Policy Research Working Paper 11090 Effective Tax Rates, Firm Size and the Global Minimum Tax Pierre Bachas Anne Brockmeyer Roel Dom Camille Semelet Economic Policy Global Department & A verified reproducibility package for this paper is Development Research Group available at http://reproducibility.worldbank.org, March 2025 click here for direct access. Policy Research Working Paper 11090 Abstract This paper documents new facts on corporate taxation and in all sample countries, over a quarter of top firms face the revenue potential of corporate minimum taxes, leverag- an effective rate below 15 percent, challenging the simple ing firm-level tax returns from 16 countries. First, effective tax haven versus non-haven dichotomy. Third, a simple tax rates follow a humped-shaped pattern with firm size: 15 percent domestic minimum tax for the top 1 percent small firms benefit from reduced rates, while large firms firms could raise corporate taxes by 14 percent on average take up tax incentives, leaving mid-sized firms with the across countries, absent behavioral responses. In contrast, highest effective rates. On average, the effective tax rate for the global minimum top-up tax would only raise a quarter the largest 1 percent of firms is 2.2 percentage points lower of this revenue due to its generous deductions and smaller than the average effective tax rate for the top decile of firms. number of firms in scope. Second, although statutory tax rates are above 15 percent This paper is a product of the Economic Policy Global Department and 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 abrockmeyer@worldbank.org and pbachas@worldbank.org. A verified reproducibility package for this paper is available at http://reproducibility.worldbank.org, click here for direct access. RESEA CY LI R CH PO TRANSPARENT ANALYSIS S W R R E O KI P NG PA 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 Effective Tax Rates, Firm Size and the Global Minimum Tax Pierre Bachas, Anne Brockmeyer, Roel Dom & Camille Semelet* JEL Classification: H25, H32, O23 Keywords: corporate tax, tax expenditures, effective tax rate, global minimum tax, multinationals * Pierre Bachas: World Bank and EU Tax Observatory pbachas@worldbank.org. Anne Brockmeyer: IFS, UCL, World Bank and CEPR, abrockmeyer@worldbank.org. Roel Dom: University of Antwerp roel.dom@uantwerpen.be. Camille Semelet: University of Munich, ifo Institute semelet@ifo.de. We gratefully acknowledge funding from the Global Tax Program, the Knowledge for Change Trust Fund and the Innovations in Tax Analytics Program at the World Bank. Bachas’ time was partly funded by a grant from the Norwegian Agency for Development Cooperation (NORAD, grant no. QZA-22/0011). Brockmeyer’s time was partly funded by UKAID from the UK government through the Centre for Tax Analysis in Developing Countries (TaxDev) and a UKRI Future Leaders Fellowship (grant reference MR/V025058/1). The findings, interpretations, and conclusions expressed in this work do not reflect the views of the World Bank, its Board of Executive Directors, or the governments that they represent. All errors are our own. We thank the revenue authorities in all 16 countries as well as several World Bank colleagues for an excellent collaboration. In particular, we thank Antonio Giraldi, Christoph Ungerer, David Fernando Pineda Pinto, Denis Mukama, Dora A. Vivas Perez de Gonzalez, Enamorado Irias, Fernando Pelaez Longinotti, Innocente Murasi, John Karangwa, Jose C. Bermudez Sanchez, Jose F. Suriano Buezo, Marc Schiffbauer, Maria Rodriguez Quezada, Milan Lakicevic, Naomi Alexander, Pedro Rafael Zuniga Figueroa, Roldan Manuel, Sol Mascarenhas, Gabriel Oqueli, Theonille Mukamana, David Suarez Castellanos and Juan Camilo Obando. We are grateful to Santiago Cesteros and Rafael Vilarouca for excellent research assistance, as well as to Alipio Ferreira, Adrienne Lees, Giulia Mascagni, Kyle McNabb, Vedanth Nair, Edris Seid, Ben Waltmann for their support with the analysis in selected countries. We thank Flurim Aliu, Annette Alstadsaeter, Katy Bergstrom, Michael Best, Michael Devereux, Petr Jansk´ y, Claus Thustrup Kreiner, Rebecca Lester, John Loeser, Helen Miller, Vedanth Nair, Oyebola Okunogbe, Steven Pennings, Imran Rasul, Mahvish Shaukat, Joel Slemrod, Gabriel Zucman and Davide Zufacchi as well as seminar/conference participants at the EBRD, ifo Institute, IFS, IIPF, PSE, TaxDev, University of Munich and the World Bank for helpful comments. 1 Introduction The corporate income tax remains a key source of government revenue, particularly for developing countries where it collects up to 20% of tax revenue. Yet, many firms, and especially large corpo- rations, have paid less and less tax on their profits over the past two decades (Garcia-Bernardo et al., 2022). Profit shifting to tax havens has increased, and governments have competed to attract firms by offering generous tax incentives, such as reduced rates, income exemptions and tax cred- its (Tørsløv et al., 2020). We study which firms benefit from these tax incentives, and what the implications are for a minimum corporate tax. To answer these questions, we assemble a unique dataset of firm-level corporate income tax returns for 16 countries. We document several new facts on corporate taxation, which are broadly consistent across countries. First, tax incentives primarily benefit the largest firms: the effective tax rate (ETR) for the largest 1% of firms by revenue is on average 2.2 percentage points lower than the ETR for other top decile firms in the same country. Second, over a quarter of the largest firms face an ETR below 15%, even though our sample only contains moderate to high tax countries, with statutory rates of at least 15% in all countries, and at least 25% in 13 of the 16 countries. Thus, a minimum corporate tax of 15% would impact firms in most countries and not just in tax havens. Our simulation of a simple domestic minimum tax on the top 1% of firms, abstracting from behavioral responses, suggest that the tax could raise corporate tax revenue substantially. Third, by carefully modeling an application of the global minimum tax rules (Pillar II) for five countries, we show that direct revenue gains are modest in most countries, due to the small number of firms in scope and the generous deductions for payroll and assets (carve-outs). Our data covers 16 countries that are heterogeneous in their size and development, includ- ing countries in Africa (Ethiopia, Eswatini, Rwanda, Senegal, South Africa and Uganda), Latin America (Colombia, Costa Rica, Dominican Republic, Ecuador, Guatemala, Honduras, Jamaica, and Mexico), and Europe (Albania and Greece). These countries are not tax havens (the median country’s statutory rate is 27%, the lowest rate is 15%), and most do not headquarter large multi- nationals.1 Our administrative data include all firms filing corporate taxes, hence capturing the 1 According to CbCR 2018 aggregate data, Greece has 18 domestic MNE groups with consolidated revenue above EUR 750 million, Mexico has 69, and South Africa 58. 1 entire formal economy and presenting a much larger sample than survey or financial data, which are sparse in developing countries.2 To build an effective corporate tax rate using comparable sources and definitions, we harmo- nize key variables across countries. We define a firm’s effective tax rate as its corporate tax liability divided by its profits. Profits (or losses) equal revenue minus material, labor, operational, depreci- ation, and financial costs. Hence, to calculate profits, we deduct standard production and financial costs from revenue, but do not deduct country-specific incentives that affect the tax base or tax rate. Differences between the ETR and the statutory rate are thus due to policy-driven tax expen- ditures, which we classify in five categories: income exemptions, special deductions, preferential tax rates, tax credits and loss carryforward.3 From the micro data we confirm that total corporate tax expenditures are large (1.0% of GDP on average in our sample) and approximately match the official macroeconomic aggregate. In the first part of our analysis, we show that firm size, proxied by total revenue, is a key determinant of ETR dispersion within countries. In all countries, tax expenditures accrue dispro- portionately to small and medium firms (SMEs), and to the largest firms, such that upper mid-sized firms face the highest tax rates. ETRs peak around the ninth decile of firm size, and then decline at the top in most countries, remaining flat in the other countries. The rising slope of ETRs over the lower part of the firm-size distribution is explained by a higher propensity for smaller firms to register losses, and, once we control for losses, by reduced statutory tax rates. The ETRs of the largest 1% of firms are 2.2 percentage points below those of other top decile firms (2.6 percentage points below for the top 0.1%). This is mainly explained by their take-up of tax credits and of income exemptions, not by differential patterns of loss-making or by reduced statutory tax rates. The lower ETRs among top 1% firms are also partly explained by location, which might capture tax advantages of special economic zones, whenever those are not directly observable on the tax return. The drop in ETRs at the top holds across industries, when computing ETRs over several years, and with alternative definitions of firm size. 2 Tax return data miss the informal sector, but still have much wider coverage than financial data, which in most developing countries cover only the largest firms. Survey data often have poor coverage among the largest firms. Firm censuses, where they exist, cover all firms but rarely contain data on tax liabilities and profits. 3 Some countries in our sample do not permit loss carryforward. At the same time, loss carryforward is not always con- sidered as a tax expenditure category. We hence document the robustness of our results to excluding loss carryforward from the list of tax expenditures. 2 In the second part of our analysis, we assess the scope and revenue potential of a minimum corporate tax. Reforming tax expenditures is often politically and legally difficult. A minimum tax can be a second-best tool to start leveling the playing field of taxation. Given that we observe a clear drop in ETRs at the top of the firm-size distribution, we begin by simulating the mechanical revenue gains from applying a 15% domestic minimum tax on the top 1% firms in each country, not allowing any exemptions. Many of the largest firms in each country face an ETR below 15%: across the 16 countries, the mean (median) share of top 1% firms paying less than 15% in taxes is 28.2% (27.8%). Conditional on paying an ETR below 15%, the mean (median) ETR is only 3.1% (2.4%). As a result, we find that a domestic minimum tax on the largest 1% of firms could raise revenue substantially: absent behavioral responses, CIT revenue could increase by 14.2% (7.8%) in the average (median) country. Yet, the international community has now enacted a much more complex global minimum tax (Pillar II of the OECD/G20 Inclusive Framework). The global minimum tax (GMT) collec- tion rules incentivize all countries to tax the profits of multinational firms in scope at a rate of at least 15%, as the firm’s affiliates would otherwise be taxed by another country.4 The GMT rules markedly differ from our hypothetical domestic minimum tax: the GMT only applies to sub- sidiaries of large MNEs (with global revenue above 750 million EUR), permits generous carve-outs (deductions) for tangible assets and payroll, applies at the group level (enabling consolidation of profits across firms in a group), and does not consider refundable tax credits as a tax incentive. In the final part of the paper, we carefully model the GMT rules that apply to firms of foreign ultimate parent entities (UPEs) in five countries where the data permits it, and contrast the revenue gain from the GMT to the simple domestic minimum tax. We do not simulate potential revenue gains that would arise from a top-up tax collected from activities of domestic UPEs abroad. First, this would require extensive data on MNE activities across jurisdictions. Second, only three coun- tries in our sample—Mexico, Greece, and South Africa—have domestic UPEs that are in scope of the GMT and could, in principle, collect top-up taxes through the Income Inclusion Rule (IIR) if their MNE groups own low-taxed subsidiaries abroad. Few estimates of the revenue gains from the GMT exist, in part because the Country-by- Country Reporting (CbCR) micro data on profits of MNEs is rarely available. We fill this gap 4 They would be taxed either by the headquarter country or by another country in which the firm has affiliates. 3 with two novel methods. We first identify via Orbis the subsidiaries of in-scope MNEs and their group structure. For our five countries, the number of in-scope subsidiaries in Orbis is close to those in the publicly available aggregate CbCR data. The list of Orbis subsidiaries is then merged with CIT returns, with a 65% match rate. The imperfect match implies that our first method yields a lower bound on the GMT revenue potential. To address this, we develop a second method: we build a list of foreign-owned firms by combining the Orbis-matched data and foreign ownership data from the tax administration. We select the N foreign-owned firms with highest top-up profits, where N is the number of firms reported as being in scope in the aggregate CbCR data for our countries. By selecting the foreign-owned firms with highest top-up tax, we estimate an upper bound on the revenue potential of the GMT. Our estimated revenue gains for the GMT are modest, and much smaller than those for the simple domestic minimum tax in the same countries.5 Costa Rica could experience a sizable 22.6% increase in CIT revenue from the GMT, but revenue gains in Greece, Honduras, Jamaica and South Africa would be much smaller, ranging from 0.3% to 6.9% of CIT revenue. The lower and upper bound estimates are fairly close to each other. What explains the lower revenue potential of the GMT? First, the firms in scope of the GMT are fewer in number and smaller than top 1% firms, as the GMT excludes domestic standalone firms. Second, the GMT tax base is narrower: it allows for group consolidation and the deferral of tax assets. In addition, the GMT allows firms to deduct a share of their payroll and tangible assets, which considerably reduces their tax base. Our study only offers a positive description of corporate taxation. We do not model behav- ioral responses, in part due to the large uncertainty around firms’ and countries’ responses to the fast changing global environment. The simple domestic minimum tax for large firms presents a benchmark for the mechanical revenue potential of an uncoordinated minimum tax. Yet, beyond differences in scope and tax base definition, the GMT is a coordinated effort to reduce profit shift- ing incentives, which could raise tax revenue in all non-haven countries. In this paper, we do not consider changes to firms’ profit shifting incentives, which would mainly benefit large high-income countries (see Ferrari et al. 2022; Devereux 2023). Further, while our paper shows that tax incen- tives disproportionally benefit the largest firms, it does not take a stance on welfare: it is possible 5 For Greece and South Africa where domestic UPEs could generate tax revenue from the IIR, those estimates should be considered as a lower bound. 4 that the ETR-firm-size patterns we document correspond to governments’ objectives. The paper is organized as follows. Section 1.1 places our work in the literature. Section 2 describes our data and method. Section 3 documents the ETR-firm-size relationship and analyzes the drop in ETRs at the top. Section 4 examines the potential revenue gains from a domestic minimum tax. Section 5 models the global minimum tax in selected countries. Section 6 concludes with a discussion of policy implications. 1.1 Related Literature Effective Tax Rates Our work relates to the literature at the intersection of economics and ac- counting that calculate firms’ effective tax rates (ETRs) (Devereux and Griffith, 1998, 2003). ETRs capture how corporate tax burdens differ from what is implied by statutory tax rates. The litera- ture distinguishes between forward-looking ETRs (considering future tax burdens based on the tax code) and backward-looking ETRs (considering realized burdens based on taxes paid). To cal- y 2022 for a summary), culate backward-looking ETRs, existing studies (see Table A.1 and Jansk´ mainly use financial accounting data, which has limitations due to differences between financial and tax accounting rules (Graham et al., 2012). Moreover, the coverage of financial data is partial, especially for low and middle-income countries. Our work advances this literature by using admin- istrative corporate tax returns, enabling a more precise measure across a diverse set of countries. The OECD recently started publishing aggregate data from country-by-country reporting (CbCR), e et al., 2022; OECD, providing an alternative to estimate ETRs of MNEs and their affiliates (Barak´ y, 2024).6 CbCR data is organized in bilat- 2020; Hugger et al., 2023; Garcia-Bernardo and Jansk` eral matrices by ultimate parent entity (UPE) and affiliate, containing jurisdiction-level informa- tion on MNE activity, profits, and taxes. Although their publication is a major step forward, CbCR data, by construction, only contain affiliates of large MNEs and the aggregate data does not permit a granular examination of firm level ETRs. Further, the data suffers from two well-documented issues: a lack of inter-temporal adjustments and double-counting of profits, which can bias ETR estimates (Blouin and Robinson, 2020). The release of micro CbCR data and addressing the afore- mentioned issues could permit, at term, a precise estimation of the effective taxation of MNEs. 6 Under BEPS Action 13, large MNEs are required to file a country-by-country (CbC) report with their global allocation of income, profit, and taxes paid in countries where they operate. This report is shared with tax administrations. 