Lashed to the Mast? The Politics of NDC Pension Reform

Chapter 14 Lashed to the Mast? The Politics of NDC Pension Reform Sarah M. Brooks and R. Kent Weaver* IN THE FINAL DECADES OF THE TWENTIETH CENTURY, governments around the world began to dramatically change the form and function of old-age pension systems. These reforms were generally motivated by a combination of population aging, slowing economic growth, and tightening budget constraints; together these led both to rising pension costs and to declining resources with which governments could finance old-age pension liabilities.1 Many governments responded to these pressures by revising the parameters of traditional defined benefit pay-as-you-go (PAYG) public pension systems with incremental changes to contribution or benefit rates, retirement age, or indexation rules to keep pension systems in line with changing economic and demographic trends and state fiscal capacity. Other governments, however, opted for more fundamental, structural revisions of the design and objectives of old-age pension systems through the creation of mandatory individual, financial defined contribution (FDC) pension schemes, wherein pension benefits are based on individual contributions to a (typically) privately managed pension fund, and the market return to capital on those funds. Throughout the 1990s, the ideas and technology behind funded pension schemes were disseminated broadly throughout the world. By the end of the decade, more than 20 countries, from South America to East Asia, Europe, and the Former Soviet Union had adopted funded, defined contribution pension schemes either as the dominant “pillar” in mandatory pension schemes, or as part of multipillar structural reforms.2 By the middle of the 1990s, however, a new model of structural pension reform—the non-financial (or notional) defined contribution (NDC) scheme—had also emerged.3 The * Sarah M. Brooks is assistant professor of political science at Ohio State University; R. Kent Weaver is professor of public policy and government at Georgetown University and a senior fellow in the Governance Studies Program at the Brookings Institution. The research reported herein was partially funded pursuant to a grant from the U.S. Social Security Administration (SSA) funded as part of the Retirement Research Consortium at Boston College. The opinions and conclusions are solely those of the authors and should not be construed as representing the opinions or policy of the SSA or any agency of the federal government. The authors would like to thank Daniele Franco, Agneta Kruse, and Edward Palmer for extensive and helpful comments on an earlier draft of this chapter. 345 346 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES NDC approach was the key feature of a national pension reform in Sweden that was adopted as framework legislation in 1994 and as final legislation in 1998. Italy, Latvia, and Poland followed Sweden’s lead. NDC pensions combine the PAYG financing that is characteristic of traditional defined benefit (DB) schemes with the defined contribution (DC) structure of individual accounts. NDC schemes tie benefits closely to individual contribution history over an entire working life, but credit those contributions with a notional interest rate tied to wage growth or overall economic growth rather than providing a return on specific financial assets. Notional accounts contain no real capital that can be claimed at retirement as a lump sum or that can purchase an annuity in the private market. Instead, at the time of retirement, the government converts the notional account balance into an annuity on the basis of cohort life expectancy, and finances this benefit on a PAYG basis. Although NDC pension schemes represent an important departure from both the incremental reforms of DB schemes and from individual account, prefunded DC models, many of the individual elements of the NDC scheme design that allow it to control costs are not new; they have been utilized on an ad hoc and often temporary basis in incremental reforms to PAYG-DB pension schemes. The originality of the NDC model lies primarily in combining those elements into a coherent package with a clear policy objective, and imbuing that package with presumed perpetuity. A number of “good policy” arguments have been advanced for adopting NDC measures. First, proponents cite the enhanced “fairness” associated with the DC formula over DB schemes that base retirement benefits on the last salary or last few years’ wages, which tend to redistribute implicitly toward high-income workers with steeper earnings profiles. NDC schemes make any redistribution more explicit by removing it from the main NDC pillar of the pension scheme and creating a separate scheme with the objective of reducing poverty. Second, supporters of NDC-based reforms argue that the tighter link between contributions and benefits signals to workers that it is important to work longer to secure an adequate pension, thus enhancing system financing along with individual responsibility and work effort. Third, NDC-based reforms ensure a long-term balance between pension contributions and payouts. Fourth, NDC reforms do not expose individuals to shortand medium-term fluctuations in market returns and annuity prices that may arise in FDC individual accounts. Fifth—and of particular importance to government finance ministries—NDC pension reforms may provide a fiscally attractive alternative to funding the transition to funded DC plans because they do not require government to finance benefits for a transitional generation as a country moves to a system of funded individual accounts. A final potential advantage offered by NDC reform is the perception of permanency, and hence credibility, that it imparts to the social security system. By reducing the need for governments to intervene regularly to adjust pension system parameters, the “political risk” of policy change in response to political pressures associated with public pension systems is significantly reduced. Indeed, the repeated tinkering with rules of DB systems in many countries has made those systems much less of a “defined benefit” in practice than in theory, especially for future retirees whose benefits are most likely to be affected. The sense of ownership felt by workers who receive regular statements tracking their rising notional account balances reinforces this sense of permanency of the NDC scheme. The resulting sense of property rights may create a “lock-in” effect that deters politicians from intervening to arbitrarily reduce or confiscate the notional capital accumulated in the NDC accounts. On the negative side, however, because NDC schemes accommodate increasing longevity completely through benefit reductions, stabilizing pension contribution rates L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 347 will lead to gradual erosion of pension values as populations age if workers do not postpone their retirement. This discussion of potential “lock in” and political risk suggests that political calculations as well as policy advantages and disadvantages may play a very important role in policy makers’ calculations on NDC-based reforms. Indeed, the central working hypothesis in this chapter is that NDC-based pension systems offer important political advantages to politicians in an era when credit-claiming opportunities in pension policy are few and blame-avoiding incentives are strong. In particular, NDC-based systems: • Provide a “clean hands” mechanism that lowers replacement rates as the system is phased in while allowing politicians to avoid blame. • Obfuscate the degree of future retrenchment because it is not known in advance, but rather depends on future economic and demographic developments. • Avoid having to repeatedly deal with pension retrenchment and refinancing in the future. In short, the complexity and automaticity of the NDC scheme creates important opportunities to limit traceability and blame for benefit retrenchment: automatic adjustment mechanisms based on economic and demographic trends absolve politicians of responsibility for potential future benefit reductions. Like Ulysses in resisting the Sirens, governments may be able to “lash themselves to mast” of a fixed contribution rate and automatic adjustment mechanisms, resisting temptations to pay unsustainable benefits to current and future retirees. The adoption of NDC schemes may also generate new political risks, however, because they accommodate increased liabilities (such as those from unanticipated longevity gains) or revenue shortfalls (for example, those due to wage decline) by reducing benefits. To the extent that benefit values for younger birth cohorts fall short of public expectations about the absolute and relative value of pensions, politicians may confront a powerful political backlash as populations age. This may in turn lead to a “loosening of the lashes” on contribution rates and automatic adjustments, to contributions to the system from the general government budget, and/or to pressure for enhanced “social protection” pensions outside the NDC pillar. The spread of NDC-based pension reforms thus raises important questions both for social science theory and for pension policy making. From the perspective of social science theory, NDC-based pension reforms represent an important example of what Jon Elster has called “pre-commitment” or “self-binding” efforts to limit future options in a way that furthers their long-term interests. Indeed, Elster uses the Ulysses and the Sirens metaphor in much of his writing on precommitment (Elster 1979, 2000). NDC-based pension reforms can also be seen as a combination of several blame-avoiding strategies that allows politicians to reconcile their policy and political objectives by making politically risky decisions through what Kent Weaver (1986, 1988) has called “automatic government” mechanisms rather than by requiring politicians to make those decisions openly. Thus NDC systems do not represent a departure from the common practice in PAYG-DB pension systems of making hard-to-detect revisions, such as revisions to indexation rates, in order to achieve fiscal and policy goals. Examining the adoption and implementation of NDC-based pension reforms provides an opportunity to examine how agenda-limiting precommitment mechanisms are adopted and how they operate in practice. Two questions are addressed in this study. First, are NDC-based reforms more likely to get on the agenda and win adoption in some countries than in others? If so, is it the characteristics of a country’s current pension system, the political ideology of elites, or the characteristics of the political system—or some 348 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES combination of these and other factors—that determines whether countries consider and adopt NDC-based reforms? Second, do NDC-based reforms actually succeed in de-politicizing painful and costly pension retrenchment decisions and limiting blame to incumbent politicians, or do they have a tendency to spark resistance that undercuts their intended effects? In the first section of the chapter we briefly outline the characteristics of NDC pension systems, arguing that NDC pension schemes should be seen as a set of principles that may be more or less closely followed in practice. The second section develops a framework for analyzing why NDC pension systems have increasingly been on the agenda in recent years and discusses conditions that facilitate considering and adopting NDC-based reform, including the complex question of why there has been a stronger move toward adopting NDC in some countries and regions than others. The third section of the chapter examines implementation challenges that may arise in NDC systems as they are adopted and mature, as well as their political sustainability. The final section of the chapter assesses the prospects for a further spread of NDC pension systems. In short, the chapter asks whether and under what conditions adoption of NDC pensions is likely to be an effective means of allowing politicians to “lash themselves to the mast” of a stabilized contribution rate and depoliticize the process of pension retrenchment. This, in turn, plays into the broader question of whether NDC-based reforms are likely to play a major role in addressing the immense aging issue facing both developed and developing societies in coming decades. How NDC Works Although the NDC system represents a synthesis of the PAYG-DB and FDC systems, it differs from these in the way it apportions risk and reward, and in its likely political consequences. Although notional accounts share with FDC schemes a tight link between pension benefits and individual contributions, NDC systems are by design not advance funded.4 As noted above, NDC schemes also differ markedly from FDC systems in their treatment of capital market risks. NDC systems diminish individuals’ exposure to fluctuations in market rates of return and annuity prices associated with privately managed funded-DC schemes by using a notional interest rate. Workers in NDC schemes continue to be exposed to significant demographic risks, however, such as unanticipated gains in longevity during working life, and to sustained declines in fertility around the time of retirement. By shrinking the overall contribution base to the pension system, such trends would cause declines in pension benefits to maintain overall financial balance of the NDC system.5 At the same time, the annual indexation of annuities to wage growth exposes retired workers to the risk of declines in benefits if wages and productivity fall. NDC systems also differ from traditional DB schemes in important ways, as shown in the first and fourth columns of table 14,1, which show different gradations of DB and NDC pension schemes. But as the second and third columns of the table show, a number of “middle positions” between DB and NDC pension schemes are possible on many of these key elements of pension system design. Some elements associated with NDC pension systems, such as life expectancy adjustments, have been enacted as ad hoc reforms of existing DB pension systems (column 2). And NDC-based systems sometimes take a “weak” or “partial” form, with one or more provisions that make them less than fully self-sustaining or inclusive (column 3). Table 14.1 divides these elements of pension system structure into four categories: structural features, coverage, time horizon, and exclusivity. Table 14.1. Non-financial Pension Provisions As a Continuum from Defined Benefit to Defined Contribution Provision Structural features Funding Defined benefit Middle position DB reforms “Weak” or “Partial” NDC “Strong” or “full” NDC Entirely PAYG May have some advanced funding, usually for liquidity purposes Entirely PAYG Reserves built up for large demographic cohorts, to hold transfers from general budget and for long-term system balance Years of earnings and contributions incorporated in benefit rule Can include anything from no work (but residency or nationality) requirement to specified years of contributions (including “best years” and “final salary” arrangements) Replacement rate based on number of years that corresponds to the entire working life for typical worker May include some nonfinanced credits for nonwork activities Full link between contributions and earnings Life expectancy at retirement No provision in benefit calculation Inclusion of “demographic factor” in benefit calculation to fully or partially compensate for population aging Infrequent or incomplete (for example, exclusion of postretirement) adjustment for increases in longevity; no automatic mechanism to correct for incomplete changes Benefit levels adjust fully (including current retirees) and automatically for increases in longevity Retirement age Fixed standard retirement age (may include actuarial adjustments for earlier or later retirement; may be ad hoc increases in retirement age over time) Flexible retirement age and/or automatic increases in retirement age to fully or partially compensate for longevity increases Minimum age to claim benefit, but no standard retirement age. Inadequate adjustment for earlier or later retirement Minimum age to claim benefit, but no standard retirement age. Partial retirement possible. Full actuarial adjustment for earlier or later and partial retirement Table 14.1. (continued) Provision Defined benefit Middle position DB reforms “Weak” or “Partial” NDC “Strong” or “full” NDC Inflation and economic growth Benefits adjusted for inflation and/or earnings increases, sometimes with ad hoc changes to restrain costs Benefits adjusted for inflation and/or earnings increases; indexation rule incorporates brakes for poor economic performance and high inflation, which may or may not be followed in practice Benefits adjusted with inflation and/or earnings. Incomplete brakes for poor economic performance and/ or government makes ad hoc adjustments in brake to enhance electoral prospects Benefits adjusted with the internal rate of return, which is tied fully to earnings growth Financing May include payroll taxes and/or general revenues; revenues adjusted to needs of PAYG system on automatic or ad hoc basis Payroll taxes fixed “in theory” at maximum target level (for example, below 10 percent in Canada, 20 percent in Germany) but may be revised in practice Payroll tax rate with part of contribution rate used to cover pension rights granted but not covered by individual’s own contribution Fixed payroll tax rate; all contributions paid give an NDC account value, and all noncontributory rights are financed externally (for example, with general revenues) Redistribution across generations First generations generally winners as in “Ponzi” scheme May be restricted within DB tiers by lowering replacement rates and tying contribution rates to level needed to be self-sustaining over long term Credits given to some cohorts for which no contributions were made; possible generational redistribution due to shifting labor market Possible generational redistribution due to shifting labor market Redistribution within generations Permitted within DB tier, based on limited number of “high years” on which benefits are based, higher replacement rates for low earners, credits given for which no contributions have been made May be restricted within DB tiers by increasing number of years on which benefits are replaced, eliminating or reducing replacement rate differentials and limiting credits for nonwage activity Some credits given for activities for which no contributions were made Barred within NDC tier unless financed by payments from government or others with respect to nonemployment activity. Exception: NDC implicitly redistributes from men to women and other population subgroups with longer life expectancy if single annuitization table is used Coverage No NDC coverage Not applicable Only some sectors (for example, private sector) are covered by NDC system All workers in specified age cohorts are covered by NDC system Time horizon No coverage of NDC system Later cohorts of workers covered by “quasi-NDC” reforms (for example, lifeexpectancy adjustments to benefits or retirement age) Only some cohorts (for example, new labor market entrants or those under age 50) covered by NDC system All employed workers, including those in labor force at time NDC introduced, covered by NDC system Exclusivity No NDC pension tier Not applicable NDC-based pension is only one of several public pension tiers NDC is only public pension tier Source: Compiled by authors. P ENSION R EFORM : I SSUES 352 AND P ROSPECTS FOR NDC S CHEMES Structural Features A wide range of variation is possible on multiple dimensions of the structural features of pension systems (table 14.2). For example, while most non-financial DB pension schemes operate solely on a PAYG basis, the Swedish NDC