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) Recommendation Report Analyzing the Size and Effects of Tax Optimization and Bunching with a Microsimulation Tool Output 4 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 Development/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 Table of Contents Preface .......................................................................................................................................................... 5 Abbreviations and acronyms ........................................................................................................................ 7 Executive summary ....................................................................................................................................... 8 Chapter I. Introduction ............................................................................................................................... 10 Chapter II. Government priorities ............................................................................................................... 12 Chapter III. The Tax System in Lithuania ..................................................................................................... 14 1. Personal income tax in Lithuania .................................................................................................... 15 2. Personal income tax system in the rest of the EU .......................................................................... 15 3. Tax systems for legal entities in Lithuania ...................................................................................... 17 4. Issues related to the Lithuanian personal income tax system........................................................ 18 Chapter IV. Microsimulation Analysis of Lithuania’s Income Tax System .................................................. 21 What is a microsimulation analysis? ................................................................................................... 21 How was the base scenario model developed?.................................................................................. 21 Main logistic challenges ...................................................................................................................... 22 1. Microsimulation Analysis of Individual Income Taxation ............................................................... 22 Revenue effect .................................................................................................................................... 24 Progressivity effect ............................................................................................................................. 28 2. Microsimulation of Legal Entities Income Taxation ........................................................................ 30 Chapter V. Summary of Findings and Recommendations .......................................................................... 34 Key findings ............................................................................................................................................. 34 Main recommendations.......................................................................................................................... 35 A. Recommendations for the overall tax structure ................................................................................ 35 B. Recommendations for the PIT system ................................................................................................ 35 C. Recommendations for the tax system for legal entities ..................................................................... 36 List of Figures Figure 1. Tax-to-GDP ratio in Lithuania, OECD and Baltics (1995-2020)..................................................... 11 Figure 2. Average PIT paid annually by salaried employees and self-employed ........................................ 19 Figure 3. Number of firms in all categories that are below the EUR 300,000 threshold ............................ 20 Figure 4. Effect of unifying rate to 15 percent for all legal entities ............................................................ 31 Figure 5. Effect of unifying rate to 15 percent for all small partnerships ................................................... 32 3 List of Tables Table 1.Tax-to-GDP ratio in EU Member States, 2020................................................................................ 14 Table 2. Tax-to-GDP ratio for main taxes, Lithuania and EU average, 2020 ............................................... 15 Table 3. Differential treatment of personal income in the EU and in Lithuania......................................... 16 Table 4. General explanation of the scenarios and the objectives ............................................................. 23 Table 5. Lithuania Base Scenario................................................................................................................. 25 Table 6. Latvian Scenario ............................................................................................................................ 25 Table 7. Belgian Scenario ............................................................................................................................ 26 Table 8. Portuguese Scenario ..................................................................................................................... 26 Table 9. Greek scenario............................................................................................................................... 27 Table 10. Spanish scenario .......................................................................................................................... 27 Table 11. Portuguese Scenario with differentiated rate for interest and dividends .................................. 28 Table 12. Distributional analysis using the Portuguese scenario in the Lithuanian model ........................ 29 Table 13. Average effective tax rates for comprehensive and differentiated Portuguese scenarios ........ 29 Table 14. General explanation of the scenarios and the objectives for legal entities................................ 31 Table 15. Results of scenarios including (i) unified rate of 15 percent and (i) dual CIT rates on marginal income ........................................................................................................................................................ 33 List of Boxes Box 1. Tax policy priorities of Lithuanian Government............................................................................... 12 4 Preface This Recommendation Report Analyzing the Size and Effects of Tax Optimization and Bunching with a Microsimulation Tool covers the deliverable 4 of the technical assistance under the project “Lithuania: Micro Enterprises and Self-employed Tax & Regulatory Assessment for Removing Hurdles to Growth.â€? The project is funded by the European Union via the Directorate-General for Structural Reform Support (DG-Reform). The project consists of four Outputs. • Output 1 – Report Analyzing the Size and Effects of Tax Optimization and Bunching, with a Microsimulation Tool. • Output 2 – Report Assessing the Impacts of Tax Optimization and Bunching as well as Enterprise Formalization on Productivity and Growth of MEs and Self-employed, as well as the Overall Economic Growth in Lithuania. • Output 3 requires delivering workshops and training and providing technical support for the tools developed by the World Bank. • Output 4 provides finding and recommendations. The overall outcome expected from this project is to enhance the tools and skills of Lithuanian authorities to carry out a systematic assessment of the impacts of tax optimization and bunching behavior for SMEs and self-employed derived from tax discrepancy and different regulations, and to build options for a more informed policy choice. Originally scheduled to commence in May 2020, the start of the project was delayed first because of change in government in Lithuania, and then because of the COVID-19 pandemic, during which period several preparatory virtual meetings were held with the Ministry of Economy and Innovation (MoEI), the State Tax Inspectorate (STI) and the Innovation Agency Lithuania. Thereafter, several virtual meetings were held with Lithuanian authorities, including Statistics Lithuania, for data collection and construction of databases. The complex legal provisions were examined thoroughly, and virtual stakeholder consultations were held. Building on these consultations and data construction, microsimulation base models for legal entities and individual taxpayers were developed. Several scenarios, discussed later in this report, were then constructed. Following the easing of COVID-19 related travel restrictions, the World Bank team visited Vilnius in May 2022 and again in August 2022. During these visits, the microsimulation exercise was continued in the secure environment of Statistics Lithuania. The team also conducted workshops and training sessions to explain the simulation tool and its working, and also held discussions with various stakeholders, including the Prime Minister’s Office, the Ministry of Finance, the MoEI, the Parliamentary Budget Committee and the Central Bank of Lithuania. Feedback obtained during these meetings was incorporated into the design of the model. 5 This report has been produced by the World Bank team led by Mr. Alberto Leyton, Lead Public Sector Specialist (co-Task Team Leader), Mrs. Cristina Savescu, Senior Economist (co-Task Team Leader), Mr. Munawer Khwaja Sultan (Technical Lead, Senior Tax Expert), Ms. Dzelila Kramer (Microsimulation Modeling Expert) and Mr. Vedad Ramovic (Python Specialist). 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 counterparts 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 Department of Statistics - Edita BaltuÅ¡e, Aleksandra GoluboviÄ?, Gita Literske, Vadimas Ivanovas, Darius Abazorius and Jonas BaÄ?elis - 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, Simona 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 1 and 2: Report Assessing the Impacts of Tax Optimization and Bunching as well as Enterprise Formalization on Productivity and Growth of MEs and Self-employed, as well as the Overall Economic Growth in Lithuania. This final version of the report has benefited from numerous comments from Lithuanian authorities, as well as the European Commission and World Bank officials. 