Chapter 2. First layer of social protection in old age

This chapter focuses on first-tier pensions in Portugal. These schemes help protect those with short contribution periods or low pension entitlements more generally against old-age poverty. In Portugal the main components of first-tier pensions consist of: the minimum pension from the earnings-related scheme, the means-tested old-age social pension and a supplement (Solidarity Supplement for the Elderly, CSI) for those whose total income is below a certain threshold taking into account various resources including descendants’ income. The chapter also investigates the interaction of the different layers of first-tier pensions and describes old-age poverty, old-age inequality and pension coverage gaps in Portugal. It concludes with policy recommendations to improve first-tier pensions.


2.1. Introduction

The pay-as-you-go defined benefit pension scheme in Portugal links benefits with past earnings. Eligibility for these pension benefits requires to have reached 15 years of contributions. Moreover, even when the minimum years of contributions are reached it can still be that earnings-related benefits are at a low level and need to be topped up.

First-tier pension schemes – the first layer of social protection in old age – help protect those with short contribution periods or low pension entitlements more generally against old-age poverty. In Portugal the main components of first-tier pensions consist of: the minimum pension from the earnings-related scheme – between 21% and 30% of the average wage; the means-tested old-age social pension at 11% of the average wage; and a supplement (Solidarity Supplement for the Elderly, CSI) up to 29% of the average wage for those whose total income is below a certain threshold taking into account various resources including descendants’ income. In addition to these three big components there are several additional supplements, including one which level varies whether the individual is younger or older than 70.

All first-tier pensions in Portugal are tax-financed. This is not unusual since first-tier pensions are not earnings-related and are therefore not closely linked to contributions. More precisely, the full old-age social pension is financed by taxes. Minimum-pension payments are the difference between the minimum pension level someone is eligible for based on her or his contribution period and the earnings-related entitlements. It is this top-up which is tax financed. This works similarly for the CSI.

When looking at the aggregate numbers, both in terms of number of recipients and of benefit levels, minimum pensions are by far the largest component. The minimum pension top-ups represented 0.5% of GDP in 20161, while total spending on minimum pensions including the earnings-related components was 2.6% of GDP. This compares to spending levels for the social pension and the CSI of 0.04% and 0.11% of GDP, respectively.

The nexus of first-tier pensions provides almost every individual older than the retirement age having no earnings-related pension with some income. However, the variety of schemes and eligibility conditions and the often complex interaction between them results in a difficult to navigate social protection system with large differences in outcomes between seemingly similar people.

This chapter focuses on schemes protecting the most vulnerable among the elderly in Portugal. The next section deals with old-age poverty and inequality followed by a description of coverage gaps and periods of low income. Then the design of first-tier pensions in Portugal is described focusing first on minimum pensions and next on safety-net features. This is followed by an investigation into the interaction of the different layers of first-tier pensions. The last section concludes with some policy options.

2.2. Old-age poverty and inequality

Relative old-age poverty and old-age inequality have fallen in Portugal over the past decade, but Portugal remains one of the most unequal European countries, both for the population as a whole and for the elderly in particular. The economic crisis which started in 2008 halted the steady decrease in both income inequality and old-age poverty, and increased working-age poverty. Therefore, despite the undeniable progress made at the beginning of this century, a significant share of the older population is still at risk of poverty and future elderly generations might face higher inequality (OECD, 2017[1]). European Commission (2018[2]) notes that, despite an improvement of retirement income adequacy between 2008 and 2016, the pension system does not ensure an adequate protection for all the elderly against poverty and social exclusion.

2.2.1. Relative old-age poverty in Portugal is slightly lower than the OECD average

The old-age poverty rate – defined as the share of individuals older than 65 having less than half the median disposable income for the total population – was more than halved, from 17.1% in 2004 to 8.1% in 2011, before rising again to 10.8% in 2015.2 It is slightly lower than in the OECD on average. It is also lower than the poverty rate for the entire Portuguese population. This relative poverty line amounted in Portugal to EUR 4 960 per year or EUR 413 per month in 2015. Breaking it down by age groups, the poverty rate among 66-75 year-olds was 9.4% compared to 11.1% for the OECD in 2015, while, for those older than 75, it is higher at 12.5% compared to the OECD average of 14.4% (Figure 2.1).

Figure 2.1. Portuguese relative old-age poverty rates are below OECD average
Poverty rates % of the population subgroup, 2015

Note: 2015 or latest data available. Relative old-age poverty defined as households receiving less than half the median disposable income.

Source: OECD Income Distribution Database.


Severe material and social deprivation at older ages, a notion that is more closely related to absolute poverty, paint a less favourable situation for Portugal than reflected by relative poverty. More than one in five retired women is identified as being severely materially and socially deprived (Figure 2.2).3 Among European OECD countries, only Greece, Hungary, Latvia and Lithuania have higher rates. As is standard, for both retired men and the employed the rates are lower, but they are still well above the OECD26 average (for those countries with comparable data) and the EU28 average.

Figure 2.2. Material and social deprivation is common among retired women in Portugal
Rates of material and social deprivation, %, 2016

Note: Severe material and social deprivation rate is defined as the proportion of people living in households that cannot afford at least five of thirteen essential items.

Source: Eurostat.


2.2.2. Old-age inequality is high in Portugal

Working-age inequality often transmits to old age. Adults with stable jobs, higher incomes and better access to health care are more likely to retire with adequate income (OECD, 2017[1]). People with unstable careers, low incomes and low-quality jobs, by contrast, tend to be in worse health, prone to unemployment and at greater risk of poverty throughout their lives and into old age. Temporary contracts are widespread in Portugal and the share of temporary contracts remained much higher than in the OECD on average (Chapter 5) despite the recovery from the economic crisis. Moreover, pension rules were tightened to improve financial sustainability and deal with public finance pressure (Chapter 3). Therefore, it is likely that old-age inequality will become a more pressing issue in the future.

