Disentangling annuities and transfers: The case of Greek retirement benefits

1. Introduction

Until 2010, conventional wisdom had it that retirement pension benefits, once awarded, could not realistically be cut; at least not in nominal terms (which in Europe, in an era of low inflation, amounts to the same thing as real terms). The political obstacles are numerous: on the one hand, the elderly make up a large and growing part of the electorate; on the other hand, younger votes are likely to view a cut in pensions as a betrayal of the social contract (Boeri et al. 2001). As a result of that, governments inheriting (or helping to build) pension systems that turn out to be unsustainable, making excessive promises on future generations, find themselves with little room for maneuver. Reforming pensions, though difficult, is still possible, but the political imperative is to spare current retirees from the pain of adjustment, placing the entire burden on the shoulders of workers, current and future. In 2010-2013, under the terms of economic adjustment programmes, Greek pension benefits were cut, and rather drastically at that. That contrasted with what occurred in Southern European countries attempting fiscal consolidation at about the same time, like Portugal (where cuts were significantly less severe), or Spain (where pension benefits were frozen in nominal terms, as indexation was temporarily suspended) (Guillén et al. forthcoming).

This paper assesses the effects of the cuts on intra- and intergenerational equity. Specifically, it draws on a unique administrative dataset (a large sample of workers retiring in 2008) to estimate the relative importance of annuities and transfers in Greek retirement benefits, before and after the 2010-2013 cuts.

Since one of the main objectives of a pension system is to redistribute income over the life span, its equity effects can be best examined by adopting a longitudinal approach. This approach compares the balance between total contributions paid and total benefits received by individuals throughout the course of their lives. If the present value of lifetime benefits is equal to the present value of lifetime contributions, then retirement benefits can be said to be actuarially fair, equivalent to the annuities offered by private insurers. In contrast, when lifetime benefits exceed lifetime contributions, current pensions receive an implicit transfer, i.e. a subsidy from the rest of society and/or by future generations. That transfer can be negative, in which case it is current pensioners who are paying a subsidy to the rest of society and/or future generations. In this research, lifetime retirement benefits are calculated both according to the rules in place at the time of retirement, and after the spending cuts introduced in 2010-2013, under the fiscal adjustment policies specified in the country’s bailout agreements of 2010 and 2012.

Identifying the relative weight of annuities and transfers in Greek retirement benefits is important. Establishing that current pension rules severely violate inter-generational equity would strengthen the case for pension reform. On the contrary, finding that the relevant transfers are insignificant would inevitably call into question the legitimacy of benefit cuts, suggesting that pensioners have actually “earned” their pensions through contributions paid when they worked by themselves and their employers, in which case attempts to cut back their entitlements would amount to a breach of the implicit social contract.

Moreover, even when net transfers are zero on aggregate, they may still be significant for different groups of pensioners. Departures from actuarial fairness may be entirely consistent with stated public policy goals: this is the case when low lifetime earners are awarded pensions over and above their lifetime contributions, since without the implicit transfer the level of benefit would fail to keep them out of poverty. Alternatively, implicit transfers may be perverse, as when workers with identical contribution histories end up receiving significantly different pension benefits (which violates horizontal equity), or when the direction of redistribution runs from low to high earners (which violates vertical equity). Hence, determining the pattern of annuities versus transfers and clarifying the nature of intra-generational redistribution between groups is of policy relevance.

The main findings of this research can be summarised as follows. Pre-crisis, main old-age pensions provided by IKA (Greece’s largest social security organisation) exceeded their actuarially fair level by wide margins, seriously violating intergenerational equity. The spending cuts introduced under fiscal adjustment, even though savage, improved the progressivity of Greek pensions, significantly reducing the implicit transfer paid to retirees on higher pensions. We estimate the average transfer at 50.4% pre-crisis, vs. 37.0% when the 2010-2013 cuts are accounted for. With respect to intragenerational equity, the Greek pension system treats favourably workers retiring earlier with fewer contributions, a feature only slightly mitigated under fiscal consolidation.

The structure of the paper is the following: section 2 presents the main features of the Greek pension system; section 3 offers key insights from the literature; section 4 explains our methodology; section 5 reports our results and discusses our main findings; section 6 reflects on the policy implications of our findings, on the limitations of our approach, and on issues for further research.

2. The Greek pensions system

Pensions represent the backbone of the Greek social protection system. In 2016, they provided households with 27.7% of their total disposable income, while all other social benefits accounted for a mere 4.5% (ElStat, 2018). Retirement pensions in Greece are public, pay-as-you-go. Until 2017, they were paid out by a large number of social security organisations, known as ‘funds’, created along occupational lines. ¹ In a context of large institutional fragmentation, pension entitlements vary widely between different occupational groups, cohorts and other individual characteristics. As a result of that, workers with identical contributory records may be eligible for very unequal pension benefits.

