Turning potential retirees into workers: pension reform act of 1999 and beyond in Turkey

TURNING POTENTIAL RETIREES INTO

WORKERS:

PENSION REFORM ACT OF 1999

AND BEYOND IN TURKEY

Serdar Sayan

Bilkent University, Turkey

ABSTRACT

Turkey's publicly managed PA YG pension system experienced serious financial difficultie during the past decade, forcing policy makers to legi late a pension reform act in 1999. Since past reductions in statuto,y entitlement ages for pension benefits were among the prima1y reasons behind the losses generated by the system, there was an evident need to extend average contribution periods. Accordingly, the major change introduced through the reform act of 1999 turned out to be a significant increase in entitlement ages for female/male workers from 3 /43 to 58/60 to become effective after a transition period of eight years. Yet, in Februa,y 2001, the Turkish Constitutional Court ruled that the gradual increases in the entitlement age should be made smoother within ix months.

This paper investigates the implications of alternative schemes to increase statuto,y enhtlement ages over time for the retiree-to-worker ratios and the composition of working age population. For this purpose, a genetic algorithm-based search routine is employed and po sible rime paths of worker-to-retiree ratios over the 2000-2060 period are identified under alternative cenarios. using projection data for workers and retirees covered by SSK. the largest state-run pension institution in the country. The result indicate that the Turkish Constitutional Court's decision requiring the gradual increases in entitlement ages to be slowed down would not only affect the total supply of labor to the formal sectors but would also change the age composition of workforce.

I.

Introduction

Turkey's pay-as-you-go (PA YG) pension system run by the state experienced seriou financial difficultie during the past decade. Like other PAYG schemes, the Turkish system has been financing pension (old-age) benefits to current retirees out of contributions collected from currently active worker since the 1940s. The system began to generate significant losses from 1990 on, since the generosity of benefits relative to contributions was retained even after the ratio of acti e workers to retirees started to decline at an increasing rate following the 1980s. The rapidly growin contribution of pension deficits to public sector's borrowing requirements proved to be

unsustainable by the end of the 1990s, necessitating a pension reform (Akmaz, 1999; Topal, 1999; Sayan and Kiraci, 2001 a and 2001 b; Teksoz and Sayan, 2001 ).

The crisis of Turkish pension system came at a time when PA YG pension system in several

other countries were encountering financial difficulties of their own, due mainly to the decreasin~ worker-to-retiree (or active-passive) ratios (Kohl and O'Brien, 1998). Differently than the e countries where decreasing worker-to-retiree ratios resulted from population aging, however the fall in this ratio in Turkey was due primarily to exceptionally low entitlement ages for pension benefits (minimum retirement age). 1n fact, the pension regulations allowed female/male worker to

begin collecting retirement benefit as early as 38/43 years of age. 1 _{The mismanagement of pension} 1 Even though rela1ively few people (less than 5% in 1995) actually took retire younger than 40 years of age. more than

Turning Potential Retirees imo Workers: Pension Reform Act ofl999 and Beyond in Turkey

funds resulting from actuarilly unsound entitlement ages and other populi t practices introduced by policy makers gave rise to the pension crisis of the 1990 although Turkey still had a predominantly young population (Sayan, 2001).2

When the losses reached at magnitudes jeopardizing the solvency of the system, the

government prepared a proposal for a major pension reform. The evident need to extend the

average contribution period (and shorten the benefit collection period) made the entitlement age a

critical parameter for pension reform. In the early drafts of the pension refom1 bill, the entitlement

age was set at 63/65 for female/male workers, representing a substantial increase over the existing

levels. The drafts also included adjustments in the values of contribution and replacement rates, the

other two critical program parameters affecting pension balances. Under inten ive pressw·e from

trade unions and other interest groups, however, the government modified the propo al to lower the

entitlement ages to 58 and 60 for female and male workers, respectively, and allowed for a

transition period of 8 years for making these effective.

