Policy Research Working Paper 10639 Revealing 21% of GDP in Hidden Assets Evidence from Argentina Juliana Londoño-Vélez Dario Tortarolo Development Economics Development Research Group December 2023 Policy Research Working Paper 10639 Abstract Despite substantial offshore tax evasion, Argentines dis- improved, expanding the tax bases for both wealth tax and closed assets worth 21 percent of GDP under a tax amnesty capital income tax, especially at the top. The subsequent in 2016. This paper studies how enforcement initiatives tax hike on foreign assets in 2019 boosted tax progressivity, impact individuals’ tax behavior, tax progressivity, and raising the effective tax rate for the wealthiest 0.1 percent revenue collection. Offshore tax evasion is concentrated of adults, and established Argentina’s wealth tax as one of among the wealthiest 0.1 percent of adults. Tax compliance the most successful globally in revenue generation. This paper is a product of the Development Research Group, Development Economics. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at dtortarolo@worldbank.org. 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 Revealing 21% of GDP in Hidden Assets: Evidence from Argentina Juliana Londoño-Vélez and Dario Tortarolo∗ Key words: tax evasion, offshore wealth, amnesties, enforcement, Argentina JEL codes: H26, H31, D31 ∗ Londoño-Vélez: Assistant Professor of Economics, UCLA, Bunche Hall 8283, 315 Portola Plaza, 90095, Los Angeles, California, United States (j.londonovelez@econ.ucla.edu). Tortarolo: Economist, World Bank, Development Research Group (dtortarolo@worldbank.org). We thank Estefanía Saravia Gómez for excel- lent research assistance. We are grateful to Annette Alstadsæter, Gabriel Zucman, Emmanuel Saez, Niels Johannesen, Finn Tarp, Thomas Piketty, Ségal Le Guern Herry, Carlos Scartascini, David Szakonyi, Daniel Weishaar, David Splinter, Ignacio Flores, and Roberto Arias for their valuable comments and seminar partici- pants at 2022 NBER Public Economics Program Meeting, EU Tax Observatory at PSE, Skatteforsk (NMBU), The World Bank Tax Conference, UNU-WIDER, IFS-UCL-LSE/STICERD Development Economics Semi- nar, and the 78th IIPF Congress. We gratefully acknowledge financial support from UNU-WIDER. Finally, Londoño-Vélez also thanks the UCLA Society of Hellman Fellows for their support. The views expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the Interna- tional 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. 1 Introduction Offshore tax evasion poses a severe challenge for tax policy (EU Tax Observatory, 2024; Slemrod, 2019; Tørsløv et al., 2022). Approximately 8% of households’ financial wealth is held in tax havens (Zucman, 2013), with the wealthiest 1% of the population owning the majority of these offshore assets (Alstadsæter et al., 2019; Guyton et al., 2021; Johannesen et al., 2023; Londoño-Vélez and Ávila-Mahecha, 2021). A significant portion of them goes undisclosed and untaxed by their respective home authorities, eroding tax revenue and hampering tax progressivity. To encourage tax evaders to reveal their foreign incomes and assets, governments worldwide have introduced voluntary disclosure programs that offer reduced penalties in exchange for tax compliance (OECD, 2015). In addition to these pro- grams, authorities have implemented various measures to combat cross-border tax eva- sion and promote transparency in the global financial system, including automatic tax information exchange agreements (TIEAs), the US Foreign Account Tax Compliance Act (FATCA), and the global multilateral network for the automatic exchange of financial ac- count data known as the Common Reporting Standard (CRS). Understanding how these policies can effectively work together to boost tax com- pliance, revenue, and progressivity is crucial for the conduct of tax policy. However, ex- tracting meaningful insights from the diverse experiences of various countries has proven challenging due to differences in enforcement capabilities across jurisdictions, baseline evasion rates, and program design. Furthermore, most voluntary disclosure programs researchers have examined, such as Alstadsæter et al. (2019, 2022) and Johannesen et al. (2020), took place before recent advancements in international tax coordination. These earlier programs witnessed limited participation and disclosures of income and assets that represented only a negligible fraction of countries’ gross domestic product (GDP). Consequently, we know little about evaders’ willingness to comply with tax regulations in the current landscape of global tax coordination. This paper overcomes these challenges by studying how tax enforcement initiatives impact individuals’ tax behavior, tax progressivity, and revenue collection, focusing on Argentina. The country is particularly well suited to examining these issues for several reasons. First, there is a lot at stake: the equivalent of 36.5% of the country’s GDP was held offshore before these initiatives (Figure A.1), nearly four times the world average (Alstadsæter et al., 2018). Second, Argentina’s rich policy variation makes it the world’s largest natural experiment for enforcement policies. Over the past few decades, both left- and right-wing governments have implemented voluntary disclosure programs, adapt- ing their scope and features. Despite previous unsuccessful attempts, a policy package 1 introduced in 2016 achieved remarkable success by bringing previously undisclosed for- eign and domestic assets into tax compliance, amounting to a cumulative total of 21% of GDP and standing out as one of the most successful tax amnesty initiatives globally. The extensive scale of Argentina’s asset disclosures provides an opportunity to utilize straight- forward methodologies to unpack the effects of changes in tax enforcement. Third, Argen- tines are required to report their domestic and foreign assets annually to the tax authority due to the existence of a gross wealth tax on individuals and firms, enabling us to track declared income and wealth over time. Specifically, we use two decades of detailed tax tabulations and employ the generalized Pareto interpolation method to characterize the distributions of income and wealth. We begin by leveraging the disclosures made under the 2016 amnesty to shed light on the prevalence, nature, and distribution of tax evasion. We find substantial offshore evasion: about 255,000 Argentines admitted to concealing assets under the program, and strikingly, more than 80% of these assets were hidden abroad, mainly in the US and tax havens or low-tax jurisdictions like Uruguay, Switzerland, and the British Virgin Islands. Regarding the types of assets disclosed, over half of the disclosed foreign assets consisted of stocks, one-quarter were held in bank accounts and various currencies, and one-tenth were real estate properties. Furthermore, our analysis reveals that offshore tax evasion is predominantly concentrated among the wealthiest individuals, especially within the top 0.1% of the wealth distribution. Next, we examine impacts on tax compliance up to five years later. We find that the asset disclosures resulted in enduring increases in reported wealth. The number of Argen- tines declaring foreign assets in their wealth tax returns tripled after 2016, and the value of reported foreign assets quadrupled from Arg$250 billion in 2015 (4.3% of GDP) to Arg$1 trillion in 2016 (16.5% of GDP) and reached Arg$1.25 trillion in 2019 (20.7% of GDP), bringing the country closer to the macro estimates of offshore wealth by Alstadsæter et al. (2018).1 Furthermore, the improved reporting of foreign assets significantly expanded the tax bases for both wealth tax and capital income tax. The wealth tax revenue more than doubled, with the amnesty’s one-time penalty recouping the entire wealth tax rev- enue lost due to tax evasion since 2002. Additionally, taxpayers also reported the income generated from the disclosed assets more truthfully, declaring three times more capital income. Again, this expanded tax base endured even five years later, indicating a lasting improvement in tax compliance. 1 While individuals who start reporting their wealth on their tax return without participating in this amnesty could, in principle, be driving the increase, we find that virtually all disclosures went through the tax amnesty program, unlike the "quiet disclosures" in the US (Johannesen et al., 2020). 2 The enforcement policy had important distributional implications. Since offshore tax evasion was predominantly concentrated among the wealthiest individuals, the reported wealth of the top 0.1% of adults experienced substantial growth. These high-net-worth individuals, who tend to own most of their assets offshore, reported more than twice as many assets five years later. This had the added effect of increasing the amount of income taxes paid by the top 0.1% of income earners by 60% between 2015 and 2021. Having broadened the wealth tax base, Argentina sought to maximize its potential by implementing higher tax rates on offshore wealth. In 2019, the government increased the top marginal tax rate for foreign assets to 2.25%, the highest rate observed in the past three decades, while the rate for domestic assets remained at 1.25%. In the final part of the paper, we explore the outcomes of this tax rate hike, specifically focusing on the progressivity of the tax system, revenue generation, and the repatriation of foreign capital. We find that the combination of improved reporting of foreign assets and the higher tax rates on these assets enhanced tax progressivity, resulting in a substantial rise in the effective tax rate for the wealthiest 0.1% of adults. As a result, Argentina’s wealth tax generated significant revenue, amounting to nearly 0.8% of GDP in 2019, establishing Argentina’s wealth tax as one of the most successful globally in terms of revenue generation. Interestingly, the higher tax rates on foreign assets influenced the behavior of indi- viduals in the top 0.1%. Some of them opted to decrease their reported assets and bring back a portion of their offshore holdings. However, for the majority of Argentines outside this wealthiest 0.1%, it seems they chose to keep their assets abroad. This implies that tax- ation was not their primary motivation for holding offshore assets. Instead, it is likely that their reasons encompass factors such as seeking a form of financial security against eco- nomic volatility, fluctuations in exchange rates, risks associated with inflation, currency controls, and the pursuit of potentially higher investment returns. Our research indicates that policy packages can make a meaningful difference in combating tax evasion, offering valuable insights for policymakers in countries adopting similar combinations of international enforcement measures. Specifically, the 2016 en- forcement policy appears to have been more effective than previous initiatives because it was introduced alongside Argentina’s announcements of international tax coordination efforts, including automatic TIEAs with critical partners such as Uruguay, Switzerland, and the US and active involvement in the CRS. These announcements, combined with the Panama Papers leak in the same year, likely created the perception among taxpayers that opportunities for tax evasion had substantially diminished compared to previous years. Moreover, the policy’s success can be attributed to its appeal to the elite. It was supported by generous tax incentives, including a temporary reduction in the wealth tax rate, and was 3 backed by an effective communication campaign led by a pro-market, business-friendly government. In light of our findings, we offer a clear recommendation to policymakers: a tax amnesty program should be accompanied by additional enforcement measures, well- structured tax incentives, and robust advertising efforts to maximize participation rates. In addition to these policy recommendations, our findings also contribute to a grow- ing scholarly literature on offshore evasion (Alstadsæter et al., 2019; Guyton et al., 2021; Johannesen et al., 2023; Zucman, 2015) and policies to combat this kind of evasion, includ- ing voluntary disclosure programs and international tax agreements (Alstadsæter et al., 2022; Johannesen et al., 2020; Langenmayr, 2017; Leenders et al., 2023; Londoño-Vélez and Ávila-Mahecha, 2021). Argentina is an interesting laboratory because it offers the world’s largest natural experiment with tax amnesties. Notably, the most recent policy was im- plemented alongside an announcement of enhanced tax coordination with the US and tax havens or low-tax jurisdictions. Our findings demonstrate that substantial disclosures of offshore assets can take place in this current landscape of global tax coordination, even in a country with a lot at stake and a history of unsuccessful amnesty programs. The remainder of this paper is organized as follows. Section 2 describes Argentina’s wealth tax system and its recent enforcement initiatives, as well as the data and methodo- logy used for our analysis. Section 3 presents the results on the effectiveness of Argentina’s enforcement initiatives in revealing hidden assets, especially those of wealthy taxpayers, and expanding the tax base. Section 4 discusses how Argentina leveraged this expanded tax base by raising tax rates on offshore wealth starting in 2019 and presents its impacts on tax progressivity, tax revenue, and repatriation decisions. Finally, Section 5 concludes. 