2. Description of the Korean public pension system

The Korean public pension system consists of three main components. Firstly, an old-age safety net, the Basic Pension, provides a means-tested benefit to 70% of the population aged 65 and over in 2020. Secondly, the contributory National Pension Scheme (NPS) covers private-sector workers since its introduction in 1988 and, in principle, all the self-employed since 1999 (Chapter 1). NPS benefits are currently paid from current contributions and accumulated financial assets, but those assets are projected to be depleted by 2057. The NPS is a defined benefit scheme with a strongly redistributive benefit formula in which the accrual has two equally weighted components: one is based on the average income of all participants while the other is calculated based on individual contribution records. Thirdly, three special regimes are in place covering civil servant (Government Employees Pension Scheme, GEPS), the military (Military Personnel Pension Scheme, MPPS), workers in private schools (Private School Teachers Pension Scheme, PSTPS) and workers in the Special Post Offices (Special Post Office Pension Scheme, SPOPS).

The chapter is organised as follows. The next section presents some old-age stylised facts in Korea. Section 2.3 describes the rules of the NPS. Section 2.4 discusses the four special regimes. Section 2.5 highlights the differences that exist within the NPS between coverage for employees and those classified as self-employed. This is followed by one section covering survivor pensions and gender gaps in wages and pensions.

Given that the main component of the pension system was only introduced in 1988, pension expenditures in Korea are still relatively low in international comparison (Figure 2.1). With increasing numbers of individuals becoming eligible for a pension from the NPS, there has been a significant increase in pension spending from a very low level of 1.9% of GDP in 2000 to 4.0% currently. Mexico and Türkiye also had big increases in expenditure due to both population ageing and increased coverage, as well as low retirement ages in the case of Türkiye. Now, only Mexico spends less on pensions than Korea, while the OECD average is much larger at 9.2% of GDP.

As gradually more individuals will become entitled to a pension in Korea, spending is set to increase further as a share of GDP. By 2050, it is projected to reach 6% of GDP (OECD, 2021[1]) (spending projections are covered in greater detail in Chapter 3).

The distribution of spending by components has changed over recent years with the introduction of the new basic old-age pension (BOAP) and the increasing number of new retirees becoming eligible for at least a partial pension from the NPS (Figure 2.2). In 2009, both the NPS and the Government Employees Pension Scheme (GEPS) spending amounted to around 0.6% of GDP, with the BOAP at under 0.3%. However, by 2020 the NPS benefits amounted to over 1.2% of GDP with expenditure set to increase markedly over the next couple of decades, while the GEPS expenditures also increased to 0.9% of GDP. With the 2014 reform to the BOAP effectively doubling the value of the benefit, expenditures increased substantially and now stand at 1.0% of GDP. Moreover, the Private School Teachers Pension Scheme (PSTPS), the Military Personnel Pension Scheme (MPPS) and the Special Post Office Pension Scheme (SPOPS) represented 0.12%, 0.17% and 0.002% in 2009, respectively, and amounted to 0.20%, 0.18% and 0.0028% respectively in 2020.

The current normal retirement age of 62 years in Korea is low in comparison to other OECD countries, with an average of around 64 years currently and a high of 67 years in a few countries. The OECD defines the normal retirement age in a given country as the age of eligibility of pensions without penalty from all schemes combined, based on a full career from age 22. Nevertheless, many workers in Korea continue in employment until very late in life. On average workers in Korea exited the labour market at 65.7 years for men and 64.9 years for women in 2020, compared to an OECD average of 63.8 and 62.4 years, respectively (Figure 1.13 in Chapter 1). Only Japan, New Zealand and Sweden had higher average exit ages for women.

As the pension system is still maturing in Korea, the statutory retirement age plays a much lesser role in influencing retirement decisions than in other OECD countries. The gap between the average effective exit age and the normal retirement age is thus amongst the highest in Korea at 4 years for men and 3 years for women. Closely related with such a high labour-market exit age, employment rates of older workers are also high in Korea. While the employment rate of workers aged 55 to 64 is slightly above the OECD average at 67% (Figure 1.10 in Chapter 1), Korea comes top when considering even older ages (Figure 2.3). Among individuals aged between 70 and 74 years over one-third are in employment, with Japan being the only other country above 30% and the OECD average at only 11%.

Most workers in Korea quit their main job relatively early (Table 1.1 in Chapter 1). This is heavily influenced by mandatory retirement ages set within firms, which limit the adverse impact of the seniority wage system on both labour cost at older ages and firms’ competitiveness. In Japan, too, it is common for employers to enforce a mandatory retirement policy (Box 2.1).

Figure 2.4 shows the average predicted increase in wages of workers in their fifties, both in the public and private sector, when they go from 10 to 20 years of tenure across the OECD, controlling for other relevant factors such as gender, education, experience and occupation. The impact of seniority on wages is considerably higher in Korea (15.1%) than in the OECD (5.9%), with Türkiye and Japan having the highest seniority-driven wage increase after Korea (Figure 2.4).

In Korea, some employers unilaterally decide the rules establishing a mandatory retirement policy while others deliberate them through collective bargaining. These rules automatically terminate the employment contract of any worker reaching a specified age limit. This kind of job termination is lawful in Korea unless the rules are stipulated based on discrimination related to gender, nationality, religion or social status, or in an arbitrary way (Article 23 of the Labour Standards Act) (OECD, 2019[2]).

