Tax (Non)-Compliance Research: A New Zealand Perspective Norman - - PowerPoint PPT Presentation

Tax (Non)-Compliance Research:

A New Zealand Perspective

Norman Gemmell

Chair in Public Finance Victoria University of Wellington

Cash & Hidden Economy – International Revenue Conference Auckland, April 2014

www.nzpublicfinance.com

Introduction

Question: What do we know (and not know!) about taxpayers’ compliance responses to changes in tax policy and compliance efforts? Three research areas:

• 1. Taxpayers’ responses to taxation
• 2. Taxpayers’ responses to audits
• 3. Lessons for ‘tax gap’ measures

• 1. Taxpayers’ responses to

taxation

Taxpayers’ responses to taxation

An (Australian) illustration:

• Australian vehicle fringe benefit tax (FBT) applies to private kms. travelled

using company-provided vehicle.

• Till 2011, tax rate applied to kms travelled at three different tax rates:

lower tax rate if travel more (above 3 annual ‘kms travelled’ thresholds) Some ‘facts’: (Henry Review (2009) data for 2007-08)

• Total of around 15,000 cars (reported for FBT)
• On average FBT payers say they travel  22,000 (private) kms/year.
• Some as low as 3,000 kms., some over 40,000 kms.

Question: In absence of tax, what pattern of kms. travelled would we expect?  Perhaps a Normal Distribution around the average (22,000kms) ?

Behavioural responses: Australian Vehicle FBT

 Perhaps something like this …?

100 200 300 400 500 600 700 800 900 1,000 100 200 300 400 500 600 700 800 900 1,000

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000 41,000

Number of cars Number of cars Annualised distance travelled in the FBT year (km)

Number of vehicles by kilometers travelled

Total no. of cars  15,000

500 1,000 1,500 2,000 2,500 500 1,000 1,500 2,000 2,500 1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000 Number of cars Number of cars Annualised distance travelled in the FBT year (km)

Number of vehicles by kilometers travelled

Total no. of cars  15,000

Actually observe this…

15k 25k 40k

Three thresholds:

Lesson: Our ability to enforce compliance depends on tax policy design

Measuring taxpayers’ response ‘elasticities’

Economic researchers measure responses formally by the: ‘elasticity of taxable income’ (ETI) = percentage change in taxable income in response to a 1% change in the ‘net-of-tax’ rate (1 – t)

• e.g. if top income tax rate: t = 0.5 then (1 – t) = 0.5
• At t = 0.5:

ETI = 1 implies: a 10% decrease in t, increases taxable income by 10% [t: 0.50 to 0.45 ; (1- t): 0.50 to 0.55] ETI = 0.5 implies: a 10% decrease in t, increases taxable income by 5%

• ETI values are specific to each tax structure - depend on policy settings,

legal regime, compliance effort etc.

• International estimates of ETI  0.1 to >1.0. Generally around 0.2 – 0.4
• What about New Zealand’s ETI values?

Measuring NZ income tax responses

• NZ income tax: top rate raised in 2001 from 33% to 39% for

incomes > $60k ( 10% fall in (1 – t): 67% to 61%)

• Introduction of 39% rate:

Estimate ETIs for top income decile & percentile ETIs are higher in the short-run

(95% confidence interval in brackets)

Measuring NZ income tax responses

• NZ income tax: top rate raised in 2001 from 33% to 39% for

incomes > $60k ( 10% fall in (1 – t): 67% to 61%)

• Introduction of 39% rate:

Estimate ETIs for top income decile & percentile ETIs are higher in the short-run Recent ETI regression estimates: Average across all taxpayers 0.68 Taxpayers with ‘non-W&S income’ 0.51 Taxpayers without ‘non-W&S income’ 0.19 Taxpayers with non-W&S income: All taxable income 0.22 Non-W&S income 2.84

Measuring NZ income tax responses

• Other evidence of responses to tax rate/threshold changes:

‘Spikes’ in the income distribution …

200 400 600 800 1,000 1,200 1,400 1,600

20,000 23,000 26,000 29,000 32,000 35,000 38,000 41,000 44,000 47,000 50,000 53,000 56,000 59,000 62,000 65,000 68,000 71,000 74,000 77,000 80,000 83,000 86,000 89,000 92,000 95,000 98,000 101,000 104,000 107,000 110,000 113,000 116,000 119,000 122,000 125,000 128,000 131,000 134,000 137,000 140,000 143,000 146,000 149,000

Aggregate taxable income ($m) Taxable Income Band ($)

2002 1999

Measuring NZ income tax responses

• Other evidence of responses to tax rate/threshold changes:

‘Spikes’ in the income distribution …

500 1,000 1,500 2,000 2,500 $20,000 $25,000 $30,000 $35,000 $40,000 $45,000 $50,000 $55,000 $60,000 $65,000 $70,000 $75,000 $80,000 $85,000 $90,000 $95,000 $100,000 $105,000 $110,000 $115,000 $120,000 $125,000 $130,000 $135,000 $140,000 $145,000 $150,000 Total taxable income ($m) taxable income band

