Technology and corruption: experimental evidence from the introduction of tax e-filing in Tajikistan

Tajikistan adopted a new tax code that went into effect on January 1, 2013. The new code contains a number of reforms aimed at reducing compliance costs for businesses and stimulating business formalization and growth. This study focuses on evaluating one aspect of this reform package: the electronic filing of taxes.
Tajikistan is a fitting context to study the impact of electronic tax filing on compliance costs and (perceptions of) corruption in tax administration. In the absence of e-filing, businesses submit their monthly tax declarations in person at local tax offices with otherwise productive time spent waiting in line for multiple checks and signatures. As such, it has one of the lowest rankings on the Ease of Paying Taxes (178/189) and ranks 143/189 on the overall Ease of Doing Business in 2014. The introduction of e-filing in this context has the potential to transform the tax administration process and thus reduce the tax compliance burden which constitutes a critical component of the business environment.
Corruption is also a major concern in Tajikistan. Enterprise Survey 2013 reports that 41 percent of firms expect to give gifts in meetings with tax officials and CPIA rates Tajikistan as 2.5/6 in its transparency, accountability and corruption in the public sector. E-filing is very relevant in this context because it presents an opportunity to reduce the frequency of interactions between taxpayers and tax officials and thus minimize opportunities for rent-seeking behavior.
The Tax Committee has several motivations for introducing e-filing:
- To improve service delivery and eliminate long wait times for submission of declarations, and resultantly promote voluntary compliance and increase tax revenues
- To reduce corruption by limiting the amount of interactions between taxpayers and the Tax Committee
- To improve the quality of tax records by reducing the mistakes made by clerks with large data entry burdens
- To improve the efficiency of tax administration by releasing officials from routine work to focus on higher value activities.

To encourage e-filing take-up, the tax authority has conducted public information campaigns and trainings for business associations and removed all registration fees. However, take-up remains lower than expected and this study seeks to increase understanding about which measures would increase utilization. In addition, the evaluation will shed light on the extent to which the anticipated benefits of e-filing adoption are being realized.

Interviews with business owners as well as Tax Committee officials indicate that firms are not using e-filing for a variety of reasons, including:
i) Lack of Awareness: Some firms are unaware of the option to file electronically
ii) Lack of Trust: Some firms are aware of the system but are concerned about the security of information submitted online, or worry that e-filing will increase their likelihood of being audited
iii) Procrastination: some firms are willing to e-file but they do not due to procrastination
iv) Supply Side Problems: some firms are willing to e-file but the registration process is too complicated or time consuming
v) Lack of Infrastructure: Some firms would like to e-file but are not able to due to a lack of access to needed infrastructure and skills (such as computers and internet connection)
vi) Preference for Direct Interaction with Inspectors: Some other firms do not wish to file online because they prefer to deal directly with the same tax inspector on a regular basis for the submission of their tax declarations. This could be for benign reasons such as having someone crosscheck their submissions, or for purposes of attempting to evade tax obligations

This evaluation tests two different types of incentives that aim to address the first four barriers:
- Providing information on e-filing and training on how to use it, to increase awareness and trust in the system;
- Providing logistical support for registration to solve some of the supply side problem. It may also solve the procrastination issue.

2014-10-25

2015-02-14

This study on e-filing will evaluate two broad sets of questions. The first set of questions examines factors affecting the adoption of electronic filing in Tajikistan:
i. What is the impact of providing information and training about e-filing on adoption?
ii. What is the additional impact of helping firms to register?
iii. What characteristics of a firm are associated with adoption?
The second set of questions examines the impact of e-filing on the following outcomes:
i. What is the impact of e-filing adoption on (perceptions of) corruption?
ii. What is the impact of e-filing on compliance costs and tax behavior of firms?
iii. What is the impact of e-filing on tax administration processes?

Perceived corruption d using a standardized index based on different types of questions on corruptions:
- Indirect questions about corruptions (asking about whether other firms are corrupt or see corruption)
- general questions about honesty of tax officials
We will also separately look at outcomes measuring extortion from tax official and outcomes measuring collusion between firms and tax officials (in order to evade taxes).
In addition, we will use list randomization to ask about direct corruption

Tax behaviors and in particular amount of tax declared and paid will be measured using administrative data from the Tax Committee.
We will complement administrative data with declared survey data on tax behaviors.

Compliance costs will be measured using survey data. We will look at three main outcomes:
- Total amount of salary paid to staffs working on tax related activities in 2015.
- The total amount of time spent on tax related activities in 2015.
- The total amount of time spent by CEOs on tax related activities in 2015.

Tax administrative processes will be measured using administrative data and qualitative data (interviews with tax officials).

