ADMINISTRATION AGREEMENT BETWEEN THE EUROPEAN COMMISSION ON BEHALF OF THE EUROPEAN UNION AND THE INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT TSI Project 20LT09 Micro Enterprises and Self-employed Tax Regulatory Assesment for Removing Hurdles to Growth Lithuania (EUROPE AND CENTRAL ASIA) Report Assessing the Impacts of Tax Optimization and Bunching in MEs and Self-employed and Legal Entities Responses to Size-based Tax Rates in Lithuania Output 1 & 2 December 2022 Project carried out with funding by the European Union in cooperation with the European Commission’s DG REFORM 1 DISCLAIMER This document was produced with the financial assistance of the European Union. The views expressed herein can in no way be taken to reflect the official opinion of the European Union. This report is a product of the International Bank for Reconstruction and Devel- opment/The World Bank. The findings, interpretation and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of the World Bank, the European Commission or the Government of Lithuania. The World Bank does not guarantee the accuracy of the data included in this work. Copyright Statement The material in this publication is copyrighted. Copying and/or transmitting portions of this work without permission may be a violation of applicable laws. 2 Contents 1 Preface 4 2 Introduction 7 2.1 What can we learn from the experience of other countries? . . . . . 7 3 Personal income taxes in Lithuania: empirical facts 10 4 2019 tax reform: Income shifting incentives? 26 5 Conclusion 30 References 32 A Additional figures and tables 35 B Data sources 37 B.1 Definition of variables . . . . . . . . . . . . . . . . . . . . . . . . . . 37 B.2 Definition of owner-managers . . . . . . . . . . . . . . . . . . . . . 39 3 1 Preface These reports correspond to the Outputs 1 and 2 of the TSI Project 20LT09: Micro Enterprises and Self-employed Tax Regulatory Assessment for Removing Hurdles to Growth in Lithuania. The two reports assess the impacts of tax optimization and bunching in micro enterprises and self-employed and legal entities and their responses to size-based tax rates in Lithuania, including the extent of the problem and drivers of tax optimization and bunching. To carry out the analysis the project team reviewed existing legislation and regulations to understand the overall context and the resulting incentive structure, it analyzed tax and other necessary data by the taxpayers to document existing trends and behaviors. As part of the project the team also analyzed relevant international experience and country cases to help benchmark the most relevant aspects of the system. Reports analysing the size and effects of tax optimisation, bunching and income shifting The reports identify the most important factors driving tax optimization and the provisions that are not aligned with international best practice that may lead to inefficiencies in the tax system, lack of neutrality and potential loopholes that impact on taxpayer behaviour, including on self-employed and MEs. The report on legal entities studies their response to the existence of differential tax rates based on firm size. While special treatments are intended to provide support for entrepreneurs and small and medium enterprises (SMEs) they may also provide the wrong incentives for some firms who may try to remain small in order to take advantage of the special treatment. The report documents the existence of a static response to differential tax rates in the short run and provides suggestive evidence that these effects extend to the medium- to long-term. The report on Personal Income Taxes and the incentives to shift income in Lithuania provides new empirical facts about personal income taxation (PIT) in Lithuania, with a focus on the implications of a schedular tax system for equity and efficiency. Different tax rates depending on the source of their income and differentiated rates encourage tax optimization and undermine the efficiency and equity of tax systems. The report documents the importance of self-employment income sources and how they vary according to income and focuses on owner- managers of small partnerships and they behaviour to minimize their tax burden. Separately, the project team has developed an analytical microsimulation model (part of Output 1) that can assist policy makers and stakeholders in simulating the taxpayer behaviour resulting in the existing optimization and bunching. The mi- crosimulation tool was used for comparisons with selected peer countries to simulate 4 possible effects of changes in the tax structure, tax parameters, or legal provisions related to tax optimisation and bunching. Data limitations, including added complications due to the COVID-19 pandemic that limited on-site missions did not allow the project team to collect the necessary data for estimating the impact of tax optimisation and bunching on overall produc- tivity and growth of MEs and self-employed using a macro-micro economic model (using the input-output matrix) differentiating by type and size of economic players in the most relevant sectors in the economy to estimate the economic productivity related to the size of economic units and their contribution to the economic out- put. The team documents however the existing practices of tax optimisation and bunching as well as other responses to regulation on growth, both for the ME and self-employed sectors, pointing to economic opportunity costs resulting from tax- payer practices attributed to the underlying incentives in the tax and regulatory systems. Micro level analysis on firm level productivity and identification of tax optimisation and bunching scenarios was impeded by lack of detailed data on the different cost variables at firm level, precluding the team from doing projections of productivity gains at the firm level. The reports have been produced by the World Bank team led by Mr. Alberto Leyton, Lead Public Sector Specialist (co-Task Team Leader) and Mrs. Cristina Savescu, Senior Economist (co-Task Team Leader). The authors of the reports are Thiago De Gouvea Scot de Arruda (Economist) and Pablo Garriga (Economist). The World Bank team would like to acknowledge and thank Mr. Vytas Adomaitis in the Innovation Agency of Lithuania and Mr. Vaidas Navickas, Adviser to the Prime Minister for their guidance, feedback, advice and hospitality, as well as coun- terparts in the Ministry of Finance, the Ministry of Economy and Innovation, the Ministry of Social Security and Labour, and the Central Bank of Lithuania for their feedback. The team would also like to express its gratitude to officials in the Depart- ment of Statistics - Edita Baltuše, Aleksandra Golubovič, Gita Literske, Vadimas Ivanovas, Darius Abazorius and Jonas Bačelis - for hosting the World Bank team in their secured office and for useful guidance in navigating through the data in the Palantir platform. Finally, the team would like to thank Kestutis Lisauskas, Si- mona Poceviciute, Akvile Laugalyte from Ernst Young for their support and advise for the interpretation and understanding of the specificities of the Lithuanian tax system. This report should be read together with the report on Outputs 4: Recommen- dation Report Analyzing the Size and Effects of Tax Optimization and Bunching with a Microsimulation Tool. This final version of the report has benefited from nu- 5 merous comments from Lithuanian authorities, as well as the European Commission and World Bank officials. 