Expenditure on pensions as a share of GDP is the highest in the EU: in 2016, it stood at 17.7%, five percentage points above the EU-28 average (Eurostat, 2019). Relative poverty in old age has fallen significantly since the late 1990s and is now below the EU-28 average (12.4% vs 15% in 2017). Nevertheless, as a result of the collapse in living standards since 2010, the proportion of the elderly with incomes below 60% of the 2005 median in 2017 was three times as high in Greece as in the EU-28 as a whole (36% vs 11.2%). Arguably, the Greek pension system is failing to deploy the large resources it commands to meet fundamental distributional objectives.

Pension reform has been high on the agenda since the early 1990s, when it first became evident that Greek pensions were neither fiscally sustainable nor socially just. The 1992 reform created a less segmented and significantly less generous system for workers entering the labour market from 1993, but left largely intact the favourable conditions enjoyed by those already in employment. ² In other words, while it addressed some of the inequities of the previous system, the 1992 reform introduced a new dimension of discrimination: pre- vs. post-1993 entrants to the labour market (IMF, 1992; Mylonas & de la Maisonneuve, 1999).

A decade later, the 2002 pension reform set gross replacement rates for all dependent workers at 70% of pensionable income, redefined as the best five out of the ten years before retirement. The reform failed to equalise the age of retirement, or pension entitlements between employees and the self-employed (Featherstone, 2005; Tinios, 2005).

In 2008, another piece of legislation tackled the fragmentation of Greek pensions by consolidating the 155 social insurance funds into 13 (five providing main pensions, eight supplementary ones, and two separation payments). This consolidation was mostly on paper, since many of the merged funds continued to collect contributions and pay out benefits according to pre-existing rules (Tinios, 2010).

In 2010, pension reform returned to the top of the political agenda in the context of the economic crisis. The Memorandum of Economic and Financial Policies that was signed in return for the €110 billion loan agreed on with the European Commission, the European Central Bank and the International Monetary Fund contained the blueprint of a drastic pension reform. The retirement age was set to 65 for all workers with a contributory record of at least 15 years, or 60 for those with a contributory record of at least 40 years. In 2012, the retirement age was raised by two more years. Favourable conditions continued to apply for workers in hard and arduous occupations (typically workers in construction, heavy industry etc.) and for women with dependent children.

Crucially, the 2010 reform had established a new structure combining a tax-funded basic pension with a contributory proportional pension. The new system was due to come into force in January 2015, but the general election of that month disrupted its introduction. The incoming government repealed the 2010 law and brought to parliament a new one. The 2016 reform accelerated the consolidation of Greek pensions, unifying entitlements across categories, and between future retirees and those already retired. It also further reduced accrual rates, especially for those with a long contributory record. At the same time, the 2016 reform reinstated the sharp division between those with at least 15 years of contributions, and those without. The latter may only be eligible for a means-tested national pension at €384 per month, with their contributions counting for nothing. ³ Moreover, the reform abolished the minimum benefit paid to low earners meeting the contributory requirements for a main old age pension (€487 for single IKA pensioners in 2016). A summary of the main characteristics of Greek pensions under the terms of the various reforms can be found in the Online Appendix (Table A1).

Further to reforms, Greek pensions have also been subject to various forms of cuts, under the terms of the 2010 bailout agreement and its subsequent revisions. Nominal reductions in main pensions took place in 2010-2013, and varied with benefit level, and sometimes pensioner age. ⁴ Christmas, Easter and summer bonuses, amounting to two months’ pay (the so-called 13th and 14th pension), introduced back in 1978 were abolished in 2010, which accounted for a 14.3% nominal cut. They were replaced by flat-rate vacation allowances totaling €800 a year, payable to pensioners aged over 60. In January 2013, these allowances were also abolished. A number of special levies on pension incomes (labelled “Pensioners’ solidarity contributions”) were also introduced. The cumulative effect of all fiscal adjustment measures was considerable, with reductions ranging from 14.3% to 45.9%, for pensioners on €600 and €3,000 per month respectively in 2009.

In December 2013, IKA was providing approximately 985 thousand primary pensions, accounting for 27% of the overall monthly pension bill. It was followed by the funds for farmers and self-employed, which provided 715 and 355 thousand primary pensions respectively, jointly accounting for 25% of the total expenditure. Primary pensions were paid by the state budget to 453 thousand retired civil servants; these corresponded to 20% of the overall monthly bill. Finally, around 1.5 million supplementary pensions were also paid, which accounted for 12% of the monthly pension expenditure (Ministry of Labour, Social Security and Welfare, 2013).