scheme incorporates “buffer funds” that prefund some benefits and protect against small dips in contributions and demographic bulges in the population of retirees.6 Indeed, proponents of NDC argue that buffer funds should in principle be included in an NDC scheme (see Palmer 2006a). Table 14.2. The NDC Pension Continuum in Practice Provision Sweden (NDC initiator; full NDC) Poland (full NDC) Germany (middle position DB) Structural features Advanced funding PAYG plus buffer fund partially accumulated under old pension system PAYG plus buffer fund (surplus in 1st pillar + privatization revenue + 1% temporary contribution) PAYG with general revenues and “eco-tax” as well as payroll tax; small “sustainability reserve” Life expectancy at retirement Unisex, account balances and benefits adjusted both before and after retirement Unisex, but calculated for life expectancy at the age of retirement Benefits adjust for changes in system dependency ratio Retirement age Flexible, with NDC and FDC pensions drawable no earlier than age 61; partial withdrawal possible minimum: 60 (women) 65 (men) Normal retirement age of 65; incentives for early retirement being reduced Inflation and economic growth Account balances and benefits adjusted for wage growth (initial benefit is higher than actuarial amount and adjusted for wage growth minus 1.6%) Accumulation is wage growth plus labor force growth; annuities indexed to consumer prices, unless real wages are falling, in which case they are uprated in line with nominal wages (that is, cut in real terms) Benefits adjusted for wage growth Payroll tax rate Fixed at 16% for NDC tier and 2.5% for FDC tier 12.22% to NDC 7.3% to FDC Government commitment to hold contribution rate to no higher than 20% through 2020 and 22% through 2030 Redistribution across generations Eliminated once NDC system is fully phased in Minimal, through minimum pension guarantee Modest, through DB formula Redistribution within generations In pension annuity, from shorter-lived to longerlived pensioners, and from men to women In pension annuity, from shorter-lived to longerlived pensioners, and from men to women In pension annuity, from shorter-lived to longerlived pensioners, and from men to women L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 353 Table 14.2. (continued) Provision Sweden (NDC initiator; full NDC) Poland (full NDC) Germany (middle position DB) Coverage Universal Not universal; separate programs for farmers and uniformed services Universal Time horizon Persons born 1938–1953 receive benefits partially in old system. Persons born 1954 and later receive benefits entirely in new system New system mandatory for all born beginning in 1949; FDC mandatory for people born after December 31, 1968 Phases in beginning in 2005 Exclusivity No. Combined with smaller FDC tier and inflation-indexed guarantee pension for those with low lifetime earnings No. Combined with FDC system, plus tax-financed minimum pension supplement (for minimum 25 years’ contributions) if NDC+FDC annuities are below minimum No. Quasi-mandatory private pillar and “zero-pillar” added Source: Information on Poland is from Góra and Rutkowski (2000) and Chloń-Domi ńczak and Góra (2006); information on Germany is from Börsch-Supan and Wilke (2006); information on Sweden is from Sweden, National Social Insurance Board (2004). With respect to population aging—and in particular rising life expectancy at retirement—PAYG-DB schemes generally do not provide for automatic adjustments to the changing demographic, social, and economic context in which the system is embedded. Thus, as populations age, the schemes may require periodic government intervention to adjust benefit levels or retirement ages. NDC schemes automatically accommodate changes in life expectancy by calculating annuities on the basis of individual accumulations and life expectancy at the time of retirement. But such measures can also be included in DB schemes to partially or fully adjust for population aging, for example the Kohl government’s “demographic factor” enacted as part of a short-lived pension reform and the “sustainability factor” enacted by the Schröder government.7 Similarly, public PAYG-DB schemes generally include a fixed standard retirement age, usually with some sort of adjustment (which varies greatly across countries in its actuarial accuracy) for earlier or later retirement. A number of countries have raised the standard age in recent years in response to population aging. The Swedish NDC scheme, on the other hand, has replaced the standard retirement age with flexible retirement age (pensions can be drawn no earlier than age 61), no upward age at which pension rights can be earned, and with the benefit based on the life expectancy of the retiree’s birth cohort at the time of retirement. But NDC-like proposals have also been made for retirement age changes in public PAYG-DB, such as proposals in the United States that would raise the age for receipt of “full” social security entitlements as longevity increases. Perhaps the most distinctive attribute of NDC pension systems is the fixing of a longterm contribution rate and the dependence of pension adjustments on the size of the wage base. With this “lashing to the mast” of contribution rates, future adjustments to keep pension systems sound must in theory be made on the benefit side. For DB pensions operated on PAYG basis, on the other hand, payroll tax contribution rates and benefit levels are usu- 354 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES ally adjusted on an ad hoc basis as current funding needs change. A number of countries also use general revenues or some other form of revenue (for example, the German ecotax) to finance part of their PAYG-DB pension system costs. Given increased longevity in most countries and flagging employment growth in a number of countries, there has been strong upward pressure on DB pension contribution rates over the past 30 years.8 In response to this trend, many countries have in recent years tried to stabilize contribution rates to their DB pensions through a variety of mechanisms. In Canada, for example, the federal and provincial governments have pledged to keep payroll taxes under 10 percent in the long term; if benefit costs are projected to exceed that target within a specified projection period, a combination of benefit cuts and contribution rate increases is automatically triggered.9 Germany, where pension contribution rates grew to over 20 percent of earnings in recent years, has also acted to try to stabilize contribution rates at no more than 20 percent through the year 2020 and 22 percent through 2030.10 In the United States, congressional Republicans made it very clear at the time of the 1983 Social Security rescue package that they would tolerate no further increases in contribution rates, and there have been none.11 In short, even without the explicit “lashing to the mast” of contribution rates associated with NDC pension reforms, public non-financial DB systems can and have undertaken a number of actions to stabilize contribution rates. Non-financial DB and NDC pension systems also differ significantly in their potential for intergenerational redistribution. DB pensions typically offer a much higher rate of return on contributions to the first generations in the program, when the ratio of contributors to beneficiaries is much higher than in later generations (especially as longevity increases). NDC pensions, on the other hand, are intended to restrict intergenerational redistribution. However, several PAYG-DB pension schemes, notably in Canada and the United States, have also moved to restrict intergenerational redistribution by making very long term projections of contribution rates and benefit costs and developing contribution rates and benefit levels that are intended to be stable over that projection period with the support of reserve funds.12 In principle, NDC schemes treat all contributions to the system the same way in terms of accruing credits (with adjustments for wage or—as in Italy—GDP growth over time): only actual contributions accrue credits, and all credits result in equal payouts (actuarially adjusted for life expectancy at retirement). If credits are granted for other reasons, they must be accompanied by financing from general tax revenues, or the financial equilibrium of the system will be jeopardized. Even NDC systems may contain some elements of intragenerational redistribution in payout, however, notably in the use of gender-neutral annuitization tables that do not reflect real gender differences in life expectancy. Many PAYG-DB schemes include more complex patterns of intragenerational redistribution. Some countries, notably the United States, offer higher replacement rates for the first dollars of earnings, effectively offering a higher replacement rate for low earners. Many PAYG-DB schemes also base benefits on a contributor’s highest or final number of earnings years (for example, the highest three, fifteen, or twenty years). To the extent that the benefit formula is based on final salary or only the latest years’ earnings, it privileges workers with a steeper career-earnings profile. Such workers are typically the most affluent and better educated in society. In recent years, public PAYG-DB pension systems in many countries have been altered to limit redistribution within generations in some ways, for example by lengthening the number of contribution years used in calculating pension entitlements or by flattening replacement rate differentials across earnings levels. But there has also been a widespread expansion in noncontributory credits given for caregiving in a number of countries. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 355 The use by NDC schemes of a benefit formula based on contributions over the entire career both diminishes the degree of regressive redistribution caused by final salary benefit formulas and at the same time eliminates the possibility for progressive redistribution of pension benefits. The overall progressivity of an NDC-based retirement income system thus depends critically on the existence of mechanisms within or outside the NDC component to progressively reapportion risk and benefits to those who are the least well off. This can be done in either of two ways. First, contributions can be made for individuals that increase their benefit entitlement. In Sweden, for example, contributions are made into the system for periods spent in providing childcare, military service, unemployment, sickness and disability covered by public social insurance, and for compensated parental leave and higher education. The money to finance these credits is transferred from the general budget to the NDC reserve fund on a yearly basis. Partial NDC reforms have imputed credits without financing them immediately, but these elements require a future tax-based transfer burden to make the system sustainable. Second, changes may be made in other tiers of the retirement income system that add or “reinvent” existing redistributive elements, such as by topping-up benefits to a minimally guaranteed level. To date, all countries introducing NDC schemes have constructed a minimum benefit guarantee financed from general revenues. Indeed, without an adjustment of other pension tiers to ease the transition to NDC, the distributional change resulting from a change to NDC is likely to impose such heavy costs on some groups that its chances of winning adoption and being sustained are substantially reduced. Time Horizon Pension reforms can also be distinguished by their phase-in periods. The U.S. increase of retirement age from 65 to 67, for example, is being phased in over almost 40 years from the time it was enacted to the time it will be fully in effect. To date, countries implementing NDC reforms also show great differences in the time horizon over which they phase in their reforms. NDC reforms can be applied to rights acquired under the previous system, if adequate contribution records exist.13 In the Swedish reform legislated between 1994 and 1998, for example, all workers born after 1953 are entirely in the new system; earlier cohorts born between 1937 and 1953 are partially in the new system. In Italy, on the other hand, only new entrants to the labor market will have their benefits fully calculated under the new NDC rules. A longer phase-in period is likely to weaken opposition from the most attentive publics—the elderly and the near-elderly—but it also lessens near-term budgetary savings from reform. Coverage Countries can also differ in the degree to which the NDC system will apply to all members of the workforce when it is fully phased in. Excluding key sectors (for example, the military, or, as in Poland, farmers) from a shift to NDC may help governments manage the politics of reform, although doing so clearly raises questions of equitable treatment. It will also mean that career changes into a sector covered by NDC can entail losing benefit privileges in the sector from which workers migrate. Exclusivity Most countries have multipillar pension systems. Thus even when an NDC system is fully phased in, it may not be the sole, or even the largest, source of public pension income. The 1994 pension reform in Sweden, for example, created a multipillar pension scheme in which the average wage worker could expect to receive a retirement pension from an NDC 356 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES account and a mandatory individual financial DC account. There is also a garantipension financed from general revenues for low-wage workers and workers with interrupted earnings histories, which tops up their accumulated pension accounts to a socially acceptable minimum. The NDC tier is dominant, however—financed by a 16 percent payroll contribution, compared with a contribution of 2.5 percent of covered wages into financial (that is, funded) individual accounts. The Polish government similarly adopted a mixed NDC and funded-DC scheme, wherein 15 percent of covered wages are credited to the first (NDC) pillar and indexed with a notional interest rate equal to growth in the covered wage bill. Nine percent of covered wages are transferred to the funded scheme in Poland, and invested according to market principles.14 The Latvian pension reform also combines an NDC pillar with a financial DC scheme. The NDC pillar was financed with contributions of 20 percent of covered wages from the outset. The individual financial account scheme, which came into force in 2001, initially received 2 percent of covered wages, while contributions to the NDC component decreased to 18 percent. Following a transition schedule specified in the law, the split will be 10 percent and 10 percent by 2010.15 Overall, the preceding discussion suggests that a political analysis of NDC pension systems must recognize that the boundaries between and NDC and non-financial DB pensions are fuzzy rather than sharp. Not only have certain elements of NDC, such as life-expectancy adjustments, been used since the mid-1990s on an ad hoc basis in incremental DB reforms, but also countries such as Italy have adopted partial NDC reforms in ways that may be unstable in the medium or long terms. In the case of Italy, for example, the government imputes contributions to notional accounts for some groups higher than those that are actually made. At the same time, the Italian system does not have automatic stabilizers, nor does it automatically take into account pre- or postretirement increases in longevity. Finally, Italy has established an inflexible rate of return on accumulated contributions, raising questions about the medium- to long-term stability of the system.16 If incomplete adoption of NDC is possible, so too, presumably, is an incomplete dismantling: thus an analysis of NDC politics must also include an assessment of the risks of an ad hoc unraveling of NDC reforms. Despite the tighter link between contributions and benefits, NDC schemes may nonetheless bear the heavy imprint of political values, objectives, and concessions to powerful domestic political groups. Governments may intervene in the accumulation phase of the NDC system, for instance, to credit socially valued activities that are not rewarded by markets, such as the provision of credits for child rearing, education, and military service. At the same time, powerful groups such as the military or specific industrial sectors may claim privileges within the NDC framework to the extent that governments credit notional accounts for workers in these sectors to finance special benefits such as early retirement.17 The use of unisex mortality tables likewise represents a political decision to redistribute from men to women. At the same time, NDC systems usually redistribute credits of workers who die before retirement to other workers within the same cohort, rather than to the workers’ surviving family members. This redistribution of credits represents a political judgment in which the claim of survivors within the family is weighed against those of workers within the cohort, to whom such credits are apportioned. The Politics of NDC Innovation, Diffusion, and Adoption The discussion above suggests some important general propositions about the politics of adopting NDC-focused pension reforms. First, it suggests that although NDC-based reforms have some features that may be politically attractive for reelection oriented politi- L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 357 cians—in particular, such reforms enhance the capacity for politicians to distance themselves from long-term benefit and eligibility cuts—they are not without political drawbacks as well. Second, it suggests that the attitudes of important actors in pension reform politics— notably finance ministries, trade unions, and groups representing retirees—are likely to be heavily influenced by what else is in the pension reform package, particularly by what protections are afforded to the aged through the provision of income-tested pensions or supplemental DC pensions to compensate for any losses, and who is expected to pay for the costs for those parts of the reform package. Third, this analysis suggests that proponents of an NDC-based reform may be able to affect the political prospects for reform through specific design features in the reform. In particular, the pace of transition strongly influences the degree and visibility of the material and distributive effects of NDC reforms, and thus the political viability of these measures. If a reform is phased in quickly, the terms of the new scheme are not only more costly and transparent, they are also at greater risk of political backlash. Other features of NDC program design that are likely to affect a proposal’s political prospects include the specific provisions of other pension tiers and how they compensate any losers from the shift to NDC (for example, the creation of new income-tested pensions), as well as which groups (for example, civil servants or the military), if any, are excluded from this shift. There may also be significant battles over whether and how much to provide credits for non-wage-related benefits within an NDC pension scheme, as well as over the question of who should pay for those credits. Finally, the politics of NDC adoption and rejection is likely to be a moving target. How politicians perceive and react to NDC proposals is likely to evolve over time as the concept is diffused more broadly among technocratic elites and as experience with implementing NDC-based pension regimes grows. This very general overview of the calculus of key actors does not take us very far, however, toward understanding why NDC pension reforms are likely to get on the government’s agenda and win adoption in some countries and not in others, let alone whether an NDC-based reform will be sustained once it has