Policy Research Working Paper                                10655




  How Can Lower-Income Countries Collect
              More Taxes?
       The Role of Technology, Tax Agents, and Politics

                             Oyebola Okunogbe
                              Gabriel Tourek




Development Economics                           A verified reproducibility package for this paper is
Development Research Group                      available at http://reproducibility.worldbank.org,

December 2023                                   click here for direct access.
Policy Research Working Paper 10655


  Abstract
 This paper examines recent evidence on tax administration                          officials shape their performance. The paper discusses the
 interventions aimed at increasing tax revenues in lower-in-                        promise and pitfalls of interventions in these two areas and
 come countries. It focuses on two major themes: the use of                         the strategic interactions between them. Lastly, it empha-
 new information technology tools—for identifying taxable                           sizes the importance of political incentives and considers
 entities, verifying tax liabilities, and ensuring collection of                    the conditions under which governments choose to invest
 liabilities—and how the deployment and incentives of tax                           in tax capacity and expand tax collection.




 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. A verified reproducibility package for this paper is available at http://
 reproducibility.worldbank.org, click here for direct access.




         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
How Can Lower-Income Countries Collect More Taxes? The Role of
               Technology, Tax Agents, and Politics1


                    Oyebola Okunogbe (World Bank Development Research Group)
                                   Gabriel Tourek (University of Pittsburg)




JEL: H2, H83, O1, O14
Keywords: tax revenue, tax administration, information technology, tax officials, political
economy of taxation




1
 This paper is forthcoming in the Journal of Economic Perspectives. We are grateful to the editors, Erik Hurst, Nina
Pavcnik, and Timothy Taylor, for their helpful feedback that improved this piece. We thank Archita Misra and
Kairat Umargaliyev for remarkable research assistance. We are grateful to Pierre Bachas, Anne Brockmeyer,
Matthew Collin, Anders Jensen, Wilson Prichard, and Mahvish Shaukat, for feedback and helpful conversations.
The findings, interpretations, and conclusions expressed in this paper, including any errors, are entirely ours and do
not necessarily represent the views of the World Bank Group, its Executive Directors or the countries they represent.
Introduction
Taxes are the main source of government revenue in most countries and provide funding for public
investments in human capital, infrastructure, and social insurance. Increasing tax revenues is thus
a major policy goal in many low- and lower-middle-income countries that collect a low share of
their GDP as tax revenues. In 2019, for example, the average tax-to-GDP ratio was 12 percent for
low-income countries, 18 percent for lower-middle-income countries, 21 percent for upper-
middle-income countries, and 30 percent for high-income countries (UNU-WIDER Government
Revenue Dataset 2022). Tax-to-GDP ratios among lower-income countries, in fact, resemble
those of modern high-income countries at similar stages of development: among 18 developed
countries with available historical data, tax-to-GDP ratios in 1919 averaged 12 percent (Besley
and Persson 2019; Mitchell 2007). To account for why developing countries tax so little, Besley
and Persson (2014) argued that low taxation is an outgrowth of deeper economic and institutional
factors constraining development, concluding that “the most important challenge is taking steps
that encourage development, rather than special measures focused exclusively on improving the
tax system.”

This view might suggest that tax capacity should expand with economic development. However,
over the last 30 years, the relationship between GDP growth and taxation levels among low- and
lower-middle-income countries has remained essentially flat. Figure 1 plots the relationship
between percentage changes from 1990-2000 to 2010-2019 in GDP per capita and tax revenues as
a share of GDP for such countries. This pattern suggests economic growth does not automatically
generate increases in tax revenue, a point Besley and Persson (2014) also note. Rather,
governments must invest in improving the tax system to take advantage of expansions in the tax
base that result from increased economic activity.

Improving a tax system requires investments in the ability of the tax administration to carry out its
three core tasks: identifying taxpayers, determining how much they owe, and bringing these
liabilities into state coffers (Cotton and Dark 2017). Following the framework in Okunogbe (2021),
we will refer to these three dimensions of tax capacity as identification, detection and collection
capacity. Identification capacity is the ability to uniquely identify individuals (as well as assets or
entities) that are liable to taxation through comprehensive registries and databases. Detection
capacity entails verifying the amounts of tax liability, often utilizing data from third parties.
Collection capacity is the ability to ensure the payment of liabilities. It involves the use of billing
                                                                                                     2
and payment systems to facilitate tax transactions, as well as the imposition of penalties in the face
of non-compliance. While policy variables such as tax instruments and tax rates are undoubtedly
crucial to the performance of a tax system, this paper focuses on inputs to tax administration. To a
large extent, lower-income countries have adopted tax instruments and rates that resemble those
in higher-income countries. That low tax revenues persist among the former underscores the value
of examining gaps in administrative capacity.

Figure 1: GDP per capita and Tax/GDP, 1990 to 2019




Data source: UNU-WIDER Government Revenue Dataset (2022) and UNU-WIDER World Income Inequality
Database (2022).
Notes: This figure displays the relationship between changes in tax revenues as a share of GDP (y-axis) and GDP per
capita (x-axis) for low-income and lower-middle-income countries by country income group. Each measure is
calculated as the difference between values calculated separately for the periods 1990 to 2000 and 2010 to 2019,
divided by the value for the base period (1990 to 2000). We truncate the second period in 2019 to omit the period of
the COVID-19 pandemic. We pool annual observations by country within each period as measures for individual
years are frequently missing due to intermittent collection of these measures. Values within each period are calculated
as the mean of annual observations for each nation within that period. The tax revenues to GDP measure corresponds
to tax revenues as a share of GDP within a given year, excluding grants and social security. The income categories
are drawn from the World Bank country classifications by income level, as applied in the UNU-WIDER datasets.
Low-income countries are displayed as green circles, and lower-middle income countries are displayed as blue
triangles. The dashed blue line of best fit is estimated using all displayed observations and has slope 0.04 and robust
standard error 0.16.



                                                                                                                     3
Over the last decade, with the growing availability of high-quality administrative tax data and a
rise in partnerships between researchers and governments, there has been a tremendous increase
in the amount of evidence on the impact of different interventions to improve tax capacity and
increase tax revenues in low- and middle-income countries. This paper examines two major themes
from that body of evidence: first, the deployment of new information technology tools to facilitate
identification, detection and collection, and second, the role of tax officials, both in harnessing the
potential of technology as well as in complementing it. Despite the availability of viable
interventions, an important question is under which conditions governments will choose to invest
in tax capacity and expand tax collection. In turn, we will discuss how the current level of taxation,
available technology tools, and extent of political competition may affect how the government
chooses the nature and level of taxation.


The Role of Technology
Electronic systems that collect, generate, process, and store vast amounts of data can support the
implementation of a wide set of tax policy instruments, such as the value-added-tax, income tax
withholding, customs, and property tax. In lower-income countries, tax modernization programs
often seek to support the tax authority in a transition away from relying on manual processes or
outdated technology and in-person interactions with taxpayers towards automated, online
processes. Of the taxation-related projects approved in 116 countries by the World Bank between
2010 and 2022, 91 countries (78 percent) had a project that included a tax modernization or
information technology component, according to our own analysis of World Bank projects.
Examples of these projects are integrated tax administration systems for domestic taxes and
customs, electronic registration of taxpayers, electronic tax filing and payment, and risk-based
audit selection. 2
A well-integrated information technology system enables the tax authority to uniquely identify
taxpayers, process information received from taxpayers and third parties to verify liabilities and
compliance, and facilitate the remittance of funds to the treasury. We discuss below the use of
technology for these three dimensions of tax capacity.
Identification Capacity: Taxpayer Identification and Registration
Identifying taxpayers and developing a taxpayer registry are first-order challenges in many low-
and middle-income countries, where proper identification systems often do not exist for
individuals nor businesses, properties, and other taxable entities or tax bases. In many countries,
people who do not have existing national identification are issued a “Tax ID number” when they
enter the tax system. Having a Tax ID that is not linked to an individual’s national identity may
serve as a stopgap measure for tax authorities with the most rudimentary systems but may also
offer scope for tax avoidance. For example, without a link to personal identity, there is a danger
that owners of firms that have accumulated significant tax liabilities could simply shut them down
and open new ones to avoid tax penalties. Anecdotally, such maneuvers likely account in part for

2
 Okunogbe and Santoro (2023a) provide a more comprehensive review of the use of technology in tax
administration. Okunogbe and Santoro (2023b) provide applications for African countries.
                                                                                                     4
the high firm death rates observed in low- and middle-income countries (McKenzie and Paffhausen
2019). The lack of identity systems in many low- and middle-income countries is mirrored in low
coverage of personal income taxes, which account on average for 26 percent of tax revenues in
high-income countries but only 14 percent in low-income countries (UNU-WIDER Government
Revenue Dataset 2022).
Recent technology advances have significantly reduced the costs of identifying individuals at scale
and linking identity information across different government functions and registries. For example,
countries can enroll millions of people in national systems with biometric identification (like
fingerprints or retinal scans) across a nation within a few months. 3 Such identification can then
serve purposes not only related to taxation but also to other government functions like verifying
eligibility for public benefits and establishing a financial credit history. In the case of real estate
taxation, GIS-enabled systems can capture property location and features at scale (Knebelmann
2022), facilitating the administration of real estate taxes as well as the provision of public services
like postal service, electricity, and sewerage.
In an interesting case, Ghana started identifying taxpayers using the “Ghana Card” national
identification system in 2021 in place of tax identification numbers issued by the tax authority.
Ghanaian tax officials reported that, as a result, they could identify 85 percent of Ghanaians
compared to 4 percent under the tax-ID-based system, and the number of filers increased from 4
million to 6.6 million within a few months (GhanaWeb 2022, Ghana News Agency 2021). While
this episode does not provide causal evidence, it certainly suggests that an integrated identification
system across key sectors of the economy can broaden the reach of the tax authority and may
transform a nation’s revenue potential.
However, information technologies are not silver bullets. Too often, tax authorities make large
investments in taxpayer identification and registration but fail to observe commensurate increases
in tax receipts. Means to act on such advances are pivotal. An illustrative case in this vein comes
from a randomized experiment in Liberia (Okunogbe 2021). In 2014, the Liberian Revenue
Authority undertook a block-by-block digital mapping exercise of a pilot area in the capital,
Monrovia, to construct an electronic database of properties and the identities of owners. This
database then served as the basis for evaluating distinct tax enforcement measures. Simply
informing taxpayers that the tax authority had identified them as the owners of the properties in
question (using a notice that included their name and a photo of the property) did not shift the
likelihood that tax liabilities were paid (relative to a generic notice). However, combining this
identification signal with information about penalties for delinquency more than quadrupled the
payment rate. A second experiment sent a signal that the stated penalties would be enforced on
defaulters, which achieved a further increase in compliance. This example highlights that tax
authorities must complement technology-assisted identification and registration with enforcement
actions to translate those investments into tax receipts.