2010 2008

Measuring NZ income tax responses

• Other evidence of responses to 2001 top rate change:

The growth rate of top incomes & trusts after 2001

(Tax Working Group, 2009)

… Trustee income rises faster after 39% tax rate applies to beneficiary income (33% for trustees)

Measuring NZ income tax responses

• Other evidence of responses to 2001 top rate change:

The growth rate of top incomes & trusts after 2001

(Tax Working Group, 2009)

• 2. Taxpayers’ responses to audits

Taxpayers’ responses to audits

• Audits are designed to raise more revenue!
• Well-targeted audits should do just that!
• But with random audits? … more compliant taxpayers likely to be

audited

• When audits reach a verdict of ‘compliant’, how do those

taxpayers respond after the audit?

What if:

• The audit fails to find avoidance/evasion?
• The audit is only partially successful – identifies just some evasion?

 The audit process & the extent of audit ‘success’ conveys information to compliant and non-compliant taxpayers  Examine evidence from UK self-assessment income tax audits

Taxpayers’ responses to audits

• We expect different reactions to audits depending on:

– whether audit verdict is ‘compliant’ or ‘non-compliant’ – random audits are conducted ‘with/without replacement’

Gemmell & Ratto (2012):

• Examined: effect of random audits in 2000 on before/after declared

incomes [1997-99 to 2001-03]

• Compare audited ‘treatment’ group with ‘control’ group of non-audited.
• Examine medium & small businesses, and personal taxpayers
• Estimate ‘preventive yield’:

additional revenue (+ or -) resulting from audit.

Taxpayers’ responses to audits

• We expect different reactions to audits depending on:

– whether audit verdict is ‘compliant’ or ‘non-compliant’ – random audits are conducted ‘with/without replacement’

Taxpayers’ responses to audits

Taxpayers’ responses to audits

• 3. Lessons for tax gap measures

Tax gap measures

Tax gap measures capture differences between tax paid and ‘theoretical’ tax liability. G = T* – T = t(B* – B)

where B* (B) is the theoretical (observed) tax base; T = tax revenue

B = qB*

where q is the fraction declared for tax [(1 – q) is ‘hidden’] 

T = tqB* and G = t(1 – q)B*

• Conventional tax gap measures based on answering question:

“If q is increased to 1, how much ‘missing revenue’ would be raised?”

• But crucially assumes B* remains unchanged by changes in q

‘effective tax rate’

Tobacco Excise Example

But: Many cigarettes purchased illegally at $2 will no longer be bought when successful compliance effort pushes up the price to $3. i.e. Anyone who values cigarettes between $2 and $3 will smoke less 1 million @ $3 Sales $3 million Tax Revenue $1 million Missing Tax Revenue $ ½ million Tax gap $½ million 33% (0.5/1.5). If smuggled cigarette sales cut in half when extra $1 tax applied to all (& formerly legal cigarettes unaffected):  ‘True’ tax gap = $250,000. i.e. only half of conventional tax gap … depends on which taxpayers’ respond to changes in compliance effort. ½ million smuggled @ $2

Tax gap measures

Observed/declared tax base @ ‘D’ (when q = 0.5); Total tax base @ ‘C’ (= ‘H’)

Tax gap measures

Tax (base) gap = G (= HF) ; declared base = q1B* ; hidden base = (1 - q1)B*

Tax gap measures

A higher effective tax rate (due to higher q) reduces the tax base from H to E. Raising q reduces both tax gap measures to zero (at H or E) but quite different revenues

Tax gap measures

With a sufficiently large response, the tax base may be lower when q = 1 (E below F)

Tax gap measures

C E = “change in tax base in response to a change in the compliance rate” A form of ETI

Amending conventional tax gap measures

Where does this leave us?

• Are taxpayer responses to changes in q like their responses to changes in t?

– Evidence from audit responses etc. may help

• If so, ETI estimates &/or observed audit responses may give some clues to

the potential loss of total tax base when full compliance achieved.

• Conventional tax gap: ‘compliance ETI’ = zero. Surely this is wrong?
• For some taxes this clearly not true: smuggled goods (tobacco, alcohol),

corporate tax?

• Corporate tax: do some governments ‘turn a blind eye’ to tax avoidance

(e.g. profit-shifting) because otherwise the total tax base would be lower?

Conclusions

1. Research has now identified a variety of taxpayer behavioural responses to tax policy (e.g. tax rates) and compliance enforcement 2. Compliance enforcement, not just tax policy, should take these into account 3. The consequences of these behavioural responses can be quite different depending on:

• Are they ‘real’ (changed economic activity) or ‘apparent’ (‘rearranged’ activity)?
• Is avoided tax paid under a different (lower rate) tax?

4. Measuring tax gaps: finding hidden economic activity may reduce total economic activity (as well as the hidden share)

Questions?

www.nzpublicfinance.com