In 2013, a new regulation on mandatory retirement age was enacted setting a floor of 60 years that employers must abide by. From 2016, the new rule is effective for state-owned or state-controlled companies, as well as private firms with at least 300 employees, extended to all employers in 2017. Prior to 2013, there was no legislation and employers were incentivised by wage subsidies to raise their mandatory retirement age on a voluntary basis.

From 2006 to 2018, encouraging employers to retain their older workers was mainly pursued through subsidies, of three kinds: the subsidy for the wage-peak system, a working-hour reduction subsidy and a subsidy for extended employment of older workers (Table 2.1). The subsidy for the wage peak system was introduced in 2006 to tackle early retirement practices. It assumed that the seniority-based wage system drives early retirement practices and that a wage cut for older workers would provide incentives for firms to retain older workers after the mandatory retirement age. In 2016, the subsidy provided KRW 30.8 billion to 6 683 employees (e.g. 0.63% of the total expenditure on unemployment benefits).

The working-hour reduction subsidy started in 2016 and was a new allowance for workers aged 50 or more. To extend employment of older workers, the government also provided subsidies to employers who i) hire workers who have retired because of the mandatory retirement age, ii) raise the retirement age above age 60, or iii) abolish the mandatory retirement age. In 2016, 2 496 companies received KRW 23.1 billion as subsidies. Since 2017, the mandatory retirement age, at age 60, has been effective for all firms, with a staggered introduction from 2013. All of three subsidies to retain the employment of older workers were eliminated in 2018 (MOEL, 2015[4]).

Due in part to the effect of the subsidy, by 2019, half of large firms with more than 300 employees introduced a wage peak scheme. In contrast, only 21.5% of firms with less than 300 employees introduced one because of the strong resistance by employees against reduced wages (MOEL, 2019[6]).

Beyond formal mandatory retirement, it is also common for employers in Korea to encourage their older employees to voluntarily leave their job before reaching the mandatory retirement age through a mutual agreement, the so-called honorary retirement. Hence, workers tend to “retire” early from their main job (most of them before reaching age 55) and find new employment in poor quality, highly insecure and low-paid jobs or become self-employed (OECD, 2018[5]). Honorary retirement is still used by employers to bypass the new regulation on the mandatory retirement age. In firms having honorary retirement, there are no other options offered by the company and basically all employees “accept” the terms and quit their job at the honorary retirement age. If they do not accept honorary retirement, they are likely to be assigned to a different job, be transferred and subject to all sorts of harassment (JILPT, 2011[7]).

Many large firms use honorary retirement. Among companies with more than 500 employees using honorary retirement, the average honorary retirement age was 51.8 years in 2015 for office workers and 54.3 years for factory workers, compared with the mandatory retirement age which was still 58.5 years in 2015. Among companies with 100-299 employees where the labour cost of older workers relative to younger workers is lower than in large firms as seniority wage is less important, there are very little differences between the mandatory retirement age and the honorary retirement age, suggesting that honorary retirement plays a small role in these firms (KLI, 2016[8]). After relatively early retirement from the main job, most Korean older workers have to find another job after they reach the mandatory retirement age in order to supplement insufficient old-age benefits.

One striking feature in Korea is that work income represents by far the largest proportion of income of the elderly at 49%, which includes earnings from both dependent and self-employment (Figure 2.5). Only Mexico is in a similar situation, with Chile being close to, highlighting that the elderly in both countries are disproportionally dependent on employment in later life. Across the OECD as a whole only one-quarter of old-age income comes from employment.

Conversely, public transfers (e.g. earnings-related pensions, resource-tested benefits) only account for 30% of income of older people in Korea, with nothing recorded from private occupational transfers (e.g. pensions, severance payments, death grants). By contrast, public transfers and private occupational transfers account for two-thirds – 57% and 7%, respectively – of the total income of older people on average among OECD countries (Figure 2.5).

Although individuals tend to work at very old ages in Korea, the total disposable income of the elderly relative to the population as a whole, is much lower than in other OECD countries. Those aged 65 or over have, on average, an income equal to 68% of that of the population, well below both Japan and the OECD average, at 85% and 88%, respectively (Figure 2.6). By comparison those aged 65 or over in Costa Rica, France, Israel, Italy and Luxembourg have incomes close to 100% or more of that of the rest of the population.

While as in all countries income tends to fall with age after 65 in Korea, the age gradient of income is particularly steep. On average, those older than 75 have an income that is 22% lower than the group aged 66-75 (59% vs 75% of average income in the whole population). Those in the 66-75 age group have higher pensions as they have been able to contribute for longer during their working lives. For comparison, the gap is only 13 percentage points on average in the OECD (80% vs 93%). This is despite record-high employment at older ages in Korea. The main reason is that over two-thirds of workers over age 60 in Korea can only find temporary or part-time employment (Min and Cho, 2018[12]). This makes it difficult to have a guaranteed source of income in later life. For those unable to find paid employment they turn to self-employment and, as will be shown below, the incomes of the self-employed tend to be well below that of employees. With a lack of pension provision older people in Korea are in a precarious position.

Such a discrepancy in income among the 66-75 and 76+ age groups contributes to the large income inequality among those aged 65+ in Korea. The Gini coefficient – a measure of income inequality defined between 0 (complete equality between all) and 1 (complete inequality, i.e. one person receives all income) – among those aged 65 and over is equal to 0.376, much higher than the OECD average of 0.310 (Figure 2.7). Other OECD countries where old-age income inequality is very high are the United States (0.421), Chile (0.441), Mexico (0.473) and Costa Rica (0.502). In the majority of OECD countries, income inequality, measured by the Gini coefficient, is significantly lower among those aged 65 than for the population as a whole.