5 Firm Size and Tax Minimization By unpacking the determinants of the heterogeneity in ETRs across the firm size-distribution, we contribute to the literature on tax minimization. Tax avoidance, notably via profit shifting, and tax evasion have received most attention (see Slemrod 2019).7 This literature tends to find that tax minimization varies with firm size: evasion rates decrease with firm size (Basri et al., 2021; Bachas et al., 2019; Best et al., 2022), while avoidance increases (Gumpert et al., 2016; Davies et al., 2018; Wier and Erasmus, 2023). We are the first to document that the availability and take-up of tax expenditures also exhibits a size dependence, with a marked drop in ETRs at the top, in several countries and with comparable data.8 The limited work on tax expenditures points to their importance: Egger et al. (2020) show that larger firms’ bargaining power enables them to negotiate tax advantages from governments. Garcia-Bernardo et al. (2022) find that only a moderate share of the recent fall in global ETRs is due to profit shifting, with the remainder due to domestic tax policy such as tax expenditures. Klemm (2010) shows the relevance of tax expenditures in selected low-income countries. Minimum Taxes Finally, we contribute to the growing literature on corporate minimum taxes, and the revenue effects of a Global Minimum Tax (GMT) on MNEs. Most studies use theoretical frameworks to assess the impact of the GMT (Devereux et al., 2020; Cobham et al., 2021; Hin- alez Cabral, 2022; Ferrari et al., 2022; Janeba and driks and Nishimura, 2022; Hanappi and Gonz´ Schjelderup, 2023; Hebous and Keen, 2023; Haufler and Kato, 2024; Hines Jr, 2024). A few stud- e et al., 2022; ies model the GMT rules in detail and estimate revenue gains (OECD, 2020; Barak´ Hugger et al., 2024). We uncover large mechanical revenue gains from a 15% domestic minimum tax on the top 1% firms. Yet, our estimates for the GMT are much lower, and below those based e et al. (2022) estimate that European countries could raise CIT revenue by on CbCR data: Barak´ 16%, while Hugger et al. (2024) find gains of 6% to 8% of CIT revenue, with two-thirds of those gains attributed to direct top-up taxation, and the remaining third due to reduced profit shifting. However, given the limitations of the aggregate CbCR data discussed above, these studies may 7 The extensive literature on profit shifting can be categorized into micro-studies that identify specific channels, such as transfer mis-pricing for goods and services, location of intangible assets and patents, debt shifting (e.g. Desai et al. 2004; Dischinger and Riedel 2011; Karkinsky and Riedel 2012; Buettner et al. 2012; Wamser 2014; Egger and Wamser 2015; Gumpert et al. 2016; Cristea and Nguyen 2016; Davies et al. 2018; Beer et al. 2020; Bilicka 2019; Clifford 2019; Wier 2020; Liu et al. 2020; De Mooij and Liu 2020, 2021; Hebous and Johannesen 2021); and macro-studies assessing aggregate profits shifted (e.g. Tørsløv et al. 2022; Wier and Zucman 2022; Fuest et al. 2022). 8 Other work studies the relationship between ETRs and firm size in a single country, summarized in Table A.1. 6 overestimate the tax gains of the GMT. By drawing on micro tax data, we can apply firm-level ad- justments with respect to carve-outs, the de minimis rule, and prior losses. Our estimated revenue gains suggest that previous studies may be optimistic. 2 Data and Methodology 2.1 Data We assemble a novel dataset of corporate tax records from 16 countries, listed in Table 1. The dataset includes countries in Africa, Latin America, and Europe, and covers a wide range of income levels (from Ethiopia with a GDP per capita of 700 USD to Greece with a GDP per capita of 19,000 USD) and population sizes (from Eswatini with 1.2 million inhabitants to Mexico with 128 million inhabitants). Each country’s data includes all corporate tax returns in the country, over a five-year span on average. In the main analysis, we focus on the latest pre-COVID cross-section, typically 2019, but for some countries we use data from earlier years.9 Administrative tax data record firms’ reported taxable income, costs, and all tax exemptions which allows for a breakdown of the tax burden. Unincorporated firms and firms filing under simplified regimes are excluded, as their tax treatment differs across countries, and information on profits is often missing. Appendix C details each country’s tax system and our treatment of any special tax regimes. Although administrative micro tax data are increasingly available (Pomeranz and Vila-Belda, 2019; Mascagni et al., 2016), our study is one of the first to use such data across many countries. The number of firms in each dataset correlates with a country’s size and income level, ranging from 5,700 firms in Senegal to 404,000 firms in Colombia. The share of profitable firms varies from around 50% in Mexico to 85% in Colombia. By definition, the data include formal incorporated firms filing the corporate income tax. Table A.2 shows that, compared to firm registries (which typically include unincorporated firms), the number of firms captured in CIT ranges from less than 10% in the lowest-income countries (e.g. Senegal, Rwanda) to 80% in richer countries (e.g Albania, Greece). Yet, the ratio of total revenues in the micro data to the GDP of the country is typically high (median country at 95%), although it is lower in the lowest-income countries. This 9 We decided to use 2019 as our baseline, instead of 2020 or 2021, to prevent the data from reflecting the effects of the COVID-19 pandemic and its tax relief measures, which we study in a related paper (Bachas et al., 2024). 7 reflects the fact that, even in countries with a vast unincorporated sector, production is concentrated within large corporations.10 Our sample of countries features moderate to high statutory corporate tax rates. The lowest statutory rate is 15% (in Albania), and in most countries, the rate is between 25% and 30%. Given that the global minimum tax was set at 15%, we would not expect the minimum tax to have a direct impact on these countries in the absence of generous tax incentives. Although countries were not randomly chosen, our sample appears representative of non-tax- haven low and middle-income countries in terms of size, GDP per capita, and statutory tax rates. These specific countries are included because they reached out for advice, or because we worked with their tax administration on other analyses. Inclusion in the sample hence reflects either a proactive interest by countries or an openness to make micro data available for analytics. The absence of Asian countries hints at more limited access to confidential tax data in these countries. 2.2 Methodology Objective We aim to compute ETRs that are comparable across countries with different tax systems and tax reporting requirements. Thus we select concepts from the CIT returns that are consistently used across countries. To the best of our ability we aim to distinguish variables that measure ‘standard’ deductions—allowed in all countries and for all firms, and reflecting economic costs—and variables measuring tax expenditures, which can be country and firm-specific. Accounting Concepts We consider the concepts that can be consistently measured across coun- tries’ tax returns, and the relationships between them in Figure 1. Total revenue is composed of sales plus other incomes (e.g. non-operating incomes, rents, interests). The deductions that firms typically subtract from their revenue to calculate their profits include the cost of material inputs, labor, and capital costs, as well as financial costs, and depreciation of capital. We call the differ- ence between revenues and costs “profit” (or loss). This concept is not always directly reported in tax returns as a line item, but can always be reconstructed. We consider that this measure of profit is the best proxy for economic profit that can be constructed from tax return data. Our profit measure differs from taxable profit. To derive taxable profit, one needs to exclude 10 An alternative view is that these countries’ GDP fails to fully incorporate informal firms’ activity. 8 tax-exempt incomes; reintegrate non-tax-deductible costs; apply tax incentives, capital allowances, and other deductions; and account for loss carryforward. This yields the taxable profits, i.e. the tax base. Multiplying the tax base by the statutory tax rate yields the gross tax liability.11 Tax credits are then deducted from the gross tax liability, to obtain the net tax liability. Computing Effective Tax Rates We define a firm’s ETR as the net tax liability divided by the profit. The numerator, net tax liability, is the tax due net of any after-tax deductions and credits, but ignoring advanced payments and withholding of taxes already paid. Hence, any deduction that is subtracted either from the tax base (proxied in our methodology by profit) or from the gross tax liability is taken into account as a tax expenditure that lowers the effective tax burden that a firm faces. By taking the ratio of the net tax liability over profit, we obtain an ETR which reflects the gap between the statutory and effective tax rate due to tax expenditures.12 This ETR measure is transparent and arguably comparable across countries. To construct it we distinguish two concepts: standard deductions, which should reflect economic costs (material, la- bor, operating costs, and depreciation), and tax expenditures, which should reflect policy choices. Standard deductions are removed from the tax base to calculate profit (our ETR denominator) but tax expenditures are not removed. Drawing the line between standard deductions and tax expen- ditures requires choices. For example, the time schedule of capital depreciation can vary across countries and asset classes, raising the issue of whether accelerated depreciation should be con- sidered a standard deduction or a tax expenditure (we decided on the latter). Standard deductions tend to be homogeneously defined, while the concepts included under tax expenditures are country- specific: e.g. the definition of non-deductible costs and exempt incomes, investment incentives, reduced tax rates as a function of sales or activity.13 We categorize tax expenditures into five cat- egories: income exemptions, special deductions, reduced rates, tax credits and loss carryforward. The latter are not permitted in several sample countries, and when permitted, regulations differ on the amount and duration allowed. We show robustness to their exclusion as a tax expenditure. 11 We validate the data cleaning and variable construction process by ensuring that when we divide the gross tax liability by the tax base we obtain the statutory tax rate. 12 Figure A.1 shows an example of how these concepts are observed on Rwanda’s tax return form. 13 Previous studies used a range of denominators, including EBIT (Adhikari et al., 2006); EBITDA (Laz˘ ar, 2014); gross income excluding variable costs (Nicod` eme, 2002); and measures with ad-hoc adjustments of taxable income (Wu et al., 2012); see Table A.1. Our ETR measure allows all costs, including depreciation, operational and financial expenses, to be deducted from the denominator. 9 Tax Evasion A potential caveat is that profit (our denominator) may deviate from true economic profit due to tax evasion and avoidance. Under imperfect enforcement, firms may under-report sales, inflate costs, and shift profits abroad, which lowers reported profit and the net tax liability. Our measure does not capture the extent to which firms lower their tax burden through these chan- nels: it is an upper bound on the true ETR that firms face on their realized (instead of reported) profits. Relatedly, firms’ decision to use tax expenditures could interact with tax evasion and avoidance decisions.14 The data do not allow us to investigate these issues further, as evasion is unobservable, and we cannot estimate shifted profits. Instead, we capture the importance of tax expenditures as a share of firms’ reported profit, holding evasion and profit-shifting fixed.15 3 Effective Tax Rates and Firm Size 3.1 Aggregate Tax Expenditures We construct firm-level effective tax rates (ETRs) with the data and methods described in section 2. The last two columns of Table 1 show for each country the top statutory tax rate, the average ETR for all firms (imputing a zero ETR for loss-making firms), and the average ETR for profitable firms. We use the difference between the top statutory rate and the firm-specific ETRs multiplied by their profits to compute firm-level tax expenditures. By aggregating across firms, we obtain total corporate tax expenditures, which we express as a share of GDP. This measures the forgone revenue due to CIT expenditures, absent behavioral responses. Figure A.2a displays our aggregate CIT expenditure estimates, ranking countries by GDP per capita. For our sample countries, tax expenditures represent 1.04% of their GDP on average. Albania, Ethiopia and Guatemala’s CIT expenditures in the micro data are close to zero, while all other countries’ tax expenditures are over 0.5% of GDP. We also include data from an additional 64 countries, using the Global Tax Expenditure Database which assembles data from official tax 14 For example, tax planning could impact both international profit shifting and tax incentives take-up. By computing an ETR measure based on profit (after the deduction of all standard production costs), we do not consider tax avoidance taking place before this stage, for example through capital depreciation rules. 15 Dyreng et al. (2017) show that ETRs have decreased in the same way for domestic and multinational firms, suggest- ing MNEs’ ability to shift profits may not be a first-order difference between domestic firms and MNEs. 10 expenditure reports (Redonda, von Haldenwang and Aliu, 2022).16 The average CIT expenditure in this extended sample of 80 countries is 0.68% of GDP, slightly lower than in our sample. Figure A.2b shows the correlation between our CIT expenditures estimates and those from government reports. The correlation is high but not perfect: our method yields higher estimates on average, especially for African countries. We note that governments’ methods to compute CIT expenditures differ across countries: for example, reduced rates are not counted as expenditures everywhere—they tend to be counted in high-income countries but rarely in developing countries. In addition, tax expenditure reports in low-income countries often do not cover all special tax regimes, sometimes explicitly stating this exclusion.17 The perception of data reporting weak- nesses is reinforced by the observation that poorer countries display lower tax expenditures than richer countries, which could reflect incomplete measurement. Our CIT expenditure estimates are arguably more comparable across countries. 3.2 Country-Level ETR-Firm-Size Curves Which firms benefit from the sizable tax expenditures measured above? We rank firms based on their revenue within their country, and assign each firm to a percentile of revenue from 1 to 100.18 To study the largest firms, we further separate the top 1 percentile of firms into five bins, each representing 0.2% of firms. We then measure the average ETR across firms in each quantile. We compute average ETRs by quantiles for two samples of firms. The first considers all firms, includ- ing zero-profit and loss-making firms to which we assign a zero ETR. The second is restricted to profitable firms, using the same revenue cutoffs between quantiles as in the first sample. Figure 2 shows the pattern of ETRs in each of the 16 countries, first for all firms, including loss makers. We rank countries by their top statutory tax rate (STR), ranging from 15% for Albania to 35% in Colombia, with a median at 29%. We show the average ETR in each firm-size quantile and a polynomial fit capturing the ETR-firm-size relationship. The gray shaded area corresponds to the top 1% of firm size, which is graphically expanded to zoom in on the largest firms’ ETRs. We observe two main patterns. First, in every country, the ETR rises between the first and ninth 16 Following international standards, tax expenditures should be computed yearly. In practice, developing countries do not systematically produce these reports, and computation methods vary significantly across countries. 17 For example, Table 1 of Rwanda’s 2021 tax expenditure report prepared by the Finance Ministry qualifies its income tax expenditure estimate by stating that it “excludes some tax expenditures that are not currently measurable”. 18 Table A.3 shows the number of firms per percentile for each country. 11 decile of firm size. Second, the relationship between firm size and ETRs always flattens and often reverses once we reach the top decile. In most countries, the largest firms (top 1%) pay a lower ETR than other top decile firms. In tandem, these patterns produce a humped-shaped relationship between firm size and ETRs in most countries, such that firms at around the ninth decile of the size distribution face the highest ETR. Which factors explain the relationship between ETRs and firm size? We first analyze the role of loss-making firms (to which we assigned an ETR of zero) and of reduced statutory tax rates, which benefit smaller firms in several countries. To examine this graphically, we restrict the sample to profitable firms only, and compute the firm-specific gap ET Ri − ST Ri , where ST Ri is the firm’s statutory tax rate. Depending on the country, this may be a function of its revenue or its profits. We then take the average of the firm-specific tax rate gaps at each quantile of the size distribution, in each country, keeping the revenue quantiles fixed based on the full population. Figure 3 plots the average tax rate gaps as a function of firm size quantiles.19 The ETR-firm- size relationship is much flatter than in Figure 2: in 11 of the 16 countries the relationship is flat between the first and ninth decile. The flattening of the slope is due in part to the fact that smaller firms are more likely to report zero or negative profits (and be assigned a zero ETR), and partly due to reduced STRs for small firms in several countries (Albania, Costa Rica, Ecuador, South Africa). Figure A.4 shows the large reduction in the share of unprofitable firms along the firm size distribution. At the top of the distribution, however, the ETR still drops in most countries: with the exception of the lowest tax countries (Albania, Guatemala and Jamaica, plotted first) plus Ethiopia, profitable firms in the top 1% pay a lower ETR than other top decile firms. The four countries without a drop in the ETR at the top display a flat relationship.20 3.3 Robustness of the ETR-Firm-Size Relation The ETR-firm-size pattern we document is not driven by methodological choices: the pattern repli- cates within broad economic sectors, with a multi-year time horizon to define the ETR, and with alternative firm size measures. To succinctly display several robustness dimensions, we summarize 19 Figure A.3 plots the same firm-size-ETR patterns as in Figure 2 for profitable firms only. 20 Figure A.5 shows how the average firm-specific tax gaps and average ETR correlate with country characteristics and macroeconomic variables in the 16 countries; in particular tax gaps are larger in high STR countries. 