Legislated in September 1999 following heated debates between the government and the

opposition group inside and outside the Parliament, the parametric reform bill was an imperative

step to control the rapid growth in pension (and hence, social security) deficits in the short-run and

to reduce them in the medium-run. Since the complete elimination of deficit over the short- to medium-run would have required more radical measures (which were not necessarily feasible politically), the act had been intended to be just the first stage of a larger reform to overhaul the

system.3 _{The next step towards the completion of reform process is the planned introduction of}

money purchase pension scheme to be managed by the private sector (IMF. 2000). While the legal

and regulatory framework has already been drawn up through a bill currently awaiting ratification

of the Turkish parliament (Teksoz and Sayan, 200 l ), a recent ruling by the Turkish Constitutional

Court (or the Supreme Court) required that the process of reform be reconsidered. In February

200 I the Constitutional Court ruled. following it examination of an appeal by the deputies from

main opposition party in the Parliament that the bill's scheme for gradual increases in the

entitlement age during the transition period violated the constitutional rights of workers in certain

age groups, and hence, the equality principle. The Constitutional Court gave the government six

months to make arrangements so as to make sure that the transition to new entitlement ages will be

smoother.

The purpo e of this paper is to investigate possible, alternative configurations of contribution

and replacement rates and statutory entitlement ages that could be considered for a parametric

pension reform involving such a smoother transition, and to di cuss some of the implications of

introducing these alternatives for the retiree-to-worker ratios or the composition of working age

population in Turkey. The paran1etric reform alternatives are identified through the genetic

algorithm-based search routine developed by Sayan and Turhan-Sayan (2000) by using the

pre-reform projections for Social Security Institution (SSK) - the largest of three state-run pension

institutions in the country, covering more than 40% of Turkish population (Topal, 1999). The

computational exercises in the paper look at the paths of worker-to-retiree ratios are likely to take

under alternative simulation scenarios over the period from 2000 to 2060.

Turkish experience with pension crisis and reform i interesting and worth studying for

several reasons. Firstly, the pension crisi bit Turkey much earlier than expected - when the

country still had a predominantly young population, due primarily to mismanagement practices

including the entitlement ages that were set exceptionally low by international standards, rather

than population aging (Sayan, 200 I· Sayan and Kiraci 200 I a and 200 I b; Topal, 1999). The

critical role played by the statutory entitlement age bas set the pension refonn in thi country

institutions, the median retirement age has been significantly lower than OECD average (ILO. 1996: Gillion and Cichon.

1996).

2 _{In 1995. 43% of Turki h population wa younger than 20 years of age, 61 % wa younger than 30 and almost 76% was} rounger than 40 ((LO, 1996).

The results in ayan and T urhan-Sayan (2000) provide evidence that the parametric reform of 1999 is not likely to

Serdar Sayan

apart from the experience in many other countries, making it particularly instructive for countries where pension laws facilitate early retirement -such as Kuwait and other GCC countries (Sirageldin and Al-Rahmani, 1999). Secondly, the governments' failure to enforce laws regulating the pension system in the country has led to substantial leakages from the system,

particularly in the form of un(der)reported wage payments and delayed transfer of contributions collected by employers to social security institutions (ILO, 1996). It was shown in Sayan (1999) that maintaining any given pension balance required higher entitlement ages and/or higher (lower) contribution (replacement) rates at the current magnitudes of these leakages in Turkey. So, the recent bill has devoted considerable space to new measure introduced to enforce the rules.4 Finally, the pension reform process is still continuing in Turkey, despite the set back by the recent decision of the Constitutional Court. The second stage of pension refom1 process will eventually take off, enabling individual workers to purchase additional coverage through privately managed defined-contribution (money purchase) retirement plans to complement the compulsory basic scheme offered by the state.

While the resolution of compliance problem and the transition to privately offered plans will have important implications for labor markets in Turkey (Sayan, 1999; Teksoz and Sayan, 2001),

their exact outcomes will depend largely on the effects of parametric reform currently under way. It is therefore necessary to explore the directions that this reform process might take following the ruling by the Constitutional Court first, and the present paper tackles this issue.

The rest of the paper is organized as follows. The next section briefly overviews the structure and the problems of Turkish pension ystem. Section III describes the approach employed in the paper to investigate alternative reform strategies for controlling the losses generated by the state-run pension system through adjustments in the value of pension program parameters, and report results. Section IV discusses the possible effects of reform on the composition of working age population in Turkey.