2 Context, data, and methodology 2.1 Wealth taxation in Argentina Argentina has levied a recurrent wealth tax on individuals and firms since 1991 (Law 23.966). Unlike other wealth-taxing countries, Argentina taxes all gross assets and does not allow discounting debt from the wealth tax base. The tax base includes all world- wide assets—that is, assets held domestically and abroad—on 31 December. The broad tax base includes real estate, vehicles, boats, foreign currency, cash, checking account bal- ances, shares, and some securities. There are two exemptions during our study period: (1) savings accounts and term deposits held at Argentine banks and (2) securities, bonds, or other negotiable instruments issued by the public sector. Figure 1, which plots Argentina’s wealth tax schedule since 1991, shows signifi- 4 cant variation in who pays the wealth tax and the wealth tax rate, resulting from Ar- gentina’s frequent tax reform episodes and high-inflation spells generating substantial ‘bracket creep.’ For example, between 2007 and 2015, Argentina’s annual inflation rate ranged from 10% to 40% (Figure A.2). Argentina nominally defined the wealth tax’s fil- ing threshold and exemption threshold, so inflation tripled the number of taxpayers filing and paying the wealth tax during that period, peaking at over 750,000 individuals or 2.5% of all adults aged 20 and above (Figure A.3). Additionally, Figure 1 illustrates Argentina’s wealth tax rates, which have fluctuated between 0.25% and 2.25%. From 2007 and 2015, the wealth tax schedule consisted of four tax rates spanning from 0.5% to 1.25%. In 2016, the Macri administration introduced a tax reform that streamlined the rates into a single rate of 0.75% in 2016, 0.5% in 2017, and 0.25% in 2018. However, in December 2018, the same administration rescinded its promise to eliminate the wealth tax, replacing the single tax rate with three progressive rates ranging from 0.25% to 0.75%, which would take effect in 2019. Subsequently, the new Fernández administration altered the wealth tax schedule in 2019, introducing eight new rates based on the asset’s location and increasing the tax rates. Notably, the top rate was set at 2.25% for foreign assets, whereas the maximum rate was 1.25% for domestic assets. In 2021, the administration aimed to narrow the gap in tax treatment between foreign and domestic assets by raising the top tax rates on domestic assets. 2.2 A brief history of Argentina’s recent experience with amnesties Argentina has a history of tax amnesties, varying substantially in policy design, contextual features, and revenue collection. Right- and left-wing governments have implemented five different tax amnesties since the country’s return to democracy in 1983. We focus on Argentina’s last three amnesties, which took place within seven years: the Fernández de Kirchner administration implemented two amnesties in 2009 and 2013–15, and Macri implemented one amnesty in 2016. As summarized in Table A.1, these amnesties differed in their effectiveness and how much revenue they collected: the Fernández de Kirchner amnesties revealed assets worth 0.5–1.3% of GDP but had little impact on tax revenue. In comparison, Macri’s program disclosed assets worth 21% of GDP and raised 1.8% of GDP in revenue from penalties. Furthermore, the three schemes varied radically in their scope, penalty rate, repatriation requirement, the availability of cross-country TIEAs, and whether compliant taxpayers were awarded tax benefits, among other things. We describe Macri’s 2016 amnesty program in the remainder of this section and will compare it with the previous two amnesties in Section 3.4. 5 Passed on 29 June 2016, Macri’s temporary tax amnesty took place for nine months, from August 2016 to March 2017 (Law 27.260). It allowed Argentine residents and com- panies to disclose undeclared foreign or domestic assets and currencies held as of 22 July 2016. The program granted participants tax and non-tax benefits. Before the amnesty, evaders caught cheating on their wealth and income tax duties paid 2–10 times the taxes evaded and could be subject to imprisonment. By contrast, the amnesty established a lower rate, depending on the asset type, size, and disclosure date. Specifically, real estate assets paid 5% of the asset’s value.2 For all other assets, the penalty varied with the dis- closed amount: 0% if less than US$19,000, 5% between US$19,000 and US$50,000, and 10% above US$50,000. (The latter increased to 15% for assets disclosed after 31 December 2016 to encourage early participation.) However, participants could waive this one-time tax by investing one-third of the disclosed assets in special Treasury bonds or domestic mutual funds for five years. In addition, the program forgave all liability for taxes and fines and granted participants protection from most types of legal prosecution. We highlight four features of Macri’s amnesty program. First, the program rewarded compliant taxpayers to safe keep tax morale while slashing wealth taxes to entice evaders to come forward. On the one hand, the government rewarded so-called ‘compliant’ taxpayers—those who filed the wealth tax in 2014 and 2015 and did not participate in the amnesty—by exempting them from the wealth tax in 2016, 2017, and 2018. On the other hand, the government lowered the wealth tax rate for amnesty participants: the average tax rates were replaced by marginal tax rates and slashed from 1.25% in 2015 to 0.75% in 2016, 0.5% in 2017, and 0.25% in 2018 (Figure 1). Furthermore, there were talks about repeal- ing the wealth tax for all taxpayers starting in 2019. At the time, Macri’s commitment to reduce and eventually abolish wealth taxation seemed credible: Macri represented a new pro-market and business-friendly government, supported by Argentina’s elite (Sturzeneg- ger, 2019). Second, the Argentine government used the amnesty program’s ‘special tax’ rev- enue to fund its public pension system. Officially named the ‘National Program of His- torical Reparation for Retirees,’ the program earmarked its revenue to finance reparations to pensioners for unpaid benefits, increase some existing benefits, and fund a new non- contributory pension. In practice, the first chapter of the tax bill restored pension benefits between 1995 and 2008 for approximately 2.3 million people. The second chapter of the bill, called ‘Tax Amnesty Regime,’ established the amnesty program to generate the rev- 2 Figure A.4 presents an advertisement used by AFIP (Administración Federal de Ingresos Públicos) to encourage participants to disclose under the amnesty program. The ad compares the penalty using an Arg$3 million property unreported for five years as a hypothetical example: only Arg$150,000 under the amnesty, com- pared to Arg$6 million outside the amnesty. 6 enue needed to fund pension debts and benefits. To illustrate this, Figure A.6 shows an advertisement used by AFIP for this purpose, encouraging Argentines to report their as- sets to “contribute to the country" and “achieve better pensions." Third, the amnesty program was salient. Argentina’s tax authority (Administración Federal de Ingresos Públicos, or AFIP for its Spanish acronym) led a massive advertising campaign. For instance, three large banners encouraging evaders to disclose their hidden assets were hung at the entrance of AFIP’s headquarters in Buenos Aires (Figure A.5). Moreover, AFIP’s website featured advertisements promoting disclosures of hidden assets (Figures A.6 and A.7). Fourth, the threat of detection became more credible after Argentina signed numer- ous tax information exchange agreements in 2016, including treaties with its most relevant tax havens, like Uruguay and Switzerland, as well as with Brazil, Chile, and the US (Fig- ure A.8 plots a timeline of these events). In addition, the Panama Papers were leaked two months before the adoption of the amnesty program, further raising the perceived threat of detection and its salience, as proxied by Google’s search interest (Figure A.9). The official reports by AFIP, reproduced in Table A.2, shed light on the magnitude of disclosures. Nearly 255,000 people and firms participated in Macri’s amnesty program. Participants revealed assets worth US$117 billion under the scheme, representing 21% of Argentina’s GDP in 2016. The success of Argentina’s amnesty exceeded the government’s initial revenue projections by sixfold (Telam, 2016). Four-fifths of the disclosed assets were abroad, and the remainder in Argentina. Nearly half of the assets disclosed were foreign stock and other investments (10% of GDP). Almost one-quarter represented de- posits in foreign bank accounts and currencies (5% of GDP). About 6% corresponded to undeclared cash (1% of GDP).3 Lastly, almost one-fifth came from real estate (4% of GDP), corresponding to 167,000 previously hidden properties. The penalties raised US$9.5 bil- lion in revenue, equivalent to 1.8% of GDP (AFIP, 2017). The amount disclosed and rev- enue collected from penalties place Argentina’s 2016 program as one of the world’s most successful tax amnesties. 2.3 Data and methodology We use information from statistical yearbooks provided by AFIP, representing detailed tabulated data from tax returns for the country’s wealth, income, value-added, and pay- roll taxes for FYs 2002–21. Our primary analysis uses data from the wealth tax, including the tabulations with information on the number of filers and taxpayers, the wealth value, 3 Participants deposited this money in a special bank account and reported it to the tax authority. 7 the tax base, and the tax liability. These tabulations also decompose this information by gender, the location of the asset (domestic versus foreign), the type of asset, the industry sector, and many wealth brackets. In addition, we use information from the personal in- come tax and its four components: rental income, capital income, business income, and labor income. The income tax tabulations include the number of filers and taxpayers and information on the asset value, debts, and net worth. This information is also reported by brackets of total income. The extensive scale of Argentina’s asset disclosures and the detailed tax tabula- tions allow us to utilize straightforward methodologies to unpack the effects of the pol- icy changes. Furthermore, we examine the distributional implications of the enforcement policies using the Pareto interpolation methodology to characterize, visualize, and esti- mate distributions of income or wealth. In particular, we use the generalized Pareto in- terpolation method developed by Blanchet et al. (2022) to flexibly reconstruct continuous distributions of income and assets and obtain precise series of the income and asset dis- tribution. This method is substantially more precise than the alternatives commonly used in the literature and can often be more precise than non-exhaustive individual microdata. 3 The effectiveness of Argentina’s enforcement initiatives This section examines the effectiveness of Argentina’s tax enforcement initiatives. Section 3.1 sheds light on the prevalence and nature of tax evasion. Section 3.2 discusses the dis- tributional patterns of tax evasion. Sections 3.3 analyzes the fiscal externalities that asset revelations have on the wealth and capital income tax bases. Finally, Section 3.4 discusses why the 2016 policy was so much more effective than previous enforcement efforts. 