convex expansion of budget set, will (weakly) bring more off A and into work.

leisure After- tax income l=24 A B C D E U0 U1

ECON 551: Lecture 8b 16 of 44

What happens to an interior-solution guy?

Could go anywhere, depending on where Z is, and nature of preferences.  In parallel segment DC there is just income effect—less work.  In BC and DE, there are income and substitution effects; potentially offsetting.

leisure After- tax income l=24 A B C D E U0 Z ??

ECON 551: Lecture 8b 17 of 44

Agenda

• 1. Overview
• 2. Basic models
• 3. Saez 2002
• 4. Evidence on in-work tax credits

ECON 551: Lecture 8b 18 of 44

Saez (2002): Transfers to low-income earners

Question asked by paper: under what conditions should we prefer a Negative Income Tax vs. an Earned Income Tax Credit?  Negative Income Tax: Give an initial transfer to zero earners; tax it back gradually.  EITC: Wage subsidy for lower earners, eventually clawed back as incomes grow.

ECON 551: Lecture 8b 19 of 44

Participation vs intensive elasticities

Key insight of model:  With fixed cost of work, may be optimal for some to choose corner solution of no work.  But, with right incentives, they may ‘jump’ into workforce.  If things aren’t continuous and smooth in the labour market, we need to augment Mirrlees (1971) to account for these kinds of costs and jumps. Definitions: Participation tax rate:  What is the change in taxes between working at a certain point and not working, divided by the wage.  Not distinct from an ‘average tax rate’. Marginal tax rate:  What you pay on the next dollar of earnings. Participation and intensive margin elasticities follow in the same way.

ECON 551: Lecture 8b 20 of 44

Optimal income taxes with participation margin

Here’s a stripped down version of what’s in Saez (2002) that captures the essential insight.  Individual has skill z  Tax function 𝑈(𝑨). After-tax income 𝑨 − 𝑈(𝑨).  With no work, gets −𝑈(0).  No savings, so after tax income is consumed.  Fixed cost of work q.  Utility given by 𝑣 = 𝑑 − 𝑟. (Consumption less cost of work) The work decision rule is very simple: If 𝑨 − 𝑈(𝑨) + 𝑈(0) > 𝑟 then work. If 𝑨 − 𝑈(𝑨) + 𝑈(0) < 𝑟 then don’t work. If q is distributed according to CDF 𝑄(𝑟|𝑨), then number of individuals who work is 𝑄(𝑨 − 𝑈(𝑨) + 𝑈(0)|𝑨)

ECON 551: Lecture 8b 21 of 44

Implementing the perturbation approach:

Let’s define an elasticity: how does participation change with net return to work? 𝜃(𝑨) = 𝑨 − 𝑈(𝑨) + 𝑈(0) 𝑄 ∙ 𝜖𝑄 𝜖𝑟 Consider a small increase in taxes at a particular skill level z: 𝑒𝑈.  This affects only those with skill level z. There are three effects we need to balance for the optimal tax rate to be in equilibrium:

• 1. Mechanical effect: higher taxes for those at skill z who still work.
• 2. Welfare effect: we care that those at skill z and working are paying higher taxes.
• 3. Behavioural effect: some workers at skill z now choose not to work.

In equilibrium, all three of these should balance, so that any change nets to zero across these three effects: 𝑒𝑁 + 𝑒𝑋 + 𝑒𝐶 = 0 Let’s walk through these one at a time.

ECON 551: Lecture 8b 22 of 44

Mechanical effect

We get extra tax revenue from those who still work. 𝑒𝑁 = 𝑄(𝑟|𝑨)𝑒𝑈 We get the change in taxes dT multiplied by the proportion of those at skill z who are working.