The study evaluates two different packages of incentives by randomly allocating firms into three groups as follows:
• Group A: Training and Logistical Support: Taxpayers assigned to this group received information about the availability of e-filing, registration procedures and instructions on how to use the system. This training, delivered in group sessions, addresses the non-material factors that constrain the willingness and ability of firms to e-file (such as lack of awareness, lack of training and lack of trust in the system). Following the training session, trained specialists followed up with the firm to gather required documents for submission to the Tax Committee to complete the e-filing registration. This incentive tackles the non-material constraints to e-filing including the hassle factor of having to register and procrastination issues. Taxpayers in this group also received a general training on taxation (not directly connected to e-filling).

• Group B: Training Only: Taxpayers assigned to this group only received the information and training on how to e-file. Unlike group A, they did not receive logistical help with registering. Taxpayers in this group also received a general training on taxation.

• Group C: Control Group: This group of firms serves as a comparison group and did not receive any incentive to e-file. However, following the exact same protocol as for firms in group A and B, firms in this group received a general training on taxation.

Only firms that attended the training were included in the study population. In order to avoid firms choosing to participate in a training based on their treatment allocation, the exact same protocol was used to invite firms to the trainings.

The 2004 firms were randomly allocated into two treatment groups and one control group:
- In group A, 800 firms were sampled and 594 attended to the training.
- In group B, 400 firms were sampled and 296 attended to the training.
- In group C, 800 firms were sampled and 608 attended to the training.

Sampling protocol:
The evaluation draws from the universe of firms in Dushanbe which belong to the Tax Committee database. All legal entities and individual entrepreneurs which are (i) simplified tax regime payers (ii) have been active in the system for at least 2 years (i.e. not new enterprises or bankrupt ones) and (iii) are not currently e-filling were eligible for the study. There were 5,218 firms in the Tax Committee database that met these three criteria.
A list of 2,004 firms was randomly selected from this overall population with stratification on status of the firm (legal entities or individual enterprises) and “rayon” (administrative units dividing Dushanbe in four). This sample size of 2,000 firms was determined after discussions with Tax Committee and i-SYS (the implementing partner). This number corresponds to the expected number of firms that need to be contacted by i-SYS to obtain the attendance of 1500 firms at a training session. Since we expect the program to be more effective on legal entities which are usually bigger firms than individual enterprises, we oversampled legal entities to have 75 percent of legal entities and 25 percent of individual enterprises in the study population.

Key data sources:
The primary sources of data are:
(1) Data from the baseline survey conducted prior to the program implementation. This self-administered survey was filed by each study participant at the beginning of the training sessions. It contains basic information on businesses, including questions on attitudes toward tax administration and perceived corruption.
(2) Data from a follow-up survey (currently scheduled to begin January 2016, corresponding to one year since the program ended). This survey will be administered in person to each firm in the sample at the firm's premises.

This will be supplemented by i) administrative data on tax behaviors (including usage of e-filing); ii) data from the company in charge of implementing the program; iii) data on program cost and iii) data from qualitative interviews with implementing agencies and program beneficiaries.

Key research questions:
- Can technological innovations such as e-filing of tax declarations reduce corruption by reducing interactions between firms and tax officials?
- Can technological innovations such as e-filing reduce tax evasion through a reduction of collusion between firms and tax officials and an increased level of the monitoring?

Randomization was done in office by a computer, with stratification on status of the firm (legal entities or individual enterprises), rayon (four rayons in Dushanbe) and firm sector of activity (trade, services and manufacture and other sector).

Randomization was done at the firm level.

No

1500 firms

1500 firms

- Group A (training and logistical help to register): 600 firms
- Group B: (training only): 300 firms
- Group C (control): 600 firms

The sample size of 1,500 firms has been calculated to meet two goals: - High statistical power to detect small changes in E-filing; - Sufficient statistical power to analyze the effect of e-filing on tax compliance costs, firm perception of corruption, behaviors and performances, assuming that the program has sufficient effect on e-filing take up (i.e., e-filing increases at least by 50 percentage points). Using baseline data on some important outcomes, the power calculation gives the following results for the following outcomes: - Total monthly amount of time spent during in visits to tax committee: mean at baseline : 172 minutes, SD= 73,3, minimum detectable effect when comparing group A and control: 2.9 min (assuming 50pp difference in take up rates); and 5.65 mins when comparing group B and control (assuming a 25% difference in take up). - Direct question on corruption faced by other firms: mean at baseline : 0.18 minutes, SD= 0.38, minimum detectable effect when comparing group A and control: 0.21pp (assuming 50pp difference in take up rates); and 0.41pp when comparing group B and control (assuming a 25% difference in take up). - Indirect question on corruption faced by other firms: mean at baseline : 0.49 minutes, SD= 0.5, minimum detectable effect when comparing group A and control: 0.24pp (assuming 50pp difference in take up rates); and 0.47pp when comparing group B and control (assuming a 25% difference in take up). Actual power will be greater once we allow for i) the use of randomization strata fixed effects (Bruhn and McKenzie, 2009); and ii) control for the lagged dependent variable (McKenzie, 2012). As a result, we are confident that the study has sufficient power as implemented to detect effects that are of economically meaningful size