3.1 Revealing foreign and domestic assets Figure 2 illustrates the number of individuals who filed wealth tax declarations. The series is indexed to 100 in 2015, before the 2016 amnesty. This figure reveals several noteworthy findings. First, neither the 2009 nor the amnesties in 2013–15 had any noticeable impact on the number of people reporting their assets to the tax authority. In stark contrast, there was a remarkable 310% surge in the number of taxpayers who declared ownership of foreign assets in 2016. This substantial increase reflects the success of Argentina’s tax amnesty in encouraging individuals who possessed foreign assets to disclose them. Second, this surge in the reporting of foreign assets continued even five years later, while the number of 8 taxpayers reporting assets within Argentina remained relatively constant after 2016.4 This pattern highlights that offshore evasion had been the predominant form of tax evasion in the country. In addition, tax evaders disclosed a significantly higher value of offshore assets. Fig- ure 3 presents the cumulative value of both domestic and foreign wealth in constant 2015 pesos. Figure 3(a) shows this series indexed to 100 in 2015, while Figure 3(b) provides the absolute values. The value of domestic and foreign assets remained relatively stable between 2002 and 2015. However, foreign assets saw a substantial 311% increase in 2016, quadrupling from Arg$250 billion in 2015 (4.3% of GDP) to Arg$1 trillion in 2016 (16.5% of GDP) and further climbing to Arg$1.25 trillion in 2019 (20.7% of GDP). In contrast, domestic assets only experienced a modest 13% increase (Table 1). Furthermore, foreign assets accounted for 83% of the overall increase in reported assets between 2015 and 2016. As a result, by 2019, Argentine wealth taxpayers declared owning an equivalent amount of wealth both domestically and offshore. Figures 4 and 5 break down these findings based on the type of assets. In 2016, there was a substantial increase in the number of individuals who reported owning for- eign stocks and real estate, with a growth of nearly 500% compared to 2015. Similarly, the number of people declaring ownership of foreign bank deposits, currencies, real rights, and vehicles also saw significant growth, ranging from 150% to almost 400%.5 Addition- ally, the value of reported foreign real estate surged by an astonishing 1044% in compar- ison to 2015. Furthermore, the declared holdings of foreign stock, bank deposits, cur- rencies, real rights, and vehicles increased by 366%, 344%, 341%, and 230%, respectively (Table 1). Although the absolute increase for domestic real rights and stocks was minimal and accounted for just 0% of the total asset increase, Figure 5 illustrates that their rise was substantial relative to their initial, smaller base. Figure 6 highlights that the most significant absolute increase in the total assets re- ported to the Argentine tax authority came from foreign stocks and investments, bank deposits, and cash holdings. Notably, the amount declared of foreign stocks stands out, constituting nearly half of the entire change in reported assets between 2015 and 2016 (Table 1). This figure is remarkable, with over Arg$400 billion or approximately US$30 4 Figure A.10 displays the number of tax returns reporting foreign assets in levels. The number of wealth tax returns reporting foreign assets quadrupled from 28,816 to 118,368 between 2015 and 2016. 5 Real rights give holders a right to do something with or on the subject property (stronger than the owner’s right) and include ownership, use, pledge, usufruct, mortgage, and predial servitude. 9 billion disclosed to the authorities in foreign stocks, equivalent to 9.9% of the 2016 GDP.6 In a similar vein, nearly a quarter of the overall change in reported assets from 2015 to 2016 stemmed from foreign bank accounts, amounting to over Arg$360 billion or around US$25 billion. Another quarter of the total change in reported assets is attributed to real estate, evenly split between foreign and domestic properties. To our knowledge, no other tax amnesty, whether attempted by any previous Argentine administration or in another country, has yielded such a substantial amount of disclosed assets. A substantial portion of these assets were concealed in countries traditionally re- garded as tax havens or those with low tax rates, such as Switzerland, the British Virgin Islands, and Uruguay. Perhaps surprisingly, a significant portion—30% of foreign stocks, 45% of foreign bank accounts, and 37% of foreign real estate—was actually situated in the US (Figure A.11). 3.2 Disclosures by top wealth groups Wealth disclosures can have significant distributional implications if tax evasion is con- centrated among the wealthiest individuals. To explore this possibility, we analyze the wealth distribution and compare the average assets owned by the wealthiest Argentines over time. More specifically, we rank individuals based on the assets they report each year (meaning the groups consist of different people each year) and compare the average re- ported assets for each group before and after 2016. To examine the extreme upper end of the distribution, we break down the top 2% of tax units, which includes individuals aged 20 and above, into bins of increasing assets all the way up to the top 0.01%: P98 to P99, P99 to P99.5, P99.5 to P99.9, P99.9 to P99.95, P99.95 to P99.99, and P99.99. To assess which groups experienced the most significant increases in reported assets, we begin by comparing the assets declared annually by different percentile groups before and after 2016. Figure 7 shows that individuals below the top 1% displayed a moderate 13% increase in their average declared assets in 2016 compared to 2015. However, the wealthiest 0.5% of taxpayers reported a considerably higher level of capital after 2016, with a particularly noteworthy increase among the top 0.1%, where their assets more than doubled. Indeed, the increase in reported assets is highly concentrated at the top, with the wealthiest 0.1% accounting for almost two-thirds of the overall increase between 2015 and 6 These values, measured in constant 2015 pesos, are roughly equivalent to Arg$755 billion or US$50 billion in 2016 pesos, which aligns more closely with the official figures reported by AFIP (Table A.2). It is important to note that our data is based on individuals who filed the wealth tax returns. but individuals below the wealth tax filing threshold can also voluntarily disclose their assets. Furthermore, the figures indicate that the majority of disclosures occurred within the context of the tax amnesty program, which is in contrast to the ’quiet disclosures’ that happened in the US, as reported by Johannesen et al. (2020). 10 2016 (Table 2). Furthermore, by 2019, Figure 7 shows that these high-wealth individuals continued to report significantly larger assets, with their declared holdings nearly tripling compared to 2015. One concern in this analysis revolves around the possibility that disclosing assets might lead to a reshuffling of individuals across different percentile groups over time. To address this potential concern, we utilize data from Argentina’s income tax returns, where individuals must report assets held in both the current and previous tax years. This approach lets us compare asset changes while keeping individuals’ income rankings constant.7 The results are depicted in Figure 8, highlighting that the most significant in- creases in reported assets are primarily observed among individuals in the top 0.1% of the income distribution. In this group, reported assets more than doubled, with these individ- uals reporting roughly one-third more assets in 2016 than in 2015. Interestingly, the scale of the asset increase reported by Argentina’s richest individuals aligns with what has been observed in other regions such as Colombia (Londoño-Vélez and Ávila-Mahecha, 2021), Scandinavia (Alstadsæter et al., 2019), and the Netherlands (Leenders et al., 2023). Lastly, Figure A.13 compares high-net-worth individuals’ likelihood of reporting a foreign asset, and the share of foreign assets declared. This information is presented sep- arately by the top fractile groups, decomposing the top 1% of the wealth distribution into bins of increasing assets.8 High-net-worth Argentines held most of their wealth offshore: three-quarters of individuals in the top 0.01% reported a foreign asset in 2015, and their foreign assets comprised over two-thirds of all their assets. Moreover, after 2016, virtu- ally all individuals in this group reported owning foreign assets, and foreign assets rose to represent four-fifths of all assets. Furthermore, a minority of individuals in the next 0.09% declared their foreign assets to Argentine authorities in 2015, so they experienced the largest increases in foreign asset disclosures. Overall, the share of foreign assets de- clared by the top 0.1% tripled. 3.3 Expanding the wealth and capital income tax bases The above results showed that the 2016 enforcement initiatives revealed substantial as- sets held by Argentines domestically and offshore. This section shows that, as a result, 7 For example, the 2016 income tax tabulation, which ranks individuals based on their income, includes data on assets held in 2015 and 2016 for each income bracket. 8 Argentina’s wealth tabulations rank individuals based on total assets. To recover the share of foreign assets (and the effective tax rate) for each fractile, we cumulate the amount of foreign assets (and wealth taxes) by total assets, interpolate the cumulative function for each fractile with a smooth cubic spline function, and differentiate the interpolated function. Reassuringly, a linear interpolation delivers virtually the same results. 11 Argentina’s wealth and capital income tax bases dramatically expanded. Figure 9 plots the total value of wealth reported by wealth tax filers in constant 2015 pesos. Total declared wealth increased by 60% in 2016 compared to 2015, from Arg$1,500 billion to Arg$2,400 billion (Figure A.14) or from US$116 billion to US$186 billion using the market exchange rate (Figure A.15). Moreover, declared wealth remained more than 50% greater five years later. The persistent effects of wealth disclosures can be attributed to the fact that wealth, unlike income, is a stock. This enables authorities to compare reported amounts in different years. As a result, once an asset is disclosed, it becomes risky for the taxpayer to backtrack and underreport it (Garbinti et al., 2023; Londoño-Vélez and Ávila- Mahecha, 2023). Ceteris paribus, an expanded tax base will boost the wealth tax revenue. However, Argentina combined the 2016 amnesty program with (1) an exemption of ‘compliant’ tax- payers from the wealth tax, (2) a progressive reduction of wealth tax rates, (3) a switch from average to marginal rates, and (4) higher filing thresholds. Therefore, we simu- late the counterfactual revenue authorities would have mechanically collected without the 2016 policy to examine this effect. We assume that, first, declared wealth would have remained the same in constant pesos in 2016 as in 2015, absent the 2016 enforcement ini- tiatives. This assumption is plausible, as the stock of reported wealth evolved stably in the 14 years preceding these initiatives (Figure 9). Next, we compute the 2016 wealth tax base, defined as wealth exceeding the exemption threshold, by subtracting 2016’s new ex- emption threshold of Arg$800,000 from the simulated amount of reported wealth. Lastly, we multiply the simulated wealth tax base by the 2016 wealth tax rate of 0.75% to obtain the counterfactual wealth tax revenue and perform a similar procedure for 2017 and 2018, when the tax rate was 0.5% and 0.25%, respectively.9 The results of this exercise are pre- sented in Figure 10(a). The figure indicates that Argentina’s enforcement initiatives more than doubled the wealth tax revenue by 165–180% from 2016 to 2018. To estimate the wealth tax revenue lost due to tax evasion in Argentina, as revealed in 2016, we perform a similar calculation. We start by summing up the value of assets disclosed within each tax bracket for the year 2016. Then, we assume that these assets should have been reported in previous years. To achieve this, we adjust the tax brackets for inflation and apply the corresponding tax rates for each bracket and each tax year. The results of this analysis, depicted in Figure 10(b), reveal the amount of wealth tax revenue that Argentina could have collected if the assets disclosed in 2016 had been reported and 9 We ignore taxpayers responding to the change in the wealth tax rates by changing their reported wealth. If taxpayers respond to the reduced tax rates of 2016, 2017, and 2018 by reporting more wealth, this could confound part of the revenue effect. Similarly, the wealth tax hike in 2019 might induce some taxpayers to report less wealth, meaning tax revenue would have been higher absent the tax change. 