Despite coming down from higher levels, Portugal still has the highest level of old-age income inequality among OECD EU countries, as measured by the Gini coefficient - which varies between 0 if everybody had the same income and 1 if all of the income went to only one person. The Gini coefficient among those over 65 was at 0.34 in 2015 (down from 0.37 in 2004) against about 0.30 in the OECD on average, ranging from 0.18 in the Czech Republic to 0.44 in Chile (Figure 2.3).4

In Portugal, the Gini coefficient is similar for older people as for the entire population. Only a few countries including Iceland, Slovenia, Sweden and Switzerland share such a pattern (Figure 2.3). This is relatively unusual because pension systems tend to level off inequality as people disengage from the labour market. Moreover, in Portugal, the high coverage of the minimum pension (Section 2.4) helps limit old-age inequality.

Earnings and income inequality in Portugal rose throughout the 1980s and 1990s before coming down slightly in the early 2000s (Cardoso, 1998[3]; Arnold and Rodrigues, 2015[4]). As OECD (2017[1]) shows, Portugal is one of the countries that largely transmits changes in working-age inequality into old-age inequality. Even when abstracting from career interruptions, more than 90% of increases in wage inequality are passed through into pension inequality against 67% for the OECD on average. This is because progressivity in the Portuguese pension system is very limited for full-career workers with replacement rates being similar for high and low earners, as in Italy, the Netherlands and Poland among others.

Figure 2.3. Old-age income inequality is high in Portugal
Gini coefficient (after taxes and transfers), 2015

Source: OECD Income distribution database.


2.3. Coverage gaps and low earnings

Two of the main reasons for having to rely on first-tier pensions are coverage gaps from the earnings-related schemes and low earnings throughout the career. On both fronts Portugal scores around the OECD average and significantly better than other Southern European countries.

2.3.1. Coverage gaps

One reason why people eventually might have to rely on first-tier pensions is limited career length. To be eligible for a contributory pension in Portugal someone needs to contribute for at least 15 years. Most people in Portugal do reach at least 15 years of contributions. By age 60, only 7% of the population had not reached that number of years in 2015, slightly below the 8% share for those aged 50 (Figure 2.4). These shares are close to the OECD average but much lower than in other Southern European countries and the Netherlands.

Figure 2.4. The incidence of very short careers is not particularly high in Portugal
Share of the population with less than 15 years of work experience, 2015

Note: Share of people with specific work experience based on survey data. This work experience might deviate from the social security contribution record as working in the shadow economy or abroad does not induce any domestic social security contributions while some non-employment spells (childcare, unemployment, disability) do. Due to data availability, Germany and Denmark are excluded from the OECD (unweighted) average.

Source: EU-SILC 2015.


Overall, the share of people reaching 15 years of contribution at different ages has been stable across cohorts (Figure 5.3 in Chapter 5). Moreover, among those who contribute, the age at which someone can be expected to have reached 15 years of contributions has been relatively stable over the past decade, close to 40 years. Indeed, on average both men and women between the ages 35 and 39 have reached 14 years of contributions in 2006 and in 2016 (Figure 2.5).

For the age groups 40 and over the average years of contributions has increased significantly, especially for women. This increase was realised despite the negative effect of the crisis on contribution records. However, the data for older age groups should be treated with caution since contribution records are missing – i.e. just not available even if there were actual contributions - for a significant part of the older population. At retirement those with missing records have to prove their contribution histories by other means, for instance by showing payslips. The reported increase in contribution years across cohorts is therefore likely to be overestimated.

More precisely, with the above caveat, around the statutory retirement age (65 in 2006 and 66 and two months in 2016) the average years of contributions sharply increased by about 8.2 years for women and 7.7 years for men between 2006 and 2016. Especially for women this makes a big difference since in 2006 they barely exceeded the minimum threshold of 15 years on average. By contrast, younger age groups have slightly lower average years of contributions in 2016 compared to 2006. This is likely caused by a combination of later entry because of longer education and falling employment rates among the youth, who were hard hit by the financial crisis.

Figure 2.5. The average years of contributions increased for most age groups, 2006-16
Average years of contributions by 5-year age group and gender for those with contribution records

Note: Records for older cohorts are incomplete. Average years of contributions are therefore likely underestimated.

Source: Portuguese social security data.


2.3.2. Low earnings

The share of low-wage earners came down sharply in Portugal. Employees earning less than two-thirds of the median wage represented 12% of total employees in 2014 against 21% in 2006 (Figure 2.6). This means that the share of low-wage earners in Portugal is now below the OECD28 average of 15%. This decrease may be partially explained by the substantial rise in the minimum wage from 47% to 58% of the median wage while the OECD average ratio is 51% (Figure 2.7).

Figure 2.6. Share of employees with low earnings has decreased significantly since 2006
Share of employees earning less than two-thirds of the median wage, %

Note: Only companies with 10 employees or more are included.

Source: Eurostat, Structure of earnings survey.


Figure 2.7. The minimum wage is relatively high in Portugal
Minimum wage as % of the median wage, 2016

Source: OECD Employment and Labour Market Statistics.


Even for someone who contributed for a large part of the working life, low lifetime earnings might still mean reliance on first-tier pensions in old age, specifically the minimum pension. Moreover, there is a wide disparity in the share of low earners by age, type of contract, education and gender. For example, employees under 30 are more likely to be low earners (Figure 2.8). This can simply reflect a life-cycle effect. However, current younger workers have been severely hit by the global crisis in Portugal, and it is therefore crucial to avoid hysteresis effects which raise the risks of getting stuck in low paying jobs and of relying on minimum pensions in old age. Also unsurprisingly, low-educated individuals and employees on contracts of limited duration are more likely to be low earners. Consequently limited duration contracts are likely to have a detrimental effect on pension entitlements. Finally, women are much more likely to be low earners, with a huge gender pay gap reflecting both differences in hourly wages and in working time. All these groups are therefore more likely to be vulnerable in old age and have to rely on minimum pensions.

