Using Movement of Exemption Cutoff to Estimate Tax Evasion: Evidence from Pakistan Mazhar Waseem University of Manchester April 2016
Introduction • How large is tax evasion in developing economies? • Extremely limited evidence because evasion is not observed • Approaches that uncover evasion credibly are costly 1 / 50
Existing Approaches • Randomized audit programme • Used by IRS and academic researchers (Kleven et al. 2011; Slemrod et al . 2001) • Problems • Not all tax evasion can be detected by audit • Political and administrative costs • Natural experiments • Used by Marion & Muehlegger (2008); Gorodnichenko et al . (2009) • Problems • Needs special tax variation or institutional features 2 / 50
This Paper • Contributes a simple approach to estimate evasion from behavioural responses to tax reforms • Estimates that at least 70% of self-employment and 1% of wage income is evaded in Pakistan 3 / 50
Outline Introduction Conceptual Framework Institutional Background Testing Predictions of the Model Estimating Tax Evasion Robustness and Heterogeneity Conclusions
Outline Introduction Conceptual Framework Institutional Background Testing Predictions of the Model Estimating Tax Evasion Robustness and Heterogeneity Conclusions
Estimating Tax Evasion • Two main challenges in estimating tax evasion from behavioural responses to tax changes 1 Real and evasion margins are conflated 2 Even if evasion margin can be separated only changes and not levels can be identified • I tackle these challenges by exploiting the asymmetry of effort and evasion costs 4 / 50
Effort and Evasion Costs • Disutility of effort is a smooth continuous function of effort • Costs of evasion are reverse-L-shaped: extremely low up to a given threshold and then rise sharply with evasion • The probability a government can detect evasion is low if income is self-reported and close to one if it is third-party reported • it is rarely feasible to evade completely even if all income is self-reported 5 / 50
Evasion Costs g � e � g e e 6 / 50
Model Setup • Standard model of labour supply and tax compliance • Agents work l hours at fixed wage rate w • Can evade e units of income at resource costs � if e ≤ e g Γ( e ) = g ( e ) if e > e . • Choose l and e to solve max u ( c, l, e ) = c − ψ ( l ) − Γ( e ) l,e s . t . c = (1 − τ )( wl − e ) + e 7 / 50
Optimal Evasion • Assume threshold e is large relative to fixed cost g g ≤ τe • Utility maximization ψ ′ ( l ) = w (1 − τ ) g ′ ( e ) ≥ τ • Optimal Evasion � 0 if τ = 0 e ( τ ) = e + κ ( τ ) if τ > 0 8 / 50
Predictions 1 Reported earnings will be a discontinuous function of tax rate � if τ = 0 wl 0 z ( τ ) = wl ( τ ) − e − κ ( τ ) if τ > 0 2 Evasion will be high even when tax rate is fairly close to zero e ( τ = ǫ ) = e ǫ > 0 9 / 50
Predictions τ τ B ' ∆ z A τ A τ B ∆ z B z( τ ) wl 0 -e wl 0 10 / 50
Estimating Tax Evasion • Consider two discrete tax reforms ∆ τ A ( τ A → 0) and ′ ∆ τ B ( τ B → τ B ) • Earnings responses induced by these reforms are given by ∆ z A ( τ A → 0) = w ∆ l ( τ A → 0) − ∆ e ( τ A → 0) ′ ′ ′ ∆ z B ( τ B → τ B ) = w ∆ l ( τ B → τ B ) − ∆ e ( τ B → τ B ) 11 / 50
Estimating Tax Evasion • If the two tax reforms are • similar-sized ∆ τ A = ∆ τ B = ∆ τ • small ∆ τ ≈ 0 ′ ′ • sufficiently close to each other τ A ≈ τ B • Then ′ ∆ l ( τ A → 0) ≈ ∆ l ( τ B → τ B ) ′ ∆ κ ( τ A → 0) ≈ ∆ κ ( τ B → τ B ) • And ′ ∆ z A ( τ A → 0) − ∆ z B ( τ B → τ B ) ≈ e 12 / 50
Outline Introduction Conceptual Framework Institutional Background Testing Predictions of the Model Estimating Tax Evasion Robustness and Heterogeneity Conclusions
Personal Income Tax in Pakistan • Two income tax schedules one each for self-employed and wage earners • Schedules not indexed to inflation and bracket boundaries are moved every two to three years to avoid bracket creep • Between 2006 and 2011 • Both schedules were comprehensively revised once • Exemption cutoff for self-employed was moved twice • Exemption cutoff for wage-earners was moved four times • Schedule assignment rule Wage Income > 0 . 5 × Taxable Income = ⇒ Wage Earner 13 / 50