12 taxed in previous years. By comparing the red and blue series, our calculations indicate that Argentina could have collected approximately 75% more wealth tax revenue in 2015 if these assets had been reported as they should have been. In terms of the total revenue loss for the period spanning from 2002 to 2015, this amounts to USD 8.4 billion or 1.8% of GDP in 2015. Interestingly, the penalty for participating in the 2016 amnesty contributed roughly 1.8% of GDP. This suggests that Argentina, through the amnesty fee, managed to recoup the entire sum of forgone revenue from the period spanning 2002 to 2015. Additionally, the improved compliance with the wealth tax should lead individuals who had previously concealed their assets to declare the return of these assets, leading to an increase in reported capital income. According to Argentine law, individuals are required to report both their foreign and domestic income in their personal income tax returns. They are subject to taxation based on their worldwide income and may obtain a foreign tax credit for taxes paid on income from foreign sources. Consequently, we antic- ipate that asset disclosures can potentially raise capital income taxation. To gauge the impact on compliance with the capital income tax, we turn to the per- sonal income tax data. Since participants in the tax amnesty disclosed their previously undeclared assets held as of July 22, 2016, and the amnesty continued until March 2017, we would expect reported capital income to start increasing from 2016 and fully reflect the asset disclosures by 2017. Figure 11(a) compares the number of taxpayers reporting some capital income, while Figure 11(b) compares the total reported capital income amount. The data shows no sig- nificant changes in these series before 2016, followed by a substantial increase in reported capital income starting in 2016. Specifically, the number of taxpayers reporting some capi- tal income doubled, and the capital income tax base tripled during this period. In contrast, the other three sources of income—wages, rental income, and business income—remained relatively unchanged after 2016. These patterns align with the notion that foreign and do- mestic capital, which previously generated taxable income but had been left undeclared before the amnesty, is reported more accurately following the program. Importantly, these improvements in reporting persisted for at least five years later.10 Given that most capital disclosures were made by individuals in the top 0.1% of the 10 Despite foreign rental income being subject to taxation in Argentina, Figure 11 reveals that there was no notable surge in reported foreign rental income after 2016. This can be attributed to multiple factors. Firstly, as demonstrated in Section 3.1, the disclosed foreign real estate’s absolute value is relatively small, with most disclosed foreign assets consisting of stocks, investments, and bank deposits. Secondly, given that most rental income typically originates within the country, the limited amount of foreign rental income disclosed may not be substantial enough to significantly impact the total rental income illustrated in Figure 11. Lastly, it is plausible that many foreign properties did not generate rental income; for example, they may have been unoccupied or not rented out during the relevant period. 13 income distribution, one could anticipate that most revelations of capital income would also be concentrated within this group. This is indeed reflected in Figure 12(a), which tracks the capital income share over time. The figure breaks down the top 1% of the in- come distribution into three subgroups, ranked by decreasing income: the top 0.1%, the following 0.4% (P99.5–P99.9), and the next 0.5% (P99–P99.5).11 In 2015, around 5% of the income for the top 0.1% was derived from capital. However, by 2017, this percentage had tripled to 15% as these individuals reported more capital income. In contrast, the capital income share remained relatively stable for individuals below the top 0.1%. Remarkably, five years after the amnesty, individuals in the top 0.1% of the income distribution saw their capital income share exceed 20%, signifying improved income tax compliance at the top. As a consequence of the increased disclosure of capital income, Figure 12(b) demon- strates that the income tax paid by the top 0.1% surged by 60% during the period from 2015 to 2021. Consequently, the top 0.1% of income earners contributed more to the total personal income tax burden after 2016. While the top 0.1% contributed a quarter of all personal income taxes in 2015, this proportion increased to approximately 40% in 2021 (Figure A.16). These results emphasize the significant impact of disclosing assets in in- creasing income tax revenue collected from the country’s richest individuals. 3.4 Discussion Why was the 2016 combined policy so much more effective than previous enforcement efforts? We highlight four main features of the former’s policy design and context. 1. The perceived threat of detection. Tax evaders are more likely to follow tax regulations if they believe there is a significant risk of being caught cheating. This risk depends on the government’s ability to detect offshore tax evasion. Automatic TIEAs have been praised as effective tools for improving evasion detection, thereby making the threat of detection more credible. As a result, governments may introduce voluntary disclosure programs to capitalize on the increased cooperation between tax administrations and the availability of information about financial accounts held abroad (OECD, 2015). However, during the 2009 and 2013-2015 amnesty periods, the detection threat was not credible in Argentina. This was because Argentina had not signed TIEAs before 11 In contrast to wealth tax returns, income tax returns lack detailed disaggregation for income groups beyond the top 0.1% of the income distribution in 2017 and 2018. Consequently, we do not provide detailed break- downs beyond the top 0.1%. Additionally, it is worth noting that Argentina’s income tax filing thresholds remained unadjusted for inflation until 2016. This led to a larger number of individuals below the top 1% of income earners being required to file income tax returns during the period from 2010 to 2015. As a result of this shift in the composition of tax filers below the top 1%, we present information for percentiles 99 and above. 14 2014, making it difficult for authorities to track foreign financial accounts. In contrast, Argentina became part of the OECD Automatic Exchange of Information initiative in October 2014 and committed to exchanging information through the CRS by Septem- ber 2017. Furthermore, in 2016, Argentina signed bilateral TIEAs with key tax haven countries such as Uruguay and Switzerland, as well as with the United States, Chile, and Brazil. The announcement of these agreements played a crucial role in the tax au- thority’s efforts to encourage participation in the 2016 amnesty program and featured prominently in the administration’s advertisement campaign (Figure A.8). Argentina’s TIEA announcement may have compelled evaders to declare their as- sets truthfully. Interestingly, a significant number of assets disclosed under the 2016 amnesty program were found to be located in these jurisdictions: 71% of foreign stocks were disclosed in the US, Switzerland, and the British Virgin Islands, 86% of foreign bank accounts were disclosed in the US, Switzerland, and Uruguay, and 86% of for- eign real estate was disclosed in Uruguay and the US (Figure A.11), consistent with taxpayers responding to the perceived threat of detection. Additionally, the release of the Panama Papers a few months before the amnesty program began may have fur- ther heightened the perception that the opportunities for tax evasion had significantly narrowed in 2016 compared to previous years. However, the increase in taxpayers’ willingness to comply with tax regulations due to automatic TIEAs cannot fully explain the substantial rise in tax compliance. There are several key reasons for this. First, more than 3% of Argentina’s GDP was disclosed in domestic assets, even though there were no changes in domestic enforcement mea- sures. Second, Argentines disclosed 2% of GDP in foreign real estate assets (Table A.2) even though the CRS and other automatic TIEAs worldwide currently focus solely on financial accounts and exclude real estate from the scope of the information exchange. Third, approximately 5.8% of Argentina’s GDP was disclosed in assets held in the US, even though the US Internal Revenue Service only shares information related to tax- able income, with no effective enforcement mechanism in place for other disclosures. Fourth, the actual impact of the automatic TIEAs on detecting and penalizing evasion remained minimal until after the tax amnesty concluded in 2017 since CRS and all bi- lateral tax agreements did not become operational before 2018. In fact, it would take several years before the TIEAs would be used by Argentina to enforce taxes (La Nacion, 2021). In summary, while the perceived threat of detection may have played a role, it is possible that taxpayers react to the mere announcement of a TIEA, regardless of its 15 direct impact on their specific assets. This behavior aligns with findings by Bergolo et al. (2023) showing that taxpayers respond to the perceived threat of detection regardless of the actual probability of being caught. 2. The tax incentives. Tax administrations often create voluntary disclosure programs that offer tax benefits to encourage tax evaders to come forward without negatively af- fecting the tax morale of compliant taxpayers. However, if there is a perception that tax evaders can receive more favorable terms through these programs than honest tax- payers, it might unintentionally lead to increased non-compliance (Langenmayr, 2017; OECD, 2015). In the case of Argentina’s 2016 amnesty program, a balance was struck by pro- viding attractive incentives to encourage tax evaders to participate while ensuring the support and compliance of honest taxpayers. On one hand, the program offered a gradual reduction in wealth tax and even proposed abolishing it from January 2019 onwards. This was intended to entice tax evaders to participate because their partici- pation would not offset their future wealth tax obligations. At that time, the govern- ment’s commitment to reducing and eventually eliminating wealth taxation seemed credible, as Macri’s government was perceived as new and pro-market, gaining wide acceptance among Argentina’s elite. On the other hand, the program imposed the high- est penalty rate, reaching up to 15%, in contrast to the 8% rate in 2009 and 0% in 2013-15 (Table A.1).12 Additionally, the government rewarded ‘compliant’ wealth taxpayers by exempting them from the wealth tax in 2016, 2017, and 2018. This combination of fea- tures may have led to a high level of participation in the program without undermining the compliance of non-evaders. 3. A favorable political economy. Political alignment and individuals’ attitudes toward the government can play a significant role in shaping taxpayers’ decisions regarding tax evasion (Cullen et al., 2021). During the 2009 and 2013-2015 tax amnesties, Argen- tines had low confidence in the left-wing government of Fernández de Kirchner (Fig- ure A.20). There was a noticeable increase in confidence when Macri assumed office as president. Furthermore, this confidence remained high when Macri introduced the amnesty program shortly thereafter. Unlike his predecessor, the fact that a pro-market and business-friendly president was implementing the program may have encouraged 12 Indeed, there is evidence that evaders are sensitive to penalty rates: most disclosures of assets took place in December 2016, before the highest penalty fee would increase from 10% to 15% (Figure A.19). 16 wealthy Argentines to participate.13 Additionally, Argentina used the amnesty program as a means to generate revenue to support increased pension benefits for elderly citizens. The legislation specified that revenue from the amnesty’s ‘special tax’ would be allocated to fund these retirees, as we analyze in Appendix B. The government actively promoted this idea as part of its campaign to motivate tax evaders to come forward, framing the amnesty program as a way for people to contribute to improving pension benefits for senior citizens (as shown in Figure A.6). Consequently, this earmarking of funds may have garnered taxpayer support for the amnesty program. 4. High salience and low compliance costs. Argentina aimed to boost participation in the tax amnesty through an extensive information campaign. Furthermore, authori- ties took significant steps to simplify the disclosure process, making it straightforward for individuals to participate. For instance, they provided detailed, step-by-step guide- lines on taking part, shared instructional videos on platforms like YouTube, and even developed a dedicated app to help people calculate their participation tax penalty (as shown in Figure A.7). These efforts likely played a significant role in driving high levels of participation. 4 Increasing tax rates with an expanded tax base Having successfully obtained substantial disclosures of foreign assets, Argentina sought to make the most of this expanded tax base by raising tax rates on offshore wealth. This change began in 2019 when the wealth tax schedule was adjusted based on the asset’s loca- tion. Notably, the top marginal tax rate for foreign assets was increased to 2.25%, marking the highest rate seen in the past three decades. In contrast, it was 1.25% for domestic assets. The following sections delve into the repercussions of these changes. Section 4.1 will fo- cus on how these adjustments influenced the progressivity of the tax system and revenue collection, while Section 4.2 will explore the impact on taxpayers’ decisions regarding the repatriation of their assets. 