6 Personal Income Taxes and Incentives to Shift income in Lithuania 2 Introduction This report provides new empirical facts about personal income taxation (PIT) in Lithuania, with a focus on the implications of a schedular tax system for equity and efficiency. Individuals in Lithuania pay very different tax rates depending on the source of their income. Since 2020, for example, workers face a progressive tax schedule on their wages, with two rates of 20% and 32% and a minimum non-taxable income. Self-employed workers, on the other hand, can declare income from several different sources that are taxed differentially. Income from Individual Activity Certificates (IAC), for example, are taxed at a progressive schedule ranging from 5% to 15%. Income from Business Certificates (BC), a cat- egory that encompasses dozen of activities and can be claimed up to €45,000, pay no income tax to the central government, instead paying local fixed fees at the local level. There are also different tax rates for dividends (15%) and several sources of capital gains(15% to 20%), a more common tax differentiation for income from capital. Differentiated rates encourage tax optimization and undermine the efficiency and equity of tax systems. As we discuss in detail below, several studies have shown that taxpayers, particularly those with higher incomes and com- plex earning composition, are sophisticate and exploit tax differential to minimize tax liabilities. In this report we first provide a descriptive analysis of the the im- portance of self-employment income sources and show how they vary according to income and its implications for equity. We then focus on a small set of taxpay- ers, owner-managers of small partnerships, and document that these individuals understand the implication of differentiated rates and report wage income so as to minimize their tax burden. Since the participation of small partnerships has steadily increased in the last decade, this is likely to be a growing issue going forward. 2.1 What can we learn from the experience of other coun- tries? The issue of income shifting, or the relabeling of income sources to exploit tax benefits, is well-known and extensively studied. Following major reforms to corporate and personal income taxes in the United States in the 7 1980s, Slemrod (1995) and Gordon and Slemrod (1998) noted the incentives to shift income from the corporate to the non-corporate sector to benefit from lower tax rates. This type of behavior has several important implications. First, it clearly shows that taxpayers are keenly aware of the tax system and will often use complex schemes to face lower rates. Second, as noted by Gordon and Slemrod (1998), it "plays havoc with the usual interpretation of many kinds of data, because it blurs the return to capital and the return to labor". As an example, observing a large decrease in personal labor income as a response to higher tax rates could be explained not by a reduced labor supply, as classical models in economics would predict, but as a relabeling of labor income as capital income through the use of corporations. Recent studies have used rigorous empirical methods and rich admin- istrative data to quantify these income shifting responses. While the issue of income shifting was debated in the 1990s, the evidence to precisely quantify those responses was largely absent. Taking advantage of the increasing availability of ad- ministrative data to researchers and advances in econometric methods, since then several studies have rigorously quantified this behavior. Romanov (2006), for exam- ple, studied a tax reform in Israel that abolished the ceiling for social contributions and therefore increased contributions for high-earners. They document that high- income individuals responded by opening over 4,500 companies (5% addition to the corporate sector). This strategy was widespread among professionals, specialists and physicians who belong to the upper percentile of the income distribution, and for whom income shifting was not only desirable but also feasible. Alstadsæter and Jacob (2016) study the 2006 Swedish tax reform which reduced the effective dividend tax rate by 10 percentage points for owners of closely held corporations (CHC). They show that, compared to a similar group of companies not affected by the reform, earned income decreased by 6 percentage points and dividend income increased in the same magnitude. When dividend taxes fell, owner-managers shifted their reported income from salaries to dividends to benefit from the lower taxes.1 Harju and Matikka (2016) study a reform that increased dividend rates in Finland and note the opposite effect, an increase in wage distribution of business owners. More than that, they compute individual-specific "optimal wages" to minimize tax burden and find that individuals distribute exactly those amounts. In the United Kingdom, Smith, Pope, and Miller (2021) show that owner-managers smooth their declared income over time to avoid higher marginal tax rates, and that these re- 1 In a companion paper studying the same reform, Jacob (2021) shows that cutting dividend taxes did not increase firm investment, employment or real activity on average. The entire effect of the reform seemed to be making dividend distribution more attractive. 8 tained profits inside companies are held in liquid assets instead of being reinvested. These income shifting activities can have important implications for the structure of legal entities and even the measurement of macroeco- nomic aggregates. In the United States, the steady rise of S-corporations (pass- through entities that do not pay corporate income taxes) and partnerships since the 1980s is linked to tax benefits relative to C-corporations, that pay corporate taxes. Since partnerships are not considered corporations, this has important implications for the measurement of corporate sector activity, for example. One crucial dimen- sion in which income shifting has already been shown to affect the measurement of macroeconomic aggregates is in the distinction of labor and capital income. In the United States, the share of value-added going to labor in the form of wages has steadily declined since the 1980s, in favor of income going to capital. Smith et al. (2022) estimate that up to 30% of that decline might be a mismeasurement of labor income, due to the rise of S-corporations, for which tax incentives encourage income to be labeled as capital, and the rise of partnerships outside the corpo- rate sector. To show that owner-managers seem to relabel labor income as capital income, they study firms that changed their corporate form from C-corporations to S-corporations, and document that these changes are immediately accompanied by a substantial fall in labor compensation and a commensurate increase in capital compensation. These changes are even more pronounced in sectors that are "capital- light", like consulting and financial services, where there are often few workers and thus relabeling is more likely. Governments can design rules to limit income shifting for owner- managers, but these can be challenging to enforce. In the United States, the Internal Revenue Service (IRS) determines that the wage compensation of owner- managers must be "reasonable compensation for services rendered" (Nelson 2016; Auten, Splinter, and Nelson 2016), although enforcement seems to be limited, par- ticularly since the rule is not precise. In Sweden, Alstadsæter and Jacob (2016) report that there exists a cap on the funds that owners of closely-held corporations can distribute as dividends—past those levels, dividends are taxed as earned (labor) income at higher rates. Harju and Matikka (2016) discuss the complex rules that apply to taxation of dividends vs. labor income for owner-managers, the Finnish system includes provisions such that dividends distributed by firms deemed to have very high returns compared to their net assets are taxed as earned income. Income shifting has important implications for the efficiency and eq- uity of tax systems. On the efficiency front, the fact that individuals engage in complex relabeling of income and that firms choose their corporate form due 9 to tax incentives likely involves socially wasteful investments, such as the hiring of accountants and lawyers specialized in these tax-optimization strategies. It might also involve inefficient allocation of capital, if it flows to entities with tax-preferred status instead of the most productive firms in the economy. Finally, it also un- dermines the goals of a tax system that might intend to provide preferential tax treatment for real capital income. While classical results in economic theory sug- gested that the optimal tax rate on capital income should be zero (Atkinson and Stiglitz 1976), recent research provides several reasons why that should not be the case and capital should be taxed at positive rates (e.g. Saez and Stantcheva 2018). One of them is precisely because declared capital income might be labor income in disguise, in which case tax authorities would want to tax that income at a positive rate to recover some of the shifted tax liabilities. As Piketty, Saez, and Zucman (2022) note, "the frontier between capital and labor income flows is often fuzzy, thereby lending support to a broad-based, comprehensive income tax to reduce tax avoidance opportunities". On the equity side, these types of complex income shift- ing activities are not available to wage workers (who often have a sole source of income, and whose taxes are retained at source in most cases) nor to less sophis- ticate self-employed individuals, who might lack the resources to hire specialists to minimize their tax burden. In that sense, these relabeling activities are likely regressive, decreasing the effective tax rate paid by individuals with higher incomes. 