been adopted. To understand these cross-national variations and how they are likely to evolve over time it is important first to introduce contextual factors that vary across countries, notably the severity of the challenges that extant pension systems face, the weight of various social actors, and the barriers that a political system imposes to instituting major reforms. Second, it is important to take a process-oriented approach, distinguishing between the forces that are critical at different stages of NDC politics: (1) initial development of the NDC idea and its first application, (2) diffusion of the concept to the agendas of other countries, (3) adoption or rejection of the NDC idea once it is on a government’s agenda in a specific country, and (4) sustaining an NDC reform once it is in place. Third, it is important to examine whether NDC is adopted fully or partially—that is, to examine the structural features, coverage, time horizon, and exclusivity of NDC-based schemes in particular countries. Thus we can imagine a continuum of outcomes ranging from early “full” adoption of NDC-based reform to partial and/or later adoption, to consideration and rejection of NDC-based reforms, or to the issue never even making it onto the agenda, as shown in table 14.3. (Of course, countries may shift over time between categories.) Since the NDC “innovation” is now well developed, we will focus on the diffusion of NDC-based reforms and their full or partial adoption or rejection.18 Sustaining implementation of NDC-based reforms will be considered in the next section of the chapter. Table 14.3. Potential Relationships between Structural Variables and NDC Choice Constraints on NDC adoption and diffusion Innovation Early “full” adoption Partial and/or nonexclusive adoption Later adoption Consideration but rejection Never reaches agenda + Very strong demographic aging pressures + Very strong demographic aging pressures + Strong demographic aging pressures + Moderate but increasing demographic aging pressures + Moderate demographic aging pressures + Weak demographic aging pressures Fiscal pressure + Very high budget deficits and debt/GDP ratio + High budget deficits and debt/GDP ratio + Funded DC likely to play relatively larger role than NDC in multitier system where budget deficits and debt/GDP ratio are low + Moderate but increasing budget deficits and debt/GDP ratio + Declining budget pressure and debt/GDP ratio + Low budget deficits and debt/GDP ratio Payroll tax rates + Very high payroll tax rates + Very high payroll tax rates + High payroll tax rates + High or rising payroll tax rates – Low payroll tax rates or no payroll tax + Political forces favoring statefocused public pension system are strong – Funded DC rather than NDC more likely to be adopted where ideologically conservative forces are very strong + Funded DC likely to play relatively larger role than NDC in multitier system where ideologically conservative forces are fairly strong + Parties favoring continued state focus in public pension system gain strength after period of stalemate + Funded DC rather than NDC more likely to be adopted where ideologically conservative forces are very strong Economic/ demographic constraints Aging pressure Ideational forces Strength of domestic forces favoring market-based pension reform Participation in regional networks where NDC ideas common Not applicable + Participation in regional networks with earlier adopters; elites view earlier adopters as peer countries + Participation in regional networks with earlier adopters; elites view earlier adopters as peer countries + Country does not participate in regional networks including early adopters = Regional network effect should dissipate as NDC ideas are fully diffused internationally Interaction with international financial institutions and other supranational organizations Not applicable + Supranational institutions: (1) press for pension expenditure reductions and (2) introduce and support NDC-based reforms + Supranational institutions: (1) press for pension expenditure reductions and (2) introduce and support NDC-based reforms + Supranational institutions press for pension expenditure cuts and support NDCbased reforms, but have limited bargaining leverage + Strong earningsrelated component in pension system + High replacement rates + Strong earningsrelated component in pension system + High replacement rates + Mix of redistribution and earningsrelated principles in current public pension system Replacement rates + High replacement rates + High replacement rates Payroll tax records + Complete payroll tax records + Complete payroll tax records Policy feedbacks Strong earningsrelated pension tier + Country is outside regional networks of early NDC adopters = Regional network effect should dissipate as NDC ideas are fully diffused internationally + Supranational institutions press for pension expenditure cuts and support NDC-based reforms, but have limited bargaining leverage + Low contact with IFIs – IFIs support FDCbased reforms only – Flat-rate benefit in current public pension scheme + Prior adoption of FDC reform + Flat-rate benefit in current DB + Medium replacement rates + Complete or improving payroll tax records + Incomplete but improving payroll tax records + Low replacement rates – Incomplete payroll tax records – Incomplete or nonexistent payroll tax records Table 14.3. (continued) Constraints on NDC adoption and diffusion Political/partisan constraints Multiple veto points in political system Availability of cartelizing mechanisms Innovation Early “full” adoption Partial and/or nonexclusive adoption Later adoption Consideration but rejection – Multiple veto points in political system – Multiple veto points in political system + Multiple veto points in political system + Multiple veto points in political system + Multiple veto points in political system + Strong political mechanisms available to overcome interest group opposition to losses resulting from NDC-based reform + Strong political mechanisms available to overcome interest group opposition to losses resulting from NDC-based reform + Moderately strong political mechanisms available to overcome interest group opposition to losses resulting from NDC-based reform + Moderately strong political mechanisms available to overcome interest group opposition to losses resulting from NDC-based reform + Weak political mechanisms available to overcome interest group opposition to losses resulting from NDC-based reform Never reaches agenda + Weak or nonexistent political mechanisms available to overcome interest group opposition to losses resulting from NDC-based reform Source: Compiled by authors. Note: + condition makes it more likely that country will be in this category; – condition makes it less likely that country will be in this category; = condition that is likely to become less important as NDC innovation is diffused; blank cells indicate no hypothesized relationship between this condition and outcome category. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 361 Agenda Setting and Adoption A critical component of any policy reform process is formulating the policy problem in such a way that the issue achieves agenda status. To Kingdon (1995), policy proposals are likely to reach and remain on a government’s agenda only if they appear to be broadly congruent with the values of policy makers and the public, as well as politically feasible, affordable, and likely to make some improvement in a perceived policy problem. Of course, successful coupling of these dimensions is typically the work of a skilled policy entrepreneur. To Orenstein (2000), such “proposal actors” are critical in establishing the “intellectual agenda” of a reform. In addition to putting reform on the political agenda, proposal actors introduce new policy innovations and delimit the feasible range of policy outcomes for the country.19 Examining structural pension reforms in Eastern Europe and Central Asia, Orenstein found that where there are fewer, and more like-minded, proposal actors, the possibility for adopting an ambitious structural reform was greatly enhanced. Importantly, however, while incorporating a broader range of interests in the early stages of reform (such as where disparate interest groups and government ministries advance reform proposals) tends to diminish the magnitude and pace of reform, to the extent that broad agreement (or “buy-in”) is achieved during the agenda-setting phase, both the implementation and the longer-term prospects for sustainability are enhanced.20 Accordingly, we examine in the next sections the key factors shaping whether and to what extent countries move toward NDC pension reform. These include (1) whether the problem or challenge of structural pension reform is brought onto the national agenda and how the problem is defined by key actors, (2) whether the specific policy alternative of NDC pension reform is one of the options considered by policy makers, and (3) how policy feedbacks from existing pension regimes as well as (4) the structure of political institutions shape both the alternatives considered and the prospects for NDC adoption. In examining the first factor, we look at the extent to which governments are pressed to reform state pension schemes by varying degrees of demographic change, budgetary constraints, and macroeconomic pressures. Second, we discuss how exposure to, and hence adoption of, NDC ideas are influenced by the strength of various domestic ideological groupings as well as countries’ interaction with supranational institutions such as the European Union and the World Bank. In looking at the third factor, we examine how feedbacks from prior policy choices may make considering and adopting NDC-based reforms more or less likely, such as by shaping how the problem and its feasible solutions are perceived, and by influencing the political coalitions that have developed around current policies. Last, we examine how the structure of political institutions and political competition affect both the political advantages and costs associated with NDC systems, and hence the likely barriers to policy change.21 Table 14.3 outlines hypothesized relationships between specific variables and several possible outcomes (for example, first innovation with NDC, early adoption, partial adoption, consideration and rejection of NDC, NDC never reaching agenda), while table 14.4 develops an informal “scorecard” of the relationship between causes and outcomes. Specifically, the table highlights the hypothesized relationship between our key variables and the outcomes achieved by several countries with respect to NDC reform (for example, with Sweden as innovator, Italy as a partial adopter of NDC, Uruguay as a country that rejected NDC in favor of a mix of DB and DC, and the United States representing a country where NDC has not yet reached the agenda). Economic-Demographic Pressures for System Change The wave of structural pension reforms that began in the 1980s and swelled in the 1990s has been widely attributed to significant demographic and economic changes, which Table 14.4. Country Characteristics and NDC Reform Outcomes Country Outcome Sweden Full NDC system adopted Economic/demographic constraints Aging pressure (% population 65+ Very high (17.4%) in parentheses) Italy Germany Uruguay United States Partial NDC system adopted with long phase-in Elements of NDC adopted, retracted, and readopted Mixed DB-funded DC reform NDC not on agenda Very high (17.8%) Very high (16.1%) High (12.9%) High (12.3%) Fiscal pressure Very strong in early 1990s Very strong Very strong Very strong Moderate Payroll tax rates (total payroll tax rates in parentheses) High (26.09%) Very high (41.11%) Very high (40.91%) Very high (35.5%) Moderate (22.7%) Weak Weak Fairly weak Weak Strong Interaction with IFIs and other supranational organizations No No interaction with IFIs, but European Union pressure to reduce pension spending No interaction with IFIs, but European Union pressure to reduce pension spending Yes, with Inter-American Development Bank, which presented NDC option in 1992. No Participation in regional networks where NDC ideas are common Became hub of network Yes Yes No No Policy feedbacks Strong earnings-related pension tier Yes Yes Yes Yes Yes Replacement rates Very high Very high Very high High Moderate Complete payroll tax records Yes Yes Yes No Yes Political/partisan constraints Multiple veto points in political system? Weak Strong Weak Strong Strong Strong Mixed Declining Weak Weak Ideational forces Strength of domestic forces favoring market-based pension reform Mechanisms for overcoming policy gridlock? Source: For data on aging and payroll tax rates: U.S. Social Security Administration, Social Security Programs Throughout the World, most recent editions. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 363 raised the specter (and reality) of wide deficits in social security budgets and tighter pressures on overall government spending. These developments have clearly had a major impact on pension reform agendas. Aging populations are a critical source of pressure for change in pension schemes. Most countries operate their pension systems on a PAYG basis, even where there is a dedicated payroll tax for pensions. Falling birthrates and life expectancy increases mean that there are fewer workers to support each retiree—a trend that is only expected to increase in future years. Demographic challenges vary significantly across nations, however.22 In general, we would expect that countries that have had a high percentage of their populations over age 65 for an extended period would be most likely to have exhausted incremental retrenchment and contribution reforms and be more likely to consider NDC and other sorts of fundamental pension restructuring initiatives. Consistent with this hypothesis, many of the early adopters of NDC (notably Sweden and Italy) are among the world’s oldest societies. Fiscal concerns also increased pressure for austerity in public pension systems. Rising government deficits and debt to GDP ratios are clearly likely to raise pressures for pension reform. But the effects of such a fiscal crunch are complex. Indeed, a quantitative study of 57 developed and developing countries has shown that countries with a high public debt to GDP ratio are less likely to privatize their pension programs (at least when pension liabilities are low or moderate), because they cannot afford the transitional costs associated with moving from a public PAYG to fully funded individual accounts. 23 For example, Pinheiro (2004, pp. 112, 134) argues that Brazil’s rejection of privatization in the late 1990s stemmed in large part from the high transition costs implied by such a measure. A plausible hypothesis is that a shift to NDC is most likely to at least get on government’s agenda where (1) a government faces severe fiscal pressure, and (2) a high debt to GDP ratio and/or current fiscal pressures inhibit a move to funded DC. Strong concerns about economic competitiveness sparked in particular by high payroll tax rates may also increase interest in NDC-based reforms. Higher capital tax rates are also severely constrained as a revenue source.24 NDC-based reforms offer the opportunity to “lash policy to the mast” of a permanently fixed pension payroll tax, while the exact implications of that change for individual beneficiaries is made contingent on future economic and demographic developments. In general, we would expect countries that already have very high payroll tax rates along with the prospect of rapidly aging populations to be most likely to consider NDC. Countries without a preexisting payroll tax structure, like New Zealand, would be likely to find the transition to NDC politically and administratively very difficult. The data in table 14.4 suggest that consideration of NDC-based reforms is associated with countries with high or very high overall payroll tax rates. Overall, there does appear to be an association between strong demographic, fiscal, and payroll tax rate pressures and the likelihood that an NDC-based reform reaches a country’s policy agenda. Nevertheless, it is probably most accurate to infer from existing evidence that economic and demographic pressures are helpful in explaining when pension reform generally achieves agenda status, and when the alternative of a radical shift toward privatized individual accounts is less plausible. Thus, although demographic and economic pressures may have some impact on which policy responses are considered, they are insufficient to explain when NDC-based reforms in particular, rather than some other type of restructuring measure, are seriously considered or adopted. Ideational and International Forces In order to understand how and when the particular policy innovation of NDC pension reform is likely to be adopted in a given country, we must examine the ideational and 364 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES international forces shaping the “policy stream,” or range of options that are included in and excluded from serious policy debate.25 At least three types of forces are important to the flow of pension reform ideas. First, the “intellectual agenda” is likely to be influenced in a given country by the strength of the domestic political base for conservative ideological critics of public PAYG pensions. Specifically, we expect that proposals for pension privatization are likely to dominate debates over pension restructuring in countries where both the overall ideological climate and the ideology of the governing political elites is suspicious of governmental action and has strong faith in market solutions to policy problems. NDC solutions, by contrast, are more likely to be on the agenda where pressures for pension restructuring are strong but the national (and governing political party or coalition) ideological climate is more favorable to government and less favorable to privatization and individual accounts. Intermediate outcomes (for example, greater or lesser roles for NDC and FDC in a multipillar system) may occur when forces are more evenly balanced in a political system. Such prolonged haggling can in fact be observed in negotiations between the Social Democrats and nonsocialist parties over the relative sizes of the NDC and “premium pension” (FDC individual account) tiers within the new Swedish pension system.26 In addition to domestic sources of pension reform ideas, the policy models that gain agenda status may be “cued” from a second source: archetypal policy models or practices to which policy makers are exposed through informal regional networks of specialized policy elites.27 Beyond the cognitive ease of looking to relevant examples rather than “reinventing the wheel,” competitive and status concerns have entered prominently in the adoption of structural pension reforms in the past decade.28 In particular, government technocrats draw information about the viability of a policy model from the experiences of relevant or “peer” nations. Such countries may be those that share similar structural conditions and policy regimes, or they may be those with common demographic, economic, or political structures. At the same time, peer nations may be those with which policy elites interact on a regular basis through regional organizations such as the European Union and the OECD, which actively promote policy diffusion—and harmonization— across member nations.29 All of these factors suggest that, at least initially, the dissemination of policy models is likely to be heavily regionalized—and available evidence confirms this hypothesis.30 Just as the Latin American nations looked more to the Chilean model than to pension reformers in other regions, early adoption of the NDC system has been concentrated in Europe, with Italy, Latvia, and Poland among the early adopters, while