Policy Research Working Paper                    9848




 The Promise and Limitations of Information
      Technology for Tax Mobilization
                             Oyebola Okunogbe
                              Fabrizio Santoro




Development Economics
Development Research Group
July 2022
Policy Research Working Paper 9848


  Abstract
  Tax revenue in many low- and middle-income countries is                            use of technology for tax administration. It highlights the
  inadequate for funding investments in public goods and                             potential of technology to improve tax collection by helping
  human capital. With high levels of informality and limited                         to identify the tax base, monitor compliance and facilitate
  state capacity, many tax authorities have difficulty deter-                        compliance. It also identifies possible limitations to the use
  mining the true tax base and collecting taxes efficiently                          of technology arising from inadequate infrastructure and
  and equitably. Tax authorities are increasingly adopting new                       connectivity, lack of adoption or resistance by taxpayers and
  technologies to improve administrative processes, reduce                           tax collectors, lack of institutional mainstreaming, and an
  taxpayer compliance costs, and enhance their overall effec-                        unsupportive regulatory environment.
  tiveness. This paper reviews the recent literature on the




 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/prwp. The authors may
 be contacted at ookunogbe@worldbank.org or F.Santoro@ids.ac.uk.




          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
The Promise and Limitations of Information Technology
                for Tax Mobilization∗
                        Oyebola Okunogbe, World Bank
        Fabrizio Santoro, International Centre for Tax and Development




Originally published in the Policy Research Working Paper Series on November 2021. This version is
updated on July 2022.
To obtain the originally published version, please email prwp@worldbank.org.


Keywords: Technology, taxation, tax administration, tax compliance
JEL codes: H20, O33




   ∗
    Okunogbe ookunogbe@worldbank.org; Santoro F.Santoro@ids.ac.uk; We thank Moyo Arewa, Christine
Awiti, Anne Brockmeyer, Giulia Mascagni, Laura Munoz, Giovanni Occhiali, Dina Pomeranz and Abebe
Shimeles for helpful comments. This paper was supported by the African Economic Research Consortium
(AERC). Santoro acknowledges support from the Gates Foundation through the International Center for
Tax and Development DIGITAX program. Kairat Umargaliyev provided invaluable research assistance. The
findings, interpretations, and conclusions expressed in this paper do not necessarily represent the views of
the World Bank Group, its Executive Directors or the countries they represent.
   Increasing tax revenue is an important policy goal in many low- and middle-income coun-
tries (LMICs). On average, low-income countries collect 11 percent of GDP as taxes com-
pared to 22 percent in high-income countries (UNU-WIDER Government Revenue Dataset,
2021). Beyond funding public investments such as infrastructure, health and education, tax-
ation is also important for providing redistribution, and reducing reliance on foreign aid and
natural resource revenues. However, with high levels of informality and limited state capac-
ity in LMICs (Besley and Persson, 2009), many tax authorities have difficulty determining
the true tax base and collecting taxes efficiently and equitably.
   Over the last few decades, tax systems around the world have increasingly adopted new
technologies to improve administrative processes, reduce taxpayer compliance costs, and en-
hance their overall effectiveness at collecting tax revenue. These developments come in the
context of digital transformation across all facets of life: from government to business to
interpersonal relationships and much more, an increasing number of interactions are con-
ducted virtually given the rapid development and expansion of digital infrastructure. For
example, in 2020, 51 percent of the global population was connected to the internet. This
figure was 7 percent only 20 years earlier (Our World in Data/ World Bank indicators).
   Given the rapid expansion in the use of information technology by tax authorities, this
paper provides a review of the recent evidence from LMICs. We draw on a burgeoning
literature on the role of technology in tax administration to highlight key lessons on max-
imizing the potential of technology. We also discuss complementary reforms that must be
undertaken to ensure that investments in technology yield the desired results for tax mobi-
lization. While we include papers that provide descriptive evidence from many contexts, we
emphasize papers that use rigorous methods to establish a causal impact of the introduction
of technology and highlight the methods used when describing these papers.
   In the first part of the paper, we consider how technology may transform three core
functions of tax administration. First, technology may be used to define compliance, that
is, to strengthen the capacity of tax administration to identify the tax base, or the pool


                                              2
of subjects liable to remit a given tax amount. This could happen, for instance, when
technology enables the creation of a new database of taxable subjects (such as individuals or
properties). Similarly, technology could help define the actual amount to be remitted more
transparently by relying on the automatic capture of transactions as they occur, through
electronic fiscal devices (EFDs) or electronic payments.
   Second and relatedly, technology may be used to monitor compliance. Modern data
analytic tools could significantly boost the capacity of the tax administration to monitor
and enforce compliance, through the automation of crosschecks of self-reported tax liability
with other data sources. Technology-based tools also allow for analyzing large amounts of
data to support risk-based audits which would result in more targeted and transparent audit
selection.
   Third, technology may be used to facilitate compliance, that is, to simplify processes and
improve taxpayer experience. Technology could significantly reduce tax compliance costs,
which are often disproportionately borne by smaller taxpayers. For example, e-filing (often
accompanied by e-payment) of taxes can make the taxpaying experience less time consuming
and more consistent across taxpayers. As we would see, certain technologies may address
more than one of these three uses.
   While technology may also be used to improve other core functions of tax administration,
such as the capacity to perform more timely data-driven decision-making, and to run more
elaborate statistical analysis and forecasting, these internal functions are beyond the scope of
this paper. Instead, we focus on those IT solutions directly affecting taxpayers’ interactions
with the tax administration, potentially shaping their compliance.
   In the second part of the paper, we consider factors that may prevent the realization of the
potential of technology to improve tax mobilization. First, lack of appropriate infrastructure
and internet connectivity will hinder effective deployment of the technologies. Second, users
(taxpayers and tax officials) may avoid or actively resist the introduction of a new technology
due to lack of awareness and training, high adoption costs or loss of opportunity for unofficial


                                               3
behaviors. Third, the absence of a clear organizational strategy that involves high-level
change management and a proper sequencing of the digitization of different functions and
processes can hamper the effectiveness of technologies that are introduced. Fourth, without
a functional legal system to enforce tax laws, the revenue authority may be unable to collect
tax liabilities that have been uncovered with the use of technology. Relatedly, there needs
to be an appropriate regulatory framework that allows a balance between data privacy and
security and the use of new data and technology for tax purposes.
   While there are many applications of technology in tax administration, this paper focuses
on foundational technological innovations which are usually adopted at the core of the digital
journey of a tax administration, such as e-filing and e-payment, electronic fiscal devices
(including electronic sales registers and electronic billing machines), IT solutions used for
taxpayer registration and communication, as well as tax data crosschecks and third-party
data analytics. We highlight these types of technology as they communicate with each
other and are likely to perform best when included in the same technological transformation
package – usually supported by an integrated (and automated) tax administration system
(ITAS). For example, the huge amount of data produced by EFDs is unlikely to be fully
tapped without an automated and integrated data management system on the receiving
end. Likewise, without a flexible, adaptive, timely response from the authority (which is
often enabled by ITAS), e-filing is unlikely to be successful (Santoro et al., forthcoming).
   Focusing on these foundational technologies also enables us to draw on an increasing,
robust body of evidence that has been produced on them in the last few years. We expect
that the lessons from reviewing the existing evidence on the effectiveness of technologies
that are more widely used in LMICs will help in understanding the potential impact and
constraints in the adoption of more sophisticated technologies that have been introduced in
some other contexts, such as the automatic withholding of taxes from sales payments made
via credit card (Brockmeyer and Hernandez, 2016), automatic pre-filling of tax returns using




                                              4
third-party information, and the use of blockchain to verify transactions and tax records.1
       This paper contributes to both the academic literature as well as policy conversations
in LMICs. It provides a review of the booming literature on the use of technology in tax
administration (Bird and Zolt, 2008; OECD, 2021; World Bank, 2016; Kochanova, Hasnain
and Larson, 2020; Bellon et al., 2019; Santoro et al., forthcoming) and identifies outstanding
gaps in the literature for future research. It therefore complements other reviews on taxation
in developing countries (e.g. Mascagni (2018), Slemrod et al. (2017), Pomeranz and Vila-
Belda (2019)).
       More broadly, the paper adds to the literature on the use of technology in governance.
The evidence on the effectiveness of investment in e-government on strengthening government
capacity to deliver services, reduce corruption, and expand citizen participation is still rather
mixed (World Bank, 2016). On the one hand, there is evidence of effective deployment of
digital public service provision in procurement (Lewis-Faupel et al., 2016), education (Aker
and Ksoll, 2019; Escueta et al., 2017), voting (Callen and Long, 2015; Fujiwara, 2015; Callen
et al., 2016; Aker, Collier and Vicente, 2017) and social protection (Aker et al., 2016; Banerjee
et al., 2020; Muralidharan, Niehaus and Sukhtankar, 2016). On the other hand, better
information technology does not necessarily translate into improved behaviors and welfare,
especially in health (Nglazi et al., 2013, provides a review) and agriculture (Aker, Ghosh
and Burrell, 2016; Courtois and Subervie, 2015; Hildebrandt et al., 2014; Nakasone, Torero
and Minten, 2014; Mitra et al., 2013; Camacho and Conover, 2010; Fafchamps and Minten,
2012; Casaburi et al., 2014). This paper examines several studies where technology has
led to improvements in government service delivery but also describes conditions that may
preclude successful outcomes. Relatedly, the paper also addresses basic questions of digital
technology access and usability (Aker and Cariolle, 2020; Santoro et al., forthcoming).
       For policy, the paper provides guidance to policy makers as tax authorities increasingly
   1
      Blockchain technology would enable a variety of services (such as e-registration with government entities,
e-filing of taxes, and sharing of data among government agencies) under a strong security model where all
information is digitally signed and encrypted and all incoming data are authenticated and logged (Martinovic,
Kello and Sluganovic, 2017).


                                                       5
adopt electronic systems in interacting with taxpayers in a post-COVID world where remote
interactions are more prevalent. It distills lessons from the academic literature on the vast
potential of technology as well as highlights the necessary technical, institutional and political
factors for a successful shift to technology. In particular, it emphasizes how technology
adoption must be contextualized within the current administrative capacity of the country.
Lastly, by highlighting the potential policy impact of rigorous and timely policy research, in
many cases made possible by the rich data produced from the use of technology, this paper
also seeks to encourage further collaboration between policymakers and researchers.



