Reforming Social Security: Lessons from Sweden
By Paul Donahue
Retirement Section News, April 2024
It is universally understood that the United States’ program of payment of wage-related, tax-funded retirement benefits is out of balance. In 2033, the date currently projected for exhaustion of the Social Security Trust Fund that serves as the program’s reserve, projected tax revenues would provide approximately 77% of projected scheduled benefits.[1] Pension reform is a daunting challenge everywhere. The political landscape is a critical consideration in fashioning effective reform. For example, although France was (and remains) one of the state retirement pension schemes most desperately in need of reform, the Macron government needed to invoke Article 49.3 of the French constitution to enact relatively modest reforms without a vote by the National Assembly.[2] This led to days of sometimes violent protests and, according to some observers, very nearly caused the collapse of the government.[3]
However, both in terms of the underlying demographics and level of current spending, the challenges in the United States are significantly lower than in most European Union or OECD countries. US Social Security benefits are first of all relatively modest in comparison to other developed countries. Only 12 OECD countries replace smaller percentages of the average worker’s income than the United States, while 26 replace more. Sweden is among the countries that replace a higher percentage of the average worker’s income (65.3%) than the US (50.5%).[4] Budget challenges are also increasing for the average OECD countries, as the average percentage of GDP spent on Social Security-style benefits is expected to rise, from 8.9% (from 2020–2023) to 10.2% (by 2050).[5] The same study cites a more modest increase in the US (from 5.2% to 6.1%) while also projecting a decrease for Sweden (from 7.6% to 7.0%). The impact of an aging population is more significant among most of the larger OECD countries than for the United States. For example, among Canada, China, France, Italy, India, Germany, the United Kingdom, Japan and the United States, only China and India have a lower ratio of people over age 65 to people aged 20 to 65 than the United States.[6]
In the October 2023 Retirement Section News, I analyzed the adequacy of Social Security in meeting goals.[7] In this second of a projected series of five essays, I examine pension reform in Sweden and discuss what lessons we might apply to social security reform in the United States based on the Swedish experience.
Sweden
Sweden is the country that has made the most comprehensive and successful effort at reform. I examine the challenges that Sweden faced and the solutions Sweden found for those challenges, which are widely regarded as successful.
Demographics
Many demographic variables affect the adequacy of retirement benefit funding. The higher the number of children born per woman (fertility rate), the more younger workers there will be whose pension contributions will support payment of retirement benefits. Sweden has a relatively high fertility rate by European standards, 1.9 children per woman.[8] Since the 1960s, net migration to Sweden has been regulated to control population growth.[9] Nevertheless, Sweden’s labor force is projected to grow through 2062.[10]
Obviously too, the more people of every age who work, the broader the retirement contribution base. In 2022, Swedish labor force participation was among the highest at every age. In fact, 87% of those aged 25–54 worked, tied for fourth behind only much less developed Columbia, Hungary, and Czechia. Also, 77% of persons aged 55–64 were employed, fourth highest among OECD countries, and 28% of those aged 65–69, a little above the OECD median.[11]
The cost of benefits depends on the age at which people leave the workforce and the period of time that retirees will collect benefits. In Sweden, the normal retirement age is 65, and the effective age of labor market exit is 65.5 for men (1.1 years above the OECD average) and 64.5 for women (1.4 years above the OECD average).[12] In Sweden, the life expectancy remaining after labor market exit is 19.5 years for men and 22.6 years for women.[13]
The ratio of those collecting benefits to those still working is also important to the sustainability of systems that are substantially “pay-as-you-go” funded. OECD statistics provide ratios of the “old age” (by assumption “non-working”) population (ages 65 and higher) to the “working age” population (ages 20–64).[14] In the past, this probably overstated the number of workers per retired person. In 2020, the average age of labor market exit for the OECD was 63.8 for men and 62.4 for women.[15] By 2022, those numbers had risen to 64.4 for men and 63.1 for women. However, this metric works reasonably well for Sweden, with the age of labor market exit very close to 65 for both men and women. By contrast, for Japan, with labor market exit ages of 68.3 for men and 67 for women,[16] very likely the direction where all countries of the industrial west are headed, the traditional statistic is less useful. That said, the ratio for Sweden is projected to be 46.0% in 2052, and 60.4% in 2082, both very significantly below the projections for the OECD in aggregate (53.8% in 2052 and 66.1% in 2082).[17]
Pre-reform State Provision of Pensions