3 Personal income taxes in Lithuania: empirical facts In this section we use administrative data on personal income tax (PIT) filers to characterize implications of the schedular tax system in Lithua- nia. Our main sample consists of all PIT forms filed between 2016 and 2020. In Table 1 we provide key descriptive statistics of the sample. Here we highlight some features of the data; a much more detailed discussion of the database construction is provided in Appendix B. First, we note that the number of individuals filing each year is very stable at 1.4 million—approximately 60% of the population aged 15 and above.2 The stability of filing numbers over time is likely explained in part by the ease to file taxes: most sources of income including wages, dividends and interest are pre-filled for the taxpayer (OECD 2022). Second, as in most countries wages are the most prevalent source of income: over 85% of taxpayers filing taxes every year 2 According to the World Bank’s World Development Indicators, Lithuania’s population around that period was 2.8 million and 15% were aged 0 to 14 years. 10 claim income from wages, whereas the number claiming income from dividends for example is always below 5%. Two sources of self-employment income that receive preferential treatment are income from IACs, reported by 8-11% of individuals, and income from BCs claimed by 5-6% of individuals. As we discuss below, while the overall claiming of these income sources is rel- atively small on aggregate, their use varies considerably across the income distri- bution. This is particularly clear for dividend income: while the overall average dividend income is less than €650 in 2020, the average among those claiming div- idends is €26,000, significantly more than the €15,000 average of wage income for wage earners. The averages for income from IAC and BC are also meaningful at €8,000 and €6,600 in 2020, respectively. Finally, we note that individuals receiving income from partnerships are a very small fraction of total filers, close to 1%, but that income type is also high among those declaring positive amounts. We discuss the issue of partnership incomes in more detail in a separate section. Table 1: Descriptive statistics 2016 2017 2018 2019 2020 Receives wage (%) 0.85 0.86 0.87 0.87 0.87 Receives IAC income (%) 0.08 0.08 0.09 0.10 0.11 Receives BC income (%) 0.06 0.06 0.06 0.05 0.05 Member of partnership (%) 0.01 0.01 0.01 0.01 0.01 Receives dividend income (%) 0.03 0.03 0.03 0.03 0.03 Average Income (Euro) Employment related income 6,837 7,545 8,434 11,959 13,218 Dividend income 462 529 560 661 649 IAC income 471 604 579 665 835 Business certificate income 339 359 373 369 307 Partnership income 53 57 61 69 68 Average positive Income (Euro) Employment related income (positive) 7,992 8,790 9,831 13,637 15,068 Dividend income (positive) 17,725 20,277 21,085 25,731 25,825 IAC income (positive) 6,241 7,359 6,438 6,823 7,922 Business certificate income (positive) 5,413 5,804 6,216 7,294 6,606 Partnership income (positive) 9,649 10,382 11,443 9,198 9,482 1,445,470 1,455,507 1,439,327 1,454,997 1,403,839 Note: This table provides descriptive statistics of personal income tax data. The first panel presents shares of each indicator while the second and third panels present averages for each variable. We provide more detail about database construction and variables in Appendix A. The use of income facing preferential tax rates is not even across the income distribution. Some forms of income are often concentrated at the top: dividends, for example, are rarely received by lower-income individuals but are very relevant among high-income earners. In the case of Lithuania, it is less clear ex-ante how the usage of income sources such as BCs or IACs is distributed. We use the 11 PIT microdata to provide evidence in that direction. The first step we take is to construct a measure of taxable income for all individuals, which includes all income subject to taxes plus income from BCs, which only pay local taxes/fees.3 We then use this measure to order individuals from the highest to lowest income.4 All our results are presented in some aggregated form, so here we explain the two main aggregations we provide. First, we provide results across the percentiles of income distribution: we create 100 groups of approximately 14,000 individuals each, start- ing from those with lowest income all the way to the 14,000 with highest income. The second aggregation recognizes that a large part of total income is concentrated at the very top of the distribution, so we provide a more fine-grained view of indi- viduals at the top. For that, we create 37 groups determined in the following way: we divide the bottom 90% of earners in 9 equal groups of approximately 140,000 individuals; we divide the top 10% earners in 9 equal groups of 14,000 individuals, with the exception of the top 1% which we divide in 9 groups of 1,400 individuals; finally we divide the top 0.1% in 10 groups of 140 individuals, such that the final group contains, approximately, the 140 individuals with highest incomes. This is an approach similar to several studies that study inequality with a focus on the top of the income distribution (Alvaredo et al. 2020; Herault et al. 2021). In Figure 2 we present a first description of these grouped data for 2020, showing the minimum amount of annual income required to be in each group.5 Panel (a) presents the data in 100 percentiles while panel (b) shows the more detailed data on the top of the distribution. In panel (a) we observe that the tax filer with median income declares annual taxable income of approximately €11,000.6 To be at the top 25% of the distribution, the annual income rises to almost €20,000 and to close to €30,000 at the top 10%. At the very top, declaring annual income above €86,000 would put an individual among the 1% highest earners. Panel (b) provides more detail about the income levels at the very top. While €86,000 puts an individual at 3 More details about these measures are provided in the appendix. 4 We note that these are measures of individual incomes, not household income. Most measures of poverty and inequality take into account the sum of household income and then proceed to distribute it among its members. Since taxes are filed individually and we do not observe the composition of households, all our measures consider only income filed by an individual. 5 In all figures we abstain from presenting results for the bottom 10% of the income distribution. Administrative tax data is an excellent source to understand the top of the income distribution, but not ideal to understand the bottom—where factors such as informality or household composition might play a very important role. For that reason, we consider that results derived from the lowest earners in the tax data (the annual income for the bottom 10% is lower than €2,000) are not reliable to inform our analysis. Best practices to obtain a comprehensive income distribution include combining administrative tax data at the top of the distribution with household survey data for the bottom (Blanchet, Flores, and Morgan 2018). 