II.

Turkish

Pension Crisis and the Reform Act of 1999

Covering working individuals against loss of wage/salary income due to aging is a primary task for governments. For thjs reason, compulsory old-age insurance coverage in many countries is provided through publicly managed pension funds which are. typically run as Pay-As-You-Go (PA YG) schemes. The publicly managed pen ion system in Turkey is made up of three different and distinct branches each operating on a PA YG basis: SSK, Bag-Kur and Emekli Sandigi (hereafter, ES). Of these, SSK (Social Security Institution) provides social security -including old-age in urance (pension and health) coverage- to blue-collar workers employed in the public sector and blue-and white-collar workers in the private sector and covers more than 40% of population. Bag-Kur

covering farmers, artisans and other elf-employed people, and ES, the pension fund administration for whjte-collar workers employed.by local and central governments, each has a membership share between 15 and 20% of population.5 _{Prior to the legislation of 1999, the losse generated by these} three branches together corresponded to one of the major components of public sector deficits funded directly by the Trea ury.

Total social security deficits -the largest part of which comes from pension deficits- also kept growing rapidly both in absolute value and in proportion to the GDP (G P) starting from the beginning of the 1990s (Table I). Projections for the period between 1995 and 2050 indicated that the GDP shares of individual deficits by these institution would rise from 1.21 % to 7.48% for SSK;

from 0.53% to 1.83% for ES, and from 0.07% to 0.75% for Bag-Kur if the pension parameters used until 1999 were to be maintained (ILO, 1996).

; Even though the compliance ha been an issue also in other countries uch as Germany (Boll, Raffelhuschen and Walliser. 1994 ), the pre-refom1 difference between actual and potential levels of contribution receipts wa e ti mated to be as high a 20% in Turkey (Sayan. 1999).

Turning Po1e111ial Re1irees info Workers: Pension Reform Ac! of 1999 and Beyond in Turkey

Table 1: Growth of Social Security Deficit in Turkey

Share of Share of Total

Years Soc. Sec. Deficit in Transfers to Soc. Sec.

Total Budget Deficit ( %) Institutions

In GNP ( %) 1993 17 1.20 1994 20 1.13 1995 36 1.44 1996 26 2.16 1997 34 2.55 1998 35 2.85 SOURCE: Ayas ( 1998), pp. 48-49.

Given that Turkey still has a predominantly young population, thi unexpectedly early advent of the cri is was a result, first and foremost, of the populist moves of legislatures and governments which, among other things, lowered the statutory entitlement age (minimum retirement age) at

variou points in time (Sayan and Kenc, 1999; Sayan, 200 l ). In 1965, for example, a law was enacted to reduce the retirement age for female workers covered by the SSK from 60 to 55. Another

law passed in 1969 broke the link between eligibility for retirement and the age of SSK workers

and introduced the number of days for which old-age insurance premiums (i.e., pension

contribution ) were paid as the relevant criteria, essentially enabling people to retire at very young ages. More recently in 1991, a governmental decree lowered the minimum contribution period required for retirement even further. Just before the legislation of the pen ion reform act of

eptember 1999, it was possible for a female (male) worker to begin collecting pension benefits as early a 38 (43) years ofage (Sayan, 2001).

For many of the workers taking an early retirement, opportunities existed to find second jobs,

enabling them to earn a second income in addition to their pensions, sometimes in the informal ector. As a result, low entitlement ages led to low actual retirement ages. Despite difference

across social security agencies, the average actual retirement age have stayed very low as compared to other countrie . While nearly 65 years of age is typical of other OECD countries, the median retirement age in Turkey was about 53 for people covered by SSK and ES, and slightly higher than 60 for Bag-Kur (Gillion and Cichon, 1996).

In addition to reduction in the entitlement ages, eleven different laws pa ed by the Turki h Parliament after 1969 extended pension coverage to previously uninsured people from different occupational categories, on the condition that they would pay modest amounts of lump sum contributions, causing, in turn, many relatively young people to begin collecting benefits (Sayan and Kenc, 1999; Sayan, 200 I).