1     Potential of Technology in Taxation

1.1    Define Compliance

Information technology is crucial for addressing the “information problem,” a core challenge
for tax administration (Alm, 2021). In order to tax, the government must be able to iden-
tify the tax base. Prior research has shown that for taxes that are subject to third-party
reporting, such as wages paid to employees and capital income paid to clients, tax evasion
is limited. However, evasion is substantially higher for taxes not subject to third party
reporting, where taxpayers self-declare their tax liability (Kleven et al., 2011).
    Tax authorities can deploy appropriate technology systems to obtain information on tax-
payer liabilities that may otherwise be hidden or manipulated by taxpayers. For third-party
information, technology can enable the efficient submission of large amounts of data from
a reporting entity (such as an employer or financial company). Technology may also facil-
itate linking databases that provide information on the different sources of taxable income
or assets of a taxpayer. The examples below will highlight the use of technology in defining
the tax base in different settings. First, in addressing the challenge of information on trans-
actions that may otherwise go unreported, we examine the use of electronic fiscal devices
that report information on sales as they occur as well as electronic payments that provide a


                                                6
paper trail of a firm’s income. In addition, we examine the use of technology in improving
taxpayer registration, with a specific focus on how IT is used in real estate tax administration
to systematically capture information on tax liabilities that are observable but not codified.


1.1.1   Obtaining Real-time Transaction Information with Electronic Fiscal De-
        vices

Electronic fiscal devices (EFDs) are machines that automatically record transactions as they
are performed and communicate this information to the tax administration through either a
mobile network or the internet. In some cases, these machines also print out receipts. Such
unprecedented flow of information holds the potential of assisting the tax administration in
knowing more about the taxpayer population.
   Ali et al. (2021) exploits the staggered rollout of EFDs across VAT-registered firms over
12 years in Ethiopia to evaluate the impact of such technology on both VAT tax payments
and proxies of the tax base. They show that EFDs significantly increase reported sales
and tax payments, while also leading to positive impacts on actual output – proxied by
employment rates. Importantly, the tax base does not appear to be negatively affected as
the paper does not find any significant relationship between EFD adoption and net entry
rate of firms.
   Mascagni, Mengistu and Woldeyes (2021) provides further evidence from the adoption of
EFDs in Ethiopia. Using a difference-in-differences approach, the paper considers the fact
that EFDs may have compliance effects across other tax types beyond VAT, following the
recent shift towards studying compliance spillovers across taxes for a complete evaluation
                             opez-Luzuriaga and Scartascini, 2019). Like Ali et al. (2021),
of any policy intervention (L´
the paper finds a significant increase in VAT payments but also in income tax payments. In
addition, the paper documents that the EFDs significantly improve the accuracy of firms’
reporting, measured by the discrepancies in reported sales between VAT and income tax
returns. On average from 2010 to 2014, 60 percent of firms reported discrepancies in turnover


                                               7
in their declarations. Such discrepancies more than halve after the introduction of EFDs as
a result of the combination of improved record-keeping practices and increased perception
of detection of misreporting.
   In addition to the positive evidence above, there are some more nuanced considerations.
First, while technology could help better identify transaction volumes and tax payable, as
shown in Ethiopia, behavioral and strategic responses from taxpayers could produce per-
verse effects. Mascagni, Mengistu and Woldeyes (2021) show that the increase in VAT sales
as a result of EFD adoption is offset by simultaneous and proportionally larger increase in
deductible costs thus substantially reducing the potential revenue gains. Similar offsetting
responses are found in the broader tax literature on firms (Ariel (2012); Carrillo, Pomeranz
and Singhal (2017); Slemrod et al. (2017); Mascagni and Nell (2022)). Second, while tech-
nology could be an effective tool to identify the tax base and the tax due more accurately,
it proves to be ineffective in those contexts lacking robust traditional enforcement. While
new technology could generate new data, it is only effective if it spurs adequate deterrence
actions from the authority, which in turn depend on the resources available for traditional
enforcement.


1.1.2    Electronic Payments for Transactions

While electronic payments such as credit cards, debit cards and mobile money are not tech-
nologies developed specifically for taxation, they play an important role in providing a paper
trail for revenues received by firms, thereby providing information on tax liabilities.2 Un-
locking the full potential of such data, however, relies on formal data sharing agreements
between revenue authorities and commercial banks and credit cards companies. Section 2.4
discusses the need for a supportive regulatory framework to facilitate such data sharing.
   In West Bengal, India, Das et al. (2022) finds that the 2016 demonetization policy re-
sulted in a higher reliance on electronic payments which in turn increased the amount of sales
   2
     As an extreme application with potential negative economic and equity implications, some countries
have introduced legislation to impose taxes on all mobile money transactions.


                                                  8
firms reported for tax purposes. By using administrative data on firms’ VAT tax returns
and exploiting the variation in the intensity of the demonetization shock across local areas,
the authors find that a 10 percent increase in electronic payments in an area increases aver-
age reported sales by 0.3 percent. The paper finds tentative evidence regarding the effect on
total tax liabilities: the effects are of similar magnitude as the effect on sales but are not pre-
cisely estimated. The authors argue, albeit cautiously, that the higher visibility of electronic
transactions reduces scope for evasion, as traders have less room for sales underreporting.
                                         aenz Somarriba (2022) use both regression dis-
   Along the same lines, Brockmeyer and S´
continuity and difference-in-difference estimations to evaluate the impact of VAT rebates for
consumers using electronic payments instead of cash in Uruguay. Using transaction-level
data on all electronic transactions and monthly firm-level VAT declarations, they examine
the impact of the rebates on credit and debit card transactions and on tax compliance. The
VAT rebate was preceded by other incentives such as lowering commissions for POS us-
age, reducing the tax withholding rates applied by card companies, and providing subsidies
to firms for POS rental, which were all part of Uruguay’s Financial Inclusion Reform, a
government-wide effort to foster electronic payments in trade and move away from cash.
   First, the paper shows that the incentives translate into an immediate 50 percent increase
in the number of debit and credit card transactions, and a 30 percent increase in the volume
of card transactions – mostly spurred by customers’ responses. On the other hand, firms
are largely unresponsive to the incentive: only a small fraction of eligible firms took up the
subsidy, 80 percent of whom already had a POS machine when taking up the subsidy. This
means that the revenue authority did not gain significant additional information on firms
previously out of their radar (extensive margin), as the vast majority of the adopters were
already visible. In terms of impact on compliance, the policy effect on reported output VAT
and net VAT liability is very small and statistically indistinguishable from zero.




                                                9
1.1.3   Using Technology for Taxpayer Registration

Technology could also significantly assist tax administration in registering taxpayers and,
more importantly, uniquely identifying them. In this sense, integrated tax administration
systems (ITAS) are crucial in streamlining the registration process. Unsurprisingly, one of
the first modules usually implemented when an ITAS is launched is the registration module,
enabling e-registrations. In the absence of a well-functioning, automated system, revenue au-
thorities often implement suboptimal practices which translate into the existence of parallel
registries, often kept on Excel spreadsheets, and, ultimately lead to administrative confu-
sion, duplication of work and difficulty in monitoring the taxpayer base (Ligomeka, 2019;
Stewart-Wilson, 2020). In light of the recent wave of automation in tax administration, more
research is needed to explore impacts on internal efficiency.
   Despite these positive considerations that can be made in support of technology in regis-
tration, technology-enabled registrations could take place over-aggressively in LMICs, along
with the increasing drive towards formalizing as much as possible, often to reach given an-
nual targets (Moore, 2020). However, it is still unclear whether mass registration would
translate into higher compliance. If anything, abundant evidence is being produced on the
poor impacts on revenues due to widespread non-filing and nil-filing. Across a variety of con-
texts, such as Costa Rica (Brockmeyer et al., 2019); Eswatini (Santoro and Mdluli, 2019),
Ethiopia (Mascagni and Mengistu, 2019), Guatemala (Kettle et al., 2016), Rwanda Mascagni,
Mukama and Santoro (2019), Uganda (Almunia et al., 2021), and Venezuela (Ortega and
Scartascini, 2015), about 20 – 50 percent of registered taxpayers either do not file taxes or file
zero taxes. New entrants also have higher compliance costs on average (Mascagni, Santoro
and Mukama, 2019).


1.1.4   Using Technology for Property Registration

Technology may also improve administration of real estate tax and the registration of prop-
erties. Knebelmann (2022) documents experiences from over 15 property tax digitalization


                                               10
projects in countries in Africa, Latin America, the Caribbean, and Asia and highlights how
satellite imagery tools and geo-referenced data systems can help in detecting, registering,
characterizing and valuing properties.
   Okunogbe (2021) provides an example of a low-cost technology investment for strength-
ening real estate tax registration and payment in Monrovia, Liberia. Only about 5 percent of
residential property owners were estimated to be on the tax register (Okunogbe, 2013). To
address the absence of a comprehensive property cadastral, the Liberia Revenue Authority
(LRA) recruited and trained young people to use open-source software on tablets to capture
the location, ownership, photograph and observable features of properties.
   Using the newly created property database, the LRA sent out four types of notices to
request payment of real estate taxes in a randomized experiment: i) a “Plain Notice” that
provided information on the property tax requirement and the procedures to pay; ii) a
“Detection Notice” that included identifying information from the new database, namely
the property owner’s name and a photograph of the property; iii) a “Penalty Notice” that
provided information on the legal consequences of not complying; and iv) a “Detection
and Penalty Notice” that combined the features of these two notices. The Detection and
Penalty notice tripled the property registration and tax payment rate among recipients,
relative to Plain notice registration rate of 4 percent and payment rate of 2 percent. The
Detection notice and the Penalty notice alone did not have a statistically significant impact
on registration or payment, indicative of the complementarities between technology-enabled
detection and traditional penalties and enforcement.
   Another illustrative example of the role of technology in identifying properties comes
from Dzansi et al. (2022) in Ghana. A municipal government in the Greater Accra region
adopted a new technology that included a geospatial database of properties and a revenue
management software. The property database was created by combining high-resolution
aerial photographs with digital registry maps. The revenue management software produced
bills and recorded payments for each property. The study implements a randomized ex-


                                             11
periment whereby a randomly selected group of tax collectors who deliver bills and collect
payments were given tablets that provided navigational directions to help the collector locate
designated properties. Tax collectors using the tablets delivered 27 percent more bills and
collected 103 percent more tax revenues than collectors in the control group.
   There are also more sophisticated tools that may help the tax authority go beyond iden-
tifying properties to also determining their valuation. For example, with machine learning,
local authorities may use satellite imagery to identify and measure properties and then de-
ploy teams to ground truth the data and collect additional data on observable characteristics
such as the quality of walls, roofs, and windows. Combining this property information with
market data on rental values may then be used to calculate the taxable value of properties
using a simplified “points based” system (Fish, 2018). Grieco et al. (2019) document how,
using this approach, the Freetown City Council in Sierra Leone almost doubled the number
on the tax register from 57,000 to about 110,000 properties and produced much higher tax
bills for the most valuable properties, owned by the wealthiest taxpayers, which have been
historically under-taxed. Given the importance of complementary automation to strengthen
tax administration, this reform also introduced technology to improve other aspects of the
real estate tax administration process beyond data collection to valuation, such as billing,
payments, appeals and enforcement.