6 We present summary of income levels and composition of income across the distribution in Table 2 and Table 3 12 the top 1%, they would need more than triple that amount or almost €300,000 to be at the top 0.1% and at least €1.18 million in order to be among the top 0.01% with highest incomes on that year.7 7 We note these levels refer to taxable income and therefore are not comprehensive nor can be used to compare inequality levels with different sources, which often will use distinct definitions of income. 13 Figure 2: Income levels across the distribution (2020) (a) Percentiles of income distribution (b) Top quantiles of income distribution Note: These figures partition the income distribution for the year 2020 into quantiles and show the minimum amount of income required to classify a taxpayer into each bin. Panel (a) partitions the population into percentiles. Panel (b) repeats the same exercise but providing more detail about the income levels at the very top: it divides the bottom 90% of earners in 9 equal groups of approximately 140,000 individuals; the the top 10% earners in 9 equal groups of 14,000 individuals, with the exception of the top 1% which are divided in 9 groups of 1,400 individuals; finally the top 0.1% are divided in 10 groups of 140 individuals. The bottom 10% of the income distribution is excluded from both calculations. The y-axes are presented in log scale. Having provided a broad overview of the income levels across the distribution, 14 we now describe first the usage of each income source and then their relevance in amounts. In Figure 3 we show, for each of the groupings, the share of individuals that claim income from four different sources: wages, BCs, IACs and dividends. These categories are not exhaustive—many more income categories can be declared for PIT in Lithuania. But, for most of the distribution, these sources encompass the vast majority of total income declared. First, across the distribution most taxpayers claim some form of wage income—above 90% in general but still over 80% at the very top. Income from dividends, on the other hand, is claimed by almost no taxpayer all the way to the top 10%, when it starts to slowly increase to about 20% at the top 1% and rises to almost four in five taxpayers among the very top. Income from IAC, on the other hand, is quite relevant both at the lower-end of the distribution, where close to 20% of taxpayers claim it, and also at the top where again close to 20% of high-earners also claim income from IAC origin. A similar pattern, although at lower levels of usage, is seen for income from BC: around 10% of taxpayers at the lower end of the distribution and 7% of those at the upper part claim income from BC. In Figure 4 we zoom in on the usage of each income source separately, to allow for a better visualization of the patterns across the distribution.8 This pattern of high usage of "self-employment" incomes like BC and IAC at the tails of the distribution is consistent with the type of activities often performed under these income codes. Until 2017, for example, activities under IAC were taxed either at 5% for most activities or at 15% for activities related to "high-earners", such as legal services, engineering, medical services, etc. 8 In these graphs we can clearly observe a "peak" in the number of wage workers close to the 30th percentile of the distribution and an equivalent valley in the share of individuals with IAC and BC income. In 2020, income at the 30th percentile was approximately the annualized value of the minimum wage - so that group is disproportionately composed of wage workers receiving the minimum wage. For the same reasons, the share of individuals receiving wages declines quickly below that level, since all full-time wage workers should in theory be receiving the minimum wage or more. 15 Figure 3: Income sources across the distribution (2020) (a) Percentiles of income distribution (b) Top quantiles of income distribution Note: These figures partition the income distribution for the year 2020 into quantiles and show the share of taxpayers in each bin that claim income from wages, dividends, BC or IAC. These categories are not exhaustive and the categories are not mutually exclusive as a taxpayer may claim income from multiple sources. Panel (a) partitions the population into percentiles. Panel (b) repeats the same exercise but providing more detail about the income levels at the very top; the quantiles in this case are defined in the same way as in Figure 2. 16 Figure 4: Income sources across the distribution (2020) 1 .2 Share with any wage income Share with any IAC income .9 .15 .8 .1 .05 .7 0 20 40 60 80 100 0 20 40 60 80 100 Income percentile Income percentile (a) Wages (b) IAC .4 .1 Share with any dividend income Share with any BC income .3 .08 .2 .04 .06 .1 .02 0 0 20 40 60 80 100 0 20 40 60 80 100 Income percentile Income percentile (c) BC (d) Dividends Note: These figures partition the income distribution for the year 2020 into percentiles and show the share of taxpayers in each bin that claim income from wages (panel a) IAC (panel b), BC (panel c), and dividends (panel d). These categories are not exhaustive nor mutually exclusive as a taxpayer may claim income from multiple sources. The bottom 10% of the income distribution is excluded from the plots. While income from BCs is negligible at the top of the distribution, income from IACs is meaningful. After presenting the share of individuals that claim income from each source, in Figure 5 we show what is the share of each income source in the total income for each group across the distribution. Mirroring the previous results, in panel (a) we present results by percentiles while in panel (b) we break down the top of the distribution in more detail. Consistent with the fact that wage workers are the majority across the distribution, between the percentiles 30 and 80 wage income is equivalent to approximately 90% of total reported income, with the remaining being IAC (3-4%), BC (1-2%) and other incomes. In figure (a) we can see that both BC and IAC are much more relevant below the 30th percentile, where we observe fewer wage workers; and that IAC also has an important partici- pation at the very top, above the 90th percentile. To have a better sense of what is happening at the top, on panel (b) we show that IAC often accounts for up to 10% of total income within the top 1% of the distribution, while BC is negligible—since 17 income from BC only receives preferential taxation up to €45,000, it becomes a very small share of total incomes as we approach the top of the distribution. At those levels, income from dividends and capital gains become much more relevant: among the top 0.1%, income from wages is often less than 20% of total income while dividends reach around 40% at the very top. As we discuss below, at least for one group of taxpayers—owner-managers of partnerships—evidence suggests that part of dividends are not returns to capital but returns to labor, instead distributed as dividends to minimize tax liability. 18 Figure 5: Income composition across the distribution (2020) (a) Percentiles of income distribution (b) Top quantiles of income distribution Note: These figures partition the income distribution for the year 2020 into quantiles and show the share of total income coming from: wages, dividends, capital gains, BC, IAC or other income sources. Panel (a) partitions the population into percentiles. Panel (b) repeats the same exercise but providing more detail about the income levels at the very top; the quantiles in this case are defined in the same way as in Figure 2. The share of total income by source is computed as the ratio between the amount of income declared from each source and the total amount of income, by quantiles. 