Figure 2.8. Share of low earners by characteristics
Low earnings defined as earning less than two-thirds of the median wage, 2014

Source: Eurostat, Structure of earnings survey.


2.4. Minimum pension

The minimum pension covers those who are eligible for an earnings-related pension but whose entitlements fall below a certain threshold. In Portugal, the minimum pension benefit tops up the earnings-related pension to given thresholds which depend on the length of the contribution period.

Those who have contributed for at least 15 years to the main earnings-related pension scheme (regime geral de Segurança Social) are eligible to a minimum pension. The minimum contribution period for an old-age pension in general is also 15 years, but used to be 10 years before 1994. Those who reached this minimum requirement before the law changed in 1994 are still eligible for a minimum pension.5

The minimum pension level increases with the contribution period with the full minimum pension obtained with 31 years of contributions or more. The required number of years to get a full minimum pension is average compared to other OECD countries (Figure 2.9). In Portugal, the criteria used to validate contributions is accommodative as 120 days of contributions count as one contributory year.6 Multiple years with less than 120 days of contributions can be aggregated to form a full contributory year.

Figure 2.9. Minimum years of contributions to get a minimum pension, 2016

Source: OECD Pensions at a Glance (2017).


Table 2.1 shows how the length of the career affects the minimum pension amount. Someone with a career length of 31 years or longer gets a 38% higher minimum pension than someone with a career length of 15 years. Periods of non-employment covered under the social security systems – such as maternity, paternity, adoption, unemployment, sickness and family care – are considered contributory periods.7

Table 2.1. Minimum pension amounts, 2018

Contribution period

Annual minimum pension amount

10 – 15 years

EUR 3 767.12 (21% of average wage)

15 – 20 years

EUR 3 951.64 (22% of average wage)

21 – 30 years

EUR 4 360.58 (24% of average wage)

31 and more years

EUR 5 450.76 (30% of average wage)

Note: For those who fulfilled the previous minimum years of contributions requirement before 1994 a minimum pension can be obtained with between 10–15 years of contributions (EUR 3 767.12, or 21% of the average wage). The average wage in 2018 is based on preliminary data. The annual amounts take into account 14 payments, 12 for each month and two extra in June and December.

The level of the full minimum pension is significantly higher as a share of the average wage than in most OECD countries, 30% versus 24% on average for countries having a minimum pension scheme (Figure 2.10), which is one-quarter higher. Minimum pensions range from about 6% of the average wage in Latvia to 41% in the Slovak Republic and Turkey. Belgium and Mexico have a similar full minimum pension level as Portugal, relative to the average wage, but with a longer and shorter required contribution period, respectively.

Figure 2.10. Level of minimum pension
Minimum pension as % of average wage, 2016

Source: OECD Pensions at a Glance (2017).


The number of beneficiaries of the minimum pension is high in Portugal. In 2016, about 823 000 people aged 65+ received a minimum pension, which represents 38% of the 65+ population: about one-quarter of men and half of women aged 65 and over received a minimum pension (Figure 2.11). Among the four levels of the minimum pension, the larger share of women (18.7% of those older than 65) received the lowest minimum pension, meaning they had less than 15 years of contributions. However, the number of people receiving the lowest level is rapidly declining since fewer people are eligible for this minimum pension level since 1994.8 The three higher levels have all slightly gained in importance for women since 2006. The share of men older than 65 receiving the minimum pension has steadily declined from about 35% in 2006 to about 25% ten years later. The majority of men now receiving the minimum pension have validated more than 20 years of contribution.

In terms of the design of the minimum pension scheme, OECD countries can be roughly divided into four categories. The most common category - the Czech Republic, Hungary, Italy, Mexico, Poland, Slovenia, Spain and Turkey - only has only one minimum pension level. The minimum years of contributions that is needed is between 15 and 35 years of contributions depending on the country. Second, France and Switzerland have linearly increasing minimum pensions from the first quarter or year of contributions, respectively. Each additional quarter or year earns the same proportion of the full minimum pension. Third, other countries, like Belgium, Luxembourg and the Slovak Republic, have a combination of both: a minimum number of years of contributions needs to be reached to be eligible after which the benefit increases linearly (for the Slovak Republic at two different rates depending on the career length). Finally, Portugal and Latvia are the only two OECD countries that have discrete steps – both Portugal and Latvia have four levels. In Portugal this will be reduced to three once the last person covered under previous eligibility criteria from before 1994 passes away (i.e. the 10-15 year step will be eliminated).

Figure 2.11. A large share of the older population receives minimum pensions
Share of 65+ population receiving a minimum pension, by pension level and gender

Source: Own calculations based on data from the Ministry of Labour, Solidarity and Social Security, Portugal.


Discrete steps create discontinuities whereby a small difference in the contributory period can generate a big change in the benefit level. Thus, in Portugal, the minimum pension does not reward each additional year of contribution, therefore distorting labour supply incentives for low earners. Only those with contributory periods just below the thresholds of 15, 21 and 31 years benefit from working longer in terms of the minimum pension.

This was not always the case. While after 1974 a single minimum pension existed at 50% of the minimum wage, in 1997 18 different levels were introduced, and one additional lower level was added in 2002 for those with fewer than 15 years of contributions (Figure 2.12). Each year (between 30 and 40 years) or every two years (between 15 and 30 years) of additional contributions increased the minimum pension. These 19 different levels were reduced to four in 2016, and will be further reduced to three over time with the elimination of the lowest level.