12 individual country patterns by constructing a synthetic average country: we take the average of the ETR at each quantile across countries, weighing countries equally. Figure 4a displays the aver- age ETR across the firm-size distribution (for profitable firms) for the 16 sample countries. The synthetic average country repeats the humped-shaped pattern of rising ETRs between the first and ninth decile, and the marked fall in ETRs at the top, observed in the majority of sample countries. Figure 4b shows the ETR-firm-size relationship after assigning firms to the four main sectors of activity: agriculture (primary), industry and construction (secondary), retail, and services. In each sector, we observe a humped-shaped pattern similar to the aggregate pattern. Industrial firms show the largest drop in ETRs at the top, but the reduction is also large for firms in services. Retail and agricultural firms display smaller ETR reductions at the top. Figure 4c shows ETRs computed over multiple years (ranking firms based on revenue from the most recent year). Multi-year ETRs are constructed by taking the sum of tax liability over several years and dividing it by the sum of profits over the same years. We can construct two-year ETRs in all countries, and five-year ETRs in 12 countries. The sample of countries thus slightly changes with the time horizon we consider, but results are consistent across time horizons, and very similar to our main results, even though profits and losses are now averaged over the period. Each multi-year ETR measure displays a clear drop at the top percentile of the size distribution. Our results are also robust to alternative definitions of firm size, such as payroll and total assets. Figure A.6 shows that the firm-size-ETR patterns look very similar when using these alternative size measures, even though the different size measures are imperfectly correlated. Finally, we address the concern that larger firms, which are often subject to public financial reporting rules, report profits more accurately to ensure consistency between their financial and tax statements (Badertscher et al. 2019, Hoppes et al. 2020). This behavior could mechanically create a drop in ETRs at the top of the distribution. While we do not have access to firms’ financial statements, we can show that the ratio between profits and assets is flat at the top of the firm-size distribution, both within the top decile and across the top three deciles (Figure A.7). This suggests that differential misreporting below the top of the firm-size distribution is not a first order concern. 13 3.4 Tax Reductions for the 1 Percent Largest Firms Firms in the top 1% of the size distribution are of systemic importance: in our sample they account for 54% of revenue reported in the CIT declarations, 59% of profits, and 56% of corporate taxes on average (Table A.4). We thus examine in a regression setting the role of firm characteristics and of the different types of tax expenditures in accounting for the drop in ETRs at the top. Due to the nonlinearity of the ETR-firm-size pattern, we restrict this analysis to firms in the top 10% of size in each country, and consider the impact on the ETR of belonging to the top 1%. In the median country by number of CIT returns, 23,000 firms file the CIT, so the top 1% corresponds to 230 firms (see Table A.3).21 Using the most recent cross-section of profitable firms, we estimate the following model: T op1 ET Ri = γ0 + γ1 Di + γk Xk,i + ϵi , (1) T op1 where ET Ri is the effective tax rate of firm i, and Di is a dummy equal to 1 if firm i belongs to the top 1% of revenue in its country. The coefficient γ0 measures the average ETR for firms between the 90th and 99th size percentile, and γ1 measures the difference in the ETR between the top 1% firms (above the 99th percentile) and to other top decile firms. Xk,i is a vector of firm-specific variables including firm characteristics and dummies for the different types of tax expenditures that take value one if a firm reports a non-zero amount.22 We estimate equation 1 separately for each country and display the average of the γ1 coeffi- cients across countries in Table 2 (country-specific results are shown in Table A.7). Column 1 does not include any controls: on average across countries, firms in the top 1% of the size distribution pay 2.2 percentage points less in taxes than other top decile firms. The coefficient on the top 1% dummy is negative in 14 countries, and statistically significantly so in 10 out of 16 countries.23 Column 2 controls for firm characteristics including sector, location, firm age, and foreign 21 We also run analyses for the bottom 90% of the firm-size distribution, where we regress the ETR on the percentile of revenue to account for the increasing trend of ETRs with firm size. Results are shown in Table A.5 and confirm the importance of reduced tax rates for the progressivity of the ETR, as discussed in Section 3.2. 22 We prefer comparing firms in the top 1% to other firms in the top decile rather than estimating the ETR-firm- size gradient across quantiles within the top decile, as it is transparent and reduces the need for functional form assumptions. We show robustness to other specifications in Table A.6. 23 Table A.6 shows that the ETR advantage at the top rises if we expand the estimation window (comparing the top 1% to the top 20%) and keeps on rising within the top 1% (the average coefficient on the top 0.1% dummy is 2.6). 14 ownership where available. Columns 3 to 7 control for different types of tax expenditures one by one: reduced tax rates, exempted incomes, special deductions, re-timing (i.e. loss carryforward), and tax credits. The two most important individual factors for the ETR differential at the top 1% are tax credits and firm characteristics. On average the coefficient on the top 1% dummy shrinks by over 30% when either is controlled for (columns 2 and 7).24 Firm characteristics (sector, location, and age) might capture tax provisions that are not listed on the tax form, such as special economic zones or support for young firms. Exempt income also explains some of the variation in ETRs at the top of the size distribution (column 4), followed in decreasing order of importance by special deductions (column 5), re-timing provisions (column 6), and reduced statutory rates (column 3). Column 8 controls for all explanatory variables at once: the average coefficient on the top 1% dummy is reduced by three-quarters, but some of the variation remains unexplained, and in four countries, the top 1% dummy coefficient is still significant. Our inability to fully explain the variation could be because we only control for each tax in- centive with a dummy variable, instead of using amounts. Taking into account amounts, i.e. fully decomposing the STR-ETR difference into its drivers, is not possible with our methodology and with the requirement that tax expenditure concepts are harmonized across countries. This is be- cause some tax expenditures are deducted from the tax base while others are deducted from the tax liability, and these concepts interact. A full decomposition of the ETR would require country- specific exercises and distract from our focus on the ETR-firm-size pattern across countries. Tax Expenditures Types and Rationales Table A.8 lists the existence of each tax expenditure type by country (from the tax form): tax credits—a key provision explaining the ETR drop at the top—and income exemptions are available in 11 and 12 of the 16 sample countries, respectively. In Albania and Ethiopia, where neither are available, the ETR-firm-size profiles are flat. Ideally, we would categorize tax expenditures based on their economic rationale. This is challenging with the line items on the tax return: many provisions have generic names and refer to laws stating multiple goals. To clarify the intention of tax expenditures, we collected each country’s official tax expenditure report. A qualitative analysis of individual tax provisions suggests that the main 24 Importantly, tax credits do not represent a compensation for taxes already paid elsewhere as shown by Figure A.8. 15 goals are to attract foreign direct investments, create job, and develop specific regions.25 These tax expenditures are often available via special economic zones (SEZs). Indeed, the largest provisions listed in the tax expenditure reports for Costa Rica, the Dominican Republic, Ecuador, and Hon- duras concerns tax credits and income exemptions for SEZ firms.26 Tax provisions for SEZs also seem important in African countries with large tax gaps (Eswatini, Senegal, Rwanda, and Uganda). 4 A Simple Minimum Tax on Corporations 4.1 Motivation Section 3.4 showed that the largest firms benefit from a lower ETR. Such ETR differentials by firm size can be detrimental to both production efficiency and equity. For policymakers desiring to level the playing field, the most direct approach would be eliminating tax credits that disproportionately benefit the largest firms. However, eliminating these tax credits is often politically, legally, and ad- ministratively difficult. A simpler policy that would achieve the same goal is to impose a minimum tax on profits for a subset of firms, e.g. firms in the top 1% of the size distribution. Such a policy is a second-best tool in that it does not directly address the root cause of differences in the ETR by firm size–e.g. the tax credits–but achieves a minimum taxation for targeted firms. The approach is similar in spirit to the global corporate minimum tax, but much simpler, as will become clear when we discuss the complex rules of the global minimum tax in Section 5. 4.2 Revenue Potential We model a simple 15% domestic minimum tax on profits, applied to all firms in the top 1% of size within each country. While the top 1% of size includes a higher share of MNEs than other firm-size percentiles, the majority of firms in the top 1% are actually domestic standalone firms (Table A.9, column 3). Applying a minimum tax broadly to include domestic firms and subsidiaries of MNEs seems desirable, both for tax competition and for efficiency reasons. We apply the minimum tax at the firm level (and not the group level), as information on groups is often missing, and rules on 25 Other aims of tax expenditures are to address externalities (environmental, health) and reduce employment discrim- ination, but these account for a much smaller share of GDP. 26 Not all firms in SEZs are required to file corporate taxes. Thus, some firms might be absent from our dataset altogether. Including these firms in the analysis would likely strengthen our results. 16 consolidation for tax filing vary across countries. Figure 5a, shows for each country the share of firms with an ETR below 15% among profitable top 1% firms, ranking countries in descending order of their statutory tax rates. Across countries, 28% of top firms on average face an ETR below 15% on average. The variation in this share is limited: the largest share of 46% is observed in Colombia and the lowest share of 6% in Guatemala, while the share is 25% in the median country. Figure 5b, shows that conditional on paying an ETR below 15%, firms’ reported ETRs tend to be very low: 2.95% in the median country. These two panels highlight that although our sample does not contain tax haven countries (the median statutory rate is 28.5%), many large firms still pay an ETR below 15%, which is often close to zero. This challenges the standard dichotomy between haven and non-haven countries: in practice, many moderate to high-tax countries provide generous tax expenditures, such that a substantial share of the largest firms pay almost zero taxes. As a consequence, a minimum tax could directly impact most countries, and push them to reconsider their tax incentive policies. To understand the revenue potential of a simple minimum tax, Figure 5c plots the share of aggregate profits that would be affected. This equals the profits of top 1% firms with an ETR below 15% over the profits of all firms filling the CIT. We then simulate the mechanical revenue collection from applying a 15% minimum tax to top 1% firms with an ETR below 15%, computed as their profits times the difference between 15% and their ETR.27 We assume that firm behavior remains unchanged.28 Figure 5d plots the hypothetical revenue collection of the simple domestic minimum tax, ex- pressed as a percentage of baseline CIT revenue from all firms. The simulations predict an average rise in corporate tax revenue of 29%, with heterogeneity across countries. In half of the countries, a 15% minimum tax would raise at least 10% more revenue, and in a third of countries, at least 20% more revenue. While the average ETR among top 1% firms with an ETR below 15% is quite homogeneous across countries, the exposed profits and revenue potential of a domestic minimum tax are more heterogeneous. The variation is in part driven by the dispersion of the profit distribu- tion: in small countries, a few large firms can account for a very high share of profits. 27 For loss-making firms, the additional tax payment due to the minimum tax is set to zero. 28 As previously discussed, if incentives to avoid taxes and shop around jurisdictions for lower rates are curbed by the GMT, real economic activity might decrease globally (and some profits could be reallocated from low-tax to higher-tax jurisdictions), although the low rate of 15% partly mitigates this concern. 17 In summary, we show that a quarter of the largest firms’ ETRs are below 15% in our sample of moderate to high-tax countries, and many firms pay almost zero tax. A simple 15% minimum tax on the top 1% of firms could raise revenue by more than 10% in half of these countries. However, this simple domestic minimum tax would not be considered as a “qualified domestic minimum tax” under the Global Minimum Tax rules, which are more complex and restrictive than our simulations. We now turn to modeling the detailed application of the pillar 2 rules in a subset of countries and compare the resulting top-up tax revenue to the revenue potential of the simple minimum tax. 5 The Global Minimum Tax (Pillar 2) 5.1 The Global Minimum Tax in Theory In 2021, 137 country-members of the OECD/G20 Inclusive Framework on base erosion and profit shifting (BEPS) reached an agreement to overhaul international tax rules starting in 2024. A key piece of this agreement is a 15% global minimum tax (GMT) on corporations.29 The global min- imum tax is applied to very large MNEs—those with a global consolidated revenue above 750 million EUR in two out of the last four years. Affiliates for these MNEs are labeled as in scope of the GMT. The GMT base is GloBE (global anti-base erosion) income, calculated at the group- jurisdiction level. A number of additional provisions affect a firm’s GMT liability, such as a re- duction in the GMT tax base linked to the real economic presence in a country (substance-based carve-outs), and the possibility to further lower the tax base by transforming existing tax credits into so-called qualifying tax credits that meet a set of criteria. We discuss these provisions in more detail below. The GMT is enforced via a set of interlocking collection rules. First, the country in which in- scope MNE affiliates are registered could apply the tax through a Qualifying Domestic Minimum Top-up Tax (QDMTT). If the source country does not apply the top-up tax, the country in which the MNE is headquartered could instead tax the under-taxed profits through the Income Inclusion Rule (IRR). And finally, if neither the country hosting the MNE affiliate nor the headquarter jurisdiction apply the top-up tax, the Under-Taxed Profits Rule (UTPR) would apply. This rule gives taxing 29 The OECD statement on the agreement can be read here and the GMT GloBE rules here. 18 rights to other countries in which the in-scope MNE has affiliates. These interlocking collection rules ensure that all countries have incentives to participate in the GMT by translating its provisions to domestic legislation. The status quo would imply foregoing tax revenue, as in-scope MNEs would be taxed anyways, with the revenue collected by other countries. Table A.10 shows that 39 countries already have adopted the GMT and 16 have prepared draft legislation. While most are high-income countries, ten low and middle-income countries either have implemented the GMT or are considering implementation. How might the global minimum tax impact our sample of countries? With the exception of Mexico and South Africa, these countries rarely headquarter large MNEs and are thus unlikely to e et al., 2022). Yet, benefit from claims to under-taxed profits via the Income Inclusion Rule (Barak´ the GMT could allow these countries to raise the ETRs on large firms to 15% through a QDMTT. If the GMT is applied widely, a rate increase up to 15% in a source country would merely redistribute the top-up tax gains from the MNE’s headquarter country to the source country, without deterring investment.30 We hence consider the scope and revenue implications of the global minimum tax through the lens of a QDMTT, which calculates the revenue potential in respect of subsidiaries of foreign MNEs operating in source countries. It does not include the revenue potential from an Income Inclusion Rule. 5.2 The Global Minimum Tax in Practice We now describe how we apply the GMT rules to simulate QDMTT revenue in a subset of five countries: Costa Rica, Greece, Honduras, Jamaica and South Africa. The first step is to identify the firms in scope of the GMT. Tax administrations do not always know which firms are subsidiaries of MNEs, and rarely know about the global activities of MNEs headquartered abroad. The Country by Country Reporting (CbCR) data is designed to close this gap: it mandates that MNEs with global revenue above 750 million euros report to their headquarter country the breakdown of sales, profits, and taxes paid by country of operation. In turn, MNEs’ headquarter countries should share this data bilaterally with all source countries. Yet, countries in our sample do not currently use the 30 The OECD published a paper specifically to advise developing countries on how to reform their tax expenditures in light of the new agreement (OECD, 2022). It states that “Pillar Two and the GloBE Rules, in particular, should empower governments to pursue tax reform and remove tax incentives where the costs outweigh the benefits from such incentives”, and later, “Given the global character of Pillar Two, inaction would only lead to forgone revenues.”