                                                  WPS8452


Policy Research Working Paper                    8452




      Technology, Taxation, and Corruption
   Evidence from the Introduction of Electronic Tax Filing

                             Oyebola Okunogbe
                              Victor Pouliquen




Development Economics
Development Research Group
May 2018
Policy Research Working Paper 8452


  Abstract
  Many e-government initiatives introduce technology                                 e-filing on tax or bribe payments, but significant hetero-
  to improve efficiency and avoid potential human bias.                              geneity exists across firms by their baseline likelihood of
  Electronic tax filing (e-filing) is an important exam-                             tax evasion. Among firms previously more likely to evade,
  ple, as developing countries increasingly adopt online                             e-filing doubles tax payments, likely by disrupting col-
  submission of tax declarations to replace in-person sub-                           lusion with officials. Conversely, among firms less likely
  mission to tax officials. This paper examines the impact                           to have been evading, e-filing reduces tax payments, sug-
  of e-filing on compliance costs, tax payments, and bribe                           gesting that officials had previously required them to pay
  payments using experimental variation and data from                                more. These firms also pay fewer bribes, as e-filing reduces
  Tajikistan firms. Firms that e-file have lower compliance                          opportunity for extortion. In all, the results indicate that
  costs, spending five fewer hours each month on fulfilling                          e-filing reduces compliance costs and makes the distribu-
  tax obligations. There are no significant average effects of                       tion of tax payments across firms arguably more equitable.




  This paper is a product of the Development Research Group, Development Economics. It is part of a larger effort by the
  World Bank to provide open access to its research and make a contribution to development policy discussions around the
  world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/research. The authors
  may be contacted at ookunogbe@worldbank.org.




          The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
          issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
          names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
          of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
          its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


                                                        Produced by the Research Support Team
   Technology, Taxation, and Corruption: Evidence
    from the Introduction of Electronic Tax Filing∗
                          Oyebola Okunogbea            and Victor Pouliquenb




Keywords: Technology, taxation, corruption, e-government, electronic tax ﬁling (e-ﬁling)
JEL codes: D73, H26, O33




   ∗
    We thank Emil Abdykalykov, Penelope Fidas, Alisher Isaev, Rustam Karimov, Vazha Nadareishvili,
Dmitry Pyatachenko, Inomjon Sadulloev, World Bank Group (WBG) Tajikistan country oﬃce, and partners
from the Tax Committee of Tajikistan for their invaluable cooperation. We are also grateful for input from
Alejandra M. Alcantara, Anne Brockmeyer, Ana Goicoechea, Marc Gurgand, Asim Khwaja, Liam Wren Lewis,
Christopher D. Miller, Jennifer Murtazashvili, Dina Pomeranz and several seminar participants. The study
protocol received approval from Harvard Institutional Review Board. The study was registered on the AEA
RCT registry website (ID: AEARCTR-0000914). The research for this paper received funding from the Impact
Program managed by the WBG Trade and Competitiveness Global Practice and the Impact Evaluation to
Development Impact Trust Fund (i2i) managed under the WBG Development Impact Evaluation Unit. The
ﬁndings, interpretations, and conclusions expressed in this paper, including any errors, are entirely ours and do
not necessarily represent the views of the Tax Committee of Tajikistan, the World Bank Group, its Executive
Directors or the countries they represent.
a
  World Bank Development Research Group
b
  Paris School of Economics
1        Introduction

Technology is transforming the ways governments deliver services to and interact with

citizens across various sectors, from public ﬁnancial management to social programs to

elections and many others (World Development Report 2016). These e-government ini-

tiatives typically seek to improve service delivery and eﬃciency. Often, they also aim to

combat corruption by automating systems and reducing oﬃcials’ discretion.

        Tax administration is an important application of e-government in developing countries.

Tax systems in these countries are often characterized by frequent interactions between

tax oﬃcials and taxpayers because, given limited data sources to monitor compliance, tax

oﬃcials are often responsible for direct, in-person monitoring of taxpayers. This makes the

system more vulnerable to corruption, both in terms of collusion between tax collectors

and taxpayers to evade taxes, as well as extortion of taxpayers by tax oﬃcials. These

manual systems also typically result in high compliance costs which may further discourage

compliance. These features of the tax system may contribute to the observed low ﬁscal

capacity in many developing countries (Gordon and Li, 2009, Besley and Persson, 2014)

that limits their ability to provide services and growth-promoting public goods.

        A number of governments have responded by adopting electronic ﬁling (henceforth e-

ﬁling) of taxes. E-ﬁling refers broadly to online submission of tax declarations, typically

replacing in-person submission to tax oﬃcials.1 As of 2015, 32 percent of developing coun-

tries had introduced e-ﬁling and its prevalence continues to grow rapidly (World Develop-

ment Report 2016).2 The most common feature of tax reforms reported in the 2015 World

Bank Doing Business Indicators was the introduction or enhancement of electronic systems

for ﬁling and paying taxes, with 26 economies implementing such changes (World Bank,
    1
    Mail submission is also possible but less common in low income countries with unreliable postal systems.
    2
    E-ﬁling use increases with national income. It is present in 85 percent of high-income countries and in
65 percent of middle-income countries (World Development Report 2016).



                                                     2
2016). Given the rising prevalence of e-ﬁling, it is increasingly important to understand its

impact.

      In this paper, we examine the impact of e-ﬁling adoption using data from small and

medium sized businesses in Dushanbe, the capital city of Tajikistan. Since e-ﬁling was

available to all ﬁrms, we implement an encouragement design whereby we provide a ran-

domly selected group of ﬁrms in-depth training on e-ﬁling as well as logistical help in

completing the diﬀerent steps involved in the e-ﬁling registration process. Ninety-three

percent of ﬁrms in this group adopt e-ﬁling, compared with 60 percent in the comparison

group. We use this diﬀerence to estimate the impact of e-ﬁling adoption on ﬁrms.3

      E-ﬁling may improve tax administration in multiple ways. First, online submission may

reduce tax compliance costs by removing the need for time-consuming, anxiety-inducing

visits to the tax oﬃce, freeing up resources for productive activities. Second, by reducing

the frequency of in-person interactions between taxpayers and tax oﬃcials, e-ﬁling may

reduce collusion between the two parties to reduce tax payments. Third, by removing the

need for a tax oﬃcial to verify submission of tax declarations, e-ﬁling may also protect

taxpayers from tax oﬃcials holding them up and extorting them. In addition, e-ﬁling may

produce system-wide eﬃciency gains by removing the need for physical collection of forms

and data entry. Lastly, by making tax information submitted by taxpayers immediately

available electronically, e-ﬁling may increase the ease of data analysis for monitoring.

      On the other hand, it is possible that e-ﬁling may not deliver these expected bene-

ﬁts or may even lead to worse outcomes, especially in low income settings with limited

information technology coverage and low capacity in other aspects of tax administration.

For example, certain taxpayers may experience signiﬁcant costs of adoption rendering e-

ﬁling inaccessible to them (Yilmaz and Coolidge, 2013). For small businesses with limited
  3
      Given the ﬁrm-level design, we are unable to examine system-wide eﬃciency impacts of e-ﬁling.




                                                    3
third-party reporting or other means of verifying income, tax oﬃcials may have important

private information gathered from ﬁeld visits and frequent interactions with ﬁrms which

they use in verifying ﬁling submissions. Since e-ﬁling removes this ex-ante check, it may

lead to lower tax revenues.

   While e-ﬁling may beneﬁt taxpayers by decreasing compliance costs and reducing ex-

posure to extortion, it may also constrain them by reducing opportunities for collusion and

tax evasion. As such, the adoption decision will depend on the relative expected costs and

beneﬁts to each ﬁrm. This suggests that there may be important selection eﬀects in the

adoption of e-ﬁling, as well as heterogeneous eﬀects of e-ﬁling on ﬁrms by their tax com-

pliance at baseline. Since tax evasion is not directly observable, we use a risk proﬁle score

developed by the tax authority as an indicator of likelihood of evasion. Supporting the

use of this proxy, we observe that conditional on observable ﬁrm characteristics including

industry, number of employees and turnover, higher risk scores are correlated with lower

tax payments and with higher audit adjustments and penalties. For ease of exposition,

we consider ﬁrms above the median risk score to be “high-risk ﬁrms” and those below the

median to be “low-risk ﬁrms.”

   We ﬁnd evidence of selection eﬀects in e-ﬁling adoption: one standard deviation increase

in a ﬁrm’s risk score is associated with a 7 percentage point decrease in its likelihood of e-

ﬁling. Further, among the treatment group for whom adoption was almost universal, high-

risk ﬁrms are 13 percentage points more likely to stop using e-ﬁling within six months.

This pattern is consistent with these high-risk ﬁrms deriving more beneﬁt from dealing

directly with tax inspectors under paper ﬁling, presumably to reduce their tax liabilities.

   Next, we examine the impact on our main outcomes (compliance costs, tax payments,

and unoﬃcial payments). Firms that e-ﬁle because of our intervention save almost ﬁve

hours on average every month, about 15 percent of the total amount of time spent on



                                              4
tax-related activities. Overall, we ﬁnd no signiﬁcant eﬀects on unoﬃcial payments or on

the amount of taxes paid, but this result masks substantial heterogeneity across ﬁrms.

Among high-risk ﬁrms, e-ﬁling leads to a signiﬁcant increase in taxes paid. This result

is consistent with high-risk ﬁrms losing opportunities to collude with oﬃcials to reduce

their taxes, an interpretation further supported by evidence of changes in tax preparation

procedures of high-risk ﬁrms. Over time, this impact on tax payment fades as more high-

risk ﬁrms revert to paper ﬁling. In line with this, there is no signiﬁcant impact of e-ﬁling

on unoﬃcial payments measured one year after the intervention.

   Among low-risk ﬁrms, e-ﬁling adoption leads to a decrease in the amount of taxes paid

and in the likelihood of paying bribes. The coexistence of these two eﬀects suggests that at

baseline, tax oﬃcials had played an important monitoring role in ensuring that these ﬁrms

paid a minimum amount of tax while also extracting bribes from them. Importantly, these

opposing eﬀects on tax payments by high-risk and low-risk ﬁrms lead to closing the revenue

gap otherwise observable between these two groups. This result holds when controlling for

observable diﬀerences between high- and low-risk ﬁrms. As such, while the intervention

does not change the amount of taxes paid, it changes the distribution of tax payments

across ﬁrms in a way that is arguably more equitable, given the evidence of higher evasion

among high-risk ﬁrms and reduction of unoﬃcial payments among low-risk ﬁrms.

   This paper contributes to diﬀerent areas of the literature: First, it adds to a grow-

ing body of work on the role of technology in enhancing state capacity. Recent studies

have examined the impact of e-government initiatives in increasing transparency, reduc-

ing leakages and monitoring compliance in diﬀerent government functions ranging from

procurement (Lewis-Faupel et al., 2016) to elections (Fujiwara, 2015) to social programs

(Banerjee et al., 2017, Muralidharan et al., 2016) and many others (Olken and Pande, 2012,

World Development Report 2016). In the ﬁeld of taxation, Ali et al. (2015) and Eissa et al.



                                             5
(2015) examine the impact of electronic billing machines on tax compliance in Ethiopia and

Rwanda respectively. While much of the existing literature across sectors indicates posi-

tive eﬀects, this paper highlights the fact that technology may have heterogenous eﬀects

on users based on their compliance behavior prior to the introduction of the technology.

Further, when adoption is voluntary, ﬁrms and individuals on whom the technology may

have the greatest monitoring impact may be the least likely to adopt.

   Second, this paper contributes to the rapidly growing literature on taxation in develop-

ing countries. To date, this literature has focused largely on taxpayer responses to diﬀerent

types of messaging, third-party reporting and tax policy changes (See Slemrod, 2017 for

a review), devoting less attention to interactions between tax oﬃcials and taxpayers. No-

table exceptions are Khan et al. (2016) and Amodio et al. (2017), who study the impact

of performance pay for tax collectors on revenue and bribe payments. Whereas those inter-

ventions sought to change the incentives of tax collectors, this paper focuses on changing

the opportunities for collusion and extortion between tax oﬃcials and taxpayers. Third,

and more broadly, this paper contributes to our understanding of corruption and of the

eﬀectiveness of strategies to reduce it (see Olken and Pande, 2012 for a review). Similar

to Sequeira and Djankov (2014), it diﬀerentiates between collusive and coercive corruption

and the diﬀerential response of ﬁrms to both.

   Lastly, this study provides some of the ﬁrst evidence on the causal impact of e-ﬁling.

The only other study we are aware of is Kochanova et al. (2016) which uses variation

in when diﬀerent countries introduced e-ﬁling and ﬁrm-level outcome data from diﬀerent

waves of the World Bank Enterprise Surveys to determine that, on average, e-ﬁling reduces

compliance costs and bribe solicitation, but provides no results on the impact on tax

revenue. By using ﬁrm-level variation in e-ﬁling adoption that arises from a randomized

experiment, this paper avoids endogeneity bias that may be otherwise present.



                                             6
        The rest of this paper proceeds as follows. Section 2 provides background on the

tax administration system and the introduction of electronic tax ﬁling in Tajikistan while

Section 3 provides details on the experiment design and program implementation. Section 4

outlines a simple conceptual framework for the impact of e-ﬁling and the adoption decision

of ﬁrms. Section 5 describes the data and Section 6 outlines the empirical speciﬁcation.

Section 7 provides results on e-ﬁling adoption, as well as results on the impact of e-ﬁling

on ﬁrms. Section 8 concludes and outlines policy implications.



2        Context: Tax Administration and Electronic Tax Filing

         in Tajikistan

Like many developing countries, Tajikistan4 faces signiﬁcant challenges with improving

tax administration and reducing corruption, making it an interesting setting to study the

introduction of e-ﬁling. Firms face high tax compliance costs: In addition to the turnover

tax (ﬁled quarterly),5 small and medium ﬁrms must also ﬁle monthly declarations for

income tax withholdings on employee salaries and social insurance tax for employees.6 In

addition, they are obliged to conduct a quarterly reconciliation of their tax accounts.

        In the absence of e-ﬁling, ﬁrms submit their tax declarations in person at local tax of-

ﬁces, spending otherwise productive time waiting in line for multiple checks and signatures

from tax oﬃcials. On average, ﬁrms in the study sample report spending six hours each

month fulﬁlling tax obligations (excluding accounting), with about three hours going to-

wards visits to the tax oﬃce. Furthermore, corruption is a major concern in Tajikistan. At
    4
     With a 2015 GDP per capita (current USD) of $919, Tajikistan is classiﬁed as a low-income country in
Central Asia (World Development Indicators, 2015).
   5
     Firms in the study sample are drawn from the simpliﬁed tax regime and as such are subject to a
turnover tax rather than a corporate income tax.
   6
     The tax rates are 6 percent of turnover, 25 percent of salaries for the social tax, and 13 percent of
salaries for the income tax.



                                                    7
the time this project commenced, the World Bank Enterprise Survey (2013) indicated that

32 percent of ﬁrms expect to give gifts in meetings with tax oﬃcials and 37 percent expect

to give gifts to any public oﬃcials to “get things done.” Opportunities for corruption in

this context may be reinforced by the fact that tax oﬃcials are responsible for a portfolio

of taxpayers, which means that they interact with the same taxpayers frequently.