Figure 2.12. The number of minimum pension levels was reduced over the years
Level of minimum pension, EUR per year

Note: Before 2006, 19 different minimum pension levels can be distinguished by contribution years: up to 15, 15-16, 17-18, 19-20, 21-22, 23-24, 25-26, 27-28, 29-30, 31, 32, 33, 34, 35, 36, 37, 38, 39 and 40 years of contributions.

Source: Ministry of Labour, Solidarity and Social Security, Portugal.


2.4.1. Indexation of minimum pensions

Before 2007, minimum pensions were indexed in a discretionary manner (as were all other pensions in payment), taking into account real GDP growth and price inflation but leaving the ultimate decision up to policy makers. However, since 2007 detailed rules are in place for the indexation of most pension benefits. Exceptionally, the values for most pensions were frozen during the crisis years as part of the structural adjustment programme of the Troika, with the exception of some of the lowest pensions to avoid significant real cuts for the poorest pensioners.

Central to the values of the various benefits is the so-called Social Support Index (Indexante dos Apoios Sociais or IAS). The IAS was established in 2007 and serves as a point of reference for determining, calculating and updating social security benefits. Its monthly value was EUR 429 in 2018. The different levels of the minimum pension, which in principle move in line with the IAS, for 15-20 years, 21-30 years and 31 and more years are currently equal to 66%, 72% and 90% of the IAS, respectively.

The indexation of the IAS depends in a complex way on real GDP growth and inflation (Figure 2.13). If real GDP growth is less than 2%, the IAS (and therefore the minimum pension) is indexed with inflation. If real GDP growth is between 2% and 2.5% the IAS is indexed with inflation plus 0.5 percentage point, if real GDP growth exceeds 2.5% the IAS is updated with inflation plus 20% of real GDP growth (Table 3.2, Chapter 3).9 Here again, this way of indexing benefits creates an undesirable discontinuity: a jump when real GDP growth is larger than the artificial 2% mark on top of the change in the slope once the 2.5% mark is exceeded. Overall, an indexation rule equal to inflation plus 15% of real GDP growth will be close to the current indexation while avoiding this weakness: this corresponds to the linear rule illustrated in Figure 2.14. It would result in slightly more generous indexation for real GDP growth rates of less than 2% offset by slightly less generous indexation for higher growth.

Figure 2.13. Indexation of minimum pensions

Note: The current rule is depicted by the lines with the step increase at 2% real GDP growth (dashed lines). The proposed linear rule (the solid line) depicts an indexation rule of inflation plus 15% of real GDP growth.

Source: Ministry of Labour, Solidarity and Social Security, Portugal and own calculations.


As a long-term trend, productivity gains result in increasing real wages. Given current indexation rules, this implies that minimum pensions will gradually fall relative to wages.10 Indexing minimum pensions to average wages is the only way to avoid this decline in “replacement rates” over time. However, wage indexation is costly in the long term. The longer someone lives, the larger the difference between benefits with wage and with price indexation becomes. Increasing life expectancy therefore makes this even more costly.

In this context, a good compromise worth considering is to distinguish the indexation of the minimum pension benefit that is paid at the time of retiring from the indexation of pension in payments. The former should be indexed to wages to avoid declining relative levels. The latter should be determined globally by the trade-off between income adequacy and financial cost. The negative side of such a framework is that recipients have a different benefit level depending on when they retired. But this reflects cohort effects driven by improving standards of living over time from one generation to the next.

The following example illustrates the extent to which the minimum pension level could fall in relative terms given current indexation rules. Based on the economic assumptions used in the OECD pension model, annual real wages grow by 1.25% in the long term, which implies that real GDP growth is below 2% and therefore that the IAS evolves in line with prices. For someone entering the labour market now, this means that at the time of retirement the full minimum pension will be reduced to 17% of the average wage against 30% today (Figure 2.14). If the indexation rules were applied over time, the minimum pension would, as shown below, lose most of its relevance.

Figure 2.14. Current indexation rules are likely to almost halve minimum pensions over time
Level of minimum pension, % of average wage, 2065 projections

Note: Real wages are assumed to grow with 1.25% per year, and GDP and the wage bill (W*L) are assumed to grow at the same rate. Since labour supply is expected to shrink due to ageing – for example based on UN projections the population aged between 20 and 69 years is projected to shrink by 35% within half a century - this means that GDP growth would be below 1.25%. Therefore, current indexation rules imply that minimum pensions are projected to be indexed by prices only.

Source: OECD calculations


2.4.2. Minimum pension levels and earnings-related entitlements

Assuming that the minimum wage and the minimum pensions remain at their current levels relative to average wages, someone with a full career at the minimum wage will roughly receive an earnings-related pension equal to the highest minimum pension: the lower two lines in Figure 2.15 coincide after a 42-year career.11 For shorter careers at the minimum wage, the minimum pension effectively tops up the earnings-related part. Hence, unless someone has a full career, minimum-wage earners will benefit from the top-up built into the minimum pensions.

For anyone earning the average wage or above it is unlikely that the minimum pension will be relevant. The entitlement of an average-wage worker (top line in Figure 2.15) will exceed the minimum pension level after 17 years of contributions, i.e. shortly after being eligible to the contributory pension. Entitlements steadily increase until someone reaches 40 years of contributions, after which accrual ceases but the pension increases because of more favourable reference earnings (Chapter 3 for a detailed description of the earnings-related pension). For someone in an intermediate situation earning three-quarters of the average wage, minimum pension levels are exceeded after 21 years of contributions.