Coupled with such developments, reductions in entitlement ages have created a strain on pension balances of the ocial ecurity y tern by extending the average retirement period and giving rise to a trend of rapidly decreasing ratios of contributors to pensioners from one year to another. The contributor to pensioner ratios (also called active/passive ratios) were alarming: The ratio for SSK stood at 1.9 in 1995 representing about one-third of the OECD average of 6, or a more than 3 times reduction a compared to its 1975 value of 6.3. The respective values for ES and

Bag-Kur were about 1.6 and 2.9 (Sayan, 200 I). Even a more gloomy picture was painted by the projected values of thi ratio very clearly indicating the unsustainability of policies in use before

I 999: the contributor to pensioner ratio was projected to fall below I for all three institutions (0.87 for SSK, 0.78 for ES and 0.71 for Bag-Kur) by 2050, had the pre-1999 contribution/replacement

Serdar Sayan

rate and statutory entitlement age configuration remained effective (ILO, 1996). Figure 1 shows the

projections for SSK over the period 1995-2060.

Figure 1: Projected Trend of Number of Active Workers per Retiree

under the pre-1999 Pension Parameter Configuration

4.00 C '\;j 0:: 3 .00 41 !: ~ 2.00 .!.. 41 .lo: 'Q 1 .00

3:

I I I I I I I I I I I

--•--~---•---~--~--~--~--

1 I I I I I I

-

•---~--·--~--~---~--~--•-

I I I I I I I

-

~-

I

-I I I I I I I I I I I I I I I I I I - .J _ - - ' - - -L - -1. - - J -- .J - - _I_ - - L - - l - -J - - -'-- -L -- .L. - - .1 - -.J _. I I I I I I I I __ I _ _ I _ _ _ 1 _ __ I __ I _ _ I _ _ I - _ _ _ _ _ _ _ I I I I I I I I I I I I I I I I I I 0 .00 + -- - r - -r---r---r- . - - - . - - - r- ...--r-- - . --r---r- . - - - r - - - r- ...--1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 206 Years

The populist interventions to the ystem have not stopped at disturbing actuarial balances. The

governments' direct control over publicly managed pension institutions enabled them to u e surplus

funds generated by the pension system before 1990 in financing agricultural support purchases and

the losses of certain state-owned enterprises preventing the surplus funds from being inve ted to

earn competitive rates of return, and further contributed to financial difficulties of the pension

system in the 1990s. The introduction of lump sum transfers to retirees not tied to previou ly paid

contributions under populist motives also helped deepen the crisi (Sayan, 1999 and 200 l ).

La t but not least, actual contribution receipts ( old-age premia collection) have consistently

been lower than potential levels. Despite the availability since the early 1940s of quite

comprehensive laws regulating the pension ystem in the country, the governments' failure to

enforce these laws has led to substantial leakages from the system, particularly in the form of

un(der)reported payments to labor by employers. The employers' failure to comply with the legal

requirements concerning timely transfer of contributions collected to social ecurity institutions has

also contributed to the growth in pension deficits (ILO, 1996; Gillion and Cichon, 1996; Sayan,

1999 and 200 I).

As a result, the social security system began to place ever increasing financial burdens upon the Treasury: SSK, for example, began to receive Treasury support to be able to make monthly

pension payment in 1994. This was the first year when the institution asked for such support

following its establishment in 1945, and the amount of upport needed has increased ever since (Sayan, 200 I).

The rapidly growing contribution of social security system to public sector imbalances made the need for reform increasingly visible, eventually proving to be unsustainable by the end of the

1990s when the whole social security ystem in the country came close to a financial collapse under

the burden placed upon the system mainly by young retiree (Topal, 1999; Akmaz, 1999; Sayan and

Kiraci, 200 I a and 200 I b ). Even in the 1990s, the pension system absorbed most of the direct

subsidies and transfers by the government and the total deficit of three pension systems was

projected to increase from 1.8% to I 0.1 % of GDP by the year 2050. Perhaps even more striking! .

the accumulated deficit was projected to reach to 316 per cent of the GDP until 2050 (ILO, 1996).