      These two concerns—improving service delivery for taxpayers by eliminating long wait

times for submission of declarations and curtailing corruption by reducing the frequency

of interactions with tax oﬃcials—were among the primary reasons the tax authority in

Tajikistan introduced e-ﬁling. By making it easier for people to ﬁle and pay taxes, and by

closing oﬀ opportunities for corruption, the government expected e-ﬁling would ultimately

lead to increased voluntary compliance and thus increased tax revenues. Other motives

for introducing e-ﬁling are to improve the quality of tax records by reducing arithmetic

mistakes by ﬁrms7 and data entry mistakes by clerks and to improve the eﬃciency of tax

administration by releasing oﬃcials from routine work to focus on higher-value activities.

      In order to ﬁle taxes online, ﬁrms must ﬁrst register for e-ﬁling by submitting application

documents to the tax authority. After processing and approval of the application, the ﬁrm

receives an e-token containing its digital signature which it must use to verify its online

declarations. Once a ﬁrm registers for e-ﬁling, it can ﬁle all taxes online using the e-ﬁling

software either on its own computer or at public e-ﬁling terminals located in tax oﬃces.8

However, ﬁrms are still required to visit the tax oﬃce quarterly to reconcile their accounts.

      Adoption of e-ﬁling by ﬁrms was slower than expected following its introduction in

2012. Initially, ﬁrms had to purchase e-tokens for $40 plus an additional $40 registration

fee. Even after the elimination of these fees, only about 30 percent of ﬁrms registered
  7
    The online system has in-built formulas to perform calculations automatically based on inputted entries.
  8
    Online payment of taxes through e-banking is also available to ﬁrms but is separate from the e-ﬁling
system: ﬁrms that e-ﬁle do not have to pay online and vice versa. Although this paper does not study the
impact of electronic tax payment, we observe that 66 percent of ﬁrms that e-ﬁle also pay online.



                                                     8
for e-ﬁling. The low take-up was puzzling given the anticipated beneﬁts from adoption,

prompting signiﬁcant interest in understanding the constraints to adoption and potential

ways to address them. Focus group meetings and interviews with both business owners

and tax oﬃcials indicated ﬁrms were not using e-ﬁling for a variety of reasons, including:

lack of awareness; lack of trust in the system’s reliability and in the security of information

submitted online; diﬃculties navigating the registration process; and lack of access to

computers and internet.9 In addition, some ﬁrms indicated that they did not ﬁle online

because they preferred to deal directly with the same tax oﬃcial on a regular basis when

submitting their tax declarations (perhaps for benign reasons such as having someone

trusted crosscheck their submissions, or perhaps intending to evade tax obligations).



3    Experiment Design

The experimental treatment arms of this study aimed to address those barriers to e-ﬁling

identiﬁed by ﬁrms. Firms were randomly assigned into two treatment groups (and one

control group). In the intensive treatment arm (Group A), ﬁrms received training and

information about e-ﬁling intended to increase awareness and trust in the system. In

the training session, they learned about e-ﬁling availability, its beneﬁts, and registration

procedures, and participated in an interactive demonstration of the e-ﬁling system. In

addition, these ﬁrms received logistical support in registering for e-ﬁling to mitigate the

hindrances ﬁrms willing to use e-ﬁling face in accessing the system: a representative of the

implementing partner helped ﬁrms complete all the steps required for registration.

    Firms in the second treatment arm (Group B) received an identical e-ﬁling training

session but they did not receive the logistical help for registration. In the control group
   9
     From the baseline survey, 50 percent of ﬁrms in our study sample were unaware of the possibility of
e-ﬁling and 37 percent did not have high-speed internet on their premises.




                                                   9
(Group C), ﬁrms did not receive any e-ﬁling training. However, to hold the delivery format

of the treatments constant, these ﬁrms received general training on taxation not speciﬁc to

e-ﬁling.10 Rather, the training included a review of diﬀerent tax laws and procedures. Due

to a requirement by the Tax Committee, this general tax training included one statement

about the availability of e-ﬁling on a slide listing the three modes of ﬁling taxes: “by

paper, by mail and electronically.” In addition, ﬁrms in this group would have been aware

of the existence of e-ﬁling from a reference to it in the invitation materials and from

some questions in the baseline survey. This observation is important for interpreting our

results since it means Group C is not a pure control group but rather a group with a light

information treatment.

      This design allows us to answer two broad sets of questions. The ﬁrst set examines

ﬁrms’ decisions to adopt e-ﬁling: What is the impact of providing information and training

about e-ﬁling? What is the additional impact of helping ﬁrms to register? What other ﬁrm

characteristics predict e-ﬁling adoption? The second set of questions focuses on the impact

of using electronic ﬁling on ﬁrm outcomes, in particular, compliance costs, tax payments

and unoﬃcial payments.


3.1      Sampling and Randomization

The study draws from the universe of ﬁrms in Dushanbe that are registered in the tax

authority database. All legal entities and individual entrepreneurs that are (i) simpliﬁed

tax regime payers, (ii) have been active in the system for at least two years (i.e. not new

enterprises or liquidated ones), and (iii) not currently e-ﬁling, were eligible for the study.

There were 5,218 ﬁrms in the tax database meeting these three criteria.

      We randomly selected a list of 2,004 ﬁrms from this overall population with stratiﬁcation
 10
      Groups A and B also received this general tax training during their sessions.




                                                     10
on status of the ﬁrm (legal entities or individual enterprises) and rayon (tax district). Based

on discussions with the tax authority and the implementing partner on expected response

rates, we estimated that we needed to contact 2,000 ﬁrms to have 1,500 ﬁrms attend the

training sessions. Since we expected the intervention to be more eﬀective on legal entities

which are usually bigger ﬁrms than individual enterprises, we over-sampled legal entities

to achieve a relative proportion of 75 percent of legal entities to 25 percent of individual

enterprises in the study population. We randomly allocated the 2,004 ﬁrms into three

groups: 802 into Group A, 400 into Group B and 802 into Group C.11 We conducted

the randomization by computer using STATA and stratiﬁed using legal status, sector of

activity and rayon (tax district). Figure 1 illustrates the experimental design and the

sampling strategy.


3.2     Program Implementation

The training programs and logistical support were delivered by a Dushanbe-based ﬁrm

with the support of the tax authority from October to December 2014. Firms were invited

to attend a general training on taxation through telephone calls by trained operators.12

Although we assigned ﬁrms randomly into the treatment and control groups before the

trainings (to facilitate scheduling into the diﬀerent training sessions), all ﬁrms received

invitations following exactly the same script (see Appendix A for the text of the script).

Out of the 2,004 ﬁrms in the study database, 1,722 (86 percent) received telephone calls

and 1,498 (75 percent of the full sample and 87 percent of those invited) attended training.

These response rates were balanced across the three treatment groups.

      Since the invitations to all ﬁrms followed the same protocol, each ﬁrm had to decide
 11
    More ﬁrms were allocated to Groups A and C than to Group B in order to increase our statistical
power to measure the impact of e-ﬁling on ﬁrms adopting it since we expected the Group A treatment to
have the greatest impact on e-ﬁling adoption.
 12
    The tax authority periodically organizes events and trainings for ﬁrms, so this program was not unusual.



                                                    11
whether to attend the training before learning its treatment group. Therefore, we are

conﬁdent that the decision to attend training is orthogonal to treatment status and did

not introduce selection bias into the study design. As such, we use the sample of the 1,498

ﬁrms that attended training in studying the program’s impact, with the caveat that this

could slightly limit the external validity of our ﬁndings.

    The training sessions took place either on the tax authority’s premises or in the imple-

menting partner’s oﬃce. At the beginning of training (before revealing treatment status),

the baseline survey was self-administered (completed on paper forms by ﬁrm representa-

tives) with detailed instructions and examples provided by implementing partner trainers.

    Groups A and B received identical training which included both a general tax presen-

tation and an e-ﬁling presentation and demonstration, with a question-and-answer session

at the end. The training for Group C (control group) diﬀered in that it did not include

the e-ﬁling presentation and demonstration. On average, training lasted for two hours

in Groups A and B and for one hour in Group C. A few days after each training, the

implementing partner called back all ﬁrms in Group A and assisted interested ﬁrms in

registering for e-ﬁling. This logistical assistance involved visiting each ﬁrm, collecting all

necessary documentation, submitting it to the tax authority on the ﬁrm’s behalf, obtaining

the e-token, and installing the software.



4    Conceptual Framework

In this section, we outline a simple framework for thinking about a ﬁrm’s decision to adopt

e-ﬁling. A ﬁrm will switch from paper ﬁling to e-ﬁling if the expected discounted value of

the beneﬁts associated with e-ﬁling adoption is greater than the cost of adoption. That is,




                                             12
if:

            T
                 βδ t E ((CCp,t − CCe,t ) + (U Pp,t − U Pe,t ) + (T axp,t − T axe,t )) > ACe           (1)
           t=1

      where CCp,t and CCe,t are compliance costs a ﬁrm would face at time t under paper

ﬁling and e-ﬁling respectively. Similarly, U Pp,t and U Pe,t are unoﬃcial payments a ﬁrm

would make to tax oﬃcials at time t under paper ﬁling and e-ﬁling and T axp,t and T axe,t

are taxes a ﬁrm would pay at time t under paper ﬁling and e-ﬁling. ACe is the one-

time adoption cost associated with switching to e-ﬁling. β and δ are the parameters for

present-bias and discount factor respectively.

      We expect that CCp,t − CCe,t > 0, that is, we expect that compliance costs will be

lower for ﬁrms under e-ﬁling since there is no longer any need to spend time traveling to the

tax oﬃce (except for those using public terminals) and obtaining approval.13 Further, we

expect that the larger the diﬀerence, that is, the greater the potential savings in compliance

costs from e-ﬁling, the more likely a ﬁrm will be to adopt e-ﬁling (for example, ﬁrms with

high paper ﬁling compliance costs).

      We hypothesize that the change in ﬁrms’ unoﬃcial payments to tax oﬃcials, U Pp,t −

U Pe,t , will be positive or null because the reduced frequency of interactions between tax-

payers and tax oﬃcials under e-ﬁling provides fewer opportunities for unoﬃcial behaviors.14

Qualitative evidence from focus group discussions and interviews indicate that at least two

diﬀerent types of unoﬃcial payments may occur during interactions between taxpayers and
  13
     E-ﬁling does not necessarily reduce compliance costs for ﬁrms in all contexts. For example, if ﬁrms
submit both electronic and paper documents due to legal requirements or lack of trust in the system (Yilmaz
and Coolidge, 2013), they may indeed face higher compliance costs. However, we do not think this should
be the case in our context given that it is not technically possible to submit a paper and an electronic
declaration simultaneously.
  14
     Sequeira and Djankov (2014) ﬁnd signiﬁcantly higher levels of customs corruption in Maputo ports
relative to Durban ports. Online processing of clearance documentation in Durban results in minimal
interactions between clearing agents and customs agents compared to higher levels of interaction with
in-person submissions in Maputo.



                                                    13
tax oﬃcials: (i) coercion (tax oﬃcials demanding unoﬃcial payments or creating adminis-

trative hurdles to put ﬁrms in a position such that they would voluntarily oﬀer a gift), and

(ii) collusion (tax oﬃcials and taxpayers working together to help taxpayers evade taxes,

usually with an unoﬃcial payment for tax oﬃcials). Bribe coercion will be more diﬃcult

under e-ﬁling since oﬃcials will be less able to hold up taxpayers, causing them to submit

their declarations late and incur ﬁnes. The greater the savings from fewer coerced unoﬃcial

payments, the more likely ﬁrms are to e-ﬁle. In addition, collusion may be more diﬃcult

to coordinate if submissions are now done electronically. For ﬁrms that voluntarily make

unoﬃcial payments to reduce their tax liability, their likelihood of e-ﬁling will depend on

the combination of savings from both unoﬃcial payments and tax payments. Since e-ﬁling

reduces but does not eliminate interactions with tax oﬃcials, these impacts may be muted

if tax oﬃcials and/or taxpayers adjust their behavior to compensate for fewer opportunities

to interact.

    The impact on tax payments as a result of e-ﬁling, T axp,t − T axe,t , is ambiguous as

it depends on ﬁrms’ experiences prior to e-ﬁling. Under paper ﬁling, tax oﬃcials review

ﬁrms’ declarations and sign before the declaration is considered accepted. Eliminating

this interaction between taxpayers and tax oﬃcials through e-ﬁling can aﬀect the amount

of tax a ﬁrm pays in at least two ways.15 First, tax payments may decrease for ﬁrms

whom tax oﬃcials had previously constrained regarding the minimum amount they could

declare. The review during in-person submission may serve as a mini-audit and help to

limit evasion as the tax oﬃcial may have private information about ﬁrms which may be

useful in estimating how much taxes the ﬁrms should pay. Alternatively, this check could be

an opportunity for tax oﬃcials to force ﬁrms to declare more than their actual tax liability.
  15
     It is also possible for lower compliance costs to increase voluntary compliance (by increasing tax morale)
or to increase proﬁts (by reducing business costs), which could both lead to higher taxes paid. However, we
ﬁnd no evidence for improved attitudes on taxation issues and the lower compliance costs are small relative
to the tax amounts; hence, we do not emphasize these channels.



                                                      14
In interviews and focus group discussions, some ﬁrms mentioned that a major consequence

of e-ﬁling was the ability to avoid oﬃcials forcing these increases in tax payment.16 In

either case, for ﬁrms in this category, the introduction of e-ﬁling may lead to lower taxes

paid, T axp,t − T axe,t > 0, since tax oﬃcials will no longer constrain the minimum amount

of taxes ﬁrms can declare at the time of submission. The greater the reduction in tax paid

under e-ﬁling, the more likely the ﬁrm will be to adopt e-ﬁling.

      The second channel by which in-person meetings may aﬀect taxes paid is that rather

than monitoring the taxes declared by ﬁrms, tax oﬃcials may actively collude with ﬁrms

and assist them in completing their declarations to minimize their tax liability. This type

of collusion typically involves an unoﬃcial payment to the tax oﬃcial. The introduction of

e-ﬁling increases the transaction costs for this collusion because forms submitted online are

not easily changeable (unlike paper declarations that can be easily redone). For ﬁrms in

this category, e-ﬁling may lead to higher taxes paid, T axp,t − T axe,t < 0, and so these ﬁrms

will be less likely to adopt e-ﬁling. These two channels may be self-reinforcing because

tax oﬃcials face incentives to meet revenue targets. If a tax oﬃcial is colluding with some

taxpayers to reduce their tax payments, he may need to compensate by collecting higher

tax revenue from others.

      Since we are unable to directly observe tax evasion and unoﬃcial payments, it is diﬃcult

to know which ﬁrms fall into each of the two categories described above. In subsequent

analyses of the heterogeneous impact of e-ﬁling, we distinguish ﬁrms in the two categories

using a risk index developed by the tax authority to identify ﬁrms by their likelihood of

tax evasion. Section 5 (Data) provides a description of this index. As such, we expect that
 16
    Some examples of comments are: “[Tax inspectors] do not care whether we make a proﬁt or how much
was our turnover during the reporting period, they insist on an amount to fulﬁll their targets. So when
ﬁling in person, sometimes they reject a declaration if the amount is too small. Now, by ﬁling online, you
don’t depend on tax inspectors.” “After moving to e-ﬁling, we’re more certain of the amount we need to
pay. Before, during paper-ﬁling, each time we didn’t know what ﬁgures the tax inspector will want us to
pay for that ﬁling period.”



