SOCIAL PROTECTION & JOBS DISCUSSION PAPER No. 1928 | April 2019 Pensions in a Globalizing World: How Do (N)DC and (N)DB Schemes Fare and Compare on Portability and Taxation? Bernd Genser and Robert Holzmann © 2019 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: +1 (202) 473 1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. RIGHTS AND PERMISSIONS The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: +1 (202) 522 2625; e-mail: pubrights@worldbank.org. Abstract: Pensions and broader forms of retirement income do not stop at national borders. As part of globalization, individuals increasingly spend part of their working or retirement life abroad but want to keep or move their acquired rights, accumulated retirement assets, or benefits in payment freely across borders. This raises the issue of the portability and taxation of cross-border pensions in accumulation and disbursement. This paper addresses both portability and taxation issues from the angle of which type of pension scheme – defined benefits (DB) or defined contributions (DC) – is more aligned with globalization in establishing individual fairness, fiscal fairness, and bureaucratic efficiency. The paper shows that DC schemes tend to dominate DB schemes both at the level of portability and taxation. Key words: Benefit Portability, Cross-Border Pensions, Front- and Backloaded Taxation, Comprehensive Income Taxation, Expenditure Taxation JEL codes: F22, H21, H55 2 Acknowledgments This paper is written for Progress and Challenges of Nonfinancial Defined Contribution Pension Schemes, Volume 2: Addressing Gender, Administration, and Communication, edited by Robert Holzmann, Edward Palmer, Robert Palacios, and Stefano Sacchi, to be published by the World Bank in autumn 2019. We are grateful to Csaba Feher and Will Price for very helpful comments and suggestions and to Amy Gautam for first-rate copy editing. A first version of the paper was presented at the NDC III conference in Rome, October 5 –6, 2017, and we are thankful to the participants for their comments and encouragement. The views expressed herein are those of the authors and do not necessarily reflect the views of the institutions they are associated with or the views of the World Bank. 3 Abbreviations and Acronyms BA Bilateral Agreement DB Defined Benefit DC Defined Contribution DTA Double Taxation Agreement EU European Union MA Multilateral Arrangement NDB Nonfinancial Defined Benefit NDC Nonfinancial Defined Contribution OECD Organisation for Economic Co-operation and Development 4 Table of Contents 1. Introduction ...................................................................................................................... 7 2. The rise of international labor and benefit mobility ........................................................ 8 3. Portability issues: Objectives, instruments, and DB–DC comparison ............................ 10 3.1. Objectives of portability............................................................................................... 11 3.2. Instruments of portability ............................................................................................ 13 3.2.1. Portability arrangements between countries....................................................... 13 3.2.2. Multinational private sector providers ................................................................. 16 3.2.3. Changes in benefit design ..................................................................................... 16 3.3. NDB and NDC schemes compared ............................................................................... 18 4. The taxation of cross-border pensions: Facts, issues, and suggested solutions ............ 22 4.1. The state of taxation of cross-border pensions ........................................................... 23 4.2. The incompatibility of deferred income taxation and OECD model tax convention .. 28 4.3. A new framework for pretaxed pension income ......................................................... 30 4.3.1. The starting position ............................................................................................. 30 4.3.2. The proposal ......................................................................................................... 31 4.3.3. Three tax payment options ................................................................................... 33 5. Frontloaded taxation, payment options, and DB and DC in comparison ....................... 36 5.1. The frontloaded tax assessment and immediate payment option ............................. 36 5.2. The frontloaded tax assessment and deferred payment option................................. 37 5.3. The frontloaded tax assessment and distributed payment option ............................. 39 6. Conclusions ..................................................................................................................... 41 5 References .............................................................................................................................. 44 Appendix 1. Two-letter country abbreviations subject to ISO code 3166 ........................ 47 Appendix 2. OECD Model Tax Convention on Income and on Capital .............................. 