6 Abbreviations and acronyms AMS Average Monthly Salary AW Average Wage CIT Corporate Income Tax EC European Commission EATR Effective Average Tax Rate ETR Effective Tax Rate EU European Union EUR Euro GMT Global Minimum Tax IAC Individual Activities Certificate IBRD International Bank for Reconstruction and Development MEs Micro Enterprises MoEI Ministry of Economy and Innovation NPV Net Present Value NTMI Non-Taxable Minimum Income OECD Organization for Economic Co-operation and Development SMEs Small and Medium Enterprises RRF Recovery and Resilience Facility (of the EU) SRPS Structural Reform Support Programme SSC Social Security Contribution STI State Tax Inspectorate PIT Personal Income Tax VAT Value-added tax WB World Bank 7 Executive summary i. This Executive Summary lays out the key findings and recommendations of the report prepared by the World Bank under the technical assistance project “Lithuania: Micro Enterprises and Self- employed Tax & Regulatory Assessment for Removing Hurdles to Growth.â€? The report corresponds to Output 4 of the project. It analyzes the size of tax optimization and bunching and the effects, with a microsimulation tool. It also provides recommendations, including proposed appropriate indicators and benchmarks, policy options for achieving tax neutrality and recommendations on reform sequencing. These recommendations are relevant for, and consistent with, Lithuania’s commitment to the EU Directive on Global Minimum Taxation and to the EU Recovery and Resilience Facility (RRF). ii. The Lithuanian tax system and regulatory environment are characterized by multiple enterprise formalization options and by differentiated (schedular) taxation of different sources of income and forms of activity. Differentiated application of personal income tax, social security contributions, and corporate income tax to various sources of income (employment, self-employment, dividends, rents, interest income) and forms of taxable activities create discrepancies in tax rates, which encourage entrepreneurs to tax optimize in some cases at the expense of business growth (as well as economic growth if Micro Enterprises (MEs) are less productive and add less value to GDP than larger enterprises). Size-based tax preferences create growth barriers by introducing a sharp increase in marginal tax rate beyond the preferential threshold. These tax preferences can create incentives for MEs to remain small and discourage self-employed from growing into larger, incorporated entities thereby forgoing economies of scale. iii. Neutrality is one of the key principles of tax policy and improving the neutrality of the tax system – with the underlying improvement in efficiency and equity - would require a move away from a highly schedular tax system, as is the case in Lithuania. Such a system makes it more difficult to achieve a more progressive tax system (fairness/equity principle), because it does not lend itself for introducing a progressive rate structure since income is not aggregated across different income types and thus the tax liability is not necessarily correlated with the ability to pay. Furthermore, on efficiency grounds, a schedular system introduces distortions. While most countries have nominally global tax systems, the reality is that all tax systems provide for different treatment for different types of income (such as income from capital gains, owner-occupied housing, and retirement savings). iv. The microsimulation tool aims to assist Lithuanian authorities in evaluating the impacts of tax optimization and tax bunching behavior of taxpayers, in order to inform tax policy design. The scenarios presented in the model using other eurozone countries are illustrative and not recommendations. The main goal is to leave behind an analytical tool that will support the development of options that would allow for increased tax neutrality and encourage efficiency, productivity growth and equity, as well as diminishing the shadow economy. v. Lithuania is one of the EU countries with the highest level of income disparity, and a high level of inequality. Two main reasons for that are that Lithuania’s tax system lacks progressivity and has a schedular taxation system where labor income is taxed higher than capital income. Additionally, individuals with lower income bear a higher effective tax rate, based on the fact that there are only two 8 tax brackets for employment income with a high threshold. This has implications for Lithuania’s commitment to foster sustainable and inclusive growth under the RRF. vi. The report lays out some other key findings. First, Lithuania’s tax-to-GDP ratio is significantly lower than the EU averages. This has relevance for Lithuania’s commitment to the EU Directive on the GMT. Lower domestic revenue mobilization limits the country’s ability to invest in innovation and social welfare, like healthcare and education. Second, the dual rate system creates a disincentive for businesses to grow since beyond the preferential taxation threshold they would face a much higher rate on the entire income. The analysis documents that Lithuanian SMEs are “bunchingâ€? at a preferential taxation threshold, which is detrimental to growth and productivity. vii. Ensuring a higher level of domestic revenue mobilization would require gradually improving the tax-to-GDP ratio, in small incremental steps, to bring it as close to the EU average as feasible. Since the low tax-to-GDP ratio is largely the result of low PIT-to-GDP and CIT-to-GDP ratios, one way of achieving a higher intake is to increase these two ratios to move them closer to the EU average and help Lithuania meet its EU Directive commitment of ensuring a GMT. viii. Another recommendation is replacing the extremely schedular PIT system with a more comprehensive income approach in which most sources of income are aggregated. Such an approach allows then to introduce a more progressive rate structure. The microsimulation model can be used to examine the effect of different options for income brackets. Removing income disparities will address Lithuania’s commitment to the RRF pillar on achieving inclusive growth. The CIT microsimulation model can also be used to examine various options for the CIT rate structure. A key recommendation is to unify the rate structure for all legal entities to 15 percent. If unifying the rate structure is not feasible in the short term, an alternative is to start with the standard rate of 15 percent as a marginal rate for income about a desired income threshold, reducing the incentive to bunch at the threshold. Subsequently, within the next couple of years using the 15 percent rate as marginal rate, will achieve a full unification of the CIT rate can be considered. 9 Chapter I. Introduction 1. Despite Lithuania’s strong economic performance in recent years, the size of the informal economy remains large in international comparison. Informality as a share of GDP in Lithuania has been in line with that of Estonia up to 2017 but it has since diverged, moving notably higher in recent years. According to OECD’s Interim Report (July 2022), informality in Lithuania is partly driven by a part of salary being in the form of an undeclared cash payment, which is of greater concern than other forms of under- reporting.1 2. Lithuania has a modern OECD economy trapped in a transition-economy style income taxation system which is one of the factors for its low productivity. In 2021, labor productivity in Lithuania was much lower than the OECD average and ranked 25th out of 38 OECD countries.2 Lithuania has since the 2009 global financial crisis experienced a decrease in the average annual productivity growth rate (11.0 percent in 2002-2008 and 5.2 percent in 2009-2017). During the same period (2009-2017), the number of Micro Enterprises (MEs) (defined by 0-9 employees) has increased by 30 percent, while at the same time the numbers of larger enterprises have decreased or remained stable. The European Commission's Lithuania Country Report for 20223 recommended that resource productivity could be considerably improved. The efficient use of resources is key to ensuring competitiveness. 3. The 2022 Country Report mentions that social and economic disparities are pronounced, and regional convergence remains a challenge. Among other reasons, the schedular system of taxation perpetuates that as labor income (wages) is taxed at relatively higher rates than capital income (business profit, dividends, interest, rent, royalties). The introduction of a progressive Personal Income Tax (PIT) system for labor income, when coupled with the introduction of the social security contributions (SSC) ceiling system, has not translated in any significant reduction in economic disparities. 4. The personal income tax system is not progressive at the top, with just two rates, at a relatively high threshold, compared to the rest of the OECD. The entry level tax rate for low net income taxpayers is relatively high compared to the OECD average, which results in lower incentive to work and lower productivity.4 Low-income second earners have lower incentives to enter work at short unemployment spells but quite strong incentives at long unemployment spells. Disincentives to enter work appear a more important area of policy concern than disincentives to progress in work.5, 6 Countries can increase 1 OECD (2022). “Strengthening the design of the tax, social security contributions and social benefits system in Lithuania to reduce inequality and poverty and stimulate economic growth.â€? Interim Report. July 2022. 2 Ibid. 3 European Commission. 2022 Country Report – Lithuania. https://ec.europa.eu/info/system/files/2022-european-semester- country-report-lithuania_en.pdf. Brussels. May 2022. 4 NBER (2004). Les Picker. The Digest. Effect of Taxes on Labor, December 2004. The paper highlights that higher tax rates on labor income lead to less work time in the legal market sector, a larger underground economy, and smaller shares of national output and employment in industries that rely heavily on low-wage, low-skill labor inputs. 