1.2    Monitor Compliance

With the increased use of technology in defining the tax base as described in the previous
section, troves of data are constantly being generated and may be used by tax authorities
to monitor taxpayers’ compliance and detect evasion when it occurs. However, many tax
authorities in LMICs do not do routinely use this information for automated large-scale
analyses but only access third-party information on a case-by-case basis. Modern technology
provides cutting-edge data analytics tools to analyze these data and detect non-compliance.
This section describes two applications: automated cross-checks of third-party reports such


                                             12
as the VAT and, more broadly, using taxpayer characteristics to predict non-compliance and
allocate audits.


1.2.1   Automated Cross-checks of Third-Party Information

Third-party information refers to when the tax authority receives information about a taxable
transaction from multiple sources. For example, VAT is supposed to be self-enforcing because
government receives a report of a transaction from both buyer and seller who have opposite
incentives.
   Numerous studies document large discrepancies between figures reported by buyers and
sellers for the same VAT transactions as well as within-firm discrepancies where figures re-
ported through different channels by the same firm differ. In Uganda, Almunia et al. (2021)
show that sellers and buyers report different amounts in 79 percent of trading pair-month
observations. Interestingly, for about a fourth of firms, their overall reporting behavior is
“disadvantageous,” due to reporting errors in VAT returns that increase the firm’s overall lia-
bility and potentially leave money on the table. The paper estimates that VAT misreporting
cost Uganda USD 446 million in revenue from 2013-2016.
   Similarly, in Ecuador, Carrillo, Pomeranz and Singhal (2017) show that self-reported
revenue on corporate income tax returns is lower than third-party reports in 24 percent of
filings. In Rwanda, Mascagni, Mukama and Santoro (2019) find that 43 percent of VAT
payers either over- (18 percent) or under-report (25 percent) their VAT sales compared to
what is automatically recorded in the Electronic Billing Machines (EBMs). In Ethiopia,
Mascagni, Dom and Santoro (2021) document that in 60 percent of cases, total sales from
VAT returns are different from the same figure from the profit tax returns.
   Given these large discrepancies, systematically identifying them through crosschecks of
data sources seem to be a low-hanging fruit that could help recover, at least partially, the
foregone revenues. However, administrative complexity, low technical capacity, and low en-
forcement capacity for following up on uncovered discrepancies often result in these potential


                                              13
gains not being materialized (Steenbergen, 2017; Carrillo, Pomeranz and Singhal, 2017).
   More promising evidence on the use of technology in monitoring compliance by effectively
using third-party information come from countries that are further along the technology
journey. Fan et al. (2018) evaluate the impact of computerized invoices, or digital invoice
encryption, as launched in 2001, on Chinese manufacturing firms for the period 1998-2007.
Such technology implied that invoices became more difficult to falsify or forge. It also
improved the quality and speed of data flows to the tax administration, as opposed to a
manual system in which invoices were paper-based and lacked a rigorous anti-counterfeit
technology. The authors make use of a difference-in-differences design comparing outcomes
before and after the 2001 reform for firms more or less affected by the policy, as proxied by
the ratio of non-deductible inputs over outputs – with a higher ratio implying a more intense
exposure to the policy. They find that e-invoicing explains 14- 23 percent of cumulative VAT
revenues in 2002-2007, a remarkable increase due to improved technology in recording VAT
transactions. They explain such finding by a reduction in deductible inputs, which resonates
with the fact that e-invoicing significantly made it more difficult to falsify deductible claims.
   Another positive example of how technology may help monitor compliance comes from
Delhi in India where Mittal and Mahajan (2017) study the impact of a policy improving the
quality of VAT-related information available to the revenue authority. They consider a policy
reform mandating VAT filers to report firm-level transaction details with tax-identification
numbers for all transacting partners. This intervention, against a previous context in which
only aggregate VAT returns were submitted and evasion was uncovered only through costly
audits, enabled the revenue authority (in principle) to easily perform computerized cross-
checks between buyer and seller reports and automatically flag discrepancies.
   Using data on the universe of VAT returns over five years, the authors adopt a difference-
in-differences strategy exploiting the timing of the policy and comparing wholesalers, who are
more likely to be affected by the policy since they are more likely to sell to other registered
firms, and retailers, for which the third-party verification mechanism is more likely to break


                                               14
down as final customers do not have to report their purchases. The authors find that the
reform led to a 29 percent increase in average tax collections from wholesalers relative to
retailers. Importantly, the impact is largely driven by the largest wholesalers, who are in
turn monitored by a special unit dedicated to large taxpayers. This result suggests that
targeted tax enforcement can be successful when accompanied by third-party information,
with potential benefits to the overall equity of the tax system.
       In Pakistan as well, Shah (2020) examines the effectiveness of exploiting third-party
information to uncover VAT evasion and documents how the revenue authority curbed fake
VAT input claims by a computerized VAT invoice validation mechanism. More specifically,
it evaluates the impact of the introduction of a software-based system to accept or reject
VAT input tax credits. Such intervention can be considered as a proxy for the increase in the
capacity of tax administration to process the VAT information flow or paper trail, enabled
by investment in IT and replacing traditional, tax officials based, enforcement mechanisms.
The paper relies on administrative tax data for the universe of VAT returns filed in Pakistan
over 8 years and uses a difference-in-differences strategy, in which non-exporting firms, the
treated group, are compared over time to exporting firms, the control group, for which a
virtually identical level of scrutiny existed prior to the reform.
       Results indicate a 50 percent decline for treated firms subject to the VAT input automatic
checks. Importantly, success has been possible only after a decade-long period of transition
towards a fully computerized infrastructure, funded by donors. Also, the reform received
wide institutional and political support from both the government and the constitution, an
aspect that is often missing in low-capacity countries.3
       Besides information from trade partners, payments records from credit card companies
also present a source of third party information that tax authorities can use to cross-check
tax declarations and detect non-compliance. Dogan (2011) describes the introduction of a
data warehouse in Turkey that enabled access to credit card information from commercial
   3
    In Pakistan, there were constitutional provisions which restricted judiciary from staying the recovery of
evaded tax.


                                                     15
banks. After the introduction of the data warehouse, the revenue authority more heavily
relied on systematic data matching, mostly between VAT returns and credit card sales data,
and data-driven risk management to uncover large evasion. Although the paper does not
have a rigorous impact evaluation methodology, the change in patterns is remarkable. In
June 2008, just before the launch of the new technology, up to 140,000 registered taxpayers
did not report credit card sales information in their VAT returns, with 60,000 (100,000)
taxpayers showing discrepancies in more than 20 (5) percent of transactions. One year after
the introduction of the data warehouse, fewer than 20,000 taxpayers had a discrepancy rate
of greater than 20 percent. This example illustrates how technology – a well-functioning
integrated and automated data warehouse – supported by political will, can also foster inter-
institutional data sharing agreements, and create a data-oriented cross-agency administrative
ecosystem.


1.2.2   Risk-based Audits

Many tax authorities compile data from several dimensions of a taxpayer’s behavior to gen-
erate a risk profile, that reflects the likelihood that a taxpayer is evading (Khwaja, Awasthi
and Loeprick, 2011). Selecting audit cases using this data-driven approach leads to more
targeted and efficient audits as risk-based audits allocate the limited resources of the tax
authority to cases with higher likelihood of recovering significant tax revenue. As a result,
compliant firms also face a lower monitoring burden as they are less likely to be audited.
   Discrepancies from third-party reports discussed in 1.2.1 are a key indicator that may
be included in the algorithm that is used to predict firm risk. Other risk criteria commonly
incorporated into risk models are 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). In
some cases, tax authorities use machine learning methods such as a random forest algorithm
to predict firm future risk of evasion while in other cases, a more explicit formula with


                                             16
different weights given to each indicator is used. Bachas et al. (2021) report that both
methods predicted historical evasion with similar levels of accuracy for audits in Senegal.
However, the method with explicit indicators was simpler and more transparent to use by
tax inspectors compared to machine learning methods.


1.3     Facilitate Compliance

An important application of technology in taxation is to improve taxpayers’ experience
by making it easy for them to obtain information and fulfill their tax obligations. Many
tax authorities are increasingly adopting a customer service orientation towards taxpayers
(OECD, 2021). One initial step taken by tax authorities is to provide detailed instructions,
forms and guidance on their website on the compliance steps and requirements for each
type of tax. A further step is to make it possible for taxpayers to complete transactions
online, such as the ability to file and pay taxes online. Tax authorities may then also have
individual taxpayer accounts that contain the taxpayer’s transaction history with the tax
authority. The tax authority may also use electronic means of communicating with taxpayers
such as SMS and email reminders, payment confirmation and announcements.
   A major consequence of these technology-driven tools is that they allow taxpayers to
interact with the tax system in an impersonal way. In their absence, taxpayers would likely
be subject to the discretion of tax officials responsible for providing information, conducting
transactions and monitoring them in person. We consider three specific technologies: e-filing
and e-payment, electronic billing machines, and online communication.


