# 1. Mandatory earnings-related pensions

In the past decades the Czech society and economy and with it the Czech pension system have been significantly transformed. In 1989, Czechoslovakia returned to democracy in the Velvet Revolution, in 1993 the Czech Republic and the Slovak Republic split into two independent countries and, in 2004, the Czech Republic joined the European Union.

The pension reform process started immediately after the political changes in 1989.1 Czechoslovakia wanted to create a unified social security system, which would provide health, sickness and pension benefits. The three labour categories that existed under communist rule were unified and the self-employed were included in the pension system. In addition, the pension system shifted from a tax-financed to a contribution-financed scheme in 1992. The inception of the new pension system took place in 1995 with the enacting of the Pension Insurance Act. Since then there have been significant reforms in 2003, 2008, 2011 and 2016.

While various changes have been implemented over the last 25 years, the architecture and principles in the design of the Czech pension system have been mainly unaltered since the 1995 Pension Insurance Act. The mandatory contributory system consists of an earnings-related component and a basic, flat-rate component. The earnings-related component is calculated by multiplying the reference wage with total accrual. A very progressive formula is used to calculate the reference wage as a function of wages throughout the career, effectively reducing accrual rates for high average wages. Some non-employment spells count towards accrual while others do not and enter on top as zeros in the calculation of the reference (average) wage. Everyone who is eligible for the earnings-related component receives the basic pension.

In 2019, the Minister of Labour and Social Affairs established the Commission for Fair Pensions, tasked with reforming the pension system to ensure a financially sustainable pension system that at the same time delivers adequate pensions. The Commission consists of representatives from academia, public bodies, social partners, political parties and interest groups. In January 2020, it published its first pension reform proposal.

The structure of the rest of the chapter is as follows: in Section 1.2 the most recent changes in the Czech pension landscape are described. Section 1.3 discusses current outcomes of the Czech pension system. Section 1.4 describes and evaluates its rules while Section 1.5 describes the proposal for a pension reform by the Czech Commission on Fair Pensions. Finally, 1.6 concludes and presents policy options.

Currently the population of the Czech Republic is slightly older than the OECD on average. Population ageing is accelerating in the Czech Republic at a similar pace as the OECD on average. Over the last 40 years, the old-age to working-age ratio – the number of people older than 65 years per 100 people of working age (20 to 64 years) – increased by a little more than 40% in the Czech Republic from 24 in 1980 to 34 in 2020 (Figure 1.1). By comparison, the old-age to working-age ratio for the OECD on average rose from 20 to 31 over the same period. Over the next 40 years, it will rise in the Czech Republic by more than 75% to a projected 60 in 2060 (against 58 for the OECD on average). This rapid shift in the demographic structure is the result of rising life expectancy and low fertility rates.

The working-age population (20-64) is projected to decrease by 10% in the OECD on average by 2060, i.e. by 0.26% per year. However, the fall will be almost double that in the Czech Republic (Figure 1.2). This might have a significant impact on both output growth and on the financing of its pay-as-you-go (PAYGO) system. Lower labour force growth negatively affects internal rates of return of PAYGO pensions, which might weigh on both pension adequacy and financial sustainability.

However, changes in the size of the 20-64 population is a demographic measure only, based on fixed age boundaries. Effective age boundaries, based on effective age of entry in and exit from the labour market, can change as a result of education and pension policies in particular, including measures affecting retirement ages. Moreover, employment rates within specific age groups can change, which might offset part of the negative effects of a shrinking working-age population. Employment of older workers is more directly influenced by pension policies, and is often critical as large margins exist in many countries to expand employment.

Since 2000, labour market participation among older individuals has increased sharply in the Czech Republic while unemployment among this group has remained low. This is a major achievement which will improve pension adequacy while somewhat alleviating pressures on pension finances.2 The employment rate among individuals aged 55 to 64 increased by almost 30 percentage points, from 36.3% in 2000 to 65.1% in 2018, compared to an increase for the OECD on average from 43.9% to 61.5% (Figure 1.3). During the same period, the employment rate among people aged between 25 and 54 in the Czech Republic increased from 81.6% to 87.5%.

Despite this sharp increase, employment rates fall sharply after age 60 in the Czech Republic (Figure 6.6 in (OECD, 2019[1])). The employment rate for the age group 55-59 is well above the OECD average (86% vs 73%), but employment rates for the age groups 60-64 and 65-69 are only 46% and 14% in the Czech Republic, respectively. The OECD average falls by less to 50% and 22%. More than 70% of the age group 60-64 are employed in Iceland, New Zealand and Sweden, and more than 50% of the age group 65-69 in Iceland. Only a small part of the decline in employment after age 60 in the Czech Republic, as in most OECD countries, can be explained by the deterioration of health with age (Chapter 5 in (OECD, 2017[2]). There are therefore large potential gains to be made in the Czech Republic.

As a result of the sharp fall in employment rates after age 60, the effective age of labour market exit is still low in the Czech Republic. In 2018, men exited the labour market on average at exactly the normal retirement age of 63.2 years (Figure 1.4) while women exited two years earlier on average at 61.3 years. Women with children being allowed to retire earlier contributes to this. The effective age of labour market exit was substantially higher in the OECD on average, at 65.4 and 63.7 years for men and women, respectively.3

Over the last decades, the main reforms of the Czech pension system took place in 1995, 2003, 2008, 2011 and 2016. Table 1.1 provides a summary.

In a major overhaul of the Czech pension system, the 1995 Pension Insurance Act established a two-tier contributory benefit structure: a flat-rate part (contribution-based basic pension) and an earnings-related part. The retirement age was gradually increased, from 60 years for men and from 53-57 years for women, depending on the number of children, to 62 and 57-61 years in 2007, respectively.

At the same time, two generous early retirement possibilities were introduced. One was available two years before the statutory retirement age, with the benefit being reduced until reaching the retirement age by 1% for each 90 days of early retirement. Alternatively, early retirement was possible three years before the statutory retirement age with a permanent reduction of 0.6% of the reference wage for each 90 days of early retirement. The penalties were increased in 2001 to 1.3% for temporary benefit reductions and to 0.9% for permanent benefit reduction for each 90 days. The temporary benefit reduction option was abolished in 2007.

In 1995, the reference wage for the earnings-related component was the average wage over the last ten years before retirement. This period has been increasing by one year each year until reaching the average lifetime wage in 2032. Past wages were and still are uprated with average-wage growth.

Before 1995, pensions in payment were indexed on a discretionary basis. In 1995, a minimum indexation formula was introduced, indexing pensions with 100% inflation and one-third of real-wage growth. However, the government could still increase pension benefits with more than the statutory minimum without any upper limit. In 1998, as part of public saving measures, the indexation rule became less generous, using 70% inflation plus one-third of real-wage growth. This was reverted to 100% inflation and one-third real wage growth in 2003.

In the same year, it was decided that the retirement age would be raised to 63 years for both men and women without children, starting the increase from 2007. Pension contributions were raised to 28%. As for self-employed workers, who were previously allowed to decide the contribution base themselves, a minimum contribution base was introduced and gradually increased between 2004 and 2006 from 35% to 50% of taxable income.

In 2008, the minimum years of contributions to receive an old age pension were raised from 25 to 35 years when including some eligible validated periods or to 30 years without. Prior to 1996, secondary education and tertiary education were considered validated periods, and prior to 2010 tertiary education only was considered a validated period. The retirement age was gradually raised to 65 years for men, women without children and women with one child, while lower retirement ages were maintained for women with two or more children. Early retirement can be taken three years before the statutory retirement age or from age 60, whichever is earlier. This means that the number of years of early retirement gradually increases from three to five years as the statutory retirement age approaches 65. As for late retirement, further benefit increases were introduced when people combine pension receipt and work after the statutory retirement age. The age of entitlement to a survivor pension for women (age 55) and men (age 58) was unified and set at four years before the statutory retirement age for men of the same birth year. A cap was introduced for annual pension contributions at 48 times the monthly average wage (i.e. 4 times the annual average wage in case of a constant monthly wage).

New reforms were decided as the thresholds used in the reference-wage formula to reduce accruals for high wages were contested as discrimination against higher income categories at the Constitutional Court in 2011. The Constitutional Court ruled that social insurance has a different function from private insurance and that therefore redistribution within the system is not unconstitutional. However, it also ruled that the income thresholds in the progressive reference-wage formula were in violation of the constitutionally guaranteed right to adequate income replacement. It expressly stated that a solution should be found by adjusting existing parameters rather than a systemic change. As a response, the pension law was amended. The first threshold below which 100% of the wage is taken into account was gradually raised from 35% of the average wage in 2011 to 44% in 2015, and the second threshold was raised from 90% to 400% of the average wage. However, the share of the wage taken into account between the first and second threshold was reduced from 30% to 26%, and above the second threshold wages were not taken into account (previously 10% was taken into account). The thresholds, which previously were fixed on a nominal basis and discretionarily increased, were fixed as a share of the average wage. Similarly, the flat-rate component was fixed at 9% of the average wage.