Figure 2.15. Minimum pension and full-career earnings-related pensions
Entitlements by contribution period, % of average wage, in gross terms

Note: The minimum wage is 42% of the average wage in every period considered. For this chart it is also assumed that minimum pension levels evolve over time in line with wages. Projections are based on someone entering the labour market at age 20 (period 0) in 2016 and retiring at age 68 (period 48).

Source: OECD pension model.


2.5. Old-age safety net

The old-age safety net helps protect those with no employment history or very short contribution periods. The Portuguese old-age safety net consists of a wide range of schemes. The most important are the old-age social pension (Pensão Social de Velhice) and the Solidarity Supplement for the Elderly (CSI). The old-age social pension is a flat-rate benefit for people with no or very low income while the CSI is a top-up for people with income up to a higher threshold but with broader means-testing. Over the last decade, between 1% and 1.5% received the old-age social pension and between 8% and 12% of people older than 65 received the CSI. On top of the social pension someone automatically receives the extraordinary solidarity supplement (CES) and can also potentially get dependency complements, additional health benefits and financial support to pay for energy cost.

Some of these benefits can or cannot be combined. This mix of safety net features can be difficult to navigate through for eligible persons. To apply for these benefits potential recipients have to fill out an array of forms. Since means-testing applies to most of these safety-net features this can lead to significant delays in receiving the benefits. European Commission (2018[2]) suggests to strengthen the transparency, harmonisation and simplification of the pension system, namely as regards the coexistence of diverse means-tested mechanisms, and to assess the efficacy of the existing means-tested benefits as regards their major goal of tackling poverty among the elderly.

2.5.1. Old-age social pension

For those who are not eligible to a contributory pension, eligibility to the social pension requires to:

  • have reached the official retirement age.

  • have a monthly gross income not exceeding 40% (for a single) or 60% (for a couple) of the IAS (EUR 428.90 in 2018). This corresponds to less than EUR 171.56 (or 13.2% of the gross average wage) for a single person and EUR 257.34 for a couple.

  • be a national citizen residing in Portugal or be a foreign citizen residing in Portugal and covered by the EU social security regulations (EU Member States, Iceland, Liechtenstein, Norway and Switzerland) or international social security instruments in force in Portugal (Australia, Brazil, Cape Verde and Canada).

  • not be covered by any compulsory social protection scheme or transitional rural schemes.

The monthly amount of the old-age social pension in 2018 was EUR 207.01, its value being linked to the IAS. This means that, given current IAS indexation rules, the value of the old-age social pension will gradually fall relative to wages. The old-age social pension is topped up by the extraordinary solidarity supplement (CES), which varies according to age (EUR18.02 per month for someone younger than 70, EUR 36.02 from the age of 70). Therefore, the full benefit of the old-age social pension with supplement is EUR 225.03 for someone younger than 70 and EUR 243.03 for someone older than 70. This corresponds to 17.3% and 18.7% of the gross average wage, respectively.

The old–age social pension in Portugal is not a top-up but a fixed amount, regardless of how much someone is under the income threshold. This can lead to the strange situation in which someone with income just below the threshold ends up with a higher income than someone with income just above the threshold. In addition, the CES top-up makes the old-age social pension unnecessarily complicated. Even if it were considered politically desirable to have a higher benefit for those older than 70, at least the top-up paid to those younger than 70 could be merged with the social pension. The reason why the CES is separated from the old-age social pension is that other social security benefits are linked to the value of the old-age social pension. An increase in the old-age social pension therefore automatically leads to an increase in the linked social security benefits. This should be reformed as, given that the receipt of the CES is automatic, the integration would simplify matters.

2.5.2. Solidarity Supplement for the Elderly (CSI)

In addition to the old-age social pension the Solidarity Supplement for the Elderly (CSI) – a cash support paid monthly to low-income elderly people – was introduced in 2006. To be eligible for the CSI someone must:

  • have reached the official retirement age.12

  • have lived in Portugal for at least six consecutive years on the date of application.

  • have an annual income of less than EUR 5 175.82 in 2018.

  • have an annual income of less than EUR 9 057.69 for a couple in addition to the individual threshold in 2018.

The CSI is a top-up, paid up to the means-testing threshold of EUR 5 175.82 (28.5% of the gross average wage) for a single and EUR 9 057.69 for a couple (50% of the gross average wage). These figures are updated on a discretionary basis.13

The CSI means-testing takes into account not only all income sources of the household but also the income of descendants with whom the applicant is in contact with (regardless of whether they live in the same household). The means-testing of the CSI is therefore more extensive than a typical old-age safety net, the reasoning being that descendants with high income can financially take care of their parents.14 Around 18% of the beneficiaries who received the CSI in 2018 had their entitlements reduced because of their descendants’ income.

The law is meant to convey the idea that it is socially just to oblige children who can afford it to take care of their parents. However, this might in practice lead to a perpetuation of low disposable income from one generation to the next within the same family. Children with parents with high pensions will never have to contribute to their care while children with poorer parents will have to contribute financially if they managed to climb the social and economic ladder. This makes the existing link between parents’ income and children’s income analysed in OECD (2018[5]) even stronger, running against the potential objective of enhancing social mobility. By contrast, financing such benefit by (progressive) taxes ensures that the broadest shoulders bear the greatest burden without having to resort to means-testing of descendants.

The eligibility criteria above are listed in a simplified way for presentation purposes. In fact, the real list is very cumbersome, and the induced complexity in trying to understand it is amplified as available official information through the law, the social security website and official guidelines is not presented in the same way. Hence, for the elderly, applying to the CSI and determining with which benefits it can be combined can be a daunting task since receipt of the CSI depends on many criteria.15 Having another instrument, and such an intricate one, on top of the other old-age safety nets creates complexity, calling for a simplification of the system.