3. The literature

The importance of disentangling the annuity from the transfer component of public pensions has been acknowledged both in the academic literature (Feldstein and Siebert, 2002) and in policy documents (Queisser and Whitehouse, 2006; World Bank, 1994). Nonetheless, empirical research on the issue remains relatively limited, hampered by the lack of data (except in few countries, such as the US and Germany). To bridge the gap, simulation techniques have often been used to generate lifetime income flows of hypothetical workers, which are then used to compute the net present value of lifetime contributions and pension benefits.

A comprehensive review of the existing empirical studies in Europe and the US can be found in Grammenos et al. (2006). The estimates of these studies rely on different sets of assumptions and make use of different methodological strategies. Keeping that in mind, some key insights from that literature are as follows. To start with, as regards intragenerational redistribution, most pension systems appear to favour female over male workers. This is found to be the case in the Czech Republic (Klazar and Slintáková, 2012), France (Colin et al., 1999), Germany (Börsch-Supan and Reil-Held, 2001), the Netherlands (Nelissen, 1995) and Spain (Bandrés and Cuenca, 1999). The Spanish, French, German and Dutch pension systems also seem to favour married over single contributors. Low-income pensioners are identified as net gainers from the US, most Latin American and European pension schemes. ⁵ The exceptions are Italy and Spain, where transfers seem to follow regressive patterns, from low to high-income groups (Borella and Moscarola, 2006; Gil and Lopez-Casasnovas, 1999). With respect to intergenerational redistribution, average transfers are found to be positive in all pension systems studied, implying that current pensions are (on average) above their actuarial fair level. Finally, recent pension reforms appear to have weakened the progressive redistributive pattern of most pension systems, and to have reduced the transfer component of pensions.

Moving to the case of Greece, Mylonas and de la Maisonneuve (1999) calculated the present value of lifetime pension benefits and contributions, and the annual rate of return on contributions, under various hypothetical work and retirement scenarios, before and after the 1992 reform. The baseline hypothetical scenario involved male workers in five industries, starting their career at age 25 and retiring after 35 years, having equivalent earnings (growing annually by 2% in real terms), and making pension contributions to the main social insurance fund of each sector. Benefits were assumed to be indexed to inflation, remain constant in real terms, and be paid out for 15 years. They found contribution and replacement rates were very high by international standards, and differed widely across industries, the present value of benefits exceeding that of contributions by wide margins. In the baseline scenario, civil servants seemed to have the most generous main pension scheme, with a rate of return (i.e. the discount rate at which the present value of lifetime benefits equaled the present value of lifetime contributions) of 4.6. Civil servants were followed in terms of generosity by public utility employees, farmers, and self-employed workers. Workers in private firms faced the least favourable rate of return (0.9). Even though pension rules for post-1992 entrants to the labour market were considerably less generous, lifetime benefits continued to exceed lifetime contributions, at an estimated internal rate of return of only 0.2.

Mylonas and de la Maisonneuve (1999) showed that workers identical in all respects apart from social insurance fund affiliation were treated very differently, and in ways that severely violated intergenerational and intragenerational equity. Our paper takes a different approach, analysing a dataset of actual (rather than hypothetical) retirees. This allows for a novel estimation of the distributional pattern of annuities and transfers in Greek pension benefits, many years and several reforms later.

4. Data and methodology

We use a representative sample of 2008 (i.e. pre-crisis) retirees provided by IKA, the country’s largest social insurance fund. The sample covers 4,795 observations (14.7% of all IKA contributors who retired on an old-age pension in 2008). It contains information on gender, age at retirement, length of contributory record, type of occupation (‘hard and arduous’ ⁶ or not), earnings class at retirement, and amount of main old-age pension awarded.

In our sample, average age at retirement was 60.8 for men and 58.4 for women, very close to the ones estimated for service workers and workers in elementary occupations by ElStat (Elstat, 2007). The distribution of retirees by age at retirement and gender is shown in the Online Appendix (Figure A1). It has three peaks for men (at age 59, 61 and 66) and three for women (at age 51, 56 and 61). As many as 87.5% of all cases retired below the age of 65 (45.9% below 60). Average monthly gross earnings on retirement were €1,595 for men and €1,276 for women. Length of contributory record ranged from 11 to 51 years. The average was 28 years of contributions for men and 22 for women. The distribution of retirees by earnings class and by number of contribution years is also presented in the Online Appendix (Figures A2 and A3).