5 OECD (2022). “Strengthening the design of the tax, social security contributions and social benefits system in Lithuania to reduce inequality and poverty and stimulate economic growth.â€? Interim Report. July 2022. 6 European Commission. Technical Support Instrument. 20LT01 – Strengthening the design of the tax, social security contributions and social benefits. 10 productivity by tackling the barriers that give rise to poor use of existing resources within countries— resource misallocation.7 5. Lithuania’s tax-to-GDP ratio is significantly lower than the OECD and EU averages (Figure 1), and lower than the average for the Baltics countries. Lithuania increased its tax-to-GDP ratio by almost 1 percentage point between 2019 and 2020, among the highest year-on-year changes in the OECD but most of the increase in Lithuania is attributable to an increase in the SSC-to-GDP ratio.8 Figure 1. Tax-to-GDP ratio in Lithuania, OECD and Baltics (1995-2020) Source: OECD Interim Report July 2022 6. Disparities in disposable income are high, with Lithuania placed among the top six OECD countries in terms of disparities. The redistributive role of taxes and transfers has been limited here.9 There are several reasons for this. Lithuania has a regressive income tax rate structure. For PIT, owners of capital income such as rent, profit, interest, royalties, and dividends are taxed at a relatively lower tax rate than taxpayers with labor income. The schedular system ensures that owners of capital continue to be taxed at a lower rate with no aggregation of comprehensive income from all sources. Likewise, the tax brackets for employment income have only two rates and a high threshold which means that many taxpayers with relatively high salary relative to the average wage, still pay tax at the lower rate. Also, the entry level tax rate is high relative to that prevailing in many EU countries which casts a heavy burden on low-income taxpayers. As discussed earlier, this creates a disincentive to enter the formal work force and contributes to labor informality of low-income wage earners. 7. Research done in Lithuania and by the World Bank demonstrate that Lithuanian SMEs are “bunchingâ€? at a preferential taxation threshold of EUR 300,000 in annual revenues for the years of 2013- 2020. “Bunchingâ€? is evidence of tax optimisation, which is detrimental to growth. 7 IMF (2017). Vitor Gaspar and Laura Jaramillo. “Designed for Growth: Taxation and Productivityâ€?. April 2017. 8 The comparison between 2019 and 2020 should be understood with caution since 2020 was an unusual year in which economic activity has been affected by the COVID-19 pandemic. 9 OECD (2022). “Strengthening the design of the tax, social security contributions and social benefits system in Lithuania to reduce inequality and poverty and stimulate economic growth.â€? Interim Report. July 2022. 11 Chapter II. Government priorities 8. To address some of the issues discussed in Chapter I above, the Government of Lithuania has put forward new policy initiatives. These priorities are presented in Box 1. The policy initiatives of the Government are commendable and challenging. They are in line with the policy recommendations resulting from the World Bank analysis. Box 1. Tax policy priorities of Lithuanian Government Reducing the VAT tax gap to 10 percent by 2030.10 A more holistic approach to public finance. A more equitable and growth-friendly tax system. Assessing the appropriateness and impact of tax incentives and reviewing special conditions that distort tax justice, creates tax arbitrage, and promotes shadow economy. Prioritizing the increase of property taxes and environmental taxes. Encouraging entrepreneurship for microenterprises. 9. After the start of this WB project, the Lithuanian government has entered into two new regional/international commitments that have a bearing on the above policy initiatives. The first commitment is to the Recovery and Resilience Facility (RRF) of the European Union11 that helps the EU emerge stronger and more resilient from the financial crisis resulting from the pandemic. Two of the six pillars have significant relevance to the WB project: (i) smart, sustainable and inclusive growth; and (ii) policies for next generation. The second commitment is to the European Council Directive of 15 December 2022 for ensuring a global minimum level of taxation for multinational and large domestic groups in the Union.12 EU Member States have until 31 December 2023 to transpose the Directive into national legislation with the rules to be applicable for fiscal years starting on or after 31 December 2023. Both these commitments, the RRF and the Global Minimum Tax (GMT) have a significant relevance on the WB recommendations for augmenting progressivity and neutrality in PIT (as a means for achieving inclusiveness and the underlying efficiency and equity goals) and increasing tax-to-GDP ratio (for achieving the GMT). 10. In order to assist the Lithuanian Government in achieving these priorities, the WB performed a series of analyses synthesized in this report. The first part consists of an analysis of the tax system in Lithuania, with a focus on personal income tax (PIT) and corporate income tax (CIT), including a comparison of the differential treatment of personal income in the EU and in Lithuania. Furthermore, the WB team identified issues related to the Lithuanian PIT system that create obstacles for the government’s priorities. These include: the regressive PIT system that creates disincentives to entering the labor market, 10 VAT gap reduction is not within the scope of this project. 11 European Commission (2020). Recovery and Resilience Facility. Brussels. April 2020. https://commission.europa.eu/business- economy-euro/economic-recovery/recovery-and-resilience-facility_en# 12 European Commission (2022). COUNCIL DIRECTIVE on ensuring a global minimum level of taxation for multinational groups in the Union. Brussels. Dec 13, 2022. https://ec.europa.eu/commission/presscorner/detail/en/ip_22_7674 12 the low tax-to-GDP ratio, the lack of tax neutrality due to the schedular system for PIT, etc. This lack of tax neutrality adversely affects the underlying efficiency and equity goals of the tax system. A similar analysis was performed for the CIT. The main issue identified as creating obstacles in this analysis was the different CIT rates based on turnover, which incentivizes enterprises to “bunchâ€? at the threshold and under-declare income. 11. Additionally, the WB team performed a microsimulation analysis of Lithuania’s income tax system. The objective of the PIT microsimulation analysis was to test several options that would address the Government’s objectives of having a more equitable and growth friendly system. Similarly, the objective of the microsimulation analysis for legal entities aimed to test options that would address the Government’s objectives for taking a more holistic approach to public finance, encouraging entrepreneurship for microenterprise and enhancing growth, and assessing the appropriateness of tax incentives that distort tax justice, create tax arbitrage and promote shadow economy. 12. Creating a more holistic approach to public finance and designing a more equitable and growth- friendly tax system will require a series of actions. The most important ones are: • a thorough analysis of the tax incentive structure and its effectiveness in achieving policy objectives. • ending (schedular) taxation that encourages informality and inhibits growth and productivity and at the same time discourages VAT registration. • improving the progressivity of the PIT system to improve horizontal and vertical equity and thereby creating a more equitable tax system which will incentivize lower income workers to join the formal labor market. • creating incentives for self-employed to incorporate and not hide behind the schedular differentiated rate structure. 13. At the same time, efforts must be made to support genuine microenterprises. Encouraging entrepreneurship for genuine microenterprises requires that the tax system remove the differentiated CIT system to curb opportunities for larger enterprises to drive microenterprises out of business by splitting and hiding as pseudo microenterprises. 14. Success will depend on weighing the various options objectively to arrive at the best solution. This report provides options for making informed policy choices by microsimulating various options and demonstrating the impact of the possible right choices. This report is not prescriptive. It does not provide a predetermined design. It gives the Government a powerful analytical tool to analyze various policy options and choose what best suits its objectives. 13 Chapter III. The Tax System in Lithuania 15. Lithuania has one of the lowest tax-to-GDP ratios13 in the European Union (EU), second only to Ireland. Compared to an EU average tax-to-GDP ratio in 2020 of 37.6 percent (37.4 percent 5-year average for 2015-2019), Lithuania has a tax-to-GDP ratio in 2020 of 31.3 percent (29.7 percent 5-year average for 2015-2019). While this is an improvement from 30.3 percent in 2019, this is still one of the lowest in the EU. In Table 1 below, the three highest and the three lowest tax-to-GDP ratios in the EU are presented.14 Table 1.Tax-to-GDP ratio in EU Member States, 2020 Tax-to-GDP ratio EU Member State (highest – (percent) lowest) Denmark 46.5 France 45.4 Belgium 43.1 EU average 37.6 Latvia 31.9 Lithuania 31.3 Ireland 20.2 Source: OECD. In terms of comparison of the tax-to-GDP ratio by major taxes, there are significant gaps in Lithuania’s tax-to-GDP for CIT, PIT and SSC compared to the EU average. (Table 2). Table 2 indicates that there are significant gaps between what Lithuania collects and what the average collection for the EU is. There is a 6.3 percent gap as a percent of GDP in overall taxes compared to the EU average, a 2.0 percent gap in PIT and a 0.7 percent gap in CIT.15 One of the main reasons for the lower tax collection is that Lithuania still uses an extremely schedular income taxation which has been abandoned long ago by most advanced countries. This system allows taxpayers to game the system by designating incomes into categories that bear the least taxes. The differentiated CIT system encourages firms to stay small by splitting into smaller firms in order to pay tax at a lower rate. Data constraints prevented an in-depth analysis of income shifting behaviors. 