. 19 CbCR micro data.31 To assess the revenue potential of the GMT, we use two alternative strategies, which rely on existing data and are replicable in many countries. The first strategy merges the CIT data with Orbis data available via institutional subscription. The second strategy uses a foreign ownership indicator from each country’s administrative tax records and the aggregate CbCR data indicating the total number of in-scope MNE affiliates, which is publicly available. We detail these two methods, and then explain how we proxy the top-up tax revenue using the CIT records. Method 1: Orbis Match Orbis contains information on the ownership structure of firms, as well as some financial data. How complete is the Orbis ownership and economic data for our sample? Table A.11 compares the number of firms and groups predicted to be in scope of the GMT in Orbis (columns 7 and 8) and the number of firms and groups in the aggregate CbCR data (columns 9 and 10), for each country. Orbis ownership data appears fairly complete: the number of in-scope subsidiaries and groups in Orbis and in the CbCR data are close for most countries. This is expected since Orbis has good coverage of firms headquartered in or with affiliates in high-income countries, and this coverage extends to these MNEs’ subsidiaries worldwide. Yet, economic activity data for subsidiaries in low and middle-income countries is almost always missing (Table A.12, columns 5 and 6). Thus, we need to use administrative tax records to capture firms’ tax bases. We extract the list of firms in scope from Orbis by identifying all MNE subsidiaries in the rele- vant countries and checking whether the MNE group meets the criteria that its global consolidated revenue was above 750 million EUR in two out of the last four years. Our partner tax adminis- trations in Costa Rica, Honduras and Jamaica then merge Orbis extraction with their CIT records, using company names. For South Africa, we did not have to perform the Orbis merge since firms report the global revenue of their MNE on their CIT declaration.32 The share of firms on the Orbis list that merge with the tax data ranges from 75% in Costa Rica to 70% in Honduras and 56% in Jamaica. At the level of the group, when at least one firm per group can be merged, the merge rates are slightly higher. However, given the imperfect merge, the number of firms that are in scope and identified in the tax administrative data is lower than the number in the aggregate CbCR data. Thus, the Orbis matched sample will yield a lower bound on the number of firms affected by the 31 Even in an EU country such as the Slovak Republic, the government did not use the micro CbCR data prior to engaging with researchers (Boukal et al., 2024). 32 For Greece, we attempted the merge based on a tax ID variable available in Orbis, but obtained a merge rate of only 39%. For Greece, we thus focus on method 2, using aggregate CbCR data. 20 GMT and on the GMT revenue potential. We use an alternative strategy to obtain an upper bound. Method 2: Aggregate CbCR Match For this strategy, we use a variable in the administrative tax records indicating whether a firm is foreign-owned (see Table B2, column 3, for the the precise source of this information, which differs across countries). For each country, we combine this sample of foreign-owned firms with the list of Orbis-matched firms (from method 1) to obtain a list of CIT filers that are subsidiaries of MNEs. We then rank firms on this list by the amount of top-up tax expected under a QDMTT (see below for how we construct this), and select the top firms, until we match the number of in-scope firms from the aggregate CbCR data. For example, the CbCR data indicates that for 2019, 703 firms are in scope in Costa Rica (Table A.11, column 9). We hence select the 703 firms with the highest potential top-up tax from the list of foreign firms. We do not know if the global revenue of the MNE group is large enough for these firms to be in scope as per the GMT rules. However, by selecting the firms with the highest top-up tax, we obtain an upper bound on the revenue potential of the GMT. Another reason why the estimate from this method would be an upper bound is that we have to calculate the top-up tax at the firm level rather than the group level for a significant share of firms, as group identifiers are only available in Orbis. Hence, we are only able to consolidate the returns–which could reduce a group’s top-up tax–for firms appearing in Orbis, but not for other foreign-owned firms. Calculation of the Top-up Tax We now present the key steps we follow to compute the top-up tax under the global minimum tax. Appendix B.2 discusses the data and definitions used in detail. 1. We consolidate firms’ outcomes at the group level, when group identifiers are available. Firms not identified as belonging to a group are considered as a one-firm group. We thereafter examine all outcomes at the group level. We apply the GMT de minimis exclusion whereby groups with profits and sales below a threshold are exempted from the top-up tax.33 2. We calculate the effective tax rate of the group. If a group had previous losses, we apply the deferred tax asset (DTA) as per the OECD GMT rules, which raises the groups’ ETR. The DTA 33 The de minimis exclusion stipulates that MNEs are not subject to top-up tax in countries in which they have an average GloBE revenue below 10 million euro and the average GloBE income tax is either negative or below 1 million euro, averaged over the last five years. 21 is essentially the tax-benefit of previous losses carried forward.34 3. We deduct the carve-outs from the tax base. This implies deducting 10% of payroll and 8% of tangible assets in the first year post implementation. Ten years after implementation, carve-outs are reduced to 5% of payroll and 5% of tangible assets. 4. We identify groups with an ETR below 15% as liable for the top-up tax. We calculate the top-up tax as profits times the difference between the ETR and 15%. 5.3 Direct Tax Revenue Gains from the Global Minimum Tax Revenue Potential Figure 6a plots the predicted revenue from the top-up tax as a share of base- line corporate income taxes, in each country. The predicted revenue is the sum of the simulated top-up tax payments by firms in scope, after applying the year 1 carve-outs, and absent any behav- ioral responses. For each country, we show the upper bound scenario (aggregate CbCR match) and the lower bound scenario (Orbis match). Figure 6b shows the number of groups liable for top-up tax in each country, and in parenthesis the number of firms belonging to those groups. The revenue gains vary substantially across countries, with a maximum gain of over 20% of CIT revenue in Costa Rica, and a minimum of less than 0.3% in South Africa. The upper and lower bound revenue gains are close to each other in most countries, except for Honduras where the lower bound is 1.66% of CIT revenue and the upper bound is 6.89%.35 The number of entities and groups in scope is closely correlated with the revenue potential, except in South Africa which has a much larger economy than the other countries and hence has a double-digit number of firms liable for top-up tax, but still a negligible revenue gain in aggregate. The estimated revenue gain is only attributable to the QDMTT and does not reflect the revenue potential of the Income Inclusion Rule in South Africa. Why are the potential revenue gains larger for Costa Rica than for the other countries? Costa Rica does not have a significantly larger number of in-scope firms as a share of its population, and the ETR among those firms is similar to that of other countries. Instead, the aggregate profit share 34 Loss making firms are able to carry forward the equivalent of 15% of their losses as ”tax assets” from the previous year. This tax asset is then added to the covered taxes of the next year, thus raising the ETR of the firm. If the resulting ETR is larger than 15%, firms can carry the remaining tax asset to the next years. 35 Note that this analysis excludes firms already paying the local Minimum Income Tax, but we include the rest of the firms. The excluded firms represent 8.9% of the total income tax revenue in Honduras in 2019. 22 of its in-scope firms is larger than in other countries (see Table B1). In addition, the carve-outs for economic substance are more limited, as we will see below. Thus, Costa Rica seems to operate as a hub for large MNEs in the region which have some real activity in the country but also engage in or are used for strategic tax minimization. The Role of Carve-outs Figure 7 shows revenue gains for different scenarios, in the order of the steps we take to compute the top-up tax, using the aggregate CbCR upper bound method. The first bar shows the revenue gains of a 15% minimum tax without any carve-outs permitted. The second bar plots the revenue gains, after applying, the carve-outs in the first year of application (10% of payroll and 8% of tangible assets). The third bar corresponds to the application of the carve-outs after ten years (5% of payroll and 5% of tangible assets). The equivalent results for the Orbis match lower bound method are shown in Figure B1. Allowing a share of the payroll and of tangible assets to be deducted from the tax base was a compromise in the negotiations for the Global Minimum Tax. The carve-outs imply that the GMT only limits “harmful” tax competition in the form of aggressive profit shifting but allows and effec- tively encourages tax competition to attract real economic activity. In Greece, Honduras, Jamaica and South Africa, the initial carve-outs reduce the revenue gains by around 60% on average. In Costa Rica, carve-outs make a much smaller dent in aggregate revenue potential, suggesting that MNE affiliates in Costa Rica have limited economic presence. In general, tangible assets play a more important role in reducing the tax base than the wage bill. We also observe substantial heterogeneity across firms, highlighting the importance of conducting analyses with micro data. Refundable Tax Credits The fourth bar in each panel in Figure 7 corresponds to a hypothetical scenario. It assumes that all existing tax credits are converted from non-refundable to refundable. A non-refundable tax credit is one which cannot lower a taxpayers’ liability below zero. A refundable tax credit instead allows such a situation in which the tax authority owes the taxpayer a refund. Refundable tax credits were permitted under GMT rules, primarily to protect green tax credits. Hence, these refundable tax credits effectively reduce a firm’s CIT liability without giving rise to GMT top-up tax. The possibility that governments may choose to convert non-refundable tax credits into refundable tax credits has been highlighted as a key loophole of the GMT, with some 23 countries already intending to convert their existing tax credits (e.g. Singapore). If we assumed that all existing tax credits were converted, and we apply the conservative carve- out rates for year 10, the revenue potential of the GMT would fall to zero in South Africa and to almost zero in Costa Rica, Honduras and Jamaica. Only Greece, which had to slash its tax credits during bail-out from the financial crisis, would be unaffected by this scenario. How likely is such a full-conversion scenario? Transforming non-refundable tax credits to re- fundable ones raises the volatility of countries’ tax revenues. This is a major concern for countries with limited fiscal space, i.e. especially for developing countries. It is also possible that countries abandon the use of their current tax credits, but instead of offering refundable tax credits, offer firm subsidies outside the tax system, to continue supporting firm production while remaining compli- ant with GMT rules. This would likely raise the net cost of subsidizing firms (i.e. it would raise CIT revenue by less than the new subsidies cost), but it seems a plausible policy choice, given the increased focus on industrial policy globally. A 20% Minimum Tax Rate The last bar in Figure 7 corresponds to a scenario with a 20% minimum tax rate, as initially proposed at the onset of the GMT negotiations. We apply the 10 year carve-outs. Unsurprisingly this scenario raises the revenue gains, but does so exponentially, as it limits the role of carve-outs, and expands the number of firms required to pay a top-up tax. 5.4 Comparison of Revenue Gains: Global versus Domestic Minimum Tax While we found large potential revenue gains from a simple domestic minimum tax, the direct revenue gains from the GMT appear much smaller, in all countries we study. In Costa Rica, the country with the highest revenue potential in both cases, the simple domestic minimum tax could raise CIT revenue by over 60%, but the GMT would realize less than a third of these gains. For South Africa, which has the lowest revenue potential under the GMT, the simple minimum tax would still increase CIT revenue by 9%, but the estimated GMT revenue increase attributable to the QDMTT is below 0.3%. Given the inability to study the revenue potential of the Income Inclusion Rule under the GMT, this represents a lower-bound estimate for South Africa. What explains these large differences in revenue gains between the two policy tools? First and foremost, the GMT affects a much smaller number of firms. On average the GMT is levied on 24 significantly fewer firms than the simple minimum tax (comparing columns 4 versus column 2 of Table A.9). In addition, firms in scope for the GMT are not necessarily among the top 1% of firms in size within a country (Table A.9, column 5). Some domestic standalone firms are much larger in both revenue and profits than GMT-liable subsidiaries. Second, the GMT base is different, as it allows firms to a) consolidate profits and losses within a group, b) carry deferred tax assets (i.e. losses in previous periods), and c) deduct a share of their wage bill and tangible assets. This means that the top-up tax which the GMT generates is smaller than the simple domestic minimum tax, even when restricting to those firms liable for both. Figure B3 shows that the consolidation at the group level weakly lowers the GMT top-up tax: this factor is not quantitatively important, as many groups have only one affiliate per country. The comparison of panel (d) in Figure 5 and in Figure A.9 shows that the deferred tax assets can be large in some countries. Finally, substance-based carve-outs are an important factor, as discussed in Section 5.3. 