Sweden introduced a modest flat benefit pension (FP) in 1913, which was supplemented by the more robust earnings-related benefit (ATP) in 1960.[18] The ATP was based on a worker’s 15 years of highest earnings but required 30 years of earnings to provide an unreduced benefit. The maximum ATP benefit replaced 60% of the benefit calculation base; the base was approximately 1.5 times the average wage. For the average worker, the combination of the FP and ATP replaced approximately 65% of average wages (while an individual’s pension could range between 30% and 90% of average wages).[19] Benefits were taxed as regular income, though people with low benefits received an extra deduction.[20]
Employer payroll taxes of 18.5% in aggregate provided primary funding for the state pensions. The payroll tax was on all earnings, despite the benefit ceiling. Financing for the FP benefit was supplemented by general tax revenue. The system was pay-as-you-go with partial funding. By 1998, the system had a reserve of five years of benefits, with 85% invested in low-risk assets, mainly government and housing bonds.[21]
In 1960, the normal retirement age was 67. In 1976, the normal retirement age was lowered to 65. Early retirement was available at age 60 with actuarially reduced benefits, and retirement could be postponed to age 70 with actuarially increased benefits. By 1998, actual retirement age for men had declined to 62, nearly converging to the retirement age for women, which had remained constant near 62. By the mid-’80s, as everywhere in the developed West, an aging population and slowing productivity had put increasing pressure on the state pension system.[22]
Pension Reform
In 1998, Sweden moved from a universal, three-pillar defined benefit (DB) state pension structure to a structure which Palmer and Könberg describe as a “three-pillar defined contribution (DC) scheme, with a DB minimum guarantee income as its foundation.”[23] I’ll discuss more about the particulars in a bit, but please note that this DC scheme is by no means the equivalent of what is considered a defined contribution program in the United States.
Broad Political Consensus
In 1992, a severe recession, which led to a steep decline in contributions to the existing pension system, drove home the need to reform the pension system to many in Sweden. Following the defeat of the Social Democratic government in elections in 1992, the newly elected government formed a Pension Working Group, with representatives of each of the parties then in Parliament, with the task of recommending pension reform. The initial legislation adopted in 1994 was endorsed by five of the seven parties then in Parliament, which represented over 80% of voters.[24] The desire to preserve the broad support for pension reform the Pension Working Group represents has led, e.g., to reversal by a Center-Right government of a decision to lower the early retirement reduction factor to 80% from 93%, which had not been approved by the five-party working group.[25]
Primary Income Related Pension: Participants’ NDC Accounts
The new primary benefit is a nonfinancial defined contribution (NDC) structure. In the US benefits landscape, the structure has nearer analogues in the deferred compensation space than in either social security of funded>private pensions, DC, or DB. Although cumbersome, I describe it as a “notional personal account accumulation structure.”
Notional accounts were created for persons born in 1938 or later. For each year from 1960 to 1994, the notional accounts were credited with 18.5% of earnings, up to a cap,[26] based on earnings data captured for the prior system. For the years 1995 to 1998, 16.5% was credited, and from 1999 onward, 16%. Each year, the notional account balance is credited with a rate of return tied to per capita wage growth by means of an index, the income index.[27] Account balances for those who die before retirement age are redistributed among remaining participants in the same age cohort.[28]
Taxes, which provide the funding basis, have a different formula than for the notional accounts, hence the distinction from a traditional DC structure. Taxes on employee income up to the earnings cap of 7% are deductible against the employee’s income tax liability. The employer pension contribution of 10.21% is not capped.[29]
Benefits at Retirement
NDC benefits are available at any time after attainment of age 61. The retiree’s account balance is converted to an annuity based on the average life expectancy for the persons of the participant’s age and an imputed rate of return of 1.6%.[30] No adjustment for life expectancy is made after age 65.[31] Benefits in pay status are initially adjusted for the amount of inflation greater or less than 1.6%, since inflation of 1.6% was included in determining the annuity factor for calculation of the initial benefit.[32]
Automatic Adjustment Mechanism
The benefits and account balances as described above are subject to an automatic adjustment designed to guarantee financial stability in case assumptions about growth of the labor force, life expectancy, age at retirement, etc., should lead to a financial imbalance, which could be positive or negative. Each year, the system administrators calculate a “balance ratio.” The balance ration is equal to 1) the contribution asset, the value based on current enrollment of future contributions, plus the assets of the buffer fund, divided by 2) the value of future pension payments.[33] When the balance ratio falls below 1, the account balance and pensions in pay status annual adjustment is reduced by the balance ratio.
