Big Data and Domestic Resource Mobilization: How Donors Can Help Developing Countries Increase Revenue
Domestic resource mobilization (DRM)1 is the capacity of a government to generate income through taxes, fees, levies, or other related resources. Increases in DRM are often achievable through the improvement of tax administration and taxpayer compliance. Increasing DRM enables developing countries to invest more of their own resources in high-priority services such as health and education, thereby reducing their dependence on donor funding for key programs. In addition, improved DRM has the potential to help countries promote good governance, strengthen domestic accountability, achieve more sustainable and inclusive growth, and reduce poverty.
The international donor community increasingly recognizes DRM as a key element of the development equation, in combination with sound public spending and a concerted fight against illicit financial flows. Last year, after more than a decade of effort to improve financing for development, the international donor community and partner countries agreed to the Addis Tax Initiative—Declaration, which prioritized DRM and committed donors to double their DRM investments in low- and middle-income countries.
The emergence of DRM as a donor priority coincides with the ascendancy of Big Data as a driving force in organizations worldwide. Big Data is bringing about revolutionary changes in various industries, for example, including healthcare and medicine, retail, and banking.
The information technology (IT) firm CRM recently reflected on how Big Data has returned “big payoffs” in five quite disparate sectors: a pizza chain using Big Data to push coupons on regular customers, based on weather expectations; a hotel combining data about weather, flight cancellations, other hotel bookings, and related information to pick up business from stranded travelers; a music company planning its releases based on iTunes information; and the big-box retailer Target using customer information to ascertain when a person is likely pregnant, then directing her to appropriate purchases.
Big Data is a somewhat amorphous concept, but the term has core elements. By means of Big Data, institutions can work with greater amounts of data from more diverse data sources, thanks to analytics enabled by specialized applications, techniques, algorithms, and software. Organizations that have embraced Big Data now have much better access to information about their industry, organizations, and customers, as well as contextual information from adjacent sectors or further afield. They can tap well-organized information from their own internal business processes and, if they have cracked the Big Data nut, they can systematically integrate this external information to meet their market needs.
The question is whether developing countries’ tax administrations can take similar advantage of the Big Data revolution to strengthen DRM. On the surface it would seem that revenue agencies could benefit from Big Data management and analytics to exploit:
- Enterprise data taken from their own sources and tax forms;
- External domestic data, such as from other parts of government (for example, customs, property maps, and commercial registries), the banking sector, international financial flows, credit card usage, and air ticket purchases;
- GPS and geographic information systems, such as google maps2; and
- International data, such as the Automatic Exchange of Information (AEOI) initiative.
However, revenue agencies in developing countries do not always have the capability to analyze huge caches of data. The data often are not cleaned or optimally organized, limiting the ability to trace patterns, identify trends, and find correlations. Too often, software is set loose on the data without the data analyst specifying parameters or models.
This paper explores how international donors can assist developing countries to take advantage of Big Data to better mobilize domestic resources.3
Big Data techniques can help developing countries compile and make connections between various types of data, and utilize these data to identify patterns and anomalous behavior—often in real time. Among other things, Big Data analytics can be used to:
- Identify surges in goods moving through ports, which might indicate smuggling.
- Detect anomalous movements in bank balances and international (potentially illicit) financial flows.
- Calculate correlations among certain economic activities and taxpayer behaviors.
- Tease out patterns in tax audits—by regional office, economic activity, regional manager, specific auditor, audit performance, and so on.
- Sniff out consumption patterns of public officials that might indicate malfeasance.
How Donors Can Help Revenue Agencies Benefit From Big Data
To mobilize revenue, tax administrations need to encourage voluntary compliance, enforce the tax law, and provide information to the fiscal system. Voluntary compliance means taxpayers fill out the required forms with accurate information and pay the tax they are supposed to pay on time. Tax administrations, by current best practice, encourage voluntary compliance by making it easy for taxpayers to comply, detecting those who do not comply, and enforcing legal sanctions against non-compliers.
Encouraging voluntary compliance is a game where risk-management techniques are applied to generate probabilistic indications of where non-compliance is happening; the tax administration then takes action to address likely non-compliers without imposing burdens on the compliant. To play this game, tax administrations need information that will help them to identify taxpayers, people who should be registered for taxation but are not, people who are registered to pay and file taxes but who have simply stopped filing, people who are filing but not paying, people who owe taxes but are in arrears, people who register but do not really exist or whose addresses and other information are bogus, and people who are filing their tax forms and returns, but whose returns include incorrect information, either intentionally or inadvertently.