6 Discussion and Policy Implications This paper presented new facts on corporate taxation drawing on administrative data from 16 coun- tries. We document a consistent inverse U-shaped relationship between firm size and effective tax rates. This is an important finding, as the size-dependence of effective tax rates has been shown to affect the efficiency, equity and effectiveness of tax policy. For instance, changes in tax-related costs can alter the distribution of sales between MNEs and their national competitors (Martin et al., 2023; Gauß et al., 2024). Similarly, lower ETRs for the largest firms may give them a competitive edge versus their smaller competitors. Reduced ETRs for top firms may also affect the income distribution, since firms’ profits accrue disproportionately to rich individuals, and the corporate income tax acts as a backstop against personal income tax evasion (Del Carmen et al., 2024). Yet, arez Serrato corporate income taxes are also, in part, passed through to workers via lower wages (Su´ and Zidar, 2016; Fuest et al., 2018). The distributional implications of our findings are complex and depend on the concentration of ownership and employment across the firm size distribution. The documented taxation patterns by firm size could also be consistent with obstacles to tax policy implementation. The drop in ETR at the top could be due to an intentional targeting of the policy to large firms or to differential take-up of incentives. If it is the latter, then it suggests that 25 small and medium enterprises struggle to claim tax credits, some of which may stimulate desirable investments such as R&D spending or greening production. Indeed, there is some evidence that the effectiveness of tax incentives may be inversely related to firm size (Appelt et al., 2020). Over a quarter of the largest 1% of firms in the 16 countries face ETRs below 15%, the min- imum rate under the GMT: the distinction between tax havens and high-tax countries is less clear than often described. Despite their relatively high statutory tax rates, non-haven countries effective tax rates are low for specific firms. These countries would face incentives to reform their tax ex- penditures under the GMT. We simulate modest revenue gains from the GMT, in contrast to larger gains from a simple domestic minimum tax. These estimates might be the best available for some countries, but can be improved on several margins. First, the tax data by itself typically does not allow for a precise identification of which firms fall under the scope of the GMT. Relying on the CbCR micro data, which is hitherto unused in our sample countries could improve the accuracy of the estimation. Second, if tax expenditures are reduced, firms might adjust their economic activity, their tax evasion strategy, and shift profits back to their headquarter countries. These behavioral responses, which our static simulations ignore, could further lower the tax gains from the GMT. Another revenue-reducing response could take the form of increased consolidation of firms into groups to lower overall tax liabilities. Third, several general equilibrium effects could occur. By curbing profit shifting, the GMT could raise revenues in most non-haven countries, and firms with some real activity in tax havens might relocate to non-haven countries (for real activity to comple- ment profit shifting, see Ferrari et al. (2022) and Davies and Scheuerer (2024)). On the other hand, the GMT could also reduce activity and taxable profits in high tax countries if firms reduce invest- arez Serrato, 2022). These effects will ments at home as a result of higher effective tax rates (Su´ crucially depend on the uncertain implementation of the GMT, including the number of countries that will ratify it, the quality of enforcement, and the development of tax strategies to escape it. References Adhikari, Ajay, Chek Derashid, and Hao Zhang, “Public policy, political connections, and effective tax rates: Longitudinal evidence from Malaysia,” Journal of Accounting and Public policy, 2006, 25 (5), 574–595. Appelt, Silvia, Matej Bajgar, Chiara Criscuolo, and Fernando Galindo-Rueda, The Effects of R&D Tax Incentives and Their Role in the Innovation Policy Mix: Findings from the OECD MicroBeRD Project, 2016-19, OECD Publishing, 2020. 26 Bachas, Pierre, Anne Brockmeyer, Pablo Garriga, and Camille Semelet, “The Impact of COVID-19 on Formal Firms: Lessons from Administrative Tax Data,” 2024. Working Paper. , Roberto N Fattal Jaef, and Anders Jensen, “Size-dependent tax enforcement and compli- ance: Global evidence and aggregate implications,” Journal of Development Economics, 2019, 140, 203–222. Badertscher, Brad A., Sharon P. Katz, Sonja Olhoft Rego, and Ryan J. Wilson, “Conforming Tax Avoidance and Capital Market Pressure,” The Accounting Review, 11 2019, 94 (6), 1–30. Barak´ e, Mona, Paul-Emmanuel Chouc, Theresa Neef, and Gabriel Zucman, “Revenue Effects of the Global Minimum Tax Under Pillar Two,” Intertax, 2022, 50, 689–710. 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, 2021, 111 (12), 3827–3871. Beer, Sebastian, Ruud de Mooij, and Li Liu, “International corporate tax avoidance: A review of the channels, magnitudes, and blind spots,” Journal of Economic Surveys, 2020, 34 (3), 660– 688. Best, Michael, Mazhar Waseem, and Jawad Shah, “Detection Without Deterrence: Long-Run Effects of Tax Audit on Firm Behavior,” Technical Report 2022. Working Paper. Bilicka, Katarzyna, “Comparing UK Tax Returns of Foreign Multinationals to Matched Domestic Firms,” American Economic Review, August 2019, 109 (8), 2921–53. Blouin, Jennifer and Leslie A Robinson, “Double counting accounting: How much profit of multinational enterprises is really in tax havens?,” Available at SSRN 3491451, 2020. Boukal, Tom´ asˇ, Petr Jansk´y, Niels Johannesen, and Miroslav Palansk´ y, “Global Minimum Tax and Profit Shifting: Evidence from Administrative Data,” 2024. Working Paper. Buettner, Thiess, Michael Overesch, Ulrich Schreiber, and Georg Wamser, “The impact of thin-capitalization rules on the capital structure of multinational firms,” Journal of Public Eco- nomics, 2012, 96 (11-12), 930–938. Carmen, Giselle Del, Santiago Garriga, Wilman Nunez, ˜ and Thiago Scot, “Two decades of top income shares in Honduras,” Technical Report, The World Bank 2024. Clifford, Sarah, “Taxing multinationals beyond borders: Financial and locational responses to CFC rules,” Journal of Public Economics, 2019, 173, 44–71. Cobham, Alex, Tommaso Faccio, Javier Garcia-Bernardo, Petr Jansk´ y, Jeffery Kadet, and Sol Picciotto, “A Practical Proposal to end Corporate Tax Abuse: METR, a Minimum Effective Tax Rate for Multinationals,” Global Policy, 2021, 13, 18–33. Cristea, Anca D and Daniel X Nguyen, “Transfer pricing by multinational firms: New evidence from foreign firm ownerships,” American Economic Journal: Economic Policy, 2016, 8 (3), 170–202. Davies, Ron and Johannes Scheuerer, “Tax Haven Usage and Employment Reallocations: Evi- dence from Norway,” Technical Report, Mimeo 2024. Davies, Ronald B., Julien Martin, Mathieu Parenti, and Farid Toubal, “Knocking on Tax Haven’s Door: Multinational Firms and Transfer Pricing,” The Review of Economics and Statis- tics, 03 2018, 100 (1), 120–134. Desai, Mihir A, C Fritz Foley, and James R Hines Jr, “A multinational perspective on capital structure choice and internal capital markets,” The Journal of finance, 2004, 59 (6), 2451–2487. Devereux, Michael P, “International tax competition and coordination with a global minimum tax,” National Tax Journal, 2023, 76 (1), 145–166. 27 and Rachel Griffith, “The taxation of discrete investment choices,” Technical Report, IFS working papers 1998. and , “Evaluating tax policy for location decisions,” International tax and public finance, 2003, 10, 107–126. , Francois Bares, Sarah Clifford, Judith Freedman, Irem Guc ¨ ¸ eri, Martin McCarthy, Mar- tin Simmler, and John Vella, “The OECD Global Anti-Base Erosion Proposal,” PwC report, Oxford University Centre for Business Taxation, 2020, 1. Dischinger, Matthias and Nadine Riedel, “Corporate taxes and the location of intangible assets within multinational firms,” Journal of Public Economics, 2011, 95 (7-8), 691–707. Dyreng, Scott D., Michelle Hanlon, Edward L. Maydew, and Jacob R. Thornock, “Changes in corporate effective tax rates over the past 25 years,” Journal of Financial Economics, 2017, 124 (3), 441–463. Egger, Peter H and Georg Wamser, “The impact of controlled foreign company legislation on real investments abroad. A multi-dimensional regression discontinuity design,” Journal of Pub- lic Economics, 2015, 129, 77–91. , Nora M Strecker, and Benedikt Zoller-Rydzek, “Estimating bargaining-related tax advan- tages of multinational firms,” Journal of International Economics, 2020, 122, 103258. Ferrari, Alessandro, S´ ebastien Lafitte, Mathieu Parenti, and Farid Toubal, “Profit Shifting Frictions and the Geography of Multinational Activity,” 2022. Fuest, Clemens, Andreas Peichl, and Sebastian Siegloch, “Do higher corporate taxes reduce wages? Micro evidence from Germany,” American Economic Review, 2018, 108 (2), 393–418. , Florian Hugger, and Florian Neumeier, “Corporate profit shifting and the role of tax havens: Evidence from German country by country reporting data,” Journal of Economic Behavior Or- ganization, 2022. Garcia-Bernardo, Javier and Petr Jansk` y, “Profit shifting of multinational corporations world- wide,” World Development, 2024, 177, 106527. , Petr Jansk´y, and Thomas Tørsløv, “Decomposing Multinational Corporations’ Declining Effective Tax Rates,” IMF Economic Review, 2022, 70 (2), 338–381. Gauß, Patrick, Michael Kortenhaus, Nadine Riedel, and Martin Simmler, “Leveling the play- ing field: Constraints on multinational profit shifting and the performance of national firms,” Journal of Public Economics, 2024, 234, 105116. Graham, John R, Jana S Raedy, and Douglas A Shackelford, “Research in accounting for income taxes,” Journal of Accounting and Economics, 2012, 53 (1-2), 412–434. Gumpert, Anna, Jr. Hines James R., and Monika Schnitzer, “Multinational Firms and Tax Havens,” The Review of Economics and Statistics, 10 2016, 98 (4), 713–727. Hanappi, Tibor and Ana Cinta Gonz´ alez Cabral, “The impact of the international tax reforms under Pillar One and Pillar Two on MNE’s investment costs,” International Tax and Public Finance, 2022, 29 (6), 1495–1526. Haufler, Andreas and Hayato Kato, “A Global Minimum Tax for Large Firms Only: Implications for Tax Competition,” arXiv preprint arXiv:2404.14302, 2024. Hebous, Shafik and Michael Keen, “Pareto-improving minimum corporate taxation,” Journal of Public Economics, 2023, 225, 104952. and Niels Johannesen, “At your service! The role of tax havens in international trade with services,” European Economic Review, 2021, 135, 103737. Hindriks, Jean and Yukihiro Nishimura, “The compliance dilemma of the global minimum tax,” 28 Technical Report, Universit´ e catholique de Louvain, Center for Operations Research 2022. Hoppes, Jeffrey, Patrick Langetieg, Edward Maydew, and Michele Mullaney, “Is Tax Plan- ning Best Done in Private?,” Technical Report 2020. Working Paper. Hugger, Felix, Ana Cinta Gonzalez Cabral, and Pierce O’Reilly, “Effective tax rates of MNEs: New evidence on global low-taxed profit,” OECD Taxation Working Papers, 2023, (67). , , Massimo Bucci, Maria Gesualdo, and Pierce O’Reilly, “The Global Minimum Tax and the taxation of MNE profit,” OECD Taxation Working Papers, 2024, (68). Janeba, Eckhard and Guttorm Schjelderup, “The global minimum tax raises more revenues than you think, or much less,” Journal of International Economics, 2023, 145, 103837. Jansk´ y, Petr, “Corporate Effective Tax Rates for Research and Policy,” Public Finance Review, 2022, 0 (0), 10911421221137203. Jr, James R Hines, “Evaluating tax harmonization,” Technical Report, National Bureau of Eco- nomic Research 2024. Karkinsky, Tom and Nadine Riedel, “Corporate taxation and the choice of patent location within multinational firms,” Journal of international Economics, 2012, 88 (1), 176–185. Klemm, Alexander, “Causes, benefits, and risks of business tax incentives,” International Tax and Public Finance, 2010, 17 (3), 315–336. Laz˘ ar, Sebastian, “Determinants of the variability of corporate effective tax rates: Evidence from Romanian listed companies,” Emerging Markets Finance and Trade, 2014, 50 (sup4), 113–131. Liu, Li, Tim Schmidt-Eisenlohr, and Dongxian Guo, “International transfer pricing and tax avoidance: Evidence from linked trade-tax statistics in the United Kingdom,” Review of Eco- nomics and Statistics, 2020, 102 (4), 766–778. Martin, Julien, Mathieu Parenti, and Farid Toubal, “Corporate tax avoidance and sales: micro evidence and aggregate implications,” 2023. Mascagni, Giulia, Nara Monkam, and Christopher Nell, “Unlocking the potential of adminis- trative data in Africa: Tax compliance and progressivity in Rwanda,” 2016. Mooij, Ruud De and Li Liu, “At a cost: The real effects of transfer pricing regulations,” IMF Economic Review, 2020, 68, 268–306. and , “At a cost: The real effects of thin capitalization rules,” Economics Letters, 2021, 200, 109745. Nicod` eme, Gaetan, “Sector and size effects on effective corporate taxation,” 2002. OECD, “Tax challenges arising from digitaliization, economic impact assessment,” Technical Re- port 2020. Report. , Tax Incentives and the Global Minimum Corporate Tax 2022. Pomeranz, Dina and Jos´ e Vila-Belda, “Taking state-capacity research to the field: Insights from collaborations with tax authorities,” Annual Review of Economics, 2019, 11, 755–781. Redonda, Agustin, Christian von Haldenwang, and Flurim Aliu, “Global Tax Expenditures Database (GTED),” March 2022. Serrato, Juan Carlos Su´ arez, “Unintended Consequences of Eliminating Tax Havens,,” R&R, American Economic Review, 2022. and Owen Zidar, “Who benefits from state corporate tax cuts? A local labor markets approach with heterogeneous firms,” American Economic Review, 2016, 106 (9), 2582–2624. Slemrod, Joel, “Tax Compliance and Enforcement,” Journal of Economic Literature, December 2019, 57 (4), 904–54. Tørsløv, Thomas, Ludvig Wier, and Gabriel Zucman, “The Missing Profits of Nations,” The 29 Review of Economic Studies, 07 2022, (0), 1–36. rdac049. Tørsløv, Thomas R, Ludvig S Wier, and Gabriel Zucman, “Externalities in International Tax Enforcement: Theory and Evidence,” Technical Report, National Bureau of Economic Research 2020. Wamser, Georg, “The impact of thin-capitalization rules on external debt usage–a propensity score matching approach,” Oxford Bulletin of Economics and Statistics, 2014, 76 (5), 764–781. Wier, Ludvig, “Tax-motivated transfer mispricing in South Africa: Direct evidence using transac- tion data,” Journal of Public Economics, 2020, 184, 104153. and Hayley Erasmus, “The dominant role of large firms in profit shifting,” IMF Economic Review, 2023, 71 (3), 791–816. Wier, Ludvig S and Gabriel Zucman, “Global profit shifting, 1975-2019,” Technical Report, National Bureau of Economic Research 2022. Wu, Wenfeng, Chongfeng Wu, Chunyang Zhou, and Jun Wu, “Political connections, tax ben- efits and firm performance: Evidence from China,” Journal of Accounting and Public policy, 2012, 31 (3), 277–300. 30 7 Tables and Figures Table 1: Descriptive Statistics on Countries and Data (1) (2) (3) (4) (5) (6) (7) (8) Country GDP pc Nbr. of Avg. Turnover Profits>0 Max Statutory Avg. ETR (%) Year (ISO Code) (const. 2010 US$) Firms (Thousand $) (%) Tax Rate (%) All/Prof. firms Albania (ALB) 2015-2019 4543.39 19237 1146.48 80.68 15 9.4/11.6 Colombia (COL) 2020-2022 6836.23 404062 1491.68 85.35 35 20.1/23.5 Costa Rica (CRI) 2018-2019 12662.42 64438 1320.35 80.7 30 14.6/18.1 Dominican Rep. (DOM) 2006-2015 6838.94 38028 1785.31 64.75 27 15.5/24 Ecuador (ECU) 2014-2019 5970.11 48477 2107.86 76.99 28 15.8/20.5 Eswatini (ESW) 2013-2018 3696.36 3805 1050.92 66.86 27.5 14.2/21.3 Ethiopia (ETH) 2010-2016 671.29 15037 2167.38 70.28 30 17.5/24.9 31 Greece (GRC) 2012-2018 18647.5 70915 NaN 58.83 29 12.9/21.9 Guatemala (GTM) 2006-2019 4263.71 22994 3133.03 67.09 25 16.4/24.4 Honduras (HND) 2014-2019 2443.88 23706 1498.2 74.62 25 20.9/25.5 Jamaica (JAM) 2018-2019 5307.51 10408 2635.69 50.89 25 8.8/17.4 Mexico (MEX) 2010-2015 10098.17 390043 3205.61 52.75 30 14.4/26.9 Rwanda (RWA) 2010-2017 769.29 5306 1070 74.84 30 20.8/27.8 Senegal (SEN) 2007-2018 1384.64 5732 3641.84 59 30 15.9/26.9 South Africa (ZAF) 2014-2019 6188.7 237438 3178.91 67.74 28 8.1/12 Uganda (UGA) 2015-2019 922.02 16083 586.99 62.61 30 13.8/22.1 Note: This table presents summary statistics on firms in the 16 countries in our data. All statistics are from administrative corporate tax records, except for the GDP per capita (column 4) which is from the World Development Indicators. Column 2 shows the years available in the data for each country. We use the most recent year to compute metrics shown in columns 3 to 8. The effective tax rate (ETR) can be larger than the statutory tax rate due to the reintegration of non-taxable deductions in the profit definition (see Figure 1). Appendix C provides additional details on each country’s corporate tax system. Additional country level data is provided in Table A.2. This table is discussed in Sections 2.1 and 2.2. Table 2: Explaining the Relationship Between Effective Tax Rates and Firm Size Within the Top Decile of Firm Size Outcome: Effective Tax Rate + Controls for firm Baseline + Dummies indicating use of tax expenditures characteristics Specification: (1) (2) (3) (4) (5) (6) (7) (8) Regressor: Dummy Top 1% (unweighted cross-country average -2.21 -1.45 -2.17 -1.86 -1.98 -2.00 -1.52 -0.56 of country-specific point estimates) N countries with 14 12 14 14 12 12 13 7 negative point estimate N countries where 10 8 10 10 8 9 8 4 lower one-sided t-test rejects null N countries 16 16 16 16 16 16 16 16 Controls: Firm characteristics × × Reduced rate dummy × × Exempt income dummy × × Special deduction dummy × × Re-timing dummy × × Tax credits dummy × × Note: This table presents regression results analyzing the drivers of the ETR-firm-size relationship among firms in the top decile of the firm-size distribution. The sample is restricted to firms within the top size decile only (revenue percentile 90 and above), so we focus on the decreasing part of the relationship between ETR and firm size. We focus on profitable firms, holding the size percentile fixed based on the full sample. We regress the ETR on a dummy tagging firms in the top one percentile of the firm-size distribution (Dummy Top 1%, equation 1). Column 1 only includes the Top 1% dummy. Column 2 controls for firm characteristics (sector dummies, capital city and location dummy, foreign ownership dummy, and firm age) where this information is available. In columns 3 to 7, we control one by one for dummy variables indicating whether or not the firm made use of each of the different tax provisions that can explain the ETR slope. In the first row of the table, we report the unweighted average of the β1 coefficients on the top 1% dummy across countries. The second row reports the number of countries for which the coefficient is negative, and the third row reports the number of countries for which a one-sided t-test rejects the null hypothesis that the coefficient is zero at a 5% significance level. Table A.5 displays the same results but focusing on firms in deciles 1-9 of the size distribution. Country-specific coefficients are detailed in Table A.7 and robustness to different choices for the main regressor (indicator for largest firms) is shown in Table A.6. See Table A.8 for details on available tax provisions by country. This table is discussed in Section 3.4. 32 Figure 1: Key Concepts and Variables Total Costs Total Revenue Material Labor Sales Other revenues (except - Capital Financial dividends) Depreciation Unspecified Other Costs = Approximation for  + Non-deductible costs Economic Profit Profit (Loss) - Exempt Income - Special Deductions:    Investment incentives    Capital allowances    Other deductions Taxable Profit (Loss) - Loss carryforwards = Tax Base x Statutory Tax Rate Gross Tax Liability - Tax Credits:   Investment credits   Export promotion   Foreign tax credits*   Other credits*  Net Tax Liability - Advanced payments - Withholdings - Other creditable    payments Tax to Remit Effective  Net Tax Liability Tax Rate = Profit Note: This figure presents the key fiscal concepts and variables used in this study, constructed in a harmonized way in 16 countries. All costs are deducted from revenue to derive the profit/loss concept which we use to compute the effective tax rate. As the denominator in our ETR measure, we use the net tax liability, which is the annual amount in corporate income tax due. For loss-making firms, the ETR is set to zero. This figure is discussed in Section 2.2. ∗ Foreign tax credits and other credits could be pre-payments instead of tax credits. In Figure A.8 we recompute the ETR without considering them as tax credit, i.e. they are not deducted from the gross tax liability. 