Public deficit were about to grow out of control unless the sy tern was not seriou ly

Turning Potemial Retirees into Workers: Pension R~fornt Act of 1999 and Beyond in Turkey

incumbent coalition government that assumed power in 1999. Since the only way to bridge the gap

between revenue and expenditures of the pension system was to increase contributions or reduce benefits by changing the existing values of pension program parameters, a reform was sure to

reduce the popularity of government. Even when political considerations are left aside, however,

the required increase (decline) in CR (RR) to eliminate the deficit wa simply outside the feasible range for these parameters (Sayan and Kiraci, 2001b). So, it was clear that an adjustment in contribution or replacement rates alone would not do the job, and the reform would have to involve an adjustment in entitlement ages, possibly along with adjustments in CR and RR.

The proposed ages in an early draft of the bill were 63/65 for female/male workers. Yet, even the talk of increases in the entitlement ages initially faced a very strong opposition from various groups including the unions and opposition parties in both right and left. By citing life expectancy figures, the unions argued that the draft indicated the government's intention to force the working masses to spend their retirement "in their graves". The opponents conveniently based their argument on estimated life expectancy at birth, even though the relevant figure was expected

duration of survival for people reaching the retirement age (Sayan, 2001 ). According to the State

In titute of Statistics (SIS) e timates, Turkish adults reaching the age of 60 in 1995 had an average expected life of 17 years (18.2 years for women and 16 years for men) whereas the average life

expectancy at birth was 68.5 (70.9 for women and 66.3 for men). These values were projected to

increase to 18.6 and 71.8, respectively, by the year 2020 (!LO, 1996).

On account of the significant contribution to pension deficits of low entitlement ages (Sayan

and Kiraci, 2001a and b), the government yielded to such intensive pressure from trade unions and other interest groups only by modifying the proposal to lower the entitlement ages to 58 and 60 for female and male workers, respectively, and keeping adjustments in the contribution and replacement rates at a minimum. In fact, the average contribution and replacement rates (which, for SSK, stood at 20% and 65%, respectively) stayed basically intact even after the refom1, and the

major change introduced by this "landmark legislation" a it is called by IMF (2000) turned out to

be the increase in statutory entitlement ages from 38/43 for female/male workers to 58/60.6 _There,

however, was a considerable number of workers who were sufficiently close to the pre-reform

entitlement age of 38/43 and planning to retire within a few years at the time of the legislation.

Since forcing these workers to postpone retirement plans for several years was neither fair, nor politically feasible, the reform act allowed for a transition period of eight years to gradually increase the retirement age over time. The scheme dictated by the new law for gradual increases in the retirement age is given in Table 2.7

6 _{nother important change introduced through the reform was tbe switch from nominal to real wage in the calculation of} ~ension benefit , as already incorporated into our model.

Since our simulation experiments were carried out without distinguishing between the gender of workers. we take the entitlement age applicable to female workers as the minimum retirement age.

Serdar Sayan

Table 2: Increases in Statutory Entitlement Ages

after the Pension Reform Act of 1999

Statutory Entitlement Years until Retirement

(Minimum Retirement) For a Worker

Years A2e A2ed 38

2000

38

0

2001

41

3

2002

43

5

2003

45

7

2004

47

9

2005

48

10

2006

49

I I

2007

50

12

2008

51

13

After

2009

58

20

III. The

Framework

for Investigation of Parametric

Reform

Alternatives

Reductions in the ratio of active workers to retirees covered signals financial trouble for PA

YG-based pension funds. Regardless of the reasons behind uch reductions, pension funds that face

sustained decreases in the number of workers relative to retirees will eventually begin to incur losse unless policy makers are willing to adjusJ amounts of contributions collected from workers

and benefits paid to retirees. Policy makers could use any combination of three policy parameters to

rehabilitate a system generating sizable lo se : The contribution rate (CR) that determines the share

of work income that the individuals will have to pay as old-age insurance premiums, the

replacement rate (RR) that detem1ines the share of average work time earnings to be replaced by

pension income, and the statutory entitlement age (A) that sets the minimum age when individuals

may begin collecting pen ions and other benefits.