3
 The Kenyan government reportedly registered 36 million citizens for its Huduma Namba digital national ID in the
space of two months (Kimani 2019).
                                                                                                               5
Detection Capacity: Third-Party Information and Verification Processes
Information technology enables a tax authority to verify tax liabilities and detect where evasion
might be occurring. In particular, technology facilitates the collection and sharing of third-party
information, which is central to modern tax administration. Third-party reporting refers to any
information provided to the tax authority about liabilities, other than the declarations taxpayers
submit directly. The classic example of third-party information in high-income countries is the
information firms provide to tax authorities on employee salaries, an innovation as yet of limited
relevance in low- and middle-income countries where over 80 percent of the labor force is
employed in the informal economy (Elgin et al. 2021). Even among those employed in the formal
sector, firms in low- and middle-income countries often report aggregate employee tax liability
without linking amounts to individual employees, who in many cases do not have tax ID numbers
(Mayega et al 2021, Mascagni and Mengistu 2019).
More evidence on how technology allows collection of third-party information in low- and middle-
income countries comes from consumption taxes. The value-added tax is increasingly widespread
and is one of the most important sources of tax revenue, accounting for on average 28 percent of
taxes collected globally (UNU-WIDER Government Revenue Dataset 2022). As Almunia et al.
(forthcoming) discuss, a value-added tax has the important inbuilt feature of collecting third-party
information on transactions as firms buy and sell across supply chains. Many countries are moving
to digitize this paper trail with the use of electronic fiscal devices, machines that automatically
record transactions as they occur and transmit this information to the tax authority through internet
or mobile networks. These tools have led to the unveiling of significant revenue previously
undisclosed to the tax authority: in Ethiopia, yielding a 48 percent increase in tax revenue
(Mascagni, Mengistu and Woldeyes 2021). The promise of digital trails does not seem limited to
low-income countries, where one might expect more “low hanging fruit” in revenue gains:
electronic invoices that made it more difficult to falsify invoices also increased value-added tax
revenues in China (Fan et al 2020), a middle-income country.
Technology can also facilitate digital trails on firms’ revenues when they receive sales payments
via credit card or other electronic payment systems. Even when these transaction records are not
automatically transmitted to the tax authorities, the existence of the digital trail may be sufficient
to induce a compliance response, as evidenced by the tax impacts resulting from India’s efforts to
shift away from paper currency (Das et al., 2023) and the introduction of incentives for credit and
debit transactions in Uruguay (Brockmeyer and Saenz Somarriba, 2022). For cross-country
evidence, Apeti and Edoh (2023) show that the adoption of mobile money increases total tax
collections in developing countries. Third-party information can also come from a wide range of
sources, such as utility companies, customs, financial records, and procurement. Sophisticated tax
administrations are increasingly linking different databases, facilitated by common and unique
identifiers for individuals and firms across sources, to get a more comprehensive picture of true
tax liability. Alerting taxpayers to the existence of these data can in turn reduce evasion (see, e.g.,
Brockmeyer et al. 2019 for experimental evidence from Costa Rica).
While an increasing number of low- and middle-income countries are taking advantage of
information systems to collect third-party information, two important steps are needed to translate
information into revenue. First, it is important for countries to use this information systematically
                                                                                                     6
in tax enforcement. For example, despite how widespread the value-added tax is, many low- and
middle-income countries do not automatically or consistently crosscheck invoices and receipts
reported by buyers and sellers for the same transactions, which can have sizeable discrepancies.
In Uganda, for example, sellers and buyers were found to report different amounts in 79 percent
of trading pair-month observations (Almunia at al., 2022). An example of the value of putting third
party information to use comes from Pakistan. After many years of collecting third party
information on firms paying value-added tax but the information to limited use, Pakistan tax
authority achieved a 50 percent reduction in value added tax credits from adopting an automated
risk analysis system that rejected suspicious claims in real time (Shah 2023).

Second, it is important to obtain information on a given tax base from multiple sources. A
consistent pattern arises in which the tax authority attempts to increase enforcement using third-
party information obtained on one margin, but taxpayers then adjust on another margin to undo
impacts on their overall liability. For example, when tax authorities are given third-party
information on revenues, taxpayers may adjust on their expenses or input deductions: Ecuadoran
firms notified that they owed more in taxes responded by reporting higher costs (Carrillo,
Pomeranz, and Singhal, 2017). Ethiopian firms responded the same way when newly mandated
electronic fiscal devices increased firms reported revenues (Mascagni, Mengistu and Woldeyes,
2021), as did Brazilian firms during an anti-tax evasion program in São Paulo, cutting the
initiative’s revenue gains in half (Naritomi, 2019). Obtaining information on each input to a tax
determination will help forestall such adjustments.

Finally, even as countries close potential loopholes within their domestic tax systems, another
potential source of revenue loss lies in international tax evasion and avoidance, such as individuals
concealing their income or wealth offshore, outside the reach of tax authorities, or multinationals
legally shifting their profits to tax havens. These tax losses can be substantial. For example, an
amnesty program on wealth taxation in Argentina revealed hidden assets equivalent to 21 percent
of GDP (Londoño-Vélez and Tortarolo 2022). Profit-shifting can be a major source of revenue
loss in lower-income countries. Data from the missingprofits.world website (based on Wier and
Zucman 2022) suggest that 26 percent of corporate taxes in Nigeria are lost due to profit shifting,
13 percent from South Africa, and 6 percent from India.
Advances in computer-based data sharing across countries, such as Tax Information Exchange
Agreement and Automatic Exchange of Information – Common Reporting Standards provide tools
to uncover these attempts at avoidance and evasion as countries can obtain information on the
overseas accounts of their citizens. Similarly, in regulating profit shifting, having access to data
from other jurisdictions enables tax authorities to compare the price at which a firm wants to
transfer an asset among its subsidiaries to a set of comparable transactions to determine whether it
complies with arms-length principles--that is, such transactions should happen under the same
conditions (including price) as it would between two separate firms. Low- and middle-income
countries face challenges gaining access to these data as signing up to these exchanges requires
countries to collect the same information in their own countries, for which they may lack capacity.
Even after granted access, low- and middle-income countries then are confronted with the crucial
step of using these third-party data effectively, through data analytics that match them to tax
records and other in-country data sources to detect evasion and avoidance.
                                                                                                   7
Collection Capacity: Electronic Tax Transactions
Another core application of information technology is the processing of tax information and
payments. For taxpayers, electronic filing and payment options reduce tax compliance costs. For
tax administrators, they reduce the risk of data entry errors and free up personnel who would
otherwise be needed for processing paper returns. The generation of tax information and tax
withholding at its source by a third-party agent – e.g., employers and financial institutions – is
another fiscal innovation that is facilitated by the availability of appropriate data systems. These
innovations appear important for tax performance globally, with evidence ranging from individual
reforms – e.g., withholding of sales tax by credit card companies in Costa Rica (Brockmeyer and
Hernandez, 2022) – to cross-country analysis comparing 100 countries at various levels of
development (Jensen, 2022). Electronic processing of tax information and payments can also
provide digitized administrative data to the tax authority that may then support the automation of
many tax processes, such as advanced analytics for risk-based audits and targeted enforcement
efforts. 4
When a tax authority more often carries out its duties through e-transactions, and less through in-
person interactions, there is also less scope for unsanctioned behaviors like extortion and collusion.
At the same time, shifts to digitization may weaken tax officials’ ability to monitor or gain
contextual knowledge about taxpayers. We return to this tradeoff later in the paper when we
discuss interactions between tax officials and technology tools.
One important concern in the use of electronic systems to facilitate compliance is the potential
impacts on equity (Bachas et al. (forthcoming) discuss equity considerations in taxation in lower-
and middle-income countries in detail). On the one hand, electronic systems may help lower-
income and less-privileged taxpayers. E-filing, e-payment, and withholding can reduce
compliance costs, often a larger proportion of income for such taxpayers. Similarly, automated
systems that log payments can facilitate installment plans to accommodate liquidity-constrained
taxpayers (for evidence from Mexico, see Brockmeyer et al., 2023). On the other hand, evidence
across different contexts suggests that taxpayers who are female, rural, less educated, or those
heading less-established companies are less likely to use e-services and, where compulsory, may
face higher costs of adoption, with evidence from small businesses in Nigeria (Efobi et al., 2019),
general use of e-tax services in Rwanda (Santoro et al, 2023), and value-added taxes in Kenya,
Tanzania (Eilu, 2018), and Rwanda (Mascagni et al., 2023). Electronic systems, coupled with other
features of the tax system can also lead to unintended consequences that disadvantage lower-
income taxpayers. For example, if taxes are over-withheld (as is common) and tax filing is optional
(due to, e.g., exemptions for taxpayers earning below a minimum threshold), non-filers will face a
higher effective average tax rate, with this impact concentrated among lower-income taxpayers
(see Hauck and Wallosek (2023) for evidence from Germany).