48 6 1. Introduction Pensions and broader forms of retirement income do not stop at national borders. As part of globalization and the increasing mobility of labor and capital, an increasing number of individuals spend at least part of their working life abroad and acquire benefit rights that they want to take home or on to a new country of work or residence. Some individuals want to spend part or all of their retirement life in places with a better climate, a lower cost of living, and/or more benign taxation of their retirement income. However, the increasing mobility of individuals before and after retirement creates issues of the portability and taxation of cross-border pensions in accumulation and disbursement. Both topics – portability and taxation – have found limited attention in pension economics so far. Simply put, full portability of pensions allows labor migrants to accumulate, keep, and/or transfer pension rights and to receive benefits in disbursement anywhere in the world. Without that ability, potential migrants may decide not to migrate, or to migrate although they risk losing their acquired rights. In the first case, international labor mobility is impeded; in the second, risk management is constrained and reduces the welfare of the migrant over his lifecycle. Such obstacles may also arise even if pension benefits are portable but other benefits are not, particularly health care benefits during retirement. The income taxation of cross-border pensions may increase or reduce individuals’ migration incentives, as the tax burden of the retired migrant abroad may rise or fall depending on the total tax burden in working and residence countries. For the relevant tax burden of the migrant’s pension, the tax treatment across his whole lifecycle matters as taxes may be levied at the time of contribution/premium payment, return receipt, and disbursement. Differences in the portability of social benefits and in the taxation of cross-border pensions raise issues of individual fairness (i.e., do I get out what I paid in, and is my tax treatment equivalent to that of a nonmobile individual?). Portability also raises issues of fiscal fairness at the country level (i.e., does the portability arrangement favor one country due to tax arrangements under double taxation treaties?). A final issue concerns the bureaucratic 7 efficiency by which individual and fiscal fairness can be achieved (i.e., how burdensome and time-consuming is tax compliance for all involved?). This paper addresses both portability and taxation issues from the angle of which type of pension scheme is more aligned with globalization by better establishing individual fairness, fiscal fairness, and bureaucratic efficiency. The focus is mostly on the benefit type – defined benefits (DB) versus defined contributions (DC) – with funding and administrative issues given secondary importance. The relevant literature on both topics is briefly summarized or referenced. Section 2 briefly establishes the facts of rising labor/benefit mobility across the world. Section 3 presents portability issues (absent taxation), how portability can be achieved, and the role of benefit types. Section 4 presents cross-border issues in case of income taxation of benefits, the current international disarray, and how it can be addressed. Section 5 extends the analysis and asks whether the type of scheme matters for the possible solutions. Section 6 summarizes and concludes on the ease of pension scheme alignment in a globalized world. 2. The rise of international labor and benefit mobility The share of individuals living outside their home country is increasing again after a temporary low in the 1970s, reaching 3.4 percent of the world population in 2017 (up from 2.3 percent in 1980), or an estimated 258 million people (United Nations 2017). Figure 2.1 presents the dynamics of the number of migrants and their share in the world population since 1960. On January 1, 2016, the number of people living in the EU-28 who were citizens of nonmember countries was 20.7 million, representing 4.1 percent of the EU-28 population, while the number of people living in the EU-28 who were born outside of the European Union (EU) was 35.1 million. In addition, 16.0 million persons were living in one of the EU member states on January 1, 2016, with the citizenship of another EU member state (Eurostat 2017). 8 Figure 2.1: Number and share of migrants in world population, 1960–2017 300 3.6% 3.4% P 250 Total migrants (in million) o 3.2% p 200 s 3.0% u h l 150 2.8% a a 2.6% t r 100 e 2.4% i 50 o 2.2% n 0 2.0% Total migrants Population share Source: United Nations 2017; Migration Policy Institute, Data Hub; authors’ comp ilation. These migrant stock numbers—impressive as they are—underestimate the underlying labor mobility dynamics, because the numbers in Figure 2.1 only capture individuals who have lived outside their traditional country of residence in the observation year. As individuals may take multiple migration spells of varying length, sometimes in different countries, the relevant number of individuals with past migration spells is significantly higher. Evidence from across the world is strong that the number of spells spent abroad is increasing. The EU figures for individuals who spend at least some of their adult life living outside their home country (as a student, intern, intra- or interfirm mobile employee, labor migrant, or “snowbird” retiree) are definitely rising and may soon be as high as one out of every five individuals (Holzmann 2015). Past labor market spells abroad translate into rising numbers of pension payments to and from abroad. For example, these amounted to about 11.1 percent of the total number of pensions paid in Germany in 2013, up from 9.8 percent in 2005. Table 2.1 details the composition and trends in former labor and more recent retirement mobility to and from Germany. 