recent German reforms contain elements of NDC, and NDC-inspired reforms are being considered in the Czech Republic and Norway. The effect of such regional networks as diffusion mechanisms may weaken over time, however, as NDC ideas become broadly disseminated through professional networks of pension experts. A third source of ideas and pressures for specific courses of pension reform transcends regional networks; examples are “supranational” institutions such as the European Union (for member countries) and international financial institutions such as the World Bank and International Monetary Fund. Scholars have long attributed a central role to international organizations in shaping domestic policy processes around the world through the diffusion of norms and ideas.31 We expect (1) that some form of structural pension reform is more likely to get on the agenda when a supranational institution sets deficit and debt reduction targets for participation in its programs, as in the case of the European Union; (2) that NDC in particular is more likely to get on a government’s agenda when a government participates in institution-sponsored networks where NDC ideas are active; and (3) L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 365 that governments are more likely to adopt an NDC-based reform when they anticipate that a shift to NDC will aid them in obtaining loan approval or another desired outcome from the institution. Supranational institutions may influence a country’s pension policy choice in several distinct ways, such as through loan conditionality (granting a loan is made contingent upon adoption of a specified policy), or by encouraging “anticipated reaction” (a country may adopt reforms that it thinks will win favorable action from the supranational institution). Alternatively, national policy elites may use perceived or alleged threats of negative actions by the supranational institution to win support for a policy change from reluctant domestic actors—in effect, making the international institution a “strategic scapegoat” in the reform process.32 Supranational institutions also may act simply as agents of knowledge transfer for “best practices” from other countries. And last, such institutions may serve as a means of harmonization, encouraging member countries to develop common practices to lower regulatory barriers to labor and capital mobility. Several of these channels can be seen in the pension austerity measures taken in European Union member countries. Perhaps most important, in countries such as Italy pension retrenchment was seen as necessary to meet the 3 percent of GDP target set for government deficit as a condition for entry into the European Monetary Union.33 However, these actions by national policy makers largely took the form of “anticipated reactions” and strategic choices designed to win the acquiescence of domestic opponents of painful pension reforms.34 The European Union has also set other requirements (for example, requiring gender neutrality in retirement ages) that have an impact on austerity policy choices and led indirectly to modest policy harmonization. Moreover, the European Union has not even attempted to harmonize most aspects of the disparate pension systems of its member countries, and where it has tried to harmonize policies directly, notably in the area of supplemental pensions, it has had little success. Although some analysts have argued that NDC-based reforms could serve as a vehicle for pension harmonization in the European Union,35 the European Commission itself has not endorsed such an approach. The absence of harmonization pressure within the European Union helps to explain the absence of overall policy convergence on NDC or any other single model for pension restructuring in Western Europe. Despite the importance of the European Union among advanced industrial nations, few institutions have rivaled the global influence of the World Bank in disseminating the ideas and technology of multipillar pension reform. The World Bank’s 1994 report Averting the Old Age Crisis brought unprecedented international attention to the issue of old-age pension reform, fundamentally transforming the way that policy makers conceptualized the issues and solutions to the challenges of old-age income reform. More concretely, the World Bank provided a knowledge base, resources to build institutional and technological capacity, financial support, and in some cases political cover (acting as “bad cop” to domestic politicians’ “good cop”) for governments to adopt some form of privately managed funded pension schemes.36 Indeed, there is little doubt that the World Bank played a powerful role in promoting the adoption of structural DC pension reforms by disseminating policy ideas and technology to government technocrats. Moreover, by associating the shift from DB to DC (both financial and notional) structures with an array of micro- and macroeconomic benefits, World Bank actors generated strong attraction to DC pension reforms among finance ministry and central bank policy makers in disparate corners of the world.37 While the International Labor Organization (ILO) and International Social Security Association (ISSA) disseminated ideas in competition with the ideas of the World Bank, 366 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES the Bank enjoyed important resource advantages over the ILO and ISSA, dedicating vastly greater human and financial resources to the research and dissemination of ideas and policy models based on the DC mechanisms.38 Indeed, following the publication of its 1994 report, the World Bank social protection group launched a multiyear dissemination project, holding workshops and training seminars around the world. It likewise fostered cross-national policy sharing through a series of annual summer workshops that became a clearinghouse for pension reform ideas, bringing together academics, social security experts, and policy makers from around the world to build connections and share ideas about pension reform. The World Bank has over the past five years shown a growing awareness of the shortcomings of FDC pensions in Latin America and elsewhere,39 and has promoted the discussion of a wide range of pension reform models—in particular NDC-based reforms—rather than a single focus on FDC schemes. Indeed, the participation of social security experts from Sweden in these summer workshops opened up the NDC reform ideas to policy makers from far outside the original networks through which these ideas were carried in Europe. More generally, existing research provides little evidence to suggest that the loan conditionality has been the primary means through which international financial institutions have promoted social policy changes, including pension privatization or NDC-based reforms.40 The predominant role of international financial institutions in influencing NDC and other pension reforms is in the agenda-setting process, by transmitting policy ideas and reform models through technocratic networks, rather than by coercing client governments to adopt specific reforms. Policy Feedbacks Once in place, public pension systems have a profound effect on subsequent battles over pension reform. The existing pension system defines not only the winners and losers of structural reform initiatives, but it also delimits the range and intensity of issues over which beneficiaries of the status quo will fight. The structure and performance of the current pension system affects several important aspects of policy choice, including public expectations about the “proper” role of the government and the “fair” level of redistribution in social security; the magnitude of vested interests in the status quo—that is, who stands to gain and lose and how much; and the cost structure and implicit pension debt associated with the current policy as well as the transitional cost of moving from an unfunded to a funded system. In short, policy feedbacks powerfully shape the agenda-setting process, making pension regime change heavily “path dependent.”41 Three policy parameters are prima facie likely to be conducive to both consideration and adoption of an NDC-based pension reform: (1) a strong earnings-related component to the pension system, (2) high implicit pension debt associated with generous replacement rates in a DB scheme, and (3) the existence of complete payroll tax records that allow compilation of lifetime earnings records for calculating benefits under an NDC scheme.42 First, NDC reforms are more likely where the existing system employs an earningsrelated benefit structure (as in a Bismarckian social insurance model), because such a regime encourages public perceptions that a fair pension scheme should allow benefits to bear some relation to contributions. If the contribution-benefit linkage is widely perceived to work poorly (for example, by delivering benefits in a distorted or unfair fashion), however, or if public confidence in the capacity of the public PAYG-DB system to pay benefits in the near and medium term has been very badly shaken, public support for tightening the link between benefits and contributions may be more easily cultivated. The structure of the old pension system also creates important financial constraints on reform options. As noted above, if payroll taxes are already high to finance a public DB L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 367 scheme, market pressures to increase flexibility and lower nonwage costs leave little room to increase contributions to social security. Adding to the constraints of high payroll taxes and well-organized constituents is the heavy implicit pension debt associated with high replacement rates in most traditional DB pension schemes.43 A large implicit pension debt makes the transition cost of a shift toward prefunding (for example, by diverting payroll contributions to individual accounts) very high. This transitional cost can in theory be financed through increased taxes, spending cuts, or issuing new debt, but political and financial constraints may place these options out of reach for many governments. In Bismarckian countries with a very large public earnings–related pension tier, pressures to reduce pension costs and reduce rather than just stabilize pension contribution rates have been especially severe. Bismarckian countries are likely to begin with incremental retrenchment and refinancing measures, but as these have been exhausted, several have turned toward more fundamental restructuring reforms to reduce current and future costs. Because the public PAYG tier was already so large, however, proposals for a mandatory occupational or personal pension individual account tier had to adapt or be “crowded out” by the double payment problem. When expanded mandatory or quasi-mandatory individual account tiers have been adopted in these countries, notably in Sweden and Germany, it has been as a relatively small supplement to a still very large public pension tier that faced severe affordability problems. As table 14.4 suggests, NDC-focused pension reforms initially emerged in countries such as Sweden and Italy that offered generous Bismarckian pensions, for it is in these countries that current and future pension-funding problems were most severe, and where the double payment problem made a DC-based alternative less feasible. In what can be called “Bismarckian Lite” countries—Canada and the United States—pension replacement rates and financing burdens are relatively modest. These countries are likely to be able to maintain their current pension structures with incremental measures for some years into the future, and NDC-based reforms have not reached the agenda. A third policy feedback that might be expected to make consideration and adoption of NDC-based reforms more likely is the existence of complete payroll records that allow retrospective calculation of lifetime earnings to establish NDC “account balances.” In practice, however, this does not appear to be the case. Countries in Eastern and Central Europe that do not have adequate retrospective wage records have developed alternative mechanisms to credit initial capital in NDC accounts.44 Political System Characteristics Specific characteristics of national political environments may also affect both the agendasetting process—that is, whether a particular reform idea develops an institutional carrier that perceives political and policy advantages to a specific pension reform proposal—and the prospects for adoption once that proposal is on the agenda. Understanding the development and diffusion of NDC-based pension reforms must begin with recognition of the opportunities and costs that these reforms impose for key actors in society. Within government, both elected politicians and finance ministries (or the equivalent ministries charged with budgetary and overall economic management) are likely to be involved in pension reform initiatives in most countries—frequently supplanting the social or labor ministries with direct responsibility for running pension programs.45 Elected politicians are driven by the powerful desire to avoid blame for costly policy adjustments.46 Altering public pension systems is especially risky, because even small but well-publicized losses among retirees or those retiring in the near future are especially likely to provoke opposition from the affected group. Moreover, in many countries, the 368 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES elderly are disproportionately likely to vote. For state leaders pressed by demographic and economic change to diminish long-term state pension liabilities, NDC reforms represent a possible way to reduce pension expenditures and commitments while avoiding blame for doing so.47 In the short term, citizens (and politicians) who are unfamiliar with actuarial principles may be confounded by the NDC scheme. The contingency of future pension benefits on demographic and economic trends also makes it difficult for citizens to predict ex ante the precise material implication of NDC shifts. Although this uncertainty could limit initial opposition to NDC, it might also foster exaggerated fears that increase opposition to a shift to NDC. Of course, voters might prefer the type of uncertainty associated with adopting NDC—cuts in benefits are likely to occur in the future, but they will be rule-based and likely to avoid complete system collapse—to the even greater uncertainty associated with future ad hoc cuts in current pension schemes that are highly likely to occur but uncertain in their timing, magnitude, and targeting. But voters are likely to underestimate the likelihood of ad hoc cuts and may exaggerate the magnitude of cutbacks that are known to be automatic but unknown in magnitude. In the longer term, NDC schemes release politicians from responsibility for benefit reductions as demographic and economic profiles change. By adjusting benefits automatically to life expectancy, not only do governments avoid the need to make such changes, but they can escape responsibility for the benefit reductions triggered by the demographic and economic trends, since these lie beyond the immediate command of government. NDC reform thus offers the appeal of a meaningful structural reform from which politicians can walk away with “clean hands.”48 Finance ministry officials have likewise played an important role in the adoption of structural pension reforms. Their attitude toward, and activism in promoting, NDC-based pension reforms is likely to be conditioned by several factors. First, the importance of finance ministries is enhanced where fiscal pressures have been a primary motivator for pension reforms, and also in countries facing strong international financial constraints, either from foreign investors, international financial institutions, or supranational organizations such as the European Union. In Poland, for example, Góra and Rutkowski (2000) argue that finance ministry officials’ demands for macroeconomic discipline in 1991 became an important force moving the country toward structural pension reform. NDC reforms thus become attractive to finance ministry officials to the extent that budgetary pressures from unfunded DB schemes collide with medium-term constraints on government’s ability to finance a privately funded-DC reform. In Brazil, although finance ministry officials favored a structural pension reform in early 1999, central bank officials strongly resisted a mixed DB and funded-DC pension reform on account of the effect of such a reform’s transition cost on short-term macroeconomic performance, which had come under intense international pressure following a currency crisis.49 However, the attitude of finance ministry officials toward NDC pension reform may also be heavily influenced by the overall design of the pension reform package: if, for example, the tighter link between contributions and benefits is accompanied by a requirement that greatly increased general government revenues be used for nonemployment credits (for example, for the care of young children) and for an expanded noncontributory pensions for low earners, then such officials are likely to balk. Political institutions may also affect the prospects for adoption of NDC proposals once they are on the agenda. First, substantial literature suggests that multiple veto points in the policy adoption process can lower both the probability of any change from the policy status quo50 and the scope of any change that is adopted.51 Passing significant pension restructuring reforms should be particularly difficult in countries with large, multiparty L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 369 governments where government leaders need to build majorities across parties occupying a broad range of ideological positions.52 But institutional fragmentation is likely to be less of a barrier in a shift to NDC than in an FDC-focused reform, because the latter is likely to impose greater costs, risks, and redistribution. Indeed, Italy, Latvia, Poland, and Sweden are all proportional representation systems in which a large number of parties held seats in the legislature at the time that an NDC or NDC-oriented reform was passed.53 In each of these countries, such reforms were adopted following intense negotiations across a diverse set of parties. Beyond the compromises embodied within the structure of the NDC scheme, each country also combined distinct pillars of the revised social security system in a way that satisfied demands of diverse coalition partners and organized interests while working within the financial and political constraints posed by the old pension system. Although such broad negotiations may diminish the magnitude and pace of reform, they may have a positive, longer-term consequence. Indeed, Orenstein (2000) finds that although countries with more “veto actors” adopted less radical pension reform, compromises made in the coalition-building process dramatically enhanced the subsequent level of support for and compliance with the pension reform in the implementation phase. When faced with political institutions that spread power broadly, governments have employed a variety of ad hoc mechanisms that can overcome these potential stumbling blocks. Such mechanisms range from technocratic governments with decree powers in Italy to informal cross-party agreements in Germany and formal multiparty working groups in Sweden.54 Clearly the availability of a multiparty working group—which in turn reflected a longer history of cross-party political cooperation—was particularly important in facilitating the Swedish NDC reform. On the other hand, inability to work across parties has clearly paralyzed Social Security reforms in the United States in the post-1983 period. Ultimately, the design of political institutions—whether concentrating power or spreading it broadly