There might be severe problems with the take-up rate, possibly due to the complexity and opacity of the CSI, which might generate large inequality even among vulnerable people. The level of the CSI benefit is around the relative poverty line defined as half the median equivalised income for the total population: for the latest year for which data are available for the median equivalised income (2015), the poverty line stood at EUR 4 960 per year while the CSI was EUR 4 909. Yet, the poverty rate of those over 65 was larger than 10% in 2015 and the poverty depth was 20.9%, meaning that those with income below the poverty line had an average income 21% lower, i.e. EUR 3 923, way below the CSI level. Somehow, even with benefits close to the poverty line the CSI does not reach a significant proportion of those who need it.

Figure 2.16 shows the benefits received from first-tier pensions for individuals who never contributed. Taking into account only the old-age social pension (plus the CES for those younger than 70), with benefits equal to 17.3% of the average wage, Portugal ranks below the OECD average of 21.0%, with a large range from less than 10% of the average wage in Hungary, Korea, Latvia, Mexico and Turkey to more than 30% in Canada, Denmark, Ireland, New Zealand and Norway (Figure 2.16).16 However, individuals with no income of their own and no descendants with sufficient income also receive the CSI, which increases the safety net to 28.5% of the average wage, well above the OECD average and a level exceeded among European countries in only Austria, Denmark, Ireland, Luxembourg and Norway.

Figure 2.16. Non-contributory pension benefit level
% of average wage, 2016

Note: The non-contributory pension benefit level assumes no other sources of income, therefore, means-testing has no impact. For Portugal, the social assistance level is the CSI threshold.

Source: OECD, Pensions at a Glance 2017.


2.6. Interactions between first-tier pension elements

The minimum pension and social safety-net levels evolve in principle in line with the IAS. This is what happened until 2010. However, as discussed above, due to the economic crisis, indexation rules were temporarily suspended with the exception of the lowest pensions, which led to a relative increase in the levels of the old-age social pension and of the lowest minimum pension compared with the highest minimum pension level (Figure 2.17). Between 2006 and 2017, the average wage was broadly stable in real terms. The minimum wage increased much faster, by 25% in real terms and the average pension, by about 10%. The CSI ceiling, which is updated on a discretionary basis, rose by about 5% (in real terms) while the full minimum pension fell by 3.5%.

Among first-tier benefits the minimum pension is by far the most commonly received. As discussed before, around 38% of people over 65 received a minimum pension in 2016, having come down from 45% in 2006 (Figure 2.18).17 In addition, after its introduction in 2006, the number of CSI beneficiaries as a share of the 65+ population quickly exceeded 12% before levelling off around 8% in 2016 and 2017. In 2017, around half of CSI recipients are receiving some minimum pension benefits. This also means that about 10% of those receiving the minimum pension are topped up by the CSI. The share of people receiving the old-age social pension is very small, between 1.2% and 1.5% of the population over 65 over the last decade, highlighting that most beneficiaries of the old-age safety nets have income between the levels of the social pension threshold and the CSI threshold.

Figure 2.17. First-tier pensions lag behind real minimum wage growth
In real terms, index 2006=100

Source: OECD and Social security statistics Portugal.


There are numerous non-contributory schemes pursuing similar objectives, and their interaction is unduly complex. Despite the CSI having different eligibility criteria it largely has the same objective as the old-age social pension. At the same time the CSI benefit ceiling exceeds most of the minimum pension levels and is about twice the level of the social pension. This gives the impression that, with the introduction of CSI in 2006, rather than adjusting existing eligibility criteria or benefit levels to changes in social and economic conditions, additional schemes have been piled up.

How the social pension, the CES and CSI complement initial income is shown in Figure 2.19, Panel A. Ignoring the means-testing feature related to the income of descendants, someone with low income levels receives both the social pension and the top up from the CSI. As income before transfers grows the flat-rate social pension stays the same, eventually leading to total income exceeding the CSI level. In total someone younger than 70 with income just below the social pension threshold (EUR 2 402) would receive a yearly social pension combined with the CES of EUR 3 150 on top, providing a total income of EUR 5 552. If someone earns slightly more they are only eligible for the CSI (again assuming no other sources of income and no income of descendants) which would top up their income to EUR 5 176, i.e. 7% less. This Figure shows that apart from this peculiarity, the social pension and the CES does not add anything to the CSI - at least in this case when descendants’ incomes are ignored.

Figure 2.18. Recipients of the minimum pension and the Solidarity Supplement
Share of population 65+

Source: Own calculations based on data from the Ministry of Labour, Solidarity and Social Security, Portugal.


Figure 2.19. Total income before and after social transfers
EUR per year

Note: CES supplement for someone younger than 70 is included in social pension. It is assumed there is no income of descendants. Panel B is for illustrative purposes only, it does not reflect policy recommendations on specific benefit levels.

Source: Own calculations based on data from the Ministry of Labour, Solidarity and Social Security, Portugal.


There is therefore a strong case for integrating the social pension, the CES and the CSI. This can easily generate the same benefit levels (ignoring descendants’ income) while drastically simplifying the whole nexus. For instance, the current level of the CSI can be taken as a top-up fully means-tested at the household level. However, the current schedule implies that pre-transfer income is withdrawn at a dissuasive rate of 100% until the CSI threshold is reached. That is, before the CSI threshold is reached a higher income before transfer does not lead to any increase in disposable income. A more appealing alternative would consist in ensuring that the income after transfer rises with the initial income by withdrawing the means-tested benefit at a lower rate against own income, as shown for illustration purposes in Figure 2.19, Panel B based on a withdrawal rate of 75%. This would provide incentives to earn income including through contributing to the earnings-related pension. This can be done either by lowering the level of the means-tested benefit – currently much higher than the OECD average (Figure 2.16) – or increasing the CSI level (and therefore minimum pensions, see below) if resources were available.