9 Table 2.1: Recipients of statutory German pensions – in Germany and abroad Number of pensioners in millions 2013 2010 2005 (in % of total pensioners) Pensioners with non-German citizenship 2.562 (100%) 2.367 (100%) 2.032 (100%) - living in Germany 1.059 (41.3%) 0.944 (39.9%) 0.774 (38.1%) - living outside Germany 1.503 (58.7%) 1.423 (60.1%) 1.258 (61.9%) Pensioners with German citizenship 22.602 22.646 22.452 (100%) (100%) (100%) - living outside Germany 0.222 (0.98%) 0.206 (0.91%) 0.170 (0.76%) Total number of pensioners 25.164 25.013 22.484 (100%) (100%) (100%) - living outside Germany 1.725 (6.85%) 1.629 (6.51%) 1.427 (5.83%) - non-German citizens living in Germany 1.059 (4.21%) 0.944 (3.77%) 0.774 (3.44) - potential recipients of cross-border 2.784 (11.1%) 2.573 (10.3%) 2.201 (9.8%) pensions Source: Genser and Holzmann 2018, based on Eurostat Online Database (June 2015). Warnes (2009) presents data for Germany, the United Kingdom, and the United States on the popularity and dynamics of their respective retirement destinations for the period mid- 1990s to 2005. His data show a dynamic similar to that presented in Table 2.1. 3. Portability issues: Objectives, instruments, and DB–DC comparison The topic of cross-border portability of pensions (and other social benefits) is a relatively new area in pension economics. While the portability of pension benefits within countries and between occupational plans has been explored for quite some time (e.g., Foster 1994), portability between countries has received little attention by economists. This field was generally left to social policy and social law experts. 10 This paper focuses on the economic issues of portability, which might be captured by the following working definition:1 “Cross-border portability of pension benefits is the ability of labor migrants to preserve, maintain, and transfer both acquired pension rights and rights in the process of being acquired from one private, occupational, or statutory pension scheme, to another independent of nationality and aligned with the country of residence. Pension rights refer, in principle, to all rights stemming from contributory payments or residence criteria in a country. Not portable typically are benefit components that are not based on contributions such as benefit top-ups for low- income individuals or minimum income guarantees.”2 Section 3 presents the economic foundation of portability based on three elements: a brief discussion of the economic objectives of international portability of pensions and more broadly of social security benefits (section 3.1); a brief presentation of the key instruments used to establish pension benefit portability (section 3.2); and an assessment of the implications for the DB–DC selection (section 3.3). 3.1. Objectives of portability Establishing portability of social benefits should be straightforward, as three key considerations—economic, social, and human rights—favor it (Holzmann and Koettl 2015). From a first-best economic point of view, individuals’ labor mobility decisions should not be hampered by the lack of portability of social benefits for which they have acquired rights. Global efficiency and global growth is increased if distortionary obstacles toward portability are absent. To ensure that international labor mobility profits the home as well as the host country, select and appropriate bilateral interventions may be necessary. 1 For early economic research on the topic see Holzmann, Koettle, and Chernetsky (2005). Later work includes Holzmann and Koettl (2015), Jousten (2015), and Holzmann (2016). 2 This definition draws on the general definition of the portability of social security benefits developed by Cruz (2004) and Holzmann, Koettle, and Chernetsky (2005). 11 The lack of benefit portability can influence labor migrants’ international mobility decisions. Workers may decide not to take a job abroad if they have to pay social security contributions in the host country but cannot profit from its benefit coverage or cannot take their acquired rights home. Nonportability is particularly relevant for the long-term benefits of pensions and health care. For pensions, this may exist due to long vesting periods of 10, 15, or more years or to restrictions on cross-border benefit payments. Access to health care services in retirement is typically linked to the eligibility of pension benefits and residence in the host country, unless cross-country legal arrangements exist. From a social policy point of view, such acquired rights are a critical element of individuals’ (or families’) lifecycle planning and social risk management. Denying