across parties and actors in government—obliges reformers to make important trade-offs in the pension reform process. Indeed, although evidence to support a systematic relationship between the concentration of political power and the depth of reform achieved remains mixed, it is clear that in the end, strategic political actors play a powerful role in bringing the issue of pension reform to the policy agenda, and mediating the competing tensions of building a broad coalition to ensure the permanency of the reform while preserving the main policy initiatives sought in the reform process. Policy Design The discussion to this point has emphasized features of the political and social context in which NDC-based pension reforms may reach the policy agenda and win adoption. But the chances of success in adopting reform may also depend on the exact provisions of an NDC reform package—notably who is covered by it, how quickly it is phased in, which groups are afforded special protections, and how much protection is provided by other pension tiers. For example, seniors’ organizations and unions dominated by older workers are more likely to oppose a shift to NDC the faster it is phased in, because faster implementation will have a greater impact on their current membership. It is no accident that pension reform initiatives of all kinds generally “grandfather” current retirees and provide generous transition rules protecting older workers from most cutbacks. NDC schemes offer reforming governments—especially those in fragmented, multiparty political systems—a way to resolve the diverse objectives of coalition partners from opposing ends of the ideological spectrum, and thus may facilitate the creation of legisla- 370 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES tive majorities behind this significant structural reform. For governments pressed by demographic and economic changes to reduce long-term state pension liabilities (in an unfunded DB pension system), the tighter link between contributions and benefits in the DC mechanism presents a key means to achieve this objective. Yet, unlike funded DC schemes, the creation of a large NDC component to mandatory pension systems will not alienate parties that are either skeptical about the privately managed pension schemes or that view a central role for the state in the pension system as the sine qua non of social security. By retaining the PAYG financing structure, and moreover by permitting legislative control over the notional interest rate, NDC reforms may conciliate the demands of parties seeking to protect bureaucratic jobs and the central role for the state in the collection and allocation of benefits. Furthermore, the aspects of NDC reforms that permit the government either to deliver privileges to constituencies such as the uniformed services, or to reward socially valued activities such as child-rearing, higher education, or military service, provides a variety of bargaining tools with which the government can propitiate interest groups or parties that oppose the NDC reform. If reformers are able to tailor NDCinspired reforms to fit local policy and local political and social conditions, they may be able to move their countries from nonadoption to at least partial adoption of NDC reforms. Evaluating Determinants and Explaining Patterns of Reform The diffusion of NDC-based pension reform is clearly a complex, multicausal process. Even the long list of potential influences on adoption or rejection of NDC reform outlined in table 14.3 does not exhaust the list of plausible causal variables influencing the spread of NDC-based reforms.55 Moreover, the patterns shown in table 14.4 do not reveal simple one-to-one relationships between specific independent variables and particular outcomes: causal effects are more aptly characterized as probabilistic and interactive rather than as individually necessary or sufficient. Sweden’s role as NDC innovator illustrates these patterns. Sweden’s serious aging problem as well as a financial crisis and resistance to higher payroll taxes helped put pension reform on the Swedish agenda in the early 1990s, as similar economic and demographic constraints did in many other countries. High technocratic capacity and a fairly even balance (and change in control of government) between Social Democratic and nonsocialist forces, as well as the more idiosyncratic factors of a strong technocratic orientation in the working group planning the pension reform and the exclusion of “social partners” from a direct role in negotiations56 all help to explain why Sweden was the first to adopt framework legislation for NDC-based reform. But Sweden’s relatively open political system, and in particular, grassroots opposition within Sweden’s powerful blue-collar labor confederation and the Social Democratic party explain why it took another four years for final legislation to be adopted rather than achieving a quick adoption. Clearly both regional networks and supranational institutions were important in explaining the diffusion of NDC-based reforms beyond its origins in Sweden to other relatively early innovators such as Latvia and Poland. Both Swedish experts and the World Bank played particularly important roles in spreading these ideas. But as Orenstein (2003, p. 174) suggests, theories that focus on the diffusion of ideas through regional networks are not very helpful in understanding when, where, and how NDC-based reforms will be altered to fit local conditions by lengthening transition periods, excluding specific groups from the reform, or deciding on the relative size of an NDC-based tier within a multipillar pension system. To answer these questions, approaches that focus on policy feedbacks, societal interests, and political conditions are needed. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 371 Policy feedbacks play a particularly important role in advanced industrial countries. Among these countries, NDC-based reforms have spread primarily among countries where Bismarckian social insurance principles of earnings replacement in the public pension system are deeply ingrained rather than those where a flat-rate system (New Zealand) or a mixed public-private system (Switzerland, the United Kingdom) is in place. Younger and less generous Bismarckian Lite pension systems (Canada and the United States) are also less likely to see NDC-based reforms on the agenda, at least in the near future. Policy feedback effects are less clear in the developing and transitional economies of Central and Eastern Europe, however, where NDC-based reforms have been enacted even in the absence of adequate wage records and a strongly earnings-related precursor system. In these countries, regional and supranational influences appear to be more important. And in Africa and East Asia, where public pension coverage rates are generally much lower, NDC-based reforms have barely made a ripple. These patterns of regionally uneven policy feedback effects are likely to change over time: as NDC-based reforms become fully diffused over the next decade or so among pension policy experts worldwide, regional network biases and supranational institution effects in adoption of NDC-based reform may weaken in developing and transitional economies. Implementation Challenges Adopting NDC-based pension reforms can be an important mechanism for restraining pension costs and generating a closer link between individuals’ contributions and benefits, as has been discussed above. But adoption of reform is only half of the battle. As Eric Patashnik (2003) has shown, major policy reforms are subject to erosion or reversal once they are enacted. Developing constituencies who have a stake in the new policy are particularly important to sustaining policy innovation. In the case of NDC-based pension reforms, erosion is probably a greater risk than outright reversal. Indeed, erosion of NDC-based reforms may be attractive to governments because it offers short-term political gains (or avoids short-term political costs), while the effects of those actions on the sustainability of the system may not be immediately obvious. As table 14.5 suggests, this erosion could occur in any of the key attributes that define an NDC-based pension system. For example, governments might choose to impute contributions or rights that have not actually been financed, either for all contributors (as in Italy) or for certain groups (for example, caregivers or university students). In countries where the government has chosen to finance all noncontributory benefits (for example, for caregivers) from the general budget, there may be a temptation to simply impute those credits during periods of fiscal stress. Statistical agencies could also be pressured not to make projections that trigger politically unpopular benefit changes—for example, projections of increased longevity or low long-term economic growth. Although this may not be a problem in countries where statistical agencies are well-established, highly professionalized, and enjoy a high degree of independence, status, and deference, it could certainly be a problem in some developing and transitional countries where none of these things are true. Governments may also be tempted to use NDC contributions to pay some of the costs for pension tiers outside the NDC system that maintain benefit adequacy for selected segments of the elderly population as NDC is phased in, rather than using them to build a buffer fund. Because NDC systems are not completely different from existing DB pensions, but there are in fact “partial” NDC positions (as shown in the fourth column in table 14.1), the big Table 14.5. Potential Erosion in NDC Pension Systems Provision Strong or full NDC Erosion possibilities Conditions facilitating erosion Advanced funding PAYG with inclusion of buffer funds in benefit calculations Government transfers funds from buffer funds to general treasury Government experiences fiscal crisis Life expectancy at retirement Future benefit levels adjust automatically for increases in longevity Government freezes annuitization tables Weak rules in place on regularity and automaticity of life-expectancy adjustments Weak autonomy of statistical agencies Retirement age No standard retirement age with full actuarial adjustment for earlier or later and partial retirement Higher nonactuarial benefits reestablished for workers who have reached a specified age or number of years in employment Strong unions; pensions become electoral issue Inflation and economic growth Benefits tied fully to economic growth Government continues inflation adjustments in annuities and account balances when economy is shrinking (for example, through cap on “brake” mechanism) Economic recession and/or decline in labor force Payroll tax rate Fixed payroll tax rate Increase in payroll tax NDC tier experiences cash-flow crisis resulting from demographic “bulge” Decrease in payroll tax rate Government seeks economic stimulus during recession Transition rules made more generous after initial NDC implementation Transition rules in place create highly visible disparities between adjacent cohorts of retirees Redistribution across generations Barred within NDC tier Redistribution within generations Barred within NDC tier unless financed by payments from government or others with respect to nonemployment activity. Exception: NDC implicitly redistributes from men to women and other population subgroups with longer life expectancy if single annuitization table is used Government imputes NDC credits for nonemployment activity rather than actually making contributions Government experiences fiscal crisis Coverage All workers in specified age cohorts are covered by NDC system Favored groups win exclusion from NDC reform after initial inclusion Favored groups have strong leverage in political system (for example, public sector unions or military) Time horizon All employed workers, including those in labor force at time NDC introduced, covered by NDC system Time horizon for phase-in of NDC pension extended Transition rules in place create highly visible disparities between adjacent cohorts of retirees Exclusivity NDC is only public pension tier New means-tested tier or minimum guarantee created or existing one expanded after implementation of NDC Income inequality or poverty among the elderly increase after implementation of NDC system (may also be pressures for expanding pension guarantee outside NDC tier if pension guarantee is not automatically adjusted for wage growth) Source: Authors. 374 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES challenge of policy makers in NDC systems is likely to be to hold the line against reforms that move from “complete” NDC provisions to middle positions (for example, retaining brakes, but with a cap; temporary increases in contribution rates; infusions of general revenue to keep real benefits from falling; imputing contributions that are purely notional rather than real; and moving from universal coverage of NDC to exclusion of privileged categories of workers). The 1995 Latvian pension reform, for example, went into effect immediately, reducing benefits for a significant portion of workers retiring in 1996 while generating vast disparities in pension benefits.57 The Latvian transition rules, however, were poorly designed and in some cases unfair for persons without work or work in the informal sector.58 This created a popular backlash and placed intense pressure on politicians to revisit the pension reform law in October 1997 and March 1998, when a series of measures were adopted that provided a more generous transition rule for computing initial capital based on old-system records and contribution records during the turbulent 1990s. In practice, this enhanced the generosity of benefits and reinstated some degree of redistribution to the system to compensate for the economic turbulence of the 1990s. In addition, politicians in Latvia voted an extra indexation of benefits for all pensioners in 1997 and 1998, reflecting an effective pensioners lobby.59 In Poland, changes made in 1999 moved new hires in the uniformed services out of the NDC pension system back into a more generous DB system enjoyed by current employees in the uniformed services.60 Several factors may lead to the erosion of NDC-based pension reforms. Most generally, the political party or coalition that instituted the reform may lose power to another political grouping that is hostile to the reform—that may indeed have used opposition to the reform as a key plank in their electoral campaign. Table 14.5 suggests several conditions that may also lead to the erosion of specific components of an NDC-based reform. As noted above, a government fiscal crisis is particularly likely to erode a government’s commitment to make actual contributions in recognition of nonemployment activities (for example, childcare years and unemployment spells). A short-term cash flow crunch as a result of a demographic “bulge” may tempt governments to raise payroll taxes rather than injecting general revenues, even though doing so increases the long-term liabilities of the system. Unrest among the military or a strike threat from public transit workers may lead to their return to a more generous DB system. A decline in the overall labor force and wage bill may leave a government unwilling to cut annuity benefits for retirees and account balances for current workers. In short, as with decisions on adopting a public pension system, economic, demographic, and societal factors are likely to affect policy implementation and final outcomes. Having an NDC-based system in place does shift the bargaining leverage in favor of those who want to put greater financial discipline in the pension system, because preventing the system’s erosion requires them merely to block changes proposed by politicians or groups catering to short-term constituency interests rather than adopting new policies. This advantage is likely to be important (1) in political systems where the governing party or coalition has sufficient agenda control to keep reform-eroding proposals off the agenda, and (2) in systems with multiple veto points, where supermajorities are usually needed to move from the default position. But where agenda control is weak and where veto points are fewer and weaker, temptations for politicians to loosen the lashes will remain strong. The most essential element to sustaining a complete NDC model once it has been adopted is likely to be the maintenance of a broad multiparty agreement in government that NDC is the right thing to do, and broad public understanding of and support for the reform. So far, this agreement within the government appears to be holding in Sweden, the home of the NDC idea. However, public opinion surveys suggest that even after several L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 375 years of multimedia public information campaigns, most Swedish citizens lack basic knowledge of the fundamental precepts (and even the name) of the NDC scheme. As Sundén (2006) argues, because DC systems make it difficult for individuals to anticipate their retirement benefits, it is important for citizens to have a clear understanding of how benefits are determined under the new system and how the new system differs from the former DB scheme if they are to make informed decisions on how much to work and save. From a political standpoint, moreover, improving public information about NDC reforms and their implications for long-term benefits may be a crucial factor in avoiding the backlash experienced in Latvia in 1996 when pension benefits diverged radically from public expectations. Whether the Swedish experience of broad political consensus sustaining an NDC-based pension system can be repeated in countries where traditions of interparty cooperation are weaker, trade unions are less cooperative, and politicians have stronger incentives to respond to independently organized groups of seniors remains to be seen. Although the public information campaign carried out by the Social Insurance Board in Sweden likewise represents a model for many countries, evidence that a majority of Swedes lack basic knowledge of the NDC component of the pension system, and that a majority also perceive that they have significant information needs, should serve as an important warning to politicians about the risk of public backlash in the future if citizens overestimate their old-age income protection and fail to save adequately.61 Moreover, if such information problems exist in small, affluent, and highly educated nations such as Sweden, successful public information campaigns will likely be even more difficult in the developing and transitional countries. Prospects and Potential Problems NDC pension reforms are likely to be an important part of pension reformers’ toolkits in the years ahead. For technocrats, NDC packages combine conceptual elegance and the promise of fiscal discipline that is seen to be lacking in most DB pension plans. For politicians, NDC plans combine an aura of fairness (in relating contributions to benefits) that can be explained to voters, as well as the prospect that those politicians will in the future be spared from dealing with politically painful benefit cutbacks and payroll tax increases. The “automaticity” of NDC is likely to give it continuing appeal to blame-avoiding politicians, especially in countries with social insurance–based systems and moderate debt to GDP ratios. This automaticity also makes NDC appealing to international financial institutions seeking to promote sustainable fiscal policies. A political analysis of NDC-based reforms suggests that there is a central contradiction in the political appeal of NDC-based reforms, however; technocrats are attracted to NDC because in its “complete” form it sends clear signals to workers on the need to work longer and