1.3.1   E-Filing and E-Payment Tax Services

Electronic filing and payment refers to digital services that allow taxpayers to (i) file their
taxes electronically – together with a range of other tax-related activities (attach annexures,
claim refunds, make appeals, update their personal details, get assistance online, check their
filing history, etc.), usually through a web portal to which they have access or through sim-


                                              17
ilar phone applications, and (ii) to pay tax liabilities online, by means of digital platforms
(mobile money, credit card or similar electronic transfers) which efficiently connect the rev-
enue authority, private banks or digital financial service providers, and the central bank.
The main promise that e-filing and e-payment hold is that they allow taxpayers to perform
such activities in a less burdensome and more transparent way – saving travel and queuing
costs, error-prone manual inputting, and avoid physical interactions with tax officials who
may ask for bribes.
   To examine to what extent these expected benefits are realized in practice, Okunogbe
and Pouliquen (2022) study the introduction of e-filing in Tajikistan. Small business owners
faced high compliance costs as they had to submit monthly tax declarations in person at the
tax office. These frequent meetings with tax officials also provided opportunity for unofficial
payments and 32 percent of taxpayers said they expect to give gifts in a meeting with tax
officials (World Bank Enterprise Surveys, 2013). The study examines the impact of e-filing
using a randomized trial in which a randomly selected group of firms were encouraged to
e-file by providing them with training on e-filing as well as logistical help to complete the
cumbersome e-filing registration process.
   The treated group e-filed at a higher rate than the comparison group. Firms that were
more likely to be evading at baseline (measured by a risk profile score used by the tax
authority) were less likely to adopt e-filing and also more likely to stop e-filing after a few
months. Tax payments doubled among these high-risk score firms. While there was no
average effect on bribes, e-filing led to an 18 percentage point reduction in bribes among
firms less likely to have been evading. More broadly, e-filing resulted in a large reduction in
compliance cost across firms, with 5 hours saved each month (about 40 percent of the time
spent on fulfilling tax obligations).
   This intervention highlights the role of technology in not only reducing compliance costs
for taxpayers, but also affecting tax payments and unofficial payments by changing the inter-
actions between taxpayers and tax officials. It also underscores the importance of thinking


                                              18
about distribution effects in the impact of technology as certain taxpayers may be more
affected than others. Lastly, it provides lessons on the take-up of technology solutions as
taxpayers may be strategic in deciding to take up and those for whom the technology may
have greatest impact may be least likely to adopt.
   In Uganda, Jouste, Nalukwago and Waiswa (2021) finds that a new e-filing system dou-
bled the number of presumptive taxpayers and increased revenues, especially when coupled
with the parallel Taxpayer Register Expansion Project that provided a one-stop shop where
tax officials offered assistance and further lower taxpayers’ compliance costs. On a similar
note, in Belarus, e-filing was part of a broader reform to lower compliance costs for citizens,
including simplifying the tax code, setting up taxpayer facilitation services, and reaching out
to the business community (World Bank, 2016).
   Non-traditional technology solutions besides ITAS-based e-filing and e-payment systems
may also be very effective in facilitating compliance and reducing opportunities for cor-
ruption in low-resource settings. During door-to-door property tax campaigns in Kananga,
                                               an et al., 2022), tax agents used handheld
Democratic Republic of Congo (Weigel, 2020; Bal´
printers to issue receipts to taxpayers who made cash payments to them instead of traveling
to the tax office. Each transaction was recorded in the device’s memory and downloaded to
the government database on a regular basis. Program supervisors confirmed that the revenue
deposited were equal to the amounts on the receipts that were issued, ensuring agents were
accountable for the cash payments received.
   The potential of e-filing in reducing compliance costs depends on the policy design itself,
as shown by Yilmaz and Coolidge (2013) who critically compare the different policy experi-
ences of South Africa, Nepal and Ukraine. Important aspects in the design of IT solutions
could have unintended consequences on taxpayers’ compliance costs, especially when solu-
tions are not customer-centric, reliable and trustworthy. E-filing is impactful in those cases,
like South Africa, in which it completely replaces paper-based filing with no additional work
required from firms. In Nepal, where e-filing was mandatory and particularly cumbersome,


                                              19
it actually increased compliance costs. In Ukraine, where businesses kept filing on paper in
parallel to e-filing due to lack of trust on the online system, compliance costs increased as
well. Similarly, Kenyan Revenue Authority (KRA) continued to maintain two concurrent
tax filing systems - manual and i-tax - even after the introduction of the latter. As these
two systems were not integrated with one another, they likely generated confusion among
taxpayers and the KRA detected discrepancies in taxpayers’ records (Maisiba and Atambo,
2016).
   Along the same lines, Kochanova, Hasnain and Larson (2020) suggest that the policy
implementation of complementary online tools such as e-payments plays a crucial role –
whereas a simpler e-filing system that does not enable online payment of taxes may not affect
compliance costs and interactions with tax officials. Likewise, the existence of an ITAS on
the tax administration side could considerably improve the e-filing experience – as taxpayers
would be able to monitor their tax obligations on their web-profile, provide feedback and
receive timely assistance. A highly policy-relevant message that can be derived from the
literature is that the success of a given technology does not only depend on the effectiveness
and design of that specific solution, but most importantly hinges on the comprehensive
implementation of a coherent set of IT measures which together address interlinked taxpayer
needs (such as file and pay online), communicate smoothly and, thus, combined have an
impact.


1.3.2     Electronic Fiscal Devices

Electronic fiscal devices, already discussed in section 1.1.1 for their potential in better defin-
ing tax liabilities, could also play a major role in terms of reducing compliance costs. The
evidence available is rather mixed. On the one hand, current research indicates that EFDs
are impactful in making the taxpayer experience less burdensome, more predictable and less
frustrating. Ethiopian firms highlighted the following benefits from EFD adoption: ‘less op-
portunity for theft’ (mentioned by 20 percent of respondents), ‘updated and easily available


                                               20
sales information’ (22 percent), ‘easy to comply with tax requirements’ (17 percent) and
‘better sales and inventory control’ (16 percent) (World Bank, 2015, p. 26).
   Similarly, evidence from a mixed methods approach - based on administrative data, focus
group discussions (FGDs) and a nationally representative survey - from Mascagni, Dom and
Santoro (2021) shows that medium and large taxpayers considerably value EBMs as a tool for
facilitating their compliance. A more sophisticated version, EBM2, which is software-based
and works on computers, is also quite appreciated by this group as it improves record-
keeping, monitoring of transactions on a daily basis and business reputation, thus attracting
new IT-savvy costumers.
   While EFDs may improve the tax paying process for many taxpayers, there is also the risk
that disadvantaged taxpayers may be left behind or may face even higher compliance costs.
Section 3.1 discusses this consideration. These technologies should therefore be coupled with
sustained taxpayer education and assistance and provide timely communication and feedback
mechanisms for taxpayers.


1.3.3   Online Communication

Linked to its role in facilitating compliance, technology also holds a great potential in terms
of communicating information around the tax system and easing the burden with complying.
By accessing e-portals, for instance, taxpayers could seek for tax-related assistance online or
browse through tax laws and similar official documents. In the large literature on tax nudges
featuring communication experiments that manipulate the message content and test differ-
ent motivations behind compliance, most projects use traditional delivery methods, such as
letters. However, increasing attention is being devoted to more modern and cheaper solu-
tions, such as SMSs and e-mails (Mascagni, 2018). More research is needed in this direction,
given the huge potential that IT-based communication retains in potentially reshaping the
way in which revenue authorities assist, educate, train their taxpayers (World Bank, 2016;
OECD, 2021).


                                              21
    An illustrative example comes from Rwanda, where in early 2016 the Rwanda Revenue
Authority sent messages to 9,000 taxpayers, aimed to encourage compliance. Mascagni
and Nell (2022) use a randomized experiment to compare the efficacy of different delivery
methods, from more traditional (physical letters) to more digitized (SMS, email). They
show that non-traditional delivery methods, specifically SMS, are highly effective and signif-
icantly increase tax due of nudged taxpayers, in contrast to physical letters. This exercise is
particularly relevant from a policy perspective, as governments in developing countries are
constantly looking for cost-effective ways to achieve policy goals, in the context of very lim-
ited capacity. In this setting, it would be difficult to scale up the delivery of physical letters
beyond a relatively small group considering the often poor quality of taxpayer addresses in
tax registries (Ortega and Scartascini, 2015; Mayega et al., 2019).



2     Limitations of Technology

While the evidence discussed above highlights the unprecedented potential of technology in
improving tax administration and revenue mobilization, it is also true that technology alone
is unlikely to be effective in a context in which pre-existing, complementary factors are not
yet in place (Bird and Zolt, 2008). This is also true in high income countries as failure to get
these preconditions right explains the poor take-up of e-government services there as well
(Gauld, Goldfinch and Horsburgh, 2010; Reddick and Turner, 2012). The relevance of these
preconditions, and the evidence produced on them, is summarized below.


2.1    Hard Infrastructure and Connectivity

Even the most user-friendly technology will hardly function if basic infrastructure such as
electricity and, for those technologies based on the Internet, stable connectivity are lacking.
Immediate implications concern the take-up and usage – neither taxpayers/users nor tax
officials can smoothly take-up and use a technology that halts in peak periods or functions


                                               22
only intermittently.
   In addition, inadequate infrastructure and connectivity could also pose important equity
repercussions, as those taxpayers most likely to be disconnected from the internet are likely
to be smaller, rural, and less-educated. In turn, their broad taxpaying experiences could be
frustrating, and their tax attitudes and perceptions could deteriorate. Particular attention
should be devoted to these unintended consequences, especially in light of the increasing
shift to e-filing and e-payment as a response to the COVID-19 pandemic - as in certain cases
such as Eswatini Santoro, Amine and Magongo (2022) and Cameroon, revenue authorities
imposed the adoption of e-tools to avoid physical interactions. However, it is unclear whether
making adoption compulsory would prove to be effective for compliance if large portions of
the taxpayer population are simply not covered by connectivity or stable electricity.
   Some descriptive evidence describes the challenges taxpayers face in adopting technology
in less IT-friendly contexts. In Zimbabwe, Obert et al. (2018) study the underutilization
of the e-filing tax system and conclude that connectivity is indeed a major challenge to
most taxpayers, with around 33 percent of respondents reporting absolutely no access to any
form of network connection. For instance, one challenge unearthed by this study was that
the server used to operate the e-filing system frequently faltered during peak periods, with
negative repercussions on taxpayers’ filing experience. Likewise, in Eilu (2018) review of six
studies on EFDs adoption in Kenya and Tanzania, poor internet connection is cited in three
of the studies, while power outages come just next.
   To address these challenges, tax authorities can proactively roll out less sophisticated
technologies with the clear policy intent of reaching less connected taxpayers and facilitating
their compliance. This is the case of Rwanda, where the revenue authority launched the M-
declaration tool as an offline declaration and payment solution for rural taxpayers, working
on simple feature phones. M-declaration has been introduced as a simpler complement to
the more IT-heavy E-tax platform, which instead works on computers and through the
internet. Solutions like M-declaration could easily address connectivity issues and enable


                                              23
digital payments through the widespread mobile money system. The survey study of Santoro
et al. (forthcoming) shows how user profiles of M-declaration and E-Tax are different by
policy design and how the two technologies get adopted by those who need them most. The
Malawi Revenue Authority also introduced a similar M-mobile application. In Tajikistan,
electronic kiosks were built and equipped in remote locations around the country for use by
taxpayers who wish to e-file but do not have access to computers (World Bank, 2018). As
these technologies are extremely new, no robust impact evaluation has yet been conducted
and future research should examine whether they alleviate the compliance costs of smaller
taxpayers.