13 The tax-to-GDP ratio includes corporate income tax, personal income tax, value-added tax, social security contributions and other taxes. 14 OECD Revenue Statistics (up to 2020 data):https://stats.oecd.org/Index.aspx?DataSetCode=RS_GBL. 15 Eurostat data (2020 14 Table 2. Tax-to-GDP ratio for main taxes, Lithuania and EU average, 2020 Tax type Average of EU Lithuania Lithuania’s gap Member State (percent of with EU (percent of GDP) GDP) average (percent of GDP) Overall 37.6 31.3 6.3 PIT 9.2 7.2 2.0 CIT 2.3 1.6 0.7 SSC 12.2 10.4 1.8 Source: OECD. 1. Personal income tax in Lithuania 16. The introduction of the progressive PIT system for wage income in 2019 has not translated to much improvement in the redistributive role of taxes. As discussed earlier, the level of progressivity in the rate structure for wage income is very low. There are only two tax brackets for wage income with a high threshold which means that many taxpayers with relatively high salary still pay tax at the lower of the two rates. Likewise, the lower of the two rates is high for low-income taxpayers and places a heavy burden on them. 17. Lithuania is one of the very few OECD countries where personal income taxation (PIT) is extremely schedular. Many EU countries have some level of schedular rates for dividends and interest but not for other sources of income. Moreover, the schedular system in Lithuania is very complex with tax rates and exemptions determined by type of income, type of activity and taxpayer category. Just by way of illustration, the PIT return form mentions 97 income codes of which 25 have been designated as either “non-taxable" or "non-taxable if limits not exceededâ€? or “non-taxable if conditions metâ€?.16 Still, with 72 existing categories of taxable income, with varying rates for different income types and taxpayer type, the rate structure is so fragmented that it does not allow for horizontal equity, where are all incomes are treated alike. This results in higher compliance costs for taxpayers as well as higher tax administrative costs. 2. Personal income tax system in the rest of the EU 18. The Lithuanian PIT system is not in line with the PIT structure in most advanced countries nor in line with the EU where most sources of net income are aggregated, and the comprehensive net income is taxed at a progressive rate structure. Meanwhile, there is a wide variation in the top marginal tax rate, from as high 55 percent and over in Denmark, France and Austria to as low as 15 percent in the Czech Republic, Hungary, and 20 percent in Latvia. The Czech Republic, Hungary, and Latvia apply their 16 Many of these are minor incomes and relate to daily allowances, awards and prizes, income from not-for-profit organizations, income from business certificates or from sports or performance activities. 15 flat personal income tax on all income earned. In contrast, Austria’s top statutory rate of 55 percent only applies to income above €1 million.17 19. Dividend and interest incomes are the only income types where most EU countries have a separate rate structure, although in most of them the withholding rate is higher than in Lithuania. The differential treatment of personal income in the EU and in Lithuania is presented in Table 3 below:18 Table 3. Differential treatment of personal income in the EU and in Lithuania Type of income Treatment in the EU Treatment in Lithuania Business income and wage Net business income and net agricultural Income from individual business activity is income incomes are aggregated with wage taxed at a rate of 5 percent to 15 percent. income and follow the same progressive rate schedule. Business income of a sole proprietor is taxed at a rate of 15 percent. Income under a business activity certificate is taxed at a rate of 0 percent if not exceeding EUR 45,000. Individual agricultural activity is taxed at a rate of 5 percent to 15 percent if the taxpayer is a VAT payer and at a rate of 0 percent if non-VAT payer. For wage income there is a complex bracket structure not based on euro value but on average monthly salary (AMS). The rate is 20 percent for income up to 60 AMS and 32 percent for income above 60 AMS. The AMS in Lithuania is calculated based on a specific formula and set annually by Lithuanian legislation. For 2020, it was set at EUR 1,241. Rental income In 18 EU countries, net rental income is 15 percent for income up to 120 AMS and 20 aggregated with wage income and percent for income exceeding 120 AMS. follows the same progressive rate schedule as wage income. Even where it is taxed separately, there is a progressive tax rate for rental income, for example in Finland and Greece.19 In Sweden rental income is taxed at a flat rate of 30 percent. Capital gains In about half the EU countries, capital 15 percent for income up to 120 AMS and 20 gains on real estate are taxed similar to percent for income exceeding 120 AMS. net rental income and included in comprehensive total income and taxed progressively. Capital gains on financial 17 Eurostat data (2020 18 Source: IBFD, European tax handbook 2017. 19 In Finland, rental income is taxed progressively at 30 percent for income up to EUR 30,000 and 34 percent above that threshold. In Greece, the progressive rates are 15 percent for rental income up to EUR 12,000 and go progressively up to 45 percent for rental income above EUR 35,000. 16 Type of income Treatment in the EU Treatment in Lithuania assets are, likewise, taxed progressively in about half the EU Member States. Interest income Most countries tax interest separately 15 percent for income up to 120 AMS and 20 with a withholding tax, because of the percent for income exceeding 120 AMS. ease of collection while some have progressive rates going up to 45 percent. Where it is taxed separately, the rate for final withholding varies from 10 percent in Latvia, 15 percent in the Czech Republic, to 25 percent in Germany and the Netherlands, 26 percent in Italy, 30 Percent in Sweden, 33 percent in Ireland and 34 percent in Finland. Dividend income Most countries tax dividend income This is taxed at a flat rate of 15 percent. separately with a withholding tax, because of the ease of collection, many with a flat rate, while some like France have progressive rates going up to 45 percent. For dividend income, the issue of economic double taxation is one of the reasons why dividend income is treated separately.20 Final withholding typically follows the same rate as that for interest. See interest rates above. 20. Thus, most EU countries aggregate incomes from wage, business and rental income into comprehensive total income and tax them at progressive rates. Except for five countries, most EU Member States levy tax on interest and dividend incomes as a final withholding at source by the payer. The rate for the final withholding varies from 10 percent to 34 percent. 3. Tax systems for legal entities in Lithuania 21. Lithuania has about 104,000 legal entities which are divided into several categories that have differentiated tax obligations. The largest group, about 83,000, are private companies followed by about 20,000 small partnerships. Other categories include agricultural companies (500 in 2020), cooperative partnerships or companies (425), public companies (266), professional partnerships of lawyers (20), municipal enterprises and state-owned companies (16), and public institutions (6). 22. Lithuania’s standard statutory corporate income tax (CIT) rate for legal entities is 15 percent which is significantly lower than the EU average statutory CIT rate of 21.7 percent. This is lower also than the worldwide average which was 23.54 percent in 2021 (measured across 180 jurisdictions). The OECD average CIT rate was 23.04 percent during the same period. Globally, 22 countries have a CIT rate 20 Dividend income is usually taxed at a lower rate than other sources of income because of the notion of economic double taxation where the same corporate profit is said to be taxed twice, once in the hand of the company and then at the hand of the shareholders. Traditionally, many countries provided individuals a “dividend imputation creditâ€? which is a dividend tax credi t to fully or partially offset the corporate income tax paid on its earnings. Some other countries provide a partial deduction of dividend paid at the corporate level, similar to a deduction of interest payment. 17 of between 30 and 35 percent and 115 countries impose a CIT rate of between 20 and 30 percent. Portugal has the highest statutory CIT rate among European OECD countries, at 31.5 percent. Germany and France follow at 29.9 percent and 28.4 percent respectively. Hungary (9 percent), Ireland (12.5 percent), and Lithuania (15 percent) have the lowest corporate income tax rates.21 23. Lithuania has a reduced rate of 5 percent for all SMEs (small partnerships/proprietorships/others) with up to EUR 300,000 in turnover and 10 or less employees. This reduces the effective average tax rate (EATR)22 to 13.5 percent, about 7 percent below the average EATR of 20.4 percent across OECD, G20 and participating Inclusive Framework jurisdictions. Lithuania has the sixth lowest EATR among the 117 participating countries.23 24. Size-based tax preferences can create growth barriers by introducing a sharp increase in marginal tax rate beyond the preferential threshold. These tax preferences create incentives for SMEs to remain small and discourage self-employed from growing into larger, incorporated entities thereby foregoing economies of scale. 4. Issues related to the Lithuanian personal income tax system 25. The disincentive for entering the labor market comes from a highly regressive income tax system. For low-income single individuals, EATRs in Lithuania are high in international comparison.24 At higher incomes, average EATRs are relatively flat. The PIT burden on low-incomes as a share of the PIT burden on high-incomes is the fifth highest in the OECD in 2021, indicating little PIT progressivity. 