x

We are happy to announce that all trial registrations will now be issued DOIs (digital object identifiers). For more information, see here.
Public Goods, Information, Trust and Tax Compliance
Last registered on August 03, 2017

Pre-Trial

Trial Information
General Information
Title
Public Goods, Information, Trust and Tax Compliance
RCT ID
AEARCTR-0002369
Initial registration date
August 02, 2017
Last updated
August 03, 2017 11:57 AM EDT
Location(s)
Primary Investigator
Affiliation
Inter-American Development Bank
Other Primary Investigator(s)
PI Affiliation
Inter-American Development Bank
Additional Trial Information
Status
On going
Start date
2017-05-01
End date
2018-12-31
Secondary IDs
Abstract
Tax morale has been shown to affect individuals’ willingness to pay taxes. There is evidence that people comply more when they see the government in action and public monies being used for the good of the community (reciprocity). There is also partial evidence that information about what the government does with the money matters too. This result has been more elusive. One possibility for the high variance in results could be that interventions have differed in the intensity of the treatments. Other possibility is that average results mask high heterogeneity across individuals based on their priors about the efficiency and efficacy of the government. These priors may be affected by their previous experience with the government. In this project we aim to disentangle among these hypotheses by evaluating the role of messages in the context of a large infrastructure campaign (i.e., high intensity treatment). We aim to evaluate the marginal effect of informing taxpayers about the use of public monies and check for heterogeneity according to the services each taxpayer receives (e.g., pave roads or dirt roads), and according to changes in their stock of public services. In terms of policy relevance, we can evaluate the marginal effect of information on top of the effect that public works would have by itself (people see the works and change their payment behavior). In this project, we also evaluate the effect of promises about future public works in the context of a local government with relatively low levels of trust but recently engaged in the expansion of public works. If people believe in promises, then governments could use them to finance future works in advance.
External Link(s)
Registration Citation
Citation
Lagomarsino, Bruno and Carlos Scartascini. 2017. "Public Goods, Information, Trust and Tax Compliance." AEA RCT Registry. August 03. https://doi.org/10.1257/rct.2369-1.0.
Former Citation
Lagomarsino, Bruno and Carlos Scartascini. 2017. "Public Goods, Information, Trust and Tax Compliance." AEA RCT Registry. August 03. https://www.socialscienceregistry.org/trials/2369/history/20135.
Experimental Details
Interventions
Intervention(s)
Messages will be delivered with the tax bills corresponding to the July-August 2017 periods. Each group defined above was randomly split between three interventions: Control, Information, and Promise. The Control group will receive the tax bill as usual. The group assigned to Information will receive the bill with an additional flier that provides infor- mation about the work performed by the government (see Figures 9 and 10). The group assigned to Promise will receive the bill with a flier that provides information about the work done by the government (see Figure 9) and the promise of future work (see Figure 11). Finally, half of the properties assigned to the information treatment were assigned to receive an additional follow-up informational treatment alongside the September and October 2017 bills.
Intervention Start Date
2017-07-01
Intervention End Date
2017-10-31
Primary Outcomes
Primary Outcomes (end points)
Our intervention is designed to provide additional evidence on the role of information on tax compliance. We will use actual payment behavior as our outcome variable. Because we are working with a property tax, the taxpayer decision is only whether to pay or not to pay (and when to pay). As such, tax compliance can be perfectly measured. We have defined compliance in several alternative ways, but always dichotomously, to ensure we capture changes in behavior as precisely as possible.
7
Our outcome variables will take value 1 if the taxpayer has paid in full the total tax liabilities during a set period of time. Because payment could potentially be made late, the dependent variables take into account the timeliness of the payment: I(Bill paid on time) takes the value of 1 if the payment took place before the second due date; I(Bill paid within 2 months) takes the value of 1 if the payment took place before two months of the second due date; I(Bill paid within 3 months) takes the value of 1 if the payment took place before three months of the second due date; I(Bill paid within 6 months) takes the value of 1 if the payment took place before six months of the second due date; and, I(Bill paid anytime) takes the value of 1 if the bill was paid (regardless the timing).