IKA records report the total number of insurees’ contribution days as well as their final earnings, but they do not contain information on earnings at each point in their career. In the absence of that information, we relied on backward induction, i.e. we created synthetic records on the basis of the following assumptions:

Following IKA’s informed suggestion, workers started their career at the age of 22 for men and 24 for women, except if in construction (in which case starting age was 18);

Disentangling annuities from transfers rests on estimating the actuarially fair component of pension benefits, i.e. the one that equalises lifetime benefits to lifetime contributions. This in turn begs two questions: first, what might be a reasonable return on contributions; second, what discount rate might be used to convert the stock of accumulated contributions at retirement into a flow of yearly payments (annuities) over the full length of life in retirement.

As regards the first question, we work with two assumptions: (a) a real rate of return of 2%, or (b) nominal rate of return equal to the yield of 1-year government bonds. ⁸ These are our baseline scenario and sensitivity check respectively. As it turns out, a real annual rate of return of 2% yielded almost twice as high a return in 1957-2007 as that of 1-year government bonds, although the two scenarios balanced out in 1982-2007.

The annual rate of return under each of the two scenarios in 1961-2007 is presented in Figure 1.

OPEN IN VIEWER

Figure 1.

Rate of return on contributions.

As regards the second question, the net present value of lifetime pension benefits was computed on the basis of the following assumptions:

The period over which pensions are paid is equal to people’s life expectancy at the age of retirement. Information on life expectancy at selected ages for both men and women can be found in the Online Appendix (Table A2). ⁹

Disentangling annuities and transfers involves comparing the actuarially fair level of pension benefits with their actual level over individuals’ lifetime. The question arises of how to project into the future the pension benefits awarded at retirement, which are depicted in our dataset. We work with two scenarios:

Pre fiscal adjustment: In the first scenario, we define lifetime pension benefits as they must have looked to their recipients in 2008, i.e. before the Greek crisis erupted. In line with the pre-austerity rules, pensions are held constant in real terms, and are paid 14 times a year.

Table 1 presents the average pension benefits per decile under each scenario in 2014, under the assumption of an actuarially fair return on contributions equal to 2% per year in real terms. The sample’s observations have been ranked in deciles according to the present value of lifetime contributions; as lifetime contributions are the same in both scenarios, this ranking guarantees that the estimated changes are not due to re-ranking effects. We find that, as a result of the 2010-2013 fiscal consolidation policies, main old-age IKA pensions were cut by 27.5% on average. Interestingly, pension benefits in deciles 1-6 were only affected by the elimination of the 13th and 14th monthly payments a year, and by the suspension of price indexation which further eroded their real value. In contrast, deciles 7-10 were also affected by additional benefit cuts implemented directly.

OPEN IN VIEWER

Table 1.

Average gross pension per decile (2014).

deciles	€ per year		% difference
	pre fiscal adjustment	post fiscal adjustment
1	6,838	5,355	-21.7
2	7,240	5,670	-21.7
3	7,719	6,045	-21.7
4	8,335	6,527	-21.7
5	9,172	7,183	-21.7
6	10,779	8,438	-21.7
7	13,542	10,485	-22.6
8	17,148	12,796	-25.4
9	23,037	15,781	-31.5
10	28,304	17,484	-38.2
All	13,210	9,576	-27.5

Notes: Deciles constructed on the basis of the present value of lifetime contributions.

Source: Own calculations, based on the IKA sample of 2008 retirees.

We define the transfer component of pension benefits as the difference between the actuarially fair part (the annuity) and the actual lifetime pension received, as a proportion of the latter. The transfer component represents the implicit subsidy paid to pensioners by taxpayers, as a share of the actual pension received. Arguably, a benchmark pension system striving to ensure intergenerational and intragenerational equity would feature with positive transfers to those in lower deciles, smoothly declining and eventually becoming negative as we move up the pension income scale, that balances out at zero on average. The transfers would have to be reasonably small, so they do not create perverse incentives for early retirement or contribution evasion. Representative hypothetical examples of modelling lifetime contributions and lifetime pension benefits are presented in the Online Appendix (Table A3).

As a short-hand way to compare actuarially fair to actual pension benefits, we also compute the internal rate of return (IRR), defined as the discount rate at which the present value of lifetime pension benefits is equal to the present value of lifetime pension contributions_. If the IRR is greater than the discount rate used to convert the stock of lifetime contributions into a flow of monthly pension benefits, then we conclude that the system pays an implicit rate of return above the actuarially fair one.

5. Results

Have IKA old-age pensioners who retired before the current crisis ‘earned’ their pensions through their (and their employers’) contributions? Or was their retirement subsidised by current and, especially, future generations of taxpayers? More formally, what proportion of their pension benefits was ‘annuities’ (i.e. corresponded to the pension contributions they and their employers had paid over their working life), and what was ‘transfers’ (i.e. amounted to a subsidy paid for by society at large)? How did the 2010-2013 fiscal consolidation affect the relative shares of annuities vs. transfers? In this section, we attempt to provide some tentative answers to these questions.