19 The multiplicity of tax rates for different incomes undermines the pro- gressivity of the tax system. Progressive tax schedules are designed to increase the effective tax rates paid by high-income individuals: marginal tax rates rise with income, so that very high-income earners are taxed at a higher rate. Whether the average effective tax rates (the ratio between taxes due and total income) indeed rise with income levels, however, is an empirical question, particularly with a system as schedular as the one in Lithuania: since tax rates differ significantly by income source, the shape of the effective rate across the income distribution will depend on the composition of incomes as well as the levels at which marginal rates increase. In Figure 6 we present the effective rates9 across the percentiles of income distribution (panel a) and also at detailed bins at the top (panel b). The first feature we note is that the effective rate steadily increases between approximately the 20th and 90th percentile. As shown in Figure 5, income from wages is dominant on that range— since a minimum non-taxable income exists and is phased-out as wages increase, the effective rate steadily increases as an increased share of labor income is taxed at the marginal rate of 20%. Second, the fact that the lowest effective rate is observed around the 20th percentile and that we see a modest increase for incomes below that is explained by the fact that the Minimum Non-Taxable Income (MNTI) for labor income in 2020 (€4,800) is equivalent to the average income at the 20th percentile (so labor income is fully exempt at that level or below) and that for individuals with lower incomes sources such as the IAC do not benefit from MNTI, so they are effectively taxed at a higher rate. The effective tax rate is highest close to the 90th percentile, where individuals in that group pay approximately 17% of their total income in taxes. Above that level we see a steady decline in the effective rate, particularly pronounced at the top 1%. We provide more details for the top of the distribution in panel (b), where we show that, within the top 1% of highest earners, the effective rate is quite stable at approximately 14%, three percentage points lower than for those individuals at the 90th percentile and similar to those individuals at the 70th percentile of the distribution. Since labor incomes are taxed at a progressive rate, we might have expected that the reverse was true, that effective tax rates increased faster at the top. The income level at which the higher marginal rate applies is important here. We highlight that in 2020 the higher marginal rate 9 We highlight that throughout this report we refer to the PIT effective tax rate, excluding social security contributions. The main reason for that is that we do not have access to social security microdata so we cannot ascertain each taxpayer’s contributions. A second reason is that the treatment of social security contributions is not straightforward, since they are a mix of deferred income and pure taxation. This is an important caveat to our results. 20 on labor income (32%) only applied to incomes above 84 average wages (AW) or approximately €104,000. We have documented in Figure 2 that this level of total income is only achieved by individuals in the top 0.7% of the distribution—less than 10,000 taxpayers. Since at the top 1% we also see other incomes becoming more relevant than labor income, in practice few individuals have any income taxed at the 32% marginal rate—meaning that we should not expect much additional progressivity at the top.10 The decrease in effective rates at the top is mainly explained by the fact that the composition of total income changes rapidly for high-income individuals, as shown in Figure 5. Within the top 0.1%, for example, dividends are a larger share of total income than wages, and the former are taxed at a flat 15% rate while the latter are taxed at 20% and 32%. Furthermore, income from IAC comprises up to 20% of total income for these high-earners and it is taxed at 5- 15% rates. Take for example individuals with total income of €300,000, which puts them in the top 0.1%. On average those individuals declare 20% of their income or €60,000 to be IAC income. In 2020, the first €20,000 would have been taxed at 5%, the amount between €20,000 and €35,000 would have been taxed at an increasing rate until 15% and the remaining €25,000 would be taxed at 15%. The effective rate on their IAC income, therefore, would be closer to 10%, substantially below both the tax rate on dividends as well as tax rates on wage income. 10 According to the United States Congressional Research Service, approximately 0.5% of US taxpayers were taxed at the higher marginal rate of 37%, but a much larger share are taxed at rates larger than 20% since there are seven tax brackets - 10%, 12%, 22%, 24%, 32%, 35% and 37%. 21 Figure 6: Effective tax rates across the distribution (2020) (a) Percentiles of income distribution (b) Top quantiles of income distribution Note: These figures partition the income distribution for the year 2020 into quantiles and calculate the effective tax rate faced by taxpayers in each bin. Panel (a) partitions the pop- ulation into percentiles. Panel (b) repeats the same exercise but providing more detail about the income levels at the very top; the quantiles in this case are defined in the same way as in Figure 2. Effective tax rates in each bin are defined as the ratio between the total amount of taxes owed and the total amount of income declared by taxpayers in each quantile. 22 Individuals with similar incomes can pay very different tax rates de- pending on the source of income received. The previous discussion has high- lighted how the average effective tax rate varies across the income distribution and how progressivity is undermined by rate differentiation. Another feature of a tax system in which a multiplicity of rates exists is that individuals with the same in- come might pay very different rates according to the source of their incomes. In Figure 7 we present measures of how much, within a percentile of income distri- bution, the effective tax rate paid varies across taxpayers. We show not only the median effective rate, but also the percentiles 10, 25, 75 and 90 of the distribution in each quantile. First, note that below the percentile 25 of income distribution the relatively high average effective rate documented before is driven by few taxpayers with higher tax rates: over 75% of taxpayers in these groups are paying zero rates. Second, over the range of percentiles 20 to 90, dispersion is limited: at percentile 50, for example, we see a tight distribution with over 65% of taxpayers paying ef- fective rates between 12% and 14%. Third, across the distribution we observe that a substantial amount of taxpayers can pay very little taxes: at the percentile 50 of total income, 10% of taxpayers are paying less than 5% in effective rates. Finally, the dispersion in effective rates becomes particularly stark at the very top: among the top 1%, 10% of taxpayers pay close to zero taxes; 25% are paying less than 12% and the top 10% pay more than 20% effective rates. The multiplicity of tax rates (for capital, wage and self-employment incomes) produces wide dispersion in effective rates paid. 23 Figure 7: Effective tax rate distribution by percentile Note: This figure presents effective tax rates (ETRs) (tax liability as a share of total income) for each percentile of the income distribution, excluding the lowest decile. We compute ETRs for each taxpayer and then compute, by each percentile, the quantiles of the distribution of ETRs - 10th percentile (p10), 25th percentile, 50th percentile (median), 75th percentile and 90th percentile. Differential tax rates for wages vs. self-employment are stark across the distribution. We first note that simply comparing tax rates paid by individ- uals receiving wage income vs. IAC income, for example, can be very misleading since the composition of income changes significantly across the income distribu- tion, so we might be comparing very different individuals. In Figure 8 we present the effective tax rate across the distribution of income, divided in deciles,11 for a select group of individuals: those receiving income almost exclusively from wages, from IAC or from BC.12 We also present the overall effective rate for each decile as a reference. First, we note that the effective rates for those earning only wages is similar to the overall rate for the majority of the distribution, with exception of the bottom 2nd decile and top decile, since these are the majority of taxpayers. Second, throughout the distribution the ETR paid by individuals who receive only IAC income is substantially lower than those receiving wage income: for individuals 11 Since we are looking at a selected number of taxpayers, we present results for deciles of income distribution instead of percentiles to reduce noise. 