26. Revenue adequacy is a major issue. As discussed earlier, Lithuania has a low tax-to-GDP ratio which is partly the result of low PIT-to-GDP and CIT-to-GDP ratios. This reduces the country’s fiscal resilience and can have serious implications on its vulnerability to regional and global shocks. 27. Effective PIT rate on labor (net income from salary) is 3.4 times higher than PIT on capital (net profit from IAC, dividends, interest, rent, capital gains, etc.) which is among the highest in the EU. Most advanced tax systems in Europe have full neutrality of 1:1.25 28. The high effective PIT rate on labor income has been caused by taxing net wage income, at a higher rate than net self-employed income. In Lithuania low and middle-income taxpayers receive more than 80 percent of their income from wages while the high-income taxpayers receive most of their income as profit from business, professions, rent and return on financial assets (dividends, interest, capital gains, etc.). Figure 2 compares the average PIT paid annually by salaried employees in Lithuania as opposed to self-employed who own an individual activity certificate. 21 Tax Foundation data: https://taxfoundation.org/corporate-tax-rates-europe-2022/#:~:text=On%20average% 2C% 20European%20OECD%20countries,was%2023.54%20percent%20in%202021. 22 The effective average tax rate (EATR) reflects the average tax contribution a firm makes on an investment project earning above-zero economic profits. It is defined as the difference in pre-tax and post-tax economic profits relative to the NPV of pre- tax income net of real economic depreciation. 23 OECD (2021). Corporate Tax Statistics. Third Edition. 24 Eurostat data (2020). 25 Eurostat data (2020). 18 Figure 2. Average PIT paid annually by salaried employees and self-employed Source: STI data and WB staff calculations.26 29. In addition to increasing disparity, and reducing the incentive to enter the labor market, the current schedular system for PIT encourages tax avoidance because wage income is declared as self- employed income to avoid tax. It also prevents formalization and growth of self-employed into incorporated entities. 30. One of the key principles of taxation is that of neutrality, and the schedular tax system violates this principle. A more neutral PIT system that does not distinguish between the type of the taxpayer or the nature of the taxable income (or tax base) will diminish the incentive to shift income and will create an incentive for individuals to enter the official labor market. This will achieve the goal of creating a more efficient and equitable tax system, and ultimately will increase domestic revenue mobilization by tackling under-declaration and tax evasion. In a neutral PIT system, the rates are the same whether one is a wage earner or self-employed. 31. For the income taxation of legal entities, differentiated rates of 5 percent and 15 percent for corporate income tax (CIT) based on turnover are not a good international practice. It could generate incentives to be in the shadow economy by creating incentives for firms to stay under the threshold of EUR 300,000 in order to pay the 5 percent tax instead of the 15 percent rate. It might also encourage large firms to split into smaller firms to take advantage of the preferential tax treatment. This leads to “bunchingâ€? of firms at threshold level increasing the shadow activity. This limits upward mobility and leads to loss of productivity. 26 Based on data for 2020 from the base scenario microsimulation model on amount of PIT paid by salaried and self-employed taxpayers divided by the number of salaried and self-employed taxpayers respectively. 19 32. Fragmentation of firms and non-incorporation to stay in special tax regimes contributes to non- compliance, especially in terms of VAT registration. It reduces the incentive to invest in technology to its fullest potential. The large shadow economy created by the “bunchingâ€? effect leads to lower VAT registration which results in both a large VAT gap and low tax-to-GDP ratio. Figure 3 below from the World Bank microsimulation analysis shows that only 582 small partnerships out of 20,000 show turnovers above the threshold of EUR 300,000. In fact, about 19,000 of the 104,000 firms show zero income. Figure 3. Number of firms in all categories that are below the EUR 300,000 threshold Only 582 out of ~ 20,000 small partnerships have turnover > EUR 300,000 20 Chapter IV. Microsimulation Analysis of Lithuania’s Income Tax System What is a microsimulation analysis? 33. Microsimulation modelling techniques are used to determine the potential impact of any fiscal or economic change on different outcome variables for each member of a representative sample, or the entire universe, as the case may be, of taxpayers. It can be thought of as an impact analysis used for an ex-ante understanding of the expected impacts of policies and economic changes, especially in cases when actual data is either asymmetric, broken, or unavailable. 34. The approach involves obtaining inputs from micro-level databases such as tax returns of individual records to see how the existing and proposed policy provisions apply to individuals and maintaining the simulated outputs for further analysis. These models, instead of working with the aggregate or tabular data, work with large micro databases (the entire universe of tax returns), taking into account the diverse circumstances and income characteristics of the relevant population. 35. It is for these reasons that micro-simulation models are used to evaluate the effects of proposed tax interventions before they are implemented in the real world. Microsimulation models help understand the impact of policy changes on taxpayer’s liability. How was the base scenario model developed? 36. In Lithuania, separate microsimulation models have been developed for taxation of individuals and for legal entities. First, the World Bank team created a data request on an Excel sheet in consultation with Lithuanian authorities for obtaining sample return data of about a thousand legal entities and later about two thousand individuals. Once the sample data were received, base scenarios, separately for individuals and for legal entities, were developed in an Excel sheet, which contained formulae for each bit of computation. 37. The base scenario was then tested against actual data. Any discrepancy in data required fine tuning of the database by revising the formulae. Assistance for revisiting the database and fine tuning the databases was provided by officials from Statistics Lithuania, the State Tax Inspectorate, and specialists from Ernst & Young. Once the base scenario was tested with the sample data and after several iterations, found to be robust data, the entire universe of return data was requested. 38. The entire universe of return data, separately for legal entities and for individuals, was then constructed on the base model. After cleaning up the database for discrepancies, again in collaboration with Lithuanian authorities, the two full models were tested again. There was an error of 0.4 percent for the PIT model and of 0.75 percent for the CIT model which are well within the 2-3 percent acceptable limit for error in tax microsimulation models. 39. Being confident of the robustness of the base scenarios, various policy intervention scenario options were tested to determine the tax revenue impact and the distributional impact of each scenario on different categories of taxpayers. Various possible scenarios were discussed with the Lithuanian authorities prior to running them. With training provided to the Lithuanian officials on running the model, 21 the Government of Lithuania can use the models in the future to test other policy options that may require simulation of revenue impact and distributional impact. Main logistic challenges 40. The confidentiality of individual tax return data is an important issue to consider. A Confidentiality Agreement had to be signed between the World Bank and Statistics Lithuania to protect data confidentiality. As a result, most of the tasks including (i) receiving sample data, (ii) testing the database, (iii) receiving all tax return data and (iv) running the base model had to be done on the premises of Statistics Lithuania on the Palantir platform. The World Bank team visited Vilnius twice, in May 2022 and July 2022, to perform these tasks. These were then transferred from Stata into the Python programming language, since the latter is the one used by Statistics Lithuania. 41. Initial data of tax returns contained a significant number of discrepancies. Some were due to taxpayers’ filing errors, others to the existing schedular PIT system in Lithuania which is exceedingly complex with about 90 categories of taxpayers and income classification with different computation formulae for each. Just by way of illustration, the Python programming required about 1200 lines of coding to build the base model. In comparison, the scenario where all incomes are aggregated into a comprehensive total income, required only about 100 lines of coding. 42. For tax return data of legal entities, the same level of data security was not needed. Accordingly, return data was sent by Statistics Lithuania online, and most of the microsimulation was accomplished offsite. 1. Microsimulation Analysis of Individual Income Taxation 43. The objective of the microsimulation analysis is to test several options that would address the objectives of the Government to have a more equitable and growth-friendly tax system. One of the objectives is to see the impact of moving from a schedular tax system to a system where all sources of incomes are aggregated into a comprehensive total income and to apply a progressive tax rate structure for this total income. This will test the impact of horizontal equity. The other major objective of the microsimulation is to try scenarios where the PIT-to-GDP ratio can be raised from 7.2 percent so as to reduce the 2.0 percent of GDP gap with the EU average of 9.2 percent. 44. The PIT burden on low-incomes as a share of the PIT burden on high-incomes is the fifth highest in the OECD in 2021, indicating very little progressivity. Since there is very little progressivity in the current system, with the tax average ETR being flat for high income taxpayers while low-income single individuals face a high average ETR by international comparison, the microsimulation analysis has tried various options for reducing the burden on low-income taxpayers and increasing the average ETR for high- income taxpayers. This is intended to show the impact of improving progressivity and enhancing vertical equity. 