In addition, it was decided to continue raising the retirement age after age 65 (reached in 2030 for men) by two months every year without any prescribed limit. The penalty rate for early retirement of more than one year increased from 0.9% to 1.2% of the reference wage for every 90 days. Finally, the government’s upwards discretion about indexation was removed.

A funded voluntary retirement savings scheme was introduced in 2013. It consisted of transferring 3 percentage points of employee’s pension contributions to a private individual retirement account, under the condition that the employee contributed an additional 2% of gross salary. The scheme was voluntary, but the decision to participate was irrevocable. Participation in the voluntary scheme meant a reduction in the mandatory earnings-related pension. Since part of total pension contributions went to a private pension fund the accrual rate of the earnings related pension was reduced from the usual 1.5% to 1.2% of the reference wage. Entry of new participants stopped in mid-2015 and the scheme was discontinued at the end of that year (see Chapter 4).

In 2016, discretion was reintroduced in the indexation rule, giving the government the option to increase pensions by up to 2.7%. In addition, from 2017, the inflation measure that is used within the indexation rule is the highest of CPI inflation and Pensioner-specific inflation. Finally, the share of real-wage growth included in the indexation formula was increased from one-third to one-half.

In 2016 as well, the cap on the future statutory retirement age was re-established at age 65, reversing the previous decision to continue raising the retirement age two months every year.

In 2018, the basic pension was raised from 9% to 10% of the average wage and a special extra bonus of CZK 1 000 (3.1% of the average wage) was introduced for all people above age 85.

In 2018, around 80% of the population aged 20-64 contributed to the pension system (Table 1.2). About 10 percentage points of those were self-employed. With increases in employment rates, the total share of the working-age population who contributed rose over time from just over 70% in 2010. Over the same period, the statutory retirement age for men rose by one year and two months and for women by two years, also contributing to this increase.

The majority of pension recipients receive an old-age pension that is claimed at or after the retirement age. In 2018, 70% of male and 73% of female pension recipients (excluding disability pensions) received an old-age pension at or after the retirement age (Figure 1.5). However, in 1996 these shares were larger than 90%. Early pension receipts have steadily become more important rising from less than 1% – as early retirement was only introduced in 1996 – to 30% and 25% for men and women, respectively. These recent shares, applying to all retirees (i.e. to the stock of recipients), are very large. Since 1996, early pensions have hovered between 20% and 30% of newly granted pensions. Finally, the share of women only receiving a survivor pension has dropped from 7% to 2%. However, the share of women receiving a survivor pension has been fairly stable around 30% (Section1.4), implying that the majority of recipients receive both their own pension and a survivor pension.

On average in OECD countries, people older than 65 have a disposable income equal to 87% of the total population (Figure 1.6). The Czech Republic is well below that with 74%, with old-age pension being the most important component of old-age income. One reason is much lower legacy pensions from the old Czechoslovakian system than new pensions combined with fewer opportunities to save while working.4 Only in Korea and the Baltic states have older people a lower relative income. Moreover, the fast wage growth over the past decades in the Czech Republic has lowered pensions relative to average income due to less than wage indexation. Wages grew by over 200% between 1996 and 2018 while pensions in payment were only indexed by a total of 150%. As in other OECD countries (with the exception of Poland), income drops further in old age, such that those older than 75 have a significantly lower income than the 66-75. In some countries, currently high old-age incomes (relative to the total population’s), compared with the OECD average, might reflect various factors, such as high contribution rates or past pension rules that were not financially sustainable.

Despite low relative income levels among the elderly, the relative old-age income poverty rate, measured as the share of people over 65 with equivalised income below half median, is relatively low in international comparison, equal to 7.4% in 2017. Moreover, income inequality in the Czech Republic is low too. In 2017, the Gini index of disposable income was below 0.20 for the population aged over 65, the lowest among OECD countries with an average of 0.31 (Chapter 3). Indeed the distribution of pension benefits is very concentrated in the Czech Republic, 83% of men’s pensions and 91% of women’s pensions lie between 25% and 50% of the gross average wage (Figure 1.7). Moreover, only about 5% of men’s pensions and 1% of women’s pensions exceed 56% of the average wage.

The average old-age pension has fallen from just below 50% of the average wage for men and 40% for women in 1996 to 43% and 36%, respectively in 2018 (Figure 1.8): real wages grew by 65% – or 2.3% per year on average – while real pensions rose by just under 50% for both men and women – or 1.8% per year. Newly granted pensions between 2002 and 2018 were 2.0 percentage points higher for men on average and 2.6 percentage points higher for women, and the pension gap between women and men has been stable at about 18%.

Older women often had shorter careers and lower wages than men, resulting in lower benefit entitlements. Using EU-SILC data, the European Commission (2020[3]) shows that, women’s average pensions in the EU were about 30% lower than the average pension for men in 2018. However, despite having a large gender pay gap the relatively strong redistribution in the Czech pension system leads to a gender pension gap that is lower at 13%5, with only Denmark, Estonia and the Slovak Republic having lower gaps. Conversely, the gender gap stood at over 35% in Austria, Germany and the Netherlands.

The mandatory Czech pension system consists of an earnings-related component and a basic, flat-rate component. The earnings-related component is calculated by multiplying the reference wage with total accrual. Some non-employment spells count towards accrual while others do not and enter on top as zeros in the calculation of the reference (average) wage. Everyone who is eligible for the earnings-related component receives the basic pension, which is equal to 10% of the average wage. The basic component is described in Chapter 3.

In the Czech Republic, pension contribution rates are 28%, split between employers (21.5%) and employees (6.5%) (Figure 1.9). These contributions are meant to finance old-age (both earnings-related and basic), survivor and disability pensions. The contribution rate in the Czech Republic is among the highest in the OECD, along with France, Hungary, Italy, the Netherlands and Poland. At the average wage, the total effective pension contribution rate equalled 18.4% on average in the OECD in 2018. There might be a bias in this comparison as in some countries the contribution rate is earmarked for old-age and survivor pensions. Even taking into account only OECD countries with contribution rates that also finance disability pensions, the Czech contribution rate is 7 percentage points higher than the average.

Pension contribution rates are part of social security contributions in the Czech Republic. In addition to pension contributions, employees and employers pay 4.5% and 9.0% contributions for health insurance, respectively. Employers pay 2.1% for sickness and 1.2% for unemployment insurance. This leads to a very high total social security contribution rate of almost 45%. In addition, pension contributions are subject to personal income tax in the Czech Republic.

Anyone earning less than CZK 3 000 per month in 2020 (8.5% of the average wage) is not subject to social security contributions and does not accrue pension entitlements. Given that the minimum wage is 40% of the average wage, this means that anyone working more than 21% of the time at the minimum wage or above contributes to the pension system.

Contributions are paid on earnings up to four times the average wage (national definition) (Section 2.2), which is equal to 3.75 times the average wage based on the OECD harmonised definition; this threshold was introduced in 2008.6 Earnings above the contribution threshold have been subject to a higher income tax rate since 2013 such that the total of the tax rate and contribution rate remains constant above the contribution threshold. The degree of progressivity of the tax system is a political choice; however, this specific nexus between contribution rates and tax rates for high earners might blur the boundaries between pension contributions and taxes.

One specificity of the Czech pension system is to accrue entitlements based on the reference wage that is computed for the whole working life, rather than accruing entitlements for each year based on earnings during that period. This makes a big difference because the reference wage is far from being a linear function of (uprated) past wages.

The earnings-related pension gives 1.5% of the reference wage for each full year of service. To calculate the reference wage and therefore pension benefits, a very progressive formula is used under which income thresholds are applied. Up to the threshold of CZK 14 388 in 2019 (44% of the average wage based on the national definition), the wage is fully taken into account. Between this threshold and the pensionable-earnings cap (CZK 130 796, 400% of the average wage), only 26% of the wage is taken into account.7 Earnings over the cap are not taken into account, neither for contributions nor for the calculation of benefits (Figure 1.10, Panel A). The average of all earned wages since 1986 are taken into account for the reference wage, uprating past wages by the growth of economy-wide average wage. Non-validated periods are included in the reference wage as zeros (see below). Lower accrual rates for higher wages lead to a reduction of the effective accrual rate as wages increase (Panel B). Up to the first threshold, the effective accrual rate is equal to the statutory accrual of 1.5%, then the effective accrual sharply drops to 0.5% at the pensionable-earnings cap, after which it gradually drops further.

Part-time work is treated in the same way as full-time work: accrual is 1.5% of the reference wage for each year of service. This means that if someone works 2.5 days a week for one year, this is counted as one year of contributions. Of course, the reference wage is lower compared to someone working full time for the same hourly wage. However, given the progressive reference wage formula, the impact on pensions is less than proportional. In extreme cases – i.e. someone working less than 21% of the time at the minimum wage – no contributions are made and no accrual takes place.

The minimum years of required coverage (work and non-work validated periods) to be eligible to an old-age pension at retirement age is 35 years, or 30 years without non-work validated periods (Figure 1.11).8 Individuals with 20 years of pension coverage or 15 years without non-work validated periods can receive a pension benefit five years later than the normal retirement age that applies to men of the same birth year. Anyone reaching 35 years of coverage within these five years can retire. However, only a few hundred new pensions are currently claimed five years after the retirement age. Voluntary contributions can be made to make up for missing years of coverage if participants fall short of the required coverage. These voluntary contributions are typically made at the minimum wage, but are not frequent.