Low relative old-age income leads to very high poverty incidence in Korea. Relative poverty rates – defined as having an income below half the national median equivalised household disposable income – are high across all age groups, ranging from 8.4% for the 26-40 age group to 52.0% for those aged 76+. They are three times the OECD average for the 66 to 75 and 76+ age groups (Figure 2.8).

In relation with high poverty rates amongst the elderly, the rate of suicide among those aged 65 or older increased from an already high level in the aftermath of the Asian economic crisis in 1997 (Figure 2.9). It then dramatically increased from 36 (per 100 000 persons) in 2000 to a peak of 82 in 2010 in the wake of the global financial crisis (far above the OECD average of 22 (OECD, 2011[13]) before declining to 47 in 2019). This suggests that given weak old-age protection, large downturns have huge and dramatic impacts on the elderly. According to a government survey of the elderly who have considered suicide, “economic hardship” is the second-most cited reason, following “disease and disability” (Statistics Korea, 2010[14]). The level of family support to the elderly has also been declining in recent years combined with many more of the elderly now living alone – private transfers, including family support, have fallen from 55% of the elderly’s income in 1990 to 45% in 2008, and by 2030 around 40% of those aged 65 and over are projected to be living alone (Jones and Urasawa, 2014[15]).

The National Pension Scheme (NPS) is a defined benefit scheme. Current contributions are sufficient to pay current benefits, and there are excesses, which are transferred to the National Reserve Fund (NRF) for investment. The NRF is projected to be fully depleted from 2057, from which point pension benefits will be paid out of contemporaneous contributions. The role of the NRF is discussed in Chapter 3.

This section describes the current system in detail, highlighting, where needed, a few key past reforms. In order to provide an overview upfront, the main pension reforms since the introduction of the scheme are summarised in Table 2.2. These reforms are discussed in greater detail in the corresponding sub-sections below.

Since the introduction of the NPS in 1988 the coverage of workers has expanded and contribution rates have been increased whilst future benefit promises have been reduced. At the beginning, the NPS covered workers in workplaces with 10 or more employees, with contributions of 3.0% of covered wages, split equally between employee and employer. The retirement age was set at 60 years and the “target” replacement rate – referred to in the NPS scheme as the replacement rate based on a 40-year career with earnings at the average of all contributors – was set at 70%. The first year where individuals are eligible to receive a pension with a 40-year career will be 2028, but none of the initial parameter values remain in place.

In 1992, four years after its introduction, the pension system was expanded to cover those with five or more employees. Until July 2006 coverage was gradually expanded to the self-employed and to those employed in smaller firms (Table 2.2). Now all workers in Korea are covered by the NPS unless of course they are covered by special schemes for civil servants, private-school employees, military personnel or employees of the special post office.

In 2000, after all the self-employed and workers in companies with more than five full-time employees were mandatorily covered, 16.2 million were enrolled, representing 77% of those in employment. Over the last 20 years these numbers have steadily increased, reaching 22.2 million and 82% of those employed at the end of 2019. For the workplace-base insured only the coverage level is slightly higher, but was still only 85% in 2017 (OECD, 2018[5]).

There has been a significant change in the composition of those covered. In 2000 of the 16.2 million covered individuals, 5.7 million were employees whilst 10.4 million were self-employed – classified as individually insured in Korea – who must pay both the employee and employer components of the contributions. By 2020 this breakdown had completely reversed with 14.3 of the 22.1 million covered now being employees and only 6.9 million being self-employed (Table 2.3). This reflects both the expanded coverage of the workplace based insured and the decline in the number of self-employed workers (Section 2.4).

Although all of the self-employed are now mandatorily enrolled in the NPS, and therefore classified as being insured, they are not all actually contributing. The income of the self-employed is self-reported and under-reporting is a substantial problem among the self-employed in Korea (UNESCAP, 2016[16]). In addition, pension contributions are exempted if there are business losses. Consequently, just under half of the self-employed are temporarily exempted from paying contributions, though the proportion has fallen recently: while in 2010, around 59% of all the self-employed were exempted from paying contributions, this fell to 54% in 2015 and further to 45% in 2020 (Figure 2.10). During the period the self-employed are exempted from paying contributions they are not gaining any pension entitlements.

As will be discussed in detail below, aggregate pensionable income affects the pensions of everyone in Korea, hence under-reporting and the large share of self-employment generate a significant impact on pension outcomes across the board. This issue is especially serious for Korea as self-employment represents one of the larger share in total employment among OECD countries.

Women contribute more and more to the pension system in Korea. Of all contributors 55% were men and 45% were women in 2020. While the number of men contributing increased by less than 10% over the last 20 years, the number of women contributing has more than doubled (Figure 2.11). Only part of the increase is due to higher employment: employment increased for men from 73.2% to 75.7% among those aged 15 to 64 years, and from 50.1% to 57.8% for women. The main source of increase was the inclusion of small workplaces with under five workers into mandatory coverage, as these workplaces are primarily staffed by women.

To encourage more workers to contribute, the government introduced the Duru Nuri Social Insurance Support Project in 2012 which pays a percentage of both the worker and the employer pension contribution. To be eligible today, individuals must be employed in workplaces with fewer than ten workers and earning under KRW 2.15 million per month, around 50% of average earnings. For those newly insured – those without a workplace insurance record in the last year – the project covers between 80% and 90% of the contribution payment, on average, for the first year. After this first year only 30% of payments are covered. However, to ensure fairness all those that were already insured but were employed in workplaces with fewer than ten workers and earning under KRW 2.15 million per month were also enrolled in the project. They receive 30% of payments from both the employee and employer from the scheme. The NPS covers around 80% of the cost of the project and the Ministry of Employment and Labour finances the remaining part.