2.2     The Human Factor

Technology is a tool, which does not live by itself but is meant to be used by people. If
people refrain from, resist, or manipulate technology, it is unlikely to produce any positive
impact. This human factor is crucial to understanding the potential of technology more
realistically. In this context, users of IT can be divided into two broad categories: taxpayers
and tax officials.


2.2.1   Taxpayers

Besides the infrastructure limitations mentioned above, taxpayers may fail to adopt a tech-
nology for several reasons. First, taxpayers may have little understanding and practical
knowledge of sophisticated innovations. Lack of awareness was a major barrier to e-filing
adoption in different contexts such as Nigeria and Tajikistan (Obert et al., 2018; Eilu, 2018;
Efobi et al., 2019; Mas’ ud, 2019). In Rwanda, focus group discussions by Mascagni, Dom
and Santoro (2021) reveal that that taxpayer confusion, complexity and practical difficulties
with the EFDs are particularly severe for smaller taxpayers. Eilu (2018) reviews the evidence
from seven studies based on the adoption experiences of firms in Kenya and Tanzania, to
conclude that the most-cited challenge with the use of electronic fiscal devices (EFD) was


                                              24
lack of training.
   Second, users may fail to adopt a new technology if sunk costs for its implementation
are too high. This was the second most-cited challenge in the Eilu (2018) review. There
could be resistance in adoption from taxpayers within certain categories having to cover
significant costs for accessing the service (Yilmaz and Coolidge, 2013). Qualitative evidence
from Rwanda (Mascagni, Dom and Santoro, 2021) documents high levels of frustration, as
taxpayers have to buy the EFDs with their own funds and also have to cover all ancillary costs
on the equipment needed for using the machines (airtime, internet connection, computers,
printers, etc.). In this sense, digital technologies can reinforce socioeconomic disparities
already present in the taxpayer population.
   Limited evidence exists on the impacts of subsidies for technology uptake, with rather
inconclusive results. Despite considering a different technology (Point of Sale (POS) ter-
                                                                  aenz Somarriba (2022)
minals for electronic payments for transactions), Brockmeyer and S´
provides some evidence on the impact of incentives, which could inform the design of incen-
tive schemes for other technology, such as EFDs. It finds that only 2 percent of eligible firms
took up the subsidy, and of these, 80 percent already had a POS. This suggests that firms
consider other factors beyond costs, such as the compliance impact of the new technology,
when deciding whether to adopt technology.
   Third, different individual taxpayer characteristics are also associated with technology
adoption. Efobi et al. (2019) finds from a survey that internal firm characteristics like
owner’s education, age of the company and use of an external auditor and computerized
accounting system are most strongly correlated with the use of e-filing and e-payment. These
characteristics may be proxies of taxpayer knowledge and ability to bear adoption costs. They
may also reflect trust in tax administration, as taxpayers could be reluctant to adopt a new
technology due to mistrust towards the tax administration, especially for concerns around
data security and confidentiality issues.
   Besides these factors affecting adoption, more IT-sophisticated taxpayers could blatantly


                                              25
manipulate the technology in order to avoid paying taxes. A tax administration whose
internal systems are not integrated and automated is open to manipulation attacks from
taxpayers. In Malawi, among many other countries, the revenue authority makes use of sev-
eral disconnected systems running in parallel, making registration, tracking and monitoring
quite challenging. Being aware of these weaknesses, taxpayers seem to use multiple TINs for
the same business to cut tax liabilities, while not being automatically detected (Ligomeka,
2019). This strategic behavior is observed in Uganda as well, despite even though the tax
authority maintains a uniform registry (Mayega et al., 2019). This highlights the fact that
even a sound integrated system, which could in theory easily identify fraud, would prove
inefficient if not used to its full potential by the tax administration.


2.2.2   Tax Officials

Tax officials as well are adopters of technology and could sub-optimally use it due to a
number of constraints. First, tax officials could resist change and prefer keeping up the
usual, often manual-based, practices. Tax officials can have a disincentive to reorganize their
work to take full advantage of digital technologies, often due to the possibility of losing the
discretionary benefits they have been accumulating with a manual, discretionary system.
However, very little causal evidence has been produced on how technology could shape the
interactions between taxpayers and tax collectors – with the exception of Okunogbe and
Pouliquen (2022) - as well as on incentive schemes for motivating tax officials in taking-up
a new technology.4
   Regarding the latter, Chalendard et al. (2020) argue that the effectiveness of technology-
enabled third-party valuation advice in Madagascar is undermined by the limited incentives
to tax officials to optimally make use of this new information. The authors find that even
though better information on import values from a third party significantly enhanced perfor-
mance of the customs department by enabling a more exact determination of tax liabilities,
  4
    Outside technology, notable exceptions are Khan, Khwaja and Olken (2016) and Amodio et al. (2021),
who study the impact of performance pay for tax collectors on revenue and bribe payments.


                                                 26
its impact on revenue appeared compromised by misaligned incentives of custom officials.
Thus, the absence of strict sanctions against opportunistic behavior and the prevalence of
rampant corrupt practices in Madagascar customs, as in customs of other low- and middle-
income countries, could explain why the provision of information to inspectors have not
yielded stronger impacts.
   Mayega et al. (2019) provides some interesting qualitative evidence on staffs’ resistance
to technology in Uganda. It documents the negligence of under-incentivized registration
staff, who do not respond to TIN duplicates warning triggered by the system, hence clogging
up the registry with duplicates. In Malawi as well, preferences for manual practices imply
that a specific, separate registry exists for large taxpayers, stored and managed on separate
Excel files, not integrated on a uniform platform and controlled by the LTU staff (Ligomeka,
2019). Anecdotal evidence suggests that these discretionary practices are commonplace in
other countries as well.
   A second major impediment for tax officials is a lack of adequate skills and training.
Many revenue authorities in LMICs routinely assign their staff to intense training periods,
often incentivized, as well as make completion of certain learning modules compulsory for
promotion. Notwithstanding, it seems that more could be done in terms of building more
specific, tailored skills on how to use technology and, correspondingly, how to manage tax
data. Observational evidence indicates that revenue authorities in low-income countries often
do not cross-check their tax records, even if it would be relatively straightforward to do so
(Kangave et al., 2016; Almunia et al., 2021; Mascagni, Mukama and Santoro, 2019). This,
for instance, is documented in Rwanda, where easily identifiable discrepancies across VAT
returns are not closed likely due to limited know-how from tax officials. Robust evidence is
lacking on this aspect but most existing studies point to the lack of IT skills as one of the
concurring causes for limited usage of different technologies, such as systematic cross-checks
        a, 2017).
(Pirttil¨




                                             27
2.3    Institutional Strategy and Change Management

In some cases, technology adoption by a tax authority is not successful because different
components are not properly sequenced. In other cases, the new IT systems are not properly
integrated into the organizational structure. While there is less evidence from causally
identified studies on these internal processes of tax administration, this section describes key
features of successful roll out of technology and provides illustrative examples from different
tax authorities.
   The adoption and implementation of technology should not happen in a single step,
motivated by a political frenzy or competitive imitation of peers, but rather within a clearly
defined, long-term transformational journey. As a first consideration, the sequencing of
technology adoption, as delineated in a strategy of structural reform, is crucial. Digitization
of functions and processes is usually pioneered by customs departments, as they tend to use
internationally standard information systems. After piloting a new automated and integrated
platform in customs (such as ASYCUDA+++ and ASYCUDA World), revenue authorities
turn to digitizing their internal functions by automating all internal processes. In this phase,
it is preferred that clear sequential actions are taken. For instance, a new integrated and
automated tax administration system is unlikely to succeed if the existing data within the
tax administration are not properly cleaned first, so as to be adequately analyzed by the
new system.
   Likewise, retraining and reengineering tax officials should come before introducing the
technology, as well as repeatedly over time. Also, there seems to be a clear sequencing in
how the different functions get automated. After customs, usually e-registration modules
are introduced next, to improve the quality of taxpayer registries. Then, the most relevant
taxes are digitalized first, with VAT and income taxes being the priority. All the remaining
taxes and fees usually get digitized at a later stage.
   Once all internal functions are integrated successfully, the next step entails providing
e-services to taxpayers. E-filing and e-payment systems could develop naturally and directly


                                              28
feed into a well-running automated tax administration system. Often, e-services are rolled
out sequentially, first imposed as a pilot on large taxpayers, and only later to the taxpayer
population at scale. More and more sophisticated modules could be added as the opportunity
arises. For instance, the M-declaration phone application introduced in Rwanda was meant
to represent an additional service building on the already existing e-filing system.
   As a second related point, technology provides the opportunity for an unparalleled cul-
tural change within the tax administration, where obsolete analog practices are abandoned
and a new mindset, based on efficiency and transparency, is embraced. Descriptive evidence
from different contexts shows that it is difficult to change the silos mentality with which
different departments within a revenue authority tend to operate. It is commonplace to see
different IT systems operating in parallel to manage taxpayer data. This simultaneous use
of distinct systems creates a host of structural challenges for the tax administration. One
major challenge is the duplication of work. Given that different units manage the systems,
another challenge pertains to the authorization required by officials in a unit to access the
data from a different unit within the same institution. As a result, officers often concentrate
solely on recording information on taxes under their mandate and ignore any other tax-
relevant data (Ligomeka, 2019).
   For example, in Ethiopia, as many as four institutions collect data on employment in-
come (the revenue authority, the Ministry of Finance and two institutions for social secu-
rity provision, each for the public and private sectors). Despite having common challenges
and aligned interests, such institutions do not communicate for both political and techni-
cal reasons (Mascagni, Santoro and Mukama, 2019). More broadly, customs and domestic
tax departments still appear not to communicate systematically. To properly address this
challenge when introducing new technology, it is acutely important to have a rigorous pro-
curement process that carefully gathers system/ business requirements, understands user
needs, and reviews business processes to ensure that the procured technology systems are
fit-for-purpose and avoid needless duplication.