Tables 2 and 3 present the average transfer components of IKA main old-age pensions pre and post fiscal adjustment, by contribution decile and by gender, under the assumption of an actuarially fair return on contributions equal to 2% per year in real terms. The differences between the two scenarios are presented in the Online Appendix (Table A4).

OPEN IN VIEWER

Table 2.

Average transfer by decile: pre fiscal adjustment scenario.

deciles	as % of actual lifetime pension received			average lifetime transfer (€)
	men	women	all
1	70.2 (67.6 - 72.8)	73.6 (72.5 - 74.6)	72.8 (71.9 - 73.8)	87,563 (85,012 - 90,113)
2	56.1 (54.8 - 57.4)	64.8 (62.8 - 66.9)	61.8 (60.4 - 63.3)	84,345 (81,663 - 87,028)
3	50.4 (49.2 - 51.6)	57.2 (56.1 - 58.4)	54.4 (53.6 - 55.5)	77,586 (75,340 - 79,831)
4	47.3 (46.3 - 48.4)	53.7 (52.6 - 54.7)	50.7 (49.9 - 51.5)	79,312 (77,148 - 81,475)
5	41.5 (40.7 - 42.3)	48.8 (47.7 - 49.9)	44.8 (44.0 - 45.6)	75,416 (73,249 - 77,583)
6	37.0 (34.9 - 39.2)	46.5 (45.4 - 47.7)	40.4 (38.9 - 41.9)	80,376 (77,574 - 83,178)
7	39.7 (38.7 - 40.6)	47.1 (45.9 - 48.3)	42.1 (41.3 - 43.9)	103,923 (100,011 - 107,836)
8	40.8 (39.8 - 41.9)	46.4 (44.1 - 48.6)	42.7 (41.6 - 43.8)	137,725 (132,211 - 143,238)
9	45.4 (44.6 - 46.3)	48.9 (47.4 - 50.4)	46.2 (45.5 - 47.0)	194,771 (188,898 - 200,645)
10	47.2 (46.4 - 47.9)	51.1 (50.0 - 52.1)	47.7 (47.0 - 48.4)	246,359 (240,888 - 251,830)
all	45.2 (44.8 - 45.7)	56.9 (56.3 - 56.9)	50.4 (50.0 - 50.8)	116,728 (114,735 - 118,722)

Note: Deciles constructed on the basis of the present value of lifetime contributions of the full sample of retirees. Confidence intervals (a = 0.05) are shown in parentheses.

Source: Own calculations, based on the IKA sample of 2008 retirees.

OPEN IN VIEWER

Table 3.

Average transfer by decile: post fiscal adjustment scenario.

deciles	as % of lifetime pension received			average lifetime transfer (€)
	men	women	all
1	64.3 (61.3 - 67.3)	67.8 (66.6 - 69.1)	67.1 (65.9 - 68.2)	66,556 (64,526 - 68,587)
2	47.1 (45.6 - 48.7)	56.8 (54.3 - 59.3)	53.4 (51.7 - 55.2)	60,173 (58,081 - 62,265)
3	40.0 (38.6 - 41.4)	47.4 (46.1 - 48.8)	44.4 (43.4 - 45.4)	52,169 (50,433 - 53,905)
4	36.0 (34.8 - 37.2)	42.8 (41.5 - 44.1)	39.6 (38.7 - 40.6)	50,996 (49,327 - 52,664)
5	28.8 (27.9 - 29.8)	36.8 (35.4 - 38.1)	32.4 (31.5 - 33.3)	45,035 (43,364 - 46,707)
6	23.2 (20.7 - 25.8)	33.9 (32.6 - 35.2)	27.0 (25.2 - 28.8)	45,183 (43,007 - 47,358)
7	25.7 (24.6 - 26.7)	34.2 (32.8 - 35.7)	28.5 (27.5 - 29.4)	57,775 (54,968 - 60,582)
8	25.1 (24.0 - 26.3)	31.9 (29.2 - 34.6)	27.4 (26.2 - 28.6)	70,955 (67,244 - 74,666)
9	25.9 (25.0 - 26.9)	31.6 (29.9 - 33.3)	27.2 (26.4 - 28.1)	85,132 (81,664 - 88,600)
10	22.3 (21.4 - 23.1)	29.3 (28.0 - 30.7)	23.3 (22.5 - 24.0)	82,093 (79,171 - 85,015)
all	29.9 (29.3 - 30.4)	46.2 (45.3 - 47.0)	37.0 (36.5 - 37.6)	61,606 (60,718 - 62,494)

Note: Deciles constructed on the basis of the present value of lifetime contributions of the full sample of retirees. Confidence intervals (a = 0.05) are shown in parentheses.