13 An illustrative instance of Macri’s pro-market approach is his swift action to remove Argentina’s foreign ex- change controls upon assuming office in December 2015. These controls, initially put in place by Fernández de Kirchner in 2011, restricted Argentines’ ability to purchase or sell foreign currency. In direct contrast to the Kirchners, Macri campaigned with the commitment to eliminate these restrictions promptly as part of his reform agenda aimed at stimulating economic growth. 17 4.1 Tax progressivity and revenue collection Improving the reporting of foreign assets can have important implications for the progres- sivity of the tax system, as offshore wealth is concentrated at the top. To examine changes in tax progressivity, Figure 13 plots the effective wealth rate between 2010 and 2021 by bins of increasing fortune, as defined previously. Improving tax compliance and raising tax rates on foreign assets substantially enhanced tax progressivity. Indeed, the effective tax rate increased significantly for all groups in 2019, but the change was particularly sub- stantial for the wealthiest 0.1% of adults. For instance, the wealthiest 0.01% experienced an eightfold increase in their effective tax rate, which jumped from 0.25% in 2018 to 2% in 2019. Similarly, the effective tax rate increased significantly for the next 0.09%. On average, the effective wealth tax rate saw a significant increase, rising from 0.25% in 2018 to 1.46% in 2019 (Figure A.17). This increase had a substantial impact on wealth tax revenue, which escalated from Arg$4.9 billion in 2018 to Arg$35.4 billion in 2019, rep- resenting a shift from 0.14% of GDP to 0.75% of GDP (Figure A.18). Notably, this places Argentina’s wealth tax among the most successful in the world in terms of generating tax revenue. Compared to the anticipated tax revenue in 2019, the significant disclosures of offshore wealth led to a remarkable increase in revenue, surpassing threefold growth, as illustrated in Figure 10.14 The expanded wealth tax base was timely as the COVID-19 crisis struck in 2020. The government used progressive wealth taxation to finance health expenses and expand the social safety net, levying a one-time wealth tax surcharge on the wealthiest 12,500 Argen- tines with assets worth more than Arg$200 million or US$2.4 million (Law 27.605). The marginal tax rates ranged from 2% to 3.5% for domestic assets and 3% to 5.25% for foreign assets. Official reports from AFIP informed that 10,000 people paid the tax by April 2021 with approximately US$80 billion in taxable assets, of which 50% were located abroad.15 This value is more than twice the US$30 billion tax base declared by the wealthiest 11,700 Argentines before the amnesty, based on our tabulations. As a result, the government col- lected US$2.8 billion in revenue, roughly equivalent to one month of Argentina’s value- added tax, the country’s largest revenue source (AFIP, Serie Anual 2021). 14 To simulate the counterfactual tax revenue in 2019, absent the disclosures, we assume that all domestic and foreign assets faced the top tax rates. This assumption is conservative because, in practice, many smaller assets faced lower tax rates, making our counterfactual revenue an upper bound and, correspondingly, our estimated revenue gain a lower bound. 15 The remaining 2,500 non-filers were actively audited by the tax authority and threat- ened with prosecution. About 1,100 taxpayers filed a lawsuit against the government and are currently being treated in court (https://www.telam.com.ar/notas/202112/ 577043-afip-aporte-solidario-extraordinario-recaudacion.html). 18 4.2 Do taxpayers repatriate assets in response to tax incentives? Argentina’s wealth tax schedule and tax allowances generated sizable incentives for the repatriation of assets. As described previously, foreign assets were subject to higher tax rates than domestic assets starting in 2019. Further, on December 28, 2019, the govern- ment announced that individuals repatriating 5% or more of their foreign assets would be subject to the lower domestic wealth tax rate for all of their assets. If Argentine taxpayers responded to these tax incentives by repatriating capital, we would observe a drop in for- eign assets starting in 2020, an increase in domestic assets, and a decrease in the average effective tax rate. Figure 3 provides evidence of asset repatriation. In the years 2020 and 2021, the value of foreign assets decreased, while domestic assets showed a slight increase. However, the growth in domestic assets falls short of fully compensating for the decline in foreign assets, which would be expected in a scenario of complete asset repatriation. Furthermore, as depicted in Figure 7, we see that this reduction in total wealth primarily stems from the top 0.1% of individuals, who faced higher tax rates starting in 2019. Within this high-net- worth group, their proportion of foreign assets decreased while their share of domestic assets increased (Figure A.13), suggesting a pattern of repatriation. Notably, for the top 0.01%, the reduction in total assets is more substantial. While part of this larger decline is mechanical (higher wealth taxes in period t lead to less wealth in t + 1), the reduction in their foreign assets is not accompanied by an increase in domestic assets, and their share of foreign assets remains constant. This suggests that the wealthiest 0.01% responded to the high tax rates on foreign assets by declaring fewer offshore assets, while the next 0.09% responded by repatriating their wealth. Below the top 0.1%, there is little discernible response. Nevertheless, given the significant contribution of the top percentile groups to total wealth tax revenue, their reduced reported wealth resulted in a decline in the wealth tax to GDP ratio after 2019, as shown in Figure A.18. While Argentines may respond to the recent tax incentives to repatriate capital, there is little evidence that they respond to repatriation clauses included in the tax amnesties, on average. In particular, the 2013–15 amnesty waived the tax amnesty participation fee for individuals who invested their disclosures in three Treasury securities. Because Trea- sury securities are exempt from Argentina’s wealth tax, the waived penalty generated a sizable tax benefit for repatriation. Notwithstanding, the amount invested in Treasury se- curities during the 2013–15 amnesty represented only 0.5% of GDP (Table A.1). In fact, 12 times less currency and fewer deposits were disclosed during this amnesty than dur- ing the 2016 scheme, despite being only a couple of months apart. Later, participants of the 2016 amnesty could also waive the penalty by investing one-third of their disclosed 19 assets in two Treasury bonds or domestic mutual funds. However, again, most Argentine evaders chose to pay the special tax and keep their assets abroad, as the head of AFIP himself later acknowledged to the press (Clarin, 2017). Why would Argentines rather pay a hefty wealth tax and keep their assets abroad? Given the seemingly negligible repatriation response for most individuals, tax incentives cannot explain why Argentines have offshore assets. Instead, they might keep their assets offshore because this allows them to insure themselves against economic volatility, cur- rency controls, high exchange rate fluctuations, and inflation spells. In addition, foreign assets might enable Argentines to access financial services they cannot get in Argentina and, presumably, obtain higher (pre-tax) returns. 5 Conclusion We examined recent changes in tax enforcement policy in Argentina, which resulted in an unprecedented disclosure of previously hidden assets. Our findings revealed that Argen- tine authorities successfully encouraged individuals to disclose their wealth, particularly financial assets held abroad. While these disclosures were widespread, they were most significant among the top 0.1% of the nation’s wealthiest individuals. This increase in total assets reported by taxpayers expanded the country’s wealth and capital income tax bases and raised tax revenue among the top 0.1%. The increased tax compliance enhanced the progressivity of the tax system and made Argentina’s wealth tax one of the world’s most successful policies in terms of revenue collection. Notably, despite significant incentives aimed at repatriating foreign assets, we observed minimal repatriation responses, except possibly among a select group of individuals at the very pinnacle of the wealth distribu- tion. Our findings provide valuable insights for developing countries that are exploring strategies for improving their domestic revenue collection, as outlined by World Bank (2023). It is widely recognized that effective tax administration and tax policy are crucial elements of a nation’s development agenda. Our findings highlight that voluntary disclo- sure programs and enforcement policies like those successfully implemented in Argentina can significantly enhance a country’s capacity to mobilize domestic resources through pro- gressive tax measures. This approach effectively achieves the dual goals of increasing gov- ernment revenue and advancing tax equity. Argentina’s diverse experiences with tax amnesties offer valuable lessons for other nations. Our research suggests that successful amnesties tend to share certain common el- ements. These include creating a credible perception of the threat of detection, designing 20 well-structured tax incentives, and conducting substantial advertising campaigns. By con- sidering these factors, other countries can potentially adapt and apply successful strategies to their unique circumstances, ultimately bolstering their domestic revenue mobilization efforts. References AFIP, “Sinceramiento Fiscal: Cierre,” https://cpcecba.org.ar/media/ download/noticias/Sinceramiento.pdf April 2017. Alstadsæter, Annette, Niels Johannesen, and Gabriel Zucman, “Who owns the wealth in tax havens? Macro evidence and implications for global inequality,” Journal of Public Economics, 2018, 162, 89–100. , , and , “Tax evasion and inequality,” American Economic Review, 2019, 109 (6), 2073– 2103. , , Segal Le Guern Herry, and Gabriel Zucman, “Tax Evasion and Tax Avoidance,” Journal of Public Economics, 2022, 206, 104587. Bergolo, Marcelo, Rodrigo Ceni, Guillermo Cruces, Matias Giaccobasso, and Ricardo Perez-Truglia, “Tax Audits as Scarecrows: Evidence from a Large-Scale Field Experi- ment,” American Economic Journal: Economic Policy, February 2023, 15 (1), 110–53. Berniell, Inés, Dolores de la Mata, and Matilde Pinto Machado, “The Impact of a Perma- nent Income Shock on the Situation of Women in the Household: The Case of a Pension Reform in Argentina,” Economic Development and Cultural Change, 2020, 68 (4), 1295– 1343. Blanchet, Thomas, Juliette Fournier, and Thomas Piketty, “Generalized Pareto Curves: Theory and Applications,” Review of Income and Wealth, 2022, 68 (1), 263–288. Bosch, Mariano and Jarret Guajardo, “Labor market impacts of non-contributory pen- sions: The case of Argentina’s Moratorium,” Technical Report, IDB Working Paper Se- ries 366 2012. Cavallo, Alberto and Manuel Bertolotto, “Filling the Gap in Argentina’s Inflation Data,” Available at SSRN: https://ssrn.com/abstract=2782104 May 2016. 21 Cullen, Julie Berry, Nicholas Turner, and Ebonya Washington, “Political Alignment, At- titudes toward Government, and Tax Evasion,” American Economic Journal: Economic Pol- icy, August 2021, 13 (3), 135–66. EU Tax Observatory, “Global Tax Evasion Report,” 2024. Garbinti, B., J. Goupille-Lebret, M. Muñoz, S. Stantcheva, and G. Zucman, “Tax Design, Information, and Elasticities: Evidence From the French Wealth Tax,” Working Paper 31333, National Bureau of Economic Research June 2023. Guyton, John, Patrick Langetieg, Daniel Reck, Max Risch, and Gabriel Zucman, “Tax Evasion at the Top of the Income Distribution: Theory and Evidence,” Working Paper 28542, National Bureau of Economic Research March 2021. Johannesen, Niels, Daniel Reck, Max Risch, Joel Slemrod, John Guyton, and Patrick Langetieg, “The Offshore World According to FATCA: New Evidence on the Foreign Wealth of U.S. Households,” Working Paper 31055, National Bureau of Economic Re- search March 2023. , Patrick Langetieg, Daniel Reck, Max Risch, and Joel Slemrod, “Taxing Hidden Wealth: The Consequences of U.S. Enforcement Initiatives on Evasive Foreign Ac- counts,” American Economic Journal: Economic Policy, 2020, 12 (3), 312–346. Langenmayr, Dominika, “Voluntary Disclosure of Evaded Taxes—Increasing Revenue, or Increasing Incentives to Evade?,” Journal of Public Economics, 2017, 171 (7), 110–125. Leenders, Wouter, Arjan Lejour, Simon Rabaté, and Maarten van ’t Riet, “Offshore tax evasion and wealth inequality: Evidence from a tax amnesty in the Netherlands,” Journal of Public Economics, 2023, 217, 104785. Londoño-Vélez, J. and J. Ávila-Mahecha, “Behavioral Responses to Wealth Taxation: Ev- idence from Colombia,” September 2023. Londoño-Vélez, Juliana and Javier Ávila-Mahecha, “Enforcing Wealth Taxes in the De- veloping World: Quasi-Experimental Evidence from Colombia,” American Economic Re- view: Insights, June 2021, 3 (2), 131–48. Ministerio de Economía, “Impuesto sobre los Bienes Personales: Evolución de Alícuotas, Mínimos no Imponibles y Mínimo Exento,” Technical Report January 2022. OECD, Update on Voluntary Disclosure Programmes: A Pathway to Tax Compliance, OECD Publishing, August 2015. 