                                                   15
ﬁrms in the ﬁrst category (non-collusive ﬁrms) will have lower risk scores as they will be

less likely to be evading taxes whereas ﬁrms in the second category (collusive ﬁrms) will

have higher risk scores. Figure 2 outlines the expected impact for each of the outcomes for

these two categories.

   Using this framework, we can also discuss the expected impact of the experimental

interventions. The two treatments aim to reduce the adoption costs, ACe of e-ﬁling: the

lower the adoption costs, the more likely a ﬁrm will switch to e-ﬁling. Although ﬁrms may

register for e-ﬁling free of charge, they would nonetheless face additional costs in access-

ing the system, such as learning how to use the technology and the time and uncertainty

costs associated with registration. The information and training component (Group B)

helps ﬁrms overcome the hurdle of understanding how to use the system while the addi-

tional logistical help with registration (Group A) reduces the time and uncertainty costs

of adoption.

   The impact of the interventions will depend on ﬁrm type. First, ﬁrms for which expected

beneﬁts of e-ﬁling are greater than the cost of adoption (for which inequality (1) holds)

will register when they learn e-ﬁling is available. Given that e-ﬁling was mentioned during

the general training provided to all ﬁrms (including the control group), all ﬁrms of this

type in the study sample should register.

   Second, ﬁrms for which expected returns from e-ﬁling adoption are positive but smaller

than the adoption cost (the left-hand term of equation (1) is positive but smaller than

ACe ) will be impacted by the two treatments designed to reduce adoption cost. As such,

we expect that the treatments will increase e-ﬁling adoption for these ﬁrms and that the

combination of training and logistical help will be more eﬀective.

   Finally, the intervention should not have any impact on ﬁrms with negative expected

returns from e-ﬁling adoption (i.e. ﬁrms for which the left-hand term of equation (1) is



                                            16
negative) as they should not be willing to register regardless of the reduction in adoption

costs.17 One exception would be if the treatment promoted the perception that e-ﬁling

was mandatory (We discuss this possibility in the section on mechanisms). Note that since

T axp,t − T axe,t is negative for these ﬁrms (since we expect that CCp,t − CCe,t > 0 and

U Pp,t − U Pe,t > 0), they are likely to be evading tax in the paper system. As a result, we

expect ﬁrms in this third category to have higher risk scores and thus to be less likely to

e-ﬁle than ﬁrms in the ﬁrst two categories.



5      Data

We rely on three main sources of quantitative data. First, we use administrative data

from the tax authority on ﬁrm characteristics (such as number of employees, legal status,

industry and district), e-ﬁling use, and tax behavior (such as monthly tax declarations and

payments). These data include all 1,498 ﬁrms in the study sample as well as the additional

506 ﬁrms originally selected but that did not participate in the training.

      The other two sources are the baseline survey of ﬁrms conducted prior to the trainings

and the endline survey conducted about one year later. The surveys include information

on ﬁrm characteristics and economic behavior, as well as experiences of ﬁrms with the tax

administration process (such as compliance costs). All 1,498 ﬁrms in the study sample

completed the baseline survey.

      The endline survey diﬀered from the baseline by including more questions and in that

enumerators administered it in person at the ﬁrm’s premises (as opposed to the brief,

self-administered baseline survey at the beginning of the training sessions). Eighty-four

percent of the study sample (1,263 ﬁrms) completed the endline survey. Twelve percent of
 17
    Note that as ACe → 0 (and t → ∞ ), inequality (1) simpliﬁes to: E (CCp + U Pp + T axp ) > E (CCe +
U Pe + T axe ), which is never true for these ﬁrms.




                                                 17
the sample did not complete the survey because the ﬁrm was liquidated, while the remaining

four percent proved untraceable or had moved outside Dushanbe. There were no signiﬁcant

diﬀerences in attrition across treatment groups at the endline survey (Appendix Table A1).

In addition to these data, the study relies on extensive interviews and focus groups with tax

oﬃcials and ﬁrms at diﬀerent stages of the project to design the intervention, understand

potential channels of impact and interpret ﬁndings.

    Table 1 shows summary statistics for some variables from the administrative data and

the baseline survey. Firms have a median of two employees (including the entrepreneur) and

an average of three. In 40 percent of ﬁrms, the entrepreneur is the only employee. About

42 percent each are in the trade and services sector. Women own eight percent of ﬁrms.

Table 1 also shows that randomization achieved balance across the diﬀerent treatment

groups for most variables.18 P-values of joint orthogonality tests obtained from regressing

the treatment variable against diﬀerent ﬁrm characteristics also indicate that the variables

are not systematically correlated with the treatment. Since the analysis will emphasize

heterogeneous results by ﬁrms’ baseline risk score, Table 1 also includes randomization

checks for the two subgroups (ﬁrms above and below the median risk score). As Section

7 will describe, the information and training only treatment has no signiﬁcant impact on

e-ﬁling adoption; therefore Table 1 combines Groups B and C as the control and compares

with Group A. Appendix Table A2 provides randomization balance tables for the original

three groups. Analogous to Table 1, Appendix Table A3 shows that randomization balance

persists in the sample surveyed at endline consistent with the non-diﬀerential attrition

across groups.

    At baseline, we measure ﬁrms’ tax compliance costs using the number of visits to the
  18
      Firms’ compliance costs are a notable exception. Firms in Group A visited the tax oﬃce 6.53 times
compared to 6.4 times in the control group. Results from the endline survey show a reduction in the number
of visits in Group A so, if anything, the results may underestimate the reduction in compliance costs due
to e-ﬁling.



                                                   18
tax oﬃce in the last six months, the amount of time spent on diﬀerent activities during

visits to the tax oﬃce,19 and the amount of time spent calculating taxes and completing

tax forms. Compliance cost measurement at endline is similar to at baseline, except that

the endline survey also includes time spent on tax accounting.

    In both baseline and endline surveys, we measure unoﬃcial payments using a survey

question asking how often it is true that ﬁrms “in their line of business” typically have to

make unoﬃcial payments to tax oﬃcials and create a dummy variable for ﬁrms saying it

is “always true” or “often true.”20

    Our main measure of unoﬃcial payments in the endline survey is a list experiment

(Kuklinski et al., 1997). In response to the same question about how many actions on a

given list a company had taken in the last year to “solve or prevent problems with the

Tax Administration,” half of the respondents receive a list of four options that are neutral

and legal.21 The other half receive a list of ﬁve options that includes “making unoﬃcial

payments” as an additional option. The diﬀerence between the average numbers selected

by ﬁrms presented with the longer list and by those presented with the shorter list gives a

measure of how common it is for ﬁrms to make unoﬃcial payments.

    To measure tax payments, we aggregate monthly payments from administrative records

from the tax authority to obtain the total amount of tax paid over the course of the year.

Given the long tail in the distribution of taxes paid, we also use the inverse hyperbolic sine

function of total taxes paid as well as the median tax amount paid in that year. As an

alternative measure, we use ﬁrms’ self-reports of their tax payments in two focal months
  19
     We ask ﬁrms separately about each of the following: Time to travel (both directions); Staying in line;
Attendance at tax inspector oﬃce for checking reports and submission; Staying in line to sign tax reports
by a head of tax oﬃce; Obtaining tax payments reconciliation act; Calculation of tax due and ﬁlling in tax
reports at tax oﬃce; and Other.
  20
     The question is worded in this manner to avoid asking ﬁrms about their own behavior as corruption is
a sensitive topic.
  21
     The four items on the short list are: 1. Received help from trade associations, 2. Had detailed
discussions with tax oﬃcials, 3. Provided additional documents, 4. Pursued court action.



                                                    19
of the year (June and December).


5.1      Risk Score

We use the ﬁrm’s risk score at baseline as a proxy for its likelihood of evading taxes. Like

many tax administrations, Tajikistan’s Tax Committee uses a risk-scoring system to assess

ﬁrm risk as part of its audit strategy.22 Risk-based audits help administrators allocate

limited resources to cases likely to yield signiﬁcant revenue while reducing the monitoring

burden on compliant ﬁrms (less likely targets for an audit). A ﬁrm’s risk score is the

product of a proprietary algorithm incorporating observed ﬁrm characteristics and results

of prior audits on other ﬁrms. Similar to those of most countries (both developed and

developing), Tajikistan’s tax authority keeps conﬁdential the details of the risk-scoring

process to prevent taxpayers engaging in strategic behavior in order to avoid an audit.

That said, risk criteria incorporated into basic models for small ﬁrms in diﬀerent countries

include, inter alia, the sector(s) of activity and business structure, quality of record keeping,

past compliance behavior, cost ratios and deductions from gross revenue, and comparisons

with industry and sector averages (Loeprick and Engelschalk, 2011).

      To investigate empirically this variable’s correlates in our setting, we regress ﬁrms’ risk

scores at baseline against diﬀerent ﬁrm characteristics from administrative and survey data.

Table 2 indicates that risk score is negatively correlated with operating in the trade sector

and with having had a recent audit, whereas it is positively correlated with claiming to

have no other employees besides the entrepreneur and with the amount of tax adjustments

and penalties from the prior audit. In addition, we ﬁnd two pieces of evidence suggesting

that the risk score is predictive of non-compliance. First, we ﬁnd that, controlling for

observable ﬁrm characteristics such as industry, gender of owner, number of employees
 22
      Khwaja et al, (2011) provide examples from dozens of countries’ experiences with risk-based audits.




                                                     20
and (survey data report of) turnover, ﬁrms with higher risk scores at baseline (2014) paid

lower taxes in 2015 (Table A4, Columns 3 and 4). Second, we ﬁnd that, controlling for ﬁrm

characteristics, risk score at baseline is positively correlated with the total amount of audit

adjustments and penalties in the following year, conditional on an audit occurring (Table

A4, Column 6). These eﬀects are statistically signiﬁcant for the inverse hyperbolic sine

function of total taxes and audit penalties, suggesting that the long tails are of particular

importance.

      The tax authority only calculates risk scores for legal entities (75 percent of the sample),

so it is unavailable for individual entities.23 Unfortunately, the risk score variable was not

available to us at the time of randomization; hence we were unable to stratify upon it.

Nevertheless, from Table 1 we ﬁnd that, within the two subgroups, most variables are

balanced, with a few exceptions: Among the low-risk group, ﬁrms in group A are smaller

and have slightly higher risk scores (both at 10-percent level).24 As with the full sample,

p-values of the joint F-test indicate that the variables are not jointly correlated with the

treatment group.



6      Empirical Speciﬁcation

We use Equation (2) below to examine the relative impact of the two treatments in pro-

moting e-ﬁling adoption as well as the ﬁrm characteristics associated with adoption:


                            Di = β0 + β1 TA,i + β2 TB,i + αXi + λSi +           i                        (2)
 23
     Since randomization was stratiﬁed on legal status, this characteristic is perfectly balanced among the
diﬀerent treatment groups.
  24
     We control for these variables in robustness checks and ﬁnd no signiﬁcant diﬀerence on our main results.




                                                     21
where Di is an indicator variable for whether a ﬁrm registers for and uses e-ﬁling, TA,i and

TB,i are indicators for the training with logistical help and the training alone treatments

respectively. β1 and β2 estimate the causal eﬀects of receiving the two treatments respec-

tively on adoption and the diﬀerence between them, β1 − β2 , estimates the diﬀerential

impact of the provision of logistical support in addition to the training. Si is a vector of

strata dummies (one dummy for each triplet of the following business characteristics): legal

status (individual entrepreneur or legal entity), sector of activity (manufacturing, trade or

services), and rayon (four tax districts in Dushanbe) (Bruhn and McKenzie, 2009). To

examine ﬁrm characteristics associated with adoption, we include in Equation (2) a range

of ﬁrm-level variables Xi , in particular, the measures of ﬁrms’ compliance costs, unoﬃcial

payments, and the proxy for their likelihood of evading taxes (as discussed above).

   To assess the impact of e-ﬁling, we can use a variant of Equation (2) above in which we

replace Di with Yi , the outcome variables of interest, to estimate the eﬀect of assignment

into either of the two treatment groups (the intent-to-treat estimate). We control for

baseline measures of outcome variables when available. In addition, we use Equation (3)

below to estimate the eﬀect of e-ﬁling on ﬁrms that adopted it as a result of the program by

using assignment to Group A (the intensive treatment arm) only as an instrumental variable

(IV) for adopting e-ﬁling (Local Average Treatment Eﬀect estimate), since assignment to

Group B has no eﬀect on e-ﬁling adoption. As such, in the IV estimates, the eﬀective

control group consists of Groups B and C (and not Group C alone).



                           Yi = β0 + γ Ef ile + πYi,t=0 + λSi +   i                      (3)

where Yi are outcome measures such as compliance costs, unoﬃcial payments, and tax

payments. Yi,t=0 are baseline measures of the outcome variable.

   We assess the diﬀerential impact of e-ﬁling by ﬁrms’ risk proﬁles (above and below


                                             22
the median of the tax authority risk score) using Equation (4) below which includes an

interaction of e-ﬁling and an indicator for a ﬁrm having below-median risk score (LowRisk

and an interaction of e-ﬁling and an indicator for a ﬁrm having above-median risk score

(HighRisk ).


 Yi = β0 + γ Ef ile ∗ LowRisk + αEf ile ∗ HighRisk + ρHighRisk + πYi,t=0 + λSi +      i   (4)


    γ and α are the coeﬃcients of interest and give the impact of adopting e-ﬁling among

low-risk ﬁrms and high-risk ﬁrms respectively. For assignment to Group A to be a valid

instrument in Equations (3) and (4), it must aﬀect outcomes only by inducing ﬁrms to e-ﬁle

and not have a direct eﬀect. For example, it must not be the case that receiving a visit

by the implementing ﬁrm (as part of the logistical help) caused ﬁrms to think that their

tax declarations were under greater scrutiny. While we are unable to explicitly rule out

a direct eﬀect, we think this is unlikely because ﬁrms were accustomed to receiving visits

from the tax authority (over two times a year on average). Further, given the opposite

pattern of results observed for high-risk and low-risk ﬁrms, the logistical help would need

to have aﬀected these ﬁrms in opposite ways.



7     Results and Discussion

7.1   Adoption of Electronic Tax Filing

By December 2015, about one year after program implementation, 93 percent of ﬁrms in

Group A had registered for e-ﬁling and used it at least once, relative to 63 percent in

Group B and 59 percent in the control group. The combination of training and logistical

support for registration was successful in increasing e-ﬁling adoption by 34 percentage

points relative to the control group (Table 3, Panel A). On the other hand, there was no


                                            23
signiﬁcant diﬀerence between the adoption rates for ﬁrms in Groups B and C, indicating

that e-ﬁling training and demonstration did not promote e-ﬁling adoption compared to the

brief mentions of e-ﬁling the ﬁrms in the control group encountered.