In comparison to other OECD countries, the Czech Republic is an outlier with the longest period to be eligible for earnings-related pensions at the statutory retirement age (Figure 1.11). Mexico comes second with “only” 24 years required. On average among OECD countries, it is equal to nine years. In many countries, it is less than one year.

At the moment the majority of people around the retirement age reach the minimum years of coverage. The inclusion of some non-employment spells in the covered period (see below) is one of the reasons why so far only few people have not reached the minimum requirements to receive a pension. The average number of years of coverage for those who claimed a pension in 2018 from the normal retirement age equals 46 years for men and 43 years for women (Figure 1.12, Panel A). For those claiming early pensions, these numbers stood only at one year less (Panel B). Only a minimal share do not reach 35 years, even among women.

In 2020, the normal retirement age in the Czech Republic is 63.5 years for men and 63 years and two months for women without children (Figure 1.13). According to the OECD definition, used for comparison purposes, the normal retirement age is the age at which individuals are eligible for retirement benefits from all pension components without penalties, assuming a full career from age 22. In the Czech Republic, the normal and the statutory retirement ages are the same. It is gradually increasing by two months per birth cohort for men until reaching age 65. For women without children, it is currently increasing by six months per birth cohort to equalise with the retirement age for men by the end of 2020.

Women who had children can retire earlier. Having one child allows a woman to retire one year earlier, for two children two years earlier, for three or four children three years earlier, and for five or more children four years earlier. The normal retirement age of women with children is also rising by six months per year until catching up with the retirement age for men. The convergence will be completed in 2037 (Figure 1.13).9

The normal retirement age was below the OECD average in 2018, and it will remain so in the future (Figure 1.14). Only seven OECD countries have and will have a normal retirement age at the same level or below the Czech one. Given current legislation, the future normal retirement age will range from 62 in Greece, Luxembourg, Slovenia and Turkey to 74 in Denmark. On average across OECD countries, it will increase from 64.2 in 2018 to 66.1 in the future – i.e. for someone having entered the labour market in 2018 and therefore retiring after 2060 -, about one year more than in the Czech Republic.

The Czech retirement age was not always projected to remain below the OECD average. In 2011, it was decided to continue raising the retirement age after age 65 by two months every year. This would have meant that for someone who entered the labour market in 2018 at age 22 the retirement age would have been 70 years and two months. However, in 2016 the decision was made to stop increasing the retirement age beyond 65.

Rather than increasing retirement ages with a fixed number of months every year, Denmark, Estonia, Finland, Italy, the Netherlands and Portugal have linked retirement ages to life expectancy. Automatic rules such as these help resist the temptation to make decisions that might be popular but ultimately unsustainable, such as lowering the retirement age from not a particularly high level while life expectancy increases. The design of the automatic link combined with the projected improvements in life expectancy implies an increase in the retirement age of about one month or less per year in Finland, Italy, the Netherlands and Portugal, and of about 1.7 months in Denmark and Estonia. With two months per year, the planned unlimited increase in the Czech Republic was thus very high in international comparison. Similar to the Czech Republic, the Slovak Republic had linked their retirement ages to life expectancy but recently backtracked by abolishing the link, while Italy temporarily suspended it for some occupations.

The majority of people in the Czech Republic claim their old-age pension exactly at the statutory retirement age. Yet, a significant share claim their pensions early while a minority claim their pension late. For men born in 1952, having retired between 2006-18, while the retirement age was 62 and 10 months, 33% early retired before age 62, more than 60% between age 62 and 63 while only 3% claimed their pension later (Figure 1.15). Among those born in 1952, 7% of women retired before the earliest statutory retirement age (56 and 4 months for women with 4 or more children) and almost all of them had retired after the statutory retirement age for women without children (i.e. beyond 64). In 2018, beyond the 1952 birth cohort, only very few people claimed their pension after age 66, i.e. less than 0.5% of new pensioners.

Future theoretical replacement rates are computed by the OECD in order to compare key outputs of pension systems across countries based on current legislation. The future replacement rate represents the level of pension benefits from mandatory public and private pension schemes relative to earnings when working. The baseline case assumes a full career starting at age 22 in 2018 until reaching the country-specific normal retirement age. This theoretical replacement rate is equal to the pension benefit at the retirement age as a percentage of the last earnings.

Looking ahead, net pension replacement rates (the net pension benefit at retirement as percentage of the last net earnings) for average wage earners are at 60% in the Czech Republic just above the OECD average (Figure 1.16). At the average wage level, net replacement rates from mandatory schemes are on average 59% in the OECD and range from below 30% in Mexico and the United Kingdom to more than 90% in Italy, Luxembourg and Turkey.

Pension contributions are subject to personal income tax in the Czech Republic, which is fairly unusual. To (partly) avoid double taxation, pensions are not taxed up to a value of CZK 439 200 corresponding to 36 months of the minimum wage (114% of the average wage), while the income tax rate is 15% above this threshold (see Annex Figure 1.A.1 for average tax rates by earnings). Given the strong redistributive nature of the Czech pension system, which drastically limits high pensions, this threshold level implies that there are no pensions taxed if that is the only source of old-age income.

Most OECD countries aim to protect low-income workers (here defined as workers earning half of average worker earnings) from old-age poverty, which results in higher replacement rates for them (68% for the OECD on average, Figure 1.16) than for average earners. This is certainly the case for the Czech Republic where net replacement rates for half-average-wage earners are high at 92%. Among other OECD countries only Denmark, Italy (based on a much higher retirement age) and Luxembourg have higher replacement rates for low earners.

By contrast, higher earners (twice the average wage) have replacement rates well below average-wage earners in the Czech Republic (41% versus 60%). On average in OECD countries, the net replacement rate at twice average earnings is 51%, somewhat below the 59% figure for average earners.

Another way to illustrate the strong redistribution within the Czech pension system even among full-career workers is to compare the internal rates of return across different earnings level throughout the career (Figure 1.17). The internal rate of return is the rate of return on paid contributions that generates the benefit provided by the pension system. A woman earning the average wage during a full career from 2018 will receive, given legislated pension rules and projected life expectancy, pension benefits ensuring a real annual return on paid contributions of 1.55% (assuming real wage growth of 1.25% per year, which is the standard assumption in the OECD pension model). For men this is lower at 1.18% given their lower life expectancy. Given pension rules, the key parameter that influences the internal rates of return is the real-wage growth assumption, with almost a one-to-one relationship.10 This means that the internal rate of return generated by the Czech pension system at the average wage is about the real-wage growth rate plus 30 basis points for women, and slightly below the real-wage growth rate for men.11

A financially sustainable PAYGO system is supposed to generate an internal return equal to the growth of the contribution base, which is closely related to, assuming a constant contribution rate, the wage bill growth. The latter is equal to wage growth (1.25% per year here in real terms here) plus employment growth; the working age population is projected to decrease by 0.5% per year on average by 2060 (Figure 1.2). Hence, if employment growth were to equal the growth of the working-age population, the natural internal rate of return of the Czech PAYGO scheme would be around 1% per year based on the assumptions of the OECD pension model, i.e. the wage-growth rate minus 0.5 percentage points.

While these numbers should be taken as orders of magnitude only, the key feature shown in Figure 1.17 is that internal rates of return on pension contributions differ widely by earnings level. By comparison to the 1.2-1.6% at the average wage, low earners (50% of the average wage) will have a real rate of return on their contributions that is larger than 2.6% while anyone earning above three times the average wage will have negative real returns based on the same assumptions.

Future replacement rates in the Czech Republic will be similar to those of current pensioners among people with uninterrupted careers. More precisely, the theoretical replacement rate of a male worker at the average wage will increase between cohorts born in 1940 and in 1996 by about 3 percentage points (Figure 1.18). This small increase stems from a longer contribution history based on the rise in the retirement age from 61 for the 1940-born cohort to 65 in the future while other reforms had only minor effects. There is no life expectancy adjustment and, despite the increase in the retirement age, the share of adult lifetime spent in retirement is projected to increase among men between the generations born in 1956 and 1996 (OECD, 2019[4]).

In other OECD countries, legislated reforms will have much stronger effects on replacement rates. In particular in Mexico, Sweden and Poland, legislated deep-cutting systemic reforms imply a shift from unsustainable DB to (notional) DC schemes and have caused or will cause a substantial decline in replacement rates. The OECD average will decrease by about 6 percentage points at normal retirement ages, or about 10%.

Since 1996, minimum indexation was first set at inflation plus one-third of real-wage growth, with the government keeping the right to index pensions above the statutory minimum, with no upper limit. In 2018, the minimum indexation was made slightly more generous to inflation plus half of real-wage growth. During some periods of tight public finances, the indexation of pensions in payment has been reduced below what is implied by the rule; between 1998-2002 and 2013-14, inflation was only taken into account for 70% and 33%, respectively.

Figure 1.19 shows that until 1999 indexation followed largely both price and wage growth given that real-wage growth in that period was subdued. Since 1999, indexation has been de facto largely in line with – even slightly larger than – inflation plus half of real-wage growth, indicating regular discretionary increases beyond the minimum indexation rule.