33 Figure 2: Effective Tax Rates and Firm Size, All Firms Countries ordered by Statutory Tax Rate ETR (All firms, incl. loss−making) Top STR Top 1% Albania Guatemala Jamaica Dominican Republic 30 20 10 0 Eswatini Ecuador South Africa Greece 30 20 10 0 Honduras Costa Rica Ethiopia Mexico 30 20 10 0 Rwanda Senegal Uganda Colombia 30 20 10 0 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 Firm Size Quantiles Note: This figure shows effective tax rates (ETRs) as a function of firm-size quantiles, for all 16 countries in our data. The gray crosses show the average ETR at each quantile. Loss-making firms are assigned a zero ETR. The blue line is a cubic smoothing spline with six knots, estimated using the R function ggformula::geom spline. Firm-size quantiles (x-axis) are based on firms’ revenue. The quantiles correspond to percentiles between the 1st and 99th percentile (white area), and to 0.2% bins between the 99th and 100th percentiles (gray shaded area). Figure A.3 replicates this figure, focusing on profitable firms only. This figure is discussed in Section 3.2. 34 Figure 3: Effective Tax Rate Minus Statutory Tax Rate, Profitable Firms Countries ordered by Statutory Tax Rate Top 1% ETR−STR (Profitable firms only) Albania Guatemala Jamaica Dominican Republic 0 −10 Eswatini Ecuador South Africa Greece 0 −10 Honduras Costa Rica Ethiopia Mexico 0 −10 Rwanda Senegal Uganda Colombia 0 −10 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 Firm Size Quantiles Note: This figure shows the difference between the ETR and the statutory tax rate (STR) as a function of firm-size quantiles, for all 16 countries in our data. The gray crosses show the average ETR-STR difference for each quantile. We include only profitable firms. The orange line is a cubic smoothing spline with six knots, estimated using the R function ggformula::geom spline. Firm-size quantiles (x-axis) are based on firms’ revenue. The quantiles correspond to percentiles between the 1st and 99th percentile (white area), and to 0.2% bins between the 99th and 100th percentiles (gray shaded area). This Figure is discussed in Section 3.2. 35 Figure 4: Effective Tax Rates and Firm Size: Robustness (a) Cross-Country Average ETR Countries ordered by Statutory Tax Rate Average ETR (Profitable firms only) 25.0 22.5 20.0 17.5 15.0 10 20 30 40 50 60 70 80 90 99.0 99.9 Firm Size Quantiles (b) ETR by Sector (c) Lifetime ETR Primary Secondary Retail Services 2−year 3−year 4−year 5−year 25.0 25.0 22.5 22.5 20.0 20.0 17.5 17.5 15.0 15.0 10 20 30 40 50 60 70 80 90 99.0 99.9 10 20 30 40 50 60 70 80 90 99.0 99.9 Firm Size Quantiles Firm Size Quantiles Note: These figures present robustness tests for the ETR-firm-size relationship. Panel (a) serves as a benchmark, presenting the average ETR by firm size for profitable firms. We take the average across countries (in Figure A.3) for each quantile, weighing countries equally, and then obtain the fit over quantiles with a cubic smoothing spline with six knots. Panel (b) shows the average ETR-firm-size relationship across countries for four large sector groups: agriculture (primary), industry and construction (secondary), retail, and services. Panel (c) presents a multi-year measure of the N N ETR (from N = 2 to N = 5 years) where the ETR for firm i is n=1 (CITi,n )/ n=1 (N etP rof itsi,n ). By construction, the different lines in Panel (c) rely on different samples, as we can compute the N-year-ETR only for firms that are in the panel at least N-1 years before the most recent cross-section. All curves are cubic smoothing splines with six knots, estimated using the R function ggformula::geom spline. The quantiles correspond to percentiles between the 1st and 99th percentile (white area), and to 0.2% bins between the 99th and 100th percentiles (grey shaded area). This figure is discussed in Section 3.3. 36 Figure 5: Scope and Tax Revenue Potential of a 15% Domestic Minimum Tax on the 1% Largest Firms (a) Share of top 1% firms with ETR below 15% (b) Average ETR among these firms Colombia 46 STR=35% Colombia 4.2 Uganda 35.2 Uganda 2.3 Senegal 30.4 Senegal 1.6 Rwanda 16.7 Rwanda 1.3 STR=30% Mexico 13.7 Mexico 4 Ethiopia 7 Ethiopia 3.4 Costa Rica 31.4 Costa Rica 1.8 Greece 20.5 STR=29% Greece 1.9 South Africa 28.1 South Africa 3 STR=28% Ecuador 16.9 Ecuador 2.9 Eswatini 38.2 STR=27.5% Eswatini 6.1 Dom. Rep. 30.3 STR=27% Dom. Rep. 2.6 Jamaica 27.8 Jamaica 6.3 Honduras 29.9 STR=25% Honduras 0.3 Guatemala 5.9 Guatemala 5.8 Albania 7.7 STR=15% Albania 11 0 20 40 60 0 5 10 15 Share of top 1% firms with ETR<15% Mean(ETR) conditional on ETR<15% (c) Profit of these firms (share of total) (d) Revenue increase (share of baseline CIT rev.) Colombia 40.5 Colombia 36.1 Uganda 36.3 Uganda 37 Senegal 30.4 Senegal 22.3 Rwanda 11.4 Rwanda 7.8 Mexico 6.8 Mexico 2.8 Ethiopia 0.9 Ethiopia 0.2 Costa Rica 61.1 Costa Rica 91 Greece 24.4 Greece 19 South Africa 16.7 South Africa 9 Ecuador 6.5 Ecuador 3.6 Eswatini 37.1 Eswatini 21.9 Dom. Rep. 25.5 Dom. Rep. 21.2 Jamaica 12 Jamaica 7 Honduras 24.5 Honduras 20 Guatemala 9 Guatemala 2.1 Albania 1.9 Albania 0.3 0 20 40 60 0 10 20 30 40 Profit cond. ETR<15%, as share of total Revenue change with domestic min. tax, share of baseline Note: Panel (a) shows the share of firms in the top 1% of the size (revenue) distribution that have an ETR below 15% in the most recent cross-section, focusing on profitable firms. Panel (b) shows the average ETR among profitable firms with an ETR below 15%, within the top 1% of firm size. Panel (c) plots the share of aggregate reported profits accounted for by profitable firms with an ETR below 15% within the top 1% of firm size. Panel (d) shows the hypothetical revenue gains from requiring all firms in the top 1% to pay an ETR of at least 15% (i.e. we simulate an ETR of 15% and the associated tax liability for profitable top 1% firms with an actual ETR below 15%), as a share of current CIT liabilities of all firms in the latest pre-COVID-19 cross-section. Figure A.9 shows the results when deferred tax assets, as defined in the global minimum tax rules, are taken into account. This figure is discussed in Section 4.2. 37 Figure 6: Global Minimum Top Up Tax Revenue Potential Year 1 Carve-outs Scenario (a) Increase in Corporate Tax Liability (b) Number of Entities & Groups (% of Aggregate CIT Revenue) Liable for Top-Up Tax Note: Dark bars indicate the CbCR method. Light bars indicate the Orbis method. Panel (a) shows the potential revenue gains from requiring the affected firms in each group to pay an ETR of at least 15% in a scenario with 10% payroll carve-outs and 8% tangible assets carve-outs (which are the carve-outs corresponding to the first year of the Global Minimum tax implementation). In Panel (b) bars refer to the number of groups that would be affected under that scenario. The numbers in parenthesis refer to the number of entities associated with such groups. *South Africa’s results not based on Orbis, rather firms are identified using the group consolidated turnover information in the administrative data. Figure B2 shows that same results for the Year 10 Carve-outs Scenario. This figure is discussed in Section 5.3. 38 Figure 7: Global Minimum Tax Revenue Potential Under Different Scenarios CbCR Method, % Increase in Aggregate CIT Revenue (a) Costa Rica (b) Greece (c) Honduras 39 (d) Jamaica (e) South Africa Note: This figure illustrates potential revenue gains from implementing a 15% minimum effective tax rate (ETR) across five scenarios. The first bar shows revenue gains from a scenario with only a de minimis exclusion and no carve-outs. The second one includes year 1 carve-outs (8% for tangible assets, 10% for payroll), while the third one uses year 10 carve-outs (5% for both). The fourth bar combines year 10 carve-outs with the possibility of using refundable credits to reduce tax liability. Finally, the last bar assumes a 20% minimum tax rate and year 10 carve-outs. This figure relies on the CbCR aggregate method to calculate the top-up tax, yielding an upper-bound on the revenue gains. See Figure B1, for the equivalent figure which relies on the Orbis lower-bound method. This figure is discussed in Section 5.3. Appendix Table of Contents A Effective Tax Rates Analysis - Additional Results 41 B Simulating the Global Minimum Tax 60 B.1 Additional Figures and Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 B.2 Data availability per Country and Definitions . . . . . . . . . . . . . . . . . . . 65 B.3 Comparison to Other Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 C Corporate Tax Data: Context and Cleaning 70 40 Appendix A: Effective Tax Rates Analysis - Additional Results Figure A.1: Example of a Corporate Tax Return Form: Rwanda 41 Note: This figure presents an example of the corporate tax return for Rwanda. Circled in color are the general concepts we use across countries, as described in Figure 1. We already exclude dividend income from total revenue, so we do not consider exempt dividend in our calculations (here line 115). This figure is discussed in Section 2.2. Figure A.2: Corporate Tax Expenditures as a Share of GDP (a) Global Comparison: (b) In-sample Comparison: CIT Expenditures vs. GDP per capita Our Estimates vs. Government Estimates Note: This figure shows estimates for the size of corporate income tax expenditures as a share of countries’ GDP per capita (2015 constant USD from the World Bank). Panel (a) shows our estimates for the 16 sample countries and estimates of CIT expenditures from 64 additional countries, available in the Global Tax Expenditure Dataset (GTED), for a total sample size of 80 countries. The latter comes from countries’ official tax expenditure reports, which are based on international standards and should be completed yearly. In practice, countries do not systematically produce tax expenditure reports, and when they do, the methods vary significantly depending on each country’s definition and statistical capacity. Panel (a) also includes two regression lines, one for each set of data points. The green dashed line results from regressing the CIT expenditure on the GDP per capita (slope 0.30, intercept -1.43) for the 16 countries in our sample. The navy dashed line represents the expanded estimate sample, including 64 additional countries (slope 0.12, intercept -0.43). For our sample countries, tax expenditures represent on average 1.04% of their GDP, compared to an average tax expenditure of 0.68% for the extended sample of 80 countries. Panel (b) shows the correlation between our estimates of tax expenditures, and government estimates, for the 15 sample countries where both are available (we miss government tax expenditure data for Ethiopia). The solid line is the 45-degree line. The dashed line represents the regression of our estimated CIT expenditures against the government’s estimates (slope 0.82, intercept 0.60). The government estimates are based on GTED, with manual adjustments based on the country’s original tax expenditure report where necessary, to include all relevant tax expenditures. To compute aggregate tax expenditures from our microdata, we first use the difference between the top statutory tax rate and the firm-specific ETRs multiplied by their profits to compute firm-level tax expenditures. We then aggregate these tax expenditures across firms, within country, and divide it by the sum of corporate tax liabilities, yielding a ratio of foregone CIT revenue to actual tax revenue. We multiply this ratio by the CIT collection as a share of GDP (using tax revenue data from Bachas et al. 2022) to obtain total tax expenditure as a share of GDP. This figure is discussed in Section 3.1. 42 Figure A.3: Effective Tax Rates and Firm Size, Profitable Firms Countries ordered by Statutory Tax Rate ETR (Profitable firms only) Top STR Top 1% Albania Guatemala Jamaica Dominican Republic 30 20 10 Eswatini Ecuador South Africa Greece 30 20 10 Honduras Costa Rica Ethiopia Mexico 30 20 10 Rwanda Senegal Uganda Colombia 30 20 10 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 Firm Size Quantiles Note: This figure is identical to Figure 2 but focuses on profitable firms only. The figure shows effective tax rates (ETRs) as a function of firm-size quantiles, for all 16 countries in our data. Size quantiles are determined based on revenue in the full population of firms (including zero-profit and loss-making firms). The gray crosses show the average ETR at each quantile. The purple line is a cubic smoothing spline with six knots, estimated using the R function ggformula::geom spline. The quantiles correspond to percentiles between the 1st and 99th (white area), and to 0.2% bins between the 99th and 100th percentiles (gray shaded area). This figure is discussed in Section 3.2. 43 Figure A.4: Share of unprofitable firms Countries ordered by Statutory Tax Rate Share of unprofitable firms Top 1% Albania Guatemala Jamaica Dominican Republic 100 75 50 25 0 Eswatini Ecuador South Africa Greece 100 75 50 25 0 Honduras Costa Rica Ethiopia Mexico 100 75 50 25 0 Rwanda Senegal Uganda Colombia 100 75 50 25 0 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 20 40 60 80 99 99.9 Firm Size Quantiles Note: This figure displays the share of unprofitable firms at each quantile, for all 16 countries in our data. The gray crosses show the share at each quantile, while the blue lines are cubic smoothing splines with six knots, estimated using the R function ggformula::geom spline. Firm-size quantiles(x-axis) are based on firms’ revenue. The quantiles correspond to percentiles between the 1st and 99th percentile(white area), and to 0.2% bins between the 99th and 100th percentiles (gray shaded area). This figure is discussed in Section 3.2. 44 Figure A.5: Cross country correlations between tax variables and macro variables log(GDP per capita) Population Max. STR Avg STR−ETR (tax gap) Avg ETR (All) Avg ETR (profitable) Coeff 500 M Rev >750 M Scope >750 M - Groups Firms Groups Albania 293147 79 0 5036 221 177 169 121 56 50 Colombia 4176803 436 13 10665 3403 3226 3171 1008 1926 887 Costa Rica 415281 135 1 1474 794 774 763 388 703 386 Dominican Republic 155716 65 2 1208 503 457 448 252 410 277 Ecuador 169088 40 2 11237 847 848 824 444 469 298 Eswatini 54707 20 0 395 162 147 143 84 106 54 Ethiopia 1448155 5 1 425 133 120 106 80 53 45 Greece 1047354 1075 20 8716 2644 2453 2392 680 1764 578 Guatemala 354048 32 0 1010 588 524 520 262 438 284 Honduras 1954 15 0 1138 304 300 295 158 216 175 Jamaica 288663 48 0 8707 187 152 151 102 117 89 Mexico 2906576 663 55 22153 11130 9937 9558 1997 11209 2093 Rwanda 51265 8 0 298 100 86 83 66 11 10 Senegal 14088 21 1 863 300 247 232 155 198 118 South Africa 3264746 1187 82 25629 10500 9238 8895 1223 6998 1111 Uganda 554171 27 0 608 247 229 225 142 163 86 Note: This table provides information on the ownership network data availability in Orbis for the sample of coun- tries included in the study. Column 1 shows the total number of firms existing in Orbis for each country. Column 2 indicates the total number of firms identified as headquarters of a multinational group in each country. Column 3 shows how many firms from the previous column are in the scope of the global minimum tax. Columns 4 to 8 display information in terms of firms identified as subsidiaries of multinational groups. Column 4 shows how many multina- tional subsidiaries are present in each country. Columns 5 and 6 present how many of these firms are subsidiaries of multinational corporations with revenues exceeding 500 million euros and 750 million euros, respectively. Column 7 shows the number of firms in scope of the global minimum tax. Column 8 shows the number of multinational groups with revenues exceeding 750 million euros operating in that country. Finally, columns 9 and 10 detail the number of subsidiaries of multinational groups and the number of multinational groups with revenues over 750 million euros, as indicated by the aggregate CbCR database. This table is discussed in Section 5.2. Table A.12: Orbis Data on Firms’ Economic Activity in LMICs is Poor (1) (2) (3) (4) (5) (6) Country Number of Subsidiaries Number of Subsidiaries Number of Firms Number of Firms Share of Subsidiaries Share of Firms with Turnover data with Turnover data with Turnover data (%) with Turnover data (%) Albania 712 5036 13612 293147 14.10 4.60 Colombia 5864 10665 1681849 4176803 55.00 40.30 Costa Rica 55 1474 167 415281 3.70 0.00 Dominican Republic 37 1208 122 155716 3.10 0.10 Ecuador 6422 11237 86825 169088 57.20 51.30 Eswatini 28 395 183 54707 7.10 0.30 Ethiopia 21 425 215 1448155 4.90 0.00 Greece 3245 8716 100160 1047354 37.20 9.60 Guatemala 41 1010 96 354048 4.10 0.00 Honduras 25 1138 54 1954 2.20 2.80 Jamaica 92 8707 113 288663 1.10 0.00 Mexico 5717 22153 225698 2906576 25.80 7.80 Rwanda 24 298 53 51265 8.10 0.10 Senegal 53 863 694 14088 6.10 4.90 South Africa 2926 25629 16280 3264746 11.40 0.50 Uganda 45 608 63 554171 7.40 0.00 Note: This table provides information on the economic activity data availability in Orbis for the sample of countries included in the study, based on data checked in October 2024. For each country, column 1 shows the number of firms identified as subsidiaries with available turnover data. Column 2 shows the total number of subsidiaries. Column 3 displays the total number of firms existing in Orbis at the country level with turnover data. Column 4 shows the total number of firms. Columns 5 and 6 are built based on the previous columns and show the share of subsidiaries and firms with available turnover data out of the total number of existing subsidiaries and firms, respectively. This table is discusssed in Section 5.2. 