45. Finally, instead of using average wage (AW) as a basis for setting tax brackets, the microsimulation will look at scenarios where the tax brackets are set as monetary (euro) values instead. This will make the front-end rate structure more transparent and easier to understand by the ordinary 22 citizen. Needless to say, at the back end, the rate structure, could probably continue to use multiples of AW as basis for various rate brackets. Table 4 below compares the current base scenario with the new scenarios and provides the reasons for the analysis. Table 4. General explanation of the scenarios and the objectives Current base scenario Different new scenarios Objective • Schedular system • Same progressive rate for Achieve horizontal and comprehensive income vertical equity through taxation progressive taxation • > 2000 rows of • 80 rows of formula in Python Reduce complexity formulae in Python • All sources of income (wage, Improve the • Different rates for business profits, dividend, progressivity of personal types of income rent, interest, capital gains) income taxation to be aggregated and taxed together Reduce barriers to entry at the lower end of the • Significantly lower rate • One scenario has labor market for individual business differentiated rates for income than for wage dividend and interest Achieve close to 8.0 income percent PIT-to-GDP ratio • Same progressive rates to as per government’s • Simplified accounting apply for all sources of reform agenda (simplified ledger or a income presumptive 30% of turnover as a tax base) • Simplified accounting or presumptive not to be applied for educated • Tax brackets are professions (e.g., doctors, Make the rate structure adjusted by MW, AW lawyers, engineers, more transparent and (the value of AW is set consultants) easier to use. each year by decree) • Tax brackets designated in a more transparent monetary terms, not based on average wage 46. There are 1,364,392 tax returns filed in 2021 for the fiscal year 2020 by individuals. A total of EUR 3,229 billion amounting to 7.2 percent of GDP was collected 2021 from individuals from all sources of income combined. 47. As it has been previously mentioned, Lithuania has a set of progressive rates for wage income and a different set of progressive rates for interest and rental incomes. The two sets of income are not aggregated into a comprehensive total income. There is a non-taxable minimum income (NTMI) of EUR 5,000 for non-wage income. 23 48. For this simulation analysis, by way of example, five eurozone countries are examined that use a comprehensive definition of income for income taxation and have progressive rates: Latvia, Belgium, Portugal, Greece and Spain.27 It needs to be reiterated that the five countries examined are purely illustrative and used for training purpose to demonstrate to the authorities how to develop their own models. The bracket thresholds in the five countries are illustrative scenarios not intended as recommendations to move to identical tax structure and thresholds. The examples of eurozone countries was intended to avoid complexity of exchange rate conversion. The model can, obviously, microsimulate any other desired rate and bracket structure, including looking at some aspects of differentiating the taxation of dividends which is a common practice in many OECD countries.28 49. Data constraints prevented an in-depth analysis of income shifting behaviors. As such, the estimates do not take into account behavioral responses and therefore could be considered lower bounds. Potential future work beyond this project could look into quantifying the size of these behavioral responses. 50. For the microsimulation exercise, the following steps were taken: • First, Lithuania’s PIT data from the base scenario model is aggregated to include all incomes to create a comprehensive income scenario. • This comprehensive income from the base model is then microsimulated by attributing the rate and bracket structure of these five EU countries.29 • For each of the five countries, two scenarios for NTMI are tested, one using the existing EUR 5,000, and the other using a higher NTMI of EUR 8,000 to take into account the fact that more incomes are aggregated into a comprehensive total income whereas under the existing system only wage income is included for determining NTMI. • The results of the microsimulation exercise are used to test whether: o (i) progressivity was improved. o (ii) the barrier to entry at the lower end of the labor market was removed. o (iii) the PIT-to-GDP ratio improved from the current 7.2 percent to about 8.0 percent. • The best fit scenario was then used to conduct a distribution analysis of the effect of the change on different income levels. Revenue effect 51. The results of the simulations are shown in the tables 5-11 below. The simulations were based on the rate structures in Latvia, Belgium, Portugal, Greece and Spain, including its effect on projected PIT 27 Data on rate structure obtained from (i) OECD Tax Database, Part 1. Taxation of Wage Income, May 2022 and (ii) Taxes in the OECD, July 2022. 28 In the simulation scenarios for PIT, we did not consider dependent deductions for children. What we have are actual expenses for some sources of income, or actual standardized 30%, whichever is greater. We also included a non-taxable minimum income that applies to everyone which is effectively part of income taxed at 0%. 29 The PIT rate structure in the tables below is rounded off to the nearest whole number. 24 revenues. As mentioned earlier in paragraph 49, the choice of the countries and their rate structures are purely illustrative and not intended as recommendations to move to identical tax structure and thresholds. Table 5. Lithuania Base Scenario Lithuania Base Scenario Lithuania’s rates and Lithuania PIT revenue PIT-to- GDP ratio brackets with EUR 5,000 NTMI30 <=60 AW 3,229 billion (i) 7.2% 20% (EUR 75,00) >60 AW 32% (EUR 75,00) Table 6. Latvian Scenario Latvian Scenario Latvia’s rates and brackets Lithuania projected Lithuania projected Projected PIT-to- PIT revenue with (i) PIT revenue with (ii) GDP ratio under EUR 5,000 NTMI EUR 8,000 NTMI the two scenarios 20% <=20,000 3,595 billion 2,953 billion (ii) 8.0% 23% 20,000 - 62,800 (iii) 6.6% 31% >62,800 52. Under this scenario, the tests for meeting criteria of government policy objective are: (i) The PIT-to-GDP ratio improved from the current 7.2 percent under EUR 5,000 NTMI but not under EUR 8,000 NTMI. (ii) Progressivity improved only with EUR 8,000 NTMI at the cost of reduced revenue. (iii) Barrier to entry remains for income below EUR 20,000 and also because the lowest marginal rate is still 20 percent. 30In general, the NTMI is around EUR 5000. In 2020, it was almost 5,000 for those with the minimum salary (if there was no additional NMTI for e.g., persons with disabilities). Not all salary earners have NTMI if their salary earnings are above the threshold. 25 Table 7. Belgian Scenario Belgian Scenario Belgium’s rates and Lithuania projected Lithuania projected Projected PIT-to- brackets PIT revenue with (i) PIT revenue with (ii) GDP ratio under EUR 5,000 NTMI EUR 8,000 NTMI the two scenarios 25% <=13540 5,484 billion 4,533 billion (i) 12.2% 40% 13,540 - 23,900 (ii) 10.1% 45% 23,900 - 41,360 50% >41360 53. Under this scenario, the tests for meeting criteria of government policy objective are: (i) the PIT-to-GDP ratio improved substantially from the current 7.2 percent under both EUR 5,000 and under EUR 8,000 NTMI. (ii) Progressivity improved. (iii) Barrier to entry increased since the lowest marginal rate is 25 percent. Table 8. Portuguese Scenario Portuguese Scenario Portugal’s rates and Lithuania projected Lithuania projected Projected PIT-to- brackets PIT revenue with (i) PIT revenue with (ii) GDP ratio under EUR 5,000 NTMI EUR 8,000 NTMI the two scenarios 15% <=7112 4,422 billion 3,658 billion (i) 9.9% 23% 7112 - 10732 (ii) 8.2% 29% 10732 - 20,322 35% 20,322 - 25,075 37% 25,075 - 36,967 45% 36,967 - 80-882 48% >80,882 54. Under this scenario, the tests for meeting criteria of government policy objective are: (i) the PIT-to-GDP ratio from the current 7.2 percent improved significantly under EUR 5,000 NTMI and adequately under the EUR 8,000 NTMI. (ii) Progressivity improved substantially because of several gradual rates. (iii) Barrier to entry reduced for low-income taxpayers. 26 Table 9. Greek scenario Greek Scenario Greece’s rates and Lithuania projected Lithuania projected Projected PIT-to- brackets PIT revenue with (i) PIT revenue with (ii) GDP ratio under EUR 5,000 NTMI EUR 8,000 NTMI the two scenarios 9% <= 10,000 3,381 billion 2,858 billion (iv) 7.5% 22% 10,000 - 20,000 (v) 6.4% 28% 20,000 - 30,000 36% 30,000 - 40,000 44% >40,000 55. Under this scenario, the tests for meeting criteria of government policy objective are: (i) the PIT-to-GDP ratio improved marginally from the current 7.2 percent under EUR 5,000 NTMI but reduced under EUR 8,000 NTMI. (ii) Progressivity improved but at the cost of reduced revenue. (iii) Barrier to entry reduced for low-income taxpayers. Table 10. Spanish scenario Spanish Scenario Spain’s rates and brackets Lithuania projected Lithuania projected Projected PIT-to- PIT revenue with (i) PIT revenue with (ii) GDP ratio under EUR 5,000 NTMI EUR 8,000 NTMI the two scenarios 19% <= 6,000 3,414 billion 2,784 billion (vi) 7.6% 21% 6,000 - 50,000 (vii) 6.2% 23% 50,000 - 200,000 26% >200,000 56. Under this scenario, the tests for meeting criteria of government policy objective are: (i) the PIT-to-GDP ratio improved marginally from the current 7.2 percent for NTMI under EUR 5,000 but reduced for NTMI under EUR 8,000. (ii) Progressivity improved but at the cost of reduced revenue. (iii) Barrier to entry reduced for low-income taxpayers. 