Different definitions allow us to measure changes in two dimensions: extensive and intensive margins. For the extensive margin, we must look at long term definitions to evaluate the rate of total payment. For the intensive margin, we must look at short term definitions to evaluate the rate of timely payment.
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
The overall remaining sample of eligible taxpayers consists of 120,807 properties. Because we are working with a sample with a high degree of heterogeneity, we divided the sample into strata to facilitate the balance between treatment and control groups. The strata are made upon the taxpayer’s average compliance for the period of interest (2014-2016) and access to public goods.
We divide the entire sample in three groups, which we characterize as (i) “never payers”; (ii) “intermittent payers”; and, (iii) “always payers”. We stratify according to these three groups because what we learned in previous field experiments.5 The group characterized as “never payers” payed less than 10% of the MVPSG’s liabilities they received between 2014 and 2016. This group is more likely to ignore the messages included in the tax bill, (they may not be receiving the bill altogether), and can in fact led to underestimate its impact due to the presence of a potential intended-to-treat downward bias. A second group of taxpayers —that we characterize as “always payers”— comprises properties of those taxpayers that have paid their taxes on time at least 90% of the time
5 Informational interventions are shown to work in the margin, but it is highly unlikely that an intervention like this will induce a radical change of state from never payer to always payer.
9
between 2014 and 2016. The remainder of the sample comprises the group of “intermittent payers”. This group encompasses properties of those taxpayers who payed between 10% and 90% of the MVPSG’s liabilities they received between 2014 and 2016, being itself also an heterogeneous group.
We also divided the sample into strata according to the access to public goods of each property. We define five categories of public goods received by a property based on two dimensions: direct public services (e.g. garbage collection and street cleaning services) and indirect ones (e.g. maintenance of public spaces and recreational and leisure activities). Properties classified as category 1 receive all indirect and direct services; properties classified as category 2 receive all indirect services and only 3 (out of 4) direct services and; properties classified as category 3 receive all indirect services and only 2 (out of 4) direct services. Properties in gated communities in Pilar are considered as category 3 since they are private neighborhoods that do not receive some public services, but in this work we treat gated communities as an extra category. Also, we define an extra category for rural properties.
Experimental Design Details
Randomization Method
To separate treatment and control groups, we run 1,000 iterations to select a random 10
draw that maximizes the balance between treatment arms and the control group. The set of pre-experimental characteristics are described in Tables 1 and 2. Description of treatment assignment and stratification in Tables 3, 4 and 5. Balancing conditions in Table 6.
Randomization Unit
Individual taxpayer
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
120,000 taxpayers
Sample size: planned number of observations
120,000 taxpayers
Sample size (or number of clusters) by treatment arms
Control 40,264
Information 40,269
Promise 40,274
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Since we have the proportion of treatments pre-defined by local authorities, we can assess the minimum detectable effect size for each treatment (MDES henceforth). Statistical power is set to 0.8 and significance to 0.05, standard values in economic literature. We conducted power calculations for each experiment separately. For each group of taxpayers we compute the MDES associated to the treatment received. Results in Table 7. We are able to identify any effect greater than 0.8 percentage points in Information and Promise treatments, and for the follow-up information treatment, we are able to identify any effect greater than 1.2 percentage points.
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
IRB Approval Date
IRB Approval Number
Analysis Plan
Analysis Plan Documents
Pre-Analysis Plan _ CardinaleL&Scartascini

MD5: e2cc1f4b44d6162639215b63252788dc

SHA1: 675f2d62f9b63fc2c7e557a32497dbd25dfa294a

Uploaded At: August 02, 2017

Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
No
Is data collection complete?
Data Publication
Data Publication
Is public data available?
No
Program Files
Program Files
Reports and Papers
Preliminary Reports
Relevant Papers