59 Appendix B: Simulating the Global Minimum Tax B.1 Additional Figures and Tables Table B1: Descriptive Statistics on Firms in Scope of the Global Minimum Tax (1) (2) (3) (4) (5) (6) Orbis Method CbCR Method Mean Percentage Percentage Mean Percentage Percentage Country Effective of of Effective of of Tax Rate Filers Profits Tax Rate Filers Profits Costa Rica 2.14 0.11 17.10 1.83 0.12 18.39 Greece NA NA NA 1.25 0.10 13.74 Honduras .85 0.08 4.87 .37 0.19 14.21 Jamaica 2.84 0.05 1.57 2.84 0.05 1.76 South Africa* .42 0.03 3.67 .52 0.03 3.74 Notes: This table provides key statistics on firms in scope of the GMT, to understand the differences in simulated revenue gains across countries. Columns display information for the sample of firms liable for top-up tax under a scenario with a 15% minimum tax rate, year 1 carve-outs, and de minimis exclusion criteria, using the Orbis method in columns 1 to 3 and the CbCR method in columns 4 to 6. Columns 1 and 4 display the mean effective tax rate. Columns 2 and 5 show the percentage of firms subject to the minimum tax relative to the total number of firms. Columns 3 and 6 display the share of profits represented by these firms relative to the total aggregate profits of all firms, in percent. *South Africa’s results not based on Orbis, rather firms are identified using the group consolidated turnover information in the administrative data. This table is discussed in Section 5.3. 60 Figure B1: Global Minimum Tax Revenue Potential Under Different Scenarios Orbis Method, % Increase in Aggregate CIT Revenue (b) Greece (c) Honduras (a) Costa Rica 61 (e) South Africa* (d) Jamaica Note: This figure illustrates potential revenue gains from implementing a 15% minimum effective tax rate (ETR) across five scenarios. The first bar shows revenue gains from a scenario with only a de minimis exclusion and no carve-outs. The second one includes year 1 carve-outs (8% for tangible assets, 10% for payroll), while the third one uses year 10 carve-outs (5% for both). The fourth bar combines year 10 carve-outs with the possibility of using qualified refundable credits to reduce tax liability. Finally, the last bar assumes a 20% minimum tax rate and year 10 carve-outs. All the revenue gains are estimated using the Orbis method. *South Africa’s results not based on Orbis, rather firms are identified using the group consolidated turnover information in the administrative data. Figure 7 shows the same results when using the CbCR method. This figure is discussed in Section 5.3. Figure B2: Global Minimum Top Up Tax Revenue Potential Year 10 Carve-outs Scenario (a) Share of Corporate Tax Liability (%) (b) Number of Groups & Entities Note: Dark bars indicate the CbCR method. Light bars indicate the Orbis method. Panel (a) shows the potential revenue gains from requiring the affected firms in each group to pay an ETR of at least 15% in a scenario with 10% payroll carve-outs and 8% tangible assets carve-outs (which are the carve-outs corresponding to the first year of the Global Minimum tax implementation). In Panel (b) bars refer to the number of groups that would be affected under that scenario. The numbers in parenthesis refer to the number of entities associated with such groups. *South Africa’s results not based on Orbis, rather firms are identified using the group consolidated turnover information in the administrative data. This figure is identical to Figure 6, except that the carve-out rates for year 10 after GMT adoption are applied. 62 Figure B3: Increase in Corporate Tax Liability (% of Aggregate CIT Revenue) Orbis Method with Year 1 Carve-Outs Note: This figure examines the role of groups (light) vs entities (dark) in generating the estimates using the Orbis method. Mechanically, the estimates are always larger when the top-up tax is calculated at the entity level rather than the group level. In Jamaica and South Africa, which have a larger average number of entities per group, the difference in the aggregate revenue gain between the entity-level and the group-level calculation is larger. The estimates assume a scenario with a 15% minimum tax rate, the applica- tion of the de minimis exclusion criteria, and year 1 (10% payroll, 5% tangible assets) carve-outs. *South Africa’s results not based on Orbis, rather firms are identified using the group consolidated turnover information in the administrative data. This figure is discussed in Section 5.4. 63 Figure B4: The Role of the De Minimis Exclusion for the Number of Groups and Share of Profits Affected by Top-up Tax (a) Orbis Method: Number of Groups (b) Orbis Method: Share of Profits (c) CbCR Method: Number of Groups (d) CbCR Method: Share of Profits Note: This figure shows the number of groups liable for top-up tax and the importance of these groups in terms of their share in aggregate profits. The dark bars are for our central scenario, without applying the de minimis exclusion. The light bars are for the scenario with the de minimis exclusion. No carve-outs are assumed in any case. Panels (a) and (b) display the results obtained by applying the Orbis method, while panels (c) and (d) are the result of implementing the CbCR method. *South Africa’s results not based on Orbis, rather firms are identified using the group consolidated turnover information in the administrative data. 64 B.2 Data availability per Country and Definitions Table B2: Data Availability for Global Minimum Tax Simulations (1) (2) (3) (4) (5) (6) (7) (8) Country Orbis Merge MNE Indicator Orbis & MNE Overlap Tangible Assets Payroll Tax Credits Refundable Tax Credits Costa 763 firms were identified as The MNE indicator was obtained There are no firms identified as Our measure of tangible assets Payroll is directly observable for Tax credits for specific tax treat- There are no clear indicators Rica multinationals in Orbis. Of from the National Registry of multinationals in Orbis that are is based on items visible in the only a small subsample of firms. ments, such as franc zones and to separately identify refundable these, 422 were matched with the Costa Rica and indicates whether also classified as foreign compa- D101 tax form. For Costa Rica We use this data and other items export incentives, are visible on tax credits. tax records. the parent company associated nies in the administrative data. we use the fixed assets line. from the D101 form and the the tax form. with each firm is a foreign com- D152 form to predict the payroll pany. 342 firms were iden- for the rest of the sample. tified as multinationals through this source. South 8,895 firms were identified as The MNE indicator is directly We were not allowed to match Our measure of tangible assets The payroll measure was ob- A list of tax credits is present in No refundable tax credits are Africa multinationals in Orbis. Though obtained from the tax return Orbis with the tax data. How- is based on items visible in the tained from the tax form, which the tax form, but not identifiable identifiable in the tax form. we were not allowed to match where firms indicate if they are ever, we identified 2,415 firms tax form, including cash hold- displays the total aggregate in the data. Thus we compute an Orbis with the tax data. Instead, part of foreign MNE. 3,413 firms in scope thanks to the tax data, ings, inventory, land properties, amount of wages and salaries aggregate amount of tax credits we used consolidated turnover were identified. against 8,895 in Orbis. and loans. paid. but we are not able to distinguish amount of groups available in them. the tax data to identify firms in scope. 2,415 firms are part of foreign groups above the EUR 750 million threshold. Honduras 295 firms were identified as 626 firms were identified as A total of 68 firms are identi- Our measure of tangible assets The payroll measure was con- Tax credits for investments and There are no clear indicators multinationals in Orbis. Of multinationals through this fied as multinationals in Orbis, is based on items visible in the structed by aggregating subitems other credits are recognizable in to separately identify refundable these, 214 were matched with the source. matching their identification as tax form, including cash hold- related to labor costs, such as the tax data. tax credits. tax records. multinational companies in the ings, investments, biological as- wages and salaries, all of which administrative data. sets, and accounts receivable. are directly visible in the tax data. Jamaica 151 firms were identified as The MNE indicator was obtained A total of 5 firms are identified as We build our measure of tan- The payroll measure was ob- Tax credits are processed sepa- Refundable tax credits are multinationals in Orbis. Of from the AT01 and AT02 Forms, multinationals in Orbis, match- gible assets based on the infor- tained from the IT02 form, rately in Section F of the IT02 clearly distinguished within the these, 106 were matched with the in which firms must indicate if ing their identification as multi- mation available on AT01 and which displays the total ag- form, and include employment IT02 form within Section F. tax records. they are either local or overseas national companies in the admin- AT02 forms. We aggregate three gregate amount of wages and tax credits, corporate tax credits, companies. 78 firms were iden- istrative data. subitems from the annual decla- salaries paid. and double taxation relief cred- tified as multinationals through rations of assets: estate in land, its, among others. this source. equipment and machinery, and other fixed assets. When no as- sets data is available, we impute the values following the proce- dure detailed below. Greece 2,392 firms were identified as 9,094 firms were identified as A total of 9 firms are identified as Tangible assets is missing for a Payroll is not directly observ- Tax credits for investments and There are no clear indicators multinationals in Orbis. Of multinationals through the admin multinationals in Orbis, match- considerable part of the data. We able. We impute payroll data other credits are recognizable in to separately identify refundable these, 948 were matched with the data. ing their identification as multi- predict it using Lasso for missing by assuming an annual minimum the tax data. tax credits. tax records. national companies in the admin- data. wage (EUR 20,000) multiplied istrative data. by the number of employees. Number of employees are pre- dicted for the missing values us- ing Lasso. Note: This table summarizes country-level data used for global minimum tax simulations. Column 2 shows the number of multinational firms identified in Orbis and matched with tax records. Columns 3 indicates how many firms were identified as multinationals through the administrative data. Column 4 indicates how many firms were identified as multinationals both in Orbis and in the tax records. Columns 5 and 6 explain how tangible assets and payroll measures were constructed, respectively. Columns 7 and 8 indicate the visibility of tax credits and refundable tax credits in tax forms. This table is discussed in Section 5.2. This section supplements Table B2 by detailing how key variables used in global minimum tax simulations are constructed and sourced from administrative data. We first discuss the firm-level foreign ownership indicator, which is crucial for the CbCR method as it combines with Orbis data to better identify multinational firms. We later address payroll expenses and tangible assets, and their coverage in tax records. We build these variables according to Pillar 2 definitions. Costa Rica • Administrative data MNE indicator. We identified 342 multinational firms in the administrative data via a binary indicator created by the National Registry of Costa Rica, a public entity respon- sible for registration and geospatial operations. This indicator enabled us to categorize firms as multinationals based on whether their parent company was local or foreign. • Tangible assets. We build our measure of tangible assets from the items available on Form D101. To do so, we select those items that are considered tangible assets within the framework of the GloBE rules. According to these rules, tangible assets usable for the computation of carve-outs include property, plant, and equipment, natural resources and the right to exploit them, and the lessee’s right of use of tangible assets located in that jurisdiction. In the case of Costa Rica, we use the fixed assets item available on Form D101. • Payroll. Payroll data is directly accessible only for a subset of large firms that submitted a more specific tax form until 2018. To address this missing information, we combine two approaches. First, we employ a Lasso methodology to estimate payroll for the remainder of the sample, com- bining data from D101 and D152 sources. The D152 form (Declaraci´ on Anual de Retenciones ´ Impuestos Unicos y Definitivos) is an annual informative declaration entities must complete when withholding taxes for various reasons. Among these, employers must withhold the correspond- ing income tax amount for wages above a certain threshold. The aggregate amount of the tax base for these withheld at source amounts (wages and salaries) provides a proxy measure of the real payroll at the firm level, which we use as input in a Lasso linear regression model together with other independent variables from the D101 form. We proceed as follows. For years with detailed corporate tax and D152 data, we run a regression of the real payroll on a series of regres- sors previously selected through a Lasso procedure. These variables include economic sector, our measure of profits, depreciation, and tax liability. Using the estimated coefficients for the subsample where we observe the real payroll, we estimate the payroll for the rest of the sample with missing data. Second, for firms lacking D152 data, we estimate payroll costs as 30% of total administrative and sales costs reported in the D101 form. This percentage is derived from the average share of wages and salaries in total administrative and sales costs for large firms that completed the more detailed tax form until 2018. These firms are the only subset with directly observable disaggregated variables, including wages and salaries expenses. Notably, the propor- tion of wage and salary costs relative to total sales and administrative costs remains consistent across income deciles for this sample. 66 Greece • Administrative data MNE indicator. The tax data includes a binary variable indicating whether the firm is foreign-owned or not. • Tangible assets. Tangible assets is only available for a small part of the sample. We estimate it for the rest of the firms with Lasso using log of turnover, sector categories, our measure of profits and foreign ownership status. • Payroll. Due to the lack of information on this item, we estimate it. We assume that firms pay workers the annual minimum wage (EUR 20,000) and multiply this amount by the number of employees. Number of employees is drawn from the data. For missing information, we also use Lasso estimation using log of turnover, sector categories, our measure of profits and foreign ownership status. Honduras • Administrative data MNE indicator. The tax data identifies foreign-owned firms with a binary indicator. • Tangible assets. Honduras’s tax data includes disaggregated items on assets. We include in our measure of tangible assets equipment, machinery, properties used as part of the firm business activity, and other assets directly involved in the economic cycle of firm. • Payroll. We build a measure of payroll costs by aggregating several wage and salary subitems directly visible in the tax data. Jamaica • Administrative data MNE indicator. The AT01 and AT02 forms were valid until 2018 and served as annual asset declarations. These forms incorporated a binary variable indicating whether a company was foreign-owned or domestic. By merging the asset data with the corporate income dataset, we identified multinationals within our data. • Tangible assets. We used data from forms AT01 and AT02 to compute tangible assets. These annual asset declarations provide disaggregated data on current and fixed assets. Despite the latest available declarations dating from 2018 and 2017, we could match 80% of the Orbis sam- ple, totaling 403 firm-year pairs over the period. To measure tangible assets, we aggregate three sub-items from the annual declarations of assets: estate in land, equipment and machinery, and other fixed assets. To address the remaining missing data, we first calculate the tangible assets to income ratio and its mean value across income quartiles for the sample with available data. We then use this ratio to impute the unobserved values. By integrating this quartile-based mean to the missing data, we reconstruct the missing tangible asset values. 