The Premium Pension
In addition to the NDC benefit component, there is a funded component, the premium pension, administered by the Premium Pension Authority (PPM). The premium pension is funded with a contribution of 2.5% of earnings up to the ceiling. The participants’ accounts are self-directed, and participants can invest in both Swedish and international funds.[34] The PPM is the record-keeper of the participant accounts. At retirement, any time at age 61 or later, the account balance will be converted to a fixed or variable annuity offered solely by the PPM. For the premium pension only, participants can elect a survivor benefit, which reduces the retirement annuity benefit.[35]
The Role of Reserve Funds
The principal pre-reform pension rogram had a reserve of 5.5 years of benefit payments at the time pension reform was enacted.[36] The significant reserve was not accidental. The Swedish Parliament intended a partially funded system, “so that future contributions needed to cover the benefit payments could be less than they would otherwise be if contributions were not made to a fund.”[37]
Although revenues from contributions have not been sufficient to provide for benefit payments since 2008, revenues from reserves have been sufficient to make up the difference and leave a small surplus.[38] This is partly because the automatic adjustment mechanism described above guards against deficits but makes no provision for surpluses. Current base line projections show steady growth in the size of the reserves.[39]
Guarantee Pension
Sweden’s retirement income program has always had the goal of providing adequate retirement income for all citizens.[40] The new pension system includes a separate guaranteed benefit to provide an adequate standard of living for persons with no or low earnings-related benefits. The guaranteed benefit is means-tested and financed by general tax revenues. Though integrated with earnings-related benefits, its general revenue funding makes it independent of them.[41] In 2016, while 32% of pensioners received at least a partial guaranteed benefit, the total amount of guaranteed benefits paid amounted to only 4% of total pension expenditures.[42]
Means-Tested Housing Allowance
The means-tested housing allowance supplements the guarantee pension to provide sufficient income for a decent standard of living for low-income pensioners. Currently, 13% of pensioners receive the means-tested housing supplement, predominantly elderly single women.[43]
Economic Status of the Elderly
About 11% of persons age 66 or older are classified by the OECD as living in poverty.[44] Both interesting and important for consideration of program adequacy, Swedish poverty rates are both lower overall and tightly clustered for all ages. Sweden’s overall population average rate is about 9.2%,[45] the rate for children aged 0-17 is about 8.8%.[46]
Benefit Financing
In 2022, Swedish pension contributions were 327 billion Swedish Kroner, of which 26 billion was a state contribution (8% of the total). Pension payments were 345 billion Swedish Kroner.[47] Private sector contributions covered 87% of the pension payments. The assets of the pension system grew by 541 million Swedish Kroner,[48] despite providing cash flow of 110 billion Swedish Kroner to support pension payments.[49] In short, the Swedish system is largely in balance, and the growth of the reserve fund might over time allow a reduction of contributions.
Lessons
Unfortunately, even though demographics are more favorable in the United States for support of government provided retirement income than even in Sweden, the Swedish outlook and attitudes that were required for state pension reform do not seem to exist in the United States. The European need for coalition government requires compromise and gives parties of the center considerable influence in shaping what the compromise will be. The general European consensus that retirees who have worked their entire lives should be able to live in dignity meant the Swedes could finance the minimum guaranteed pension and the means-tested housing allowance from general tax revenues, restricting “contributions” to the income pension. Similar support for the social security minimum benefit seems unlikely to materialize.
Funding of the premium pension proved the link that allowed overall revenue dedicated to the state pensions to rise. That is an element that could possibly play a similar role in reform of social security financing in the United States. Building a consensus for reform is the fundamental challenge, and perhaps one the professional and corporate sectors will need to achieve first, before political implementation is possible.
Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries, the newsletter editors, or the respective authors’ employers.
Paul Donahue, FSA, is a lawyer and actuary who works in New York City. Paul can be contacted at pauljdonahuefsa@aol.com.
Endnotes
[1] Status of the Social Security and Medicare Programs, A SUMMARY OF THE 2023 ANNUAL REPORTS,
Social Security and Medicare Boards of Trustees (https://www.ssa.gov/oact/trsum/).
[2]https://www.cnbc.com/2023/03/16/frances-macron-overrides-parliament-to-pass-pension-reform-bill.html.
[3]See, e.g., “French March in New Pension Protests, but Are They a Final Stand?” New York Times, June 6, 2023, https://www.nytimes.com/2023/06/06/world/europe/france-pension-plan-protests.html
[4] OECD (2023), Pensions at a Glance 2023: OECD and G20 Indicators, OECD Publishing, Paris,
https://doi.org/10.1787/678055dd-en, p. 157.
[7] The first was Adequacy of Social Security in Meeting Consensus Goals, Retirement Section News, October, 2023(https://www.soa.org/sections/retirement/retirement-newsletter/2023/october/ret-2023-10-donahue/).