4
  In addition to supporting tax compliance monitoring, data from tax systems also supports revenue forecasting and
financial reporting. Beyond tax administration applications, information generated for tax purposes can also generate
indirect benefits for the design of economic policies more generally. For example, data on firm-to-firm transaction
networks can inform about the nature of trading structures, their resilience to financial and climate shocks, and
opportunities to diversify supply chains. Likewise, measures of corporate growth – captured through income taxes –
can assist in the targeting of industrial policy.
                                                                                                                   8
Cross Country Evidence on Use of Technology for Identification, Detection and Collection
To complement these causally identified studies at the country level, we zoom out to a cross-
country description of the relationship between the extent of technology use and a country’s tax
performance using data from the OECD Inventory of Tax Technology Initiative’s 2023 Global
Survey on Digitalisation. The survey catalogs the tax technologies used in 75 countries, capturing
variables can be mapped to our framework to understand how countries use technology to boost
their identification, detection and collection capacity. 5 For identification-related technologies, the
survey collects whether the tax authority requires taxpayers to have digital identification with a
unique identity number, and whether this ID is based on government-issued documents and/or
biometric information, as well as whether taxpayers can register online. For detection, it collects
whether the tax authority receives data from third parties such as trade partners, shares databases
with other government bodies, requires taxpayers to submit e-invoices or to maintain online cash
registers that directly report their sales to the tax authority, or uses artificial intelligence to conduct
risk assessments or detect evasion. For collection capacity, the survey records whether taxpayers
can file and/or pay taxes online, and whether they can request an extension or set up a payment
arrangement online.


Figure 2: Taxation Technology and Tax-to-GDP




5
 For more information, see: https://www.oecd.org/tax/forum-on-tax-administration/tax-technology-tools-and-
digital-solutions/.
                                                                                                             9
Data Sources: Tax revenues as a share of GDP measure drawn from the International Survey on Revenue
Administration (CIAT, IMF, IOTA, OECD 2022) using values for 2018. Taxation technology measure drawn from
OECD Inventory of Tax Technology Initiatives (2023) Global Survey on Digitalisation.
Notes: This figure displays the relationship between tax revenues as a share of GDP and an index of taxation
technology for 75 individual countries. The tax revenues as a share of GDP measure is drawn from the International
Survey on Revenue Administration (CIAT, IMF, IOTA, OECD 2022) using values for 2018. We fill in missing values
for 7 out of 75 countries using UNU-WIDER (2022) data in 2018. The index of taxation technology is computed from
individual measures drawn from the OECD Inventory of Tax Technology Initiatives (ITTI) (2023) Global Survey on
Digitalisation. The index includes variables capturing tax authorities’ use of third-party information, taxpayer
identification systems, taxpayer services, and innovation in technology use. More details on these measures are
provided in the text and Appendix I. Classifications of country income level are drawn from World Bank Development
Indicators. Low-income countries are represented by green triangles, lower-middle-income by blue diamonds, upper-
middle by gray squares, and high-income by black circles. The line of best fit is estimated using all observations and
has slope 2.32 (robust standard error 1.10).



For each dimension of tax capacity, we construct an index that counts how many of the possible
technologies a country uses, as well as an overall index that sums across the three groups. 6 The
horizontal axis of Figure 2 shows the standardized scores for each country (in standard deviation
units relative to mean zero). Figure 2 reveals a positive correlation between the extent of
technology use (captured by the overall index) and tax revenue as a share of GDP, consistent with
the case-specific evidence we discuss above. We examine these relationships using exploratory
regressions reported in Appendix Table 1 and find that this correlation persists while holding
constant country income level. Further, the relationship appears driven by technology applications
for identification and detection in contrast to the collection index, which does not have a detectable
relationship with tax performance. Of course, these relationships are simply correlations and
should by no means be interpreted as causal as many factors can affect both the use of technology
as well as tax performance. Nevertheless, the patterns in cross-country data are consistent with
findings from well-identified studies at the level of individual countries.




The Role of Tax Officials
Technology supplies data and tools that, ultimately, will assist tax personnel in the work of tax
collection. What this work entails depends on the underlying tax system. In more advanced tax
systems, tax personnel are deployed mostly in the higher-level tasks of maintaining and updating
information technology systems, processing returns, analyzing data, and undertaking audits. In
less advanced systems, particularly in the context of local taxation in many lower-income
countries, tax collection proceeds manually, with officials soliciting payments and undertaking
enforcement efforts in-person.
As a starting point, we examine how the relative strength of a nation’s tax staff relates to its fiscal
capacity. Figure 3 illustrates a negative relationship between tax-to-GDP and population-to-tax-
staff, using data from the International Survey on Revenue Administration (CIAT, IMF, IOTA,

6
    Appendix I provides further details on the construction of these variables and summary statistics.
                                                                                                                   10
OECD 2022). For ease of illustration, the sample excludes small island nations – which cluster
primarily in the low population-to-staff, low tax/GDP quadrant, and where the personnel to tax
capacity relationship may fundamentally differ due to reduced scale and their prominence as tax
havens – but the observed relationship is robust to including them.


Figure 3: Population per Tax Staff and Tax-to-GDP




Data Sources: Both measures are drawn from the International Survey on Revenue Administration (CIAT, IMF, IOTA,
OECD 2022) using values for 2018.
Notes: This figure displays the relationship between tax revenues as a share of GDP and population per tax staff.
Classifications of country income level are drawn from World Bank Development Indicators (YEAR). Low-income
countries are represented by green triangles, lower-middle by blue diamonds, upper-middle by gray squares, and high-
income by black circles. The line of best fit is displayed in blue, is estimated using all observations, and has slope -
0.002 (robust standard error 0.0002) which translates to an average decrease in the tax-to-GDP ratio of 2 for every
additional 1,000 tax personnel.



The ratio of population-to-tax-staff is an order of magnitude smaller in high-income countries. For
example, the United Kingdom reported 68,722 full-time-equivalent staff in 2018, or 1,133 citizens
per staff, and Sweden 10,486 in total tax staff, or 1,064 citizens per staff. By contrast, middle-
income countries like Pakistan (10,176 population per staff) and Ghana (10,006 per staff) – and
low-income nations – e.g., DRC (12,811 population per staff) and Togo (12,582 per staff) – possess
almost systematically higher ratios. Specialization of tax staff varies with the nature of the
economy and the tax system. According to the CIAT, IMF, IOTA, OECD (2022) database, while
the shares of staff assigned to registration, service, and payment activities (at roughly 30 percent)
                                                                                                                     11
and debt collection (roughly 11 percent) is relatively constant across income groups, high-income
countries dedicate a substantially higher share of personnel to audit and investigations (28 percent)
than in low-income countries (16 percent).
Deployment of Tax Officials
The effectiveness of tax authorities depends in part on how personnel are deployed in the work of
identification, detection, and collection. While strengthening technological capacity may reduce
the need for manual functions that can be digitized, the low citizen-to-tax-staff ratios observed in
richer countries suggest that personnel remain integral to collection capacity even as economies
grow. The specialization of tax staff – and the requisite qualifications – may simply evolve as per
capita GDP rises: for example, from in-person registration and monitoring and manual checking
of returns to wrangling third-party data to detect evasion.
Another margin for adjustment lies in how personnel are deployed across tax bases, both across
the types of taxes a government collects and even the characteristics of taxpayers within bases. In
a 2002 policy experiment, Indonesia introduced a corporate tax administration reform that created
“medium taxpayer offices” to oversee enforcement of the top several hundred taxpayers in each
region. These offices tripled the staff-to-taxpayer ratios assigned to handle such firms. As a result,
tax revenue more than doubled, with evidence suggesting this increase derived from more
business-reported taxable income and wage bills being reported to the tax authority (Basri et al.
2021). The response implies a sizeable elasticity of collections with respect to the strength and
focus of personnel. 7
As the cross-country pattern in Figure 3 implies, increasing staff—especially in high-return tax
functions—could offer a powerful way to rachet up tax capacity. For example, tax authorities
might pursue endeavors with potential to generate large increases in revenue, like concentrating
monitoring efforts on large taxpayers or those in evasion-prone sectors. They might also seek to
strengthen enforcement in a broad-based way, by training staff to effectively leverage third-party
information. Tax authorities must then also weigh the benefits of such actions against the cost of
staff resources that would be expended, as well as costs associated with reallocating personnel
away from other functions.
Ultimately, strengthening tax capacity will occur iteratively, and must respond to evolving
conditions. Yet improved staff allocation can offer a feasible means of raising collections even
absent complementary reforms to tax regimes, administrative processes, or technologies. In an
experimental study in Peru, Kapon et al. (2022) show that “prioritized iterative enforcement” of
tax debts – which targets enforcement efforts by trading off expected collection and expected
capacity use – can yield collection gains relative to randomly-targeted enforcement efforts.
Evidence from the significantly lower capacity setting of the Democratic Republic of Congo
suggests revenue gains as large as 26 percent can be achieved through resource-neutral
reallocations of staff (Bergeron et al. 2022). 8 The nature of gains from assignments – in the