Today’s more advanced developing country tax administrations have been exploiting both enterprise data from their own operations and taxpayer reporting, sometimes informed by external source data, and applying risk management techniques to encourage taxpayer compliance. But none, so far, has made the leap to integrating Big Data into their day-to-day operations. IT modernization is a pre-requisite for this to happen.
Such modernization has been a key component of many U.S. Agency for International Development (USAID) tax reform efforts, for example. The list includes Liberia in the early and mid-1980s, El Salvador and Guatemala in the 1990s and early 2000s, Bosnia and Herzegovina from 2001 to 2006, Moldova from 2007 to 2009, Jordan from 2009 to 2012, and the Philippines from 2013 to 2015. A recent Organisation for Economic Co-operation and Development (OECD) discussion paper finds that “Modern information technology (IT) is a core component of any overall tax modernisation effort.”
Most IT reform efforts focus on developing specific applications such as e-declarations, building case management or workflow engines, or cleansing or replacing taxpayer registries. In some cases, such as in El Salvador in the early 1990s or Bosnia and Herzegovina in the early 2000s, donor projects helped procure and install completely new IT systems.
The Six Vs
The opportunities for Big Data to support taxation are many, but tax administrations first need to get their current IT services in order4. They need to implement e-services of all types and well-automated business processes, integrate good data analysts into everyday operations, and hold to a vision of the future that embraces evolving technology.
For a tax administration to be ready to catch the Big Data wave, it must build capacities against the six basic readiness requirements, or the six Vs: Vision, Volume, Variety, Velocity, Veracity, and Value. Donors can help the tax administration make this readiness assessment and address any gaps identified. We discuss each of these areas below.
1. Create a Big Data Vision
The tax administration requires a Chief Information Officer (CIO) on the leadership team who understands IT as well as tax administration. Able to communicate technical solutions to non-technical decision makers and line managers alike, the CIO must lead the development of a clear vision, with a strategy and detailed plan for making it real. Donors cannot play the role of the CIO but should probably refrain from encouraging a Big Data approach in the absence of such a counterpart.
Donors can be helpful to the tax administration, and especially the CIO, in its vision-building phase by helping to develop a Big Data strategic plan. Strategic planning advisors can work with counterparts to explore the opportunities for Big Data, conceptualize the institutional and data relationships that must be established, assess how Big Data will lead to further organizational change, inform the change management process, and identify the talent needed to implement the strategy. They can also provide advice and share experiences from other countries or industries where Big Data has transformed an organization.
2. Meet the Volume Requirements
The tax administration’s infrastructure and staff need to be able to handle the volume that Big Data implies. This capacity includes having the appropriate systems and methods for capturing all the right enterprise data, especially data provided by taxpayers, as well as the ability to capture other data. The tax administration needs to have the servers, switches, and other hardware to house data or access to cloud services. Most importantly, tax administration staff must be equipped with the knowledge and tools for managing big databases and creating and storing data tables and reports.
If tax forms are not predominantly electronic or based on 2d barcoding, then the tax administration is probably not ready to handle the volumes that Big Data implies. The administration should implement e-filing, e-payment, and other e-services as a first step on a long path.
3. Prepare for a Variety of Data Types
Tax administrations must be ready to integrate and use data of different sources and types: from financial data from the banking, financial, and insurance sectors to imagery data from cadasters or GPS locational information. International donors can help assess the path to managing a wide variety of data by addressing the following questions:
- Does the tax administration have, or can it get, access to external data?
- Are there legal obstacles to accessing external data?
- Does the tax administration have the systems, tools, and people to integrate a variety of data and data types into a holistic database that can serve its and others’ needs?
4. Develop Systems to Handle Velocity
Donors can help ensure the tax administration’s systems can produce, handle, and utilize high-velocity or real-time data. Real-time links between customs and internal revenues can be useful for tracking imports, exports, payments, VAT credits, and payments of excises. Real-time links with other government systems, such as the commercial register, can be extremely helpful to taxation. Real-time links to organizations outside of government, such as banks or the Automatic Exchange of Information (AEOI) Forum, can be facilitated by means of data-sharing programs5.