portability – particularly once the mobility decision has been made and cannot be reversed – increases the risk of lifecycle planning for individuals and their families and creates substantial welfare losses. For emigrants a lack of portability of acquired rights means that they can establish pension rights only in their host country. While a higher (comparable) wage rate in the host country may provide some compensation, labor emigrants will face a lower replacement rate after retirement. This typically happens for mid-career labor migrants. A migrant who plans to return home but cannot transfer pension rights acquired abroad or receive cross-border benefits will need to increase private saving or to continue working. These adjustments in lifecycle planning are beneficial, but they do not avoid welfare losses compared to the portability case. From a human rights point of view, migrants have the right to enjoy social protection according to national legislation and international conventions. These rights should carry over when individuals leave the country or change profession. A key question is whether these human rights apply only to acquired (contributory or residential) pension rights or to all social rights. As they are resource-consuming, economic and human rights tradeoffs will emerge. 12 3.2. Instruments of portability Essentially three approaches are available to establish cross-border portability of pension benefits between countries: • Binding portability arrangements between countries • Using multinational private pension providers • Changing the pension benefit design to make benefits portable without further government action 3.2.1. Portability arrangements between countries Most portability analyses and discussions focus on bilateral agreements, but the scope is much larger and includes unilateral and multilateral arrangements. Unilateral actions (UA) can be taken by the country where migrants earn labor income and are able to acquire pension rights. Examples of UAs include: • Denying migrants access to the national social security scheme3 can be substituted by giving them the option to contribute to pension systems in their home country, as is feasible in Mexico, the Philippines, and Sri Lanka. • Denying migrants access to the national security scheme can be substituted by voluntary access to either the host or the home country pension system. Enrollment in the home country pension system avoids host country constraints on cross-border benefit payments.4 • Granting migrants full access to the statutory national pension scheme as well as full exportability of eligible pension rights may establish full portability. Hence all pensioners with a contribution length beyond the vesting period keep their acquired 3 As occurs in the Gulf Cooperation Council countries for essentially all expatriates, and for some categories of foreign workers in Hong Kong, Malaysia, and Singapore. 4 The Philippines and Mexico fall somewhere between the first and second examples. The Philippines allows workers to contribute to national pension schemes but independent of access in the host country. Similarly, Mexican migrants can get access to health care benefits for a flat-rate premium (for their families left behind or themselves when they return) independent of their insurance in the host country (i.e., the United States). 13 pension rights and receive pension benefits after the minimum retirement age is reached and other eligibility conditions are fulfilled. Ineligibility typically emerges because of a contribution record below the vesting period. Bilateral agreements (BAs) are the centerpiece of current portability arrangements between countries. While they can, in principle, cover the whole range of exportable social benefits, they typically focus on long-term benefits such as old-age, survivors’, and disability pensions, and to a much lesser extent on health care benefits, if at all. 5,6 With regard to pensions, BAs can: • Focus on temporary migrants only (e.g., waiving the contribution requirement to the pension scheme in the host country while making contributions mandatory in the home country). • Establish mutual exportability of pension claims between the two countries. • Allow migrants to continue paying their social security contribution to their home country for an extended period of time. • Establish “totalization” (i.e., summing up) of the insurance periods across both countries, thus eliminating or at least reducing the binding effects of vesting periods in individual countries. • Cover all (legal or even illegal) migrants who have established acquired rights. • Establish full eligibility across the two agreement countries. • Establish benefits for migrants in the case of different benefit types between countries, such as the complex case between a residence-based basic benefit 5 For a historical and legal background on BAs, see Strban (2009). 