22 Rottenschweiler, Sergio, “Un mismo comienzo y dos caminos dispares: la Reparación Histórica y la Pensión Universal para el Adulto Mayor (2016-2019),” Revista Latinoamer- icana de Desarrollo Económico, 2020, (34), 67–92. Slemrod, Joel, “Tax Compliance and Enforcement,” Journal of Economic Literature, Decem- ber 2019, 57 (4), 904–54. Sturzenegger, Federico, “Macri’s Macro: The Elusive Road to Stability and Growth,” 2019, 2019 (2), 339–436. Project MUSE. Telam, “El sinceramiento fiscal ya alcanzo los U$S 21.863 millones,” https://www.telam.com.ar/notas/201611/ 171267-sinceramiento-fiscal-blanqueo-capiales-afip-alberto-abad. html November 2016. Tørsløv, Thomas, Ludvig Wier, and Gabriel Zucman, “The Missing Profits of Nations,” The Review of Economic Studies, 07 2022. rdac049. World Bank, World Bank Support for Domestic Revenue Mobilization, Washington, DC: World Bank, 2023. Zucman, Gabriel, “The missing wealth of nations: Are Europe and the U.S. net debtors or net creditors?,” Quarterly Journal of Economics, 2013, 128 (3), 1321—-1364. , The Hidden Wealth of Nations: The Scourge of Tax Havens, University of Chicago, 2015. 23 Figures and Tables Figure 1: Argentina’s wealth tax rates have ranged from 0.25% to 2.25% Millions of Tax Rates (%) 2015 pesos 2.25% 2 18m 1.75% Exemption 6.5m 306m 1.5 cutoff (right) Tax rates 1.5% on foreign assets 106m 1.25% Cutoff: 5m 20m 24m 1 1% Cutoff: 100k Cutoff: 2m 8.5m 12.5m 0.75% Cutoff: 200k Cutoff: 750k 800k 5m 9m 0.5 0.5% Cutoff: 102.3k Cutoff: 102.3k Cutoff: 305k 950k 2m 6m 0.25% 1.05m 0 0 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 Notes: this figure plots the wealth tax schedule in Argentina between 1991 and 2022, showing sizable varia- tion in the exemption cutoff and bracket schedule. The left axis plots the statutory wealth tax rates and as- sociated bracket cutoffs in current pesos. Because these cutoffs are nominally defined, high-inflation spells cause ‘bracket creep’: the exemption cutoff (plotted in the right axis and expressed in millions of 2015 pe- sos) dropped between 2007 and 2015. Moreover, Argentina’s wealth tax rates have historically ranged from 0.25% to 2.25%, with reforms taking place in 1995, 1999, 2007, 2016, 2019, and 2021. For instance, in 2016 Argentina replaced the bracket schedule based on four (average) tax rates with a single (marginal) rate of 0.75% and raised the filing threshold. Besides the exemption threshold, the tax has a filing threshold for people with annual gross income above the following thresholds (in pesos): 2007–14: 96k; 2015: 200k; 2016: 500k; 2017: 1m; 2018: 1.5m; 2019: 2m; 2020: 2.5m; 2021: 3.7m; 2022: 6.6m. Argentina applies differential tax rates on foreign assets since FY 2019. To determine the tax rate on foreign assets, taxpayers must first sum domestic and foreign assets and then apply the corresponding (average) rate on the total value of foreign assets. Domestic rates apply on total foreign assets when taxpayers repatriate at least 5% of those assets. Source: authors’ compilation based on Ministerio de Economía (2022). 24 Figure 2: A 310% increase in the number of taxpayers declaring foreign assets Number of tax returns (2015=100) 2009 2013-15 2016 Amnesty Amnesty Amnesty 400 Assets located abroad 300 200 Assets located in Argentina 100 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Notes: this figure plots the number of taxpayers declaring assets owned domestically and abroad. The series is indexed to equal 100 in 2015, before the 2016 amnesty. Neither the 2009 nor 2013–15 amnesties affected the number of people reporting assets to the tax authority. By contrast, there was a 310% increase in the number of taxpayers reporting to own foreign assets in 2016, which persisted even five years later, consistent with offshore evasion being the primary form of evasion. Source: authors’ calculations using data from the AFIP statistical yearbooks. 25 Figure 3: A more than 310% increase in the value of declared foreign assets (a) Relative values Wealth value (2015=100) 2009 2013-15 2016 Amnesty Amnesty Amnesty 500 400 Assets located 300 abroad 200 Assets located in Argentina 100 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 (b) Absolute values Billions of 2015 pesos 2009 2013-15 2016 Amnesty Amnesty Amnesty 1500 1250 Assets located in Argentina 1000 750 Assets located abroad 500 250 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Notes: this figure plots the total value of declared domestic and foreign wealth in constant 2015 pesos. Panel (a) expresses the series indexed to equal 100 in 2015, while Panel (b) reports the absolute values. The value of domestic and foreign assets is remarkably stable between 2002 and 2015. However, while domestic assets continued in the same trend, the value of foreign assets quadrupled from Arg$250 billion in 2015 to Arg$1 trillion in 2016 and Arg$1.25 trillion in 2019. Source: authors’ calculations using data from AFIP statistical yearbooks. 26 Figure 4: The likelihood of declaring foreign or domestic assets by asset type (a) Foreign assets Number of tax returns (2015=100) 2016 Amnesty 600 Stocks Real Estate 500 Deposits & 400 Currency Real Rights 300 Accounts Receivable 200 Vehicles 100 Other Assets 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 (b) Domestic assets Number of tax returns (2015=100) 2016 Amnesty 600 500 400 300 200 Real Rights Stocks 100 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Notes: this figure decomposes Figure 2 by asset type and plots the number of taxpayers declaring different types of foreign or domestic assets in panels (a) and (b), respectively (2015 = 100). The number of people reporting foreign stocks and real estate increased by nearly 500% in 2016 relative to 2015. Source: authors’ calculations using data from AFIP statistical yearbooks. 27 Figure 5: The relative value of declared foreign or domestic assets by asset type (a) Foreign assets Wealth value (2015=100) 1300 2016 Amnesty 1200 1100 1000 900 Real Estate 800 700 Stocks 600 Deposits & 500 Currency Real Rights 400 300 Vehicles Accounts 200 Receivable 100 Other Assets 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 (b) Domestic assets Wealth value (2015=100) 1300 2016 Amnesty 1200 1100 1000 900 800 700 600 500 400 Stocks 300 Real Rights 200 100 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Notes: this figure decomposes Figure 3 by asset type and plots the relative value of different types of foreign or domestic assets in panels (a) and (b), respectively, in constant pesos (2015 = 100). The value of reported foreign real estate increased by more than 1,000% compared to 2015. Similarly, the number of people re- porting to own foreign bank deposits and currencies, real rights and credits, and cars and boats increased by 380%, 240%, and 160%, respectively. Source: authors’ calculations using data from AFIP statistical yearbooks. 28 Figure 6: The real value of declared foreign or domestic assets by asset type (a) Foreign assets Billions of 2015 pesos 2016 Amnesty 1500 1000 Real Estate Deposits & Currency 500 Stocks 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 (b) Domestic assets Billions of 2015 pesos 2016 Amnesty 1500 Accounts Receivable Business 1000 Vehicles Deposits & Currency 500 Real Estate 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Notes: this figure plots the absolute value of the different types of foreign or domestic assets reported by taxpayers in panels (a) and (b), respectively, in constant 2015 pesos (billions). More than Arg$400 billion (US$30 billion) of foreign equities were disclosed to the authorities, equivalent to 9.9% of GDP in 2016. Likewise, more than Arg$360 billion or US$25 billion deposited in foreign bank accounts were reported in tax returns after the amnesty. Panel (b) shows a drop in domestic real estate after the 2019 reform created a separate wealth tax exemption threshold for primary residences, nine times larger than the threshold for all other assets. Source: authors’ calculations using data from AFIP statistical yearbooks. 29 Figure 7: The increase in reported assets is greater for Argentina’s wealthiest 0.1% Reported assets (2015 = 100) 2013-15 2016 300 Amnesty Amnesty 250 p99.95-p99.99 p99.9-p99.95 200 Top 0.01% p99.5-p99.9 150 p99-p99.5 100 p98-p99 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Notes: this figure compares the assets reported yearly by the wealthiest 2% of adults (aged 20 and above) separately by bins of increasing assets relative to 2015. Individuals below the top 1% had moderate increases in their average assets after the amnesty. By contrast, the wealthiest 0.5% of taxpayers declared substantially more assets after the program. In particular, the rise was remarkable among the top 0.1% who, four years after the amnesty, reported to own two to three times as much assets as before the scheme. Source: authors’ calculations using data from AFIP statistical yearbooks. 30 Figure 8: Assets roughly doubled, and the change is about one-third of their assets (a) Net asset growth between 2015 and 2016 Net YoY Assets Growth (%) 175 150 125 100 75 50 25 0 p98-p99 p99-p99.5 p99.5-p99.9 p99.9-p99.95 p99.95-p99.99 Top 0.01% Top income fractiles (b) Net asset change as a share of assets in 2016 Net ∆ Assets / Total Assets (%) 50 40 30 20 10 0 p98-p99 p99-p99.5 p99.5-p99.9 p99.9-p99.95 p99.95-p99.99 Top 0.01% Top income fractiles Notes: this figure presents the change in assets reported between 2015 and 2016 by the richest 2% of adults (aged 20 and above) separately by bins of increasing income. Panel (a) compares the change in assets between 2015 and 2016 net of the change in assets between 2014 and 2015. Panel (b) expresses this relative to the amount of assets in 2016. Source: authors’ calculations using data from AFIP statistical yearbooks. 31 Figure 9: The 2016 amnesty raised the total value of wealth reported by tax filers Assets value (2015=100) 200 2009 2013-15 2016 Amnesty Amnesty Amnesty 180 160 Reported Assets 140 120 100 80 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Notes: this figure plots the total value of wealth reported by tax filers in constant 2015 pesos (2015 = 100). Total declared wealth increased by 60% in 2016 compared to 2015 (or from US$116 billion to US$186 billion using the market exchange rate of Arg$12.9 to US$1 in December 2015). Moreover, declared wealth remained more than 50% greater five years later. Source: authors’ calculations using data from AFIP statistical yearbooks. 32 Figure 10: The 2016 amnesty raised the wealth tax revenue (a) Wealth tax revenue Billions of 2015 pesos 40 2009 2013-15 2016 2.25 Amnesty Amnesty Amnesty 30 Top wealth tax rate (right) 1.25 20 0.75 10 Wealth tax revenue (left) 0.50 0.25 Counterfactual revenue 0 (using 2015 tax base) 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 (b) Forgone wealth tax revenue Billions of 2015 pesos 40 2009 2013-15 2016 2.25 Amnesty Amnesty Amnesty 30 Top wealth Forgone revenue tax rate (right) (absent evasion) 1.25 20 0.75 10 Wealth tax revenue 0.50 0.25 0 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Notes: this figure plots the wealth tax revenue and two counterfactuals against the top wealth tax rate (right axis). Panel (a) presents the counterfactual revenue absent the 2016 enforcement initiatives, while Panel (b) plots the forgone wealth tax revenue absent evasion, as detected through the 2016 enforcement initiatives. Argentina’s policy package more than doubled the wealth tax revenue by 165–180% from 2016 to 2018. Moreover, the wealth tax revenue increased more than sevenfold from Arg$4.9 billion in 2018 to Arg$35.4 billion in 2019, after the 2019 reform increased the wealth tax rates. Relative to the counterfactual revenue in 2019, Argentina raised tax revenue more than threefold thanks to the prominent disclosures of offshore wealth it induced. Source: authors’ calculations using data from AFIP statistical yearbooks. 33 Figure 11: Capital income tax revenue increased after 2016 (a) Number of taxpayers subject to capital income tax Tax Returns 50,000 2016 Amnesty 40,000 30,000 Capital income (taxpayers) 20,000 10,000 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 (b) Capital income tax base Tax base (2015=100) 600 2016 Amnesty Capital income (tax base) 400 200 Rental income Wage income Business 0 income 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Notes: this figure compares the number of taxpayers subject to the capital income tax and the capital income tax base in panels (a) and (b), respectively. There is a meaningful increase in reported capital income start- ing in 2016 when disclosers are required to register income: the number of taxpayers reporting some capital income doubled, and the capital income tax base tripled. By contrast, none of the other three sources of in- come (wage income, business income, rental income) changed after 2016. These patterns are consistent with foreign and domestic assets, which generated taxable income and were left undeclared before the amnesty, becoming more truthfully reported after the program. Critically, these improvements persisted years after the amnesty program ended. Source: authors’ calculations using data from AFIP statistical yearbooks. 