the need to save for retirement in order to obtain an adequate pension. The problem is that transparency on these issues may kill prospects for NDC-based reform, since workers are likely to object to making these changes, especially older workers who have limited time to adjust and blue-collar workers for whom working longer may be more difficult or even impossible. Politicians, on the other hand, are likely to be attracted to NDC for precisely this reason: it can hide the magnitude of future recipient losses vis-à-vis the policy status quo because (1) from the perspective of the “average” citizen, NDC benefit determinations are more opaque than DB plans—although the former may also be less exposed to future political risk; and (2) future cutbacks are contingent on future economic and demographic developments. There are strong incentives for politicians not to be clear about the 376 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES likely effects of an NDC-based reform on individual workers if they hope to succeed in adopting and sustaining that reform. But this in turn may undercut many of the hoped-for effects of NDC-based reforms on retirement and savings behavior, and the political sustainability of the NDC reform itself. Overall, the analysis in this chapter suggests that NDC is no panacea. Six issues in particular are important in considering the prospects for NDC-based reforms. First, there are limitations to the applicability of NDC systems. For countries with the weakest states, lacking the administrative capacity to collect and maintain adequate contribution records on a consistent basis, NDC systems are not likely to be workable. At a minimum, the phase-in period in such countries will have to be very long to develop such a capacity. Moreover, in countries where public pensions have not historically had a close link to earnings (for example, countries with a flat-rate pension), a shift to an NDC is likely to impose substantial losses for some recipients. Again, a long lead time is likely to be necessary to lessen political opposition to such a shift. In countries where NDC reforms are applied retrospectively, conflicts over transition rules for crediting past contributions are likely to be intense, and subject to multiple revisions to respond to the grievances of particular constituencies. Second, there is a danger that adoptions of incomplete and flawed NDC pension plans, as in Italy, may lull politicians into a false sense of complacency, letting them think that they have “solved” their long-term pension problems when they have not. As both politicians and constituencies become more familiar with the dynamics of NDC pensions, the probability grows rather than shrinks that compromises will be built into reform packages that undercut their effectiveness and sustainability. Like a photocopy of a photocopy of a photocopy that still bears the original image in a perceptible but fuzzy way, future NDC pension regimes in some countries may be called NDC, and have many of their elements, but lack fiscal and political sustainability. Third, adoption of NDC-based pension reforms are no panacea for providing political cover for long-term pension retrenchment. It still requires that politicians refrain from the politically easy course of demanding more generous pension benefits for visible groups of constituents, and that they refrain from going along when other politicians make those demands. It also requires that politicians refrain from political interference with buffer funds where they exist. There is no such thing as a “manipulation-proof reform”—pension or otherwise—although some political institutions and policy structures may be more resistant to political risk than others. Fourth, although the novelty of NDC schemes and the contingency of future benefits on economic and demographic factors afford politicians a unique opportunity to rationalize state pensions without confronting strong opposition ex ante, these factors also enhance the need for effective political management of information and guidance of public expectations as to the value of future pension benefits. The Swedish strategy of providing workers with annual statements showing the evolution of notional account balances is emblematic of good public information and expectations management. Campaigns such as these diminish the risk that the government will face a political backlash if benefits under the NDC scheme diverge widely from public expectations. A fifth potential shortcoming of NDC-based pension reform flows directly from the periodic information about balances in individual notional accounts that in theory should be provided to participants in an NDC system. As Daniele Franco pointed out in comments on an earlier draft of this chapter: when NDC is enacted in a flawed or incomplete manner that is likely to require further changes in the future, this information can in fact make those further reforms politically much more difficult. Notional account balance L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 377 statements give workers a much greater sense of “property rights” in the balances in those accounts by making them much more visible. If a future reform, for example, were to eliminate pension rights for contributions that were credited but not actually made, the balances in the accounts would shrink. A major public uproar would likely result over politicians “stealing” their money—a politician’s worst nightmare. Similarly, if expected increases in longevity are not included in statements regularly, irregular adjustments could lead to sudden drops in the benefit flow expected from account balances when they are included. Such changes would lead both to a decline in confidence in the pension system and to political problems for politicians. Finally, while the stabilization of contribution rates remains an important concern, it should be emphasized that NDC pension systems transfer significant new risks to individuals, notably the risk of lower benefits due to poor macroeconomic performance and the risk of being unable to find appropriate (or any) work at an advanced age if workers are expected to stay employed longer to maintain a replacement rate similar to that enjoyed by current retirees.62 The risk of poverty in old age remains a significant policy risk, especially in many developing and transitional nations, and may increase under NDC-based systems over the equivalent risk in what for many countries is the most likely alternative: continued ad hoc adjustments in underfinanced DB schemes. In particular, the reluctance of politicians to make unpopular adjustments to the parameters of pension systems may raise the longer-term risk that the cost of increasing longevity will be borne primarily in the value of pension benefits, which may erode pensions below a socially accepted level.63 Thus the political advantages of NDC reforms in allowing politicians to avoid blame for costly benefit reductions may also give rise to a longer-term social cost if old-age pensions fail to provide sufficient protection against the risk of poverty in aging societies. This risk may be especially high in political systems with multiple veto points where the policy default position is privileged. Thus it is important for governments to balance the goals of stabilizing or reducing payroll taxes with other political and social objectives in designing overall pension reforms. In the Latvian case, for instance, the authors of the Latvia Human Development Report 2000/2001 observe that “[t]he average Latvian pension still remains considerably below the value of the minimum goods and services basket, although the difference between these two indicators is steadily diminishing.”64 Despite the increase in pension values, cost-saving measures passed in 1999 were expected to significantly decrease pension values in 2000 and 2001, more than half of which were already at the minimum value, due to early problems in the design of the transitional rules.65 In Latin America also, given that a large portion of pensions granted are at the minimum level, efforts to control costs through benefit reduction alone should take into consideration issues of adequacy and poverty reduction when considering how to accommodate demographic change. In these cases, raising the retirement age and promoting higher individual savings may offer alternative mechanisms for enhancing the financial stability of public pension schemes without drastically lowering benefits. Increasing the link between pension contributions and benefits will also almost certainly require supplementation of NDC benefits with “social pensions” if they are to provide adequate minimum benefits. This is true in both rich and poor countries, but it is especially true in less-developed countries and transitional economies. Social pensions complementing an NDC-based tier can be structured in several ways: as a universal pension received by all, or as a pension that is tested against income, against income and assets, or (as in Sweden), against only other pension income. It can also be financed in a variety of ways, notably through a separate payroll tax or general revenues. 378 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES Our overall conclusion is thus unsurprising but critical: NDC-based reforms are likely to work best in countries that have the political capacity to achieve and sustain a broad political agreement and the administrative capacity to produce independent forecasts of economic and demographic trends and complete and accurate records of earnings, as well as ensuring compliance and adequate understanding on the part of employers and employees. They are less likely to work well where those capacities are lacking. NDC should not, therefore, be thought of as a simple way to avoid the political dilemmas of pension politics by securely lashing politicians to the mast of an automatically adjusting pension system. Neither the mast nor the lashings are that strong, and the sirens’ call remains very powerful in many countries. Moreover, a shift from DB to NDC-based pensions, although it does lower financial risks over an FDC system and may lower political risks over those in a financially unsustainable DB scheme, is nevertheless likely to increase workers’ fears and uncertainty about their future pension incomes and create new employability risks for older workers. Thus, although NDC-based reforms can be an important part of the pension reform repertoire for both domestically based politicians and expert groups and for transnational actors (including international financial institutions), there is no substitute for a careful analysis of a country’s political and social environment and administrative capacity to determine how such a reform is likely to work on the ground. NDC at best is likely to be the least undesirable of many imperfect alternatives in achieving pension reforms that balance fiscal sustainability, benefit adequacy, and fairness within and across generations. Notes 1. See, for example, Pierson (1998, 2001). 2. See, for example, Brooks (2002) and Madrid (2003). 3. “Notional defined contribution” and “non-financial defined contribution” should be understood to have the same definition. 4. See Palmer (2006a). 5. See Palmer and Góra (2004). While Palmer and Góra (2004) emphasize the financial stability of NDC pension schemes, Kruse (2002) argues that there remains a possibility of financial instability in NDC systems unless there is an automatic balancing mechanism (as in Sweden). 6. In this respect, the Swedish NDC scheme follows the practice of the prereform Swedish earnings-related pension. But buffer funds are by no means unique to Sweden, or to NDC systems. A number of other countries faced with deteriorating demographic situations have developed such funds in their defined benefit pension programs, including the United States, Canada, and (more recently) New Zealand. See Iglesias and Palacios (2001); Palacios (2002); Weaver (2004). 7. See Börsch-Supan and Wilke (2006). 8. Blöndal and Scarpetta (1997, pp. 17–18), in a survey of 18 OECD countries, found that pension contribution rates rose from an average of 9.3 percent in 1967 to 16.5 percent in 1995; the average contribution rate was 1.88 times its 1967 level in 1995. 9. A finding of a future deficit in the CPP’s triennial review process sets in motion a process under which Ministers from Ottawa and the provinces are supposed to agree on any needed changes to keep the plan viable; if they do not agree, contribution rates will increase automatically to meet half of the anticipated deficiency (phased in over three years), and indexation of the CPP will be frozen for the next three years unless cabinet ministers agree to override these procedures. See Slater and Robson (1999, pp. 6–7). 10. See Gern (2002, p. 457). L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 379 11. See Light (1995). 12. The United States invests its reserve fund only in federal securities, however, while the Canada Pension Plan and Quebec Pension Plan invest in a broader range of financial instruments. See Weaver (2004). 13. Indeed, the absence of adequate contribution records in Brazil was a crucial obstacle to the creation of an NDC pension reform in that country. See Pinheiro and Viera (2000). 14. See Góra and Rutkowski (2000). 15. See Fox and Palmer (1999) and Palmer et al. (2006). 16. See Franco and Sartor (2006). 17. The Polish government is easing out certain early retirement privileges by financing “bridge pensions” for uniformed services, wherein the government makes additional contributions to finance a pension paid from early retirement age to the normal retirement age established under the 1997 reform law. See Góra and Rutkowski (2000). 18. A substantial literature on policy diffusion in a variety of sectors among the states within the United States suggests that innovating states—those who first design and implement a reform—are found predominantly among relatively wealthy states with high technocratic capacity and slack resources (see the discussion in Orenstein 2003). The fact that NDC was pioneered in Sweden, a wealthy country with strong technocratic capacity in the pension sector, is consistent with this line of argument. 19. See Orenstein (2000, pp. 12–13). 20. See Orenstein (2000). 21. Tavits (2003) in a review of Estonian pension reforms, categorizes causal variables affecting diffusion of innovation into three categories: internal determinants, external pressures, and lesson-drawing. In Tavits’s categories, economic/demographic, policy feedbacks, partisan/political constraints, and societal constraints would all fit within the category of internal determinants, while ideational forces would be divided into Tavits’s internal (strength of conservative forces), external and lesson-drawing (supranational institutions), and lesson-drawing (regional networks) categories. 22. See Kinsella and Velkoff (2001) and European Union, Economic Policy Committee (2001). 23. See Brooks (2002, pp. 513–5). 24. See Swank and Steinmo (2002), Swank (1998), and Garrett (1998). 25. See Kingdon (1995). 26. See Lundberg (2003). 27. See Bennett (1991), Haas (1992), and Most and Starr (1980). 28. Prominent research suggests that individuals are, in essence, “cognitive misers” who, due to their finite information-processing abilities, look to cues, heuristics, or information shortcuts to make decisions more tractable (see Tversky and Kahneman [1974], Kahneman, Slovic, and Tversky [1982]). Brooks (2005) has argued that competitiveness and status concerns were significant influences in the adoption of private pension reforms in developing countries seeking to attract foreign investment, while Hering (2004) has examined the role of such pressures in European nations seeking accession to the European Union. 29. See Hering (2004). 30. See Brooks (2005) and Madrid (2003). 31. See Keohane and Nye (1974), Finnemore (1993), Krasner (1982,1985); see also Nelson (2004) and Weyland (2004). 32. See Vreeland (1999). 33. See Schludi (2002). 34. See Franco and Sartor (2006). 380 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES 35. See Holzmann (2006). 36. Brooks (1998) found that in Argentina, to gain leverage over domestic opponents to an FDC pension reform proposal, President Menem requested that the IMF place a condition in its Letter of Intent of its March 1992 Extended Fund Facility Agreement requiring the Argentine government to pass a structural pension reform by January, 1993. For the role of international financial institutions as strategic scapegoats, see also Vreeland (1999). 37. For the attraction argument, see Brooks (2005). Among the transition countries, Bulgaria, Hungary, and Latvia received considerable technical and financial support from the World Bank through conditional loans, but only Latvia adopted the NDC model. Nevertheless, the attraction to DC structures among Eastern European and Central Asian countries is striking, with FDC reforms adopted in Estonia, Kazakhstan, Lithuania, and (in addition to NDC) in Poland, while NDC was adopted with the later addition of an FDC tier in Latvia. Whether these governments chose financial or notional DC schemes may be understood to have been shaped by a combination of financing constraints, explained above, and political dynamics, discussed below. 38. See Brooks (2005) and Orenstein (2003). 39. See in particular Gill, Packard, and Yermo (2004). 40. See Brooks (2004a) and Hunter and Brown (2000). In Uruguay, loan conditionality also was not a relevant means of policy diffusion. The Inter-American Development Bank (IDB) played a more important role in transferring ideas and technology of pension reform than in promoting any specific policy model. The research that led up to the (eventually rejected) 1992 NDC reform proposal in Uruguay was financed by a loan from the IDB and based on technical advice from a team of Latin American pension experts, including a Brazilian scholar who later participated in the design of the 1999 pension reform in Brazil that contains some elements similar to NDC. 41. See Pierson (1994, 2000). 42. Although sufficient, the existence of complete payroll tax records is unlikely to be a necessary condition. 43. See James (1998). 44. See Palmer (2006b), Chloń-Domi ńczak and Góra (2006), and Palmer et al. (2006). 45. See Orenstein (2000). 46. See Pierson and Weaver (1993). 47. Early evidence suggests that the adoption and diffusion of NDC and FDC pension reform models differ systematically in this respect, with governments being more likely to adopt NDC reforms where political institutions share power broadly across parties represented in government, and thus allow reformers to avoid direct blame for costs that become apparent after adoption (Brooks 2004b). 48. The need to “walk away with clean hands” is one problem that concerns more than simply pragmatic considerations of the likely backlash that elected politicians might incur. It reaches to a core dilemma of public ethics that suggests that in order to do the “right thing,” politicians at times must do what citizens would otherwise oppose, such as automatic reductions in pension benefits when pension system revenue declines. For the metaethical discussion of the politics of “dirty hands,” see Walzer (1973) and Parrish (2002). 49. See Brooks (2004a, p. 75). 50. See Immergut (1992) and Tsebelis (1999). 