    The large (30 percentage point) diﬀerence between the impacts of the treatments in

Groups A and B indicates that the logistical help with registration addressed an important

constraint to e-ﬁling. This may be due to a number of reasons including reducing the hassle

costs of navigating a complex registration process (Currie, 2004) or helping ﬁrms overcome

procrastination. One additional possibility is that ﬁrms in Group A may have felt coerced

to register for e-ﬁling due to the follow-up support the implementing ﬁrm oﬀered.

    The lack of a signiﬁcant diﬀerence between Group B and the control group could be

because neither of the two interventions had any impact on ﬁrms or because control group

ﬁrms’ limited exposure to information on e-ﬁling had eﬀects as strong as the e-ﬁling train-

ing. While there are currently no data available about a randomly selected pure control

group to investigate these two possibilities, we can compare the study treatment groups

to two groups of ﬁrms not included in the study: ﬁrst, those ﬁrms not contacted at all

after reaching the required number of ﬁrms,25 and second, contacted ﬁrms that declined to

participate in the training.26 In both groups, the e-ﬁling adoption rate is about two-thirds

of that of the control group at 39 percent and 44 percent respectively, suggesting that the

brief mentions of the availability of e-ﬁling in the control group had a signiﬁcant eﬀect

not dissimilar to the eﬀect of detailed training on e-ﬁling procedures and demonstration.

Indeed, 80 percent of ﬁrms in the control group indicated that they had found out about

e-ﬁling from the general tax training session they attended, only ﬁve percentage points

lower than the share of Group B ﬁrms that found out about e-ﬁling through the detailed
  25
     This group is not necessarily random because the implementing ﬁrm may have selected certain types
of ﬁrms to call before others.
  26
     This group is also not necessarily random as ﬁrms with particular characteristics may have declined to
participate.



                                                    24
training sessions (Table 3, Panel B).27

      We examine whether ﬁrms’ compliance costs, experiences with unoﬃcial payments, and

their propensity to evade taxes aﬀect their likelihood of adopting e-ﬁling as outlined in the

conceptual framework. Since almost all ﬁrms in the treatment group adopt e-ﬁling, we

run Equation 2 using only Groups B and C and present the results in Table 4 to show

which ﬁrm characteristics are associated with e-ﬁling adoption. The ﬁrst two columns of

Table 4 provide results for all ﬁrms while the third column covers only legal entities.28

We ﬁnd a positive association between e-ﬁling adoption and number of hours spent on tax

preparation, supporting the hypothesis that ﬁrms with higher compliance burdens under

paper ﬁling would be more willing to switch.

      In addition, we ﬁnd that ﬁrm size is strongly associated with e-ﬁling adoption: one-

(wo)man businesses are signiﬁcantly less likely to e-ﬁle and, among ﬁrms with employees,

e-ﬁling increases with the number of employees, suggesting that ﬁrms with more complex

taxes are more likely to adopt.

      Unlike our priors, we ﬁnd no signiﬁcant correlation between e-ﬁling adoption and ﬁrms

stating that bribe payment is common (our baseline proxy for ﬁrms that pay bribes) in

Table 4. Since this measure does not distinguish between voluntary and involuntary bribes,

we interact it with the risk score, as ﬁrms with higher risk scores are more likely to be

voluntarily paying bribes in exchange for tax evasion support from tax oﬃcials, but ﬁnd

no signiﬁcant eﬀects (regressions not shown). This could be because this measure of bribes

is too noisy (given the potential bias in responding directly to a sensitive question) or

because no relationship between e-ﬁling use and bribe payments exists once you control for

risk score and other ﬁrm characteristics.
 27
     The other main source of information about e-ﬁling for the control group was business networks (17
percent). In contrast, almost all ﬁrms in Group A learned about e-ﬁling from the intervention trainings.
  28
     Since risk score is only available for legal entities, column 2 assigns the mean value for legal entities to
individual enterprises.



                                                       25
      Importantly, Table 4 shows that ﬁrms with higher risk scores are signiﬁcantly less likely

to adopt e-ﬁling. One standard deviation increase in the risk score is associated with a

7.2 percentage point decrease in a ﬁrm’s likelihood of e-ﬁling. This result is consistent

with more risky ﬁrms preferring to deal directly with tax inspectors with whom they

have ongoing relationships and are able to collude to pay less in taxes. Among Group

A ﬁrms, this relationship does not exist (almost all ﬁrms in this group registered for e-

ﬁling). We analyze the diﬀerential selection into e-ﬁling use by ﬁrm risk by assessing

complier characteristics and calculating the relative likelihood that compliers are high-

                        orn-Steﬀen, 2008). This statistic, given by the ratio of the “ﬁrst
risk ﬁrms (Angrist and J¨

stage”29 among high-risk ﬁrms to the overall “ﬁrst stage,” (0.381/0.349=1.09) shows that

compliers are 9 percent more likely to be high-risk ﬁrms. This pattern is consistent with

the possibility that ﬁrms in Group A may have perceived the treatment as meaning e-ﬁling

was mandatory, thereby bringing in almost all ﬁrms, including those that would otherwise

not have registered voluntarily.

      Lastly, we ﬁnd that other ﬁrm characteristics, such as having a female owner, prior

exposure to e-ﬁling, a ﬁrm’s level of comfort with technology (measured by an index of

whether a ﬁrm has high-speed internet, uses email for business communications, and main-

tains accounting records electronically), and having an accountant do not predict e-ﬁling

adoption.


7.2     Impact of Electronic Tax Filing

This section examines the impact of e-ﬁling on compliance costs, unoﬃcial payments, and

tax payments as outlined in the conceptual framework. Table 5 presents estimates of the

impact of e-ﬁling using assignment to Group A as an instrumental variable for e-ﬁling
 29
   The “ﬁrst stage” is the share of compliers and is given by the adoption rate in Group A minus the
adoption rate in the control group



                                                26
use.30 The top panel shows results for the overall sample, the middle panel shows results

for legal entities alone, while the bottom panel shows heterogeneous results for ﬁrms above

and below the median risk score.

    We ﬁnd a signiﬁcant impact of e-ﬁling on compliance costs as shown in Table 5 (columns

1 and 2). Firms that e-ﬁle as a result of the program visit the tax authority 1.4 times fewer

each month. In all, e-ﬁling reduces the time ﬁrms spend on tax-related activities by 4.7

hours a month. As shown in Table 7 (column 6), this eﬀect is concentrated in activities

that involve visiting the tax authority oﬃce (submitting tax returns and obtaining the

reconciliation act). As such, e-ﬁling does fulﬁll the intended goal of reducing tax compliance

costs of ﬁrms. The time savings corresponds to 15 percent of the overall 33 hours ﬁrms

report spending on tax-related activities. A “naive” cost eﬀectiveness analysis comparing

program costs to the amount of money ﬁrms save from lower compliance costs31 (using the

average wage of an accountant in our sample) shows that potential beneﬁts from compliance

costs reduction more than compensate program costs after 7 months (Appendix C and

Appendix Table A6 detail program costs and calculations).

    Using our two measures of unoﬃcial payments, we ﬁnd no overall impact of e-ﬁling,

but observe diﬀerent patterns by ﬁrm risk proﬁle. For the survey question about how

widespread unoﬃcial payments are (Table 5 column 3), we ﬁnd no overall impact either

in the full sample or among legal entities. However, we ﬁnd a decrease of 23.6 percentage

points (only marginally statistically signiﬁcant with p=0.103) among low-risk ﬁrms and a

positive impact of 16 percentage points (not statistically signiﬁcant) among high-risk ﬁrms

and the diﬀerence between the two coeﬃcients is signiﬁcant (p=0.035).

    Similarly, using the list experiment, we ﬁnd opposite results for high- and low-risk
  30
     For simplicity, we only present the instrumental variables results. ITT results are qualitatively similar
(scaled down by a factor of about 3) and available upon request.
  31
     A full cost eﬀectiveness analysis would account for other beneﬁts such as savings in tax administration
costs.



                                                     27
ﬁrms. With the list experiment, since we measure the outcome at a group (rather than

individual) level, we are unable to use the instrumental variable framework and instead

analyze group-level diﬀerences to report intent-to-treat (ITT) estimates. Table 6 (columns

1 and 2) shows the results for the full sample. Among ﬁrms in the combined control group

(Group B and Group C), ﬁrms presented with the longer list (including “making unoﬃcial

payments” to settle problems with the Tax Authority) report carrying out 1.194 actions on

average compared with 1.119 among those presented with the shorter list. This diﬀerence

(0.075) is statistically signiﬁcant, indicating that 7.5 percent of those in the control group

reported unoﬃcial payments. In contrast, in Group A the diﬀerence is 0.02, which is not

statistically signiﬁcant. Comparing the results from the treatment and control groups in

a diﬀerence-in-diﬀerences framework, we ﬁnd a 5.5 percentage point lower prevalence of

unoﬃcial payments in Group A but this diﬀerence is not signiﬁcant.

   In columns 3-6 of Table 6 we analyze results from the list experiment separately for

low-risk and high-risk ﬁrms. Among low-risk ﬁrms, using the same framework described

above, we ﬁnd a large (18 percentage point) diﬀerence in the share of ﬁrms that paid

bribes between Group A and the combined control group, indicating that the movement

to electronic ﬁling and subsequent reduced interactions with tax collectors is eﬀective at

reducing unoﬃcial payments. On the other hand, for high-risk ﬁrms we observe a positive

but not signiﬁcant impact of assignment to Group A on unoﬃcial payments, indicating that

e-ﬁling did not have the same deterrent eﬀect on bribe payments. We discuss potential

reasons for these patterns in the following section.

   As with unoﬃcial payments, we ﬁnd no overall impact of e-ﬁling on tax payments,

but observe opposite eﬀects for ﬁrms above and below the median risk score. The top

and middle panels of Table 5 (columns 4-7) show no statistically signiﬁcant results for the

four diﬀerent measures of taxes paid (total taxes for the year from administrative records,



                                             28
its inverse hyperbolic transformation, median tax paid, as well as self-reported tax paid).

Standard errors are, however, large and we cannot reject small positive or negative impacts.

   As described in the conceptual framework, we expect that the impact of e-ﬁling on

ﬁrms will diﬀer based on ﬁrms’ experiences at baseline: ﬁrms for which meeting with a

tax inspector served as a constraint on the minimum payments they could make would

reduce their tax payments, whereas ﬁrms that received support from tax oﬃcials to reduce

their tax liability would increase their tax payments. Our heterogeneous analysis of tax

payments by ﬁrm risk of evasion at baseline provides evidence for this hypothesis, as shown

in the bottom panel of Table 5.

   All four measures of tax payments show a similar pattern of impact from e-ﬁling: there

is a decrease in tax payments among low-risk ﬁrms and an increase in taxes paid among

high-risk ﬁrms. These diﬀerences are statistically signiﬁcant for the median and inverse

hyperbolic sine transformation for low-risk ﬁrms (measures that are sensitive to having

long tails in the distribution) and for the mean, median, and self-reported survey mean for

high-risk ﬁrms.

   These impacts are quantitatively large. Taking the total taxes for the year from the

administrative data as an example, column 4 reveals an increase of TJS 25,357 ($3,600

USD) as a result of e-ﬁling among high-risk ﬁrms (p=0.046). In contrast, among low-

risk ﬁrms e-ﬁling adoption results in ﬁrms paying TJS 15,099 ($2,160 USD) less in taxes,

but this amount is not statistically signiﬁcant (for this variable). The estimated impact

on ﬁrms with above-median risk scores is signiﬁcantly diﬀerent than that for ﬁrms with

below-median risk scores at the 10 percent level (p=0.078).

   An important implication of these opposite impacts is that e-ﬁling eliminates the tax

gap that would otherwise exist between low-risk and high-risk ﬁrms. Among ﬁrms in the

combined control group, high-risk ﬁrms pay substantially lower taxes than low-risk ﬁrms, a



                                            29
statistically signiﬁcant diﬀerence of $2,300 (p=0.0018) likely reﬂecting evasion by high-risk

ﬁrms. However, among ﬁrms in Group A, the diﬀerence between high- and low-risk ﬁrms

falls to $527 and is not statistically signiﬁcant. This result implies that e-ﬁling helps to

close the revenue gap between ﬁrms that were likely to be evading taxes and those that

were not, thereby promoting horizontal equity. Although e-ﬁling does not change the total

level of taxes paid, it changes the distribution of taxes paid across ﬁrms as low-risk ﬁrms

pay less and high-risk ﬁrms pay more. As such, e-ﬁling also has a potential redistribution

eﬀect that undoes the eﬀect of tax oﬃcials colluding with high-risk ﬁrms to pay lower taxes

and constraining low-risk ﬁrms to cover the shortfall.

   We discuss below in Section 7.4 the possibility that other diﬀerences between high-risk

and low-risk ﬁrms may be driving these results. We show that the results hold when we

control for observable baseline diﬀerences between the two groups, suggesting this is not

the case.


7.3   Mechanisms

This section provides additional evidence to understand the divergent results by risk group.

The decrease in tax payment for low-risk ﬁrms is consistent with tax oﬃcials constraining

them to an acceptable amount of tax to declare under paper ﬁling. In the combined control

group, low-risk ﬁrms on average make positive payments more often than high-risk ﬁrms

(8.3 months a year in contrast to 5.1 times for high-risk ﬁrms). Table 7 (column 1) shows

that, among low-risk ﬁrms, those that e-ﬁle pay zero taxes two more months in a year on

average, but this eﬀect is not apparent for high-risk ﬁrms. The cumulative distribution

function of taxes paid (Figure 3) provides additional evidence that interactions with tax

oﬃcials were likely a constraint on the amounts low-risk ﬁrms declared. Figure 3 (Panel B)

shows that, among low-risk ﬁrms, Group A ﬁrms paid lower taxes than control group ﬁrms,



                                             30
but primarily in the lower end of the distribution, which is consistent with the existence

of a previous lower bound.

   In addition, the simultaneous reduction in both unoﬃcial payments and tax payments

for low-risk ﬁrms suggests that the unoﬃcial payments were more likely not for collusion

to pay lower taxes but for extortion. Less-frequent interaction with tax oﬃcials due to

e-ﬁling therefore reduces the occurrence of bribe-giving.

   Whereas the impact on tax payments for low-risk ﬁrms occurs primarily on the extensive

margin (number of months paying positive taxes), Table 7 (column 2) indicates that the

increase in tax payments for high-risk ﬁrms occurs on the intensive margin (the amount of

taxes paid, conditional on paying in a given month). Similarly, the cumulative distribution

function shows that, among high-risk ﬁrms, the higher tax payment among Group A ﬁrms

relative to the combined control group occurs mostly on the upper end of the distribution

(Figure 3, panel C). This pattern would be consistent with high-risk ﬁrms—both treatment

and control—not being constrained by tax oﬃcials on the minimum amount of tax they

could pay (unlike low-risk ﬁrms). However, with e-ﬁling, treatment ﬁrms at the upper-end

of the distribution lose support to reduce high levels of tax liabilities.