                                              29
   At a higher level, the implementation of technology requires strong buy-in from key lead-
ers and a long-term strategy at the national level. The Rwandan case is very much emblem-
atic of how technological transformation has been enabled by the strong political support
from the very top government through the Vision 2020 development plan (Schreiber, 2018).
A number of concurring strategic actions have been taken in order to unleash the full poten-
tial of technology in Rwanda. First, government and donors massively invested in high-speed
fiber-optic network, serving the whole country. This directly enabled the RRA to make use
of more advanced online technologies and to eventually launch its automated data manage-
ment system. Second, a target-oriented and performance-based staff management culture
has been crucial to build higher professionalism within the tax authority – often coupled with
intense trainings on IT tools. Third, the government eliminated import taxes on comput-
ers, cell phones, and telecommunications equipment, facilitating digital inclusion. Lastly, to
facilitate compliance of micro-entrepreneurs through digital means, the Finance Ministry’s
policy committee moved quickly to create a simple, new presumptive tax for them, easily
payable through mobile phones (Schreiber, 2018). More research is needed to understand
how the political vision of a country could spur successful technological transformation.


2.4    Regulatory Framework

A broader key consideration is the regulatory and legal framework in which technology exists.
As technological innovations develop at a faster and faster pace and tax administrations
keep updating themselves, the same should be true for the legal system. However, the
latter is typically much slower to respond and adapt to change. On the one hand, revenue
authorities could strengthen themselves through the use of technology. On the other hand,
the ultimate ability to enforce, prosecute, impose sanctions and collect taxes often lies outside
the authority itself. Hence, countries need a functional, modern legal system enabling the
technology-enhanced tax administration to fully benefit from this transformation.
   There are many ways in which the legal system could hamper the potential of technology


                                               30
in tax administration. For instance, technology could empower revenue authorities in their
enforcement capacity, helping them with assessing the underlining true liability, tracking
defaulters, and identifying blatant evaders. All these efforts are however likely to be un-
fruitful if actual prosecution does not materialize due to long delays in courts. Likewise,
technology could spur mass registration and formalization, which however would not deliver
better compliance outcomes if the tax code is silent about tax obligations of new entrants
(Mascagni, Mengistu and Woldeyes, 2021). Furthermore, confusion could arise about the
legal requirements to use a given technology, such as e-filing. In a situation where e-filing
is made compulsory by decision of the revenue authority but where the national laws are
still ambiguous on whether the adoption of a digital practice is mandatory or voluntary, or
for whom, navigating the regulations could lead to higher compliance costs of taxpayers,
alongside greater mistrust and frustration.
   On top of that, gaps in regulation could arise when it comes to co-using a given technology
across different public institutions. As mentioned above, data sharing between revenue
authorities and a range of public and private actors does not happen systematically, often
due to privacy and confidentiality concerns. One possible approach is for policy makers
in a country to work towards setting up a central automated platform which would be
accessible from a variety of government institutions and banks for identification of taxpayers
and cross-checking of information. Robust evidence on the effectiveness of such mechanisms
is needed to determine how effective this approach may be. Along the same lines, in order
to boost tax compliance, higher income countries have enacted specific legislation whereby
processors of debit or credit cards and electronic payment systems are mandated to report
to the tax authorities the gross receipts of the enterprises accepting these forms of payments.
Legislation in lower income countries is still mostly undefined.
   Lastly, a crucial consideration is that as technology evolves, adjustments in the regu-
latory framework are needed in order to protect citizens from unintended consequences in
terms of privacy, confidentiality, data leakages and cybersecurity. Accessing third-party data


                                              31
from, say, digital financial services providers comes with risks of breaching confidentiality
agreement with the clients of those services. Likewise, the institution of national ID systems
or other electronic systems at scale in contexts which are not fully democratic poses new
risk in terms of the government ability to surveil and restrict citizens’ rights. In these cases,
a conducive regulatory framework needs to be in place, allowing keeping a balance between
data protection and security and the use of new data and technology for tax purposes. In
sum, technology needs to be thought of as a unique catalyst for broader regulatory reforms
within public institutions and an opportunity to modernize tax codes, eventually promoting
the country to a higher development trajectory.



3     Conclusion

This paper describes the potential that information technology provides for tax mobilization,
as well as the challenges that still exist in low- and middle-income countries in order to tap
into it. First, technology could be a powerful tool for understanding the tax base: tax
authorities can use technology-based tools to collect information to identify taxable entities
(such as individuals or property during registration drives) as well as to collect information
on the tax liability, for example, through the use of electronic billing machines that record
sales transactions, or from third-party sources like employers, vendors, customers or financial
institutions.
    Second, technology also contributes to strengthening the capacity for tax monitoring.
It provides data analytic tools such as machine learning for collecting and analyzing large
amounts of data to detect inconsistencies, such as mismatches between sales reported for
income tax to those recorded for VAT, or between VAT sales and purchases reported by sellers
and buyers. By using different electronic sources of information on taxpayers’ behavior, tax
authorities can systematically assign risk of tax evasion to different firms to prioritize audits
to maximize revenue recovery.



                                               32
   Third, technology is also valuable for improving service delivery to taxpayers. Technology
solutions like electronic filing and payment can reduce the time and other compliance costs
borne by taxpayers. Electronic billing machines reduce the costs of compiling and submitting
information. Electronic communication tools (such as email and SMS) provide a timely and
cost-effective way of keeping taxpayers informed of their tax requirements and account status.
These technologies also reduce the level of in-person interactions between taxpayers and tax
officials, thus reducing opportunities for extortion and collusion.
   A common theme to these uses of technology is that they move the tax system towards
becoming more standardized and delivering a consistent experience across taxpayers and
away from the use of tax official discretion in the treatment of taxpayers. At the same
time, several country examples highlight a range of challenges associated with the adoption
of technology in taxation. First, given the infrastructure deficit in many low- and middle-
income countries, access to internet and electricity may hinder the effective adoption of
technology by taxpayers, especially the disadvantaged, raising equity concerns.
   Second, no matter how sophisticated an IT system is, it is only as effective as its users
can employ it. On the one hand, taxpayers may be hesitant to use technology tools due to
lack of information and trust, security concerns or high costs of adoption. Taxpayers could
also deliberately respond to technology in perverse ways, by adjusting multiple margins in
their tax returns and thus muting the impacts on revenue mobilization. Tax authorities
have responded by conducting regular trainings for taxpayers and in some cases subsidiz-
ing the costs of technology adoption or rolling out less sophisticated alternatives to serve
disadvantaged taxpayers.
   On the other hand, tax officials may not maximize the use of IT tools due to lack of skills
and training or may actively resist to avoid losing the private benefits they received from
the influence they had under a manual system. Tax authorities must ensure that their IT
agenda focuses on empowering staff members with the skills and training they need, while
also being as manipulation-proof as possible. Related to that, tax administrations should


                                              33
be equipped with sufficient resources to conduct traditional strategies in enforcement and
monitoring, to which IT-enabled audit solutions could contribute.
   Third, technology reform must not be a stand-alone project but should be part of a
broad institutional strategy, where complementary reforms are designed. It requires a clear
sequencing of reforms to avoid redundancies and disorder. For example, it is important to
coordinate human resource needs (recruitment and training) with hardware and infrastruc-
ture needs (e.g. internet, electricity, computers) alongside investment in different software
and complementary e-solutions. Different tax modules should be sequenced so they build
on one another. Importantly, adequate investments in traditional tools for monitoring and
enforcement should be planned for as well.
   Fourth, the success of technology reforms in taxation is dependent on the broader reg-
ulatory environment. In many countries, the legal authority to enforce, prosecute, impose
sanctions to recover delinquent taxes does not lie within the tax authority but with the
judiciary, which may be slow and inefficient. Also, the regulatory framework needs to be
updated on data sharing between institutions and private actors, and provisions to protect
citizens’ rights. As such, technology-based reforms in tax administration must be backed
by a functional, modern legal system to realize the full revenue gains and ensure a fair and
equitable tax system.
   The recent upsurge in the literature on taxation in low and middle-income countries is, in
large part, due to the availability of administrative tax data in electronic format. As countries
continue on their technology transformation journey, there are two key opportunities for
collaborating with researchers. First, tax authorities can work with researchers in real-time
to understand the impact of different technology interventions they adopt to guide expansion
or modifications to the technology. Second, tax authorities can collaborate with researchers
more broadly to understand the implications of different tax policy choices for efficiency
and equity through the analysis of tax data in electronic format. Beyond providing rigorous
evidence on government policies to advance both research and policy, these partnerships


                                               34
may also have beneficial spillover effects. Academic researchers typically have state of the
art data analytic tools that may not only be used on the specific collaborative project but can
also translate into upgrading the capacity of officials in the tax agency. Pomeranz and Vila-
Belda (2019) provide examples of successful collaborations and tips for mutually beneficial
engagements.
   Avenues for future research include better understanding the role of tax officials in the ef-
fective deployment of technology, the effectiveness of “low-tech” technologies such as mobile
taxation, and further applications of machine learning and artificial intelligence in moni-
toring and facilitating compliance. In the long run, it will be interesting to observe how
technology goes beyond changing tax administration to change tax policy: with better tools
for identifying the tax base and monitoring compliance, low- and middle- income countries
may change the mix of tax instruments and rely more on payroll and personal income taxes
as in high-income countries.