51. Other aspects of political institutions may also affect capacity for policy change. For example, countries that have relatively short electoral cycles may find it particularly difficult to make changes that impose visible losses on retirees and those approaching retirement. See the discussions in Pal and Weaver (2003) and Bonoli (2000). L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 381 52. See for example Immergut (1992), Kay (1998), and Brooks (2002). The advantages of concentrated power and minimal veto points may, however, be at least partially offset by the concentration of accountability in political systems: because voters know that it is the governing party that is imposing losses, governing parties may be reluctant to undertake initiatives that are very likely to incur retribution at the next election (Pierson and Weaver 1993). Because future losses may be obscured ex ante in NDC reforms, accountability concerns may be partly allayed. Moreover, even governing parties with very strong formal powers may refrain from enacting policies that are likely to be reversed by a later government. 53. Italy adoped a mixed-member majoritarian (MMM) electoral system in 1993. The new electoral system “did not reduce party fragmentation, but it did provide powerful incentives for parties to enter into pre-electoral coalitions.” (D’Alimonte 2001, p. 323). 54. See Schludi (2002, chapter 9). 55. For example, in an analysis of the determinants of differences between pension reform in Estonia and Latvia, Tavits (2003) argues that a record of successful domestically initiated economic reform outside the pension sector in Estonia gave Estonian political elites greater confidence that they could successfully design and implement a homegrown pension reform, while a weaker Latvian record in this regard made them more inclined to borrow heavily from the Swedish NDC model. An extension of this argument would be that in countries where previous policy borrowing from supranational organizations or through regional networks in other sectors has been judged by political elites to be successful, they may be inclined to do it again in the pension sector. 56. See Lundberg (2003) and Lindbom (2001). 57. Bite and Zagorskis (2003, p. 41) report that pensions calculated in January 1997 ranged from as low as 8 lats to over 1,000 lats, which was “shocking to the society” that was accustomed to the traditional equalizing role of social security. In fact, no one actually received such a low pension due to the system’s guarantee rule. The high pensions were not the result of NDC rules, but to a generous conversion of special privileges acquired under the old system, such as for Latvians forced into exile in Siberia for a large part of their lives. See Palmer et al. (2006). 58. See Palmer et al. (2006). 59. See Palmer et al. (2006). 60. See Chloń-Domi ńczak and Góra (2006). 61. See Sundén (2006). 62. See Scherman (2003). 63. Diamond argues that while this diminishes the risk of excessive legislative intervention, the solvency promoted by automatic benefit reductions may enhance the risk that governments intervene too infrequently to balance the demographic adjustment to the contribution formula as well; see Diamond (2000, p. 86). 64. See Bite and Zagorskis (2003, p. 63). 65. See Bite and Zagorskis (2003, p. 64). References Bennett, C. J. 1991. “What Is Policy Convergence and What Causes It?” British Journal of Political Science 21 (April): 215–33. Bite, I., and V. Zagorskis. 2003. Latvia Country Study, Study on the Social Protection Systems in the 13 Applicant Countries, carried out for the Commission by GVG, http://europa.eu.int/comm/employment_social/index_en.htm. 382 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES Blöndal, S., and S. Scarpetta. 1997. “Early Retirement in OECD Countries: The Role of Social Security Systems,” OECD Economic Studies 29: 7–53. Bonoli, G. 2000. The Politics of Pension Reform. Cambridge: Cambridge University Press. Börsch-Supan, A., and C. B. Wilke. 2006. “The German Public Pension System: How It Will Become an NDC System Look-Alike.” In Pension Reform: Issues and Prospects for NonFinancial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 22. Washington, DC: World Bank. Brooks, S. 1998. “Political Dynamics of Pension Reform in Argentina.” Paper presented at the International Meeting of the Latin American Studies Association, Chicago, Illinois, September 25. ———. 2002. “Social Protection and Economic Integration: The Politics of Pension Reform in an Era of Capital Mobility,” Comparative Political Studies 35 (5): 491–523. ———. 2004a. “International Financial Institutions and the Diffusion of Foreign Models for Social Security Reform in Latin America.” Learning from Foreign Models in Latin American Policy Reform, ed. K. Weyland, 53–80. Washington, DC: Woodrow Wilson Center Press and Johns Hopkins University Press. ———. 2004b. “A Competing Risks Model of Structural Pension Reform,” Paper presented at the annual meeting of the American Political Science Association, Chicago, Illinois, September 2–5. ———. 2005. “Interdependent and Domestic Sources of Policy Change: The Diffusion of Pension Privatization around the World.” International Studies Quarterly 49 (2): 273–94. Chloń-Domi ńczak, A., and M. Góra. 2006. “The NDC System in Poland: Assessment after Five Years.” In Pension Reform: Issues and Prospects for Non-Financial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 16. Washington, DC: World Bank. D’Alimonte, R. 2001. “Mixed Electoral Rules, Partisan Realignment, and Party System Change in Italy.” In Mixed-Member Electoral Systems: The Best of Both Worlds? ed. M. Soberg Shugart and M. P. Wattenberg, 323–50. New York: Oxford University Press. Diamond, P. 2000. Social Security Reform. Oxford: Oxford University Press. Elster, J. 1979. Ulysses and the Sirens. Cambridge, U.K.: Cambridge University Press. ———. 2000. Ulysses Unbound: Studies in Rationality, Precommitment, and Constraints. Cambridge, U.K.: Cambridge University Press. European Union, Economic Policy Committee. 2001. Budgetary Challenges Posed by Ageing Populations. Brussels: EPC/ECFIN/655-EN-fin, October 24. Finnemore, M. 1993. “International Organizations as Teachers of Norms.” International Organization 47 (4): 565–99. Fox, L., and E. Palmer. 1999. “Latvian Pension Reform.” Social Protection Discussion Paper No. 9922, World Bank, Washington, DC. Franco, D., and N. Sartor. 2006. “NDC in Italy: Unsatisfactory Present, Uncertain Future.” In Pension Reform: Issues and Prospects for Non-Financial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 18. Washington, DC: World Bank. Garrett, G.. 1998. Partisan Politics in the Global Economy. Cambridge, U.K.: Cambridge University Press. Gern, K.-J. 2002. “Recent Developments in Old Age Pension Systems: An International Overview.” In Social Security Pension Reform in Europe, ed. M. Feldstein and H. Siebert, 439–78. Chicago: University of Chicago Press. Gill, I. S., T. Packard, and J. Yermo. 2004. Keeping the Promise of Old Age Income Security in Latin America: A Regional Study of Social Security Reform. Washington, DC: World Bank. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 383 Góra, M., and M. Rutkowski. 2000. “The Quest for Pension Reform: Poland’s Security through Diversity.” Working Paper 286, Washington, DC, World Bank. Haas, P. M.1992. “Introduction: Epistemic Communities and International Policy Coordination.” International Organization 46 (1): 1–35. Hering, M. 2004. “Rough Transition: Institutional Change in Germany’s ‘Frozen’ Welfare State.” Ph.D. dissertation, Johns Hopkins University, Baltimore. Holzmann, R. 2006. “Toward a Coordinated Pension System in Europe: Rationale and Potential Structure.” In Pension Reform: Issues and Prospects for Non-Financial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 11. Washington, DC: World Bank. Hunter, W., and D. S. Brown. 2000. “World Bank Directives, Domestic Interests, and the Politics of Human Capital Investment in Latin America.” Comparative Political Studies 33 (1): 113–43. Iglesias, A., and R. J. Palacios. 2001. “Managing Public Pension Reserves: Evidence from the International Experience.” In New Ideas About Old Age Security, ed. R. Holzmann and R. Stiglitz, 213–53. Washington, DC: World Bank. Immergut, E. 1992. “The Rules of the Game: The Logic of Health Policy-Making in France, Switzerland and Sweden.” In Structuring Politics: Historical Institutionalism in Comparative Perspective, ed. K. Thelen and S. Steinmo, 57–89. New York: Cambridge University Press. James, E. 1998. “The Political Economy of Social Security Reform: A Cross-Country Review.” Annals of Public and Cooperative Economics 69 (4). Kahneman, D., P. Slovic, and A. Tversky. 1982. Judgment under Uncertainty: Heuristics and Biases. New York: Cambridge University Press. Kay, S. 1998. “Politics and Social Security Reform in the Southern Cone and Brazil.” Ph.D. dissertation, Department of Political Science, University of California, Los Angeles. Keohane, R. O., and J. Nye. 1974. “Transgovernmental Relations and International Organizations.” World Politics 27 (1): 39–62. Kingdon, J. 1995. Agendas, Alternatives, and Public Policies, second edition. New York: HarperCollins. Kinsella, K., and V. A. Velkoff. 2001. An Aging World: 2001. U.S. Census Bureau, Series P95/01-1. Washington, DC: U.S. Government Printing Office. Krasner, S. 1985. Structural Conflict: The Third World Against Global Liberalism. Berkeley, CA: University of California Press. Krasner, S. 1982. International Regimes. Ithaca, NY: Cornell University Press. Kruse, A. 2002. “Ageing Populations and Intergenerational Risk-Sharing in PAYG Pension Schemes.” Working Paper 18, Department of Economics, Lund University, Sweden. Available at www.nek.lu.se/publications. Light, P. 1995. Still Artful Work: The Continuing Politics of Social Security Reform, New York: McGraw-Hill. Lindbom, A. 2001. “De borgerliga partierna och pensionsreformen.” In Hur blev den stora kompromissn mojlig. Politiken bakom den svenska pensionsreformen, ed. Joakim Palme, 50–87. Stockholm: Pensionsforum. Lundberg, U. 2003. Juvelen I kronan: Social demokraterna och den allmänna pensionen. Stockholm: Hjalmarson and Högberg. Madrid, R. 2003. Retiring the State: The Politics of Pension Privatization in Latin America and Beyond. Stanford: Stanford University Press. Most, B. A., and H. Starr. 1980. “Diffusion, Reinforcement, Geopolitics, and the Spread of War.” The American Political Science Review 74(4): 932–46. 384 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES Nelson, J. M. 2004. “External Models, International Influence, and the Politics of Social Sector Reforms.”In Learning from Foreign Models in Latin American Policy Reform, ed. K. Weyland, 35–52. Washington, DC: Woodrow Wilson Center Press and Johns Hopkins University Press. Orenstein, M. A. 2000. “How Politics and Institutions Affect Pension Reform in Three Postcommunist Countries.” Policy Research Working Paper 2310, World Bank, Washington, DC. ———. 2003. “Mapping the Diffusion of Pension Innovation.” In Pension Reform in Europe: Process and Progress, ed. Robert Holzmann, Mitchell Orenstein, and Michal Rutkowski, 171–93. Washington, DC: World Bank. Pal, L., and R. K. Weaver. 2003. The Government Taketh Away: The Politics of Pain in the United States and Canada. Washington, DC: Georgetown University Press. Palacios, R. J. 2002. “Managing Public Pension Reserves Part II: Lessons from Five Recent OECD Initiatives.” Social Protection Discussion Paper 0219, World Bank, Washington, DC. Palmer, E. 2006a. “What Is NDC?” In Pension Reform: Issues and Prospects for Non-Financial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 2. Washington, DC: World Bank. Palmer, E. 2006b. “Conversion to NDCs—Issues and Models.” In Pension Reform: Issues and Prospects for Non-Financial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 9. Washington, DC: World Bank. Palmer, E., and M. Góra. 2004. “Shifting Perspectives in Pensions.” Discussion Paper 1369, Institute for Labor Studies, Bonn. Palmer, E., S. Stabina, I. Svensson, and I. Vanovska. 2006. “NDC Strategy in Latvia: Implementation and Prospects for the Future.” In Pension Reform: Issues and Prospects for Non-Financial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 15. Washington, DC: World Bank. Parrish, J. M. 2002. From Dirty Hands to the Invisible Hand: Paradoxes of Political Ethics. Ph.D. Dissertation, Department of Government, Harvard University. Patashnik, E. 2003. “After the Public Interest Prevails: The Political Sustainability of Policy Reform.” Governance (16) 2: 203–34. Pierson, P. D. 1994. Dismantling the Welfare State. Cambridge: Cambridge University Press. ———. 1998. “Irresistible Forces, Immovable Objects: Post-Industrial Welfare States Confronting Permanent Austerity.” Journal of European Public Policy 5 (4): 539–60. ———. 2000. “Increasing Returns, Path Dependence, and the Study of Politics.” American Political Science Review 94 (2): 251–67. ———. 2001. The New Politics of the Welfare State. Oxford: Oxford University Press. Pierson, P. D., and R. K. Weaver. 1993. “Imposing Losses in Pension Policy.” In Do Institutions Matter? ed. R. Kent Weaver and Bert A. Rockman, 110–50. Washington, DC: The Brookings Institution. Pinheiro, V., and S. Viera. 2000. “Reforma previsional en Brasil: La nueva regla para el cálculo de los beneficios.” Serie Financiamiento del Desarrolo No. 97, CEPAL, Santiago de Chile. Pinheiro, V. 2004. “The Politics of Social Security Reform in Brazil.” In Learning from Foreign Models in Latin American Policy Reform, ed. Kurt Weyland, 110–38. Washington, DC: Woodrow Wilson Center Press and Johns Hopkins University Press. Scherman, K.G. 2003. “Old Age Security in Sweden, Quo Vadis?” Paper delivered at the Forschungsnetzwerkes Alterssichung, Erfurt, December 4–5. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 385 Schludi, M. 2002. “The Reform of Bismarckian Pension Systems.” Ph.D. dissertation, Humbold University, Berlin. Slater, D. W., and W. B. P. Robson. 1999. Building a Stronger Pillar: The Changing Shape of the Canada Pension Plan. Toronto: C.D. Howe Research Institute. Sundén, A. 2006. “How Much Do People Need to Know about Their Pensions and What Do They Know?” In Pension Reform: Issues and Prospects for Non-Financial Defined Contribution (NDC) Schemes, ed. R. Holzmann and E. Palmer, chapter 13. Washington, DC: World Bank. Swank, D. 1998. “Funding the Welfare State: Globalization and the Taxation of Business in Advanced Market Economies” Political Studies 46 (4): 671–92. Swank, D., and S. Steinmo. 2002. “The New Political Economy of Taxation in Advanced Capitalist Democracies.” American Journal of Political Science 46 (3): 477–89. Sweden, National Social Insurance Board. 2004. The Swedish Pension System Annual Report 2003, Stockholm: Riksförsäkringsverket (National Social Insurance Board). Tavits, M. 2003. “Policy Learning and Uncertainty: The Case of Pension Reform in Estonia and Latvia.” Policy Studies Journal 31 (4): 643–60. Tsebelis, G. 1999. “Veto Players and Law Production in Parliamentary Democracies: An Empirical Analysis.” American Political Science Review 93 (3): 591–608. Tversky, A., and D. Kahneman. 1974. “Judgement Under Uncertainty: Heuristics and Biases.” Science 185: 1124–31. Vreeland, J. 1999. “The IMF: Lender of Last Resort or Scapegoat?” Midwest Political Science Association Annual Meeting, Chicago, IL, April 15–17. Walzer, M. 1973. “Political Action: The Problem of Dirty Hands.” Philosophy and Public Affairs 2 (2): 160–80. Weaver, R. K. 1986. “The Politics of Blame Avoidance.” Journal of Public Policy 6 (4): 371–98. ———. 1988. Automatic Government: The Politics of Indexation. Washington, DC: The Brookings Institution. ———. 2004. “Whose Money Is It Anyway? Governance and Social Investment in Collective Investment Funds.” In Rethinking the Welfare State: The Political Economy of Pension Reform, ed. M. Rein and W. Schmähl, 294–316. Cheltenham: Edward Elgar. Weyland, K. 2004. “Conclusion: Lessons About Learning in Latin American Policy Reform.” In Learning from Foreign Models in Latin American Policy Reform, ed. K. Weyland, 241–83. Washington, DC: Woodrow Wilson Center Press and Johns Hopkins University Press. Discussion of “Lashed to the Mast? The Politics of NDC Pension Reform” Daniele Franco* THE CHAPTER BY SARAH BROOKS AND KENT WEAVER PROVIDES a broad overview of the economic and political factors underlying the introduction of NDC pensions. The analysis is based on an extensive survey of policy developments at the country level and of the relevant economic literature. I share most of the views and the conclusions of the authors. I will focus my comments on three aspects that may require some integration of the analysis of the authors. First, the authors note that the shift from a traditional defined benefit (DB) system to a non-financial defined contribution (NDC) system can diminish the visibility of benefit cuts: the NDC system is a new system, it is complex. As the authors note, it can even appear opaque. Benefits are contingent on several possible events. This may imply, in particular, that the introduction of the NDC system can determine benefits cuts that are not noticed by citizens. These considerations are surely reasonable, although they do not apply to the case of Italy, where the reform was not exploited to cut future spending. However, this lack of visibility may disappear once the NDC system is introduced. In fact, the NDC framework makes the pay-as-you-go (PAYG) liabilities more explicit than they were before: each citizen is informed about her or his pension wealth. Therefore, NDC liabilities can be perceived as a harder form of public debt than the non-explicitly specified liabilities of more traditional PAYG schemes. This development is probably not problematic if the parameters of the system are sustainable and if corrective mechanisms keep the system in balance, which is basically the case in Sweden. It may be problematic, however, if the design of the system is incomplete or the system involves spending and revenue levels that may not be sustainable in an open economy. This is probably the case in Italy. Defaulting the PAYG debt via further pension reforms becomes more visible and politically more difficult because individuals are more aware of their pension wealth than in a traditional PAYG scheme. Moreover, financial markets can perceive a higher default risk. This higher risk may be particularly problematic for emerging market economies in which a relatively small public debt can sometimes be difficult to manage. These considerations may suggest some additional caution in advocating the introduction of NDC systems in these countries: revenue levels should be clearly sustainable and corrective mechanisms should be very effective. * Daniele Franco is director of the Public Finance Division in the Research Department of the Bank of Italy. The opinions expressed are those of the author and do not involve the responsibility of the Banca d’Italia. 