7
  This elasticity can be interpreted as a form the “enforcement elasticity of tax revenue” conceptualized by Slemrod
and Keen (2017) as an input to identifying optimal interventions by a tax administration.
8
  Even larger gains may derive from engaging non-traditional personnel in tax collection: in the same setting in the
Democratic Republic of Congo, engaging local chiefs in taxation generated a 44 percent increase in revenues,
                                                                                                                  12
Congolese case, resulting from high-ability types being more effective when paired with similarly
high-ability teammates, particularly when collecting from households with higher payment
likelihoods – will likely vary across contexts, but could be fruitfully identified through policy
experimentation.
In short, the available evidence is consistent with more staff being able to bring in more revenues,
while reallocation of tax officials towards higher return activities can yield non-trivial revenue
gains, potentially achieved at low cost, for resource-constrained authorities.
Incentives of Tax Officials
The incentives that tax officials encounter will shape their effectiveness, and by extension, a tax
authority’s collection capacity. Ample evidence from the public sector suggests incentives can
modulate what types of individuals accept roles within bureaucracies and how well they perform.
Higher wages or better career benefits offered for public sector positions can attract more able
applicants without compromising motivation (as Dal Bo et al. (2013) show in Mexico and Ashraf
et al. (2020) in Zambia).
In theory, when designing incentive structures, tax authorities should seek to trade off gains in tax
collections against the costs of monetary and non-monetary incentives for staff. In practice, tax
administrations operate within government bureaucracies that may constrain options through rigid
salary structures, limits on performance incentives, and political considerations. Within the scope
of feasible incentive structures, how much can we expect the incentives facing tax officials to
matter?
One factor consists in how elastic collections are to effort and skills – or other factors like
dishonesty – within a certain job. For tax officials engaged in customs, as one example, there might
seem less scope for pure effort to impact tax revenues (because of rote processing of total
shipments), but larger scope for skills like attentiveness (monitoring product misclassification and
underdeclaration to evade tariffs) and dishonesty (bribes in exchange for misclassification) to
influence trade tax collections. For service staff tasked with registering informal firms, sheer effort
in identifying, locating, and communicating the process of formalization to such businesses may
matter comparatively more.
In lower-income countries, the work of taxation still largely involves in-person interactions with
taxpayers. On average, 74 percent of firms in low-income countries say they are required to meet
with tax officials and on average 3.4 meetings are held per year, compared to 26 percent of firms
and 1.8 meetings in high-income countries (World Bank Enterprise Surveys). Tax collectors
receive payments, often in cash, from individuals and firms, audits require inspection visits to
premises rather than examining taxpayer financial accounts, and registration drives involving
physical outreach are periodically undertaken to extend the tax net over sizeable informal
economies. The interpersonal nature of this work – along with officials’ discretion over outcomes
– naturally increases opportunities for greater revenue but also corruption.



suggesting that governments with low staff capacity can realize revenue gains by involving agents outside the
official state administration (Balán et al., 2022).
                                                                                                                13
Consider an example of how these tensions emerge from the lower middle-income country of
Pakistan. In the largest province – Punjab – tax officials are directly responsible for the assessment
of property taxes. Tax staff manually construct the property register, including updating for new
construction or changes in property values, and wield considerable discretion in determining tax
liability through the application of valuation tables and granting of exemptions. At the same time,
the wages of these officials are not linked to performance and few audit mechanisms exist, creating
scope for low tax collections and corruption, achieved via coercion or collusive agreements
between tax officials and taxpayers. This context mirrors the nature of local tax administration in
many low- and middle-income countries.
On one hand, tying officials’ incentives more closely to revenue generation could raise collections;
on the other hand, raising the return to formal collections could also increase officials’ bargaining
power when extracting bribes (in exchange for undervaluation or misclassification) from property
owners. To assess how performance incentives affect tax collections, the Punjab government ran
an experiment in collaboration with researchers in 2011 providing bonuses to tax officials in some
jurisdictions – in the form of a share of revenue raised for increasing collections (Khan et al.
2016). 9 Revenue in jurisdictions with the new performance pay scheme grew by close to 50
percent. The gains derived from a small number of high-value properties being newly taxed at their
true value. At the same time, most properties in incentivized areas did not pay more, but reported
paying higher bribes. Were properties more homogenously low in value, it is unclear whether the
average effect of performance incentives of revenues would have been positive, though the average
value of bribes would still likely have increased. A subsequent scheme to incentivize revenue
generation through merit-based assignment of postings in Punjab – under which we might expect
collusive corruption to be less attractive as a substitute for the reward of obtaining a preferred
posting to a new area – resulted in even larger increases in revenue (Khan et al. 2019).
The example from Pakistan demonstrates that in settings where scope for collusive corruption is
high, the objective function of tax officials should be conceptualized as including their personal
gain from wages and formal incentives as well as those from bribes, subject to the costs of being
detected. The integration of technology into tax collection may impact this calculus in several
ways, as we consider next.


Interaction of Tax Officials and Technology

Thus far, we have discussed the role of technology and tax staff as independent forces. In reality,
the two may act as substitutes or complements within specific tax functions. On the one hand,
technology may almost wholly obviate the need for staff in certain functions (e.g., a shift to online
tax registration removes the need for manual registration staff) but may also increase the relative
return to deploying staff in other roles. For example, the adoption of automatic personal income

9
 The revenue-based incentive scheme rewarded revenue increases over predetermined benchmarks based on
collections in previous years. Bonuses were inversely scaled to the size of jurisdictions to reflect the greater
difficulty of increasing revenues in smaller jurisdictions. Additional incentive schemes adjusted performance
bonuses based on taxpayer satisfaction and assessment accuracy or evaluations of performance determined by a
committee of senior tax officials.
                                                                                                                   14
tax payment through withholding by employers observed in the second half of the 20th century
among high-income nations (Besley and Persson 2014) likely freed up resources dedicated to
marshaling tax collections through visits, in-person collections, and so on. This shift could account
to some extent for the greater shares of staff in audit and investigation roles in higher-income
countries relative to lower-income countries we note above.
On the other hand, technology can clearly offer productivity gains for staff by expediting formerly
cumbersome processes – like managing tax records or checking returns for errors – and facilitating
reallocation of effort towards higher-return activities. The introduction of a digital tax collection
technology within a Ghanaian municipality sheds light on this dynamic. In 2021, the municipal
tax authority provided a tablet-based technology containing a geospatial database of properties and
revenue management software that helped property tax collectors more easily locate properties in
the field. A randomly selected subset of local property tax collectors – whose task is to deliver tax
bills and collect payments manually – were provided with this new technology. These collectors
delivered almost one-third more bills and collected double the amount of revenues compared to
collectors persisting in the old, entirely manual system (Dzansi et al. 2022). The timeline of the
evolution of visits and payments suggests that the technology assisted collectors in learning more
quickly which households were willing and able to make payments and to target their in-person
collection efforts accordingly. Technology can thus have high returns in even low resource, low
compliance settings.
However, tax authorities face two major challenges. The first is how best to strike a balance
between using technology to improve on potential error, bias, or manipulation in the work of
human agents –– while retaining the value of rich contextual knowledge of an environment that
tax officials can acquire from repeated interactions with taxpayers. The tradeoffs involved likely
vary with the nature of the task, even within the same setting. As an example, in Senegal, property
assessment officials using a “discretionary” method produced real estate valuations that with more
regressive tax implications than those produced by a computer algorithm (Knebelmann et al.,
2023), while officials charged with identifying firms with high evasion for audit performed better
than a newly-introduced algorithm directed toward the same purpose (Bachas et al., 2023).
An example from Tajikistan, in which a set of small- and medium-sized businesses in were
experimentally encouraged to adopt e-filing (Okunogbe and Pouliquen 2022), highlights the
tensions around using technology to replace interactions between taxpayers and officials. Firms
that adopted the electronic filing system – replacing a process by which taxes were filed in-person
with tax officials – reduced their time spent on tax matters by 40 percent and also freed up the time
of tax officials previously tasked with receiving declarations. However, e-filing did not change the
average amount of taxes or bribes paid by firms, due to counterbalancing patterns among two types
of businesses. One type of firm, identified as being more likely to evade pre-adoption,
approximately doubled their amount of taxes paid. The study suggests that e-filing disrupted
previous collusion between these firms and tax officials that had reduced tax liabilities. For the
other type of firms, those deemed less likely to evade, e-filing reduced taxes paid, which is
consistent with e-filing leading to less direct monitoring by officials who could have had private



                                                                                                  15
information to enforce compliance with firms' true liability. 10 In addition, firms flagged as less
likely to evade taxes paid fewer bribes, presumably because e-filing reduced extortion
opportunities. This case thus provides a nuanced view of how new technology systems can lead to
divergent impacts depending on the nature of existing relationships and interactions between tax
officials and taxpayers.
A second major challenge is that tax officials often resist, manipulate, and in some cases sabotage
technology reforms that constrain or monitor their behavior. After all, digitization of information
permits greater scrutiny of staff activities, through automatic flagging of errors or data
manipulation, generation of reports, and investigation of specific activities (for example, verifying
that the outcomes of audits can be replicated with official records). However, in practice, these
systems are often difficult to implement successfully, similar to findings among civil servants in
other sectors (for example, Banerjee, Duflo and Glennerster (2008) document public sector nurses
evading monitoring efforts in India).
A revealing case comes from customs officers in Madagascar. An information technology system
was developed to assign customs inspectors randomly to customs declarations, at which import
shipments are reviewed and tariffs are assessed. However, in practice, inspectors were able to
contravene the official process to assign declarations to particular staff (Chalendard et al. 2023).
The manipulation of the assignment process suggests that import brokers bribed staff to be paired
with a preferred inspector, who would then facilitate evasion. Upon discovering this scheme,
customs management outsourced the assignment process to a third party that used its own software
to assign inspectors to declarations randomly. Even then, after a few months inspectors began
withholding a portion of the declarations shared with the third party in order to continue to
circumvent random assignment. This case drives home the strength of the incentives that tax
officials often possess to engage in corruption and the need for bureaucratic reforms to create
alternative incentives (e.g., by meaningfully raising the cost of being caught).