                                              35
References
Aker, Jenny C, and Christopher Ksoll. 2019. “Call Me Educated: Evidence from a
 Mobile Phone Experiment in Niger.” Economics of Education Review, 72: 239–257.
Aker, Jenny C, and J. Cariolle. 2020. “The Use of Digital for Public Service Provision
 in Sub-Saharan Africa.” Ferdi Policy brief B209.
Aker, Jenny C, Ishita Ghosh, and Jenna Burrell. 2016. “The Promise (and Pitfalls)
 of ICT for Agriculture Initiatives.” Agricultural Economics, 47(S1): 35–48.
Aker, Jenny C., Paul Collier, and Pedro C. Vicente. 2017. “Is Information Power?
 Using Mobile Phones and Free Newspapers during an Election in Mozambique.” The
 Review of Economics and Statistics, 99(2): 185–200.
Aker, Jenny C, Rachid Boumnijel, Amanda McClelland, and Niall Tierney. 2016.
 “Payment Mechanisms and Antipoverty Programs: Evidence from a Mobile Money Cash
 Transfer Experiment in Niger.” Economic Development and Cultural Change, 65(1): 1–37.
Ali, Merima, Abdulaziz B Shifa, Abebe Shimeles, and Firew Woldeyes. 2021.
 “Building Fiscal Capacity: Evidence on the Role of Information Technology.” National
 Tax Journal, 74(3): 591–620.
Alm, James. 2021. “Tax Evasion, Technology, and Inequality.” Economics of Governance,
 1–23.
Almunia, Miguel, Jonas Hjort, Justine Knebelmann, and Lin Tian. 2021. “Strategic
 or Confused Firms? Evidence from “Missing” Transactions in Uganda.” National Bureau
 of Economic Research.
Amodio, Francesco, Jieun Choi, Giacomo De Giorgi, and Aminur Rahman. 2021.
 “Bribes vs. Taxes: Market Structure and Incentives.” Journal of Comparative Economics.
Ariel, Barak. 2012. “Deterrence and Moral Persuasion Effects on Corporate Tax Compli-
 ance: Findings from a Randomized Controlled Trial.” Criminology, 50(1): 27–69.
Bachas, Pierre, Anne Brockmeyer, Al´    ıpio Ferreira Cantisani, and Bassirou Sarr.
 2021. “When it comes to tax enforcement, data and technology aren’t always the answer.”
 https: // www .opml .co .uk/ blog/ when-it-comes-to-tax-enforcement-data-and-
 technology-aren-t-always-the-answer .
   an, Pablo, Augustin Bergeron, Gabriel Tourek, and Jonathan L. Weigel. 2022.
Bal´
 “Local Elites as State Capacity: How City Chiefs Use Local Information to Increase
 Tax Compliance in the Democratic Republic of the Congo.” American Economic Review,
 112(3): 762–97.
Banerjee, Abhijit, Esther Duflo, Clement Imbert, Santhosh Mathew, and Rohini
 Pande. 2020. “E-governance, Accountability, and Leakage in Public Programs: Exper-
 imental Evidence from a Financial Management Reform in India.” American Economic
 Journal: Applied Economics, 12(4): 39–72.

                                          36
Bellon, Matthieu, Jillie Chang, Era Dabla-Norris, Salma Khalid, Frederico Lima,
 Enrique Rojas, and Pilar Villena. 2019. Digitalization to Improve Tax Compliance:
 Evidence from VAT E-Invoicing in Peru. International Monetary Fund.

Besley, Timothy, and Torsten Persson. 2009. “The Origins of State Capacity: Property
 Rights, Taxation, and Politics.” American Economic Review, 99(4): 1218–44.

Bird, Richard M, and Eric M Zolt. 2008. “Technology and Taxation in Developing
  Countries: From Hand to Mouse.” National Tax Journal, 61(4): 791–821.

Brockmeyer, Anne, and Magaly S´    aenz Somarriba. 2022. “Electronic Payment Tech-
 nology and Tax Compliance: Evidence from Uruguay’s Financial Inclusion Reform.” World
 Bank Policy Research Working Paper WPS 9947, World Bank Group, Washington, D.C.

Brockmeyer, Anne, and Marco Hernandez. 2016. “Taxation, Information, and With-
 holding: Evidence from Costa Rica.” World Bank Policy Research Working Paper WPS
 7600, World Bank Group, Washington, D.C.

Brockmeyer, Anne, Spencer Smith, Marco Hernandez, and Stewart Kettle. 2019.
 “Casting a Wider Tax Net: Experimental Evidence from Costa Rica.” American Economic
 Journal: Economic Policy, 11(3): 55–87.

Callen, Michael, and James D Long. 2015. “Institutional Corruption and Election
 Fraud: Evidence from a Field Experiment in Afghanistan.” American Economic Review,
 105(1): 354–81.

Callen, Michael, Clark C Gibson, Danielle F Jung, and James D Long. 2016. “Im-
 proving Electoral Integrity with Information and Communications Technology.” Journal
 of Experimental Political Science, 3(1): 4–17.

Camacho, Adriana, and Emily Conover. 2010. “The Impact of Receiving Price and
 Climate Information in the Agricultural Sector.” Working Paper, Centro de Estudios sobre
 Desarrollo Economico Universidad de los Andes, Bogota, Colombia.

Carrillo, Paul, Dina Pomeranz, and Monica Singhal. 2017. “Dodging the Taxman:
 Firm Misreporting and Limits to Tax Enforcement.” American Economic Journal: Applied
 Economics, 9(2): 144–64.

Casaburi, Lorenzo, Michael Kremer, Sendhil Mullainathan, and Ravindra Ram-
 rattan. 2014. “Harnessing ICT to Increase Agricultural Production: Evidence from
 Kenya.” Harvard University.

Chalendard, Cyril, Alice Duhaut, Ana M Fernandes, Aaditya Mattoo, Gael Ra-
 balland, and Bob Rijkers. 2020. “Does Better Information Curb Customs Fraud?”
 World Bank Policy Research Working Paper 9254, World Bank Group.

Courtois, Pierre, and Julie Subervie. 2015. “Farmer Bargaining Power and Market
 Information Services.” American Journal of Agricultural Economics, 97(3): 953–977.


                                           37
Das, Satadru, Lucie Gadenne, Tushar Nandi, and Ross Warwick. 2022. “Does
 going cashless make you tax-rich? Evidence from India’s demonetization experiment.”
 The Institute for Fiscal Studies, Working Paper W22/03.

Dogan, Ugur. 2011. “Tools for Risk Management: The Case of Turkey.” In Risk-Based
 Tax Audits: Approaches and Country Experiences. , ed. Sultan Khwaja Munawer, Awasthi
 Rajul and Jan Loeprick. Directions in Development Series. Washington, DC:The World
 Bank.

Dzansi, James, Anders Jensen, David Lagakos, and Henry Telli. 2022. “Technology
 and Local State Capacity: Evidence from Ghana.” National Bureau of Economic Research,
 Working Paper 29923.

Efobi, Uchenna, Ibukun Beecroft, Tanankem Belmondo, and Amelia Katan. 2019.
  “Small Business Use of the Integrated Tax Administration System in Nigeria.” African Tax
  Administration Paper 8, International Centre for Tax and Development. Brighton: ICTD.

Eilu, Emmanuel. 2018. “Adoption of Electronic Fiscal Devices (EFDs) for Value-Added
  Tax (VAT) Collection in Kenya and Tanzania: a Systematic Review.” The African Journal
  of Information and Communication, 22: 111–134.

Escueta, Maya, Vincent Quan, Andre Joshua Nickow, and Philip Oreopoulos.
  2017. “Education Technology: An Evidence-based Review.” NBER Working Paper 23744,
  National Bureau of Economic Research, Cambridge, MA.

Fafchamps, Marcel, and Bart Minten. 2012. “Impact of SMS-based Agricultural Infor-
  mation on Indian Farmers.” The World Bank Economic Review, 26(3): 383–414.

Fan, Haichao, Yu Liu, Nancy Qian, and Jaya Wen. 2018. “Computerizing VAT
  Invoices in China.” NBER Working Paper 24414, National Bureau of Economic Research.

Fish, Paul. 2018. “Practical Guidance Note: Training Manual for Implementing Property
  Tax Reform with a Points-Based Valuation.” African Tax Administration Paper 2, Inter-
  national Centre for Tax and Development. Brighton: ICTD.

Fujiwara, Thomas. 2015. “Voting Technology, Political Responsiveness, and Infant Health:
  Evidence from Brazil.” Econometrica, 83(2): 423–464.

Gauld, Robin, Shaun Goldfinch, and Simon Horsburgh. 2010. “Do They Want it?
 Do They Use it? The ‘Demand-Side’ of E-Government in Australia and New Zealand.”
 Government Information Quarterly, 27(2): 177–186.

Grieco, Kevin, Amou Bakarr Kamara, Niccol´         o F Meriggi, Julian Michel, Wilson
 Prichard, and Graeme Stewart-Wilson. 2019. “Simplifying Property Tax Adminis-
 tration in Africa: Piloting a Points-Based Valuation in Freetown, Sierra Leone.” Summary
 Brief 19, International Centre for Tax and Development. Brighton: ICTD.




                                           38
Hildebrandt, N, Y Nyarko, G Romagnoli, and E Soldani. 2014. “Information is
 Power? Impact of an SMS-based Market Information System on Farmers in Ghana. Ac-
 cessed July 2016.”
Jouste, Maria, Milly I Nalukwago, and Ronald Waiswa. 2021. “Do Tax Adminis-
  trative Interventions Targeted at Small Businesses Improve Tax Compliance and Revenue
  Collection? Evidence from Ugandan Administrative Tax Data.” UNU-WIDER Working
  Paper 2021/17.
Kangave, Jalia, Suzan Nakato, Ronald Waiswa, and Patrick Zzimbe. 2016. “Boost-
 ing Revenue Collection through Taxing High Net Worth Individuals: The Case of Uganda.”
 Working Paper 45, International Centre for Tax and Development.
Kettle, Stewart, Marco Hernandez, Simon Ruda, and Michael Sanders. 2016. “Be-
 havioral Interventions in Tax Compliance : Evidence from Guatemala.” Policy Research
 Working Paper WPS 7690, World Bank Group, Washington, D.C.
Khan, Adnan Q, Asim I Khwaja, and Benjamin A Olken. 2016. “Tax Farming
 Redux: Experimental Evidence on Performance Pay for Tax Collectors.” The Quarterly
 Journal of Economics, 131(1): 219–271.
Khwaja, Munawer Sultan, Rajul Awasthi, and Jan Loeprick. 2011. Risk-based Tax
 Audits: Approaches and Country Experiences. World Bank Publications.
Kleven, Henrik Jacobsen, Martin B Knudsen, Claus Thustrup Kreiner, Søren
 Pedersen, and Emmanuel Saez. 2011. “Unwilling or Unable to Cheat? Evidence from
 a Tax Audit Experiment in Denmark.” Econometrica, 79(3): 651–692.
Knebelmann, Justine. 2022. “Digitalisation of property taxation in developing countries.”
 ODI Report. London: ODI (www.odi.org/publications/digitalisation-of-property-taxation-
 in-developing-countries-recent-advances-and-remainingchallenges).
Kochanova, Anna, Zahid Hasnain, and Bradley Larson. 2020. “Does E-Government
 Improve Government Capacity? Evidence from Tax Compliance Costs, Tax Revenue, and
 Public Procurement Competitiveness.” The World Bank Economic Review, 34(1): 101–
 120.
Lewis-Faupel, Sean, Yusuf Neggers, Benjamin A Olken, and Rohini Pande. 2016.
  “Can Electronic Procurement Improve Infrastructure Provision? Evidence from Public
  Works in India and Indonesia.” American Economic Journal: Economic Policy, 8(3): 258–
  83.
Ligomeka, Waziona. 2019. “Assessing the Performance of African Tax Administrations:
  A Malawian Puzzle.” African Tax Administration Paper 14, Brighton: ICTD.
Loeprick, Jan, and Michael Engelschalk. 2011. “Simplified risk scoring for SMEs.”
  In Risk-Based Tax Audits: Approaches and Country Experiences. , ed. Sultan Khwaja
  Munawer, Awasthi Rajul and Jan Loeprick, 45–54. Directions in Development Series.
  Washington, DC:World Bank Publications.