Source: Own calculations, based on the IKA sample of 2008 retirees.

We observe that the transfer component of pensions is significant in both scenarios. Before the fiscal adjustment, 99.5% of all IKA retirees received an implicit subsidy from current and future taxpayers. The few cases of pensioners who did not receive any implicit subsidy were mostly people who retired at a very late age. On average, just over half (50.4%) of IKA pension benefits were transfers, not annuities. The value of the average transfer on a lifetime basis was approximately €116,700. The 2010-2013 measures have reduced the transfer component to 37%, and the value of the average transfer to €59,400. The overwhelming proportion of IKA retirees (99%) continued to receive a transfer. The additional retirees who ended up subsidising the system were people with long contributory histories paying special levies on pensions.

Focusing on gender differences, we find that the transfer component was significantly larger for women than for men (57% vs. 45% pre fiscal adjustment, 46% vs. 30% post fiscal adjustment). Longer life spans for women explain only part of the difference. The other part is due to retirement rules favouring the early exit of women (especially of mothers) from the labour market. The gender gap grew under fiscal adjustment because cuts fell most heavily on those on higher pensions, who tended to be men.

Turning to the distributional pattern of transfers, we find that, for those on the lowest 10% of the distribution, transfers accounted for almost three-quarters of the amount of benefit received (computed on a lifetime basis). This is mostly due to the minimum pension mechanism, intended to ensure that low earners meeting the contributory requirements received at least a certain level of benefit. More specifically, the minimum pension (€487 per month in 2008) was paid to new retirees who would have otherwise been eligible for a lower ‘organic’ pension (i.e. on the basis of their earnings and contributory record alone). At low earnings, the ‘organic’ pension only started to exceed the level of the minimum pension at 30 contributory years. As many as 19% of all IKA retirees in our sample received the minimum pension. As noted by Mylonas and de la Maisonneuve (1999, p.6): “The perverse incentive to retire at the minimum pension, with far less than the normal 35 years of contributions, lowers the effective contribution rate by far more than the effective replacement rate.” In view of that, the share of transfers in pension benefits received was highest at the low end of their distribution.

Pre-fiscal adjustment, the transfer component declined monotonically, from 73% in decile 1 to 40% in decile 6, then rose again to 48% for decile 10 (largely populated by workers with a longer contributory record). ¹¹ The 2010-2013 measures broadly reversed this pattern: the transfer component decreases as we move from decile 1 to decile 6, remains constant (at 27-28%) in deciles 6-9, then falls to 23% in decile 10 (i.e. less than half what it was pre fiscal adjustment).

Figures 2 and 3 show the position of all observations in our dataset in terms of lifetime contributions and lifetime pension benefits, under each scenario. The black solid line (the 45-degree line, if the axes were drawn of equal length) represents actuarial fairness, when lifetime contributions equal lifetime benefits. OLS regression lines are also fitted, with lifetime contributions explaining around 85% of the variability in lifetime transfers. ¹² Fitted lines in both figures are over the 45-degree line, implying positive transfers (as actual pensions exceeded their actuarially fair level), while their slope decreases as we move from pre- to post-fiscal adjustment (as the cuts reduced transfers, without fully restoring actuarial fairness).

OPEN IN VIEWER

Figure 2.

Scatter plot and fitted OLS regression line: pre-fiscal adjustment

OPEN IN VIEWER

Figure 3.

Scatter plot and fitted OLS regression line: post-fiscal adjustment

The high variability of transfers between different population sub-categories is clearly depicted in Table 4. Our estimates suggest that the transfer component of pensions is proportionally larger the lower the age at retirement. Even though the average transfer components are falling for all age clusters as we move from the pre- to the post-fiscal adjustment scenario, the difference in the transfer parts between those retiring at the age of 55 or earlier and those retiring after they reach the age of 65 remains relatively stable (around 12 percentage points).

OPEN IN VIEWER

Table 4.