57. The result of this analysis shows that the Portuguese PIT rate structure fulfills all the three criteria of the test. The Portuguese structure is shown just as an example, where, under the new illustrative system, we see a significant improvement in progressivity, with average effective tax rates (AETR) for lower income taxpayers has come down and that for high income taxpayers, it has gone up. Under the current Lithuanian system, which is regressive, AETR for lower income taxpayers is higher than that for high income taxpayers. With an NTMI of EUR 8,000, the proposed goal of a PIT-to-GDP ratio of 27 8.2 percent is achieved, which is slightly over the desired 8.0 percent. Progressivity improves without loss to PIT revenue. The barrier to entry is reduced because of a low entry rate of 15 percent for income up to EUR 7112, after taking into account the NTMI of either EUR 5,000 or EUR 8,000. Thus effectively, income up to EUR 15,112 (7112+8,000) is taxed at a marginal rate of 15 percent after allowing for the NTMI. 58. Since the Portuguese scenario achieves the Government’s objectives, the team tried to introduce comprehensive rate structure scenarios for all income except interest and dividends AND differentiated tax rates for interest and dividends which a majority of EU countries use. In the new simulation, the team has used two scenarios, one with a differentiated rate of 20 percent for interest and dividend, and the other with a differentiated rate of 25 percent. Table 11. Portuguese Scenario with differentiated rate for interest and dividends Portuguese Scenario with Differentiated Rate for Interest and Dividends Portugal’s rates and Lithuania projected PIT Lithuania projected PIT Projected PIT-to- brackets for all incomes revenue with EUR revenue with EUR GDP ratio under except interest and 5,000 NTM and 5,000 NTM and the two scenarios dividends (i) 20% rate for interest (i) 25% rate for interest and dividends and dividends 15% <=7112 3,446 billion 3,498 billion (iii) 7.7% 23% 7112 - 10732 (iv) 7.8% 29% 10732 - 20,322 35% 20,322 - 25,075 37% 25,075 - 36,967 45% 36,967 - 80-882 48% >80,882 Progressivity effect 59. For the analysis of the distribution effect of the simulated change, this exercise uses the Portuguese scenario since it meets all the criteria of improved progressivity, higher PIT-to-GDP ratio, and reduced barrier to entry. Evidently, any other scenario can be chosen to assess the progressivity effect but given that some of the criteria are not met in the other scenarios, it will give the same distributional effect as indicated in the tests above. 28 Table 12. Distributional analysis using the Portuguese scenario in the Lithuanian model Distributional Analysis of the Portuguese Scenario Income Number of Distribution PIT burden Average tax per Effective PIT rate levels taxpayers of taxpayers Million EUR person EUR EUR by income Current New Change Current New Current New level <= 25000 1,129,414 82.8% 1,483 806 -46% 1,313 713 12.5% 6.8% 25K – 50K 174,700 12.8% 924 1,012 10% 5,290 5,793 17.2% 18.8% 50K – 75K 31,513 2.3% 264 447 69% 8,393 14,189 16.3% 27.7% 75K – 100K 10,951 0.8% 112 235 110% 10,213 21,430 15.2% 32.0% 100K-150K 8,360 0.6% 118 248 110% 14,112 29,610 16.8% 35.2% 150K-200K 3,349 0.2% 58 137 135% 17,467 41,023 16.1% 38.0% 200K-250K 1,712 0.1% 36 90 149% 21,085 52,496 15.9% 39.7% >250,000 4,393 0.3% 233 711 205% 53,137 161,847 14.6% 44.9% 60. In the scenarios where comprehensive rate structure is used for all income except interest and dividends AND differentiated tax rates for interest and dividends, a different distributional pattern can be observed. Table 13 presents the pattern. Table 13. Average effective tax rates for comprehensive and differentiated Portuguese scenarios Distributional Analysis of the Portuguese Scenario under Comprehensive and Differentiated Rates Income Number Distribution Average Effective PIT rate under different Portuguese scenarios levels of of Current Portuguese New Portuguese New Portuguese EUR taxpayers taxpayers Lithuanian scenario with scenario BUT scenario BUT by income system all sources of with interest and with interest and level incomes under dividends taxed dividends taxed comprehensive separately at separately at total income 20% 25% <= 25000 1,129,414 82.8% 12.5% 6.8% 6.8% 6.8% 25K – 50K 174,700 12.8% 17.2% 18.8% 18.7% 18.7% 50K – 75K 31,513 2.3% 16.3% 27.7% 26.9% 27.1% 75K – 100K 10,951 0.8% 15.2% 32.0% 30.1% 30.6% 100K-150K 8,360 0.6% 16.8% 35.2% 32.0% 32.6% 150K-200K 3,349 0.2% 16.1% 38.0% 33.5% 34.4% 200K-250K 1,712 0.1% 15.9% 39.7% 33.4% 34.6% >250,000 4,393 0.3% 14.6% 44.9% 35.0% 36.8% 61. This analysis demonstrates (Tables 12 and 13) some interesting features of the Lithuanian PIT structure: • More than 95 percent of taxpayers earn less than EUR 50,000. • Conversely, only 5 percent of taxpayers show income above EUR 50,000. 29 • One major inequity issue noticed here is that taxpayers earning net income between EUR 25,000 and 50,000 currently have a higher effective PIT rate (17.2 percent) than those with net income of more than EUR 150,000 (16.1 percent) and this regressivity continues for income between EUR 200,000 and 250,000 (15.9 percent) and further for income above EUR 250,000 (14.6 percent). • This unfairness is largely due to the fact that high income individuals earn a much higher percentage of their total income from profits and gains of business and profession and from other capital incomes, which in Lithuania are taxed at a much lower rate. • The new scenario will remove the regressivity in the PIT rate structure and bring greater progressivity in the system for all individuals. 2. Microsimulation of Legal Entities Income Taxation 62. The WB team also performed a microsimulation of legal entities income taxation. The objective of the microsimulation analysis for legal entities was to test several options that would address the objectives of the Government to encourage entrepreneurship for microenterprises and promote their growth, as well as assess the appropriateness of tax incentives that create tax inequality and promote the shadow economy.31 63. In order to ensure that there are incentives for microenterprises to grow into big firms and for self-employed to formalize and incorporate, the dual rate system needs to be phased out. The reason for this is that it encourages firms to stay small and creates bunching at the threshold of EUR 300,000. 64. In the microsimulation for legal entities, two types of scenarios have been designed. The first is one in which there is integration to a single rate of 15 percent for all enterprises. In the second one a 15 percent rate has been made marginal, which means that until the threshold firms pay 5 percent and above the threshold, it pays 15 percent on the tax base above the threshold. 65. In Lithuania, the income tax rates are dependent on a turnover threshold of EUR 300,000. In determining the tax threshold, the good practice is to have the rate or rates linked to the tax base, which is income or profit, not turnover. Having turnover as threshold for designating SMEs and providing facilities like simpler recordkeeping and fewer filing submission is a normal practice. However, for determining the tax due, the base is multiplied with the rate. 66. Another objective of the microsimulation is to try scenarios where the CIT-to-GDP ratio can be raised to reduce the gap between the Lithuanian CIT-to-GDP ratio (1.6 percent) and the EU average (2.3 percent). Table 14 below presents a general explanation of the scenarios. 31 For CIT simulations, we have kept triple the amount of R&D expenditure, as in the current system. 30 Table 14. General explanation of the scenarios and the objectives for legal entities Current base scenario Different new scenarios Objective • 5% reduced rate for • Common standard 15% rate for Reduce barriers to entry turnover below EUR all legal entities at the microenterprises 300,000 and up to 10 to grow into larger • Optional scenario with 15% employees and a enterprises seamlessly. standard rate as a marginal rate, standard 15% rate for i.e., 15% rate applies only for As gradually step into all others incomes above the threshold the standard rate to reduce shock for micro enterprises • Build scenarios to raise Achieve close to 2.0 additional revenue from CIT percent CIT-to-GDP rate • Threshold based on • Threshold based on income for Tax base for CIT should turnover for application application of reduced rate have income threshold of reduced rate 67. Figure 4 shows that unifying the tax rate for all legal entities to 15 percent will affect mainly firms below the EUR 300,000 threshold and has little or no effect on firms above the threshold. For firms with turnover up to EUR 150,000 the overall tax burden will increase from EUR 17.6 million to EUR 43.9 million. For firms with turnover between EUR 150,000 to 300,000, the overall tax burden will increase from EUR 20.1 million to EUR 44.2 million. Figure 4. Effect of unifying rate to 15 percent for all legal entities 31 68. Figure 5 focuses on the effect of unifying the rate to 15 percent for small partnerships which are the main beneficiaries of the reduce rate. Here also, the scenario will affect mainly firms below the EUR 300,000 threshold and has little or no effect on firms above the threshold. Figure 5. Effect of unifying rate to 15 percent for all small partnerships 69. For small partnerships with turnover up to EUR 150,000 the overall tax burden, resulting from unifying the rate to 15 percent for all legal entities, will increase from EUR 3.3 million to EUR 12.1 million. By unifying the rate to 15 percent, the CIT -to-GDP ratio will increase marginally from 1.60 percent to 1.63 percent. For small partnerships with turnover between EUR 150,000 to EUR 300,000, the overall tax burden will increase from EUR 1.8 million to EUR 5.6 million. The overall average annual tax paid by small partnerships will increase from EUR 266 to EUR 912. 70. Table 15 shows: (i) a scenario where a unified rate of 15 percent is applied for all legal entities; and (ii) the results of several scenarios where the 15 percent standard rate is a marginal rate, i.e., it applies for net income above the threshold. Also, except in the base scenario, where the threshold is turnover based, in all other scenarios it is based on a net income threshold of EUR 38,500 to move from the 5 percent to the 15 percent rate. The net income threshold of EUR 38,500 was taken as the average net income of all returns where the turnover was between EUR 200,000 and EUR 300,000. The revenue and the CIT-to-GDP ratios for each scenario are produced in order to determine which scenario will achieve the target of getting closer to the EU average of 2.3 percent for the CIT-to-GDP ratio. 