Political Incentives and Constraints on Tax Collection
Taxation is ultimately the coercion of resources from private citizens to the state, and, beyond the
tools and personnel of a tax administration, high level political support is arguably the most
important determinant of effective taxation. There is substantial evidence from low- and middle-
income countries of tax liabilities that are known to the tax authority that go uncollected, despite
the availability of relatively low-cost interventions that could be adopted. For example, for
property taxes – where the tax base is visible and immobile, and thus presumably easy-to-tax –
collection rates range between 5 and 16 percent in cities in Haiti (Krause, 2020), Liberia
(Okunogbe 2021), Senegal (Knebelmann 2023), Ghana (Dzansi et al 2022), DRC (Bergeron et al.,
2022), and Uganda (Manwaring and Regan 2023). This suggests that, in many cases, there is
insufficient political motivation to pursue the collection of these liabilities by enforcing tax laws
and imposing financial penalties and legal prosecution on non-compliant taxpayers, some of whom
may be economically and politically powerful. A full discussion of the politics of taxation,

10
  Some firms reported that this effect resulted from tax officials no longer being able to force them to pay more than
their true liability to meet officials’ revenue targets.
                                                                                                                   16
including bargains with elite taxpayers, reform design to manage resistance from the general
population, and politics within the tax administration is beyond the scope of this paper and is
extensively discussed in the political science literature (e.g. Martin 2023, Moore et al 2018,
Pritchard 2019, Brautigam et al 2008, among others). Instead, we focus the discussion below on
evidence relating to factors that determine political choices to invest in tax capacity or to deploy
existing capacity.
Broadly, we can consider politicians as weighing the expected benefits and costs of taxation for
re-election prospects. On the one hand, higher tax revenues may support re-election by funding
the provision of public services, infrastructure investments, social protection, and other public
programs that citizens value. On the other, taxing citizens may harm re-election by raising citizen
expectations and demands for accountability. Interestingly, this accountability effect can be
triggered by simply demanding taxes regardless of whether those taxes are paid. A study of a door-
to-door property tax campaign in Kananga, DRC, found higher levels of citizen engagement even
among those who were visited by tax authorities but did not comply (Weigel, 2020). More broadly,
there exists cross-country evidence that tax collection declines prior to competitive elections,
suggesting that governments act strategically to avoid negative backlash from citizens (Prichard,
2018).
We discuss below three key factors that may shape the cost-benefit calculus of politicians in
determining how much taxation to pursue: alternative revenue sources, political competition, and
available technologies for reducing the salience and cost of collection. Alternative revenue sources
may outright dampen the impetus to tax. Oil-rich states are a prominent example of a low-tax, low-
accountability setting, where oil revenues support a high level of public spending without the need
to tax citizens. For instance, oil rich gulf countries collect on average 3.6 percent of GDP as taxes
(UNU-WIDER Government Revenue Dataset 2022). At the subnational level, transfers from the
central government may dampen the motivation for local taxation. In Brazil, jurisdictions hit by a
negative shock to federal transfers resulting from an update in their population count increased
local tax collection by approximately 30 percent (Ferraz et al., 2023). Localities achieved these
expansions of tax collection by broadening the tax base through increased investment in tax
capacity, like improving property-tax registers and higher spending on tax agents.
The remarkable expansion of tax collection in Lagos State, Nigeria, offers an interesting historical
case indicating the role of political competition, alongside other factors, in generating political
support for taxation. Nigeria’s government has a strong reliance on oil revenues and
correspondingly low levels of non-resource tax collection, with only about 6 percent of GDP
collected in non-oil taxes. Following the return to democracy in 1999, Lagos state recorded a
dramatic growth in annual tax revenues (Figure 4), from about ₦30 billion in 1999 to over ₦150
billion in 2011 (measured in Nigerian naira in 2010 prices), 11 amounting to a fivefold increase in
a little over a decade, before the growth tapers off. This revenue growth was greater than in the
rest of the country.




 At 2010 exchange rate of 150 Naira to 1 USD, this is equivalent to growth from 200 million USD to 1 billion
11

USD.
                                                                                                               17
Analyses of this case have highlighted political, economic and personal factors (De Gramont 2015,
Gaspar et al 2016, Cheeseman and De Gramont 2017, Bodea and LeBas, 2016). The ruling party
in Lagos was an opposition party in stiff competition with the central government, and, in the
nascent democracy, was keen to show that it could provide a better model of government and
public services. Providing a steady source of domestic revenues was thus important not only for
meeting the severe needs of road infrastructure, waste management, and security in a congested
city, but also in securing funding to consolidate and expand political influence. Although the drive
to increase taxation was already underway, the need for tax revenues was underscored in 2004
when the central government withdrew government transfers to Lagos local governments
following a dispute over the creation of new local governments. The social conditions in Lagos,
with a strong middle class and active civil society, pushed the local government to deliver public
goods to win elections rather than the commonplace strategy of relying on patronage politics.
Because Lagos is one of the most populous states and the commercial center of the country, the
robust formal sector could support an expansion of personal income taxation through employer
withholding. There were also widespread allegations that the leadership of Lagos state held a
personal stake in the consulting company that was contracted to process the electronic tax
payments, including a lawsuit by the company founder against the former governor of Lagos
(Olawoyin 2020). Thus, personal benefit and the ability to solidify political power through access
to finances may have provided an added incentive.


Figure 4: Tax Revenues in Lagos State




Data Source: Lagos State Government Digest of Statistics 2006, 2010, 2013.


                                                                                                 18
Notes: Adjusted for inflation using the World Bank World Development Indicators Consumer Price Index with 2010
as the base year. Data on tax revenue is from the Lagos State Government Digest of Statistics 2006, 2010, 2013. The
period between the red vertical lines indicates when the federal government withheld transfers from Lagos local
governments.



In Lagos, the specific actions taken by the tax administration focused on deploying technology by
introducing electronic payments and tax withholding, by creating an autonomous tax agency
separate from the civil service, and by hiring and incentivizing skilled professionals. There was
political support for enforcing tax laws, like imposing fines and sealing delinquent businesses. In
addition, there was significant taxpayer education and outreach, and publicity campaigns to foster
tax morale by linking the visible public infrastructure investments to the expansion in tax
payments. Overall, this case suggests that success in increasing tax collection requires a
combination of factors, including a well-motivated government and an ability to deploy diverse
tools, particularly those relating to technology and personnel.
Another key factor that may influence a government’s decision of how much to enforce taxes is
the availability of technology tools that reduce the political and economic costs of taxation by
making tax collection more automated and less salient. For example, using employers to withhold
personal income taxes removes the need to require citizens to remit payments, and consumption
taxes like the value-added tax may be less salient to citizens. Such tools or instruments may lead
to less political resistance—but they only work well if the government has access to appropriate
technology tools for implementing them. As an example from the United States context, Cabral
and Hoxby (2012) show that property taxes are less salient to property owners who use the
technology-aided practice of bundling their taxes with their monthly mortgage payment through a
tax escrow account, compared to other homeowners with mortgages who make one or two large
direct tax payments a year to the government. Areas with higher rates of escrow use also have
higher property tax rates, suggesting that the lower salience and automation of collection allows
those jurisdictions to collect more revenues.
One contemporary example of how the availability of technology can influence the tax choices of
governments in low- and middle-income countries is the emergence of taxes on mobile money
transactions in a number of African countries. Digital payments via mobile phones have seen
substantial growth in the region, serving as a major driver of financial inclusion (Suri and Jack
2016). For governments, a major justification for taxes on mobile money transactions is to expand
the tax net over large informal sectors that are traditionally outside the current reach of tax
authorities. A mobile money tax is relatively easy to monitor and can be collected by
telecommunication companies and service providers, without the need for a tax authority to make
costly investments in identification, detection, and collection capacity. Critics have called it a “lazy
tax” (Karombo 2022). Tellingly, the salience of the tax has led to stiff public opposition, with some
countries having to reduce the tax rate (for example, Ghana, Tanzania, Uganda) and others to
completely abandon it (like Malawi).
Lastly, the decision of whether or not to pursue investments in tax capacity may depend on the
current levels of tax collection. In settings with low levels of tax collection, initial investments in
tax capacity – such as building a tax database or establishing technology for tax monitoring –

                                                                                                                19
typically have high financial costs but may yield only marginal revenue gains initially. As a result,
at the early stages of trying to expand taxation, governments in this position can be caught in a
bind: they do not yet have sufficient resources to significantly expand public good provision to
respond to the increased demands, public scrutiny, and backlash that strengthening tax capacity
might trigger among citizens. As such, political factors may contribute to the persistence of low
tax-low accountability traps (Besley and Persson 2009).


Discussion
Transforming a nation’s tax capacity requires intentional investments in the development of tax
systems. Based on the growing body of evidence on this topic, we have highlighted the role that
two factors – information technology and tax officials – can play in this process in low- and
middle-income countries. We have discussed many of the relevant studies, and in Figure 5, we
attempt to summarize recent work showing the range impacts from successful interventions.

Figure 5 displays effect sizes observed in experimental or quasi-experimental studies published
since 2011 that examine tax interventions in low- and middle-income countries. We include five
types of interventions that map to our framework and preceding sections: 1) identification focuses
on interventions that help to identify or register taxpayers; 2) detection/third-party information
refers to interventions that provide information on the liabilities of taxpayers; 3) collection-
facilitation are interventions that make it easier to pay taxes; 4) collection-enforcement are
interventions focused on the penalties for non-compliance; and 5) tax officials- incentives/
deployment are interventions in these areas. Panel A shows effects on the extensive margin of
compliance, the percentage of taxpayers that pay the tax, with effects measured in percentage point
changes relative to the mean of the comparison group. It includes 19 interventions across 13
studies. Panel B shows the gains in tax revenues, expressed as a percent of the comparison sample
mean. It includes 39 separate interventions across 26 different studies. As the goal is to show the
relative effect sizes of successful interventions, in both panels, we include only headline effects
that are statistically significant at the 10 percent level or lower. In addition, because low initial
levels may magnify gains when expressed in percentage terms, darker dots indicate higher values
of comparison means: that is, a given change is judged more substantial if achieved relative to a
larger reference level.

The studies summarized in Figure 5, as well as the broader case study evidence discussed
throughout the paper, suggest a wide menu of tax administration choices available to low- and
middle-income countries. In Panel A, effect sizes are broadly similar across intervention
categories, suggesting that extensive margin compliance can be increased from a variety of inputs
to strengthening capacity. 12 Panel B shows that the largest revenue gains observed within this set
of studies come from interventions aimed at improving collection via enforcement, deployment of
tax officials (largest gains observed in this category are from studies of deployment rather than
incentives), and improving detection via third-party information. In the most striking cases,

12
  The outlier, with a negative impact, is the Tajikistan e-filing result (Okunogbe and Pouliquen 2022) discussed
previously, which likely resulted from reduced scrutiny by officials.
                                                                                                                   20
revenues have more than doubled or even tripled, even at substantial levels of existing tax
collection.
Of course, in interpreting the results in Figure 5, it is important to remember that many of these
studies were done in local areas, not entire countries, and cover a range of different taxes. The
interventions considered by this body of work are also a function of where research has thus far
been feasible, and of where the interests of researchers have aligned with those of policymakers.
Thus, both the evidence summarized in Figure 5 and the case studies we have called attention to
in this paper are not meant to suggest a comprehensive nor “one size fits all” approach to tax
administration choices for developing countries. After all, deliberations about tax mobilization will
inevitably confront an array of issues, both economic and political. Instead, we intend our
discussion to offer inspiration – and caveats – for governments and tax authorities considering the
value of technologies and what implications their use carries for the direction of tax personnel.