Average transfer component of pension benefits by group.

	as % of actual lifetime pension received		share of all cases (%)
	pre-fiscal adjustment	post-fiscal adjustment
age at retirement
<=54	61.8 (60.5 - 63.1)	50.6 (48.9 - 52.3)	7.4
55-59	51.2 (50.7 - 51.6)	36.9 (36.3 - 37.5)	38.5
60-64	48.6 (47.9 - 49.2)	35.3 (34.4 - 36.2)	41.6
>=65	47.2 (45.5 - 48.9)	35.2 (33.1 - 37.3)	12.5
contribution years
<=15	62.9 (62.1 - 63.7)	54.9 (54.0 - 55.9)	15.5
16-20	56.5 (55.7 - 57.4)	46.8 (45.8 - 47.8)	22.1
21-25	46.8 (45.9 - 47.6)	34.6 (33.5 - 35.6)	16.8
25-20	41.9 (40.9 - 42.9)	27.9 (27.6 - 29.0)	13.9
31-35	44.3 (43.4 - 45.1)	26.9 (25.9 - 27.9)	17.9
>=36	47.3 (46.2 - 48.4)	26.7 (25.4 - 28.0)	13.9
hard and arduous
no	53.1 (52.6 - 53.7)	40.5 (39.8 - 42.1)	64.2
yes	45.4 (44.9 - 45.9)	30.8 (30.2 - 31.4)	35.8
all	50.4 (50.0 - 50.8)	37.0 (36.5 - 37.6)	100.0

Note: Estimated on the basis of a real rate of return on contributions equal to 2%. Confidence intervals (a = 0.05) are shown in parentheses.

The pre-fiscal adjustment system seems to be benefiting both those with the least contribution years as well as those with the most. This picture changes drastically after the implementation of 2010-2013 measures; the non-contributory part of pensions for those with the longest contribution histories (i.e. more than 31 years) decreases substantially to 27 percent approximately, i.e. 10 percentage points below the estimated average of this scenario. The transfer component of retirees previously employed in hard and arduous occupations is 5 (6) percentage points below the average in the pre- (post-) fiscal adjustment scenario.

The results shown so far have relied on the assumption of a real rate of return equal to 2% in real terms. Using the yield of 1-year government bonds would result in higher estimates of average transfers (from 50.4% to 53.6% pre fiscal adjustment, and from 37.0% to 41.1% post fiscal adjustment), but would not qualitatively affect the pattern observed.

As seen in Figure 4, the estimated internal rate of return on lifetime contributions confirms our analysis. At 10.7% pre-fiscal adjustment (7.9% post-fiscal adjustment) on average, IKA pensioners enjoyed (and still do) a far higher rate of return on contributions than what the bonds market might have paid. Moreover, unlike what usually happens in financial markets, those on lower pensions actually did considerably better than those on higher ones.

OPEN IN VIEWER

Figure 4.

Internal rates of return.

Even if the return on contributions at IKA were assessed against a discount rate equal to that of 32-year government bonds (4.95% in 2008), it would still be above the actuarially fair level. Nevertheless, adopting such a yardstick would result in near actuarial fairness for retirees in the upper half of the benefit distribution (computed on a lifetime basis).

As a sensitivity check, we converted lifetime contributions to present values using a rate of return equal to that of 1-year government bonds. In that case, the estimated average rate of return would be slightly higher (11.1% vs. 10.7% and 8.3% vs. 7.9% pre- and post-fiscal adjustment respectively), distributed along the same lines.

6. Concluding remarks

The purpose of this paper is to investigate the relative importance of annuities and transfers in Greek retirement benefits and to assess, using a longitudinal approach, their impact on intra- and intergenerational equity. This compares the balance of lifetime contributions paid to lifetime benefits received, the difference between the two being the implicit transfer or subsidy from current and future taxpayers. Based on a representative sample of IKA retirees in the year 2008, the average transfer and internal rates of return of main old age pensions were calculated according to the pension formulae in place both before and after the fiscal consolidation of 2010-2013.

We find that the vast majority of IKA retirees receive sizeable net transfers. In other words, the present value of lifetime pension benefits largely outweighs the present value of lifetime social insurance contributions. This implies that the underlying pension rules seriously deviate from actuarial fairness, violating intergenerational equity. The widespread perception that pensioners have fully earned their retirement benefits is not supported by the evidence. Large transfers are needed, over and above the contributions built up by the pensioners themselves (and their employers) before they retired. Until 2010, the difference was partly made up from taxation, and partly from borrowing. Since then, with the Greek government committed under the terms of the economic adjustment programmes to return a primary surplus for many years to come, the scope for borrowing is drastically limited. Transfers to pensioners can now only be financed out of the gross earnings of workers (and their employers), on top of those normally required for an actuarially fair PAYG pension system.

Our analysis shows that the departure from actuarial fairness remains considerable: 2010-2013 measures improved intergenerational equity by reducing (not fully eliminating) transfers from just over 50% to 37% on average. Relative to pre fiscal adjustment, when the distribution of transfers by lifetime contributions was U-shaped, favouring high earners with long careers as well as low earners with short ones, the cuts reduced implicit subsidies to the former. Even though there might be cases where subsidising all pensions, rather than just the low ones, can be justified (when for example pay-as-you-go pension systems have not yet reached maturity, as in most European countries post-WWII, and/or when pensioners/the elderly are poorer than workers/the young), none of these conditions hold in Greece at the present time.