Early retirement in the Czech Republic is allowed up to three years before reaching the statutory retirement age or at 60 years of age, whichever is earlier. The minimum requirements for early retirement are the same as for normal retirement: 35 years of coverage is necessary or 30 years of paid contributions without validated periods. Compared to other OECD countries, the Czech Republic has a relatively low early retirement age of 60 years. For someone entering the labour market at age 22 the early retirement age was 61.2 years in 2018 on average among the 31 OECD countries that have a specific minimum retirement age for mandatory earnings-related pensions (Figure 1.20).

Given longer life expectancy, the age of 60 for the future eligibility to early retirement is too low. This age reference contributes to shaping social norms and influencing behaviours by both employees and employers about working at older ages; it is not consistent with other efforts to enhance the labour supply of older workers.

Every started 90-day period of early retirement decreases pension entitlements. Each of the first four 90-day periods (i.e. roughly a year) lowers pension benefits by 0.9% of the reference wage; retiring another four 90-day periods earlier decreases benefits by 1.2% of the reference wage each; and, any additional 90-day period decreases benefits by a large 1.5% of the reference wage. Combining work and early retirement is not permitted in the Czech Republic. Pensioners have to suspend pension receipts if they want to start working again before reaching the statutory retirement age.

For someone entering the labour market at age 22 in 2018, retiring more than one year early results in large pension losses compared with other OECD countries (Figure 1.21). In the Czech Republic, given the increasing penalty by years of early retirement plus additional losses from accruing less pension rights, the pension loss (compared with someone working until the normal retirement age) increases from 6.8% for retiring one year early to about 8.7% per year of early retirement for retiring five years early (in the future, i.e. 5*8.7%=43.3% in total).

Deferring a pension and continuing to work may increase pension benefits through two mechanisms, depending on the specific design of pension systems. First, as the expected time spent in retirement is then reduced, existing accrued entitlements can finance higher monthly benefits that are paid over a shorter period. Second, continuing to work leads in most OECD countries to additional contributions and entitlements. In the Czech Republic, as in most countries with defined benefit schemes, the first mechanism operates through bonuses on past entitlements. On top of additional entitlements for working longer (second mechanism), deferring retirement increases the earnings-related pension component by 1.5% of the reference wage for each 90-day period of postponement. In principle, deferral is possible indefinitely. In practice, few people defer more than a couple of months.

Figure 1.22 shows the impact for a full-career worker of deferring pensions and continuing to work on monthly benefits summed over all pension schemes. Across OECD countries, the combined overall increase – from the deferral rate, additional entitlements and benefit indexation – averages about 7.5% per year of deferral, and depends only slightly on the length of the deferral. The increase in the Czech Republic is just slightly higher at 8%, which is very close to what is implied by actuarial neutrality (OECD, 2017[5]). This suggests that there is limited financial advantage or disadvantage from the pension point of view between retiring at the retirement age and continuing to work and deferring pensions – and that the retirement decision is broadly neutral for public finances. However, Figure 1.15 above shows that few people extend their working life beyond the statutory retirement age. This means that the statutory age plays a big role beyond the financial implications in shaping social norms about when to retire in the Czech Republic and/or that workers value retiring as soon as possible without penalty well beyond what actuarial neutrality suggests.

There are no restrictions to combine pension receipt and work after the retirement age. In this case, the earnings-related pension is increased by 0.4% of the reference wage per year but no regular accrual takes place. Partial pension receipt is also possible. Someone can draw half of their pension and keep working without restrictions. In this case, the earnings-related pension is increased by 1.5% of the reference wage per half a year of work.

To determine accruals and the reference wage the Czech pension system distinguishes three types of periods (Table 1.3). First, there is time working. Each period leads to additional accruals while earned wages are included in the calculation of the reference (average) wage. Second, there are so-called (non-work) validated periods (usually referred to as non-contributory periods), which lead to accruals but are excluded from the reference (average) wage. Third, there are non-validated periods, referred to as uninsured periods in the Czech pension system. These periods do not lead to accruals and are in addition included as zero earnings in the calculation of the reference (average) wage.

Schematically this can be shown as:

The system therefore heavily penalises non-validated periods: not only do non-validated periods lead to no accrual (as is common in many countries), they also lower the reference wage, hence actually diminishing “acquired” entitlements.

The non-work validated periods leading to accruals are:

1. 1. Periods of childcare for children up to four years of age

2. 2. Periods of care for a dependent person in their household

3. 3. Compulsory military service

4. 4. Spells of unemployment with unemployment benefit

5. 5. Up to three years of unemployment without unemployment benefit

6. 6. Period of disability of the 3rd degree.

Periods in education are excluded from benefit calculations, including in the computation of the reference wage. Prior to 1996, secondary education and tertiary education were considered validated periods, and prior to 2010 tertiary education only was considered a validated period.

Spells related to care and military service (1-3) are fully taken into account while all other spells (4-6) count only for 80%. In addition, the duration of unemployment benefits varies with age. Therefore, credits to the pension system for unemployment vary with age. Five months of unemployment benefits are available before age 50, eight months from 50 to 55 and 11 months for the over 55s. In addition, up to three years spent unemployed without unemployment benefits are also credited. However, only one of those years can be credited before the age of 55. Hence, for people unemployed from age 55, more than three years can be credited in terms of accruals (80% * (three years + 11 months)) with the reference wage not being affected.

While periods not spent working can significantly affect pension benefits, there are thus big differences between the different types of non-employment periods. First, validated periods lead to lower if any losses than non-validated periods. Second, lower reference-wage brackets typically incur larger losses while within each bracket higher wage earners incur larger losses (Box 1.1 for details).

One example of a non-employment period that leads to limited or no losses in the Czech Republic is childcare. Time spent on childcare for children up to four years of age are fully credited in terms of accrual and are excluded from the reference wage. Figure 1.23 shows the impact of career breaks due to childcare on pension benefits under the assumptions of the OECD pension model.12 More precisely, it compares pension benefits for women who stop working during five years from age 30 to care for their two children born when the mother was aged 30 and 32 (at age 35 they are assumed to resume full-time work until the normal retirement age) compared with the full-career case.

In the Czech Republic, due to generous childcare credits, such a career break has no impact on future pensions whatever the earnings level. By contrast, this career break generates a loss of 4% on pensions on average across OECD countries for average-wage workers. In countries with funded DC systems, with a one-to-one relationship between actual contributions and pensions, such as Mexico, Australia and Chile the losses from childcare are much larger, resulting in more than 12% lower pensions at the average-wage level.

Spells of unemployment often lower pension benefits. Figure 1.24 illustrates the impact across countries of career breaks due to unemployment on pension benefits for the case of men being unemployed during five years from age 35. They are then assumed to resume full-time work until normal retirement age.

Spells of unemployment lead to larger pension losses in the Czech Republic than in many other OECD countries. First, only 80% of the time spent in unemployment is credited towards pensions. Second, once unemployment benefits run out only 80% of one year of unemployment without benefits until age 55 is credited. Finally, once these two options are exhausted no accrual takes place and all additional time out of work is then included as zeros in the reference wage. Therefore, a five-year unemployment break before age 55 leads to only just over a year of credits: 80% of five months of unemployment benefit receipts plus 80% of one year of unemployment without benefits.13 Moreover, the other three years and seven months, which are not credited at all, and only those, lower the reference wage that is computed when retiring. In the case defined above, this leads to a 10% lower pension for the average-wage worker in the Czech Republic (Figure 1.24). In the OECD on average, it is only 6% lower. Low-wage earners with unemployment breaks in both the OECD on average and the Czech Republic are slightly less affected in terms of relative pensions.

Longer spells of unemployment in combination with later entry into the labour market lead to larger pension losses. The assumptions in Figure 1.25 are largely the same as in the case above but instead of a five-year break, now a break of ten years is considered with the person entering the labour market at age 27 instead of 22. Since at age 65 this person will not have reached the necessary 30 years of employment or 35 years of insurance, she will have to work two more years to get a pension. For average-wage earners, the pension will be 30% lower than for a full-career worker. Most of the period spent out of work will not be validated for, which means both no accrual and a lower the reference wage. For low earners the loss is slightly larger (Figure 1.25). This is uncommon and due to the peculiar and complex interactions between the way the reference wage is computed to account for non-validated periods and the accrual rules (Box 1.1). The impact is significantly larger than in the OECD on average where the pension is 24% lower at the average wage and 18% lower for low earners.

Box 1.1 discusses in greater detail the mix of reference-wage and period effects that is at play. Both differ depending on the reference-wage bracket in the benefit formula. Looking at the different cases above by wage level, the losses in pension benefits follow a very erratic pattern. Box 1.1 also shows that pension losses are much lower for unemployment break after age 55.

As part of the economic support measures during the COVID-19 global health emergency, the Czech government covered part of the labour cost. Employers have applied for a subsidy to cover part of wages of employees who cannot work because of shutdowns imposed by the government or because of low demand. Employers still pay the wage in full directly to the employee while the government refunds part of the wage cost. In terms of pension contributions and pension accruals nothing changes compared to the pre-crisis situation for individual who kept their job.