32 Table 15. Results of scenarios including (i) unified rate of 15 percent and (i) dual CIT rates on marginal income Scenario Reduced Standard Threshold Revenue CIT-to-GDP rate rate euro euro percent Baseline – current 5% 15% 300,000 888,635,000 1.60 turnover Scenario A – single rate 15% - 944,998,720 1.70 Scenario 1 -marginal 5% 15% 38,500 850,363,392 1.53 income Scenario 2 - marginal 7% 15% 38,500 869,290,432 1.57 income Scenario 3 - marginal 10% 15% 38,500 897,681, 024 1.62 income Scenario 4 – marginal 5% 20% 38,500 1,118,045,184 2.01 income Scenario 5 - marginal 10% 20% 38,500 1,165,362,944 2.10 income Different new scenarios The results show that Scenario A (unified rate) will result in raising CIT-to-GDP ratio from 1.60 currently to 1.70. Scenarios 1 and 2 would be revenue negative. . Scenario A with unified rate is the most appropriate since it removes the bunching incentive all together. Scenario 3 will be revenue neutral. Scenarios 4 and 5 are revenue positive and achieve a CIT-to-GDP ratio of over 2 percent. However, Scenario 4 creates a large difference between the reduced rate and the standard rate. Of the scenarios with marginal rates, Scenario 5 seems to be the most appropriate because it the reduced rate is not too low, just half the standard rate and it achieves an appreciable improvement in CIT-to-GDP ratio. In all cases, the standard rate is on the marginal income and not the total income. The preferred option is still to have a unified rate. 33 Chapter V. Summary of Findings and Recommendations 71. The purpose of this microsimulation exercise is to provide the Lithuanian authorities with an analytical tool to carry out a systematic assessment of the impacts of tax optimization behaviour for SMEs and self-employed derived from tax discrepancy and different regulations, and to build options for a more informed policy choice. It allows the authorities to consider options for addressing some of the key challenges in the Lithuanian income tax system that affect productivity and growth of the microenterprises and create barriers to entering the formal labor market by low-income workers. Key findings 72. As discussed earlier, Lithuania has among the highest economic disparity in the EU. This is partly perpetuated by the absence of progressivity and the schedular system of taxation where labor income (wages) is taxed at a relatively higher rate than capital income (profits and gains of business and professions, dividends, interest, rent, and royalties). Moreover, there are only two tax brackets for employment income with a high threshold, which means that even taxpayers with a relatively high salary, still pay tax at a relatively lower rate. Under these circumstances, low-income individuals currently shoulder much of the tax burden. 73. The results of the microsimulation exercise reconfirm the high level of inequality resulting from the current system. Under the current base scenario, taxpayers earning more than EUR 250,000 have a much lower effective PIT rate (14.6 percent) than taxpayers earning between EUR 25,000 and 50,000 (17.2 percent). Consequently, the redistributive role of taxes and transfers has been limited. Also, the entry level tax rate for low-income taxpayers is high relatively to the practice in many EU countries. This creates a disincentive for low-income workers to enter the formal work force and pushes many into the shadow economy, further lowering productivity. 74. Another issue that was examined during the microsimulation exercise was the fact that Lithuania’s tax-to-GDP ratio (31.3 percent) is significantly lower than the EU averages (37.6 percent). This has translated into both PIT-to-GDP and CIT-to-GDP ratios being lower than the EU average. Greater investment in innovation and social infrastructure would require the Lithuanian budget to have a larger tax-to-GDP ratio, closer to the EU average. This has to be considered in relation to Lithuania’s commitment to the EU Directive on GMT and to the RRF. 75. As regards the income taxation of legal entities, the dual rate system creates a disincentive for businesses to grow and face a much higher rate on the entire income. There is a reduced rate of 5 percent for enterprises with turnover up to EUR 300,000 and a standard rate of 15 percent for those above this threshold. The analysis demonstrates that Lithuanian SMEs are “bunchingâ€? at a preferential taxation threshold of EUR 300,000. “Bunchingâ€? is evidence of tax optimisation, which is detrimental to efficiency, equity, productivity and growth. 76. These challenges have been highlighted in the Government’s fiscal priorities which are commendable. This microsimulation analysis provides the Government with a menu of options and their impact on revenue and distribution. Microsimulation is a powerful tool to analyse the policy options. It can be used in the future also to test the impact of various new options that the government may consider. Success will depend on weighing the various options objectively to arrive at the best outcome. 34 Main recommendations 77. The following recommendations are made with a main focus to help improve the productivity and growth of the private sector while preserving or improving the tax collection levels. This will be achieved by: (i) reducing barriers to formalization for low-end workers; (ii) removing the regressive nature of PIT taxation and introducing progressivity therein; (iii) improving transparency in the rate structure; and (iv) removing the incentive for bunching by enterprises. 78. The recommendations below relate to the overall rate structure, the PIT system and the taxation of legal entities. It also provides the sequencing for the proposed reforms. It is advised that the Lithuanian authorities use the microsimulation models to their full extent to determine the appropriate levels of taxation that would best achieve the Government’s stated goals and the commitments to the European Commission. A. Recommendations for the overall tax structure 79. Ensuring a higher level of domestic revenue mobilization that is consistent with growth objectives would require a gradual improvement of the tax-to-GDP ratio, in small incremental steps, from the current 31.3 percent to bring it closer to the EU average. The increase in tax-to-GDP would be most effective by the combined increase of PIT-to-GDP and CIT-to-GDP ratios. This would also align to the objective of removing hurdles to growth for microenterprises and individuals as described below. Therefore, the optimal scenario is improving the PIT-to-GDP ratio from the current 7.2 percent to move it closer to the EU average rate, coupled with an improvement in the CIT-to-GDP ratio from the current 1.6 percent to bring it closer to the EU average of 2.3 percent. This has to be considered in relation to Lithuania’s commitment to the EU Directive on GMT. B. Recommendations for the PIT system 80. It is recommended to remove the extremely schedular PIT system and introduce a more comprehensive income approach where most sources of income are aggregated. Additionally, it is recommended to introduce a significantly more progressive rate structure. This means (i) reducing the rates and brackets at the lower end; (ii) increasing the rates at the upper end; and (iii) introducing three to four income brackets instead of the current two, to improve progressivity. 81. The microsimulation model can be used to examine the merits of different options for income brackets. Additionally, monetary euro values for the income brackets can be used instead of the currently used multiples of AW. This will provide simplicity and greater transparency in tax computation at the front end. At the back end, AW could still be used to refine the bracket structure from time to time for inflation adjustment. 82. Microsimulation done on the rate structure of several eurozone countries applied to the Lithuanian tax database shows that the Portuguese PIT system best achieves all the desired objectives. In the scenario, using Portuguese structure is not a recommendation, just an illustration. Under the new illustrative scenario, average effective tax rates for lower income taxpayers has come down and that for high income taxpayers, it has gone up. It is recommended to simulate possible scenarios approximating 35 the Portuguese scenario which gives the best results for improving progressivity, removing the barrier to entry at the lower income level, improving productivity and growth, and reducing informality. C. Recommendations for the tax system for legal entities 83. The microsimulation model can be used to examine various options for the CIT rate structure. It is recommended to unify the rate structure for all legal entities to 15 percent. The microsimulation analysis shows that for microenterprises with turnover of EUR 300,000 it will increase their annual income tax payment from the current EUR 266 to EUR 912 or monthly payment from EUR 22 to EUR 76. In euro terms the increase is insignificant, but the greatest advantage this will bring is to remove the bunching phenomenon and the barrier it creates to growth of small enterprises to larger firms. 84. If unifying the rate structure is not immediately feasible, a second-best option is to take a gradual approach. The approach will require starting with the standard rate of 15 percent as a marginal rate for income about a desired income threshold. This will reduce, but not remove, the incentive to bunch at the threshold. 85. It is recommended to forgo the practice of taxing income according to different rates based on a turnover threshold. Even though this practice can be beneficial for providing non-tax facilities and support to microenterprises, it is not a good practice for the purpose of computing taxable income. If at all a dual rate structure is maintained, it is recommended to use an equivalent income threshold. After a few years of using the 15 percent rate as marginal rate, a full unification of the CIT rate can be implemented. 36