When growing pension deficits make such a parametric pension reform necessary, the immediate problem before policy makers will be the identification of pension program parameter configurations needed to achieve a target balance over a specified period of time and under certain demographic and political constraints.9

8 _{In many countrie . employers are required to supplement employee contributions that are typically collected in the form of} eayroll taxes.

New parameter value chosen to achieve the targeted level of (intertemporal) pension balance must be compatible with

demographically realistic contribution and retirement period and not require radical reductions in the living standard of

Turning Pote111ia/ Retirees i1110 Workers: Pension Reform Act of'l999 and Beyond in Turkey

Formally the problem facing the policy makers is to legislate a set of program parameters,

[CR, RR, A], so as to achieve a targeted balance between total (inflation-adjusted) present discounted value (PDV) of future (projected) pension payments that need to be made to retirees and

contribution receipts to be collected from active workers. Since making a legislated increase in

entitlement age effective for everyone immediately would mean forcing thousands of workers who

made their retirement plans under the pre-refonn configuration of parameters to stay at work for

several more years than they originally intended to, however, a once-and-for-all increase in the

entitlement ages would not be fair, nor feasible politically. So, the configuration to be picked must

be of the form [CR, RR, A1] and allow for statutory entitlement age to be increased gradually over

time (Sayan and Kiraci, 2001a; Sayan and Turhan-Sayan, 2000). The rest of this section first lays

out the problem in mathematical terms and then, briefly describes the genetic algorithm approach

used here to identify alternative parametric reform strategies for SSK.

For a given period from

t

0 to

r

,

the difference, D, between total PDVs of future (projected)

pension payments and contribution receipts can formally be defined as

L

r 1

I

'

'

-

L

,

,r.

m

D

=

RR rw r - CR rw w

/=lo (1 +

a)1

'

(

u=A, 0,1 U.I a=ao U,/ O.I)

(l) where CR RR A1 r _.a., rwa.l a

Average contribution rate -including the supplementary contributions that the

employer are legally required to make (O<CR< I),

Average replacement rate showing the percentage of pension income to replace the

average wage/salary income earned prior to retirement (O<RR< I),

Statutory entitlement (minimum retirement) age at time t ( A1

<

mwa

,

where mwa is

maximum working age),

Discount rate,

Projected number of contributing workers at the age of a at time t,

Average real wage income projected to be earned by active workers at the age of a at

time t -adjusted for the earnings cap or wage ceiling that determines the maximum

amount out of which contributions are collected,

Projected number of pensioner who are aged a at time t

Average real work time income of pensioners aged a at time t,

Age index running from the beginning of working life, a0 , to le, life expectancy in

years (le>mwa),

Time index running from the initial period, t0, to

r

,

the end of model horizon, and

t'=t-t0 .

By (1), total pension payments to be made by the pension fund at any year tare given by the

product of r.0 .1 the number of retirees distinguished by their ages, with the applicable pension for

the corresponding age group. The applicable pensions at time t. in tum, are calculated through an

indexation scheme requiring that each retiree be paid a certain proportion, RR, of the average of

configurations might further be re tricted by additional constraints and different priorities assigned to variou outcome of refonn. Within this context. the riming of reform relative to election cycle and the implied distribution of the burden

between the working young and retired elderly may be important con iderations for policy makers in deciding whether or not a proposed configuration should be legi lated (Sayan and Kiraci, 200 I a; Sayan, 200 I). Yet another consideration in picking the parameter configuration to be introduced may be given to the intergenerational aspect of the distribution of reform' burden on presently living and future generations (see. for example. Boll, et.al.. 1994).

Serdar Saya11

real wages earned prior to retirement. Each active worker aged a at time t on the other hand, is paid

1'\110 ., and a fraction, CR, of this income adjusted for the wage ceiling is collected as contributions.

The pension system is said to have a surplus (deficit) when the total PDY of contribution

receipts exceeds that of pension payments or when D < 0 (D > 0). Given that publicly managed

pension systems do not typically aim to run surpluses,10 _{we may let DT}

-

o

_{represent the target value}

of pen ion deficit that the policy makers are willing to fund out of the Treasury. Then, D -DT may

be taken as the objective function in a constrained minimization problem requiring to find CR and

RR values and A,, minimum retirement ages to be introduced at different points in time, so as to

keep Das close to DT as possible under exogenously given, projected values of real wages, retiree

and active worker populations and other relevant constraints.