Figure 5: Impacts of Tax Interventions
A. Compliance Impacts




                                                                                                  21
B. Revenue Impacts




Source: The full list of studies and results compiled by the authors is available in Appendix II.
Notes: This figure displays the effect sizes of tax interventions from existing studies. The full list of studies and
corresponding details can be found in the replication materials. The y-axis displays the type of interventions
considered, followed by the number of effect sizes included for each type (denoted by n). For Panel A, the x-axis
displays the change in tax compliance in percentage points. For Panel B, the x-axis displays changes in revenues in
percentages. The opacity of encircled dots is relative to the size of the control means (denoted in % for Panel A, and
in US dollars for Panel B) – that is, darker dots indicate higher value control means. In both the panels, each
intervention type is captured by a different color of the encircled dots. Panel A includes 19 interventions compiled
from 13 studies, and Panel B includes 39 interventions compiled from 26 studies.



There remains much to learn beyond our current understanding of how these factors contribute to
tax capacity. One clear area for future exploration in particular is understanding how to take
advantage of potential complementarities between technology and tax officials, where staff
recruitment and skills will likely play important roles. Moreover, technological advances will
undoubtably pose evolving challenges for taxation, including how to tax digital activities and
incorporate digital currencies. Finally, given how fundamental political support is for deploying
tax capacity, it is imperative to build our understanding of the factors that shape how governments
choose when, how, and how much to tax.




                                                                                                                   22
References

Almunia, Miguel, Jonas Hjort, Justine Knebelmann, and Lin Tian. "Strategic or confused firms?
Evidence from “missing” transactions in Uganda." Review of Economics and Statistics (2022): 1-
35.

Almunia, Miguel, Anne Brockmeyer, Giulia Mascagni, Vedanth Nair and Mazhar Waseem. “Does
the value-added tax add value? Empirics meets theory.” Journal of Economic Perspectives
(forthcoming).

Apeti, Ablam Estel and Denise Edoh Eyah. “Tax revenue and mobile money in developing
countries.” Journal of Development Economics, 161, (2023) 103014.

Ashraf, Nava, Oriana Bandiera, Edward Davenport, and Scott S. Lee. "Losing prosociality in the
quest for talent? Sorting, selection, and productivity in the delivery of public services." American
Economic Review 110, no. 5 (2020): 1355-1394.

Bachas, Pierre, Lucie Gadenne, and Anders Jensen. “The equity of tax systems in low and middle
income countries.” Journal of Economic Perspectives, (forthcoming).

Bachas, Pierre, Anne Brockmeyer, Alipio Ferreira, and Bassirou Sarr. “How to target enforcement
at scale? Evidence from tax audits in Senegal.” Working Paper, (2023).

Balán, Pablo, Augustin Bergeron, Gabriel Tourek, and Jonathan Weigel. “Local Elites as State
Capacity: How City Chiefs Use Local Information to Increase Tax Compliance in the D.R.
Congo.” American Economic Review, 112(3) (2022): 762-797.

Banerjee, Abhijit V., Esther Duflo, and Rachel Glennerster. "Putting a band-aid on a corpse:
incentives for nurses in the Indian public health care system." Journal of the European Economic
Association 6, no. 2-3 (2008): 487-500. https://doi.org/10.1162/JEEA.2008.6.2-3.487

Basri, M. Chatib, Mayara Felix, Rema Hanna, and Benjamin A. Olken. "Tax administration versus
tax rates: evidence from corporate taxation in Indonesia." American Economic Review111, no. 12
(2021): 3827-3871.

Bergeron, Augustin, Gabriel Tourek, and Jonathan Weigel. "The state capacity ceiling on tax rates:
Evidence from randomized tax abatements in the DRC." Working Paper, (2021).

Besley, Timothy, and Torsten Persson. "The origins of state capacity: Property rights, taxation, and
politics." American economic review 99, no. 4 (2009): 1218-1244.

Besley, Timothy, and Torsten Persson. "Why do developing countries tax so little?." Journal of
economic perspectives 28, no. 4 (2014): 99-120. DOI: 10.1257/jep.28.4.99

Bodea, Cristina, and Adrienne LeBas. "The origins of voluntary compliance: attitudes toward
taxation in urban Nigeria." British Journal of Political Science 46, no. 1 (2016): 215-238.


                                                                                                 23
Brautigam, Deborah, Odd-Helge Fjeldstad, and Mick Moore, eds. Taxation and state-building in
developing countries: Capacity and consent. Cambridge University Press, 2008.

Brockmeyer, Anne, Alejandro Estefan, Karina Ramírez Arras, and Juan Carlos Suárez
Serrato. “Taxing property in developing countries: theory and evidence from Mexico.” National
Bureau of Economic Research Working Paper 28637, (2023).

Brockmeyer, Anne, and Magaly Sáenz Somarriba. "Electronic payment technology and tax
compliance: Evidence from Uruguay's financial inclusion reform." Working Paper, (2022).

Brockmeyer, Anne, and Marco Hernandez. "Taxation, information, and withholding: evidence
from Costa Rica." Working Paper, (2022).

Brockmeyer, Anne, Spencer Smith, Marco Hernandez, and Stewart Kettle. "Casting a wider tax
net: Experimental evidence from Costa Rica." American Economic Journal: Economic Policy 11,
no. 3 (2019): 55-87. https://doi.org/10.1257/pol.20160589

Cabral, Marika, and Caroline Hoxby. “The hated property tax: Salience, tax rates, and tax revolts.”
No. w18514. National Bureau of Economic Research Working paper 18514, (2012).

Carrillo, Paul, Dina Pomeranz, and Monica Singhal. "Dodging the taxman: Firm misreporting and
limits to tax enforcement." American Economic Journal: Applied Economics 9, no. 2 (2017): 144-
164.

Chalendard, Cyril, Ana M. Fernandes, Gael Raballand, and Bob Rijkers. "Corruption in
customs." The Quarterly Journal of Economics 138, no. 1 (2023): 575-636.
https://doi.org/10.1093/qje/qjac032

Cheeseman, Nic, and Diane De Gramont. "Managing a mega-city: learning the lessons from
Lagos." Oxford      Review     of   Economic Policy 33, no.   3    (2017):    457-477.
https://doi.org/10.1093/oxrep/grx033

CIAT, IMF, IOTA, OECD (2022), International Survey on Revenue Administration: 2020 and
2021, https://data.rafit.org/?sk=8b008788-ebde-4d61-bc90-7438d6aa12dc&sId=1637191076670
Accessed: June 2, 2023.

Cotton, Ms Margaret, and Gregory Dark. “Use of technology in tax administrations 2: Core
information technology systems in tax administrations.” International Monetary Fund, (2017).

Dal Bó, Ernesto, Frederico Finan, and Martín A. Rossi. "Strengthening state capabilities: The role
of financial incentives in the call to public service." The Quarterly Journal of Economics 128, no.
3 (2013): 1169-1218.

Das, Satadru, Lucie Gadenne, Tushar Nandi, and Ross Warwick. "Does going cashless make you
tax-rich? Evidence from India’s demonetization experiment." Journal of Public Economics 224
(2023): 104907. https://doi.org/10.1016/j.jpubeco.2023.104907


                                                                                                24
de Gramont, Diane. ‘Governing Lagos: Unlocking the Politics of Reform’, Carnegie Paper,
Carnegie           Endowment         for         International       Peace,           (2015).
https://carnegieendowment.org/2015/01/12/governing-lagos-unlocking-politics-of-reform-pub-
57671

Dzansi, James, Anders Jensen, David Lagakos, and Henry Telli. "Technology and local state
capacity: Evidence from Ghana." National Bureau of Economic Research, Working Paper 29923,
(2022).

Efobi, Uchenna, Ibukun Beecroft, Tanankem Belmondo, and Amelia Katan. "Small business use
of the integrated tax administration system in Nigeria.", African Tax Administration Paper 8,
(2019), International Centre for Tax and Development. Brighton: ICTD.

Eilu, Emmanuel. "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 (2018): 111-134.

Elgin, Ceyhun, M. Ayhan Kose, Franziska Ohnsorge, and Shu Yu. "Understanding informality."
(2021) CERP Discussion Paper 16497, Centre for Economic Policy Research, London.

Fan, Haichao, Yu Liu, Nancy Qian, and Jaya Wen. “Computerizing VAT invoices in China.” NBER
Working Paper 24414. Cambridge, MA: National Bureau of Economic Research. (2020).

Ferraz, Claudio, Dirk Foremy, and Juan Santini. “Economic slumps and fiscal capacity: Evidence
from Brazil”, Working Paper, (2023).

Gaspar, Vitor, Laura Jaramillo, and Mr Philippe Wingender. “Political institutions, state building,
and tax capacity: crossing the tipping point.” International Monetary Fund Working Paper 16/233,
(2016).

Ghana News Agency. “GRA identifies over 14 million tax payers after integration of Ghana
Card/TIN.” 22 July 2021, Accessed: 14 July 2023: https://gna.org.gh/2021/07/gra-identifies-over-
14-million-tax-payers-after-integration-of-ghana-cardtin/

GhanaWeb. “Ghana card shoots up Ghanaians with TIN from 4% to 85% – Bawumia.” 20 July
2022, Accessed: 14 July 2023: https://www.ghanaweb.com/GhanaHomePage/business/Ghana-
card-shoots-up-Ghanaians-with-TIN-from-4-to-85-Bawumia-1586516

Hauck, Tobias and Luisa Wallossek. “Optional (Non-)Filing and Effective Taxation” Working
Paper, (2023)

Jensen, Anders. "Employment structure and the rise of the modern tax system." American
Economic Review 112, no. 1 (2022): 213-234.