   For high-risk ﬁrms, the increase in tax payments is consistent with their losing the

opportunity to collude with tax oﬃcials to reduce their tax liabilities. From anecdotal

evidence, one of the deterrents to evasion under e-ﬁling is higher transaction costs for

collusion, because after a ﬁrm submits a declaration online it is diﬃcult to change it

(unlike declarations ﬁled on paper, which are easy to change). From the survey data, we

ﬁnd some additional evidence consistent with high-risk ﬁrms becoming less able to rely

on assistance from colluding with tax oﬃcials. We ﬁnd that the amount of time ﬁrms

spent daily collating records increases by 1.2 hours among more risky ﬁrms, while among

low-risk ﬁrms there is no signiﬁcant impact (p=0.042 for the diﬀerence between the two



                                              31
coeﬃcients) (Table 7 column 5), suggesting that high-risk ﬁrms may now need to pay more

attention to preparing their tax declaration submissions themselves.

    The e-ﬁling enrollment behavior of high-risk ﬁrms provides additional insight into the

diﬀerential impact on tax payment. As noted in the discussion of e-ﬁling adoption results,

risk score is negatively correlated with e-ﬁling adoption in the combined control group,

suggesting that ﬁrms with high risk scores anticipated receiving a negative impact. In

addition, we ﬁnd evidence that ﬁrms in the treatment group are more likely to stop e-ﬁling

during the course of the year, an eﬀect concentrated among high-risk ﬁrms: the drop-out

rate for high-risk ﬁrms is 21 percent, relative to 9 percent for low-risk ﬁrms (Appendix

Table A7).32 In contrast, in the control group, the drop-out rates for high risk and low-

risk ﬁrms are similar at 3 percent. This ﬁnding is consistent with the notion that ﬁrms

in Group A may have interpreted the logistical support for e-ﬁling registration to mean

that registration was compulsory. Given that e-ﬁling led to an increase in tax payment for

high-risk ﬁrms, it is not surprising that they were more likely to stop using it.

    In line with the decline in e-ﬁling among Group A high-risk ﬁrms, we examine whether

the impact of e-ﬁling on tax payments fades over time. It does. Whereas there is a strong

and statistically signiﬁcant increase in tax payments for Group A high-risk ﬁrms from

January to June, the analogous result for July to December is smaller and loses statistical

signiﬁcance (Table 7, columns 3 and 4). In contrast, given the more modest drop in e-ﬁling

among Group A low-risk ﬁrms, the impact on taxes paid from January to June is similar to

the impact from July to December and remains statistically signiﬁcant for the tax payment

measures for which it was statistically signiﬁcant for the full period in Table 5 (regressions

available on request).
  32
     Appendix Table A7 shows the diﬀerences in the drop-out rate only among ﬁrms that registered for
e-ﬁling and, as such, the coeﬃcients are descriptive, not causal. The analysis continues only until June
2015 because the data from July to December do not report e-ﬁling use for each month.




                                                  32
   The abandonment of e-ﬁling by Group A high-risk ﬁrms can also explain the result that,

unlike for low-risk ﬁrms, among high-risk ﬁrms no reduction in bribe payments occurs.

Since the survey measuring this outcome took place in January/ February 2016, one year

after program implementation, we have only a one-time measure of this outcome, which

occurs after a sizable proportion of the high-risk ﬁrms have dropped out of e-ﬁling. Since

the impact on tax payments has dissipated by then, it is reasonable to assume that any

collusion arrangements have resumed, such that we see no negative impact of assignment to

the treated group upon the prevalence of bribes. Instead, from the list experiment (Table

6), we observe relatively high levels of bribes (10-14 percent) among high-risk ﬁrms in both

treated and control groups.


7.4   Robustness Checks

One potential concern is that, since the baseline risk proﬁle scores of ﬁrms is correlated with

other ﬁrm characteristics (Table 2), if the impacts of e-ﬁling adoption are heterogeneous

across these characteristics then initial diﬀerences in these characteristics between low-

and high-risk ﬁrms may be driving our heterogeneous analysis by risk score. For example,

suppose that e-ﬁling has greater impacts on ﬁrms in a certain sector and results in those

ﬁrms paying higher taxes. If ﬁrms in that sector also have higher risk scores, this eﬀect

could drive the heterogeneous results found by risk proﬁle score.

   To examine this possibility, we allow the eﬀect of e-ﬁling adoption to vary with other

ﬁrm characteristics. To do so, we estimate the following equation for a set of H potential




                                              33
explanators Kh :33

                                                                  H
    Yi =β0 + γ Ef ile ∗ LowRisk + αEf ile ∗ HighRisk +                 πh Ef ile ∗ LowRisk ∗ Kh,i
                                                                 h=1
                                                                                                    (5)
              H                                     H
          +         µh Ef ile ∗ HighRisk ∗ Kh,i +         ϕh Kh,i + ρHighRisk + λXi +     i
              h=1                                   h=1


      As in previous analyses, Ef ile is instrumented by assignment to group A. For the

variables Kh , we use variables that are signiﬁcantly correlated with the risk scores and

across which the eﬀects of e-ﬁling adoption could plausibly be heterogeneous. We use the

following variables: operating in the trade sector (this sector of activity may beneﬁt more

from e-ﬁling adoption), declaring no employees (one-(wo)man businesses may have simpler

taxes and beneﬁt less from e-ﬁling), rayon (interactions with oﬃcials may vary across

diﬀerent tax oﬃces) and monthly amount of time spent on tax related activities (ﬁrms with

high tax compliance costs may beneﬁt more from e-ﬁling). All variables Kh are demeaned

so we can still interpret γ and α as the impact of e-ﬁling on low-risk and high-risk ﬁrms

respectively, measured at the mean of the other variables.

      Results of these estimations are presented in the Appendix Table A8 and show that our

main results are robust to the introduction of these additional controls. γ and α coeﬃcients

have similar magnitude and statistical signiﬁcance as in Table 5.



8      Conclusion and Policy Implications

This paper presents some of the earliest evidence on the impact of electronic tax ﬁling.

We ﬁnd that ﬁrms’ baseline likelihood of tax evasion is an important determinant of their

willingness to use e-ﬁling, as well as of the impact e-ﬁling has on them. For ﬁrms in a
 33
      We follow the same strategy as in De Mel et al. (2009).



                                                     34
baseline collusive equilibrium with tax oﬃcials, e-ﬁling disrupts that collusive relationship

and causes them to pay signiﬁcantly higher taxes. These high-risk ﬁrms are also less likely

to sign up for e-ﬁling voluntarily and, when they do (potentially due to perceived compul-

sion), they are more likely to stop e-ﬁling, indicating a preference to continue interacting

directly with tax oﬃcials. On the other hand, for ﬁrms in a non-collusive equilibrium, e-

ﬁling results in lower tax payments as tax inspectors no longer constrain them to pay higher

taxes (possibly higher than their true liability). At the same time, low-risk ﬁrms beneﬁt

from e-ﬁling by becoming less exposed to coercive demands for unoﬃcial payments. These

opposing patterns of impact on tax payments close the revenue gap that would otherwise

exist between high-risk and low-risk ﬁrms.

   Over time, e-ﬁling use has continued to expand in Tajikistan. The overall impact on tax

revenue from the introduction of e-ﬁling (beyond the encouragement design considered in

this study) would depend on the distribution of high-risk and low-risk ﬁrms among the total

population that adopts e-ﬁling and the relative magnitudes of the respective increase and

decrease in tax payment among these two groups. Beyond the impact on tax payments,

the tax authority stands to derive signiﬁcant system-wide eﬃciency gains. Back-of-the-

envelope calculations indicate that, at current e-ﬁling rates, the tax authority frees up over

5,800 tax-oﬃcial-hours each month that can be reallocated to other activities likely higher

in marginal productivity than receiving tax declarations.34

   This study’s ﬁndings provide lessons extending beyond e-ﬁling to other technologies

designed to increase eﬃciency and reduce corruption. First, our results highlight the im-

portance of selection among potential users. Those who anticipate negative beneﬁts from

the new technology due to expected constraints on their behavior may be the least likely
  34
     This calculation assumes that a tax oﬃcial spends six minutes on each tax declaration. This is a
conservative estimate as the ﬁrms in our baseline survey report tax oﬃcials spending an average of 18
minutes checking their reports.




                                                 35
to adopt. However, this may be the population segment of greatest interest. As a result,

governments must weigh carefully the capacity constraints of potential users in adopting

a new technology, with the potential of risky types opting out, when deciding whether

to make new technologies mandatory. In Tajikistan, while e-ﬁling was not mandatory,

updated policies require that ﬁrms which start e-ﬁling may not revert to paper ﬁling.

   Second, our results underscore the possibility of heterogeneous eﬀects of technology.

The impact of technology introduced to replace human discretion will depend on the way

that discretion was previously exercised. If discretion produces worse outcomes, technol-

ogy may improve outcomes (as we ﬁnd with higher tax payments among high-risk ﬁrms

and lower bribe payments among low-risk ﬁrms). However, if discretion produces desirable

outcomes (such as monitoring ﬁrms), technology may have an unanticipated eﬀect of pro-

ducing less desirable outcomes. Lastly, on a related note, our results on the distributional

consequences of e-ﬁling emphasize the potential of technology to correct possible patterns

of bias resulting from human discretion.




                                            36
References
[1] Ali, M., Shifa, A., Shimeles, A., and Woldeyes, F.B. (2015). “Information Technology
    and Fiscal Capacity in a Developing Country: Evidence from Ethiopia.” ICTD Working
    Paper 31.
[2] Amodio, F., Choi, J., De Giorgi, G., and Rahman, A. (2017). “Bribes vs. Taxes: Market
    Structure and Incentives.” Working Paper.
[3] Angrist, J., and J¨orn-Steﬀen, P. (2008). “Mostly Harmless Econometrics: An Em-
    piriscist’s Companion” Princeton Univercity Press.
[4] Banerjee, A., Duﬂo, E., Imbert, C., Mathew, S., and Pande, R. (2017). “E-governance,
    Accountability, and Leakage in Public Programs: Experimental Evidence from a Fi-
    nancial Management Reform in India.” Working Paper.
[5] Besley, T. and Persson, T. (2014). “Why Do Developing Countries Tax So Little?”
    Journal of Economic Perspectives, 28(4): 99-120.
[6] Bruhn, M. and McKenzie, D. (2009). “In Pursuit of Balance: Randomization in Practice
    in Development Field Experiments.” American Economic Journal: Applied Economics,
    1(4): 200-232.
[7] Currie, J. (2004). “The Take Up of Social Beneﬁts.” NBER Working Papers 10488,
    National Bureau of Economic Research.
[8] De Mel, S., McKenzie, D., Woodruﬀ, C. (2009). “Are Women more Credit Constrained?
    Experimental Evidence on Gender and Microenterprise Returns.” American Economic
    Journal: Applied Economics, 1(3): 1–32.
[9] Eissa, N. and Zeitlin, A., with Karpe, S. and Murray, S. (2015). “The Incidence and Im-
    pact of Electronic Billing Machines for VAT in Rwanda.” International Growth Center
    report.
[10] Fujiwara, T. (2015). “Voting Technology, Political Responsiveness, and Infant Health:
    Evidence from Brazil.” Econometrica, 83(2): 423-464.
[11] Gordon, R. and Li, W. (2009).“Tax Structures in Developing Countries: Many Puzzles
    and a Possible Explanation,” Journal of Public Economics, 93(7): 855–866.
[12] Khan, A., Khwaja, A. and Olken, B. (2016). “Tax Farming Redux: Experimental
    Evidence on Performance Pay for Tax Collectors.” Quarterly Journal of Economics,
    131(1): 219–271.
[13] Khwaja, M.S., Awasthi, R. and Loeprick, J. (2011). Risk-Based Tax Audits: Ap-
    proaches and Country Experiences. Washington, DC: The World Bank.

                                            37
[14] Kochanova, A., Hasnain, Z. and Larson, B. (2016). “Does E-Government Improve
    Government Capacity? Evidence from Tax Administration and Public Procurement.”
    Background Paper to the 2016 World Development Report. World Bank Policy Research
    Working Paper 7657.

[15] Kuklinski, J. H., Sniderman, P. M., Knight, K., Piazza, T., Tetlock, P. E., Lawrence,
    G. R. and Mellers, B. (1997). “Racial Prejudice and Attitudes toward Aﬃrmative Ac-
    tion.” American Journal of Political Science, 41: 402–19.

[16] Lewis-Faupel, S., Neggers, Y., Olken, B. A. and Pande, R. (2016). “Can Electronic
    Procurement Improve Infrastructure Provision? Evidence from Public Works in India
    and Indonesia.” American Economic Journal: Economic Policy, 8(3): 258-83.

[17] Loeprick, J. and Engelschalk, M. (2011). Simpliﬁed Risk scoring for SMEs. In Khwaja,
    M.S., Awasthi, R. and Loeprick, J., (Eds). Risk-Based Tax Audits: Approaches and
    Country Experiences (pp. 45-54). Washington, DC: The World Bank.

[18] Muralidharan, K., Niehaus, P., and Sukhtankar, S. (2016). “Building State Capacity:
    Evidence from Biometric Smartcards in India.” American Economic Review, 106(10):
    2895-2929.

[19] Olken, B. and Pande, R. (2012). “Corruption in Developing Countries.” Annual Review
    of Economics, 4: 479–509.

[20] Slemrod, J. (2017). “Tax Compliance and Enforcement.” Working Draft.

[21] Sequeira, S. and Djankov, S. (2014). “Corruption and Firm Behavior: Evidence from
    African Ports.” Journal of International Economics, 94(2): 277-294.

[22] The World Bank Enterprise Survey 2013.

[23] The World Bank. 2016. “Doing Business 2016: Measuring Regulatory Quality and
    Eﬃciency.” Washington D.C.

[24] The World Bank Group (2016). World Development Report 2016: Digital Dividends.
    Washington, DC: World Bank.

[25] Yilmaz, F. and Coolidge, J. (2013) “Can E-Filing Reduce Tax Compliance Costs in
    Developing Countries?” World Bank Policy Research Working Paper 6647.