When considering this problem within the context of parametric reform alternatives for SSK

over the 2000-2060 period, DT was set equal to zero since the losses generated by the SSK (and by

the ocial security system at large) had already reached alarming proportions prior to the reform act

of 1999. The cho en target level of pension deficit es entially keeps the existing debt of the SSK

from growing any further (or maintains the real debt stock of the SSK at its 2000 level) and this

treatment is consistent with the rationale of the reform of 1999. With DT=O, minimizing D -DT is

equivalent to minimizing D and since the sy tern should not be allowed to generate large surpluses,

I

D

(

should be brought as close to zero as possible implying that the problem can formally be defined

as I

Mtnmu. . . ze

I

D

I

=

12060

L,

- - -

I

.

(RR

L,

1,1101r01

-

CR

L,

rwu,wu,)

/e - ""'"

I

A,.CR.RR 1=2000 (I +a)' o=A., . . a=Uo . .

!

wa., for ao:;; a< A1, Vr

subject to Pa,t

=

11·0 ., +r0 ,1 for A, :;; a:;; mwa, Vr

r0 ,, for mwa <a:;; le, Vt

(2)

O<CRL :;;cR:;;CRH <I , O<RRL :c;RR:;;RRH <1

where CR L (RR L) and CRH ( RR H) respectively denote the low and high values within the range of which policy makers would like to adjust the contribution rate (replacement rate), and

pa

1

'

represents the projected year t population of the people in age group a (

a

;:::

ao )

and covered by

the SSK. Hence the equality constraint in (2) states that anyone who is between the minimum retirement age A, and maxi.mum working age,

m

w

a

,

at time t must either be a worker or a retiree, whereas any member of this group who, at time t, is younger (older) than A, (mwa) will necessarily be a worker (retiree).

pa

t is determined solely by the age composition of society and workforce

participation behavior that are exogenously given by the expected nature of demographic transition

and the labor market condition , and hence, is beyond the control of policy makers and SSK

authorities (Sayan and Kiraci, 2001 b).

Given the way minimum retirement age, A,, defines the limits for relevant age brackets in the

equality constraint, however, any increase in A, introduced as part of a pension refom1 would

increase the future size of working population at the expense of the retiree population. So, even

10 Because many governments view pension systems a a channel to make income transfer to working and/or retiree populations. they would often be willing to allow public pension system to run ""reasonable'· deficits -with the definition of rea onable varying across governments and macroeconomic condition of the country in question.

11 With the ab olute value operator, the global minimum for

J

oj

would be zero. Hence, the formulation of the objecti,e

function rule out the possibility that the SSK would run a significant surplus while providing pension coverage over th

Turning Po1e111ial Re1irees i1110 Workers:

Pension Rejom1 Ac1 of /999 and Beyond in Turkey

though the policy makers/pension authorities can not control

p

a,t itself, they may affect its

composition by increasing minimum retirement age in such a way to reduce the number of retirees to collect benefits -while simultaneously increasing the number of workers to pay contributions.

So, during the search for the [CR, RR, A,] configuration which will minimize the total pension

deficit

I

D

I

defined in the objective function in (2), any increase counterfactually introduced to the

minimum retirement age requires that initial demographic projections be manipulated in such a way

to transfer retirees younger than the value of A, under consideration to working population so as to

satisfy the constraint in (2). To do that, the time horizon for the computational exercises was chosen

to be the 61-year period between 2000 and 2060. Projections on active worker/retiree populations

by ages (aggregated over genders) were then used to construct a (mwa-a0- J)x61 matrix for the working population, and a (le-A0+J)x61 matrix for the retiree population over 2000-2060, where Ao is the initial value of minimum retirement age. Based on data on SSK membership in ILO ( 1996),

maximum working age, mwa, was taken to be 75 and the earliest working age, a0, to be 15.