Kapon, Samuel, Lucia Del Carpio, and Sylvain Chassang. “Using Divide-and-Conquer to Improve
Tax Collection”. National Bureau of Economic Research Working Paper 30218, (2022).


                                                                                                25
Karombo, Tawanda. ““It’s a lazy tax”: Why African governments’ obsession with mobile money
could backfire.” (2022). https://restofworld.org/2022/how-mobile-money-became-the-new-cash-
cow-for-african-governments-but-at-a-cost/

Khan, Adnan Q., Asim I. Khwaja, and Benjamin A. Olken. "Tax farming redux: Experimental
evidence on performance pay for tax collectors." The Quarterly Journal of Economics 131, no. 1
(2016): 219-271.

Khan, Adnan Q., Asim Ijaz Khwaja, and Benjamin A. Olken. "Making moves matter: Experimental
evidence on incentivizing bureaucrats through performance-based postings." American Economic
Review 109, no. 1 (2019): 237-270.

Kimani, Trizza. “36 million registered for Huduma Namba.” The Star. May 24, 2019.
https://www.the-star.co.ke/news/2019-05-24-36-million-registered-for-huduma-namba/.
Accessed November 2023.

Knebelmann, Justine, Victor Pouliquen, and Bassirou Sarr. “Discretion versus algorithms:
Bureaucrats and tax equity in Senegal.” Working Paper, (2023)

Knebelmann, Justine. "Digitalisation of property taxation in developing countries." ODI Report.
London: ODI (2022).

Lagos State Ministry of Economic Planning and Budget. “Digest of Statistics 2006”. Retrieved 7
Nov. 2023 from https://mepb.lagosstate.gov.ng/lbs-publication/

Lagos State Ministry of Economic Planning and Budget. “Digest of Statistics 2010”. Retrieved 7
Nov. 2023 from https://mepb.lagosstate.gov.ng/lbs-publication/

Lagos State Ministry of Economic Planning and Budget. “Digest of Statistics 2013”. Retrieved 7
Nov. 2023 from https://mepb.lagosstate.gov.ng/lbs-publication/

Londoño-Vélez, Juliana, and Dario Tortarolo. “Revealing 21 per cent of GDP in hidden assets:
Evidence from Argentina's tax amnesties.” World Institute for Development Economic Research
(UNU-WIDER) Working Paper 2022-103, (2022).

Martin, Lucy. Strategic Taxation: Fiscal Capacity and Accountability in African States.
Forthcoming, Oxford University Press. https://global.oup.com/academic/product/strategic-
taxation-9780197672648

Mascagni, Giulia, and Andualem Mengistu. "Effective tax rates and firm size in
Ethiopia." Development Policy Review 37 (2019): O248-O273.

Mascagni, Giulia, Andualem T. Mengistu, and Firew B. Woldeyes. "Can ICTs increase tax
compliance? Evidence on taxpayer responses to technological innovation in Ethiopia." Journal of
Economic Behavior & Organization 189 (2021): 172-193.



                                                                                            26
Mascagni, Giulia, Roel Dom, Fabrizio Santoro, and Denis Mukama. "The VAT in practice: equity,
enforcement, and complexity." International Tax and Public Finance 30, no. 2 (2023): 525-563

Mayega, Jova, Ronald Waiswa, Jane Nabuyondo, and Milly Isingoma Nalukwago. "How clean are
our taxpayer returns? Data management in Uganda Revenue Authority." African Tax
Administration Paper 24, (2021), Brighton: ICTD

McKenzie, David, and Anna Luisa Paffhausen. "Small firm death in developing countries." Review
of economics and statistics 101, no. 4 (2019): 645-657. https://doi.org/10.1162/rest_a_00798

Mitchell, Brian R. International Historical Statistics: Volume 1: Africa, Asia and Oceania 1750–
2005; Volume 2: The Americas, 1750–2005; Volume 3: Europe, 1750–2005, Palgrave Macmillan
(2007).

Moore, Mick, Wilson Prichard, and Odd-Helge Fjeldstad. Taxing Africa: coercion, reform and
development. Bloomsbury Academic, 2018.

Naritomi, Joana. “Consumers as tax auditors.” American Economic Review, 109(9): Pages 3031-
3072, (2019).

OECD Inventory of Tax Technology Initiatives (2023). https://www.oecd.org/tax/forum-on-tax-
administration/tax-technology-tools-and-digital-solutions/ Accessed: April 27, 2023.

Okunogbe, Oyebola. “Becoming legible to the state: The role of information and collection
capacity on taxation.” World Bank Policy Research Working Paper 9852, (2021)

Okunogbe, Oyebola, and Victor Pouliquen. "Technology, taxation, and corruption: evidence from
the introduction of electronic tax filing." American Economic Journal: Economic Policy 14, no. 1
(2022): 341-372. DOI: 10.1257/pol.20200123

Okunogbe, Oyebola, and Fabrizio Santoro. “The promise and limitations of information
technology for tax mobilization.” The World Bank Research Observer, 38(2): 295-324, (2023a).
https://doi.org/10.1093/wbro/lkac008

Okunogbe, Oyebola, and Fabrizio Santoro. “Increasing tax collection in African countries: The
role of information technology.” Journal of African Economics, 32(1):Pages i57–i83 (2023b).
https://doi.org/10.1093/jae/ejac036

Olawoyin, Oladeinde. “Tinubu, Alpha-beta, others accused of fraud, tax evasion, money
laundering.” Premium Times October 8, 2020, Accessed: 14 July 2023:
https://www.premiumtimesng.com/news/headlines/419179-tinubu-alpha-beta-others-accused-of-
fraud-tax-evasion-money-laundering.html.

Prichard, Wilson. "Electoral competitiveness, tax bargaining and political incentives in developing
countries: Evidence from political budget cycles affecting taxation." British Journal of Political
Science 48, no. 2 (2018): 427-457.


                                                                                                27
Prichard, Wilson. "Tax, politics, and the social contract in Africa." In Oxford research encyclopedia
of politics. (2019).

Regan, Tanner and Priya Manwaring, 2023. "Public disclosure and tax compliance: Evidence from
Uganda," Institute for International Economic Policy Working Paper 2023-04, George Washington
University.

Santoro, Fabrizio, Adrienne Lees, Marco Carreras, Theonille Mukamana, Naphtal Hakizimana,
and Yves Nsengiyumva. 2023. "Technology and tax: Adoption and impacts of e-services in
Rwanda," ICTD Working Paper 153, Brighton: Institute of Development Studies, DOI:
10.19088/ICTD.2023.001.

Shah, Jawad A. “Using computerized information to enforce VAT: Evidence from Pakistan.”
Working Paper, (2023).

Slemrod, Joel, and Michael Keen. “Optimal tax administration.” Journal of Public Economics,
152, August (2017): 133-142.

Suri, Tavneet, and William Jack. "The long-run poverty and gender impacts of mobile
money." Science 354, no. 6317 (2016): 1288-1292.

UNU-WIDER. “Government revenue dataset”. Version 2022. https://doi.org/10.35188/UNU-
WIDER/GRD-2022. Accessed: October 14, 2022.

UNU-WIDER. “World Income Inequality Database (WIID)”. Version 30 June 2022.
https://doi.org/10.35188/UNU-WIDER/WIID-300622

Weigel, Jonathan L. "The participation dividend of taxation: How citizens in Congo engage more
with the state when it tries to tax them." The Quarterly Journal of Economics 135, no. 4 (2020):
1849-1903.

Wier, Ludvig S., and Gabriel Zucman. “Global profit shifting, 1975-2019.” National Bureau of
Economic Research Working Paper 30637, (2022).

World Bank Enterprise Surveys, http://www.enterprisesurveys.org Accessed: June 2, 2023.

World Bank. “World Development Indicators”. Version 2023. Washington, DC: The World Bank.
https://databank.worldbank.org/source/world-development-indicators Accessed: May 1, 2023.




                                                                                                  28
Appendix
Appendix I: Cross Country Evidence on Use of Technology for Identification, Detection and
Collection
As discussed under “Cross Country Evidence on Use of Technology for Identification, Detection
and Collection”, we construct indices of taxation technology computed from individual measures
drawn from the OECD Inventory of Tax Technology Initiatives (ITTI) (2023) Global Survey on
Digitalisation, which collects measures of uses of technology and data for tax purposes. More
information about the tax technology measures can be found at https://www.oecd.org/tax/forum-
on-tax-administration/tax-technology-tools-and-digital-solutions/.
We construct indices from indicators for variables included in each set, described in detail below.
To build each index we first sum the indicators and then standardize the resulting sum so that it
has a mean of zero and standard deviation equal to 1. Appendix Table 1 below summarizes the
estimates of the relationships between indices and tax-to-GDP in 2018.
   a. The taxpayer identification index is calculated from the indicators listed below. The
      unstandardized index (sum) has mean 3.51 and standard deviation 0.74.
      1. whether the tax authority requires taxpayers to have a digital ID
      2. the digital ID is built on an existing system
      3. a digital ID can be established using a unique identity number, government issued
         documents, or biometric information
      4. whether there are online services for registration for any tax type (PIT, CIT, VAT)

   b. The detection index is calculated from the indicators listed below. The unstandardized
      index (sum) has mean 4.41 and standard deviation 1.81.
      1. whether the tax authority receives data on taxpayers
      2. whether the tax authority receives data from third parties
      3. whether some taxpayers are required to submit e-invoices
      4. some taxpayers are required to maintain online cash registers
      5. the tax authority receives data from other government bodies
      6. there exists a common database across government bodies
      7. whether the tax authority uses AI to conduct risk assessments for any tax type
      8. whether the tax authority uses AI to detect evasion.