Tax Changes – Self-employed 2009–10 8 Change in Net-of-tax Rate (%) 6 4 2 0 0 50 100 150 200 250 300 350 400 450 500 Taxable Income in PKR 000s 14 / 50
Tax Changes – Self-employed 2010–11 8 Change in Net-of-tax Rate (%) 6 4 2 0 0 50 100 150 200 250 300 350 400 450 500 Taxable Income in PKR 000s 15 / 50
Tax Changes – Wage Earners 2007–08 1.5 Change in Net-of-tax Rate (%) 1 .5 0 -.5 0 100 200 300 400 500 600 700 Taxable Income in PKR 000s 16 / 50
Tax Changes – Wage Earners 2008–09 1.5 Change in Net-of-tax Rate (%) 1 .5 0 -.5 0 100 200 300 400 500 600 700 Taxable Income in PKR 000s 17 / 50
Tax Changes – Wage Earners 2009–10 1.5 Change in Net-of-tax Rate (%) 1 .5 0 -.5 0 100 200 300 400 500 600 700 Taxable Income in PKR 000s 18 / 50
Tax Changes – Wage Earners 2010–11 1.5 Change in Net-of-tax Rate (%) 1 .5 0 -.5 0 100 200 300 400 500 600 700 Taxable Income in PKR 000s 19 / 50
Data • Population of income tax return filed by self-employed and wage earners in 2006–11 • All registered taxpayers required to file return − → earnings reported at zero tax rate are observed • Primary focus: taxpayers around the exemption cutoff 20 / 50
Summary Statistics Full Sample Analysis Sample Self-Employed Wage Earners Self-Employed Wage Earners (1) (2) (3) (4) Observations 1. Self-Employment Income > 0 100 4.30 100 6.40 2. Wage Income > 0 0.60 100 0.40 100 3. Partnership Income > 0 3.50 1.10 1.80 1.20 4. Switchers 0.90 1.30 0.90 2.70 Characteristics 5. Years Registered 7.04 7.80 7.39 7.79 (4.95) (5.19) (4.83) (4.99) 6. Location 0.39 0.52 0.38 0.47 (0.49) (0.50) (0.49) (0.50) 7. Male 0.98 0.97 0.99 0.97 (0.12) (0.18) (0.11) (0.18) 8. Age 39.55 43.41 40.39 43.27 (15.48) (12.23) (15.22) (13.08) 21 / 50
Outline Introduction Conceptual Framework Institutional Background Testing Predictions of the Model Estimating Tax Evasion Robustness and Heterogeneity Conclusions
Nonparametric Evidence • Earnings responses to to-zero tax changes are an order of magnitude larger than those to similar-sized, not-to-zero tax changes • Difference between the two set of responses reflects tax evasion • Tax evasion is large even at extremely low tax rates 22 / 50
Self-Employment Income – Treatment .4 Not-to-zero tax change Average Log Change in Income .2 0 -.2 To-zero tax changes -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 2006 2007 2008 2009 2010 23 / 50
Self-Employment Income – Control .4 Average Log Change in Income .2 0 -.2 -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 2006 2007 2008 2009 2010 24 / 50
Difference-in-differences ( SE × 2009 ) .4 Not-to-zero tax change Difference-in-Differences Coefficient .2 0 -.2 To-zero tax changes -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 25 / 50
Difference-in-differences ( SE × 2010 ) .4 Not-to-zero tax change Difference-in-Differences Coefficient .2 0 -.2 To-zero tax changes -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 26 / 50
Difference-in-differences ( SE × 2009 - 10 ) .6 Difference-in-Differences Coefficient .4 Not-to-zero tax change .2 0 -.2 To-zero tax changes -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 27 / 50
Difference-in-differences ( SE × 2008 ) .4 Difference-in-Differences Coefficient .2 0 -.2 -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 28 / 50
Line Items – Sales .4 Not-to-zero tax change Average Log Change in Sales .2 0 -.2 To-zero tax changes -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 2006 2007 2008 2009 2010 29 / 50
Line Items – Costs .4 Not-to-zero tax change Average Log Change in Costs .2 0 -.2 To-zero tax changes -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 2006 2007 2008 2009 2010 30 / 50
Line Items – Profit and Loss Expenses .4 Not-to-zero tax change Average Log Change in Expenses .2 0 -.2 To-zero tax changes -.4 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 2006 2007 2008 2009 2010 31 / 50
Line Items – Imports 3 Not-to-zero tax change 2 Average Log Change in Imports 1 0 -1 To-zero -2 tax changes -3 100 150 200 250 300 350 400 450 500 Self-Employment Income in PKR 000s 2006 2007 2008 2009 2010 32 / 50
Recommend
More recommend