                                           39
 opez-Luzuriaga, Andrea, and Carlos Scartascini. 2019. “Compliance Spillovers
L´
  across Taxes: The Role of Penalties and Detection.” Journal of Economic Behavior &
  Organization, 164: 518–534.

Maisiba, Gekonge Justus, and Wallace Atambo. 2016. “Effects of Electronic-tax sys-
 tem on the Revenue Collection Efficiency of Kenya Revenue Authority: A Case of Uasin
 Gishu County.” Imperial Journal of Interdisciplinary Research (IJIR), 2(4): 815–827.

Martinovic, Ivan, Lucas Kello, and Ivo Sluganovic. 2017. “Blockchains for Govern-
 mental Services: Design Principles, Applications, and Case Studies.” Working Paper No.7,
 Centre for Technology and Global Affairs.

Mascagni, Giulia. 2018. “From the Lab to the Field: A Review of Tax Experiments.”
 Journal of Economic Surveys, 32(2): 273–301.

Mascagni, Giulia, and Andualem Mengistu. 2019. “Effective tax rates and firm size in
 Ethiopia.” Development Policy Review, 37: O248–O273.

Mascagni, Giulia, and Christopher Nell. 2022. “Tax Compliance in Rwanda: Evidence
 from a Message Field Experiment.” Economic Development and Cultural Change.

Mascagni, Giulia, Andualem T Mengistu, and Firew B Woldeyes. 2021. “Can ICTs
 Increase Tax Compliance? Evidence on Taxpayer Responses to Technological Innovation
 in Ethiopia.” Journal of Economic Behavior & Organization, 189: 172–193.

Mascagni, Giulia, Denis Mukama, and Fabrizio Santoro. 2019. “An Analysis of Dis-
 crepancies in Taxpayers’ VAT Declarations in Rwanda.” Working Paper 92, International
 Centre for Tax and Development. Brighton: ICTD.

Mascagni, Giulia, Fabrizio Santoro, and Denis Mukama. 2019. “Teach to Comply?
 Evidence from a Taxpayer Education Programme in Rwanda.” Working Paper 91, Inter-
 national Centre for Tax and Development. Brighton: ICTD.

Mascagni, Giulia, Roel Dom, and Fabrizio Santoro. 2021. “The VAT in Practice:
 Equity, Enforcement and Complexity.” Working Paper 117, International Centre for Tax
 and Development. Brighton: ICTD.

Mas’ ud, Abdulsalam. 2019. “Acceptability of e-Filing of Taxes by Micro-Entrepreneurs in
 Northwestern Nigeria.” Working Paper 96, International Centre for Tax and Development.
 Brighton: ICTD.

Mayega, Jova, Robert Ssuuna, Muhammad Mubajje, Milly I Nalukwago, and
 Lawrence Muwonge. 2019. “How Clean is our Taxpayer Register? Data Management in
 the Uganda Revenue Authority.” African Tax Administration Paper 12, Brighton: ICTD.

Mitra, Sandip, Dilip Mookherjee, Maximo Torero, Sujata Visaria, et al. 2013.
 “Asymmetric Information and Middleman Margins: An Experiment with West Bengal
 Potato Farmers.” Working Paper 401, The Bureau for Research and Economic Analysis
 of Development (BREAD).

                                           40
Mittal, Shekhar, and Aprajit Mahajan. 2017. “VAT in Emerging Economies: Does
 Third Party Verification Matter?” Working Papaer, Anderson School of Management,
 Los Angeles.

Moore, Mick. 2020. “What is Wrong with African Tax Administration?” ICTD Working
 Paper Series 111, International Centre for Tax and Development. Brighton: ICTD.

Muralidharan, Karthik, Paul Niehaus, and Sandip Sukhtankar. 2016. “Building
 State Capacity: Evidence from Biometric Smartcards in India.” American Economic Re-
 view, 106(10): 2895–2929.

Nakasone, Eduardo, Maximo Torero, and Bart Minten. 2014. “The Power of Infor-
 mation: The ICT Revolution in Agricultural Development.” Annual Review of Resource
 Economics, 6(1): 533–550.

Nglazi, Mweete D, Linda-Gail Bekker, Robin Wood, Gregory D Hussey, and
 Charles S Wiysonge. 2013. “Mobile Phone Text Messaging for Promoting Adherence to
 Anti-Tuberculosis Treatment: A Systematic Review.” BMC Infectious Diseases, 13(1): 1–
 16.

Obert, Sifile, Kotsai Rodgers, Mabvure Joseph Tendai, and Chavunduka Des-
 derio. 2018. “Effect of E-Tax Filing on Tax Compliance: A Case of Clients in Harare,
 Zimbabwe.” African Journal of Business Management, 12(11): 338–342.

OECD. 2021. “Building Tax Culture, Compliance and Citizenship:    A Global
 Source Book on Taxpayer Education, Second Edition.” OECD Publishing, Paris,
 https://doi.org/10.1787/18585eb1-en.

Okunogbe, Oyebola. 2013. “Optimising Real Estate Tax in Liberia: Implications for
 Revenue Performance and Economic Growth.” Working Paper, The International Growth
 Centre (IGC).

Okunogbe, Oyebola. 2021. “Becoming Legible to the State: The Role of Detection and
 Enforcement Capacity on Tax Compliance.” World Bank Policy Research Working Paper.

Okunogbe, Oyebola, and Victor Pouliquen. 2022. “Technology, Taxation, and Cor-
 ruption: Evidence from the Introduction of Electronic Tax Filing.” American Economic
 Journal: Economic Policy.

Ortega, Daniel, and Carlos Scartascini. 2015. “Don’t Blame the Messenger: A Field
 Experiment on Delivery Methods for Increasing Tax Compliance.” Mimeo, Inter-American
 Development Bank.

       a, Jukka. 2017. “Building up Tax Systems: Lessons from the Nordic Countries.” In
Pirttil¨
  Conference on Public Economics for Development. Embassy of Finland and UNU-WIDER.

Pomeranz, Dina, and Jos´    e Vila-Belda. 2019. “Taking State Capacity Research to the
 Field: Insights from Collaborations with Tax Authorities.” Annual Review of Economics,
 11: 755–781.

                                          41
Reddick, Christopher G, and Michael Turner. 2012. “Channel Choice and Public
 Service Delivery in Canada: Comparing E-Government to Traditional Service Delivery.”
 Government Information Quarterly, 29(1): 1–11.

Santoro, Fabrizio, Adrienne Lees, M. Carreras, and T Mukamana. forthcoming.
  “E-Tax Services in Rwanda: Uptake and Impacts.” Working Paper Series. International
  Centre for Tax and Development. Brighton: ICTD.

Santoro, Fabrizio, and Winnie Mdluli. 2019. “Nil-Filing in Eswatini: Should the Rev-
  enue Administration be Concerned?” African Tax Administration Paper 6, International
  Centre for Tax and Development. Brighton: ICTD.

Santoro, Fabrizio, Razan Amine, and Tanele Magongo. 2022. “Mandating Digital
  Tax Tools as a Response to Covid: Evidence from Eswatini.” Working Paper Series.
  International Centre for Tax and Development. Brighton: ICTD.

Schreiber, L. 2018. “A Foundation for Reconstruction: Building the Rwanda Revenue
  Authority, 2001-2017.” Innovations for Successful Societies, Princeton University, 29.

Shah, Syed Jawad Ali. 2020. “Using Computerized Information to Enforce VAT: Evidence
  from Pakistan.” Working Paper.

Slemrod, Joel, Brett Collins, Jeffrey L Hoopes, Daniel Reck, and Michael Se-
  bastiani. 2017. “Does Credit Card Information Reporting Improve Small Business Tax
  Compliance?” Journal of Public Economics, 149: 1–19.

Steenbergen, Victor. 2017. “Reaping the Benefits of Electronic Billing Machines.” Work-
  ing Paper, International Growth Centre.

Stewart-Wilson, Graeme. 2020. “Strengthening Data Management In Tax Administra-
  tion: The Case Of The Sierra Leone National Revenue Authority.” International Centre
  for Tax and Development (ICTD).

UNU-WIDER Government Revenue Dataset. 2021. “Version 2021.” https: //
 doi .org/ 10 .35188/ UNU-WIDER/ GRD-2021 .

Weigel, Jonathan L. 2020. “The participation dividend of taxation: How citizens in Congo
 engage more with the state when it tries to tax them.” The Quarterly Journal of Eco-
 nomics, 135(4): 1849–1903.

World Bank. 2015. “Tax Compliance Cost Burden and Tax Perceptions Survey in
 Ethiopia.” World Bank Group.

World Bank. 2016. World Development Report 2016: Digital Dividends. World Bank
 Group.

World Bank. 2018. “Improved Tax Administration Can Increase Private Investment
 and Boost Economic Development in Tajikistan.” Feature Story March 12, 2018. Avail-
 able at https: // www .worldbank .org/ en/ news/ feature/ 2018/ 03/ 12/ improved-

                                          42
  tax-administration-can-increase-private-investment-and-boost-economic-
  development-in-tajikistan Last accessed March 12, 2022.

World Bank Enterprise Surveys.                2013.   “Enterprise   Survey.”   http: //
 www .enterprisesurveys .org .

Yilmaz, Fatih, and Jacqueline Coolidge. 2013. “Can E-filing Reduce Tax Compliance
 Costs in Developing Countries?” World Bank Policy Research Working Paper 6647, World
 Bank Group.




                                         43