                                           38
Figure 1: Study Design and Program Implementation

                             Population of all firms in Dushanbe

                                  5,218 firms eligible firms:
       Registered in the simplified regime for two or more years and not already e-filing

              2004 firms randomly selected from the Tax Committee database

                                     Random allocation

            Group A                        Group B                    Control Group
       802 firms selected             400 firms selected             802 firms selected

    690 firms (86%) invited       332 firms (83%) invited         700 firms (87%) invited
          to a training                 to a training                   to a training

   594 firms (74%) trained:       296 firms (74%) trained:        608 firms (75%) trained:
        - General training             - General training              - General training
        -E-filing training             -E-filing training

   Logistical help with all 594
              firms




                                                   39
Figure 2: Conceptual Framework
                                                  Expected Impact of E-filing on:
                   Compliance Costs                 Unofficial payments             Tax payments
High-risk firms    Decrease                         Decrease                        Increase
                                                    (collusion)
                   Taxpayer spends less time on     Official and taxpayer have      Official and taxpayer have
                   travel and on queues             less opportunity to collude     less opportunity to collude


Low-risk firms     Decrease                         Decrease                        Decrease
                                                    (coercion)
                   Taxpayer spends less time on     Official has less opportunity   Official is less able to
                   travel and on queues             to hold up taxpayer             require taxpayer to increase
                                                                                    tax paid




                                                         40
  Figure 3: CDF of Tax Paid in 2015 (Inverse Hyperbolic Sine
Transformation) for the Whole Sample and by Risk Profile Score
                                                                     PANEL A: All Firms
                          1
                                            Control group (B&C)
                          .9                Group A
                          .8


Cumulative Probability
                          .7

                          .6

                          .5

                          .4

                          .3

                          .2

                          .1

                          0

                               0                            5                            10         15
                                      Inverse Hyperbolic Sine Transformation of Tax Paid in 2015
                               Notes: Data source: Administrative data from Tax Authority, N=1096

                                              PANEL B: Firms with Below-Median Risk Score
                          1
                                            Control group (B&C)
                          .9                Group A
                          .8
 Cumulative Probability




                          .7

                          .6

                          .5

                          .4

                          .3

                          .2

                          .1

                          0

                               0                             5                           10         15
                                      Inverse Hyperbolic Sine Transformation of Tax Paid in 2015
                               Notes: Data source: Administrative data from Tax Authority, N=547

                                              PANEL C: Firms with Above-Median Risk Score
                          1
                                            Control group (B&C)
                          .9                Group A
                          .8
Cumulative Probability




                          .7

                          .6

                          .5

                          .4

                          .3

                          .2

                          .1

                          0

                               0                                 5                            10         15
                                      Inverse Hyperbolic Sine Transformation of Tax Paid in 2015
                                                                               41
                               Notes: Data source: Administrative data from Tax Authority, N=520
Table 1 : Descriptive Statistics and Balance Checks
                                                  (1)             (2)      (3)              (4)             (5)     (6)        (7)             (8)     (9)
                                                                                       Below median of 2014 risk          Above median of 2014 risk
                                             All sample                                  profile score (low risk)           profile score (high risk)
                                    Mean [SD]                                           Mean [SD]                          Mean [SD]
                                                   Diff. in                                            Diff. in                           Diff. in
                                   Control group                                       Control group                         Control
                                                  Group A                                             Group A                            Group A
                                      (B&C)                                N             (B&C)                    N       group (B&C)                N
PANEL A: Administrative data from Tax Committee (2014)
 Legal entitiesβ                                 0.729             -      1,498              1               -      547         1               -      520
                                                [0.445]                                     [0]                                [0]
 Sector of activity is Tradeβ                    0.415             -      1,498            0.374             -      547      0.248              -      520
                                                [0.493]                                   [0.485]                           [0.432]
 Sector of activity is Servicesβ                  0.42             -      1,498            0.495             -      547      0.582              -      520
                                                [0.494]                                   [0.501]                           [0.494]
 Sector of activity is Manufacturing             0.116            0.012   1,498             0.1            -0.003   547      0.113            0.012    520
                                                [0.321]         (0.011)                   [0.301]         (0.017)           [0.317]          (0.021)
 Female owner                                    0.076            0.023   1,498             0.07            0.007   547      0.048            0.032    520
                                                [0.266]         (0.015)                   [0.255]         (0.023)           [0.215]          (0.024)
 No employee                                     0.406            -0.01   1,498            0.195            0.003   547      0.469            -0.025   520
                                                [0.491]         (0.025)                   [0.396]         (0.035)             [0.5]          (0.045)
 Number of employees                             3.287           -0.127   1,498            4.271           -0.57*   547      3.257             0.108   520
                                                [3.443]         (0.171)                    [3.71]         (0.306)           [3.781]          (0.331)
 The firms was audited in 2014                   0.037           -0.006   1,498            0.067           -0.023   547      0.016             0.014   520
                                                [0.188]         (0.009)                    [0.25]         (0.019)           [0.126]          (0.014)
 Risk profile score in 2014λ                      62.6             0.1    1,067             41.3             1.6*   547       85.1              -0.3   520
                                                 [28.9]           (1.7)                    [10.5]           (0.9)            [24.7]            (2.1)
 Risk profile score in 2014 above                0.486           -0.015   1,067               0            0.000    547         1             0.000    520
                                                  [0.5]         (0.031)                      [0]          (0.000)              [0]           (0.000)
PANEL B: Baseline Survey Data (2014)
 Firm has an accountant                          0.73            0.028    1,498            0.799          -0.032    547      0.762            0.041    520
                                                [0.444]         (0.023)                   [0.401]         (0.037)           [0.427]          (0.036)
 Share of technological practices                0.538           0.024    1,498            0.639           -0.011   547      0.665           0.074**   520
 implemented                                    [0.437]         (0.021)                   [0.422]         (0.038)           [0.402]          (0.036)
 Number of visits to tax authority office        6.382          0.13**    1,498            6.514            0.109   547      6.444            0.158    520
 in Jan-Jun 2014                                [0.948]         (0.056)                   [1.027]         (0.101)           [1.111]          (0.106)
 Time spent on tax related activities            6.185           0.131    1,483             6.26           -0.005   544      6.657            0.188    513
 during a typical month (hours)                 [2.926]         (0.162)                   [2.852]         (0.241)           [3.195]          (0.359)
 Number of times tax inspectors visited          1.324           -0.027   1,498            1.271           -0.044   547      1.199            0.022    520
 the company in Jan-Jun 2014                    [0.948]         (0.054)                   [0.871]         (0.107)           [0.879]           (0.08)
 Ever used e-filing (with another                0.14            -0.013   1,498            0.198           -0.037   547      0.177            0.000    520
 company)                                       [0.348]         (0.018)                   [0.399]         (0.034)           [0.382]          (0.035)
 Think that it is common for firms to            0.18             0.004   1,498            0.255           -0.046   547      0.174            0.019    520
 make informal payments                         [0.385]          (0.02)                   [0.437]         (0.035)           [0.379]          (0.035)
P-values of joint orthogonality testsα                  0.658                                     0.526                              0.553
Notes: Columns 1, 4, 7: Standard deviations presented in brackets. Columns 2, 5, 8: coefficients and standard errors (in parentheses) from an
OLS regression of the firm owner/firm characteristic on a treatment variable (group A), controlling for strata dummies. β: variables used for
stratification. λ: Risk profile scores only calculated for legal entities. α: P-values of joint orthogonality tests obtained from regressions of the
treatment dummy on all the variables in the right-hand column (not used for stratification) and testing for all coefficients are jointly equal to
zero. ***, **, * indicate significance at 1, 5 and 10%.



                                                                                  42
Table 2: Baseline Correlates of Risk Profile Score
                                                                 (1)        (2)          (3)         (4)
                                   Dependent variables:                                Above median of
                                                              Risk Profile Score in   Risk Profile Score in
                                                                     2014                    2014
Administrative data
Sector of activity: omitted = manufacturing
  Trade                                                        -6.161** -6.598**      -0.115** -0.120**
                                                                (2.783)   (2.748)      (0.048)  (0.047)
 Services                                                        -0.607    -0.826       -0.016   -0.019
                                                                (2.565)   (2.536)      (0.044)  (0.044)
Female owner                                                     -1.924    -2.389       -0.078   -0.085
                                                                (3.396)   (3.392)      (0.061)  (0.061)
No employee                                                   20.330*** 20.399***     0.299*** 0.300***
                                                                (2.147)   (2.138)      (0.036)  (0.036)
Number of employees                                               0.120     0.240        0.006    0.007
                                                                (0.240)   (0.238)      (0.005)  (0.005)
The firm was audited in 2014                                            -20.187***             -0.264***
                                                                          (3.231)               (0.069)
Amount of fine following an audit in 2014 (in                            0.150***              0.002***
TJS/1000)                                                                 (0.034)               (0.000)
Baseline survey data
Firm has an accountant                                           3.353      3.038        0.041     0.037
                                                               (2.089)    (2.068)      (0.037)    (0.037)
Share of technological practices implemented                     3.144      3.229        0.029     0.030
                                                               (2.030)    (2.011)      (0.038)    (0.038)
Number of visits to tax authority office in Jan-Jun 2014       -1.029*    -0.921       -0.011      -0.010
                                                               (0.586)    (0.596)      (0.013)    (0.013)
Time spent on tax related activities during a typical            0.288      0.303      0.008*     0.008*
month (hours)                                                  (0.271)    (0.269)      (0.004)    (0.004)
 Number of times tax inspectors visited the company in           0.514      0.398        0.003     0.002
Jan-Jun 2014                                                   (0.739)    (0.741)      (0.015)    (0.014)
Ever used e-filing (with another company)                        2.063      2.075       -0.000     -0.000
                                                               (2.270)    (2.261)      (0.041)    (0.041)
Think that it is common for firms to make informal              -2.594     -2.411     -0.083**   -0.080**
payments                                                       (2.052)    (2.033)      (0.036)    (0.036)
Observations                                                1,057       1,057         1,057      1,057
R-squared                                                   0.203       0.221         0.141      0.151
Mean Dependent Variable                                    63.864      63.864         0.494      0.494
Notes: OLS regressions with standard errors presented in parentheses controlling for dummies for the tax
Rayon. ***, **, * indicate significance at 1, 5 and 10%.




                                                         43
 Table 3: Impact on E-filing Adoption
                                                          Mean [SD]   Difference between          P-values of the
                                                            Control control group and […]          test Group A=
                                                             group   Group  A     Group   B   N       Group B
PANEL A: Administrative data from TC (Aug 2014-Dec 2015)
 Used E-filing                                                0.59   0.337***       0.038   1498      0.000***
                                                            [0.492]   (0.023)      (0.033)
 Used E-filing conditional on survival                       0.641    0.34***       0.042   1275      0.000***
                                                             [0.48]   (0.022)      (0.035)
 Still using e-filing in Dec 2015                            0.548   0.238***       0.046   1498      0.000***
                                                            [0.498]   (0.027)      (0.034)
PANEL B: Endline survey data (Feb 2016)
 Firm used electronic filing to submit tax reports in        0.564   0.439***       0.038   1263      0.000***
 2015                                                       [0.496]   (0.022)      (0.037)
 Found out about e-filing during intervention training       0.796   0.202***      0.054* 1263        0.000***
                                                            [0.403]   (0.018)      (0.029)
 Found out about e-filing from business network               0.17  -0.169*** -0.047* 1263            0.000***
                                                            [0.376]   (0.017)      (0.027)
 Found out from another source (other training, tax          0.033  -0.033***       -0.007 1263        0.011**
 Committee publication...)                                   [0.18]   (0.008)      (0.013)
 Notes: Column 1: Standard deviations presented in brackets. Columns 2-3: coefficients and standard errors (in
 parentheses) from an OLS regression of the firm owner/firm characteristic on treatment dummies, controlling for
 strata dummies. ***, **, * indicate significance at 1, 5 and 10%.




                                                           44
Table 4: Determinants of E-filing Adoption in Groups B and C
                                                                 (1)       (2)              (3)
                                    Dependent variable:                  Firm used E-Filing
                                                                 Group B & C           Group B & C
                                                                  All Sample        Only  Legal entities
Treatment variables
 Group B (Training alone)                                       0.026      0.033                0.015
                                                               (0.033)    (0.033)              (0.040)
Administrative data (baseline)
 Female owner                                                    -0.009   -0.009               0.010
                                                                (0.063)  (0.063)              (0.081)
 No employee                                                  -0.133*** -0.092**             -0.147***
                                                                (0.040)  (0.041)              (0.052)
 Number of employees                                           0.018*** 0.018***             0.017***
                                                                (0.005)  (0.005)              (0.005)
 The firms was audited in 2014                                  -0.034    -0.068               0.001
                                                                (0.084)  (0.085)              (0.085)
 Standardized risk profile score in 2014λ                               -0.072***            -0.060***
                                                                         (0.019)              (0.019)
Survey data (baseline)
 Firm has an accountant                                         0.059         0.062               0.039
                                                              (0.038)       (0.038)              (0.048)
 Share of technological practices implemented                   0.023         0.031               0.013
                                                              (0.043)       (0.043)              (0.049)
 Number of visits to tax authority office in Jan-Jun 2014       0.012         0.008               0.005
                                                              (0.017)       (0.017)              (0.016)
 Time spent on tax related activities during a typical       0.010**       0.011**               0.011*
 month (hours)                                                (0.005)       (0.005)              (0.006)
 Number of times tax inspectors visited the company in         -0.016        -0.018              -0.039*
 Jan-Jun 2014                                                 (0.017)       (0.017)              (0.022)
 Ever used e-filing (with another company)                     -0.018        -0.016               -0.029
                                                              (0.050)       (0.050)              (0.052)
 Think that it is common for firms to make informal            -0.006        -0.010               0.036
 payments                                                     (0.043)       (0.043)              (0.047)
Observations                                                     894           894                 634
R-squared                                                       0.173         0.188               0.161
Mean Dep Var in control                                         0.590         0.590               0.563
Notes: Robust standard errors in parentheses. Omitted group are firms in the control group that received
a general tax training. Regressions include fixed effects for strata. λ: for individual entities there are no
risk profile scores so missing values were replaced by the mean of the variable. ***, **, * indicate
significance at 1, 5 and 10%.




                                                         45
Table 5: Impact of E-filing Adoption on Main Outcomes
                                              (1)         (2)          (3)          (4)        (5)      (6)       (7)
                Dependent variables:        # visits  Total time Think that it              Inverse   Median
                                          per month spent on tax- is common      Tax paid hyperbolic of Tax Tax paid
                                            to tax      related   for firms to   in 2015α of tax paid paid in in 2015α
                                          authority activities by     make       (Admin.    in 2015   2015α (Survey
                                           office in   month in     informal        data)  (Admin. (Admin. data)
                                             2015    2015 (hours) payments        (in TJS)   data)     data)   (in TJS)
E-filing Impact, Overall Sample:
E-filing Impact (IV) (all firms)           -1.4***     -4.7***         0.07        5,857       0.087        -155    -4,395
                                            (0.1)       (1.1)         (0.084)     (8,735)     (0.631)      (474)   (22,286)

Observations                                1,263       1,252          1,263       1,498       1,498       1,498     1,263
Mean outcome control group (B&C)             0.8         33.5          0.613      28,062       8.827       1,605    32,786
Mean outcome group A                         0.3         31.9          0.632      29,996       8.83        1,551    30,806
E-filing Impact, Legal Entities Sample:
E-filing Impact (IV) (Legal Entities)  -1.5***         -4.8***         -0.025      7,217       -0.049       -216    -6,980
                                        (0.1)           (1.2)         (0.093)    (10,937)     (0.771)      (591)   (28,230)

Observations                                 934         925            934        1,096       1,096       1,096     934
Mean outcome control group (B&C)             0.8         34.1          0.617      30,971       8.404       1,660    38,192
Mean outcome group A                         0.3         32.4          0.608      33,533       8.41        1,586    35,082
Impact by Baseline Risk Profile Score:
 E-filing Impact (IV) for below-      -1.6***          -7.0***         -0.236     -15,099    -1.920**    -1,896*    -59,215
median risk score                      (0.1)            (1.9)         (0.144)    (18,966)     (0.910)    (1,023)   (51,804)
 E-filing Impact (IV) for above-           -1.3***      -2.8*          0.161     25,357**      1.060     1,145*    35,866*
median risk score                           (0.1)       (1.5)         (0.119)    (12,728)     (1.075)     (671)    (19,399)

P-val difference (Low risk / High risk)     0.031       0.085          0.035       0.078       0.034       0.013     0.081

 Observations                                 934           925           934       1,067       1,067      1,067       934
 Mean Outcome in control group (B&C) for:
  … Firms below median risk score              0.8          34.4         0.637     39,686      10.076      2,295      53,095
  … Firms above median risk score              0.8          33.7         0.596     23,645       7.151      1,090      22,241
 Mean Outcome in group A for:
  … Firms below median risk score              0.3          32.2        0.5690     36,116       9.529      1,785      37,140
  … Firms above median risk score              0.3          32.6         0.652     32,433       7.604      1,455      32,803
Note: The first two parts of the table present results from 2SLS regressions measuring the impact of e-filing adoption
(instrumented by assignment to group A) on outcomes presented in columns. The second part of the table presents results
from 2SLS regressions measuring the heterogeneous impact of e-filing adoption (instrumented by assignment to group A)
by risk profile scores. For the heterogeneous analysis, samples include only legal entities because individual enterprises do
not have risk scores. Robust standard errors in parentheses. All regressions include control dummies for strata. α: truncated
at the 99th percentile. ***, **, * indicate significance at 1, 5 and 10%.