34 Figure 12: Income tax compliance improved for the top 0.1% after 2016 (a) Capital income share Capital Income / Total Income 2013-15 2016 Amnesty Amnesty 20% Top 0.1% 15% 10% 5% p99.5-p99.9 0 p99-p99.5 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 (b) Income tax Income Tax (2015=100) 160 2013-15 2016 Amnesty Amnesty Top 0.1% 140 p99.5-p99.9 120 100 80 p99-p99.5 60 40 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Notes: this figure illustrates the capital income share and the income tax in panels (a) and (b), respectively. Within the top 1% of the income distribution, we have further divided this group into three subcategories based on decreasing income: the top 0.1%, the subsequent top 0.4% (P99.5 to P99.9), and the following 0.5% (P99 to P99.5). Between 2015 and 2021, the capital income share for the top 0.1% of the income distribu- tion saw a remarkable increase, soaring from 5% to over 20%. In contrast, the next 0.4% witnessed a more modest rise in their capital income share, while the capital income share remained relatively stable for the subsequent 0.5%. Due to the increased disclosure of capital income, the income tax paid by the top 0.1% surged by 60% during the period from 2015 to 2021. Source: authors’ calculations using data from AFIP statistical yearbooks. 35 Figure 13: An increase in the progressivity of the wealth tax in 2019 Effective Tax Rate (%) 2.25% 2013-15 2016 Amnesty Amnesty Top 0.01% p99.95-p99.99 p99.9-p99.95 1.25% p99.5-p99.9 0.75% p99-p99.5 0.5% 0.25% 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Notes: this figure compares the wealth tax liability expressed as a share of total (taxable and non-taxable) assets by the wealthiest 1% of adults (aged 20 and above) separately by bins of increasing assets. The pro- gressivity of the wealth tax increased in 2019 when Argentina taxed offshore assets at higher rates. Source: authors’ calculations using data from AFIP statistical yearbooks. Table 1: Domestic and foreign assets in 2014, 2015, and 2016 Domestic assets Foreign assets 2015 2015–14 2016–15 2016–15 2015 2015–14 2016–15 2016–15 US$ %∆ (pre) %∆ (post) % of total ∆ US$ %∆ (pre) %∆ (post) % of total ∆ (1) (2) (3) (4) (5) (6) (7) (8) Deposits and currencies 20,336 10% 1% 0% 4,736 16% 344% 23% Stocks and investments 1,020 1% 76% 1% 9,297 13% 366% 48% Real estate 37,230 -7% 24% 13% 792 20% 1044% 12% Vehicles 13,264 4% -10% -2% 17 21% 230% 0% Real rights 174 14% 71% 0% 9 13% 341% 0% Equity of companies 10,685 -9% 1% 0% 758 43% - -1% Accounts receivable 7,659 -6% 38% 4% 1,042 23% 116% 2% Other assets 7,552 -3% 10% 1% 2,221 13% -17% -1% Total assets 97,920 -2% 13% 17% 18,872 8% 311% 83% Notes: this table shows changes in foreign and domestic assets before and after Argentina’s 2016 tax amnesty. First, columns (1) and (5) report the total amount of domestic and foreign assets declared by wealth taxpay- ers in 2015. Columns (2) and (6) report the percentage change before the amnesty between 2015 and 2014, while columns (3) and (7) report the post-amnesty change between 2016 and 2015. Lastly, columns (4) and (8) express these differences relative to the total change in reported domestic and foreign assets between 2015 and 2016. Each row corresponds to a different type of asset. The last row shows the aggregate across categories. Source: authors’ calculations using data from AFIP statistical yearbooks. 36 Table 2: Reported assets by the wealthiest 2% in 2015 and 2016 Average reported assets (in 2015 USD) 2015 2016 %∆ % of total ∆ p98–p99 62,283 70,208 13% 3% p99–p99.5 121,491 150,029 23% 5% p99.5–p99.9 304,985 515,821 69% 28% p99.9–p99.95 586,091 1,234,090 111% 11% p99.95–p99.99 1,455,901 3,510,129 141% 27% p99.99–p100 (top 0.01%) 8,097,634 15,804,161 95% 26% Notes: this table compares the average assets (in 2015 US dollars) reported by the wealthiest 2% of tax units (individuals aged 20 and above) in 2015 and 2016. The table decomposes the top 2% into bins of increasing assets all the way to the top 0.01%. The last column expresses the differences relative to the total change in reported assets between 2015 and 2016. Individuals below the top 1% had moderate increases in their average assets after the 2016 amnesty. By contrast, the wealthiest 0.5% of taxpayers declared substantially more assets after the program. The number of tax units is 28,764,680 in 2015 and 29,164,076 in 2016. Source: authors’ calculations using data from AFIP statistical yearbooks. 37 ONLINE APPENDIX Appendix A Additional Tables and Figures Figure A.1: Argentina owned the equivalent of 36.5% of GDP in offshore wealth Notes: this figure shows the amount of household wealth owned offshore as a percentage of GDP, in 2007. Argentina, highlighted in red, owns the equivalent of 36.5% of GDP in offshore wealth. Source: Alstadsæter et al. (2018). I Figure A.2: Annual inflation: 2000–17 Annual Inflation (%) 40 30 20 10 0 2000m1 2002m1 2004m1 2006m1 2008m1 2010m1 2012m1 2014m1 2016m1 2018m1 Notes: this figure plots the average annual inflation rate in Argentina between 2000 and 2017. Source: authors’ compilation based on data from The Billion Prices Project at MIT (Cavallo and Bertolotto, 2016). Figure A.3: The wealth tax exemption threshold and the number of wealth tax filers and taxpayers Number of Constant individuals 2015 pesos 1.5m Exemption cutoff (right) 1.25m 1.5m Tax Filers 1m Taxpayers 1m 750k 500k 0.5m 250k 0 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Note: this figure plots the number of wealth tax filers and payers between 2002 and 2021 on the left axis, and the wealth tax exemption threshold (in 2015 pesos) on the right axis. Source: authors’ calculations using data from AFIP statistical yearbooks. II Figure A.4: An ad to encourage real estate disclosures Note: the banner presents the hypothetical case of a citizen with property worth Arg$3 million that had never been declared in their income and wealth tax returns. The left blue panel shows a 5% penalty (Arg$150,000) if the person comes forward and discloses it before 31 March 2017. The right red panel shows that the penalty increases to 202% (Arg$6 million) starting 1 April 2017 if the person does not disclose it under the amnesty and is caught by AFIP. Source: AFIP’s webpage. III Figure A.5: Three banners at the entrance of AFIP’s building in Buenos Aires Note: the banners on the left, in the center, and on the right say: ‘Pay 10% until December 31st,’ ‘Disclose your undeclared assets,’ and ‘Pay 15% until March 31st,’ respectively. Source: AFIP’s webpage. Figure A.6: An advertisement to encourage amnesty participation Note: the advertisement translates to: ‘Tax Amnesty. Declaration of assets. Report your assets, contribute to your country, we achieve better pensions. We all grow.’ Source: AFIP’s website. IV Figure A.7: Screenshots of AFIP’s website about the 2016 Amnesty Note: this figure reports screenshots of AFIP’s website regarding the 2016 tax amnesty. The top left panel reads: Tax Amnesty. How to disclose assets. Access this video-tutorial for a step-by-step guide to report your unde- clared assets and enjoy the benefits. The top right panel reads: Law 27.260. Tax Amnesty. This is an opportunity to do your part, declare all your assets, regularize your debt and, if you complied, find out about the benefits. The middle left panel reads: Tax Amnesty. Do you have undeclared cash? You have until October 31st. Don’t miss it out. You still have time! The middle right panel reads: Tax Amnesty. New App for smartphones. You can now download the tax amnesty’s App. Note also that the bottom of these four panels shows the countdown to the deadline of the amnesty program. The bottom panel shows a calculator that was made available for people to simulate the tax penalty when disclosing their assets. V Figure A.8: TIEAs signed around the 2016 amnesty program Inducción ÁMBITO INTERNACIONAL Agosto Setiembre Octubre Noviembre Diciembre Inicio del Sinceramiento Declaración Convenio de Acuerdo Acta de Acuerdo de conjunta para intercambio de bilateral para Entendimiento intercambio de el Intercambio información, el intercambio (retroactivo información de Información A partir de de información últimos 5 años) y tributaria (IRS), Tributaria. enero de 2017 A partir de Acuerdo de automático, a A partir de 2018 Intercambio requerimiento 2017 (inmuebles, y espontáneo. autos, cuentas bancarias). A partir enero de 2017 ADMINISTRACIÓN FEDERAL DE INGRESOS PÚBLICOS Note: this figure plots the timeline of TIEAs signed in 2016 between Argentina and Uruguay (September), Chile (October), Switzerland (November), Brazil (December), and the US (December). Source: AFIP’s communication campaign. Figure A.9: Google search interest Google search interest 2009 2013-15 2016 Amnesty Amnesty Amnesty 100 "Tax Amnesty" "Panama Papers" 80 60 40 20 0 2007m1 2009m1 2011m1 2013m1 2015m1 2017m1 2019m1 2021m1 Source: authors’ calculations using data from Google Trends. VI Figure A.10: Number of tax returns reporting foreign assets (levels) Number of tax returns 2009 2013-15 2016 120,000 Amnesty Amnesty Amnesty 100,000 Tax returns reporting foreign assets 80,000 60,000 40,000 20,000 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Note: this figure plots the number of taxpayers declaring foreign assets in wealth tax returns over time. Source: authors’ calculations using data from AFIP statistical yearbooks. Stocks Figure A.11: Where had the assets disclosed in 2016 been hidden? Disclosed value (% of GDP) 10 Bank accounts (deposits and cash holdings) 9.9 Total disclosed: US$ 117B (21% of GDP) 8 6 Real estate 4.7 4 2 1.8 1.9 1.4 1.2 0 .2 Foreign stocks Domestic stocks Foreign bank acc. Domestic bank acc. Foreign real est. Domestic real est. Other assets Notes: this figure plots the value of disclosed assets in the 2016 tax amnesty by type and location. Source: authors’ compilation based on official information from the national tax authority AFIP. VII Figure A.12: The richest 0.1% accounted for 30% of the entire change in reported assets between 2015 and 2016 ∆ Assets / Total ∆ Assets (%) 30 25 20 15 10 5 0 p96-p97 p97-p98 p98-p99 p99-p99.5 p99.5-p99.9 Top 0.1% Top income fractiles Notes: this figure compares the net change in assets reported between 2015 and 2016 by the richest adults (aged 20 and above) separately by bins of increasing income. The top 0.1% accounted for 30% of the entire change in reported assets between 2015 and 2016. Source: authors’ calculations using data from AFIP statistical yearbooks. VIII Figure A.13: Foreign assets reported by top percentile groups (a) Share of taxpayers reporting foreign assets % Reporting Foreign Assets 100% 2013-15 2016 Top 0.01% Amnesty Amnesty p99.95-p99.99 80% p99.9-p99.95 60% p99.5-p99.9 40% 20% p99-p99.5 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 (b) Foreign assets as a share of total assets Foreign Assets / Total Assets 100% 2013-15 2016 Amnesty Amnesty Top 0.01% 80% p99.95-p99.99 60% p99.9-p99.95 40% p99.5-p99.9 20% p99-p99.5 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Note: this figure plots foreign assets reported by the top 1% separately by groups of increasing assets. Panel (a) plots the share of individuals reporting a foreign asset, while Panel (b) plots foreign assets as a share of total assets. Nearly 100% of individuals in the wealthiest 0.01% of the distribution report a foreign asset after the 2016 amnesty program. Since other individuals in the top 0.1% did not declare foreign assets before the amnesty, they experience a large increase after the amnesty both in terms of the share reporting a foreign asset and the value of reported foreign assets. Source: authors’ calculations using data from AFIP statistical yearbooks. IX Figure A.14: Reported wealth in levels Billions of 2015 pesos 3000 2009 2013-15 2016 Amnesty Amnesty Amnesty 2500 2000 Reported Assets 1500 1000 500 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Note: this figure plots the total value of wealth reported by tax filers in constant 2015 pesos. The exchange rate was about Arg$12.9 per US$1 in December 2015. Source: authors’ calculations using data from AFIP statistical yearbooks. Figure A.15: Exchange rate: Argentine pesos per US Dollar Exchange Rate Pesos-USD 2016 70 ~20% increase ~30% increase Amnesty ~24% increase ~24% increase Jan 23, 2014 Dec 17, 2015 Aug 30, 2018 Aug 12, 2019 60 Currency controls lifted (dollar clamp) 50 40 30 20 10 0 01jan2013 01jan2014 01jan2015 01jan2016 01jan2017 01jan2018 01jan2019 01jan2020 Note: this figure plots the nominal exchange rate of Argentine pesos per US dollar between 2013 and 2020. Source: authors’ compilation based on data from the Central Bank of the Argentine Republic (BCRA). X Figure A.16: The top 0.1% of income earners contribute a higher share of income taxes Share of tax owed 50% 2013-15 2016 Amnesty Amnesty Top 0.1% 40% 30% p99.5-p99.9 20% p99-p99.5 10% 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Notes: this figure illustrates the amount of personal income tax owed by the top 1% of income earners, expressed as a proportion of the total income tax owed by all tax filers. Within this top 1%, we further divide the group into three subcategories, ranked by their decreasing income: the top 0.1%, the subsequent 0.4% (P99.5–P99.9), and the next 0.5% (P99–P99.5). These subcategory percentages do not sum to 100% since the remaining share is attributed to individuals below the 99th percentile. The share of total income tax owed by the top 0.5% decreased between 2010 and 2015, mainly due to an increase in the number of income tax filers during this period. However, the disclosures of assets under the 2016 amnesty program led to the reporting of more capital income by the top 0.1% of income earners (Figure 12). Consequently, the proportion of income tax owed by the top 0.1% witnessed a significant rise, climbing from around 25% in 2015 to approximately 40% in 2021. Source: authors’ calculations using data from AFIP statistical yearbooks. XI Figure A.17: Top statutory wealth tax rate and effective tax rate 2009 2013-15 2016 2.25% Amnesty Amnesty Amnesty Top statutory tax rate 1.25% Effective tax rate 0.75% 0.5% 0.25% 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Note: this figure plots the top statutory wealth tax rate against the average effective wealth tax rate between 2002 and 2021. The average effective wealth tax rate is the wealth tax liability divided by total (taxable and non-taxable) assets. Source: authors’ calculations using data from AFIP statistical yearbooks. Figure A.18: Wealth tax revenue to GDP ratio Wealth Tax / GDP 0.8% 2009 2013-15 2016 Amnesty Amnesty Amnesty 0.7% 0.6% 0.5% 0.4% 0.3% 0.2% 0.1% 0% 03/04 04/05 05/06 06/07 07/08 08/09 09/10 10/11 11/12 12/13 13/14 14/15 15/16 16/17 17/18 18/19 19/20 20/21 21/22 Note: this figure plots the ratio of wealth tax revenue to GDP for the period 2003–21. Source: authors’ calculations using data from AFIP statistical yearbooks. XII Figure A.19: Revenue from the 2016 amnesty’s special tax peaked in December 2016 Millions of pesos 90395 Total collected: 80,000 AR$ 148,600 mill (USD 9,522 mill) 60,000 40,000 27977 20,000 12903 7670 3053 3820 43 375 1066 0 Aug-16 Sep-16 Oct-16 Nov-16 Dec-16 Jan-17 Feb-17 Mar-17 Apr-17 Note: the amnesty took place between August 2016 and March 2017 and raised US$9.522 billion in revenue from penalties (‘special tax’). As a benchmark, this was the third largest source of tax revenue in 2016, after VAT and income tax. Arg$1,298 million were left unassigned to any month and April 2017 corresponds to late payments. Most disclosures of assets happened in December 2016, before the highest penalty fee increased from 10% to 15%, raising 61% of the special tax revenue in only one month. Source: authors’ calculations using data from AFIP statistical yearbooks. XIII Figure A.20: Confidence in government (UTDT index) Confidence in Currency Macri Govt Index 2009 controls 2013-15 took 2016 Amnesty introduced Amnesty office Amnesty 3.5 3 Mean 2.5 2.5 2 Mean 1.7 Mean 1.5 1.3 1 .5 Jan-08 Jan-10 Jan-12 Jan-14 Jan-16 Jan-18 Jan-20 Note: the UTDT index measures the evolution of public opinion about the work carried out by the national government. The scale of this confidence index varies from 0 (low) to 5 (high). Source: authors’ calculations using data from Indice de Confianza en el Gobierno. Escuela de Gobierno. Universi- dad Torcuato Di Tella (https://www.utdt.edu/icg). Table A.1: A comparison of Argentina’s recent tax amnesty programs 2009 2013–15 2016 President Fernández Fernández Macri Political inclination Left Left Right Can you disclose foreign currencies? ✓ ✓ ✓ Can you disclose assets? ✓ ✓ What is the maximum penalty? 8% 0% 15% Is there a penalty for disclosing? ✓ ✓ Is there a reduced penalty for repatriation? ✓ ✓ Is repatriation required? ✓ Is there a credible information exchange threat? ∼ ✓ Is there legal certainty? (Currency controls) ✓ How many people disclosed? 35,000 16,000 255,000 How much was disclosed? (% GDP) 1.3% 0.5% 21% Notes: this table compares the features of Argentina’s recent tax amnesty programs. The features of each amnesty were drawn from Law 26.749 for the year 2009, Law 26.860 for the period 2013–15, and Law 27.260 for the year 2016. The 2013 amnesty was meant to last three months, but was extended on nine occasions until December 2015. Source: authors’ compilation. XIV Table A.2: The 2016 amnesty according to AFIP Asset type Value % of total % of GDP (in million US$) Investments—abroad 54,999 47 10 Investments—in Argentina 860 1 0 Cash deposits—abroad 25,925 22 5 Cash deposits—in Argentina 405 0 0 National/foreign currency—in Argentina 7,344 6 1 Real estate—abroad 10,124 9 2 Real estate—in Argentina 10,434 9 2 Rest of assets 6,685 6 1 Total 116,775 100 21 Note: this table breaks down the US$116,755 disclosed in the 2016 amnesty program by type of asset. The geographic distribution of assets located abroad is the following. Investments abroad: 30% located in the US, 26% in Switzerland, and 15% in the British Virgin Islands; cash deposits abroad: 45% located in the US, 32% in Switzerland, and 9% in Uruguay; real estate abroad: 49% located in Uruguay, 37% in the US, and 4% in Brazil. The ‘rest of assets’ category includes: vehicles, boats, aeroplanes, art, jewellery, and more. The value disclosed in real estate corresponds to 167,000 properties—110,000 located in Argentina and 57,000 located abroad. Source: official information from the national tax authority AFIP. XV Appendix B Increasing transfers by earmarking revenue for pension spending As explained in Section 2, Argentina earmarked the revenue from the 2016 amnesty pro- gram’s ‘special tax’ to fund the public pension system, including reparations to pensioners for unpaid benefits and an increase in some existing benefits. In this section, we show that earmarking resulted in higher pension benefits for the elderly. For this analysis, we use data from two main sources. First, we use monthly re- tirement data from Argentina’s Social Security Administration (ANSES, for its Spanish acronym). The data consists of monthly tabulations of the number of retirees, the aver- age benefit, and the average by deciles. In Argentina the retirement benefit has two main components: a fixed universal basic amount and a variable social insurance component for persons aged 65 or older with at least 30 years of contributions. The latter is 1.5% of the insured’s average adjusted monthly earnings in the last ten years multiplied by the num- ber of years of contribution up to a maximum of 35 years. In addition, there is a minimum pension that acts as a floor, akin to minimum wages for low-skilled workers. All benefits are automatically adjusted for inflation twice a year, in March and September. Critically, the minimum pension is fixed by law. (For this reason, Appendix B leverages the fact that the minimum pension cannot be affected by the reparations program to proxy how aver- age benefits would have evolved absent the policy.) Second, we collected monthly data on the reparation spending funded by the amnesty’s revenue from a series of public gov- ernment memos; specifically, we use ANSES information from the government’s reports to Congress numbers 97, 99, 101, 103, 112, and 116. We leverage two institutional features to examine the effect of the tax amnesty on pension payouts. First, the reparations program aimed to raise pensions for those con- tributing for at least 30 years, who are eligible to receive a monthly pension benefit (based on pre-retirement income) in addition to Argentina’s monthly minimum pension benefit. By contrast, the reparations program should not affect the minimum monthly pension re- ceived by the roughly 2.5 million individuals—one in two older citizens—who contribute for fewer than 30 years (Berniell et al., 2020; Bosch and Guajardo, 2012; Rottenschweiler, 2020). Therefore, the average pension of retirees earning more than the minimum is po- tentially affected by the policy (treated), while the monthly minimum pension is not (con- trol). Second, Argentina adopted the amnesty law in June 2016, and the Social Security Administration (SSA) began accepting reparation applications from retirees three months later. Therefore, we should expect the program to increase pension payouts starting in September 2016. XVI Figure A.21 shows how retirees’ pension benefits evolve before and after the amnesty program. Panel (a) compares the minimum pension benefit (control) and the average pension above the minimum benefit (treated) before and after the adoption of the amnesty law in June 2016. Both series are expressed in constant 2015 pesos and normalized to 1 in December 2015. The two series evolve identically before the amnesty and then diverge, with the average pension substantially increasing after September 2016, when the SSA began accepting applications for pension reparations. Over 603,000 pensioners applied for reparations that month. The number of applicants doubled by November 2017 and stabilized at around 1.2 million. As a result, Figure A.21(b), which reports the difference- in-difference (DD) coefficient, shows that the difference between the two series stabilizes at around 15%.1 In addition, Figure A.21(b) superimposes the total monthly reparation spending based on official SSA reports. The series aligns closely with the DD coefficient, consistent with the amnesty program causally increasing reparation spending on pension benefits.2 In sum, by earmarking the revenue from the amnesty for Argentina’s pension reparations program, the average pension received by retirees increased by 15%. 1 In December 2017, Argentina introduced a new pension reform. Among other things, this reform revised the pension indexation formula used to calculate increases in pension benefits. As shown by the pension benefits’ step function growth in Figure A.21, the indexation system was based on semi-annual adjustments (based on growth in wages and taxes). By contrast, the 2017 reform based the system on quarterly adjust- ments (based on wage and price inflation). 2 Figure A.22 plots the evolution of average pension benefits above and below the median benefit. Since one- half of retirees receive the statutory minimum pension, they do not experience any change in their pension benefits after the amnesty. XVII Figure A.21: Earmarking amnesty revenue to fund retirees (a) Average and minimum pension benefits Avg Pension (Dec'15=1) Jun'16 Sep'16 Dec'17 1.2 amnesty law SSA accepts pension is passed reparation reform applications 1.1 Average pension 1 .9 Minimum .8 pension 2014m1 2015m1 2016m1 2017m1 2018m1 2019m1 (b) The difference between average and minimum pension and reparation spending Avg pension w.r.t. minimum Millions of (DiD coeff) Jun'16 Sep'16 Dec'17 2015 pesos 20% amnesty law SSA accepts pension 4000 is passed reparation reform applications 15% 3000 10% Reparation 2000 spending (right) 5% Avg pension 1000 relative to min pension 0% 0 2014m1 2015m1 2016m1 2017m1 2018m1 2019m1 Notes: this figure plots retirees’ pension benefits before and after the 2016 amnesty program. Panel (a) com- pares the minimum pension benefit (control) and the average pension above the minimum benefit (treated). The series are expressed in constant 2015 pesos and normalized to 1 in December 2015. The average pension substantially increased after September 2016, when the SSA began accepting applications for reparations. The number of reparation applicants stabilized at approximately 1.2 million by November 2017. As a result, panel (b), which plots the DD coefficient (left axis) against the amount spent on Argentina’s pension repa- rations program (right axis), shows that the difference between the two series also stabilizes at around 15%. Pension reparations spending increases after September 2016 and aligns closely with the DD coefficient. Source: authors’ calculations using data from ANSES. XVIII Figure A.22: Difference between average and minimum pension for the bottom and the top 50% of retirees Avg pension w.r.t. minimum (DiD coeff) Jun'16 Sep'16 Dec'17 20% amnesty law SSA accepts pension is passed reparation reform applications 15% Top 50% 10% 5% Bottom 50% 0% 2014m1 2015m1 2016m1 2017m1 2018m1 2019m1 Note: this figure shows how pensions increased after the 2016 amnesty for the top 50% of retirees earning above the minimum pension (blue line) but not for the bottom 50% who receive the minimum pension and were unaffected by the reparation program (red line). Each series plots the DD coefficient comparing the average pensions relative to the minimum pension in December 2015. Source: authors’ calculations using data from ANSES. XIX