12 We define a taxpayer as receiving income almost exclusively from one source if more than 98% of their income comes from a given source. We use that rule since individuals often will receive very little amounts of income from some sources, like interests. 24 making the median income (5th decile), for example, the ETR for wage-workers is approximately 13% vs. 4% for those receiving IAC income—a difference of 9 per- centage points. This gap is smaller for those at lower incomes and reversed for those at the 2nd decile—at that level, IAC workers actually pay a higher ETR since wage workers are paying close to zero given their minimum non-taxable income levels. At the top of the distribution, the ETR for these two groups become more similar since the marginal income tax on IAC income increases from 5 to 15% but even at the top decile individuals with similar income will have ETRs that are 6 percentage points different depending on their source of income. An opposite relationship is observed for those individuals receiving income exclusively from BC: since these are taxed at a flat rate and capped at €45,000, the ETRs are very high for individuals at the bottom of the distribution (in practice, the taxes paid can be even larger than the total amount of income) and it declines steadily as individuals earn a higher amount of BC income.13 Figure 8: Effective tax rate by type of income source Note: This figure presents effective tax rates (ETRs) (tax liability as a share of total income) for each decile of the income distribution, excluding the lowest decile. We compute ETRs separately for all taxpayers ("Overall") and for those receiving their income mostly from Wages, IAC and BC. We define an individual to receive most of their income from one source when that source responds for more than 98% of their total income. 13 We should note that the separation of BC and IAC income is not always clear. This is because income from business certificates exceeding €45,000 are taxed as IAC, so individuals might receive both income sources from the same activity. 25 4 2019 tax reform: Income shifting incentives? In this section we analyze income shifting responses of owners-managers of firms in response to the increase in tax rates to wage-related income that took place in 2019. As highlighted previously in the document, the compo- sition of incomes for taxpayers at the top of the distribution is very different from that of the majority of taxpayers as a large share of income comes from dividends and capital gains rather than wages. However, there are reasons to think that in fact part of these returns to capital are actually returns to labor relabeled to exploit lower tax rates. In this section we investigate whether a specific group of taxpayers, the owner-managers of small partnerships, relabel their labor income as capital in- come. These are individuals that own their companies and also provide labor, and therefore have the possibility to optimize their declared income to reduce their tax liability (e.g., Harju, Koivisto, and Matikka 2022). Even though small partnerships are still a small share of all legal entities in the Lithuanian economy, their partic- ipation has increased markedly in the last decade: in Figure 9 we document that the number of small partnerships went from virtually zero in 2011 to over 20,000 in 2020, when they represented almost 17% of total legal entities in the country. In that context, understanding the implications of the personal income taxation of small partnership owners is likely to become even more relevant in the near future. 26 Figure 9: Number of legal entities by type 80,000 Private companies 60,000 Number of firms 40,000 Unlimited liability Small partnerships 20,000 0 2010 2012 2014 2016 2018 2020 year Note: The figure shows the number of legal entities over time for the three main groups of firms: joint stock companies (UAB), unlimited liability companies and small partnerships. The figure reflects the number of each group of firms filing the yearly tax forms PLN204 (joint stock and small partnerships or PLN204A (unlimited liability). Our initial approach relies on understanding the changes in incentives around the 2019 PIT reform. We perform a similar exercise to that by Alstad- sæter and Jacob (2016) for Sweden, where a tax reform decreased the dividend rate for owners of certain companies and made dividend distribution more attractive in comparison to wage payments. In the case of Lithuania, the intuition is as follows. After the 2019 PIT reform, reporting income from dividends became relatively more attractive than reporting income from wages as the rates for the latter increased (from 15% to 20% for incomes below 120 average wages and 27% for incomes above) while the rates for the former remained unchanged (15%). Individuals responding to the economic incentives present in the tax system would then shift their incomes from wages to dividends.14 For our empirical analysis we first identify individuals who report income consistent with being owner-managers of partnerships before 14 We note that one feature of the 2019 reform that subdues the incentives to shift income from wages to dividends is that, for individuals in the top PIT bracket, a ceiling on the social security contributions was introduced, effectively keeping the marginal rate inclusive of SSC constant. It is still true, nonetheless, that the increase from 15% to 20% tax rate on wages still made dividends more attractive vis-a-vis wage payments. 27 the 2019 reform;15 we then keep only those individuals filing taxes every year (i.e. define a balanced panel) and follow their income declarations over time. We do not find a shift from wages to dividend income after the reform. Figure 10 shows how labor and capital income evolved over time for the group of owners-managers. While Alstadsæter and Jacob (2016) document a stark increase in dividend income and a similar decrease in wage income following a similar reform in Sweden, for owners-managers in Lithuania the figure shows the opposite effect that we would expect: after an increase in the rate faced by labor relative to capital in 2019, they report higher levels of labor income and lower levels of capital income. There are several caveats in these results. First, the 2019 reform also introduced a mandatory increase of 28.9% in all wages to make up for the change in statutory contributions from employers to employees, so this increase in wages is observed for all workers in the economy.16 Furthermore, the tax forms and existing income categories also changed in 2019, including the introduction of new income codes for dividend distribution which might affect our estimates.17 15 We do not observe ownership data, so we cannot ascertain which individuals are part- ners/owners of which firms. We infer that individuals are members of a small partnership when they declare, at any point in the period 2016-2018, income with code 02. 16 We empirically verify the implementation if the mandate in Appendix Figure 13, which plots the coefficients of a regression for log wages on year dummies. Each coefficient represents the growth rate of wages relative to 2018, within individuals. On average, wages increased by an amount that is very similar to the mandated increase. 17 If we were to follow the same income types that existed in 2018 we would naively report a fall in incomes from capital when in fact it was due to some income types ceasing to exist. See Appendix Figure 12. 