As regards intra-generational equity, the earlier system treated more favourably those who retired earlier, with fewer contributions, on a lower (often the minimum) pension. This pattern was only slightly affected by fiscal consolidation. Somewhat surprisingly, the favourable treatment of ‘hard and arduous’ occupations in terms of early retirement was found to be more than fully offset by the higher contributions paid by the workers concerned and their employers.

Within current cohorts of pensioners, women benefit considerably more than men. The reason is twofold. On the one hand, lower earnings and shorter careers for women produce a smaller annuity component, made even smaller by the minimum pension mechanism. On the other hand, women’s longer life expectancy combined with favourable rules for early retirement renders the transfer component of their lifetime benefits significantly higher than that for men. Fiscal consolidation, by falling more heavily on higher earners with longer careers, who tend to be men, have unwittingly amplified the gender gap in lifetime pension transfers. Nevertheless, the country’s very low employment rate of women, combined with the 15-year contributions record required for reaching pension eligibility results in a wide pension coverage gender gap (Burkevica et al., 2015).

A certain amount of caution is called for when interpreting the above results. The main caveats are discussed below.

As only the final earnings of retirees were available in the data, individual contributory histories were reconstructed on the basis of a backward induction, assuming that earnings grew in line with minimum wages. ¹³ To some extent this assumption is not unreasonable: the course of minimum wages broadly reflected that of the economy overall, while almost a quarter of all retirees in our sample earned wages at or just above the minimum. However, more erratic earning profiles, rising faster or slower than the minimum wage, affected by periods of inactivity or unemployment, are bound to exist. To some extent, they are also bound to cancel each other out. As the labour economics literature suggests, time spent out of work lowers the trajectory of future earnings (Kletzer, 1998). Including such spells when reconstructing individual contributory histories would lower the present value of lifetime contributions and hence further increase the estimated transfer component of pension benefits. The same would occur if trends in average earnings rather than minimum wages had been used to reconstruct individual contributory histories. As average earnings rose faster than minimum wages in 1991-2008 (when data on average earnings are available), using the former would result in lower lifetime contributions, and hence a larger transfer component.

Furthermore, our results are mainly based on the hypothesis that actuarial fairness implies a real rate of return on contributions and a discount rate of future benefits equal to 2%. ¹⁴ Other rates might conceivably have been used instead. Sensitivity analysis, applying a rate of return equal to yearly government bonds, showed that our findings are robust to that choice. In any case, we also estimate internal rates of return, which allows us to refrain from the use of specific discount rates and invites policymakers to reflect upon what an acceptable rate of return might be.

Turning to pension benefits, IKA data allowed for a detailed simulation of lifetime pension flows both in the absence and after the implementation of the 2010-2013 measures. In doing so, we made use of ElStat’s life expectancy estimates, which only account for differences in gender and age at the time of retirement. However, there is evidence that richer and better educated persons enjoy higher life expectancy (OECD, 2017b). Accounting for these differentials would tilt the curves depicted in Figure 4 in favour of higher earners with longer careers.

Finally, in the absence of data from other social insurance funds our analysis is limited to IKA old age pensions. Extending our research to more funds would allow for a more comprehensive assessment of the impact of annuities and transfers on intra-generational equity. Nevertheless, we can safely infer that the addition of the two biggest funds after IKA (i.e. for farmers and civil servants) would further reinforce the imbalance between lifetime benefits and lifetime contributions as (a) famers were not required to pay any social insurance contributions until 1987, whereas contributions for main old-age pension only became compulsory in 1998, and (b) in the case of civil servants, until 2007 the conditions for individuals who have entered the labour market before 1993 were much more favourable than private sector workers. Adding survivor and disability pensions into the frame of analysis would also increase the relative importance of transfers, as the annuity component of these pension benefits is by definition small.

Overall, our research reveals that the link between lifetime Greek pension benefits and social insurance contributions remains weak. It is not an easy task to design a pension system that is fair to current as well as future generations and one that ensures that benefits offer protection from poverty in old age and that, at the same time, enhances incentives to work and pay contributions (Shokkaert and van Parijs, 2003). Trade-offs are present; for instance, between the need to protect low pensions (deviating from actuarial fairness to enhance overall progressivity) and the desire to guarantee a decent return on contributions for all, including those who have contributed the most. Even though our estimates suggest that the 2010-2013 fiscal consolidation strengthened that link, plenty of scope for improvement remains.

Supplemental material