6 No single study (inventory) captures the content of BAs across the world or even of subregions such as Europe; to the authors’ knowledge, no single evaluation has been undertaken to assess the effectiveness of BAs and MAs. 14 country (such as Australia) and an earnings-related/contribution-based benefit country (such as Germany). Multilateral arrangements (MAs) represent a general framework of portability for a group of countries for all or a subset of social benefits. These general rules are in most cases supported by more detailed BAs. Traditional MAs have been established in Latin America (MERCOSUR) and the Caribbean (CARICOM) and in 15 French-speaking countries in Africa (CIPRES); one was recently established between Latin America and Spain and Portugal (Ibero-American Social Security Convention); and one is under development for the Association of Southeast Asian Nations (ASEAN) countries. The most developed MA is the one among EU member states (plus Norway, Liechtenstein, and Switzerland). Strictly speaking the EU arrangement is not an MA but an EU Directive that obliges EU member countries to adjust their existing regulations accordingly (i.e., to revise their existing BAs). The main objective of the Directive is to essentially make all social benefit claims portable among EU member states, including unemployment and family benefits, in order to avoid discrimination and to establish full labor mobility, one of four core freedoms of the EU Treaties.7 For portability of statutory pension benefits, exportability works well for private sector schemes in principle, but is not frictionless; hence, benefit losses are possible for those moving between countries’ public sector schemes.8 Issues emerge with the portability of occupational and personal pension schemes – as within countries – when individuals leave a DB scheme that is, for example, linked to their final salary. This also happens with DC schemes, which are essentially individual savings plans. Here the tax privileges granted at 7 The four freedoms were set out in the Treaty of Rome (1958), extended by the Single European Act (1987), and strengthened in the Lisbon Treaty (2009). 8 Both authors experienced this when leaving their (former) civil servants scheme as Austrian academics to join a similar scheme in Germany; in Austria, their acquired rights in the civil servants schemes were transferred to private sector schemes with substantial reductions in pension wealth. For one author this happened again when he left German academia to move to the World Bank in the United States. 15 the level of contribution/premium payments and rates of return received render their simple export difficult and the EU has not yet found an effective way to establish comprehensive portability (see sections 4 and 5). Even when transfers can be made, they may inhibit the original intention of the pension policy, for example, when a pension plan offers a lump sum in cash to workers when they leave the country. If there is no requirement to invest the money into another pension plan, then the likelihood is higher that the money will be partly spent on short-term consumption rather than contributing to retirement saving. So portability should ideally be portability of assets from one pension vehicle to another. 3.2.2. Multinational private sector providers A promising approach, at least for supplementary benefits, is to use the services of privately organized multinational providers (MPs). MPs exist and function well for health care benefits. For example, Cigna, a Belgium-based service provider, services World Bank staff and retirees residing in Europe, as well as staff of the European University Institute. MP arrangements have been discussed, and sometimes implemented, for supplementary pensions of international workers in multinational enterprises so that these insured persons are tied to a single pension vehicle even if they work in various countries. Multinational providers may prove superior to national providers with respect to interjurisdictional risk sharing, because of risk pooling, transmission of best practices and innovations across countries, and better information on the state of the world. 3.2.3. Changes in benefit design The key idea behind changing the benefit design is to transparently disentangle the components that are lumped together in the pseudo-actuarial benefit design of social security schemes. For all social benefits, these components are the period insurance element, the presaving element, and the redistributive element (Holzmann and Koettl 2015, 378–80). 16 The period insurance component is only valid for one period, in which it is consumed; hence, it does not require portability. This element is relevant in health insurance, does not exist in old-age pension schemes, but does exist in the form of survivors’ or disability claims if all are lumped together under one contribution rate. The presaving (or asset accumulation) component exists in all social benefit systems in one form or another. It is huge in health care and old-age benefit schemes, amounting to a high multiple of annual contributions. In a health care scheme without age-related contribution rates, this component serves to accumulate reserves for health care costs that rise with age and for catastrophic health care. In (old-age) pension schemes, the financial or nonfinancial presaving is the constituent component. Conceptually, a pure accumulation phase is followed by a decumulation phase in which annuities or phased withdrawals are paid out. The redistributive component can be thought of as the deviation between