Since the introduction of Duru Nuri, coverage has expanded rapidly (Table 2.4). However approximately 98.5% of the programme cost was a deadweight loss: the large majority of subsidised people would have been insured without the subsidy (OECD, 2018[17]). Put differently, for every 1 000 subsidised employees Duru Nuri created 15 were newly covered employees, implying a cost per new enrolment of around KRW 50 million – roughly three times the person’s annual wage.

In order to be eligible for a pension under the NPS individuals must have made at least 10 years of contributions. This is close to the OECD average; 16 countries require at least 15 years while 14 countries have no minimum period, granting eligibility as soon as any contribution is made (Figure 2.12). Like Korea, Japan and the United States also require 10 years. In Korea those who do not meet the 10-year condition receive a lump-sum payment equal to past (employee plus employer) contributions plus accrued interest.

The statutory retirement age at which someone is eligible for an old-age pension is currently 62 years. The retirement age was set at 60 years when the system was introduced, but has been increasing since 2013 and will reach 65 for those born in 1969 or after i.e. from 2034 (Table 2.5).

The normal retirement age is currently one of the lowest in the OECD and will remain low in international comparison.1 For full-career workers, the current retirement age is 64.2 years in the OECD on average for men and 63.4 for women, with only Costa Rica and Türkiye as well as Colombia but for women only having a lower retirement age than 62 years (Figure 2.13). With many countries planning to increase their retirement ages over the coming decades the average age across the OECD will increase by nearly two years, reaching 66 years for individuals who started working at age 22 in 2020. As a result, the retirement age in Korea will be one year below the OECD average despite experiencing much faster population ageing (Chapter 1). For future normal retirement ages, only four OECD countries will have retirement ages below 65 years: the Slovak Republic at 64 years and Colombia, Luxembourg and Slovenia at 62 years. By contrast, Denmark will be at 74 years with Estonia and Italy at 71 years; all three countries have linked future retirement ages to life expectancy.

Under the NPS individuals now pay a contribution of 9% of covered wages, being split equally between employees and employers. The only exception to these levels are for farmers and fishermen, which currently represent 6% of total employment (KOSIS, 2020[19]). Since July 1995 up to half of their contributions have been subsidised, in a context in which their incomes declined with large imports of agricultural products and seafood.

For workers contributions are mandatory from age 18 until reaching age 59. From age 60 onwards contributions are voluntary. Only those who have not yet reached 10 years of contribution, but would do so by age 65, can make voluntary contributions from age 60. In nearly all other OECD countries, pension contributions are mandatory until retirement age, though Japan, for example, has voluntary coverage for those aged 60 to 64 within the national pension.

The current NPS contribution rate is very low, as 9% represents less than half of the OECD average contribution rate for mandatory schemes (Figure 2.14). Only Lithuania and Mexico have a lower rate (in addition to New Zealand that does not have any mandatory contributory pensions), but the contribution level in Mexico will increase to 15% by 2030. The contribution rates in most other countries are considerably higher, for example 18% in Japan, 28% in France and 33% in Italy.

In Korea there is a wage ceiling to contributions and therefore to pensionable wage at KRW 5.03 million per month, equivalent to 131% of average earnings in 2020 as computed by the OECD. Only four countries have a lower level than Korea, with the average for Japan, for example, much higher at 237% of average earnings (Figure 2.15).

There are two separate NPS benefit components: one is based on the average earnings of all contributors and one based on individual earnings. The component that is based on the average earnings of all contributors is a flat-rate payment depending solely on the number of years of contributions. The OECD classifies this component as a contribution-based basic pension. The reference wage for this component is the average wage of all contributors over the three years prior to retirement, uprated by inflation.2 The other component is an individual earnings-related component, with an associated accrual rate. Formally, in Korea, the contribution-based basic pension and the accrual-based component are treated as one pension, with an associated “target” replacement rate at the average earnings of all contributors (A level).

Upon introduction the NPS annual accrual rate (of the earnings-related component) was 0.75% for an average earner, based on the 3% contribution rate, which also financed the contribution-based basic component. These parameter values allowed to offer a gift to the first generation of contributors, but over time such pension promises based on such a low contribution rate could not be financially sustainable. Twenty years after, in 2008, the contribution rate had been increased to 9% and the accrual rate for further contributions cut to 0.625%. Thereafter the accrual rate for each subsequent year was reduced by 0.00625 percentage points until it reaches 0.5% in 2028. Those who have contributed for 40 years will be able to retire for the first time in 2028, and for them the effective annual accrual rate will be 0.66%, but will decline over time until reaching 0.5% for those retiring from 2068 onwards (Figure 2.16).

As past earnings are uprated with average – wage growth and as the ceiling for contributions is above the average-earnings level, at around 130% of average earnings, the nominal accrual rate of 0.5% is also the future effective accrual rate of the individual-earnings component. In the OECD, only Lithuania and Estonia have lower accrual rates at 0.18% and 0.32% respectively, with Japan also being at 0.5%. Across the OECD, the average effective accrual rate is slightly above 1% (Figure 2.17).