                                                                 46
Table 6: Impact on Informal Payments Using a List Experiment
                           (1)      (2)           (3)       (4)                         (5)          (6)
                                                            Below median of 2014       Above median of 2014
                                      All Sample               risk profile score        risk profile score
                               Group B&C Group A            Group B&C Group A Group B&C Group A
Short list (4 questions without the question on informal payments):
  Mean                             1.119        1.146          1.109          1.182      1.088         1.075
  SE                                0.02        0.027           0.03          0.046      0.027         0.035
  N                                 361          246            138             99        136           80
Long list (5 questions, including the question on informal payments):
  Mean                             1.194        1.166          1.252          1.143      1.187         1.214
  SE                               0.029        0.033           0.05           0.05      0.051         0.06
  N                                 397          259            151             98        134           98
Difference short list - long list:
  Difference                      0.075**        0.02         0.143**        -0.039      0.098*       0.139*
  SE Difference                    0.036        0.042           0.06          0.068      0.057         0.074
  P-value T-test Difference        0.038        0.643          0.017          0.569      0.087         0.06
Difference in Difference with Control group:
  Difference in Difference                     -0.055                       -0.182**                   0.041
  SE Difference                                 0.056                         0.092                    0.093
  P-value T-test Difference                     0.326                         0.048                    0.658
Note: Endline survey data, Feb 2016. Columns 3-6: samples include only legal entities because individual
enterprises do not have risk scores. ***, **, * indicate significance at 1, 5 and 10%.




                                                        47
Table 7: Impact of E-filing Adoption: Mechanisms
                                             (1)         (2)       (3)       (4)         (5)          (6)             (7)
                                                     Administrative data                            Survey data
                Dependent variables:                                                    Monthly time spent by type of task:
                                           # months                                                  (hours)
                                             with a   Tax paid                                   Submit tax       Prepare
                                            positive   in 2015 Tax paid Tax paid              returns and get    primary
                                          amount of divided in Jan- in Jul-                    reconciliation documents
                                          tax paid in by # of    June     Dec         Collate act (including used for tax
                                             2015     payments 2015      2015         records travel time)       purposes
E-filing Impact, Overall Sample:
 E-filing Impact (IV) (all firms)            -0.16     2,936**    3,725      242       -0.04       -3.33***          -1.35
                                            (0.72)     (1,397)   (4,693)   (4,515)     (0.4)        (0.14)          (0.97)

Observations                                1,498       1,267     1,498     1,498      1,252         1,252           1,252
Mean outcome control group (B&C)            7.38        3,772    13,766    14,331      21.78         1.69            9.49
Mean outcome group A                        7.31        4,578    14,998    14,482      21.78         0.59            8.99
E-filing Impact, Legal Entities Sample:
 E-filing Impact (IV) (Legal Entities)   -0.42         4,077**    4,386      426        0.29       -3.54***          -1.51
                                        (0.84)         (1,897)   (5,882)   (5,647)     (0.47)       (0.18)          (1.09)

Observations                                1,096        881      1,096     1,096       925           925            925
Mean outcome control group (B&C)            6.54        4,441    14,973    16,046      22.26         1.84            9.44
Mean outcome group A                        6.44        5,606    16,536    16,295      22.36          0.6            8.89
Impact by Baseline Risk Profile Score:
 E-filing Impact (IV) for below-       -2.3*            2,669 -12,106 -9,030             -0.7       -3.9***           -2.4
median risk score                      (1.2)           (2,785) (10,145) (9,235)         (0.7)        (0.3)           (1.7)
 E-filing Impact (IV) for above-        0.9            6,270** 18,089** 7,944           1.2*        -3.2***           -0.8
median risk score                      (1.1)           (2,821) (7,169) (6,604)          (0.6)        (0.2)           (1.4)

P-val difference (Low risk / High risk)     0.051       0.382     0.016     0.139      0.042         0.025           0.479

 Observations                                 1,067        880       1,067  1,067         925          925             925
 Mean Outcome in control group (B&C) for:
  … Firms below median risk score             8.279       4,509     20,104 19,783        22.37        1.85            9.67
  … Firms above median risk score              5.11       4,344     10,460 13,073        22.13        1.82            9.18
 Mean Outcome in group A for:
  … Firms below median risk score              7.67       5,379     17,054 17,540        22.14        0.61             8.9
  … Firms above median risk score              5.47       5,926     16,786 15,766        22.61        0.58            8.870
Note: The first two parts of the table present results from 2SLS regressions measuring the impact of e-filing adoption
(instrumented by assignment to group A) on outcomes presented in columns. The second part of the table presents results
from 2SLS regressions measuring the heterogeneous impact of e-filing adoption (instrumented by assignment to group A)
by risk profile scores. For the heterogeneous analysis, samples include only legal entities because individual enterprises do
not have risk scores. Robust standard errors in parentheses. All regressions include control dummies for strata. α: truncated
at the 99th percentile. ***, **, * indicate significance at 1, 5 and 10%.




                                                                 48
Appendix Tables

Table A1: Attrition at Endline Survey
                                                   Mean [SD] Difference between                 P-values of the test:
                                                    Control control group and […]              Group A= Group A=
                                                     group   Group A Group B             N     Group B Group B= 0
Survey completed and firm still operating             0.84    0.007          0          1498    0.786        0.932
                                                    [0.366]  (0.021)     (0.026)
Survey completed and firm liquidated                 0.127    -0.008      -0.023        1498    0.506        0.586
                                                    [0.333]  (0.019)     (0.023)
Survey not completed: not available or not found     0.023    0.009       0.023*        1498    0.307        0.226
                                                     [0.15]  (0.009)     (0.014)
Survey not completed: moved to another town           0.01   -0.008*         0          1498    0.205         0.11
                                                    [0.099]  (0.004)     (0.007)
Notes: Endline survey data 2015. Column 1: Standard deviations presented in brackets. Columns 2-3: coefficients and
standard errors (in parentheses) from an OLS regression of the firm owner/firm characteristic on treatment dummies,
controlling for strata dummies. ***, **, * indicate significance at 1, 5 and 10%.




                                                             49
Table A2: Balance Checks by Initial Treatment Groups
                                                Mean [SD]         Difference between             P-values of the
                                                 Control         control group and […]           test Group A=
                                                  group
                                                   Group A                   Group B      N         Group B
PANEL A: Administrative data from Tax Committee (2014)
 Legal entitiesβ                                   0.734             -           -       1,498          -
                                                  [0.442]
 Sector of activity is Tradeβ                      0.413             -           -       1,498          -
                                                  [0.493]
 Sector of activity is Servicesβ                   0.419             -           -       1,498          -
                                                  [0.494]
 Sector of activity is Manufacturing                0.11            0.02      0.025*     1,498        0.699
                                                  [0.313]         (0.012)     (0.014)
 Female owner                                      0.072            0.027      0.012     1,498        0.474
                                                  [0.259]         (0.016)     (0.019)
 No employee                                       0.414           -0.019       -0.03    1,498        0.754
                                                  [0.493]         (0.027)     (0.033)
 Number of employees                               3.257           -0.089      0.119     1,498        0.376
                                                  [3.431]         (0.189)      (0.24)
 The firms was audited in 2014                     0.036           -0.006      0.000     1,498        0.642
                                                  [0.187]          (0.01)     (0.013)
 Risk profile score in 2014λ                        62.1              1          2.8     1,067        0.433
                                                   [28.7]           (1.8)       (2.3)
 Risk profile score in 2014 above medianλ          0.481          -0.004       0.034     1,067        0.351
                                                    [0.5]         (0.033)      (0.04)
PANEL B: Baseline Survey Data (2014)
 Firm has an accountant                             0.72         0.044*        0.047     1,498        0.903
                                                  [0.449]        (0.025)      (0.03)
 Share of technological practices                  0.547           0.016      -0.023     1,498        0.159
                                                  [0.432]        (0.023)     (0.028)
 Number of visits to tax authority office in       6.373         0.136**        0.02     1,498       0.095*
 Jan-Jun 2014                                     [0.978]        (0.061)     (0.066)
 Time spent on tax related activities during       6.043         0.279*      0.461**     1,483        0.451
 a typical month (hours)                          [2.681]        (0.166)     (0.227)
 Number of times tax inspectors visited            1.334          -0.032      -0.016     1,498        0.808
 the company in Jan-Jun 2014                      [0.989]        (0.059)     (0.063)
 Ever used e-filing (with another company)         0.127           0.000        0.04     1,498        0.111
                                                  [0.333]         (0.019)     (0.025)
 Think that it is common for firms to make         0.184           0.000       -0.014    1,498        0.591
 informal payments                                [0.388]         (0.022)     (0.026)
P-values of joint orthogonality tests:                            0.515       0.157
Notes: Column 1: Standard deviations presented in brackets. Columns 2-3: coefficients and standard errors
(in parentheses) from an OLS regression of the firm owner/firm characteristic on treatment dummies,
controlling for strata dummies. β: variables used for stratification. λ: Risk profile scores are only calculated
for legal entities. ***, **, * indicate significance at 1, 5 and 10%.
                                                            50
Table A3: Balance Checks on Endline Survey Sample
                                              (1)           (2)     (3)          (4)             (5)      (6)        (7)             (8)      (9)
                                                                            Below median of 2014 risk           Above median of 2014 risk
                                                 All sample                  profile score (low risk)             profile score (high risk)
                                       Mean [SD]                            Mean [SD]                            Mean [SD]
                                                       Diff. in                            Diff. in                             Diff. in
                                         Control                           Control group                        Control group
                                                      Group A                             Group A                              Group A
                                      group (B&C)                 N           (B&C)                   N            (B&C)                   N
PANEL A: Administrative data from Tax Committee (2014)
 Legal entitiesβ                          0.737            -    1,263             1               -       486         1               -       448
                                          [0.44]                                 [0]                                 [0]
 Sector of activity is Tradeβ             0.401            -    1,263           0.346             -       486       0.252             -       448
                                          [0.49]                               [0.477]                             [0.435]
 Sector of activity is Servicesβ          0.425            -    1,263           0.512             -       486       0.567             -       448
                                         [0.495]                               [0.501]                             [0.496]
 Sector of activity is Manufacturing      0.124         0.008 1,263             0.104            0.004    486       0.119           0.011     448
                                          [0.33]       (0.012)                 [0.306]          (0.018)            [0.324]         (0.023)
 Female owner                              0.08         0.018 1,263             0.073            0.007    486       0.044           0.021     448
                                         [0.272]       (0.017)                  [0.26]          (0.025)            [0.206]         (0.023)
 No employee                              0.392        -0.015 1,263             0.197            0.003    486       0.422           -0.04     448
                                         [0.488]       (0.026)                 [0.399]          (0.037)            [0.495]         (0.047)
 Number of employees                      3.396        -0.157 1,263             4.294          -0.703**   486       3.519            0.16     448
                                         [3.563]       (0.191)                 [3.795]          (0.321)            [3.967]         (0.375)
 The firms was audited in 2014             0.03        -0.009 1,263             0.062            -0.03*   486       0.015           0.011     448
                                         [0.172]       (0.009)                 [0.242]          (0.018)            [0.121]         (0.014)
 Risk profile score in 2014λ               61.9          -0.7    934             41.7             1.7*    486        83.6            -1.4     448
                                          [27.8]         (1.7)                  [10.5]            (0.9)             [24.1]          (2.2)
 Risk profile score in 2014 above         0.483        -0.025 934                  0             0.000    486          1            0.000     448
                                           [0.5]       (0.033)                    [0]           (0.000)               [0]          (0.000)
PANEL B: Baseline Survey Data (2014)
 Firm has an accountant                    0.73          0.04 1,263             0.796           -0.026    486       0.752           0.074*    448
                                         [0.444]       (0.025)                 [0.404]         (0.039)             [0.433]          (0.038)
 Share of technological practices         0.544         0.035 1,263             0.664           -0.026    486       0.651          0.108***   448
                                         [0.438]       (0.023)                 [0.417]          (0.04)             [0.408]          (0.039)
 Number of visits to tax authority        6.398        0.13** 1,263             6.536           0.086     486       6.459            0.183    448
 office in Jan-Jun 2014                  [0.994]       (0.062)                 [1.067]         (0.106)             [1.149]          (0.119)
 Time spent on tax related activities     6.139         0.111 1,252             6.117           0.075     483       6.686            0.019    442
 during a typical month (hours)          [2.709]       (0.162)                  [2.35]         (0.236)             [3.171]           (0.36)
 Number of times tax inspectors           1.309        -0.038 1,263              1.26           -0.061    486       1.167            0.048    448
 visited the company in Jan-Jun 2014     [0.924]       (0.059)                 [0.832]         (0.114)             [0.874]          (0.088)
 Ever used e-filing (with another         0.152         -0.01 1,263             0.208            -0.04    486       0.185            0.017    448
 company)                                [0.359]        (0.02)                 [0.406]         (0.036)             [0.389]          (0.039)
 Think that it is common for firms to     0.186         0.005 1,263              0.26           -0.053    486       0.178            0.009    448
 make informal payments                  [0.389]       (0.022)                 [0.439]         (0.037)             [0.383]          (0.037)
P-values of joint orthogonality tests:              0.558                                0.4                               0.292
Notes: Columns 1, 4, 7: Standard deviations presented in brackets. Columns 2, 5, 8: coefficients and standard errors (in parentheses)
from an OLS regression of the firm owner/firm characteristic on a treatment variable (group A), controlling for strata dummies. β:
variables used for stratification. λ: Risk profile scores only calculated for legal entities. ***, **, * indicate significance at 1, 5 and
10%.


                                                                          51
Table A4: Correlation between 2014 Risk Score and 2015 Tax Paid and Audits Results
                                      (1)       (2)         (3)         (4)           (5)                 (6)
                                                                                   Total audit Inverse hyperbolic of
            Dependent variables:                         Inverse hyperbolic of    fine in 2015 total audit fine in 2015
                                    Tax paid in 2015        tax paid in 2015     (in TJS 1000)      (in TJS 1000)
Risk Profile Score in 2014            346      539       -1.287*** -1.229***           1.8             0.037**
                                    (1,904) (2,416)       (0.119)      (0.144)        (2.1)             (0.017)
The firm was audited in 2015                                                        58.5***           10.582***
                                                                                      (9.1)             (0.119)
Controlling for:
- Trade, Services, gender, no         yes       yes         yes        yes            yes                 yes
employee, number of employees
and tax office.
- Index of profit and turnover in              yes                     yes
Jun and Dec 2015
Observations                        1,424      934        1,424        934           1,424                1,424
R-squared                           0.202     0.505       0.335       0.368          0.190                0.978
Mean Dependent Variable            33073 35969            8.561       9.118           6.9                 1.249
Notes: OLS regressions with standard errors presented in parentheses. ***, **, * indicate significance at 1, 5 and 10%.