At 28.5% of the gross average wage, the current CSI ceiling is very close to the full minimum pension which is only available after 31 years of contributions (Figure 2.20). This implies that the CSI almost eliminates entirely the step pattern of the minimum pension schedule and substantially reduces the role of minimum pensions. Indeed, since the CSI is not linked to contributions and since the highest minimum pension is only marginally higher there are very limited additional minimum pension entitlements generated by contributing on earnings. Overall, the CSI resembles a flat-rate benefit, leading to an almost doubling of benefits from the old-age social pension.

For people with less than 15 years of contributions, the social pension (plus CES) topped up with the CSI is the only source of income (Figure 2.20), when excluding other sources such as capital income and ignoring descendants’ income for the means-test. After that individuals are eligible for the first step of the minimum pension which is also topped up by the CSI. This still holds when reaching the second step of the minimum pension and it is only after 31 years that someone receives a marginally higher income solely from the full minimum pension.

In order to simplify the current framework and reward contributions, on top of the integration of the old-age social pension (plus the CES) and the CSI discussed above, the minimum pension could increase steadily such that each year or month of contribution generates some first-tier pension entitlements. This is illustrated in Figure 2.20 for example in the case where minimum pension entitlements grow linearly with the contribution period while the minimum pension at 40 years of contributions remains at the current level of the full minimum pension. This linear minimum pension is topped up by the integrated safety net (up to the first 37 years of contributions in the illustrated schedule). The shown schedule is constructed as an illustration by assuming that someone with no contributions gets a benefit of 21% the average wage, similar to the OECD average. This initial benefit is then assumed to be withdrawn at a rate of 75% against the linear minimum pension, ensuring a steady progression with the number of contribution years. In the shown schedule it is fully withdrawn after 37 years, from which total income is determined by the minimum pension only (or the earnings-related pension if higher of course). The levels of all these parameters could be set as needed, weighing pension adequacy and financial cost. Beyond enhanced work incentives, the implementation of these mechanisms would smooth the overall benefit pattern for low-income pensioners.

Figure 2.20. Only the highest level of minimum pension exceeds the CSI
First-tier pensions by contribution period, share of the average wage

Reading note: The current system is represented by the minimum pension step function, the social pension (including CES) and the CSI benefit. In the illustration shown in the chart, the minimum pension grows linearly with the career length from the first year of contributions to reach EUR 6 480 for someone with a 48-year career. On top of that for someone with no other sources of income there is a safety net top-up, starting from EUR 4 000 and withdrawn at a rate of 75% against minimum pension income. Hence, for a 25-year career, as an example, the illustrated schedule includes a minimum pension of EUR 3 375 which is topped up to the level of the integrated safety net of EUR 4 744 for this level of income (Figure 2.18, Panel B).

Source: OECD calculations.


2.7. Policy options

A significant share of the Portuguese population benefits from minimum pensions. Among them, about 10% receive additional income from the Solidarity Supplement (CSI). On top of this a wide range of safety-net benefits support the most vulnerable. These benefits often pursue similar objectives making them potentially redundant; in terms of spending at least they represent small amounts. CSI effectively plays the largest role within the old-age safety net, but eligibility criteria are complicated and the relatively high level of CSI in combination with a 100% withdrawal rate means that a wide range of pre-transfer income does not generate any increase in disposable income. In short, while first-tier pensions in Portugal improve the living standards of older people, they can be streamlined.

A comprehensive reform would consist in consolidating the old-age social pension, the extraordinary solidarity supplement (CES) and the CSI. Merging these schemes would enable every old-age person with low income levels to receive the merged benefit. The means-tested consolidated benefit should be withdrawn at a much lower rate than 100% against other income (including income from the minimum pension) to maintain incentives to contribute.

Means-testing should be done at the household level only. Means-testing the CSI not only at the household level but also including children’s and grand-children’s income might cut cost but contributes to perpetuating low disposable income across generations within the same families, makes the pension system complicated and defeats the purpose of a formal pension system. Indeed, one of the reasons to introduce a formal pension system is to avoid that people have to rely on family members in old age. In addition the administrative complexity for pensioners to apply for the CSI implies that some people who are eligible might miss out on receiving it while others might be affected by long waiting times. Furthermore, this also generates administrative costs.

The current levels of the various schemes imply that for the most part minimum pensions are offset by the CSI, leading to small additional income. Moreover, in itself, the minimum pension in Portugal provides very little additional pension benefits from working longer when workers are far from one of the career-length thresholds. In order to solve this, beyond withdrawing the consolidated safety-net benefit at a slower rate than the current 100%, the minimum pension level should increase gradually with the length of the contribution period such that each year of additional work is rewarded by a higher minimum pension. Moreover, even if average contribution periods are rising, the minimum pension (and earnings-related pensions in general) should be available at a lower level to those who have fewer than 15 years of contributions.

The indexation of the minimum pension should distinguish between uprating the initial minimum pension level that people get at the time of retirement and the indexation of pensions in payment. The uprating of the initial benefit should be closely related to average- or minimum-wage growth in order to ensure a stable level relative to labour income over time. The indexation of pensions in payment should be determined by trading-off its generosity and its financial cost. For a given total cost, a more favourable indexation implies a lower initial benefit level and vice-versa.

Key recommendations

  • Simplify non-contributory benefits to avoid the multiplication of instruments with similar objectives. In particular, merge the old-age social pension, the complement (CES), and the top-up (CSI), and remove the CSI’s means-testing to descendants’ income.

  • Avoid withdrawing safety-net benefits such as the CSI against other income at a 100% rate. Instead withdraw the CSI or ideally the consolidated benefit recommended above at a much lower rate.