Some tax administrations have their own data centers, while others rely on a government-wide data center. Today, many countries are using cloud services to readily share data in various forms.
5. Assess the Big Data Value Proposition
It is not a given that investing in Big Data will yield value for all tax administrations at all times. Individual tax administrations will have to make individual decisions on both cost-benefit and value-for-money bases, with an emphasis on increasing revenues. The value assessment should also guide choices between bespoke systems, off-the-shelf software, or freeware; between the expansion of in-house facilities and the use of cloud solutions; and between the recruitment of technical experts and the outsourcing of services or staffing.
6. Build Veracity into the Process
Data veracity is a pervasive problem: taxpayers cheat, taxpayers make errors, taxpayers provide insufficient information, data requested is greater than processing capabilities, and processes are poor—especially controls over the taxpayer registry.
The taxpayer registry lies at the very center of this information cyclone. Taxpayers, whether companies or individuals, are required to register with the tax administration at some time during their economic lives. The taxpayer registry is invariably a very large database, usually housing information about millions of taxpayers.
Attached to the registry are the tax accounts, where taxes owed and paid by the taxpayer are tracked and linked through the taxpayer’s unique taxpayer identification number. In many countries, taxpayers must file numerous forms or returns with the tax administration for each tax they pay. For instance, a company may pay estimated corporate income tax, personal income tax withholdings, and VAT, among others—each, say, four times per year. Millions of taxpayers, making millions of filings, means a very large base of information.
Donor assistance often includes cleaning up the registry: identifying and removing defunct businesses and deceased individuals, for example, or resolving duplications. Other registrations, too, such as the debt registry (that is, the registration of moneys owed to the tax administration) are often outdated and compromised by fraudulent locations and names. In addition, the vintage of obligations (the date the debt was incurred) is too often unknowable.
USAID’s Fiscal Reform II Project in Jordan, implemented by DAI, helped clean up the debt registry of that country’s Income and Sales Tax Department and was able to separate collectable from uncollectable debt by identifying its vintage and the location of the taxpayer (in or out of the country), among other factors. In this way, we reduced the seemingly enormous stock of arrears to a more manageable and minable source of revenue for tax collectors.
CNBC commentator Curtis Clark reports on cases in a Middle Eastern country and in an American Northeastern state where tax administrations have used analytics to find anomalies in tax system data. The Middle Eastern country “used predictive analytics to examine 10 years’ worth of existing tax-fraud cases. Using that data, investigators built a system that is able to spot new (likely tax fraud) cases….That country expects to boost recoveries from fraud investigations by 25 percent.” In the case of the U.S. state, the tax administration moved to rapid appraisal of tax returns using predictive analytics to detect errors and fraud in returns—saving hundreds of millions of dollars each year since.
Five Crosscutting Activities
In addition to addressing the “six Vs,” donors can help revenue agencies build their capacity to exploit Big Data in more crosscutting ways.
1. Legal and Regulatory Reform
Current laws and regulations often pose roadblocks to IT modernization and e-services implementation. For instance, the USAID-funded Business and Tax Administration Reform Project was charged with designing an e-filing system in collaboration with the Moldovan State Tax Inspectorate, only to encounter regulatory obstacles such as centralized control and high prices for e-signatures that made e-filing unattractive to most taxpayers. Donor-funded experts can help counterparts identify such legal and regulatory obstacles, facilitate dialogue on these issues, and draft model legislation that partner governments can tailor to their own national contexts to facilitate the move to Big Data.
2. Organizational Review
Much has been written over the past two decades on how tax administrations should be organized. These recommendations are usually based on received wisdom in the field of organizational development, but there is a strong consensus that tax administrations should be organized by function (accounting, customer service, audit) rather than by tax type (VAT, income tax, excises). This consensus also holds that tax administrations should organize, in part, according to taxpayer segmentation (size), or at least have a unit dedicated to large taxpayers and perhaps others for medium-sized taxpayers.
The advent of Big Data may require changes in organizational set-up. While we have little information about how tax administrations have incorporated Big Data into their organizations, we do have evidence on how this has been done in the private sector. One study by Capgemini Consulting found that while 90 percent of U.S. banks, for example, would like to integrate Big Data analytics into their everyday operations, only 37 percent actually do so. The obstacles cited include internal segregation of data and the teams charged with managing them, a shortage of skilled people, heavy time requirements, unstructured content, the complexity of databases involved, and the high costs of storing and maintaining them. Tax administrations in developing countries are likely to face similar challenges, and while there is no recipe for incorporating Big Data into tax organizations, it is clear that careful thought should be given to organizing for Big Data exploitation.