There is evidence that some employees (are forced to) claim a temporary sickness insurance benefit by declaring to be temporarily unable to work. In that case, employees receive a benefit that is not subject to social security contributions. However, the periods are validated for pension purposes, meaning that accrual takes place with the period being excluded from the reference wage.

In 2018, self-employment represented 18% of total employment in the Czech Republic, slightly more than the OECD average of 15%. The self-employed are covered by the same pension scheme as employees and the total pension contribution rate is 28% for both employees and self-employed workers. However, compared to employees, the self-employed pay lower contributions for two reasons. First, the tax code allows them to deduct between 30% and 80% (depending on someone’s occupation) of revenue as cost in order to set profit, thereby eliminating the need for cost accounting. While this simplified solution reduces the administrative burden for the self-employed, which is highly relevant for those with low income, it lowers taxes and pension contributions of many of them compared to employees with similar earnings. Second, the base for social security contributions is set at 50% of profits (with a minimum of 25% of the economy-wide average wage).

A better harmonisation of contributions between employees and the self-employed would be achieved by setting the contribution base closer to the equivalent of the gross wage for employees. For an employee, total labour cost is equal to gross wages plus employer’s social security contributions, which rate is equal to 33.8% in the Czech Republic. Hence, the gross wage is equal to 100/133.8=74.7% of labour cost. Drawing a parallel between profit of a self-employed worker and total labour cost of an employee, a true harmonisation would imply that the contribution base of a self-employed in the Czech Republic is 74.7% of profits (OECD, 2019[1]).

A lower base translates into a lower pension. After contributing what is mandatory during a full career, self-employed workers with income net of social security contributions equal to the pre-tax income of an average-wage worker can expect to receive an old-age pension equal to 83% of the pension of this worker, a ratio that is very close to the OECD average (Figure 1.27). However, given the high progressivity of the Czech pension system, setting the contribution base lower than the harmonised base (i.e. at 50% instead of 74.7%, or 75% to simplify, of profits) implies that in this case the self-employed pay only 67% – 50%/74.7% = 67% – of the contributions paid by employees with similar earnings.

This results in an implicit subsidy benefiting the self-employed as reduced contributions only partially translate into lower pensions. Subsidising the self-employed within the pension system is hard to justify. First, it provides a favourable treatment based only on the contract type. Second, it undermines pension finances or induces other workers to pay higher contributions. Finally, it creates opaque incentives on the labour market, making self-employment attractive in the short term at the cost of weak social protection and/or higher public spending in the long term (OECD, 2019[1]). If support to the self-employed is a policy objective, transparency would be greatly enhanced, and the cost of the subsidy more fairly shared, by explicitly subsidising self-employment – for example financing extra social security contributions – through taxes.

Spending on survivor pensions account for 0.6% of GDP compared with 1.0% on average in the OECD. One-third of all newly granted pensions paid to women in 2018 and 13% to men were survivor pensions (Figure 1.28). Dependent children might be eligible for a survivor pension as well (in case the child is below 26 years of age and studying), but this is beyond the scope of this chapter. In the Czech Republic, more than 90% of survivor pension recipients are widows or widowers (OECD, 2018).

The conditions for claiming a survivor pension in the Czech Republic are twofold. First, the deceased must have had entitlement to either an old-age or a disability pension. Second, the survivor must fulfil certain conditions to be able to draw the pension. The first year of survivor pension benefits are paid out to any current surviving spouse, but after that only to those fulfilling one of the following conditions:

• The recipient is taking care of a child,18

• The recipient is taking care of a handicapped relative,

• The recipient is disabled (in the highest degree of disability),

• The recipient has reached a certain age (statutory retirement age of men minus four years or statutory retirement age of self, whichever is lower).

If the recipient does not receive a pension of his/her own, the survivor benefit consists of the basic component and 50% of the earnings-related component of the deceased’s pension. In case the recipient has his/her own pension, the total benefit (own pension plus survivor benefit) consists of once the basic component, the higher earnings-related component and 50% of the lower earnings-related component. In the Czech Republic, the survivor receives a lump sum equal to one year of the survivor benefits in case of remarriage.

If a survivor does not have an own pension but the spouse had a full career at the average wage, the survivor will have a gross pension equal to 26% of the average wage in the Czech Republic, lower than the OECD average of 31% (OECD, 2018[6]). Receiving a survivor pension significantly raises the future benefits as without it the surviving spouse would in that case only receive the safety net (7% of the average wage in the future given indexation rules).

However, one specificity of the Czech Republic is that more than 95% of recipients receive their own old-age pensions. If the surviving spouse also had a full career at the average wage, the gross pension is equal to 56% of the average wage in the Czech Republic, compared with 64% in the OECD on average (Figure 1.29). In that case, survivor pension benefits add 10 percentage points in the Czech Republic, just below the 12 p.p. addition for the OECD on average.

In the Czech Republic there are two types of special regimes. One is for deep-shaft miners and one for the armed forces comprising the military, the police, the fire brigade, customs officers, judicial guards, prison guards and the secret service. Deep-shaft miners contribute to the general pension scheme administered by the Ministry of Labour and Social Affairs. They can retire seven years before the statutory retirement age once they have worked a certain number of shifts under the earth’s surface, and have a more generous benefit formula. In the future, this scheme will apply to a small share of workers as, in 2017, 26 000 people only were employed in the mining and quarrying sector (and not all of them deep-shaft mining), representing significantly less than 1% of total employment.19

Pensions for the armed forces are outside the general pension scheme and are administered by the Ministries of Defence, Interior and Justice. They are made of two components. The first is the standard pension, which follows the same rules as other workers. Usually the last employment before claiming a pension determines under which scheme the pensioner falls. The second component is the Retirement Allowance.

The Retirement Allowance for former members of the armed forces can be claimed by individuals who spent at least 15 years in service. With 15 years of service, the pension benefit is equal to 20% of the last year’s earnings. One additional year of service generates an additional benefit of: 3% of the last year’s earnings between 16 and 20 years over 15; of 2% between 21 and 25 years over 20; and, of 1% from 26 years, up to a maximum of 50% of the last year’s earnings, which is reached after 30 years.20 The Retirement Allowance is indexed in the same way as the earnings-related part of the standard pension. When claiming a standard pension after reaching the statutory retirement age, the Retirement Allowance is reduced, so that the sum of both pensions in payment is equal to the initial Retirement Allowance. That is, the sources of financing and administration change but the total income stays the same.

In January 2020, the Commission for Fair Pensions, tasked with making proposals to reform the pension system, published its first report. This section looks into the basic framework of its reform proposals. The Commission for Fair Pensions set three objectives of the reform: fairness, comprehensibility and financial sustainability. Behind the proposal, there are three different scenarios with differing degrees of generosity and financial sustainability. The main focus of the proposed reforms, however, centres around one scenario, which is discussed in this section.

Currently the earnings-related pension has a flat component of 10% of the average wage and an earnings-related part.21 The big change is to triple the level of the basic component, up to 30% of the average wage, to better reflect the minimum cost of living of pensioners. In the proposals, the basic component is indexed to wages as is currently the case. On top of the basic component, the accrual of the earnings-related component is lower than the current one while the earnings-related component is indexed in payment with prices only. These indexation rules are likely to increase the role played by the basic component as retirees age. One big simplification in the proposed system is that, each year, pension entitlements (whether gained through employment or validated periods) are calculated and communicated to participants. Currently, people only know their entitlements at the time of retirement (Section 2.4). The eligibility conditions are the same as in the current system in terms of retirement age and period of insurance.

In the proposed system, the accrual rate is also simplified and set at 0.39% applied to the full wage up to the ceiling of 400% of the average wage. This compares with 1.5% currently applied to a declining share of wages with thresholds of 44% of average wage and the 400% ceiling. The coefficient of 0.39% is equivalent to the current effective (marginal) accrual rate between the two current thresholds: within this reference-wage bracket, the current accrual rate is 1.5% on 26% of earnings (1.5%*26% =0.39%).

The new pension system would largely provide similar pension benefits for those earning more than 44% of the national average wage (Figure 1.30). The level of the basic component and the accrual rate is such that, for someone with a 41-year career, the pension benefit is the same in the current and proposed system. For those earning less than 44% of the average wage, benefits would go up since the basic component is increased. These will be largely people working part-time given that the minimum wage is 40% of the average wage, only slightly less than the current threshold.

Since the basic component does not increase with career length in both the current and proposed system and since the accrual rate for the earnings-related part is lower in the proposed system, the new benefit formula would generate lower pensions for careers longer than 41 years. To compensate for this, a complementary bonus is included in the proposed system for those with coverage of more than 41 years. Each year of insurance above 41 years is awarded with a yearly bonus of CZK 2 880 (0.75% of the average wage). As Figure 1.31 shows, the proposed reform generates higher benefits for workers with shorter careers (but still at least of 30 years). Someone with a 30-year career at the average wage would gain 5 percentage points.

In the reform proposal, credits would also be included for non-employment periods, for example periods of caring for children and dependants, basic military service and unemployment. In addition, rather than leaving the pension benefit to be calculated at the end of someone’s career by excluding these non-employment spells from the computation of the reference wage, the new system would credit non-employment periods as they occur, thereby increasing transparency. For each non-employment spell a notional wage is recorded, to which the 0.39% accrual applies, allowing participants to calculate their current entitlements at all times.