  • Lower the minimum contribution period of 15 years required for the minimum pension, adjust the benefit level accordingly and ensure that each additional year of contribution results in a higher minimum pension benefit.

  • Separate the uprating of minimum pension benefits available upon retiring, which should be as close as possible to wage indexation, from the indexation of minimum pension in payments, which should be identical to the indexation of earnings-related pensions.


[4] Arnold, J. and C. Rodrigues (2015), Reducing Inequality and Poverty in Portugal, (accessed on 16 August 2018).

[3] Cardoso, A. (1998), “EARNINGS INEQUALITY IN PORTUGAL: HIGH AND RISING?”, Review of Income and Wealth, Vol. 44/3, pp. 325-343,

[2] European Commission (2018), The 2018 pension adequacy report - Current and future income adequacy in old age in the EU. Volume II, Country profiles, European Commission,

[5] OECD (2018), A Broken Social Elevator? How to Promote Social Mobility, OECD Publishing, Paris,

[1] OECD (2017), Preventing Ageing Unequally, OECD Publishing, Paris,


← 1. Of this 0.5% of GDP, the minimum pension directly related to the earnings-related scheme represents 0.4 percentage points, while the remainder refers to minimum pensions related to survivor and disability pensions.

← 2. This poverty measure is different from the definition used by Eurostat, which uses 60 % of the national median equivalised disposable income after social transfers. The median disposable income increased a little during the crisis.

← 3. Severely affected by a lack of resources or social activities. Eurostat defines the severe material and social deprivation rate as the proportion of people living in households that cannot afford at least five of the following thirteen items: face unexpected expenses; one week annual holiday away from home; avoid arrears (in mortgage, rent, utility bills and/or hire purchase instalments); afford a meal with meat, chicken or fish or vegetarian equivalent every second day; keep their home adequately warm; a car/van for personal use; replace worn-out furniture; replace worn-out clothes with some new ones; have two pairs of properly fitting shoes; spend a small amount of money each week on him/herself (“pocket money”); have regular leisure activities; get together with friends/family for a drink/meal at least once a month; have an internet connection.

← 4. Income inequality for the working age population (15-64) followed largely old-age income inequality: coming down in the early 2000s before flattening out during the crisis.

← 5. Also those who enter retirement through invalidity need fewer minimum years of contributions, 3 and 5 years for fully and partially disabled respectively.

← 6. In the case of part-time work a full working day consists of 6 hours of work.

← 7. In the past military service was included as well.

← 8. However, anyone who fulfilled the old qualifying period criteria prior to the change in the law can still receive this minimum pension.

← 9. The brackets in the law are not literally defined in the same way, but boil down to the same formula. The law states that a) if the average real GDP growth is less than 2%, the IAS is indexed with inflation, b) if the average real GDP growth is equal to or greater than 2% and less than 3%, the IAS is indexed with inflation plus 20% real GDP growth, with a minimum of inflation plus 0.5 percentage points and c) if the average real GDP growth is equal to or higher than 3%, the IAS is indexed with inflation plus 20% real GDP growth. Part of the second bracket b) can be included in the third bracket c) since real GDP growth rates between 2.5% and 3% will push indexation over the lower limit of inflation plus 0.5 percentage points. Therefore, the rule for this range is equal to the rule for bracket c). Growth rates between 2% and 2.5% will by definition lead to indexation of inflation plus 0.5 percentage points.

← 10. Even if more recently this has not been the case (see Section 2.2)

← 11. If minimum pensions are indexed to prices only, the minimum pension would be largely irrelevant in 2065. At 12% of the average wage, the first step of the minimum pension would exceed entitlements for someone with 15 to 20 years of contributions at the minimum wage. Afterwards, entitlements would always be larger than the minimum pension.

← 12. In 2018 it was possible to combine the CSI with an early pension claimed after 2014 (Chapters 3 and 5). However, the CSI benefits was only available in 2018. This measure was introduced to compensate for the penalties levied on early pensions during the financial crisis.

← 13. The value of the CSI is updated periodically by joint ordinance of the Minister of Finance and the Minister of Labour and Social Solidarity, taking into account the evolution of prices, economic growth and the distribution of wealth.

← 14. According to Portuguese law descendants are legally obliged to take care of their parents when needed.

← 15. For instance, within the official guidelines, in the description of the eligibility criteria it seems that someone who is eligible for the social pension cannot be eligible for the CSI. The guideline (on page 5) states as a necessary condition: “ser cidadão português e não ter tido acesso à pensão social por ter rendimentos acima do valor limite de EUR 171,56 se for uma pessoa ou de EUR 257,34 se for um casal” (translation: be a Portuguese citizen without access to the social pension because of income above of the threshold of EUR 171.56 for a single or EUR 257.34 for a couple). However, the section describing which other benefits the CSI can be combined with states that the CSI can be combined with the social pension. The guidelines (on page 7) state: Pode acumular [CSI] com: 1) Pensão de Invalidez (a partir de outubro 2018) e Velhice do Regime Geral; 2) Pensão de Sobrevivência; 3) Pensão Social de Velhice; 4) Prestação Social para a Inclusão, desde que estejam reunidas as demais condições de atribuição do CSI (idade, recursos e residência em território nacional). In English: [the CSI] can be combined with 1) a disability pension (as of October 2018) and an earnings related pension; 2) a survivor pension; 3) the old-age social pension; 4) the Social Provision for Inclusion, provided that the other eligibility criteria of the CSI (age, resources and residence in national territory) are met. . In practice, the source of income is irrelevant as long as the means-testing criteria are met.

← 16. For some countries, this includes a residency-based basic pension. In these countries people accrue basic pension rights solely by living in that particular country.

← 17. Those receiving a minimum pension while being younger than 65 are not included in this statistic.

End of the section – Back to iLibrary publication page