The accrual rate that is consistent with financial sustainability depends on a range of factors. Those include the resources financing the system, referring primarily to the contribution-rate parameter, retirement ages, demographic variables and the valorisation of past wages. For example, in Italy where NDC (notional defined contribution) rules are meant to ensure financial sustainability over time, the future accrual rate is high at around 1.6, broadly consistent with a very high contribution rate of 33% and the way NDC deals with the impact of ageing.

As a result of all of the changes outlined above concerning eligibility criteria, the retirement age and accrual rates, the formula for calculating the final pension entitlement has undergone a number of changes since the pension system was first introduced.3

When the system was introduced, pension entitlement required 20 years of contributions with their pension expected to be calculated as follows:

Pension= A+0.75 B* 20%* (1+0.05 N)

where ‘A’ is the mean wage of all participants in the pension system in the three years before reaching pension age and ‘B’ is the average income of the individual during the entire insurance period, uprated with wage growth. ‘N’ is the total number of contribution years beyond the pension entitlement requirement of 20 years and 20% is the constant factor designed to give an average earner with a 40-year career a replacement rate of 70%.4 With positive real-wage growth, the replacement rate is actually slightly lower than the “target” as the three years of past earnings for ‘A’ are only uprated with inflation.

The 1998 National Pensions Reform Act reduced the pension promise for contributions made thereafter, whilst preserving the entitlements of those made previously. The pension promise for contributions made after 1998 was that a 40-year career would generate a replacement rate of 60% for an average earner. The pension formula was then adjusted accordingly, becoming:

 A+0.75 BP1P*20%+ A+BP2P*15%*(1+0.05 N)

where ‘P1’ is the number of months of contributions made before 31 December 1998, ‘P2’ the number of months of contributions made after 31 December 1998, ‘P’ is the total number of months of contribution across all periods and 15% is the constant guaranteeing a replacement rate of 60% for a 40-year career from 1999 onwards for an average earner.5

Following the latest reform to the NPS in 2007 the pension promise was again revised, with an immediate lowering of the “target” replacement rate to 50% from 1 January 2008 and then a gradual reduction in the “target” replacement rate of 0.5 percentage points every year thereafter until reaching 40% in 2028. This therefore led to the pension formula that is in place today

 A+0.75 BP1P*20%+ A+BP2P*15%+ A+BP3P*12.5%++ A+BP23P*10%*(1+0.05 N)

where ‘P3’ is the number of months of contribution made in 2008, up to ‘P23’ which is the number of months of contributions from 1 January 2028. 12.5% is the constant guaranteeing a replacement rate of 50% for a 40-year career at average earnings with the constant being reduced by 0.125 every year until reaching 10% for 2028 and beyond – the level required to guarantee a future replacement rate of 40%. Given the phasing-in of this reform, it is going to take several decades before the replacement rate for a full-career average earner stabilises at 40%, assuming there are no subsequent reforms.6 Chapter 3 analyses the replacement rates at various earnings levels and career paths in greater detail.

An additional entitlement is also given for dependents of the pension recipient, including the spouse, children up to age 19 or parents of either the pension recipient or their spouse aged at least 60 years. The age requirement for children and parents is removed if they suffer either first or second degree disability. The levels of the supplement are small, amounting to around 0.5% of average earnings for the spousal supplement and less than 0.4% for both the child and parent components.

The GEPS was the first mandatory pension scheme in Korea. It covered both public officials and military personnel when it was introduced in 1960. The latter have had their own separate pension scheme from 1963. As of December 2020 there is a total of 1.23 million contributors to the GEPS and 590 000 pension recipients, which implies a support ratio – number of contributors per pension recipient – of 2.08, much lower than that of the NPS (4.15). This results from both the NPS still having to reach maturity and because formally there was no retirement age between 1962 and 1996 in the GEPS – retirement was possible at any age once 20 years of contributions had been made -, leading to a large number of early retirees.

Initial contribution rates of the GEPS were very low. The initial rule would have entitled individuals to a pension of 2.1% of their final salary for each year of contribution, subject to a minimum period of 20 years. As with the NPS, all of these parameter values have since been changed substantially. Table 2.6 provides an overview of reforms of the GEPS, highlighting four major reforms to the system, in 1995, 2000, 2009 and 2015.

The retirement age is currently 61 years, gradually increasing to 65 in 2034. By comparison the retirement age in the NPS is 62, also increasing to 65 by 2034. Between 1996 and 2010 retirement was possible from age 50 for those employed prior to 1995 and 60 for those employed after; between 1962 and 1995 there was no legislated retirement age with only 20 years of contribution being required.

The contribution rate has sharply increased over time. The current rate is 18% of “standard monthly income”, split equally between employee and the government as the employer, having been increased many times since 1969, and from 14% as a result of the 2015 reform. The 2015 reform also reduced the ceiling to contributions from 1.8 to 1.6 times the average “standard monthly income”, which is around one-third higher than the ceiling for private-sector workers.7

Reforms to the accrual rate and the reference wage have substantially reduced the generosity of the GEPS. As part of the 2015 reform the accrual rate is being gradually lowered by 0.01 percentage point each year from 1.9% until reaching 1.7% in 2035. Prior to 2009, the accrual rate had been set at approximately 2.1%.