accumulated individual contributions (including returns) and individual pension wealth (i.e., the present value of expected future pension benefit payments) at the end of each period. The redistributive component is the consequence of a nonactuarial benefit design due to explicit or implicit redistributive considerations within the pension scheme. For individuals the redistributive component may be positive or negative. A dominantly positive redistributive component typically emerges when a pension scheme is not only fed by contributions but also receives transfers from the general budget. If the three components can be separated conceptually and technically, then benefit portability between countries is substantially facilitated: • In the most drastic separation, there is no period insurance component, as disability and survivors’ pensions are separately organized; there is no redistributive component, as all redistribution is done outside the pension scheme; and the remaining presaving component is purely actuarial and can be transferred across borders upon migration. 17 • In a less complete separation, there is again no period insurance component, and no redistributive component due to interpersonal transfers, but the presaving component is not actuarially fair due to government transfers. Although this component is ready for portability, the question that emerges is: To what extent should the transferred amount be corrected to account fairly for the presaving increment, which is financed through the budget of the source country? 3.3. NDB and NDC schemes compared Against the background of the objectives, instruments, and evaluation criteria presented above, how do DB and DC schemes compare on portability? To simplify and shorten the comparison, the focus is only on nonfinancial DB and DC schemes (NDBs and NDCs), but most results are believed to also hold for financial DBs and DCs (FDBs and FDCs).9 The following properties of NDC and NDB schemes are relevant for cross-border portability: • Ideally, an NDC scheme has no period insurance component, as disability insurance is separately financed and organized (but coordinated with the NDC scheme); long- term survivors’ benefits are financed by own accounts and shared accumulations of spouses; and short-term transitional benefits during child-rearing periods are financed by other structures and resources (Holzmann 2017; Kruse and Ståhlberg 2017). • The “textbook” NDC scheme has no redistributive component within the insurance pool, and no redistributive component of budget subsidies to support financial sustainability.10 The existing redistributive components are explicit and financed 9 A main difference may emerge between funded and unfunded provisions with regard to the actual portability of financial assets when changing residence versus the mere recognition of rights while the assets remain in the source country. The latter is always the case in unfunded provisions as the pay-as-you-go (PAYG) asset remains in the source country. In funded provisions the assets can remain in the source country (as is typically the case under FDBs) but may also be transferred to the new residence country under FDC schemes, but there is no obligation and possibly no incentives to do so. 10 Abstracting from heterogeneity in longevity, which can be corrected for (Holzmann et al. 2019). 18 through earmarked government transfers and reflect purposeful social policy objectives. These social policy objectives emerge if individuals cannot make own contributions due to disability, unemployment, maternity leave, family leave, etc., and are financed by the respective programs (typically by earmarked contributions). Beyond that one can also imagine selective matching or lump-sum contributions to individual accounts to incentivize formal labor market contributions and/or to render the NDC scheme explicitly redistributive (Holzmann, Robalino, and Winkler 2019). • Because of the above-mentioned characteristic features of an NDC scheme, the accumulated individual account values reflect own contributions, rates of return that are consistent with financial sustainability, and an external contribution earmarked for individual circumstances. Thus, these accounts are fully portable as NDC annuities or as accumulated pension wealth amounts prior to eligibility. • In a traditional NDB scheme, disability and survivors’ pensions are typically part of the old-age benefits scheme design. Survivors’ pensions lose importance under reformed NDB schemes as receipt of an own pension above a certain amount increasingly disqualifies one from receiving a widow/widower’s pension, and one’s children receive flat-rate amounts. Again, these reforms reduce the contemporaneous insurance component but do not eliminate it. No good answers arise regarding which acquired rights for these risks should be portable. • Traditional NDB schemes have a few explicit and many implicit redistributive components because of their design. Most countries also have a variable redistributive component to keep these schemes afloat. Making explicit redistributive components portable raises little objection from the perspective of family or other social policy considerations (often expressed by assimilated insurance periods