Do Rewards Work to Maintain and Increase Tax Compliance? Evidence - - PowerPoint PPT Presentation

Do Rewards Work to Maintain and Increase Tax Compliance? Evidence from the Randomiza?on of Public Goods

Paul Carrillo Edgar Castro Carlos Scartascini World Bank’s ABCDE 2016 Washington, DC 21 June 2016

The opinions expressed in this publicaDon are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent.

MoDvaDon

§ Tax evasion is high in LaDn America (50% rule) § This is even the case at the local level for property taxes

40% 45% 50% 55% 60% 65% 70%

Ene Mar May Jul Sep Nov Ene Mar May Jul Sep Nov Ene Mar May Jul Sep Nov Ene Mar May Jul Sep Nov Ene Mar May Jul Sep 2010 2011 2012 2013 2014

%

Timely Payment

Background

§ Do rewards work? § Ample literature for health and donaDons (blood, organs) § PosiDve results

§ Financial incenDves and gym aVendance (Aclan and Levy, 2013) § Financial incenDves and weight loss (John et al, 2011) § Non-cash rewards and blood donaDons (Lacetera et al, 2012, 2013)

§ But results are not independent of their design

§ Effects fade away when incenDves are removed (Aclan and Levy, 2013; John et al, 2011) § They may backfire if the financial incenDves crowd out intrinsic moDvaDons (Gneezy, Meier, and Rey-Biel, 2011)

Background

§ What about rewards and tax compliance? § LiVle empirical evidence § Dunning et al. (2015). Reward: “tax holiday”.

§ Didn’t increase for non-compliers § Decreased for compliers. Lost of habit; crowding out on intrinsic moDvaDon

§ Dwenger et al. (2015). RecogniDon and money.

§ Increased compliance among the intrinsically moDvated individuals § Decreased compliance among the extrinsically moDvated. § The type of rewards seems to play no role.

Designing a reward

§ What type of rewards may work beVer for

§ CreaDng posiDve incenDves to maintain compliance? § Having a longer term impact? § Having some spillover effects?

§ One opDon may be to influence the channels that would crowd in intrinsic moDvaDon

The effect of a “moral” reward

§ Effect of a “moral reward” on the individual’s decision problem (from Dwenger et al., 2015) u = u(c, I ∗ s) u’>0, u’’>0 π=max[u(Y −T,s), (1−θ) u(Y,0) + θ u(Y −τT,0)] (1) θ : detection-and-enforcement probability. Individuals’ perceptions about θ are heterogeneous (θ ∼ F ) τ > 1 is a penalty

The intervenDon

§ We are using a natural experiment from the City of Santa Fé

The intervenDon

§ RandomizaDon of sidewalk construcDon in the city of Santa Fe

§ TGI: real property tax § All properDes are taxed § ProporDonal to the properDes’ assessed values (by the Province) § Minimum tax: differs by district (10 districts). ¼ paid the minimum. § Taxpayers pay monthly. Bills are received every three months. § Average yearly tax: 2008: US$107; 2009: US$160 (about US $300 for the 90th perc.)

The intervenDon

§ 400 sidewalks randomly allocated between those with their taxes up to date. About 73,000 § LoVery announced on December 16, 2008 § Eligibility condiDon: all taxes paid by January 12, 2009 § LoVery took place on February 27, 2009 (performed by the provincial authority loVery in a public and televised act) § We find no staDsDcally significant differences in observables between winners and non-winners

§ Balance test § Test of the spaDal distribuDon of winners

The intervenDon

§ ConstrucDon of the sidewalk cost: about USD $1,550 (almost 10 Dmes the yearly payment) § Most of the construcDon took place between 2q.2009 and 3q. 2010 (random)

The intervenDon

The intervenDon

§ People could reject the sidewalk

Data

§ AdministraDve data for every property from 2008 to 2012 § Property characterisDcs (land, construcDon size) § Public goods provision

§ water, gas and sewerage networks, § street lightning § street surface: asphalt, concrete, gravel or dirt § public garbage recollecDon services

§ Payment history

§ Dependent variables:

§ Payment on Dme § Payment within 3 months § Payment within 6 months

§ Why would rewards have any effect?

§ Ex-ante, financial moDve § Most of the literature concentrate on this effect

§ Winners

§ Reciprocity: watching the government at work; evaluaDon of efficiency § Reciprocity: loyalty § Peer-effects: signaling to others own compliance

§ Spillover effects

§ Reciprocity: watching the government at work § Peer-effects: signal of compliance in the neighborhood

Effect of the reward

IdenDficaDon

§ Di : indicator that takes the value of one if the property was awarded a sidewalk renovaDon and zero otherwise § Xi : vector of Dme-invariant characterisDcs of property i § Zit : tax liability of property i in month t § λt are Dme fixed effects § µit is an unobserved random term § α measures the causal effect of the loVery, or the ITT effect § Tit : one if property i featured a renovated sidewalk in t.