Non-employment spells would be credited as the higher value of two options. The first option is based on the economy-wide average wage and the second option is based on previous individual earnings (Table 1.4). It is allowed to work part-time while accruing credits for childcare. The earnings from part-time work will be added to the notional wage for the non-employment spell. In addition to the accrual mentioned above during childcare, the new system also provides inflation-indexed bonuses for each child raised (CZK 6 000 or 1.5% of the average wage).22

The self-employed currently only pay mandatory contributions on half of their profit (difference between revenues and expenditure) with a fixed minimum base of 25% of the average wage (Section 2.4). As a result, the self-employed contribute very little and receive little. The reform proposal stresses the need to boost contributions and pension benefits for the self-employed. The proposal would see the calculation base increase from 50% of profits to 75%. As shown in Section 1.4 this would indeed lead to largely similar pension contributions between the self-employed and employees with similar earnings. The minimum base would increase from 25% to 40% of average wage, i.e. to the current level of the minimum wage.

In sum, the proposed pension reform greatly simplifies the system while maintaining its redistributive elements. It would significantly improve transparency and make the system more intuitive, since for each year the earned entitlements can be directly and quite easily calculated, by contrast with the currently situation. The proposal allows to inform the insured every year about their pension entitlements. This would enable them to see the impact of their current situation on entitlements, and to better anticipate their future benefits.

However, by trying to stick too closely to current outcomes and with the inclusion of additional bonuses for childcare and long careers, the pension proposal creates other complications and inequalities. The bonus for careers longer than 41 years makes the system both more complicated again with limited impact (Figure 1.31). There is no compelling reason why total accrual after 41 years should be higher than before 41 years. Moreover, the two options for non-employment credits (based on the average wage or on last earnings) also makes the system unnecessarily complicated. The same holds for adding more childcare related bonuses on top of the already generous ones.

Moreover, one big flaw in the current system, the very high minimum years of coverage needed to be eligible for a pension, is maintained. A combination of a basic component that depends on career length (Chapter 3) and eligibility to a pension with no or low minimum years of coverage required would improve the system. Figure 1.32 shows that a pro-rated basic component in combination with accruals from the first year of contributions can maintain similar benefits as those provided by the current system, consistent with the proposal of the Commission for Fair Pensions, while generating pensions for people with short careers in line with their contributions to the system. All parameters can be adjusted to a given level of total spending.

While various changes have been implemented over the last 25 years, the architecture and principles in the design of the Czech pension system have been mainly unaltered since the 1995 Pension Insurance Act. The system works relatively well in ensuring net replacement rates close to the OECD average for average-wage workers. The contribution rate is relatively high and there is a high redistribution within the pension system, at least within generations.

Redistribution is a political choice and operates through a weak link between pension contributions and future benefits: the benefit structure is highly compressed, replacement rates are high for low earners while high earners suffer from very low internal returns on their contributions. For the latter, the boundary between contributions and taxes is blurred. A discussion might be welcome about the opportunity of shifting – on top of health insurance contributions – part of the financing of some redistributive instruments, such as credits for non-employment periods or basic pensions to general taxes. This would allow lowering the currently very high social security contribution rates, increasing accrual rates, in particular for high earners, or leaving more space to develop funded pensions. Moreover, the tax regime underlying the design of the system has unusual features: income taxes are paid on both employee and employer contributions, while pension benefits are mostly untaxed. These features are, however, difficult to reform given that modifying them would generate deep transition issues across generations.

Overall, the main weaknesses of the system are twofold. The average income of retirees relative to that of the total population is relatively low despite high contribution rates; and, the building-up of pension entitlements is complex, making it very complicated for contributors to understand their accrued rights and anticipate their future pension level, and for the administration to manage the system.

The Czech Republic is an outlier in requiring that either 30 years are contributed or 35 years are validated to be eligible to an old-age pension, unless one retires five years after the statutory retirement age. This applies to both the basic and the earnings-related components. It implies that contributions might be paid during many years without generating pension entitlements. Only three other OECD countries require more than 15 years at the normal retirement age: Hungary and Italy (20 years) and Mexico (24 years).

So far, such a long required period has not raised much concern because of high recorded employment in former Czechoslovakia and the generous validation of non-contributed periods. However, the transformation of the economy from the 1990s, the welcome tightening of some instruments to validate pension entitlements, as for example the exclusion of years spent in secondary (from 1996) and tertiary education (from 2010), as well as labour market trends will raise the share of older people not covered by old-age pensions. In short, such a long required period is not suited to a modern pension system: it overly penalises workers with short careers, increases the risk of vulnerabilities faced by future retirees and seriously weakens the attractiveness or consent to contribute. Each contributed period should generate some pension entitlements. The currently long eligibility period should thus be eliminated or at least drastically reduced.

Another characteristic that almost singles out the Czech pension system is the complexity of calculating acquired pension rights. The Czech Republic is among OECD countries where, beyond the uprating of past wages, the accrued entitlements during each year of contributions cannot be determined until the career is completed. In other words, the acquired entitlements depend on the future part of the career. This complexity prevents contributors from understanding how pension entitlements are built and therefore from being able to anticipate their income in old age. It also weakens the good management of the system by making it difficult and more uncertain to project pension flows.

The complexity stems from the calculation of the reference wage, the key input to determine pension benefits. The combination of three properties in this calculation compounds the difficulty. First, effective accrual rates, which are determined based on the weights granted to wages within brackets, diminish sharply in a non-linear way with wages (even below the pensionable earnings ceiling). While this results in a very progressive scheme, it makes the overall pension formula complex and the entitlements difficult to track. Second, the reference wage used to calculate pension benefits is closely related to average lifetime wages. While using lifetime wages might be fairer, its combination with the accrual pattern comes at the price of a lack of transparency in understanding entitlements when they are supposed to accrue. This makes the Czech system unique, except for the Portuguese and Swiss systems which are, however, based on a much smoother accrual-rate pattern (OECD, 2019[7]) and the US public scheme based on the best 35 years. Third, some non-contributed periods not only do not generate entitlements but lower both already accrued and future entitlements by lowering the reference wage (“double penalty”).

There are several options to simplify the pension formula. All of them suppose first either to have a constant effective accrual rate or to eliminate the reference to average lifetime wages such that pension entitlements are clearly identified for each year based on earnings during that period. This allows to create and monitor individual accounts where pension entitlements steadily build up as in Austria. It would significantly improve transparency and make the system more intuitive, since for each year the earned entitlements can be directly and quite easily calculated. It would allow to inform the insured every year about their pension entitlements, enabling them to see the impact of their current situation on entitlements, and to better anticipate their future benefits, thereby increasing trust in the pension system. Of course, this does not prevent from accounting for earnings throughout the whole career. Only six OECD countries do not take into account the whole career.

The various options differ depending on the progressivity of accruals. The current non-linear shape may be kept while applying e.g. to monthly or yearly wages. However, this might not be the fairest solution. Indeed, if that was the case, someone who would spend the whole career earning the average wage might have a very different earnings-related pension from someone earning half the average wage during half of the career and 1.5 times the average wage during the other half, even though both would have contributed similarly and earned the same lifetime wages. An alternative would be to smooth the weighting pattern in a linear way, but this would come at the price of an additional complexity: the linear formula itself. Perhaps, the simplest solution is to opt for a constant accrual rate for all wage levels while preserving the overall progressivity of the scheme by raising the weight (i.e. the level) of the basic pension. This is what is proposed by the Commission for Fair Pensions. Their proposal simplifies greatly the core of the system while broadly maintaining its progressivity.

A more generous system would be based on a high level of basic pensions and high accruals. This would, however, generate high spending, triggering the usual trade-off between generosity and cost. A second trade-off for a given level of total spending relates to the redistribution objective as the basic pension is more important for low earners while the accrual rate plays a bigger role for high earners. Based on the policy objective, it is possible to make the system less (or more) progressive by opting for a lower (higher) level of the basic pension and a higher (lower) constant accrual rate. In international comparison, consistent with the high level of progressivity in the current system, the Commission for Fair Pensions has opted for both a basic pension level among the highest, at 30% of average wages, and the lowest accrual rate, at 0.39% (except for Lithuania with 0.24%), while the latter is larger than 1.60% in Austria, Italy, Luxembourg, Portugal, Spain and Turkey (OECD, 2019[1]). Still another possibility would be to introduce a tax-financed residence-based basic pension, on top of which contributory pension entitlements can accrue (Chapter 3).

Finally, over-penalising some non-contributed periods should be eliminated. This would directly be achieved by eliminating the way the lifetime reference wage is currently calculated. At least in the current system, non-validated periods should not be accounted for in the computation of the reference wage and should only affect pensions by lowering accruals.