The pension formula that will apply after the reform is complete is:


where n is the total number of years of contribution and w is the uprated average pay (up to the ceiling) over the whole career since the 2009 reform. Before 2009, the reference wage was computed over the final three years, and before 2000 it was the final salary.8

The pension calculation is also subject to an additional maximum career length factor as only 36 years of contributions can be considered within the calculation, therefore giving a maximum replacement rate of 61.2% after the reduction to the accrual is fully enacted. Government employees will then have a maximum replacement rate of 61.2% based on a contribution rate of 18%, whereas the long-term “target” replacement rate in the NPS of 40% is based on a contribution rate of 9%.9

Unlike the NPS which has been using indexation to the CPI since it was introduced, the indexation process of the GEPS has changed several times. Since 2009 it is set to follow CPI. Between 2000 and 2009 there was indexation to the CPI but with an additional protective element to ensure that the gap between CPI and wage growth was not too high. Prior to the 2000 reform the pensions were indexed to the salary base.

As mentioned above military personnel were initially covered under the GEPS until their own scheme was established in 1963. The rules, however, for the military pension remained closely aligned with those of civil servants until 2015, with a few notable exceptions. Firstly, once eligibility conditions have been met, retirement is possible at any age, meaning there is still no legislated retirement age, as was the case for the GEPS between 1962 and the 1995 reform. Secondly, for periods served in combat the validated period is effectively tripled as a compensation. Since 2015, the schemes have diverged further as none of the reforms of the GEPS were adopted for the MPPS, so the accrual rate remains at 1.9% and the contribution rate is still set at 14%.

In order to standardise the pension treatment of public- and private-sector teachers the Private School Teachers Pension Act was introduced in 1973 to cover private-sector teachers since 1975. Unlike with the MPPS above, the rules of the PSTPS follow those of the GEPS exactly. The source of funding for the 18% long-term contribution rate for teachers is composed of 9% from employees, 5.294% coming from the private school employer (foundation) and 3.706% from the government. However, for non-teaching staff the entire 9% is actually paid by the school foundation rather than being split between the foundation and the government.

Special post offices are those post offices that have been established by individuals at their own expense in an area where a post office does not currently exist. The staff within these post offices are subsequently treated as civil servants for pension rights and so as with the Private School Teachers Pension the SPOPS follows that of the GEPS.

Korea is one of only four countries along with Belgium, France and Germany that has maintained a completely separate pension system for public- and private-sector workers (Table 2.7). It is relatively common in the OECD to have different rules or systems for military personnel, but the trend in recent years has been for the unification of pension systems covering public- and private-sector employees. Over the last decades, several countries have indeed integrated the pension schemes covering civil servants into their private-sector pension systems: Greece, Israel, Italy, Japan, New Zealand, Portugal, Spain and Türkiye (Box 2.2). However, this integration typically applies only to new employees; existing employees are thus “grandfathered” meaning that different rules exist for public-sector workers depending on when they began their career in government.

Amongst OECD countries half now have a unified system covering both public- and private-sector workers. OECD (2016[20]) estimated the difference in pension benefits for full-career average-wage workers between the public and the private sector. A comparison of replacement rates for public-sector workers across the OECD is presented in Chapter 3.

Most pension systems were designed with the case of workers having stable full-time careers in mind. This was also true for Korea and the self-employed had to wait several years after the introduction of the NPS to be mandatorily covered. Before that, they had to choose to voluntarily contribute to the NPS, as was the case for all individuals not mandatorily covered. Non-standard forms of employment such as self-employment, part-time work and temporary work are not marginal phenomena in OECD countries, and workers in such jobs have often a lower level of old-age protection (Chapters 2 and 3 in OECD (2019[22])). This section focuses on recent trends in self-employment in Korea and describes the pension coverage of the self-employed.

As in most OECD countries, self-employment as a share of total employment has decreased in the last decades, but the fall was much sharper in Korea from 37% in 2000 to 25% in 2020. By comparison the OECD average went from 19% to 15% over the same time period (Figure 2.18). The current level remains high in Korea, ranking third of all countries in 2020, after Greece and Türkiye.

Since 2005, the contribution rate for the self-employed (individually insured) has been aligned to that for employees, i.e. at the current 9% level. As the self-employed are their own employers, they are liable for the full 9% contribution rate, i.e. both the employee and employer parts. For employees the contribution base is their gross wage, whilst for the self-employed it is business income, i.e. total revenue minus business expenses.

The enrolment of the self-employed into the NPS is markedly lower than for workers with regular employment status (UNESCAP, 2016[16]). However, in Korea, when combining employment and self-employment, only earnings from dependent work are subject to pension contributions and entitlements, irrespective of the level of income from self-employment.

Having an accurate assessment of the income of the self-employed can be difficult and under-reporting is commonplace in many countries, including Korea where around 40% of income from self-employment might be under-reported (Kim, Gibson and Chung, 2009[3]). Such a high level has large financial impacts, from uncollected taxes to possible overpayment of means-tested benefits.

For pension benefits, the impact is not only felt by the individual through lower entitlements from the earnings-related component, but also by all contributors to the NPS due to the basic pension element depending on the average income of all contributors. Indeed, if income is under-reported then the overall average is lowered. This results in a lower average pension from the collective component for everyone. Conversely, the individual who under-reports will only see the future pension reduced through the earnings-related component, which therefore does not reflect the full impact of under-reporting. This provides some incentives to under-report. In order to combat these impacts, several countries, including Korea, use a minimum contribution base for the self-employed. In Korea this base is the lower limit of earnings, equivalent to 9% of average earnings whilst other countries often use the minimum wage.

In Korea, the average income of the self-employed has actually been decreasing in recent years relative to the average employee salary, from 62% in 2000 to 7% in 2020 (Table 2.8). This contributes to lowering the overall average earning while the shrinking employment share of self-employment goes the other way, with an ambiguous overall effect.