67 • Payroll. We gather salary and wage cost data from three sources: the IT02 corporate income tax form, the Schedule 1 annex form, and the S02 form. The IT02 and Schedule 1 forms provide aggregate payroll expenses, while the S02 form offers a detailed breakdown of employee-level expenses. Due to occasional discrepancies between these sources, we use the highest reported amount as the company’s total payroll expense. South Africa • Administrative data MNE indicator. We identify foreign-owned firms through the administrative data using indicators on headquarters firm location. • Tangible assets. The South African data includes several lines of assets. Following the GloBE rules, we include in our measure of tangible assets property assets, vehicles, and fixed assets. • Payroll. We realy on an aggregate measure of payroll expenses that includes all the costs related to wages and salaries. B.3 Comparison to Other Studies Our GMT simulations complement other studies that have used aggregate data, e.g. Country-by- Country Reporting (CbCR) data, to estimate GMT revenue effects (Barak´ e et al., 2022; Cobham et al., 2021; Devereux et al., 2020; OECD, 2020; Hugger et al., 2024). For all countries except Costa Rica, our revenue gain estimates are lower than those from other studies. For instance, Barak´ e et al. (2022) estimate that European countries could gain about 16% of CIT revenue from the introduction of a top-up tax. The OECD study by Hugger et al. (2024) suggests revenue gains could reach 6.5% to 8.1% of CIT revenue. Higher gains are projected for both high- and low- income countries compared to middle-income countries. Approximately two-thirds of the GMT revenue gains can be attributed to direct top-up taxation (4.3% to 5.4% of CIT revenue), and the remaining third to reduced profit shifting. Our estimates, ranging from 0.2% to 6% of CIT revenue for Greece, Honduras, Jamaica and South Africa indicate that even the OECD projections might be overly optimistic. We now discuss the main methodological differences between the studies. Identifying In-Scope Firms The use of micro-level data enables us to precisely identify firms in scope of the GMT. Iden- tifying in-scope firms is more complicated when working with macro data. CbCR data, which underpins both the Barak´ e et al. (2022) and Hugger et al. (2024) studies, is organized into bi- lateral matrices on a UPE-affiliate basis (where UPE stands for ultimate parent entity), contain- ing jurisdictional-level information on MNE activity, profits, and taxes. Filing a CbCR report is mandatory for MNE groups with consolidated revenues of EUR 750 million or above. Thus, CbCR data, in principle, allows for convenient identification of in-scope firms, whereas relying solely on administrative tax data can pose challenges, as discussed in Section 5.2. 68 However, CbCR data is often incomplete, requiring Barak´ e et al. (2022) and Hugger et al. (2024) to impute missing data using third-party sources like ORBIS or through ad hoc extrapola- tions. This is especially true for lower-income countries where robust data is scarce. In contrast, we utilize administrative records containing the universe of CIT filers in each country. Furthermore, the semi-aggregate nature of the CbCR data precludes identifying MNE affiliates that fall below the de minimis threshold and should hence be excluded from the GMT revenue es- timation. This potentially inflates estimated revenue gains from a top-up tax. Barak´ e et al. (2022) attempt to address this by setting the top-up taxes to zero for country pairs with aggregate revenue and profits below the de minimis threshold. Nevertheless, they acknowledge that this correction changes their estimates only marginally and that it likely under-corrects for the de minimis exemp- tion. Conversely, the detailed nature of our administrative tax data allows us to assess whether the de minimis exclusion applies to each firm. Calculating the Effective Tax Rate (ETR) Determining ETRs with macro-data such as aggregate CbCR, presents challenges, notably to correctly assess profits. We highlight two well-known issues: inter-temporal adjustments, such as dealing with prior period losses, and double-counting profits. If not addressed, these issues can bias ETRs downward and overstate the revenue gains from a GMT (Blouin and Robinson, 2020). Barak´ e et al. (2022) manage inter-temporal adjustments by averaging ETRs over their sample period (2016-2017). The Hugger et al. (2024) applies a more sophisticated loss adjustment to each cell of their profit matrix based on the typical share of losses in positive profits observed in aggregated CbCR data at the affiliate jurisdiction level. However, in our data we observe significant heterogeneity across firms regarding the importance of prior period losses in determining ETRs. Accounting for this heterogeneity at the firm level, which our granular data allows, enhances the precision of our revenue gain estimates. The issue of double-counting profits, particularly intra-group dividends, is addressed similarly. Barak´ e et al. (2022) admit that their data does not fully allow for systematic correction of this is- sue. They estimate, using information from a few tax administrations, that accounting for dividend payments could reduce pre-tax profits by about 40%, lowering their estimated revenue gains for the European Union by 25%. The Hugger et al. (2024) also relies on information from tax admin- istrations where available. In the absence of such data, they base their double-counting correction on comparing ETRs of domestic MNEs with those of foreign MNEs in the same jurisdiction, as- suming that foreign entities are less impacted by double counting as intra-company dividends are more prevalent in the UPE’s jurisdiction. They apply a downward correction to domestic affiliates’ profits to align with the average ETR of foreign-owned affiliates. However, this approach is not fully satisfactory as affiliate-to-affiliate dividend payments in tiered MNE structures would still lead to double counting. Our granular administrative data, on the other hand, is not affected by the double-counting issue. We conclude that, although our study is more narrowly focused than previous studies — con- ducting only QDMTT simulations and for a select set of countries — the granular nature of our data allows us to conduct a nuanced and precise estimation of QDMTT revenue gains. 69 Appendix C: Corporate Tax Data: Context and Cleaning Some countries supplement their main corporate tax with additional taxes which are levied in place or in addition to the corporate tax. Given the country-specific nature of these tax regimes, we deal with them on a case-by-case basis. Broadly, we include in our analysis firms that belong to the main regime. We typically drop firms paying a minimum tax, and firms in simplified tax regime, which usually concern small firms and requires a different tax form. We restrict the analysis to firms whose revenue is larger than 1 to avoid distortions in the ETR computation procedure. • Albania. The statutory corporate tax rate (STR) was 15 percent during the years covered by our data (2015 to 2019). In 2019, smaller firms with revenue below 14 million ALL benefited from a 5 percent reduced tax rate, and firms with revenue below 5 million ALL were fully exempt. The relevant revenue thresholds for the rate reduction and exemption have changed over the years. • Colombia. The statutory corporate tax rate was 32 percent in 2020, 31 percent in 2021, and 35 percent in 2022. Entities located in Free Trade Zones enjoy a reduced rate of 20%. • Costa Rica. – The highest statutory corporate tax rate was 30 percent during the years covered by our data (2006 to 2019). The tax system includes two other tax brackets for smaller firms with tax rates at 10 and 20 percent. The tax rate is applied to profit, but the tax brackets are based on firms’ revenue. The tax bracket thresholds are inflation-adjusted annually. – We restrict the sample to firms that are clearly labeled as legal entities. To do so, we rely on an entity type indicator that classifies firms as either legal or natural persons. This indicator contains missing values in several cases, which are also dropped from the sample to avoid including natural persons accidentally. • The Dominican Republic. The statutory corporate tax rate was 27 percent for all firms in 2015. Over the span of our panel, the statutory rate changed several times: it was 28 percent in 2014, 29 percent from 2011 to 2013, 25 percent from 2007 to 2010, and 30 percent in 2006. • Ecuador. The STR was 22 percent from 2013 to 2017, and increased to 25% percent in 2018. In 2018, micro-firm with revenue below 1,000,000 LCU are still subject to the 22% rate, as well as firms in the mining and extractive industry. The STR can also be 28%, depending on the company’s shareholders structure (a corporate structure where at least 50% of the firms is owned by tax haven residents) and disclosure compliance (at least 50% of undisclosed shareholders). • Eswatini. The STR was 27.5 percent from 2014 to 2018, and 30 percent in 2013. • Ethiopia. The STR was 30 percent over the span of our data (2011 to 2016). We remove from the sample legal entities classified as NGOs, cooperatives, government institutions, joint ventures, micro and small enterprises, sole proprietors, and individuals. We filter these observations by relying on legal status indicator. 70 • Greece. The statutory corporate tax rate was 26 percent until 2014 and 29 percent between 2015 and 2018. • Guatemala. The STR was 25 percent over the span of our panel (2006 to 2019). Firms with a profit rate below a threshold are taxed on turnover. We do not include these firms in our analysis. Firms without a defined legal status (unclassified) are removed from the sample. • Honduras. – The STR is 25% since 2017 and was 30% from 2014 to 2016. A Solidarity Contribution tax is also applied on top for firms with a taxable income over HNL 1 million. The Solidarity Contribution tax rate is 5%. A minimum tax on turnover at a rate of 1.5 percent was applied to firms above a turnover threshold. These firms pay either the corporate income tax on profits or the tax on turnover, whichever is larger. The threshold for the minimum tax application has been gradually raised over time. As result, only 0.3 percent of firms in our sample (within the very largest firms) paid the minimum tax in 2019. We hence exclude minimum tax payers from the sample. In addition, some firms in Honduras are subject to the asset tax in lieu of the CIT: firms pay 1 percent on the excess above L3 million of their total assets. – We keep firms paying the asset tax in the sample because these firms are subject to whichever tax liability is greater–between the CIT and the Asset Tax–and we would drop a large share of the largest firms from the sample if we dropped asset tax payers. Those firms are not subject to an STR of 25%, so we compute an STR that firms are subject to in the following fashion:ST Ri = (SolidarityT axBasei ∗ 0.05 + N etT axBasei ∗ 0.25)/(N etT axBasei ). In that sense, the maximum STR for Honduras can reach 34%. The tax liability we take into account for the ETR calculation is the greater of the two tax liabilities. In the ETR calculation, if the denominator (e.g. profit) is smaller or equal to zero, but taxes paid is greater than zero (due to the asset tax), we set the ETR to the maximum STR.36 – However, we exclude firms subject to Honduras’s minimum income tax. The Honduran min- imum tax of 1% on gross income is paid by taxpayers with gross income over one billion lempiras in the prior fiscal that have a tax calculation (based on the rates established in the Honduran Income Tax Law) that results in a lower amount. • Jamaica. The statutory corporate tax rate was 25 percent during the period covered by our data, which is the rate applicable for most of the firms. Building societies pay a tax rate of 30%. Regulated companies (companies regulated by the Bank of Jamaica or other government 1 institutions) pay tax rate of 33 3 %. • Mexico. 36 Finally, in Honduras, we also drop the 11 percent of firms filing manually in 2019 (instead of online), because we do not observe profits for them. Other papers working with Honduras data do the same (Lobel et al. 2021). 71 – The STR was 30 percent for all firms during the span of our panel (2010 to 2015). The data we use for Mexico are open source and have been altered before release by the tax administration (SAT) to ensure the data are fully anonymized. First, they added an error term to all reported amounts, drawn from a mean-zero normal distribution. Then, they dropped observations for which total income was larger than three standard deviations above the median, which means in practice that the data do not contain the very top firms.37 – Firms with a turnover smaller than MXN 1,000 were removed from the sample due to incon- sistent amounts displayed in the data. – Although Mexico’s statutory tax rate was 30 percent during our study period, several firms showed effective tax rates exceeding 100% in our non-winsorized data. Therefore, we set the upper bound for winsorized ETR measures at 100% rather than 30% in order to capture this behavior in our data. • Rwanda. Rwanda’s statutory corporate tax was 30 percent during the period covered by our data (2010 to 2017). There are specific regimes for smaller firms, such as a flat tax and a lump sum tax, but we do not include these firms in the analysis. Only firms taxed under the standard CIT regime are kept in the analysis sample. • Senegal. During the period covered by our data (2010 to 2018), Senegal applied a corporate tax of 30 percent on positive taxable profits, and an alternative minimum tax of 0.5 percent of turnover on firms with negative taxable profits. The maximum amount cannot be more than XOF 5 million. • Uganda. The STR was 30 percent over the span of our data (2015 to 2019). Small firms with revenue below certain thresholds pay a simplified tax. We do not include these firms in our analysis. • South Africa. – During the analyzed period, the maximum statutory corporate tax rate was 28 percent. Eligible small businesses were subject to a graduated tax rate system, ranging from 0% to 28%. – Only legal entities are part of the analysis sample, which are identified through an entity type variable. Dormant companies are removed from the sample. We remove outlier firms from the bottom of the income distribution, i.e. firms in the first size deciles that face a STR above 10%. We also exclude micro business, body corporate, and share block companies from the analysis. While these firms are required to file annual CIT returns (ITR14), levy income received by body corporates and share block companies is typically exempt from income tax; and micro businesses with a turnover of R1 million or less can opt into the Turnover Tax system. 37 For details on how the data have been altered, see here. 72 Supplementary References Adhikari, Ajay, Chek Derashid, and Hao Zhang, “Public policy, political connections, and effective tax rates: Longitudinal evidence from Malaysia,” Journal of Accounting and Public policy, 2006, 25 (5), 574–595. Bach, Laurent, Antoine Bozio, and Cl´ ement Malgouyres, “L’h´ erog´ et´ eit´ en´ e des taux d’imposition implicites des profits en France: constats et facteurs explicatifs.” PhD dissertation, Institut des politiques publiques (IPP) 2019. Bachas, Pierre, Matthew H Fisher-Post, Anders Jensen, and Gabriel Zucman, “Globalization and Factor Income Taxation,” March 2022, (29819). Carreras, Marco, Chandu Dachapalli, and Giulia Mascagni, “Effective corporate tax burden and firm size in South Africa: A firm-level analysis,” Technical Report, United Nations Univer- sity 2017. Clark, W. Steven, “Using Micro Data to Assess Average Tax Rates,” in P.B. Sorenson, ed., Mea- suring the Tax Burden on Capital and Labor, MIT Press, 2004, pp. 319–358. Gaertner, Fabio B, Brent Glover, and Oliver Levine, “A Re-examination of Firm Size and Taxes,” Available at SSRN 3928145, 2021. Gauthier, Bernard and Mark Gersovitz, “Revenue erosion through exemption and evasion in Cameroon, 1993,” Journal of Public economics, 1997, 64 (3), 407–424. and Ritva Reinikka, “Shifting tax burdens through exemptions and evasion: An empirical investigation of Uganda,” Journal of African economies, 2006, 15 (3), 373–398. Guha, Atulan, “Company size and effective corporate tax rate: study on Indian private manufac- turing companies,” Economic and political weekly, 2007, pp. 1869–1874. Gupta, Sanjay and Kaye Newberry, “Determinants of the variability in corporate effective tax rates: Evidence from longitudinal data,” Journal of accounting and public policy, 1997, 16 (1), 1–34. Halleux, Frederic and Christian Valenduc, “Effective tax rate and the size of the company in Belgium an empirical investigation on micro-data,” Technical Report 2007. Janssen, Boudewijn, “Corporate effective tax rates in the Netherlands,” De Economist, 2005, 153 (1), 47–66. Kern, Beth B and Michael H Morris, “Taxes and firm size: the effect of tax legislation during the 1980s,” The Journal of the American Taxation Association, 1992, 14 (1), 80. Kim, Kenneth A and Piman Limpaphayom, “Taxes and firm size in Pacific-Basin emerging economies,” Journal of international accounting, auditing and taxation, 1998, 7 (1), 47–68. Laz˘ar, Sebastian, “Determinants of the variability of corporate effective tax rates: Evidence from Romanian listed companies,” Emerging Markets Finance and Trade, 2014, 50 (sup4), 113–131. 73 Lobel, Felipe, Thiago Scot, and Pedro Zuniga, “Corporate Taxation and Evasion Responses: Evidence from a Minimum Tax in Honduras,” Technical Report, Job Market Paper 2021. Mascagni, Giulia and Andualem Mengistu, “Effective tax rates and firm size in Ethiopia,” De- velopment Policy Review, 2019, 37, O248–O273. , Nara Monkam, and Christopher Nell, “Unlocking the potential of administrative data in Africa: Tax compliance and progressivity in Rwanda,” 2016. Nicod` eme, Gaetan, “Sector and size effects on effective corporate taxation,” 2002. Porcano, Thomas, “Corporate tax rates: Progressive, proportional, or regressive,” Journal of the American Taxation Association, 1986, 7 (2), 17–31. Rego, Sonja Olhoft, “Tax-avoidance activities of US multinational corporations,” Contemporary Accounting Research, 2003, 20 (4), 805–833. Richardson, Grant and Roman Lanis, “Determinants of the variability in corporate effective tax rates and tax reform: Evidence from Australia,” Journal of accounting and public policy, 2007, 26 (6), 689–704. Stickney, Clyde P and Victor E McGee, “Effective corporate tax rates the effect of size, capital intensity, leverage, and other factors,” Journal of accounting and public policy, 1982, 1 (2), 125–152. wu Wang, Shiing, “The relation between firm size and effective tax rates: A test of firms’ political success,” Accounting Review, 1991, pp. 158–169. Wu, Wenfeng, Chongfeng Wu, Chunyang Zhou, and Jun Wu, “Political connections, tax ben- efits and firm performance: Evidence from China,” Journal of Accounting and Public policy, 2012, 31 (3), 277–300. Yuzhu, Lu, Shao Liang, and Yue Zhang, “Growing “political power” of large frms and the down- ward cash ETR trend,” in “Proceedings of the Appalachian Research in Business Symposium: 9th Annual Conference, March 24-25 2022” The Appalachian Research in Business Symposium (ARBS) 2022, pp. 117–124. Zimmerman, Jerold L, “Taxes and firm size,” Journal of accounting and economics, 1983, 5, 119–149. 74