   c. The collection capacity index is calculated from the indicators listed below. The
      unstandardized index (sum) has mean 2.56 and standard deviation 1.16.
      1. taxpayers can file online for any tax type
      2. taxpayers can pay online for any tax type
      3. taxpayers can request an extension online for any tax type
      4. taxpayers can ask for a payment arrangement online for any tax type

   d. The “taxation technology” index used in Figure 2 is a summary index that includes all of
      the variables listed above. The unstandardized index (sum) has mean 10.49 and standard
      deviation 2.58.
                                                                                                29
   Appendix Table 1: Taxation Technology and Tax-to-GDP
                                              Tax Revenues as a Share of GDP (2018)
   Index of Tax Technology                       2.155*
                                              (1.133)
   Index of Identification                              1.679*
                                                        (0.858)
   Index of Detection                                                2.136**
                                                                  (1.008)
   Index of Collection                                                         0.033
                                                                               (1.031)

   Country income group fixed
                              Yes                             Yes         Yes               Yes
   effects
   Countries                  75                              75          75                75
Notes: This tables displays the estimates of the relationship between tax revenues as a share of GDP
and an indices of taxation technology for 74 individual countries. The tax revenues as a share of GDP
measure is drawn from the International Survey on Revenue Administration (CIAT, IMF, IOTA,
OECD 2022) using values for 2018. We fill in missing values for 7 out of 74 countries using UNU-
WIDER (2022) data in 2018. The indices of taxation technology is computed from individual
measures drawn from the OECD Inventory of Tax Technology Initiatives (ITTI) (2023) Global
Survey on Digitalisation. The identification index is calculated from indicators for whether the tax
authority requires taxpayers to have a digital ID, the digital ID is built on an existing system, a digital
ID can be established using a unique identity number, using government issued documents, using
biometric information, and whether there are online services for registration for any tax type (PIT,
CIT, VAT. The detection index is calculated from indicators for whether the tax authority receives
data on taxpayers, receives data from third parties, some taxpayers are required to submit e-invoices,
some taxpayers are required to maintain online cash registers, the tax authority receives data from
other government bodies, there exists a common database across government bodies, whether the tax
authority uses AI to conduct risk assessments for any tax type, and whether the tax authority uses AI
to detect evasion. The collection index is calculated from indicators for whether taxpayers can file
online for any tax type, taxpayers can pay online for any tax type, taxpayers can request an extension
online for any tax type, and whether taxpayers can ask for a payment arrangement online for any tax
type. More information about the tax technology measures can be found at
https://www.oecd.org/tax/forum-on-tax-administration/tax-technology-tools-and-digital-solutions/.
All regressions include fixed effects for country income group. Standard errors in parentheses are
robust.




                                                                                                              30
Appendix II: Impacts of Tax Interventions
Below is the list of papers included in Figure 5. Details on the results included are provided in the
online replication dataset.


Tax officials: incentives/deployment
Amodio, Francesco, Jieun Choi, Giacomo De Giorgi, and Aminur Rahman. "Bribes vs. taxes:
Market structure and incentives." Journal of Comparative Economics 50, no. 2 (2022): 435-453.
Balan, Pablo, Augustin Bergeron, Gabriel Tourek, and Jonathan L. Weigel. "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, no. 3 (2022): 762-797.
Basri, M. Chatib, Mayara Felix, Rema Hanna, and Benjamin A. Olken. "Tax administration versus
tax rates: evidence from corporate taxation in Indonesia." American Economic Review 111, no. 12
(2021): 3827-3871.
Bergeron, Augustin, Pedro Bessone, John Kabeya Kabeya, Gabriel Z. Tourek, and Jonathan L.
Weigel. “Optimal assignment of bureaucrats: Evidence from randomly assigned tax collectors in
the DRC.” No. w30413. National Bureau of Economic Research, 2022.
Dzansi, James, Anders Jensen, David Lagakos, and Henry Telli. "Technology and local state
capacity: Evidence from Ghana." National Bureau of Economic Research, Working Paper 29923,
(2022).
Khan, Adnan Q., Asim I. Khwaja, and Benjamin A. Olken. "Tax farming redux: Experimental
evidence on performance pay for tax collectors." The Quarterly Journal of Economics 131, no. 1
(2016): 219-271.
Ortega, Daniel, and Carlos Scartascini. "Don’t blame the messenger. The delivery method of a
message matters." Journal of Economic Behavior & Organization 170 (2020): 286-300.
Weigel, Jonathan L. "The participation dividend of taxation: How citizens in Congo engage more
with the state when it tries to tax them." The Quarterly Journal of Economics 135, no. 4 (2020):
1849-1903.


Identification
Jouste, Maria, Milly I. Nalukwago, and Ronald Waiswa. "Do tax administrative interventions
targeted at small businesses improve tax compliance and revenue collection?." Evidence from
Ugandan Administrative Tax Data.” (2021).
Okunogbe, Oyebola. “Becoming legible to the state : The role of identification and collection
capacity in taxation.” World Bank Policy Research Working Paper 9852, (2021)



                                                                                                  31
Detection/third-party info
Ali, Merima, Abdulaziz B. Shifa, Abebe Shimeles, and Firew Woldeyes. "Building fiscal capacity
in developing countries: Evidence on the role of information technology." National Tax
Journal 74, no. 3 (2021): 591-620.
Bellon, Matthieu, Era Dabla-Norris, Salma Khalid, and Frederico Lima. "Digitalization to improve
tax compliance: Evidence from VAT e-Invoicing in Peru." Journal of Public Economics 210
(2022): 104661.
Brockmeyer, Anne, Spencer Smith, Marco Hernandez, and Stewart Kettle. "Casting a wider tax
net: Experimental evidence from Costa Rica." American Economic Journal: Economic Policy 11,
no. 3 (2019): 55-87.
Carrillo, Paul, Dina Pomeranz, and Monica Singhal. "Dodging the taxman: Firm misreporting and
limits to tax enforcement." American Economic Journal: Applied Economics 9, no. 2 (2017): 144-
164.
Eissa, Nada, Andrew Zeitlin, Saahil Karpe, and Sally Murray. "Incidence and impact of electronic
billing machines for VAT in Rwanda." International Growth Centre Report (2014).
Fan, Haichao, Yu Liu, Nancy Qian, and Jaya Wen. “Computerizing VAT invoices in China.” No.
w24414. National Bureau of Economic Research, 2018.
Londoño-Vélez, Juliana, and Javier Ávila-Mahecha. "Can wealth taxation work in developing
countries? Quasi-experimental evidence from Colombia." In Annual Congress of the IIPF. 2018.
Mascagni, Giulia, Andualem T. Mengistu, and Firew B. Woldeyes. "Can ICTs increase tax
compliance? Evidence on taxpayer responses to technological innovation in Ethiopia." Journal of
Economic Behavior & Organization 189 (2021): 172-193.
Mittal, Shekhar, and Aprajit Mahajan. "VAT in emerging economies: Does third party verification
matter?" Available at SSRN 3029963 (2017).
Naritomi, Joana. "Consumers as tax auditors." American Economic Review 109, no. 9 (2019):
3031-3072.
Shah, Jawad. "Using computerized information            to   enforce   VAT:    Evidence    from
Pakistan." Available at SSRN 4569607 (2020).


Collection: facilitation
Das, Satadru, Lucie Gadenne, Tushar Nandi, and Ross Warwick. "Does going cashless make you
tax-rich? Evidence from India’s demonetization experiment." Journal of Public Economics 224
(2023): 104907. https://doi.org/10.1016/j.jpubeco.2023.104907
Jouste, Maria, Milly I. Nalukwago, and Ronald Waiswa. "Do tax administrative interventions
targeted at small businesses improve tax compliance and revenue collection?." Evidence from
Ugandan Administrative Tax Data.” (2021).
                                                                                             32
Mascagni, Giulia, Fabrizio Santoro, and Denis Mukama. "Teach to comply? Evidence from a
taxpayer education programme in Rwanda." (2019).
Okunogbe, Oyebola, and Victor Pouliquen. "Technology, taxation, and corruption: evidence from
the introduction of electronic tax filing." American Economic Journal: Economic Policy 14, no. 1
(2022): 341-372.


Collection: enforcement/sanctions
Aparicio, Gabriela, Paul E. Carrillo, and M. Shahe Emran. "Taxes, prisons, and CFOs: The effects
of increased punishment on corporate tax compliance in Ecuador." Available at SSRN
1808945 (2011).
Brockmeyer, Anne, Spencer Smith, Marco Hernandez, and Stewart Kettle. "Casting a wider tax
net: Experimental evidence from Costa Rica." American Economic Journal: Economic Policy 11,
no. 3 (2019): 55-87.
Castro, Lucio, and Carlos Scartascini. "Tax compliance and enforcement in the pampas evidence
from a field experiment." Journal of Economic Behavior & Organization 116 (2015): 65-82.
Holz, Justin E., John A. List, Alejandro Zentner, Marvin Cardoza, and Joaquin E. Zentner. "The
$100 million nudge: Increasing tax compliance of firms using a natural field experiment." Journal
of Public Economics 218 (2023): 104779.
López-Luzuriaga, Andrea, and Carlos Scartascini. "Compliance spillovers across taxes: The role
of penalties and detection." Journal of Economic Behavior & Organization 164 (2019): 518-534.
Mascagni, Giulia, and Christopher Nell. "Tax compliance in Rwanda: Evidence from a message
field experiment." Economic Development and Cultural Change 70, no. 2 (2022): 587-623.
Okunogbe, Oyebola. “Becoming legible to the state : The role of identification and collection
capacity in taxation.” World Bank Policy Research Working Paper 9852, (2021)
Ortega, Daniel, and Pablo Sanguinetti. "Deterrence and reciprocity effects on tax compliance:
experimental evidence from Venezuela." (2013).
Ortega, Daniel, and Carlos Scartascini. "Don’t blame the messenger. The delivery method of a
message matters." Journal of Economic Behavior & Organization 170 (2020): 286-300.




                                                                                              33