The effective retirement age is relatively low in the Czech Republic for three reasons. First, the statutory (or normal) retirement age is relatively low and the gap to the OECD average will remain in the long term even when based on legislated increases. It is therefore important to implement the planned increase of the retirement age up to age 65 as well as the welcome convergence across genders and family statuses by 2037. Second, in the Czech Republic, perhaps more than in other countries, the statutory retirement age plays a big role in influencing retirement behaviours, and few people retire after the retirement age. Third, early retirement is and will be possible from age 60, with relatively large penalties generating potentially low pensions.

With ageing prospects, extending working lives is crucial to preserve a good level of pensions while limiting financial sustainability issues. Beyond pensions, it lowers the impact of ageing on total output and ultimately on the average standard of living among the whole population. The link between the retirement age and life expectancy at older ages is the key instrument to respond to longevity trends, making increases in the retirement age conditional on effective gains in life expectancy and limiting the political cost to undertake such unpopular measures as raising the retirement age.

In 2011, the Czech Republic introduced a gradual increase in the statutory retirement age by two months per year from 2030 once the retirement age reaches 65 (for most people, Figure 1.13). This mechanism was eliminated in 2017. A link to life expectancy is a better option because if gains in life expectancy stop or slow (or become negative) the retirement age is consistently adjusted.

Possible reform options refer to the pace of the link. One baseline scenario is to maintain the relative share of time spent working and time spent in retirement. This is broadly consistent with passing about two-thirds of life expectancy gains (for example at age 65) into the retirement age. Based on current mortality projections until 2065, this would imply increasing the retirement age by slightly less than one month per year. However, other considerations might also be brought into the picture, such as opting for a faster or slower adjustment to ensure financial sustainability, depending in particular on other pension measures potentially linking benefits to demographic developments through a sustainability factor.

Not only should such a link be established, but it should be extended to the early-retirement age such that early retirement is possible only a few years before the statutory age. The age of 60 years to still be able to early retire in the future is too low.

It is often argued that such a link is regressive, penalising more over time those with low socio-economic background who tend to have a shorter life expectancy. However, it is important to distinguish the static impact – resulting from the increase in the retirement age – from the dynamic impact – the link aims at responding to overall longevity gains (Chapter 1 in (OECD, 2019[1])). Broadly shared longevity gains with unchanged retirement ages is progressive based on the same argument: they tend to benefit those with shorter expected lives relatively more. In that sense, increasing the retirement age to accompany well-shared life-expectancy gains goes towards restoring neutrality. The relevance of linking the retirement age to life expectancy would be weakened if there were a long-term increase in socio-economic differences in life expectancy. There is no evidence of such a trend in the Czech Republic.

The validation of some non-employment periods might drastically limit the impact of career breaks. Currently, these periods are implicitly credited upon retirement at the reference-wage level. With the proposed shift to simplify the system discussed above, the crediting has to be based on conditions applying when the career is interrupted. Pensionable earnings can in that case be based on either the last earnings before the interruption or some economy-wide metric at that time; the latter would be more redistributive.

Long periods are currently validated for employment breaks related to childcare. For unemployment, the period that is credited is 80% of the length of the unemployment benefit period plus some extra months for periods without unemployment benefits (Section 2.4). The length of both unemployment benefit and extra-months periods vary with age, such that much more generous credits are granted for unemployment after age 55 while the impact of unemployment-related career breaks in the middle of the career on future pensions is larger than in the OECD on average. Moreover, long-term unemployment currently has a larger impact, given the “double penalty” once unemployment benefits expire. The above proposed simplification to calculate pension benefits will eliminate the “double penalty”. In addition, the length of the credited period should not depend on age as age-specific measures contribute to stereotypes that eventually lower employment prospects at older ages, especially by encouraging early retirement. Instead, activation and re-employment support policies should be strengthened irrespective of age but closely tailored to individual needs and employment prospects (OECD, 2019[8]). Given that the eligibility period to unemployment benefits is relatively short, pensionable earnings could extend for a longer period although at a reduced rate.

Old-age social protection for the self-employed should also be improved. Despite the harmonisation of their contribution rates with those of employees, the self-employed currently pay low contributions because their contribution base is low. As a result, they accrue less pension entitlements, even if the impact is limited thanks to the progressivity of the pension system. In line with the proposal from the Commission on Fair Pensions, the contribution base should be substantially increased from the current level of 50% to 75% of profits, ensuring a better harmonisation with the equivalent of gross wages.

Such an increase would reduce inequity between different forms of work and limit the misuse of self-employment to reduce labour cost, which might develop in the future as new forms of work gain in importance (OECD, 2019[1]). This increase may be perceived as generating a too high burden of contributions paid by the self-employed. However, lower pension contributions generating lower pension entitlements should not be used as an instrument to promote self-employment. If it is a policy objective to support self-employment, the cost of such a support should be made transparent by subsidising part of these better harmonised contributions through general taxes. This would avoid that this cost is diluted in an opaque way in terms of future spending on safety nets and/or lower old-age income among the self-employed.

## References

[3] European Commission (2020), Closing the gender pension gap? - Product - Eurostat, https://ec.europa.eu/eurostat/web/products-eurostat-news/-/DDN-20200207-1 (accessed on 4 May 2020).

[9] OECD (2020), Taxing Wages 2020, OECD Publishing, Paris, https://dx.doi.org/10.1787/047072cd-en.

[7] OECD (2019), OECD Reviews of Pension Systems: Portugal, OECD Reviews of Pension Systems, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264313736-en.

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

[4] OECD (2019), Will future pensioners work for longer and retire on less?, https://www.oecd.org/pensions/public-pensions/OECD-Policy-Brief-Future-Pensioners-2019.pdf (accessed on 19 August 2019).

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

[6] OECD (2018), OECD Pension Outlook 2018, OECD Publishing, Paris.

[5] OECD (2017), Pensions at a Glance 2017: OECD and G20 Indicators, OECD Publishing, Paris, https://dx.doi.org/10.1787/pension_glance-2017-en.

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

Benefit losses from non-employment spells differ widely depending on the type of spell. In this Annex, the dynamics behind benefit losses from non-employment spells are explained.

The denotations are those used in Box 1.1. The earnings-related pension is the product of: the reference wage (), which is a function of the average wage for pension purposes, ; the total number of validated years, itself broken down into worked years ($\begin{array}{c}{N}_{w}\end{array}$) and non-worked validated years (${N}_{v}$); and, the annual accrual rate (1.5%):

Equation 1

The average wage for pension purposes is computed as total wages divided by the sum of the number of worked years (${N}_{w}$) and the number of non-validated years (${N}_{nv}$):

Equation 2

implying that non-validated years enter as zeros: , in which $w$ is the average wage while working, assumed not to be affected by non-working years throughout this Annex. Therefore, non-validated periods negatively affect the average wage for pension purposes included in the benefit formula.

The reference wage, shown in Figure 1.10, Panel A, is a function of the average wage based a non-linear weighting pattern that ensures its slow increase only as the average wage grows. More precisely, the reference wage function $F$ is piece-wise linear as a function of $\stackrel{-}{w}$, implying that it is not differentiable at all points:

Equation 3

To calculate the reference wage, a very progressive formula is used under which income thresholds are applied. Up to the threshold (${T}_{1}$) of CZK 14 388 in 2019, the wage is fully taken into account. Between this threshold and the pensionable-earnings cap (${T}_{2}$) of CZK 130 796, only 26% of the wage is taken into account. Earnings over the cap are not taken into account. This means that the derivative is equal to:

Equation 4

The total pension is the sum of the earnings-related pension and the basic pension:

Equation 5

The basic pension is only conditional on the sum of work years and non-work validated years reaching the minimum eligibility period. Throughout this Annex it is assumed that these conditions are met. Therefore, the potential effects of non-employment spells on the basic pension are ignored.

From age 18 until retirement, all years are either spent working $\left({N}_{w}\right)$, in some forms of validated non-contribution periods (${N}_{v}\right)$, non-validated periods for accruals but excluded from the computation of the reference wage (${N}_{nve}$), such as time in education and part of unemployment periods, and non-validated periods and not excluded (${N}_{nv}\right)$. It should therefore be noted that ${N}_{v}$ and ${N}_{nv}$ in the benefit Equation 1 and Equation 2 are not independent of ${N}_{w}$. The retirement age, $RA$, is thus expressed as:

$RA=18+{N}_{w}+{N}_{v}+{N}_{nve}+{N}_{nv}$ Equation 6

The effects of a change in the number of years worked (${N}_{w}$) on pension benefits can thus be studied. Assessing changes in $B$ (Equation 1) at the margin (i.e. taking the derivative):

Equation 7

This expression shows that there is a reference-wage effect (the first element in the large brackets) and a period effect (the second element in the large brackets). The reference-wage effect consists of impact driven by the change in the reference wage () as a result of the change in the average wage for pension purposes ($d\stackrel{-}{w}$), which itself might be affected by the change in the number of years worked ($d{N}_{w}$). This reference-wage effect is proportional to the total number of years worked and validated $\left(\begin{array}{c}{N}_{w}+{N}_{v}\end{array}\right)$. The period effect consists of the impact driven by the direct change in the number years worked itself (equal to $1$) and by the potential related change in the number of validated years (${dN}_{v}\right)$. This effect is proportional to the reference wage. Therefore, using Equation 3 and Equation 4 in Equation 7, the change in benefits as a result of a change in years worked can be expressed as:

Equation 8

in which:

Equation 9

Finally, expressing Equation 6 in changes gives:

$dRA={dN}_{w}+{dN}_{v}+{{dN}_{nve}+dN}_{nv}$ Equation 10

For a given retirement age, a decrease in the period of work (${dN}_{w}<0$) is offset by an increase in either validated or non-validated periods. Of course, if the total number of insured periods (${N}_{w}+{N}_{v}\right)$ falls below 35 and the total number of employed periods (${N}_{w}\right)$ falls below 30, the retirement age needs to increase to be eligible to a pension. This Annex will not analyse the case of a change in the retirement age ($dRA=0$ throughout this Annex).