and/or earlier retirement age); the problem is their costing. The implicit and often unknown redistributive components should, in principle, not become portable. But serious problems arise in establishing appropriate adjustment 19 mechanisms to account for characteristic NDB features like last salary assessment period, variable annual accrual rate, or nonactuarial decrements for earlier or later retirement. Some of these components could be mitigated, for example, by basing the pension benefit on lifetime income or introducing actuarial increments/decrements for early retirement. • As a result of the difficulties of eliminating the contemporaneous component and of reducing the redistributive component to meaningful, measurable components, the presaving component cannot be well defined; it also requires cumbersome actuarial calculations for which objective estimates are difficult if not impossible to establish. Consequently, the amount of pension benefit to be sent abroad may still require a BA to establish portability, for example if the vesting period cannot be reduced to a few months. Making the acquired rights portable before retirement will not work unless the sustainability transfers are eliminated. This is possible with the introduction of an automatic balancing mechanism, but is technically much more challenging in an NDB scheme compared to an NDC scheme. Summarizing the comparison of NDB and NDC schemes to establish cross-border portability, the following conclusions emerge: • Textbook NDC schemes promise full portability even in the absence of BAs and MAs. Full exportability of benefits in disbursement and preservation of the acquired rights are required. Full exportability can be established unilaterally; the full preservation until eligibility is a design component of an NDC scheme as the account values are annually indexed with the notional (sustainable) rate of return. • Whether acquired rights in an NDC scheme prior to eligibility should become portable and transferred in real cash is a question of convenience and reciprocity with another NDC country, as only the annual balance of notional inflows and outflows needs to be settled in cash. However, as the annuity at retirement is determined by country-specific cohort life expectancy, such portability before retirement may invite benefit arbitrage. It would not affect the source country but 20 would affect the receiving country if its cohort life expectancy was well below that of the sending country, while it offers access to groups with a higher life expectancy. • NDB schemes will always need BAs or MAs to achieve portability. But the closer NDBs are to NDCs, the simpler are cross-border portability arrangements. BAs or fully-fledged MAs will still exist for NDC corridor countries for purely administrative reasons as well as to establish portability for other benefits, such as health care. • While BAs exist between most industrialized countries, they are the exception, not the rule, between industrialized and emerging/developing economies. As a result, in 2013 only 23.3 percent of worldwide migrants lived in countries with BAs between home and host countries, and over 80 percent of these migrants were from high- income countries (Table 3.1). The global progress since 2000 – the other year for which comparable data estimates are available – has been moderate and amounts to 1.4 percentage points (Table 3.2). Further progress on BAs is likely to be slow too, as their establishment depends on demanding conditions in lower-income countries (Holzmann 2016). Table 3.1: Global migrant stock estimates by origin country income group and portability regime, 2013 Percentage per regional income group Origin country income Regime I Regime IIa Regime IIIb Regime IVc Total group (Portability) (Exportability) (No Access) (Informality) (millions) High-income non-OECD 50.7 40.2 4.3 4.8 5.1 High-income OECD 76.3 19.0 0.4 4.3 33.0 Upper-middle-income 23.3 54.4 0.5 21.8 33.6 Low-middle-income 20.2 58.5 8.7 12.6 104.8 Low-income 2.7 61.2 18.7 17.3 75.9 Total (%) 23.3 53.2 9.4 14.0 252.3 Source: Holzmann and Jacques 2018. Notes: a. Legal migrants with access to social security in the host country in the absence of a bilateral or multilateral arrangement; b. Legal migrants without access to social security in their host country; c. Undocumented immigrants. 21 Table 3.2: Global migrant stock estimates by origin country income group and portability regime; change between 2000 and 2013 Percentage points per regional income group Origin country income Regime I Regime IIa Regime IIIb Regime IVc Total group (Portability) (Exportability) (No Access) (Informality) (millions) High-income non-OECD 10.3 -14.1 0.8 3.0 -0.4 High-income OECD -8.4 5.9 -0.6 3.0 4.1 Upper-middle-income 9.6 -4.4 -0.2 -5.0 8.2 Low-middle-income 6.1 -4.2 4.2 -6.1 26.9 Low-income 1.2 -7.9 8.9 -2.2 21.8 Total (%) 1.4 -3.0 4.5 -2.9 60.6 Source: Holzmann and Jacques 2018; Holzmann, Koettl; and Chernetsky 2005. Notes: See Table 3.1. 