§ Instrumented with Di

§ β : causal effect, TOT

Outcome variable: [1] [2] [3] [4] Ya: Tax bill was paid on time

0.024** 0.026** 0.026** 0.026** (0.011) (0.011) (0.011) (0.011)

Yb: Tax bill was paid within 3 months

0.017* 0.019** 0.019** 0.019** (0.009) (0.009) (0.009) (0.009)

Yc: Tax bill was paid within 6 months

0.015*** 0.016*** 0.016*** 0.016*** (0.005) (0.005) (0.005) (0.005)

Yd: Tax bill was never paid

• 0.004*
• 0.005*
• 0.005*
• 0.005*

(0.005) (0.005) (0.005) (0.005)

Covariates Characteristics of property N Y Y Y Characteristics of tax bill N N Y Y Time Fixed effects (35) N N N Y Number of observations:

2,503,194 2,503,194 2,503,175 2,503,175

Model Specification Effect of Lottery on Tax Compliance (OLS)

Outcome variable: [1] [2] [3] [4] Ya: Tax bill was paid on time

0.059** 0.065** 0.066** 0.065** (0.028) (0.027) (0.027) (0.027)

Yb: Tax bill was paid within 3 months

0.042* 0.048** 0.048** 0.048** (0.022) (0.021) (0.021) (0.021)

Yc: Tax bill was paid within 6 months

0.036*** 0.039*** 0.039*** 0.039*** (0.013) (0.013) (0.013) (0.012)

Yd: Tax bill was never paid

• 0.011*
• 0.012*
• 0.011*
• 0.011*

(0.013) (0.013) (0.013) (0.012)

Covariates Characteristics of property N Y Y Y Characteristics of tax bill N N Y Y Time Fixed effects (35) N N N Y Results from First Stage: LM test statistic for underidentification (Anderson or Kleibergen-Paap) 193.4 193.5 193.5 193.5 p-value of underidentification LM statistic 0.000 0.000 0.000 0.000 F statistic for weak identification (Creagg-Donald or Kleibergen-Paap) 441.8 442 442 442 Number of observations:

2,503,194 2,503,194 2,503,175 2,503,175

Model Specification Effect of Sidewalk Renovation on Tax Compliance (2SLS)

• 4.0%
• 2.0%

0.0% 2.0% 4.0% 6.0% 8.0% 10.0% 12.0% Timely payment Paid 3 months Paid 6 months Unpaid

Persistence

§ The esDmated effect are an average for the whole period § How does the effect change over Dme? Is it persistent?

[1] [2] [3] [4] D x M

• 0.006*
• 0.007**
• 0.007**
• 0.007**

(0.003) (0.003) (0.003) (0.003)

D: l(Renovatted Sidewalk) 0.186** 0.207** 0.209** 0.208**

• 0.09

(0.090) (0.090) (0.089)

I: # Months Since Lottery 0.001*** 0.001*** 0.001*** 0.003*** (0.001) (0.001) (0.001) (0.001) Characteristics of property N Y Y Y Characteristics of tax bill N N Y Y Time Fixed effects (35) N N N Y Persistency of the Effect of Sidewalk Renovation on Tax Compliance (2SLS) Dependent Variable: Ya: Tax bill paid on time

Heterogeneous effects

Heterogeneous effects

[1] [2] [3] [4] D x I

• 0.617*
• 0.614*
• 0.628*
• 0.628*

(0.333) (0.329) (0.329) (0.328)

D: l(Renovatted Sidewalk)

0.555** 0.553** 0.565** 0.565** (0.270) (0.266) (0.266) (0.266)

I: Public Service Index

0.398*** 0.302*** 0.279*** 0.278*** (0.010) (0.011) (0.011) (0.011)

Characteristics of property N Y Y Y Characteristics of tax bill N N Y Y Time Fixed effects (35) N N N Y Heterogeneous Effect of Lottery on Tax Compliance (2SLS) Dependent Variable: Ya: Tax bill paid on time

Summary of the results: winners

§ Winners show higher compliance than eligible non-winners (about 7 percentage points on average over the period) § Compliance is parDcularly different immediately aoer construcDon and differences decrease slowly over Dme § The effect is larger where the sidewalk may be more noDceable

Spillovers

Eligible: § We esDmate the effects of having a winner in the block on the average compliance of eligible taxpayers in the block § Preliminary results indicate liVle to no average effects § Effects, however, seem to differ by group. Again, effects seem to be posiDve for those blocks in areas with lower public service provision Non-eligible: § TBC

Concluding remarks

§ LiVle work on the effect of rewards on compliance

§ So far, mixed results. It seems to depend on design

§ Here we have a visible, durable, public good § Reward seems to help cement the good equilibrium § The evidence seems to indicate reciprocity and peer-effects § The reward may not be cost-effecDve.

§ Could this type of reward have spillovers on other margins? Sidewalk construcDon? § TargeDng?

THANK YOU!