Historically female employment has been lower than male’s in the large majority of countries. Combined with longer life expectancies, this implies that female pensioners are typically more reliant on first-tier pensions, their partner’s pensions or ultimately survivor pensions. As a result, poverty levels are higher among older women than among older men in all OECD countries, with women over-75 more at risk of poverty than the 66-to-75-year-olds due to cohort effects – female employment was lower among the older cohorts -, the impact of pension indexation and widowhood.

Women’s average pensions were 44% lower than those of men, in Korea in 2019, well above the OECD-35 average of 26%. Only Japan has a larger gap (Figure 2.19). The large gender pension gap in Korea is partly explained by gender differences in wages, which are the largest in the OECD, equal to between two and three times the OECD average (Figure 2.20). However, as only half of pension entitlements depends on individual earnings (Section 2.3), the impact of gender wage differences on gender pensions is attenuated. With improving employment among women in many countries, their pension entitlements will increase over time, and gender differences in earnings-related pensions will tend to decrease.

Following the death of a partner, survivor pensions have two main objectives. Firstly, they protect the surviving spouse from the risk of poverty by preventing disposable income from falling to potentially very low levels. Secondly, they contribute to insuring against the decrease in disposable income after the partner’s death, in the same way as old-age pension’s help avoid a sharp drop in income when moving out of paid work upon retirement – the so-called consumption-smoothing objective aimed at maintaining standards of living. Women represent around 88% of survivor pension recipients for widowed persons on average across OECD countries, and 92% in Korea (OECD, 2018[25]). Hence, survivor pensions may play a key role in reducing gender differences in old-age disposable income.

In the NPS scheme, the survivor pension is equal to 60% of the deceased’s pension, provided that the deceased made at least 20 years of contributions (Figure 2.21). Between 10 and 19 years of contributions, 50% is paid, 40% for less than 10 years of contributions. In this latter case of less than 10 years of contribution, the initial NPS entitlement is calculated pro-rata. In the GEPS the survivor benefit is 60% irrespective of the length of contribution.

Half of OECD countries have a survivor pension equal to at least 60% of the deceased’s pension. On average in the OECD, future survivor pensions will replace 50% of the deceased’s mandatory contributory pension at retirement age, when no other income is taken into account for any household member including the survivor (OECD, 2018[25]).

Survivor benefits, in Korea, can be paid to a range of relatives but there is an order of priority: spouse, children up to the month they reach age 25 (any age if assessed with a first- or second-degree disability), parents, grandchildren up to the month they reach age 19 (any age if assessed with a first- or second-degree disability) and grandparents (including the spouse’s parents or grandparents) aged 62 or older (any age if assessed with a first- or second-degree disability).

In the majority of OECD countries the age of the surviving spouse has to be above a threshold to provide eligibility to a permanent benefit from survivor pensions. In many countries a temporary payment is made irrespective of age to help smooth the transition, but after that survivors are eligible to receiving the permanent component only once they have reached the eligibility age. However, Korea is one of 11 OECD countries for which there is no such minimum age for entitlement to permanent survivor pensions. This implies that even if the surviving spouse is young, he or she can receive permanent survivor benefits for many decades. Spending on survivor pensions represented 0.3% of GDP in Korea against 0.8% in the OECD on average.

When applying the rules for survivor pensions, assumptions need to be made about the employment history of the survivor, as this also affects the pension that would have been received before the partner’s death. For example, in Korea, a non-working spouse will receive a spousal supplement within their partner’s pension, even though they have not made any contributions in their own right. In Korea, when survivors have no personal pension entitlement, they will receive, based on current legislation, 59% of the deceased’s pension, compared to 46% on average across OECD countries (Figure 2.22). These numbers are lower than 60% for Korea and 50% for the average shown in (Figure 2.21) because of the aforementioned spousal supplement that was received upon retirement, but which is not part of the survivor pension calculation. A similar system also exists in Japan amongst others.

When the survivor is also assumed to have had a full career at average earnings then the survivor benefit falls to 12% of the deceased’s earnings-related pension in Korea, compared to an average of 22% across the OECD. In Korea, survivors are entitled to the greater of the full survivor pension and of their own pension plus 20% of the survivor pension. So for a couple with both partners with full careers at average earnings the latter is the more beneficial, supplementing the survivors own pension with 12% (= 20% * 60%) of the deceased’s pension. Overall, the survivor pension is thus withdrawn against the survivor earnings-related pension at a higher rate in Korea than in the OECD on average, at least at the average-wage level.


[26] GEPS (2016), Annual Report 2016, Government Employees Pension Service, GEPS Research Institute, https://www.geps.or.kr.

[7] JILPT (2011), Business Labour Trend Sep/2011(in Japanese), JILPT Publishing, Tokyo, https://www.jil.go.jp/kokunai/blt/backnumber/2011/09/index.html.

[15] Jones, R. and S. Urasawa (2014), “Reducing the High Rate of Poverty Among the Elderly in Korea”, OECD Economics Department Working Papers, No. 1163, OECD Publishing, Paris, https://doi.org/10.1787/5jxx054fv20v-en.

[8] KLI (2016), Korea Labor Institute , Labour Review AUG/2016(in Korean), https://www.kli.re.kr/kli/pdicalView.do?key=19&pblctListNo=8747.

[19] KOSIS (2020), Korean Statistical Information Service, https://kosis.kr/eng/statisticsList/statisticsListIndex.do?menuId=M_01_01&vwcd=MT_ETITLE&parmTabId=M_01_01&statId=1962002&themaId=#SelectStatsBoxDiv.