4. The taxation of cross-border pensions: Facts, issues, and suggested solutions The topic of taxing cross-border pensions is terra incognita in economics. No single recognized or competing paradigms explain how internationally portable pensions should be taxed. Yet countries typically have many bilateral double taxation agreements (DTAs) that include rules on how the rights are assigned to tax income from pensions and other retirement saving instruments. But this agreed tax treatment of pensions in a DTA for one migration corridor is not necessarily the same for another corridor, even if the corridor partners are neighbors. Furthermore, the tax treatment typically differs substantially across pension pillars (statutory, occupational, and personal). The guidance that exists on pensions is established in the Organisation for Economic Co-operation and Development (OECD) model tax convention on income and capital. Its relevant articles 18 and 19 suggest different tax treatment of cross-border pensions for private and public sector pensions – namely residence- versus sourced-based (see Appendix 2). Furthermore, they are also highly incomplete as they deal only with the disbursement phase of pension taxation, leaving out the contribution payment/saving and return receipt phases. Hardly any other area in economics has such a conceptual void, which has led to operational complexity and inconsistency in the taxation of cross-border pensions (Genser and Holzmann 2016, 2018). 22 This section summarizes recent attempts to highlight issues and offers a new proposal on how pensions should be taxed to address the double fairness dilemma of current pension taxation (Holzmann 2015; Genser 2015; Genser and Holzmann 2016, 2018): individuals risk unfair treatment due to the differences between and within countries, with some individuals paying the income tax on pension benefits twice – once during accumulation in the source country and again during decumulation in the residence country; others may benefit from tax exemption of pension wealth accumulation and disbursement in two countries. Of course, the latter case gives rise to tax arbitrage by strategic migration. Countries risk substantial fiscal unfairness as the current rules propose the taxation of cross- border pension benefits in the residence country, while income tax losses emerge in the source country, if income spent on contributions and income from pension wealth returns are tax-exempt. In view of the rising share of international mobility and benefit eligibility abroad (recall section 2), such a situation is unfair and unsustainable. To substantiate this proposal, this section highlights three areas: the state of taxation of cross-border pensions (section 4.1); the incompatibility of deferred income taxation and the OECD model tax convention (section 4.2); and a new framework for pretaxed pension/retirement income (section 4.3). 4.1. The state of taxation of cross-border pensions Income taxation in most OECD countries is codified according to the Schanz/Haig/Simons principle of comprehensive income taxation, which regards any annual increase in personal wealth as taxable income. This is uncontroversial for individual pension wealth accumulated in financial institutions like pension funds, insurance companies, or banks, because wealth accruals increase individuals’ ability to pay and should therefor e be taxed as a component of comprehensive income. Economically this is also true for notional pension wealth accruals within a statutory or mandatory occupational pension system, because individual pension claims under these systems increase ability to pay, although pension benefits are not capital-funded but financed on a pay-as-you-go (PAYG) basis. In fact, this difference 23 between funded and unfunded pensions has led to a different tax treatment of these pensions. To compare national pension tax practices, three phases of capital accumulation are distinguished in which income taxes can or should be levied: pension wealth accumulation through contributions or savings, returns on accumulations, and dissaving or withdrawal of pension wealth. Technically, comprehensive income taxation of savings can be characterized by a T-T-E income tax, where T indicates that the respective income flow is taxed at the going tax rate and E indicates that it is tax-exempt. With respect to old-age pensions’ comprehensive income taxation, T-T-E requires that income used to contribute to a pension system should be taxed; growing pension claims as returns to pension wealth should be taxed as well; but withdrawals of pension wealth are tax-exempt. In contrast to the comprehensive income principle, most national income tax codes tax PAYG financed pensions as E-E-T, which implies exempting income spent on pension contributions and income from accruals in pension claims and taxing withdrawals of pension benefits. While a long-lasting dispute persists among public economists whether to tax capital income according to either T-T-E (Schanz/Haig/Simons) or E-E-T (Fisher/Kaldor), tax lawyers argue that the difference in taxing pensions is nondiscriminatory if statutory pensions are preferentially taxed as (deferred) labor income and funded pensions are double taxed as capital income. A survey of pension taxation in OECD countries shows a much broader variety of tax rules for different forms of pensions (Table 4.1). To capture the different tax rules, “t” and “s” are introduced to indicate that in a certain phase of the pension cycle a lower tax rate, t