386 L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 387 Second, the chapter examines the economic and political factors underlying the introduction of NDC systems in some countries. Some of the authors’ preliminary conclusions do not seem convincing. The authors attribute a role to international economic institutions in spreading NDC systems. There is no evidence in the cases of Italy and Sweden that these institutions did have a role. Surely the European Commission did not advocate the introduction of NDC pension systems. The role of regional networks should also not be overestimated. Italy and Sweden were not influenced by neighboring countries. The introduction of the NDC system in Brazil does not seem to have been influenced by the debate that has taken place in Uruguay. Rather, the Brazilian discussion was initiated following contacts with Swedish reformers. Some caution is also warranted in drawing general conclusions about the role of economic factors and political factors. As to the role of overall budgetary conditions, Sweden, for instance, introduced the NDC system in a period in which its public finances did not face serious immediate fiscal problems. However, when the Swedish reform was formulated and legislated in 1992–1994, the impact of the recession of the early 1990s was still vivid. In Sweden, the recession had been substantially more severe than in the rest of Europe. Sweden was confronted with a serious crisis in the financial market and there was considerable discussion about long-term structural problems that would give rise to increasing debt, the most apparent structural issue being the pension system. The Swedish pension system reformers benefited greatly from this general feeling of crisis among politicians, institutional players, and the average person on the street. In Italy, the NDC system was introduced in a period characterized by continuous efforts to reduce the budget deficit. However, given the long transition phase envisaged for the introduction of the NDC system in Italy, the reform was not expected to contribute to fiscal consolidation over the short to medium term. Moreover, even in the long term it may not reduce pension spending more than under the previous set of pension rules. As to the role of new-liberal ideologies, in Italy the NDC system was introduced by a center-left government. In Sweden it was implemented by a Social Democratic government— following the consensus between conservatives, liberals, and social democrats reached in 1994; all the parties supported the reform as it went through parliament in June 1994. In sum, we do not yet have systematic indications concerning the factors underlying the introduction of NDC regimes in some countries and not in others. There are just too few available case studies. Third, the authors stress that one of the greatest advantages of NDC pensions is the perception of permanency that they provide. The system is designed to last forever. But they also note that there is “no such thing as a manipulation-proof reform” and they suggest that there is a risk of erosion of NDC rules. This discussion may remind us of the debate about fiscal rules and discretionary fiscal policy. Fiscal rules are said to make policies more time consistent and to reduce the room for short-term political exploitation of fiscal policy. Basically, one does not trust governments to always take the best discretionary decisions. However, the evidence on the effectiveness of fiscal rules at the national level is very much mixed.1 Democracies keep changing fiscal rules for a variety of reasons. Basically, governments and parliaments feel entitled to change whatever was decided previously. So, even if pension schemes are well designed and corrective mechanisms keep them in balance, we cannot expect too much in terms of the long-term durability of pension rules. This should not discourage pension experts from trying to improve the design of schemes, but it should remind them that pension schemes are also a tool in the political debate. 388 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES Note 1. See Kopits and Symansky (1998) and Banca d’Italia (2001). References Banca d’Italia, ed. 2001. Fiscal Rules. Rome, Banca d’Italia. Kopits, G., and S. Symansky. 1998. “Fiscal Policy Rules.” IMF Occasional Paper 162, International Monetary Fund, Washington, DC. Discussion of “Lashed to the Mast? The Politics of NDC Pension Reform” Agneta Kruse* SWEDEN WAS THE FIRST COUNTRY TO ADOPT A PENSION SYSTEM designed as a notional defined contribution system, that is, an unfunded (pay-as-you-go) defined contribution system, what has become known as a non-financial defined contribution (NDC). The old system, a defined benefit, price-indexed pay-as-you-go system, turned out to be both unsustainable and unfair. However, according to public choice theory—at least in its crudest form (majority rule/median voter)—a reform such as the Swedish one should not be possible. And still, there it is! I discuss this contradiction in Kruse (2003) without reaching any definite explanation or conclusion. So, it is a pleasure for me to comment on Sarah Brooks and Kent Weaver’s chapter, in which they tackle the problem from a different angle. The focus of their chapter is on reasons for adopting such a scheme and whether, once adopted, the politicians will be “lashed to the mast.” As a background, the chapter starts with a lengthy and informative description of different ways of organizing pension systems with a focus on an NDC system. The description emphasizes that it is possible to choose a system that is not a clear-cut NDC. They point out that many systems currently in use mix features from NDB (defined benefit) and NDC systems. This is an important observation and points to the fact that all reforms are path dependent; thus actual systems turn out to be mixtures. In an informative table, the NDC pension provisions are presented as a continuum. The categories presented are then used to fit in countries with different reforms. The description is ambitious as it tries to cover all possible combinations of features. The authors want to discuss pros and cons of an NDC system in order to pin down the reasons that politicians should choose such a system. For such a purpose the description occasionally becomes confusing as it is sometimes unclear about whether it is an NDC system that is being described or some specific features of an NDC system in a certain country. It is unclear whether the merits or drawbacks of NDC depend on NDC per se or the particular design (mixture) chosen in a specific country. Brooks and Weaver maintain that pension reforms are as easy to do in defined benefit (DB) systems as in an NDC system. This might be true, and there is a list of reforms done in DB systems. However, also true is that even today there is a lack of insight among workers and pensioners (voters) in many countries that have DB systems about what reforms are needed in order to get a sustainable system. The investigation by Boeri, Börsch-Supan, and Tabellini (2001) is witness to this lack, and last but not least the strikes last year in France and the mass demonstrations in Spain. More important, though, is that even if there are a lot of (limited) reforms, the question is whether that is such a good thing. Brooks and Weaver do not discuss the uncertainty these reforms bring about or the * Agneta Kruse is senior lecturer at the Department of Economics, Lund University. 389 390 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES increased planning problems it causes to people trying to maximize their lifetime utility. The merits of a stable system as a risk-reducing device are obvious, but this is not mentioned by Brooks and Weaver. This omission is probably the reason why they conclude that everything that can be achieved in an NDC system can also be achieved in a DB system. In the description of different reform options, one aspect of the difference between a DB system and a defined contribution (DC) system is missing in Brooks and Weaver’s chapter. When it comes to risk sharing between generations in the face of demographic and economic changes, there are important differences between the systems. In DB systems, all adaptation is borne by the working generation. NDC systems with their fixed contribution rates and growth index make co-living generations share the burden of lean years and the fruits of good years (see Kruse [2002] and references therein for a discussion of risk sharing between generations). Brooks and Weaver do discuss how the two systems differ when it comes to intergenerational distribution. They describe the intergenerational distribution in the following way: in a DC system “. . . benefits adjusts automatically to return financial balance to the system.” In a DB system, the benefit rate is fixed as a social objective according to Brooks and Weaver. I have no objections to this. However, what is not stressed in the chapter is that when the benefits are reduced in a DC system, it is caused by bad economic times and low growth. In such a situation, the working part of the population also gets a reduction in consumption possibilities. In a DB system, this decrease will be magnified by increases in the contribution rate. As the focus of the chapter is on the benefit side of the system, it is rather hidden that in a DB system the bill is passed on to the younger generation, sometimes even to generations not yet born, let alone not having a vote in a democratic process. Having such a strong focus on the benefit side also means that the important life-cycle aspects of a pension system are missed. When it comes to intragenerational redistribution, Brooks and Weaver find that DB systems have a “more complex pattern of redistribution.” Somehow it is often taken for granted that this means redistribution in favor of low-income groups. This seems to be the case in the U.S. system. However, very often it turns out to go in the opposite direction. In DB systems the benefit often depends on the last year’s wage or an average of the last X (X = 3, 5, or 10) years’ wages, transferring resources from flat career, blue-collar worker to white-collar worker in career jobs. This is true in many European countries, it was true in the old Swedish system, and this is why a DC system where such “perverse” redistribution does not take place may win political support. The analysis would be improved with a life cycle approach to pension systems. With such an approach the fact that everybody will eventually become a pensioner is evident; also evident is that the aim of a pension system is to help the individual even out income and consumption possibilities over the life cycle (plus risk sharing within and between generations). The chapter gives the impression that a reform is of interest only to the elderly and not to the younger generations who are still in active ages. The life cycle approach brings forward the individual’s planning problem and considers how a pension system handles risk. There are different kinds of risk—economic, demographic, and political1—affecting the rate of return the system gives. How these risks are met differ with the design of the chosen system. These risks can never be totally eliminated, but they are heavily reduced in an NDC system. However, the individual, the insured person, does not exist in the Brooks and Weaver chapter. Instead the focus is on public bodies. These are used as variables in the section on political viability of reforms. Although I miss the perspective of the individual, I find this perspective innovative and interesting. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 391 A main conclusion from the description seems to be that whatever is achievable in an NDC system is also achievable in an NDB system. If I understand it correctly, the only reason for choosing an NDC system is that the politicians than can get rid of the responsibility for retrenching the benefits. Political Viability of an NDC Reform The questions here are why the NDC gets on the agenda in some countries and not in others, and why, once on the agenda, NDC is adopted in some countries but not in others. Case studies for Brazil, Italy, Sweden, and Uruguay are used to highlight the questions. The first question—why reform gets on the agenda—is sort of an odd question in the Swedish case. The NDC system was actually invented in Sweden. On the agenda was what to do about an unsustainable system with perverse redistribution (from low-income, bluecollar workers to high-income, white-collar workers). The reform was preceded by a number of investigations by parliamentary committees. Slowly an awareness of the necessity of a reform arose. In scrutinizing the problems, a reform proposal took form, a design that years later (in 1997) came to be named NDC. Of course there have now and then been discussions and proposals in specific areas that mimics parts of the NDC system, but not a coherent proposal like the Swedish one. The description in the chapter gives the impression that there is a basket of ready-made reform options from which one can be drawn. Be that as it may, in order to answer the questions a great number of interesting hypotheses are presented. The first hypothesis is that NDC is more likely to be introduced in a country the more that country’s population has aged. The second hypothesis is that NDC is most likely in countries with severe governmental fiscal pressure and a high public debt in relation to GDP. A third one is that NDC is most likely in countries where the payroll taxes are so high that they limit economic and political sustainability (my italics).2 Also, the strength of various domestic ideological groups, plus the interaction with supranational institutions such as the European Union and the World Bank, are assumed to be influential. The hypotheses are presented and summarized in a table. The authors then proceed and give examples from countries and reforms where specific factors seem to have been at hand. The analytical approach chosen is a difficult one, as the examples often open up possibilities of counterexamples. For example, the description of the World Bank’s efforts to convince countries of the advantages of NDC must build on a misunderstanding. The 1994 World Bank publication Averting the Old Age Crisis did not advocate a NDC system. On the contrary, it took quite a few years to convince people at the World Bank (if this was ever done!) of the merits of a NDC system. So far the chapter gives only tentative answers to the hypotheses posed. A problem is, however, that the authors do not indicate what a more systematic analysis would look like. A full-fledged formal testing of the hypotheses is probably beyond the scope of the chapter. As the variables listed are quantifiable, a step in the right direction would be to have them presented for a large number of countries. A table showing population structure plus projections, public debt in relation to GDP, and perhaps also the level of payroll taxes and whether the country has adopted a pension reform and if so what kind of reform, would give a good opportunity to judge the relevance of the hypotheses. For an economist it seems odd that, when discussing the politics of pension reforms, voters are nonexistent. There is a very rich literature discussing the possibilities of pension reforms in democracies with majority voting, showing among other things how the result differs with the assumptions made. The literature contains both theoretical analysis (Browning [1975], Sjoblom [1985], and Verbon [1993] are but a few examples) and empiri- 392 P ENSION R EFORM : I SSUES AND P ROSPECTS FOR NDC S CHEMES cal studies (see Breyer and Craig [1997], Tabellini [2000], and Galasso and Profeta [2002]). One conclusion is that the age structure among voters is important, indeed significant, in explaining pension systems and reforms. However, the results from these studies do not seem to be consistent with Brooks and Weaver’s discussion. An important aspect of public pay-as-you go pension systems is that they are a social contract between generations—an implicit one, but still a contract. The generation voting for the system and introducing it hopes that coming, yet unborn, generations will fulfil the contract. The problem is that unborn generations do not have a vote in a democracy. This is one of the reasons it is so difficult to keep a pay-as-you-go DB system on an optimal level and also why a DC system (be it notional or funded) may do the trick. In Brooks and Weaver’s description it is taken for granted that the aim of adopting a NDC system is to reduce benefits and retrench the system, not to make the system robust, able to sustain in bad as well as in good times. In comparing NDC with reforms in DB systems, Brooks and Weaver say that introducing NDC is politically advantageous as the politicians can hide the retrenchment. Such a statement needs to be substantiated. Also, there is no explanation of why a politician would like to retrench the system. Why not just let the snowball keep growing and let coming generations (not having the right to vote, thus not being able to throw a politician out of office) inherit the debt? This is the intriguing question I would like to get answered. Also, here it would not be out of place to remind the readers about Disney’s (1996) ironic comments about one of the British reforms in the 1980s. According to Disney, the indexation (in a DB system) was changed in such a complicated way that “no one” realized what the effects on benefits would be! An aspect not discussed is to what extent the short time period since the first NDC system was introduced hampers the possibility of drawing conclusions. In the discussion about the diffusion of ideas, Brooks and Weaver bring forward the importance of networking for adopting a system like NDC, or the possible contagious effect of a neighborhood with an NDC system. Brooks and Weaver try to highlight the influence of these kinds of factors and it would be interesting to see these factors explored even further. There is a vast literature, perhaps mostly by economic historians such as D. North and N. Rosenberg, about the rise, spread, and development of different institutions. In most cases it does not seem to be a very speedy process. Probably there are neither enough cases, nor enough time that has elapsed, to make it possible to reject or accept this hypothesis about NDC systems. In the chapter, a large number of hypotheses is presented and discussed in the light of examples from different countries. It is, however, difficult to judge the value of the isolated examples given as the examples sometimes seem to be chosen in a rather unsystematic way. In any event, as the article contains such a great number of possible explanations, it is a plentiful source to use for future researchers wanting to perform a more formalized analysis. Notes 1. I define political risk as a decrease in the rate of return that individuals (think they) have been promised, a decrease caused by politically decided changes in the system. The ultimate risk in a pay-as-you-go system is of course that coming generations refuse to accept the social contract—that is, it turns out that you belong to the terminal generation. In this chapter there is a totally different definition of political risk: it is defined there as the risk that the politician will not be reelected. L ASHED TO THE M AST ? T HE P OLITICS OF NDC P ENSION R EFORM 393 2. These “limits” are highly dependent on a country’s history and culture. I doubt such limits are possible to establish in an objective way. Thus, it will be hard (impossible?) to accept or reject that hypothesis. References Boeri, T., A. Börsch-Supan, and G. Tabellini. 2001. “Would You Like to Shrink the Welfare State? The Opinions of European Citizens.” Economic Policy 16 (32): 7–50. Breyer, F., and B. Craig. 1997. “Voting on Social Security: Evidence from OECD Countries.” European Journal of Political Economy 13: 705–24. Browning, E. 1975. “Why the Social Insurance Budget is too Large in a Democracy.” Economic Inquiry 13: 373–88. Disney, R. 1996. “Can We Afford to Grow Older?” Cambridge, MA: MIT Press. Galasso, V., and P. Profeta. 2002. “The Political Economy of Social Security: A Survey.” European Journal of Political Economy 18: 1–29. Kruse, A. 2002. “Ageing Populations and Intergenerational Risk-Sharing in PAYG pension Schemes.” WP 2002:18, Department of Economics, Lund University. www.nek.lu.se/ publications. ———. 2003. “Svenska pensionsreformer under 1900-talet. Ett “Public Choice-perspektiv.” [Swedish Pension Reforms during the 20th Century: A Public Choice Perspective.] In 25 år i täten. En vänbok till Per Gunnar Edebalk. ed. L. Harrysson, O. Mallander, and J. Petersson Socialhögskolan, 83–98. Lund, Sweden: Lund University. Sjoblom, K. 1985. “Voting for Social Security.” Public Choice 45: 225–40. Tabellini, G. 2000. “A Positive Theory of Social Security.” Scandinavian Journal of Economics 102 (3): 123–35. Verbon, H. 1993. “Public Pensions. The Role of Public Choice and Expectations.” Journal of Population Economics 6: 523–45. World Bank. 1994. Averting the Old Age Crisis. Washington, DC: World Bank.