28 Figure 10: Income sources of owners-managers across time Note: The figure shows the evolution across time of labor and capital income for a balanced sample of taxpayers identified as owners-managers. However, the main reason for the lack of change towards dividends seems to be that owener-managers are already paying themselves the minimum wage. In Figure 11 we show the distribution of employment-related wages across years and highlight the amount that corresponds to the minimum wage with vertical dashed-lines: every year there is a large mass of owners-managers that report receiving exactly the minimum wage or slightly above it. It is unlikely that the real market value of the labor being provided by owner-managers, who are likely educated and specialized professionals, is the minimum wage. More likely, this suggests that even before the 2019 reform, these individuals were already maximiz- ing their income optimization: even though both wages and dividends were taxed equally at 15% before the reform, wages entailed social security contributions. As in other countries, the minimum wage is the lowest wage that can be reported in order to qualify a worker for employment-based social security benefits or health insurance. An additional incentive for reporting wage-income is that it is deductible from the employers’ CIT liability and, up to the amount equivalent to the mini- mum wage, exempt from paying PIT. The combination of these incentives imply that owners-managers were likely targeting the minimum wage to optimize their tax liability. This implies that there is very little space for income shifting out of labor after the 2019 reform. As we discuss in the introduction, some countries have 29 rules in place that aim to limit this type of behavior, by taxing dividends above a certain threshold as labor income, for example. Figure 11: Histogram of wage income Note: This figure shows the distribution of labor income reported by owners-managers for each of the years in our sample. Only positive values are showna and the distribution is cut at €10,000. Each income bin is of size €100. The vertical dashed-lines highlighted in red correspond to the annualized minimum wages in each of the years: €4,200 (2016); €4,560 (2017); €4,800 (2018); €6,660 (2019); and €7,284 (2020). 5 Conclusion This report uses administrative tax filings from all taxpayers in Lithuania to document how a schedular tax system has implications for the effi- ciency and equity of a tax system. The existence of different tax rates for labor and capital income is not an unusual feature of tax systems. In Lithuania, however, several different tax rates apply to labor income sources, including the existence of special regimes such as the IAC and BC. These income sources are declared by a minority of taxpayers—in the middle of the income distribution, taxpayers receiving only income from wages are dominant. But IAC and BC incomes are more impor- tant both at the bottom of the distribution and at the very top—with important implications for the effective tax rate in the tails of the income distribution. In a nutshell, these regimes, by design, allow individuals with very similar incomes to face vastly different tax liabilities—while often the concept of horizontal equity in 30 a tax system is linked to the core idea of individuals with similar incomes paying similar tax rates. These differences are particularly stark in Lithuania since they not only apply to differential rates between labor and capital,18 but also to different sources of labor income. The high level of schedularity also undermines the goals of progressive marginal rates. Since the 2019 reform the tax system has been moving towards more progressivity, with higher top marginal rates and a lower threshold to which these rates are applied. Since these rates are only applied to a subset of income sources and thresholds are still relatively high, few individuals face the highest rate. More importantly, the existence of multiple income sources taxed at different rates means that individuals can actively engage in tax planning. This is less likely for wage earners that receive the majority of income from a single source, but more likely for those with diverse income sources. The report also documents how taxpayers are keenly aware of tax incentives and optimize their declared income to minimize liabilities. When investigating the behavior of owners-managers of small partnerships, a group that has considerable leeway to decide how to compensate themselves, we show that these taxpayers will predominantly pay themselves the minimum wage and distribute the remaining income as dividends. Since the minimum wage is probably not the fair compensation for their labor, this behavior distorts our understanding of the sources of income for these individuals and, more importantly, show they are sophisticate in their understanding on how to distribute income in order to minimize tax liabilities. This phenomenon has been documented in several other settings and highlights that a highly schedular tax systems, coupled with few rules attempting to limit income shifting, provides very strong incentives not only for individuals to minimize liabilities when possible but also to choose corporate forms that provide tax advantages—likely sacrificing efficiency in this process. 18 The argument for lower taxation of capital income at the individual level is not only that they derive from savings, that might already have been taxed in the past, but also that the income distributed by a corporation, for example, is already taxed at the corporate level. 31 References Alstadsæter, Annette, and Martin Jacob. 2016. “Dividend Taxes and Income Shift- ing” [in en]. _Eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/sjoe.12148, The Scandinavian Journal of Economics 118 (4): 693–717. https://onlinelibra ry.wiley.com/doi/abs/10.1111/sjoe.12148. Alvaredo, Facundo, Anthony Atkinson, Lucas Chancel, Thomas Piketty, Emmanuel Saez, and Gabriel Zucman. 2020. Distributional National Accounts (DINA) Guidelines : Concepts and Methods used in WID.world [in en]. Technical re- port 201602. World Inequality Lab. https://ideas.repec.org/p/wel/wpaper/ 201602.html. Atkinson, A. B., and J. E. Stiglitz. 1976. “The design of tax structure: Direct versus indirect taxation” [in en]. Journal of Public Economics 6, no. 1 (July): 55–75. https://www.sciencedirect.com/science/article/pii/0047272776900414. Auten, Gerald, David Splinter, and Susan Nelson. 2016. “Reactions of High-Income Taxpayers to Major Tax Legislations” [in en]. National Tax Journal 69, no. 4 (December): 935–964. https://www.journals.uchicago.edu/doi/10.17310/ntj. 2016.4.10. Blanchet, Thomas, Ignacio Flores, and Marc Morgan. 2018. The Weight of the Rich: Improving Surveys Using Tax Data [in en]. Technical report hal-02878315. HAL. https://ideas.repec.org/p/hal/wilwps/hal-02878315.html. Gordon, Roger, and Joel Slemrod. 1998. Are "Real" Responses to Taxes Simply Income Shifting Between Corporate and Personal Tax Bases? [In en]. Technical report w6576. Cambridge, MA: National Bureau of Economic Research, May. http://www.nber.org/papers/w6576.pdf. Harju, Jarkko, Aliisa Koivisto, and Tuomas Matikka. 2022. “The effects of corpo- rate taxes on small firms” [in en]. Journal of Public Economics 212 (August): 104704. https://www.sciencedirect.com/science/article/pii/S00472727220010 62. Harju, Jarkko, and Tuomas Matikka. 2016. “Business owners and income-shifting: evidence from Finland” [in en]. Small Business Economics 46, no. 1 (January): 115–136. https://doi.org/10.1007/s11187-015-9679-4. 32 Herault, Nicolas, Dean Hyslop, Stephen P. Jenkins, and Roger Wilkins. 2021. Rising Top-Income Persistence in Australia: Evidence from Income Tax Data [in en]. SSRN Scholarly Paper. Rochester, NY, October. https://papers.ssrn .com / abstract=3934754. Jacob, Martin. 2021. “Dividend taxes, employment, and firm productivity” [in en]. Journal of Corporate Finance 69 (August): 102040. https://www.sciencedirect. com/science/article/pii/S0929119921001620. Nelson, Susan C. 2016. “Paying Themselves: S Corporation Owners and Trends in S Corporation Income, 1980-2013” [in en] (August): 29. OECD. 2022. Tax Administration 2022: Comparative Information on OECD and other Advanced and Emerging Economies [in en]. Tax Administration. OECD, June. isbn: 978-92-64-69704-1 978-92-64-41168-5 978-92-64-86659-1 978-92-64- 66895-9. https://www.oecd-ilibrary.org/taxation/tax-administration-2022_ 1e797131-en. Piketty, Thomas, Emmanuel Saez, and Gabriel Zucman. 