[8] Edward Palmer and Bo Könberg, The Swedish NDC Scheme: Success on Track with Room for Reflection, World Bank Group, April 2019, p. 24. The OECD estimate for 2022 is 1.67, only modestly higher than the Unites States or other European countries.
[9] Palmer, p. 22. Sweden experienced net migration inflow of .8% in 2022, but that significant number was below the 2013-2019 average of 1.2%. International Migration Outlook 2023, Figure 1.2 (OECD), https://www.oecd-ilibrary.org/sites/b0f40584-en/1/3/1/index.html?itemId=/content/publication/b0f40584-en&_csp_=f32aa69b63450530407ffa5853cb88a4&itemIGO=oecd&itemContentType=book#figure-d1e2772-b045ae91b8. It is hard to overstate the importance of net positive migration on retirement benefit funding stability, which is part of the reason that the situation of Japan, with only .1% positive migration, is so dire with respect to retirement benefit funding.
[10] (OECD 2023), Figure 6.5, p. 185 (https://In .oecd-ilibrary.org/finance-and-investment/pensions-at-a-glance-2023_678055dd-en).
[11] Ibid., Figure 1.7, p. 27.
[12] Ibid., Figure 6,3, p. 191.
[13] Ibid., Figure 6.15, p. 193.
[14] Ibid., Table 6.2, note, p. 185.
[15] Pensions at a Glance 2021: OECD and G20 Indicators, https://www.oecd-ilibrary.org/finance-and-investment/pensions-at-a-glance-2021_ca401ebd-en,, p. 179.
[16] Ibid., Table 6.3, page 191.
[17]Ibid., Table 6.2, page 185.
[18] Annika Sundén, The Swedish Experience with Pension Reform, Oxford Review of Economic Policy, Spring, 2006, No. 1, p. 134. (cited as Sundén, Swedish Experience), Cf. also Annika Sundén, How Will Sweden’s New Pension System Work?, An Issue in Brief, Retirement Research Center at Boston College, March 2000, Number 1, p. 4. (cited as Sundén, Issue.)
[19] The average OECD replacement rate was 57% in 2005. P. 135. As noted above, the US social security replacement rate for the average worker is 43%.
[20] Sundén, Swedish Experience, p. 135.
[21] Ibid. I omit Sundén’s discussion of collectively bargained (roughly, “private” pensions).
[23] Edward Palmer and Bo Könberg, The Swedish NDC Scheme: Success on Track with Room for Reflection, World Bank Group, April 2019, p.7.
[24] Edward Palmer, The Swedish Pension Reform Model: Framework and Issues, p. 2.
[25] Kent Weaver and Alexander Willen, The Swedish pension system after twenty years: Mid-course corrections and lessons, OECD Journal on Budgeting, vol. 2013, no. 3, 2014, p.5.
[26] The cap in 2022 was 8.01 times the income base amount. Sweden, Old Age Pension, European Commission. The income base amount for 2022 was SEK 52,500 and increase of SEK 4,500 over 2222. Statistics Sweden.
[27] Sundén, p. 8. Sundén continues: “The goals were to ensure that earned pension rights followed the growth in average wages for the active population, and that individuals’ relative income had the same effect on their pension income irrespective of when they earned it during their lifetime.” p, 8.
[29] Sundén, p. 7. Note, the failure to cap the employer contribution introduces an additional element of progressivity to the overall pension system.
[33] Ole Settergen, The Automatic Balance Mechanism of the Swedish Pension System, The National Social Insurance Board, Working Papers in Social Insurance 2001:2, p. 10.
[35] Email to me from Alexandra E. Tellender, Swedish Pension Agency, Wednesday, July 26, 2023.
[36] Karl Gustav Scherman, The Swedish Pension Reform, Issues in Social Protection, Discussion Paper 7, Social Security Department, International Labour Office, Geneva, 1999. US Social Security reserves at the end of 2022 held less than three years of benefit payments.
[37] Palmer and Könberg, p. 28, quoting Parliamentary Bill No. 1958:55,
[40] Súnden, Swedish Experience, p. 41.
[42] Palmer and Könberg, pp. 32-33.
[44] OECD (2023), Table 7.2, p. 199. The OECD defines “poverty” as less than half the average income. Ibid., p. 198. By that definition, the United States rate of elderly poverty is more than double the Swedish rate, at 23%.
[46] https://www.statista.com/statistics/264424/child-poverty-in-oecd-countries/#:~:text=Among%20the%20OECD%20countries%2C%20Costa,United%20States%20did%20the%20same.
[47] Swedish Pension Authority 2022 Annual Report, p. 100. A Google translation of the Swedish original.