With these building blocks, it is possible to study the effects of a decrease in the time spent working. The first scenario looks at an increase in time spent on child-care without any prior non-validated periods. This scenario describes an increase in a validated period that is fully reflected in accrual and excluded from the computation of the reference wage. This scenario corresponds to the five-year childcare case in Box 1.1. The second scenario looks at unemployment spells without previous non-validated periods, this corresponds to the three years and nine months unemployment case after age 55 in Box 1.1. Finally, the third scenario looks at an increase in time spent in inactivity, which corresponds to effects seen in the five-year unemployment and 10-year unemployment with late entry cases in Box 1.1.

Fully validated non-employment spells are for instance periods of childcare for a child below the age of four. In this scenario ${N}_{nv}=0$, $\frac{d{N}_{v}}{d{N}_{w}}=-1$ and $\frac{d{N}_{nv}}{d{N}_{w}}=0$, meaning there were no existing non-validated periods and all extra non-employment spells are full accounted for in validated periods. In that case Equation 8 can be written as:

Equation 11

in which:

Equation 12

Equation 13

From these equations, it is clear that there is no effect at all from the fall in employment or the increase in non-employment. Both the reference-wage effect and the period effect are zero. The reason is that, for childcare, accrual continues as before (i.e. the full period is credited), while the period is excluded from the reference wage. This means that accrual is unaffected and the reference wage is unaffected too.23

This scenario looks at partially validated non-employment spells for people without previous non-validated periods and relatively a short unemployment period such that ${dN}_{nv}=0$. Periods like this include for instance spells of unemployment. This scenario is largely the same as the one above except that in this scenario $\frac{d{N}_{v}}{d{N}_{w}}=-0.8$ (i.e. only 80% of time spent in unemployment is credited, the rest is excluded from the benefit calculation) while $\frac{d{N}_{nve}}{d{N}_{w}}=-0.2$. The change in benefits as a result of unemployment can be written as:

Equation 14

in which:

Equation 15

Equation 16

From these equations it is clear that there is only a partial period effect and no reference-wage effect.

This scenario reflects for instance periods of inactivity once unemployment benefits run out. It has ${N}_{nv}=0$ and $\frac{d{N}_{v}}{d{N}_{w}}=0$ and $\frac{d{N}_{nv}}{d{N}_{w}}=-1$. This means that in the past no time was spent in inactivity and the new non-employment spells are non-validated periods. The change in benefits can therefore be written as:

Equation 17

and:

Equation 18

The starting point ${N}_{nv}=0$ implies that initially $\stackrel{-}{w}=w$ (initially the average wage for pension purposes is equal to the average wage while working). Rewriting leads to:

Equation 19

This time both a reference-wage effect (below the second threshold) and a period effect are present. Non-validated periods lower both the reference wage and the accrual. Within the first two reference-wage brackets the penalty goes up with the wage. However, between the brackets it goes down.

## Notes

← 1. The pension system under the communist regime distinguished between three categories of occupations: high risk, medium risk and all other occupations. The high the risk category (such as miners for instance), the shorter the vesting period and the lower the retirement age. For the lowest risk category, the retirement age was fixed at age 60 for men and age 55 for women. Legislation in 1964 differentiated women’s retirement age based on number of their children. After this law, the retirement age for men continued to be age 60, and for women it was set at ages 53 to 57 depending on the number of children they raise. In addition, the vesting period was increase from 20 to 25 years.

← 2. People working longer also means higher pension entitlements, which would offset some of the rise in contributions.

← 3. However, life expectancy is lower in the Czech Republic than in the OECD on average. When accounting for differences in life expectancy between countries, remaining life expectancy at the average labour market exit age is close to the OECD average (OECD, 2019[1]). This suggests that the same length of the retirement period can be achieved in the Czech Republic as in the OECD average while having retired earlier and thus worked during a shorter career.

← 4. Legacy pensions were more generously indexed. However, this is unlikely to have offset the shortfall compared to new pensions.

← 5. This is lower than the 18% gender gap measured by MoLSA.

← 6. Average Wages are estimated by the Centre for Tax Policy and Administration at the OECD. For more information on methodology see (OECD, 2020[9]).

← 7. It should be noted that the thresholds of 44% and 400% of the national average wage correspond to 41% and 375%, respectively, when calculated using the OECD harmonised average wage.

← 8. One year of coverage consists of 365 insured days. The 365 days do not have to be consecutive or in the same calendar year.

← 9. In the past the retirement ages for women rose 4 months per year.

← 10. These numbers are based on underlying assumptions in the OECD pension model. The OECD pension model assumes an annual growth rate of 1.25%. Real-wage growth of 0.75% per year leads to a decrease in internal rate of return from 1.6% to 1.1% for women and from 1.2% to 0.8% for men.

← 11. If Czech pensioners had the same tax treatment as workers (see above for details about their favourable tax treatment), the gross return generating the same net benefit would be significantly higher than these 1.2-1.6%.

← 12. In case of prior non-validated periods there will be a small benefit loss because the non-validated periods (which are included in the reference wage as zero) will gain in relative importance.

← 13. Multiple shorter unemployment spells with periods of work in between would lead to much lower losses as in most OECD countries since unemployment benefits would be received for a longer period in total.

← 14. A fully validated five-year child-care break is possible in the case of having two children born at least one year apart.

← 15. However, losses are only absent if there had not been any prior non-validated periods (${N}_{nv}=0$). In the case of prior non-validated periods (${N}_{nv}>0\right)$, there are some losses due to an increased relative importance of inactivity years in the average wage calculation (). However, periods of child-care with prior non-validated periods have no effect on the reference wage (and therefore benefits) above the pensionable-earnings cap since average wages above the threshold are not taken into account (as long as the average wage stays above the earnings cap).

← 16. This specifically holds only for people without prior non-validated periods.

← 17. If there had been prior inactive periods, the reference wage changes.

← 18. Survivor pensions being indirectly linked to family policies is common in OECD countries. Having or caring for a dependent child can increase the survivor benefits or waive some eligibility conditions such as age or length of marriage requirements in Germany, Hungary, Israel, Portugal, the Slovak Republic, Switzerland and the United States.

← 19. Eurostat: Annual enterprise statistics by size class for special aggregates of activities (NACE Rev. 2).

← 20. For military personnel the rules are slightly different. The initial pension after 15 years of service is lower (5% of the last five years of earnings) but the increase for subsequent years in the career are higher allowing for a maximum of 55% of the last five years of earnings.

← 21. In the pension reform proposal, the committee states that the new system will consist of a zero pillar and a first pillar according to the World Bank classification. However, the reform proposal still sticks to the current set-up of a contributory basic component, which is mistakenly framed as zero pillar while it should still be regarded a first pillar pension.

← 22. The pension committee introduces three scenarios for the new pension system. In the main one (the so-called fair scenario), the increase of the basic component will increase spending. This is (partially) offset by the lower indexation of the earnings-related component. As is currently the case, any shortfall in contribution revenues has to be matched by tax financing. The pension committee also describes a possible “technical” scenario and an “austere” scenario. These scenarios mainly differ in terms of the level of the basic pension and possible bonuses in the system. In the “technical” scenario, the relative level of the basic pension decreases from 30% of the average wage to 25% of the average wage after 2030 to maintain financial sustainability estimates of the current system. This scenario of the reform also eliminates certain child-care and long-career bonuses proposed in the “fair” scenario. The so-called austere scenario largely follows the “technical” scenario except that it sets the basic pension at 22% of the average wage to begin with and reduces further child-care credits. Given the chosen names for the different scenarios of the pension reform it seems clear which scenario is preferred by the Commission for Fair Pensions.

← 23. In case childcare periods occur with previous inactivity (i.e. ${N}_{nv}>0$) the effects are slightly different. This means that in the past some periods were not validated but all extra non-employment spells are full accounted for in validated periods. This time the reference wage effect is not zero:

which means that the total effect is different too:

Increasing validated periods (i.e. decreasing working periods $d{N}_{w}<0$) for the first two reference-wage brackets will have a slightly negative effect on benefits. This is reflected in the reference wage effect only while the period effect is zero (i.e. accrual continues). The wage effect is negative since the non-validated years will gain in relative importance compared to the years spent working . The effects for the highest reference-wage bracket is zero since wages above threshold ${T}_{2}$ are excluded from the reference wage.

The negative effects within the first two reference-wage brackets increase with the wage level, but decreases from the first to the second reference-wage brackets, since only 26% of the wage above the threshold is taken into account for the reference wage.