ECON 551: Lecture 8b 23 of 44

Welfare effect

Imagine we value a dollar distributed to workers of skill z according to 𝑕(𝑨). Welfare goes down because of this tax increase. 𝑒𝑋 = −𝑒𝑁 ∙ 𝑕(𝑨) = −𝑄(𝑟|𝑨)𝑒𝑈𝑕(𝑨)

ECON 551: Lecture 8b 24 of 44

Behavioural effect

Raising the tax rate at skill level z causes some workers to drop out as they can no longer cover their fixed cost of work. This costs us revenue. To calculate the behavioural effect, we need to multiply the number of workers affected by the fiscal cost for work dissuaded. First, the number of workers. How many drop out? Drop out if dT pushes you over your q. Those between 𝑨 − 𝑈(𝑨) + 𝑈(0) − 𝑒𝑈 and the original 𝑨 − 𝑈(𝑨) + 𝑈(0) drop out. There are 𝑒𝑈

𝜖𝑄 𝜖𝑟 such workers. Now substitute in using the elasticity definition.

𝑒𝑈 𝜖𝑄 𝜖𝑟 = 𝑒𝑈𝜃𝑄 𝑨 − 𝑈(𝑨) + 𝑈(0) The cost of this is therefore: [𝑈(𝑨) − 𝑈(0)] ∙ 𝜃𝑄(𝑟|𝑨)𝑒𝑈 𝑨 − 𝑈(𝑨) + 𝑈(0)

ECON 551: Lecture 8b 25 of 44

Add up the effects

𝑒𝑁 + 𝑒𝑋 + 𝑒𝐶 = 0 𝑄(𝑟|𝑨)𝑒𝑈 − 𝑄(𝑟|𝑨)𝑒𝑈𝑕(𝑨) − [𝑈(𝑨) − 𝑈(0)] ∙ 𝜃𝑄(𝑟|𝑨)𝑒𝑈 𝑨 − 𝑈(𝑨) + 𝑈(0) = 0 Which reduces to: 1 − 𝑕(𝑨) − [𝑈(𝑨) − 𝑈(0)] ∙ 𝜃 𝑨 − 𝑈(𝑨) + 𝑈(0) = 0 Further reducing: [𝑈(𝑨) − 𝑈(0)] 𝑨 − 𝑈(𝑨) + 𝑈(0) = 1 𝜃 (1 − 𝑕(𝑨))

ECON 551: Lecture 8b 26 of 44

Interpretation

Define the participation tax rate as 𝑢(𝑨) = [𝑈(𝑨) − 𝑈(0)] 𝑨 We can now rewrite the tax formula as 𝑢(𝑨) 1 − 𝑢(𝑨) = 1 𝜃 (1 − 𝑕(𝑨)) This comes out as a very simple inverse elasticity rule, multiplied by a term accounting for the social value of workers of type z.  The higher is the participation elasticity, the lower we want the tax rate to be.

ECON 551: Lecture 8b 27 of 44

Intensive margin

Saez goes through the same analysis for the intensive margin. A key parameter of the intensive side is the intensive elasticity: 𝜂. In the appendix, this is then mapped into a more standard labour supply elasticity that will be used in the simulation of the equilibrium. 𝜁 = (1 − 𝑈) 𝑨 𝜖𝑨 𝜖(1 − 𝑈)

ECON 551: Lecture 8b 28 of 44

The simulations

The next step is the simulations of the equilibrium. The important parameters are: 𝑤: controls redistribution. +∞ is Rawlsian. 0 is equal weight. 𝜁𝑀: Intensive elasticity for those with earnings < 20,000 𝜁𝐼: Intensive elasticity for those with earnings > 20,000 𝜃: Participation elasticity.

ECON 551: Lecture 8b 29 of 44

ECON 551: Lecture 8b 30 of 44

Comments:

On the left-hand side:  Bottom vs. top has different redistribution.  With high redistribution on bottom, little else matters. Just big transfer.  On top, we see a more EITC-like shape arise when 𝜃 gets bigger. Across the top:  RHS has larger 𝜁𝑀.  Not a big difference, but slightly lower transfer On the bottom:  RHS has super low redistribution tastes  We see low transfers.  Also, EITC shape emerges with higher participation elasticity.

