Understanding Tax Evasion Through Data
A Carnegie Mellon University’s (CMU) Heinz College capstone project completed in partnership with the Open Government Partnership
By: Mina Narayanan, Sarah Sherwood, Mush Chowdhury, and Anirudh Koka
OGP’s Chief Research Officer, Joseph Foti, served as an advisor of the project.
This guest piece is part of exploratory work that the OGP Support Unit is carrying out around open government and tax to address issues of domestic resource mobilization, countering kleptocracy, and financing sustainable development. The article discusses a centralized database of publicly available information now available here, for use by open government reformers. We hope to expand this work in earnest with partners worldwide in 2022. For more information contact research@opengovpartnership.org.
According to the State of Tax Justice 2020, countries are losing over $427 billion in tax each year to international corporate tax abuse and private tax evasion. There is growing interest in addressing issues of tax using open government approaches generally — transparency, civic participation, and public accountability — and more specifically through OGP action plans. Part of getting to this is understanding the nature of the problem in OGP countries. OGP members span 78 countries and 76 local jurisdictions, making them well placed to tackle tax evasion at the national and local levels through their action plans. Beyond this, there is a need for more research into tax evasion at the national level and the specific open government mechanisms reformers can develop. In addition to a general understanding, more specific understanding is needed on how tax affects different social groups, such as women and those employed in informal or subsistence sectors.
Identifying Scale and Kinds of Tax Evasion
The 2008 financial crisis brought the issues of tax evasion and avoidance to popular consciousness and established that they are far more central to the global economy than previously understood. Since then, organizations and governments have put forth efforts to further understand these issues and their true cost to the global economy. Recently, the OECD estimated that base erosion and profit shifting practices, or tax avoidance, cost $100 to $240 billion annually in lost tax revenue, equating to 4 to 10% of the global corporate tax base. Additionally, OECD also gauged that the amount of income reallocated by multinational firms to lower-tax countries ranged from 2 to 4%.
This tremendous missing revenue has very real implications for human development. Countries need large bases of tax revenue to fund their own economic development. High rates of tax evasion make it difficult for governments to efficiently collect revenue from taxpayers, which in turn limits spending on public services. For instance, while countries in South Asia have experienced an average of 7 to 8% growth, about twice the global average, high rates of tax evasion have continued. As a result, the region’s tax-to-GDP ratios have remained relatively stagnant over the past ten years, restricting the government’s ability to fund infrastructure and social development. Clearly, the global movement for open government can help to reduce tax evasion and avoidance. In fact, doing so will be essential for other core open government goals such as sustainable development, fostering equality, countering kleptocracy, and strengthening democracy.
As OGP members evolve in their approach to tackling issues of inequality, sustainable development, and corruption, they will need information on how and where improving open government around tax can help.
Comprehensive data is needed to craft effective policy responses to a country’s own tax evasion issue, however, that data in each OGP country is sparse. This data will need to address both the actual phenomenon of tax evasion as well as how taxes are governed.
As CMU graduate students, we worked with OGP to identify the ideal data on tax evasion and better understand which datasets currently exist to improve tax transparency. Throughout the semester, we carried out a data census, evaluated the data sources, and compiled the relevant data into a comprehensive database that the OGP Support Unit could use to inform OGP members’ action planning processes, should they choose to focus on improving their respective tax evasion and transparency policies.
Constructing a Comprehensive and Effective Database
Building the dataset had three components — data identification, compilation, and standardization.
- Data identification: Tasked with a directive to find datasets related to tax evasion and illicit financial flows, we found online sources that contained information ranging from wealth inequality indices to the effectiveness of budget oversight institutions. We then carefully evaluated the integrity of these sources and selected indicators that fit into at least one of three policy clusters. The first cluster describes how a country handles its revenue problem, the second encompasses ways that countries can prevent illicit financial outflows, and the third refers to controls to prevent illicit financial inflows. We also flagged indicators to denote their relationship to the three OGP values of transparency, participation, and accountability. Our classification scheme aims to help future researchers conceptualize the expansive field of tax evasion.
- Compilation: We compiled all of this data into a database that accommodates various use cases. To this end, we included indicators that measured similar phenomena but relied on different methodologies (e.g., profit shifting in the Missing Profits Database and the State of Tax Justice Report) to capture multiple research approaches. For benchmark indicators with scores (e.g., 0 to 100 for Financial Secrecy Index), we also recorded the sub-indicators that contributed to the score to give more context to their formulation. The richness of our data allows users to identify the number of OGP members that mandate reporting of large financial transactions, which contributes to the “Banking Secrecy” score of the Financial Secrecy Index (FSI). Alternatively, a user can study the distribution of data by building a histogram, world map, or heatmap to compare the values of indicators between member nations, while scatter plots can be utilized to explore the relationships between indicators across countries.
- Standardization: While examining data on tax evasion is a useful endeavor, its utility is limited if the data is not standardized. To account for this, we developed a special naming convention to clearly identify each indicator and color-coded indicators to differentiate between those that describe the context of tax evasion in a country and those that are related to the policy clusters. To preserve different versions of the database, we built a Google Apps Script that encodes our data using a 64-bit encoding scheme and packages it into a JSON format. Anyone with access to the database can manually run the script to transform the Google Sheet into a JSON file and push it to a Github repository.
Revelations of the Data Census
Although our extensive data census resulted in a decent amount of data, we found that there were also notable gaps in the information we collected.
- Sparse participation and accountability data: For one, our visualizations of the data distributions revealed that we had collected very few participation and accountability indicators. We suspect that the type of our data sources may contribute to this uneven distribution. Many of our indicators were taken from international, well-established organizations that focus on transparency-related data instead of civil society groups that may be more familiar with participation and accountability initiatives. Because these organizations track and supply the majority of our data, we found it difficult to locate participation or accountability-related indicators that covered all OGP member countries. The one exception is the International Budget Partnership which does track disclosure, participation, and oversight on issues of revenue.
- Few sources for many indicators: After analyzing the distribution of our data, we also discovered that most of our indicators came from two sources, despite using a total of nine sources for our database. Nearly 80% of all the indicators we used came from FSI and OECD Tax Cooperation, with FSI accounting for 74% alone. Both of these sources are reputable and focus largely on tax evasion. Some of the other sources, such as the World Bank and the World Inequality Database, provided necessary context about the nature of tax evasion in OGP member countries. We also found that many of our contextual indicators were related to gender equity, whereas the theme of gender equity was absent among the more actionable indicators. So while the FSI and OECD indicators are certainly important for our database, we acknowledge their overwhelming majority in our database and advocate for other organizations to not only expand their breadth of data collection to include metrics related to tax evasion but also examine how the failure to effectively capture tax revenue affects gender and income inequality.
Recommendations for OGP
The results of our data census revealed that while the existing data on tax evasion is encouraging, it is not collected by a diverse group of organizations and it does not necessarily capture all facets of tax evasion. Moving forward, we recommend OGP to encourage its members to:
- Establish more initiatives to combat tax evasion that involve citizen participation and accountability,
- Encourage organizations at both the grassroots level and internationally to collect and publish more data on tax evasion, and
- Consistently monitor how policy levers that mitigate tax evasion affect gender equity.
With a more robust set of data, OGP can identify leaders in key performance areas and craft more precise action plans to help member countries combat tax evasion.