[11] MHLW (2019), Ministry of Health, Labour and Welfare, Employment status of elderly people 2019 (in Japanese), https://www.mhlw.go.jp/stf/houdou/0000182200_00003.html.

[9] MHLW (2017), Ministry of Health, Labour and Welfare, General Survey on Working Conditions 2017 (in Japanese), https://www.mhlw.go.jp/toukei/list/11-23.html.

[21] MHLW (2012), Ministry of Health, Labour and Welfare, Bill tabled in the Diet (2012) (in Japanese), https://www.mhlw.go.jp/topics/bukyoku/soumu/houritu/dl/180-54.pdf.

[12] Min, D. and E. Cho (2018), “Patterns in quality of life according to employment among the older adults: the Korean longitudinal study of aging (2008–2014)”, BMC Public Health, Vol. 18/1, https://doi.org/10.1186/s12889-018-5296-x.

[6] MOEL (2019), Ministry of Employment and Labor, Status of the introduction of wage peak system(in Korean), https://www.moel.go.kr/info/publict/publictDataView.do?bbs_seq=20200800349.

[4] MOEL (2015), Ministry of Employment and Labor, The effect of introducing a wage peak system, https://www.moel.go.kr/common/downloadFile.do?file_seq=21171316982&bbs_seq=1433822121810&bbs_id=LOCAL1.

[18] OECD (2021), Pensions at a Glance 2021, country profiles, http://oe.cd/pag.

[1] OECD (2021), Pensions at a Glance 2021: OECD and G20 Indicators, OECD Publishing, Paris, https://doi.org/10.1787/ca401ebd-en.

[23] OECD (2021), Towards Improved Retirement Savings Outcomes for Women, OECD Publishing, Paris, https://doi.org/10.1787/f7b48808-en.

[24] OECD (2020), OECD Employment Outlook 2020: Worker Security and the COVID-19 Crisis, OECD Publishing, Paris, https://doi.org/10.1787/1686c758-en.

[2] OECD (2019), OECD EPL Database, OECD Publishing, Paris,, https://www.oecd.org/els/emp/Korea.pdf.

[22] OECD (2019), Pensions at a Glance 2019: OECD and G20 Indicators, OECD Publishing, Paris, https://doi.org/10.1787/b6d3dcfc-en.

[3] OECD (2019), Working Better with Age, Ageing and Employment Policies, OECD Publishing,Paris, https://doi.org/10.1787/c4d4f66a-en.

[25] OECD (2018), OECD Pensions Outlook 2018, OECD Publishing, Paris, https://doi.org/10.1787/pens_outlook-2018-en.

[17] OECD (2018), Towards Better Social and Employment Security in Korea, Connecting People with Jobs, OECD Publishing, Paris, https://doi.org/10.1787/9789264288256-en.

[10] OECD (2018), Working Better with Age: Japan, Ageing and Employment Policies, OECD Publishing, Paris, https://doi.org/10.1787/9789264201996-en.

[5] OECD (2018), Working Better with Age: Korea, Ageing and Employment Policies, OECD Publishing, Paris, https://doi.org/10.1787/9789264208261-en.

[20] OECD (2016), OECD Pensions Outlook 2016, OECD Publishing, Paris, https://doi.org/10.1787/pens_outlook-2016-en.

[13] OECD (2011), Society at a Glance: Asia/Pacific 2011, OECD Publishing, Paris, https://doi.org/10.1787/9789264106154-en.

[14] Statistics Korea (2010), Outcome of Social Survey 2010: Family, Education, Health, Security and Environment, Daejeon (in Korean).

[16] UNESCAP (2016), Income Security for Older Persons in the Republic of Korea, https://www.unescap.org/resources/income-security-older-persons-republic-korea.


← 1. The OECD defines the normal retirement age in a given country as the age of eligibility of all schemes combined without penalty, based on a full career after labour market entry at age 22.

← 2. The average earnings of all contributors over the last three years are uprated by inflation growth to calculate the reference earnings. In times of positive wage growth the replacement rate will be slightly lower as a result.

← 3. The OECD calculates pensions on an annual basis using annual earnings as the reference base. Therefore the formula below have been modified from those normally referenced in Korea to account for this methodological difference, but the more commonly used formula are shown in the endnotes.

← 4. This formula is based on annual earnings and years of contribution. In Korea the calculation normally refers to monthly earnings and the formula is then given as Pension = 2.4(A+0.75B) * (1+0.05n/12), where A and B are monthly earnings and n is the number of months of contribution.

← 5. The corresponding monthly formula is Pension = [2.4(A+0.75B) ×P1/P +1.8(A+B) ×P2/P] (1+0.05n/12).

← 6. The corresponding monthly formula is Pension = [2.4(A+0.75B) ×P1/P +1.8(A+B) ×P2/P+ 1.5(A+B) ×P3/P+..+1.2(A+B)×P23/P] (1+0.05n/12).

← 7. The contribution base prior to 2009 was defined as 65% of gross pay, classified as “basic pay and a part of allowances” (GEPS, 2016[26]). In 2009 this was changed to full gross pay, therefore leading to a recalibration of the contribution rate. Prior to 2009 the equivalent contribution rate was around 11% having increased over time from about 3% of gross wage upon introduction.

← 8. To calculate the reference wage past wages were valorised with average wage growth within the civil service.

← 9. Prior to the 2015 reform only 33 years could be considered but as the accrual rate was 1.9% the maximum replacement rate was slightly higher at 62.7%.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2022

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at https://www.oecd.org/termsandconditions.