ECON 551: Lecture 8b 31 of 44

Agenda

• 1. Overview
• 2. Basic models
• 3. Saez 2002
• 4. Evidence on in-work tax credits

ECON 551: Lecture 8b 32 of 44

What does the evidence say?

 Empirical evidence on EITC started in mid 1990s.  Many studies in many countries over last 18 years.  Hotz and Scholtz (2003) review of literature makes 4 conclusions i) Positive effect on labour-force participation of single-parent households. ii) Participation effects substantial. iii) Small negative effect on second-earner participation in two-parent families iv) Small negative effect on hours; but evidence mixed.  Canadian evidence: Milligan and Stabile (2007)

• Looks at National Child Benefit Supplement introduced in 1998.
• Some provinces subtracted (clawed back) benefit from social assistance; others not.
• When clawed back from SA, only receive value of benefit when not on SA; when

working.

• Effectively allows you to carry some of your SA cheque with you if you work.

ECON 551: Lecture 8b 33 of 44

Labour supply under National Child Benefit

•  AB is social assistance. BC section is 100% tax rate of earnings under social assistance.

 Would prefer B to X0 without NCB. Does not work.  With EFG added to budget constraint, now prefers point X1. Works.

ECON 551: Lecture 8b 34 of 44

Benefit structure of National Child Benefit

 Different amounts by number of children.  Introduced in 1998.  Some provinces had clawback of social assistance; some didn’t.

ECON 551: Lecture 8b 35 of 44

What impact on Social Assistance receipt?

(Transfers major source 1/0 indicator)

ECON 551: Lecture 8b 36 of 44

What impact on Earnings receipt?

(Earnings major source 1/0 indicator)

ECON 551: Lecture 8b 37 of 44

Estimation strategy:

Implement a triple difference strategy with an instrumental variable.  Sample of single mothers.  Three dimensions: province, year, number of kids.  Control for main effects; 2nd order interactions.  Calculate individual benefits based on observed income. 𝑍

𝑞𝑧𝑙𝑗 = 𝛾0 + 𝐶𝑓𝑜𝑓𝑔𝑗𝑢𝑞𝑧𝑙𝑗𝛾1 + 𝑌𝑞𝑧𝑙𝑗𝛾2 + 𝑞𝑠𝑝𝑤𝑞𝛾3 + 𝑧𝑓𝑏𝑠 𝑧𝛾4 + 𝑙𝑗𝑒𝑡𝑙𝛾5 + 𝑓𝑞𝑧𝑙𝑗

Endogeneity: Benefits depend directly on income.  A high income person has no benefits.  A low income person has high benefits.  OLS will give a strong negative 𝛾1. Instrumental variables:  Form 𝐶𝑓𝑜𝑓𝑔𝑗𝑢𝑞𝑧𝑙 on p,y,k cells by putting a sample through tax calculator for each p,y,k combination and taking the cell mean.  Isolates the legislative variation.

ECON 551: Lecture 8b 38 of 44

Main Results:

ECON 551: Lecture 8b 39 of 44

New paper: Kleven (2019) The EITC and the Extensive Margin: A reappraisal

Re-examines every state/federal EITC expansion:  Compares single women with and without kids.  Uses CPS like others have.  An explicit event-study framework. What he finds:  Previous work has focused on 1993 reform.  Finds 1993 reform confounded by other factors, like welfare reform.  Why? Informational frictions…

ECON 551: Lecture 8b 40 of 44

ECON 551: Lecture 8b 41 of 44

ECON 551: Lecture 8b 42 of 44

ECON 551: Lecture 8b 43 of 44

ECON 551: Lecture 8b 44 of 44

Next time:

Taxing high earners. Look particularly at: Milligan and Smart (2015) “Taxation of Top Incomes in Canada”, Canadian Journal of Economics. link