                                                              52
Table A5: Impact on E-filing Adoption by Baseline Risk Profile Scores
                                                  (1)          (2)      (3)            (4)         (5)     (6)         (7)           (8)     (9)
                                                                                  Below median of 2014 risk      Above median of 2014 risk
                                                     All sample                     profile score (low risk)       profile score (high risk)
                                             Mean [SD]                             Mean [SD]                      Mean [SD]
                                                           Diff. in                              Diff. in                        Diff. in
                                               Control                               Control                        Control
                                                          Group A                               Group A                         Group A
                                            group (B&C)                  N        group (B&C)               N    group (B&C)               N
PANEL A: Administrative data from TC (Aug 2014-Dec 2015)
 Used E-filing                                  0.604       0.325*** 1,498           0.626      0.321*** 547          0.547       0.39*** 520
                                               [0.489]        (0.02)                [0.485]      (0.032)             [0.499]      (0.035)
 Used E-filing conditional on survival          0.658       0.326*** 1,275           0.681      0.32*** 488           0.592      0.365*** 474
                                               [0.475]       (0.019)                [0.467]      (0.029)             [0.492]      (0.035)
 Still using e-filing                           0.563       0.224*** 1,498           0.614      0.248*** 547          0.537      0.209*** 520
                                               [0.496]       (0.024)                [0.488]      (0.037)             [0.499]      (0.042)
PANEL B: Endline survey data (Feb 2016)
 Firm used electronic filing to submit            0.579      0.427*** 1,263          0.526        0.475*** 486           0.53       0.488*** 448
 tax reports in 2015                             [0.494]      (0.019)                [0.5]         (0.031)               [0.5]       (0.032)
 Found out about e-filing during                  0.815      0.185*** 1,263          0.792        0.202*** 486          0.796       0.211*** 448
 intervention training                           [0.388]      (0.015)               [0.406]        (0.025)             [0.403]       (0.026)
 Found out about e-filing from business           0.154     -0.154*** 1,263          0.176       -0.172*** 486          0.152       -0.158*** 448
 network                                         [0.362]      (0.014)               [0.382]        (0.024)              [0.36]       (0.024)
 Found out from another source (other              0.03     -0.031*** 1,263          0.031        -0.03*** 486          0.052       -0.053*** 448
 training, tax Committee publication...)         [0.172]      (0.006)               [0.174]         (0.01)             [0.222]       (0.014)
Notes: Columns 1, 4, 7: Standard deviations presented in brackets. Columns 2, 5, 8: coefficients and standard errors (in parentheses)
from an OLS regression of the firm outcome on a treatment variable (group A), controlling for strata dummies. Columns 4 to 9: Sample
restricted to legal entities because risk profile scores only calculated for legal entities. ***, **, * indicate significance at 1, 5 and 10%.




                                                                             53
Table A6: Cost-Effectiveness Analysis (only Accounting for Reduced Compliance Costs)
                                                                     Control
                                                                     Group          Group A        Group B
 Number of firms                                                       608            594             296
 Number of training conducted                                           12             36             16
Program costs (in USD):
 Training organization (specific by group)                            7,573          10,167          4,931
 Logistical help to register to e-filing (group A only)                  0           4,696             0
 Total program costs                                                  7,573          14,863          4,931
Cost effectiveness analysis:
 Cost per firm included in treatment (in USD)                           12             25             17
 Additional cost with respect to control Group (in USD)                                13
 Program impact on e-filing adoption (from table 3)                                   0.34
 Cost per additional e-filing adoption (in USD)                                        37
 Program impact on compliance costs (in hours saved) (from Table 5)                   -4.7
                                             α
 Amount of money saved monthly by firms (in USD)                                       5.5
 Number of months for private benefits in term of time                                  7
 saved to exceeded program costs
Note: Training costs include direct costs of organizing trainings (trainers' salaries, equipment rental,
calls to invite participants) and costs to develop training materials. Costs related to the logistical help to
register include costs to call and visit firms for the software installation and costs related to firms'
registrations. Exchange rate from Oanda.com on January 1st 2016: USD 1 = TJS 6.99. α: assuming the
wage of the person in charge of tax declaration is in average USD 178 per month or USD 1.11 per
hour.




                                                          54
Table A7: Share of Firms that Dropped Out from E-filing
                                                                                (1)
                                                         Among those who registered for e-filing before
                                                                            June 2015:
                                                          Stopped e-filing at some point after registration
 Group A x Below median risk profile score in 2014                           0.060**
                                                                              (0.030)
 Group A x Above median risk profile score in 2014                           0.172***
                                                                              (0.032)
 Above median risk profile score in 2014                                       -0.029
                                                                              (0.027)
Observations                                                                     623
R-squared                                                                       0.144
Mean Dependent variable in control group (B&C) for:
…all firms (with a risk profile score)                                                0.028
…firms below median risk profile score in 2014                                        0.031
…firms above median risk profile score in 2014                                        0.024
Mean Dependent variable in group A for:
…all firms (with a risk profile score)                                                0.151
…firms below median risk profile score in 2014                                        0.089
…firms above median risk profile score in 2014                                        0.215
Notes: OLS regressions with standard errors presented in parentheses controlling for gender of the owner, no
employees, number of employees, strata dummies and all baseline survey variables presented in Table 2. Risk
profile scores only calculated for legal entities. ***, **, * indicate significance at 1, 5 and 10%.




                                                         55
Table A8: Heterogeneous Impact by Baseline Risk Profile Score Including Additional Controls
                              (1)        (2)        (3)          (4)      (5)             (6)     (7)          (8)        (9)     (10)
                                                             α
                                            Tax paid in 2015                                 Inverse hyperbolic of tax paid in 2015
 E-filing Impact (IV) for    -12,821 -11,116     -15,762       -16,833 -10,194          -1.677* -1.676*    -1.854** -1.952* -1.543
below-median risk score     (20,536) (17,728)    (19,950)     (19,834) (20,443)         (0.925) (0.885)     (0.915)     (0.999) (1.013)
 E-filing Impact (IV) for
                          25,016** 29,879**         31,741**     26,276** 38,812** 1.026 2.031**                1.080       0.981 2.098**
above-median risk score
                          (12,738) (13,143)         (14,669)     (13,108) (15,771) (1.077) (0.881)             (1.110)     (1.093) (0.894)

P-val difference (Low        0.121      0.068         0.064        0.074        0.065   0.057     0.003         0.043       0.049     0.008
risk / High risk):
Controlling for                                  Technological Time           All                          Technological Time          All
heterogeneity of E-filing               No          Practices    spent on variables                No         Practices      spent variables
impact with variable:       Trader employee Implemented            tax     together Trader employee Implemented on tax together
Observations                 1,067     1,067          1,067       1,057     1,057      1,067     1,067          1,067        1,057 1,057
R-squared                    0.049     0.106          0.046       0.046     0.087      0.119     0.287          0.128        0.120 0.262
Note: 2SLS regressions measuring the heterogeneous impact of e-filing adoption (instrumented by assignment to group A) by risk
profile score. Samples include only legal entities because individual enterprises do not have risk scores. Robust standard errors in
parentheses. All regressions include control dummies for strata. α: truncated at the 99th percentile. ***, **, * indicate significance at 1, 5
and 10%.




                                                                           56
Appendices
A     Script Used to Invite Firms to Training
Good day. My name is “OPERATOR NAME”. We are calling you from company “IM-
PLEMENTING PARTNER NAME” LLC. Our company in cooperation with the Tax Com-
mittee and the International Finance Corporation is holding a seminar. Your company -
LLC “COMPANY NAME” is in the short-list of taxpayers we are inviting to participate
in our training. Training will take place on “DATE AND TIME” at “PLACE”. All partic-
ipants will be given training and guidance materials on tax and taxation. Some ﬁrms will
be selected to be registered for e-ﬁling system free of charge. Due to limited resources, we
cannot register all ﬁrms now, so the ﬁrms selected will be based on chance as chosen by a
computer program. In this regard, please prepare and bring the following documents:
    • Copy of the registration certiﬁcate
    • Extract from the Uniﬁed State Registry
    • Passport copy of Director
    • Company Seal
This training will provide important information but it is optional and there will be no
penalty for your ﬁrm if you do not attend. Thanks in advance for your participation.
Telephone number for inquiries and information: xxx-xx-xx


B     Measurement of Key Variables and Outcomes
Baseline administrative data:
    • Legal entities: equal 1 if the ﬁrm is a legal entity (75% of the sample) and 0 if it is
      an individual enterprise.
    • Sector of activity: Sector of activity reported by the ﬁrm when registering
    • Female owner: the ﬁrm owner is a woman
    • Number of employees: Number of employees reported by the ﬁrm to the tax authority
    • The ﬁrm was audited in 2014: equal 1 if the tax administration reported that the
      ﬁrm was audited in 2014.
    • Amount of ﬁne following an audit in 2014 (in TJS/1000): amount of ﬁne in 2014 in
      TJS/1000, as reported in administrative data. Equal zero if the ﬁrm was not audited
      in 2014.

                                             57
   • Risk score in 2014: Risk score calculated by the tax authority in 2014 (see section on
     Data for a detailed description of the risk score)

   Baseline survey data: (self-administered survey)

   • Firm has an accountant: the ﬁrm reported that an accountant (full-time or part-time)
     is responsible for tax accounting (as opposed to the ﬁrm owner himself).

   • Share of technological practices implemented: Share of technological practices im-
     plemented by the ﬁrm from the following list: “the ﬁrm has high speed internet on
     premises,” “the ﬁrm uses emails for business communication,” and “the ﬁrm main-
     tains accounting and tax records in your organization electronically using a specialized
     program (such as Excel).”

   • Number of visits to tax authority oﬃce in Jan-Jun 2014: Number of times any
     employees of the ﬁrm visited a tax authority oﬃce between January and June 2014.

   • Time spent on tax-related activities during a typical month (hours): This variable is
     the sum of the total amounts of time spent during a typical month (in January-June
     2014) on tax report preparation and on visiting a tax oﬃce (including travel time).

   • Number of times tax inspectors visited the company in Jan-Jun 2014: number re-
     ported by the ﬁrm owner

   • Ever used e-ﬁling (with another company): Equals 1 if the respondent reported that
     he/she had ever used e-ﬁling to submit a tax document with another company.

   • Think that it is common for ﬁrms to make informal payments: answered that the
     following statement is always or often true: “It is common for ﬁrms in my line of
     business to have to pay some irregular ‘additional payments/gifts’ to tax oﬃcials.”

E-ﬁling use and registration:

   • Used E-ﬁling: equal 1 if the ﬁrm used e-ﬁling at any time during the study period.

   Firm outcomes from endline survey data:

   • Number of visits per month to tax authority oﬃce in 2015: sum of all visits reported
     by the ﬁrm owner to any tax oﬃce and for any reason.

   • Total time spent on tax-related activities by month in 2015 (hours): This variable is
     the sum of the total amounts of time spent during a typical month in 2015 on the
     following activities:

        – Collate records
        – Submit tax returns (including travel time)

                                             58
        – Get the reconciliation act
        – Prepare primary documents used for tax purposes

   • Think that it is common for ﬁrms to make informal payments: answered that the
     following statement is always or often true: “It is common for other companies to
     have to pay some irregular ‘additional payments/gifts’ to tax oﬃcials.”

   • Tax paid in 2015: Total amount of tax paid in 2015, calculated using the average
     (declared) amount of tax paid across two focal months of 2015 (June and December)
     multiplied by 12.

   Firm outcomes from administrative data:

   • Tax paid in 2015: Sum of all taxes paid in 2015 using monthly administrative data.

   • Inverse hyperbolic of tax paid in 2015: Inverse hyperbolic sine transformation of tax
                                               2 + 1)1/2 ).
     paid in 2015. It is deﬁned by log (yi + (yi

   • Median of Tax paid in 2015

   • Number of months with a positive amount of tax paid in 2015

List Experiment:
Respondents were asked the following question:
“Now, I will read a list of various actions that a company can take to solve or prevent
problems with the Tax Administration. After I read the entire list, I would like you to tell
me how many of these actions your company took in 2015”.

    50 percent of the sample had to choose among a short list of answers (which did not
include unoﬃcial payments) and 50 percent among a longer list (which included unoﬃcial
payments). The selection was random and was stratiﬁed on ﬁrm legal and treatment status.

 Short list of answers (50% of sample)           Long list of answers (50% of sample)
 1. Received help from trade associations        1. Received help from trade associations
                                                 2. Made unoﬃcial payments or pro-
                                                 vided free services/goods
 2. Had detailed discussions with tax oﬃcials    3. Had detailed discussions with tax oﬃcials
 3. Provided additional documents                4. Provided additional documents
 4. Pursued court action                         5. Pursued court action




                                            59
C       Cost Eﬀectiveness Analysis Using Time Saved by Firms
Administering the program (organizing the trainings, inviting ﬁrms, and providing logis-
tical support for e-ﬁling registration to Group A ﬁrms) cost $25 per ﬁrm in Group A,
compared to $17 per ﬁrm in Group B and $12 per ﬁrm in Group C (Table A6). Given
the 34 percentage point diﬀerence in take-up between Group A and the control group, and
the $13 per ﬁrm diﬀerence in program costs, the cost per additional e-ﬁling adoption in
Group A relative to the control group is $37. The lack of any signiﬁcant diﬀerence between
adoption in Group B and the control group indicates that the relevant aspect of Group A
treatment was the logistical help with registration.
    The diﬀerence in program costs per ﬁrm between Group A and Group B (cost of
logistical support for registration) is $8 per ﬁrm. Given the 30 percentage point diﬀerence
between Group A and Group B, the cost per additional e-ﬁling adoption in Group A relative
to Group B is $27.
    We can compare the program costs to the beneﬁts accruing to ﬁrms from the reduction
in compliance costs. Data limitations preclude our calculating other potential beneﬁts of
the program such as savings in tax administration costs. In addition, from the government’s
perspective, we detect no signiﬁcant average eﬀects on tax revenue (although any revenue
impact would be a transfer from ﬁrms to the government). Table 5 estimates that ﬁrms
save 4.7 hours each month they would otherwise have spent on tax-related activities. From
survey data, the average wage of the person in charge of tax declaration in ﬁrms is $178
per month (or $1.11 per hour), creating an estimated $5.5 savings per ﬁrm each month.
As such, it would take ﬁve to seven months for private beneﬁts in terms of time saved to
exceed program costs. Although ﬁrms may not necessarily be willing to pay the full costs of
the program,35 these results provide guidance for a social planner on types of interventions
that may be considered in promoting e-ﬁling adoption. Appendix Table A6 provides details
on the cost eﬀectiveness analysis.




 35
      Take-up remained quite low when ﬁrms had to pay $40 to register and obtain a token.


                                                   60