2022. “Rethinking Capital and Wealth Taxation” [in en]. World Inequality Lab - Working Paper n 2022/18. Romanov, Dmitri. 2006. “The corporation as a tax shelter: Evidence from recent Israeli tax changes” [in en]. Journal of Public Economics 90, no. 10 (November): 1939–1954. https://www.sciencedirect.com/science/article/pii/S00472727060 00375. Saez, Emmanuel, and Stefanie Stantcheva. 2018. “A simpler theory of optimal capi- tal taxation” [in en]. Journal of Public Economics, In Honor of Sir Tony Atkin- son (1944-2017), 162 (June): 120–142. https://www.sciencedirect.com/science/ article/pii/S0047272717301688. Slemrod, Joel. 1995. “Income Creation or Income Shifting? Behavioral Responses to the Tax Reform Act of 1986.” Publisher: American Economic Association, American Economic Review 85 (2): 175–80. https : / / econpapers . repec . org / article/aeaaecrev/v_3a85_3ay_3a1995_3ai_3a2_3ap_3a175-80.htm. Smith, Kate, Thomas Pope, and Helen Miller. 2021. Intertemporal income shifting and the taxation of business owner-managers [in en]. Technical report. The IFS, December. https://ifs.org.uk/publications/15885. 33 Smith, Matthew, Danny Yagan, Owen Zidar, and Eric Zwick. 2022. “The Rise of Pass-Throughs and the Decline of the Labor Share” [in en]. American Economic Review: Insights 4, no. 3 (September): 323–340. https://www .aeaweb.org / articles?id=10.1257/aeri.20210268. 34 Appendix A Additional figures and tables Table 2: Income level and composition across distribution (percentiles) (1) (2) (3) (4) (5) Minimum income Wage income BC IAC Dividends Percentile (Euros) (%) (%) (%) (%) 20 4,799 0.76 0.07 0.11 0.00 30 7,398 0.93 0.02 0.03 0.00 40 9,241 0.93 0.02 0.03 0.00 50 11,383 0.93 0.02 0.03 0.00 60 13,983 0.91 0.03 0.03 0.00 70 17,212 0.92 0.02 0.03 0.00 80 21,412 0.92 0.02 0.03 0.00 90 28,927 0.86 0.04 0.04 0.01 95 37,965 0.85 0.02 0.05 0.01 99 62,435 0.71 0.01 0.07 0.06 100 86,353 0.32 0.00 0.12 0.25 Note: The table reports the minimum amount of yearly taxable income needed to be in each percentile (column (1)) and the composition of total income in each percentile by income source (columns 2 - 5) in 2020. The table does not include all possible sources of income so the shares do not sum to 100%. 35 Table 3: Income level and composition across distribution (top quantiles) (1) (2) (3) (4) (5) (6) Minimum income Wage income BC IAC Dividends Capital Gains Percentile (Euros) (%) (%) (%) (%) % Top 1% - 0.9% 86,353 0.62 0.01 0.10 0.08 0.13 Top 0.5% - 0.4% 123,059 0.46 0.01 0.12 0.14 0.19 Top 0.1% - 0.09% 298,707 0.23 0.00 0.22 0.26 0.20 Top 0.05% - 0.04% 446,646 0.22 0.00 0.18 0.32 0.22 Top 0.04% - 0.03% 505,692 0.16 0.00 0.14 0.42 0.17 Top 0.03% - 0.02% 598,409 0.16 0.00 0.12 0.48 0.15 Top 0.02% - 0.01% 748,417 0.12 0.00 0.08 0.43 0.27 Top 0.01% 1,186,292 0.08 0.00 0.04 0.40 0.27 Note: The table reports the minimum amount of yearly taxable income needed to be in each quantile (column (1)) and the composition of total income in each quantile by income source (columns 2 - 5) in 2020. The table does not include all possible sources of income so the shares do not sum to 100%. Figure 12: Income sources of owners-managers across time Note: The figure shows the evolution across time of labor and capital income for a balanced sample of taxpayers identified as owners-managers. The yellow line corresponds to unadjusted partnership income, in which we do not account for the change in income categories over the years. 36 Figure 13: Income sources of owners-managers across time Note: The figure shows the coefficients of a regression for log wages on year dummies for individuals whose main source of income was wages (98% or more of total income). Each coefficient represents the growth rate of wages relative to 2018, within individuals. B Data sources In this section we describe in more detail the construction of the datasets used for the analysis in this report. We focus on construction of datasets for 2019 and 2020, since most of the results in the analysis refer to this period and the tax forms, and therefore datasets, are different for the period before 2019. B.1 Definition of variables • Employment-related income is defined as the sum of incomes with codes 01, 02, and 03. • Income from Individual Activity Certificates is defined as the sum of incomes with codes 35, 93, 96 and 97, net of the allowable deductions for these same incomes. • Income from Business Certificates is defined as the sum of incomes with codes 90 and 92. Since deductions are not allowed for BC income (since it only pay local fees and not income taxes), we impute a 30% deduction to obtain net income, following the allowable deduction for IAC income. 37 • Income from capital gains is defined as the sum of incomes with codes 11, 12, 13, 14, 16, 17 and 18, net of allowable deductions for the same codes. • Other income encompasses dozen of other taxable income codes, including but not restricted to dividends (code 26), interest (codes 55,58, 59, 64 and 67), income from profit sharing (code 44), rental income (codes 23 and 24) and income from sports and performance activities (codes 51 and 52). Our main measure of taxable income, used to rank individuals across the income distribution, is the sum of the previously discussed income sources. We note some decisions we take when constructing total taxable income: • First, we include income from business certificates (net of imputed deductions) in the sum of taxable income. Even though income from BC is only "taxed" at a flat rate (which could be considered an operating fee and not a tax), we include it as we see no economic reason to treat that income differently from other sources. • Second, we do not include some income codes that are considered non-taxable, such as gifts (code 49) and sport prizes (code 30). • Finally, we do include "non-taxable" portions of income codes that are taxable. The most relevant example is the minimum non-taxable income (MNTI) for employment-related income. But several other income codes also allow for a portion of total income to be fully exempt. These are closer to tax expendi- tures and for that reason we do not net those values when calculating total taxable income. Finally, when calculating taxes owed, we directly obtain the total tax liability from the tax forms19 and add an imputed value for BC income. Since these fees vary by municipality, we use a reference value for Vilnius in 2020 (€650/year). We then compute the effective tax rate (ETR) as the ratio between tax liability and total taxable income20 19 Total tax liability is different from tax payable, since a large portion of taxes are withheld at source, so taxpayers do not have to pay them after filing. We are interested in the total tax liability, regardless of when they are paid. 20 When presenting average ETRs across quantiles, we compute ETR at each bin as the ratio between total tax liability and total income, which is equivalent to computing ETRs at individual level and the taking an weighted average using total income as weights. 38 B.2 Definition of owner-managers In order to follow owner-managers’ incomes before and after the 2019 reform, we first define individuals to be owner-managers if they received any income with code 02 in the 2016-2018 period. Income with code 02 refers to income withdrawn from sole-proprietorships or small partnerships that is considered as labor income. We then restrict the sample to owner-managers that file income taxes in every period between 2016 and 2020 and compute their total income from employment-related sources and capital income. We define employment related income as the sum of incomes with code 01, 02 and 03, meaning not only the funds withdrawn from the partnership but also any other wages they might receive. We define capital income as funds likely to be drawn from their partnerships, which include dividends and the distribution of profits from partnerships (codes 27, 28 and 29). 39