3. Human Resources and Capacity Building
To take advantage of Big Data, tax administrations should train their staff in integrating Big Data into everyday operations. This observation applies to all staff involved in general business processes, but especially to those directly involved in risk analysis and risk management. Building Big Data into core business processes will require strong IT cadres who can ensure the effective management, interaction, and administration of complex IT systems—people able not only to maintain the IT infrastructure and serve customers, but to see ahead of the tech curve and incorporate ever-changing IT capabilities into tax administration processes. Aside from the need for a visionary CIO, the tax administration will need a cadre of data scientists: experts in how to elicit useful knowledge from a wide variety of data types. Encompassing statistics, data mining, and predictive analytics, data science is a relatively new field, but experience in the private sector indicates that absent this type of expertise, Big Data analytics will not yield its full promise.
“Infrastructure is the cornerstone of Big Data architecture,” writes Eileen McNulty in Dataconomy. “Possessing the right tools for storing, processing and analyzing your data is crucial in any Big Data project.” Tax administrations will benefit from soliciting expert advice on how to choose between the various infrastructure options available to them: open source hardware, such as “Hadoop,” which is free but lacks services and assistance unless one buys higher-level support; other software packages that typically come with annual service charges included; and cloud services, which can provide an economical approach to Big Data infrastructure, allowing the tax administration to forgo heavy capital expense and pay for services as they are used.
5. Knowledge Sharing
We do not yet know enough about experiences with Big Data and tax administration, nor about how experience in other sectors can yield lessons for tax administration. Many developing country tax administrations are still limping along in their application of IT, let alone jumping on the Big Data bandwagon. But an important first step—and one that donors are well positioned to support—is establishing what we know, what we don’t know, and what we are learning from early experiments with Big Data.
Big Data has a lot to offer in terms of mobilizing revenues, but getting to the point where a tax administration is ready to reap the benefits will take effort, talent, and vision. Where the information society is a living concept, where data analysts or even data scientists are readily available, where the legal and regulatory framework facilitates data sharing, and where privacy protection is not overly burdensome, employing Big Data to increase DRM is likely to be more feasible. Donors can help eager counterparts to overcome obstacles, set targets, and implement roadmaps to Big Data and modern taxation.
As a useful follow-up to the Addis Tax Initiative commitments, donor countries should explore the potential of using Big Data for DRM, especially by bringing together experts and policy makers in Big Data, taxation, and technological modernization. Encouragingly, developing countries have shown that in other sectors—think telecom, and the mobile revolution—they are able to “leapfrog” their developed-country counterparts and rapidly catch up with the rest of world.
Could such a leap be before us now?
Donors and international institutions use DRM interchangeably for either domestic resource mobilization or domestic revenue mobilization. ↩
In Greece, one measure taken to collect revenue from wealthy but tax-averse householders was to tax private swimming pools. In this upper-middle-income country of 11 million people, only 324 pools from the smarter suburbs of Athens had been registered, and the number had not increased in a long time. Finally, as Cory Doctorow reports, the tax authorities used satellite imagery to locate 16,974 pools in these neighbourhoods. ↩
I would like to thank Charlie Gallagher, Alex Kitain, and Admir Zajmovic for their comments and guidance on the technical aspects of this paper; and Stephen Carpenter and David Hall for guidance on direction and style. All errors, however, are my own. ↩
In 2015, the International Tax Compact and KFW study, Information Technology in Tax Administration in Developing Countries, determined “readiness requirements” for undertaking to install an integrated tax administration system, based on case studies in Mozambique, Peru, Senegal, and Swaziland. Bottom line: if a country has not already exceeded these readiness requirements and implemented an integrated tax administration system, it should not yet be contemplating a Big Data solution. ↩
Established with the assistance of the OECD, the Global Forum on Transparency and Exchange of Information for Tax Purposes sets the standards for the AEOI, which in turn seeks to help countries reduce tax evasion through international cooperation in information exchange. Participating countries agree to exchange information each year on an automatic basis, without having to petition for such information. ↩