Likewise, following ILO (1996) the life expectancy, le, was set equal to 80+ allowing retirees to

continue to collect benefits by the age of 80 or more. Since ILO ( 1996) projections on working and retiree populations were based on the maintenance of the actual (pre-1999) minimum retirement

age, Ao was taken to be 38. These values yielded two demographic projection matrices for

2000-2060: WAT=[w0 .,]_61x61 and RA T=[ra.,fo,₆₁ showing the composition of the active (contributing)

worker and the retiree population by age, respectively. Other data available in ILO ( 1996) included

real wage and pension incomes of actively working and retired members of the SSK by ages, all in

terms of Turkish Liras at constant 1995 prices. Whenever nece sary, the available erie were extended into the year 2060 by taking into account the likely growth in economywide labor

productivities as well as productivity differentials due to seniority, and by using the average

replacement rates given in the same source. The resulting work and pension income matrices, WPA Y=[,-wa,,]61• 61 and RPA Y=[ rw

a.1

]

43, 61 , were then constructed and used together with the corre ponding demographic projection matrices, WAT=[w0 .,]₆₁• 61 and RAT=[r0 .1]43x61. to calculate

the initial (pre-reform) value of

I

D

I

by using the average contribution and replacement rates, and minimum retirement age prior to the reform of 1999. That is, by taking CR=0.20, RR=0.65 and Ao

=38, and assuming that 8=0.05.

To see the effects of increases in the minimum retirement age, these initial projection matrices

were regenerated for each value of A, accordingly with the constraint in (2). Once A, is

counterfactually increased from 38 to 39, for example, people initially projected to retire at the age of 38 any time during the 2000-2060 period would no longer remain eligible for retirement. To

represent such a policy switch, the first row of the initial RA T=[r0

.,

fo

,

61 matrix containing the projected retiree population before reform was replaced by zeros in all columns and the original row was added to the 24'h row of the initial matrix ( corresponding to age 39) for projected working

population, WAT=[ w0 .1]61 , 61, to modify the projected numbers of workers at the age of 38 according

to the newly introduced entitlement age. To be able to con ider minimum retirement age increases

within a reasonably large range, A, was allowed to take 32 different values between 38 and 69.

Keeping in mind the February 2001 ruling of the Constitutional Court, however, the reform

scenario considered were always chosen to require a smoother transition to higher retirement ages over the ten-year period for gradual increases in this age.

The search for optimal pension program parameters was carried out through the genetic algoritlm1-based optimization program written in MATLAB as described in Sayan and Turhan-Sayan (2000). The genetic algorithm searches for the minimum value of

I

D

I

for alternative [CR, RR,

A,] configurations by considering different feasible paths for entitlement age increases along with different CR-RR pairs and having the value of I

D

I

computed for each parameter configuration considered. The results reported below are chosen from among a larger set.

Serdar Sayan

IV.

R

es

ul

ts

This section reports results from simulation experiments, obtained under alternative range

imposed on the values that parameters of interest are allowed to take. In the ca e of CR and RR, a

single range i defined for the entire model horizon whereas the mininmm retirement age, A, was

a urned to be increased at different points in time between 2000 and 2060. This treatment is

generally compatible with pension reform practices in different countrie (Kohl and O'Brien,

1998) as well as the recent parametric reform introduced to the SSK and other branches of

Turkish pension system.

The simulation experiments are carried out under two different cenarios. In both scenarios average contribution rate is allowed to change within the range of CRL = 0.10 and CRH = 0.30,

whereas the range allowed for average replacement rate is RRL

=

0.50 and RRH

=

0.75. The

scenario are distinguished by the differing ranges allowed for changes in the entitlement age

during 2000-2010. In other words, the scenarios maintain the transition period but instead of

fixing the entitlement ages for each year within this period, they set age ranges whose upper limits

are smaller than or equal to the ages introduced by the parametric reform act of 1999 (see Table

2) as required by the Constitutional Court ruling. Given these ranges Scenario I allows for the entitlement age to be 38 or 39 during the 2000-2003 period, and to take a value between 39 and 42 over the 2004-2010 period. Scenario 2, on the other hand, defines a single range for the entire 2000-20 IO period requiring the entitlement age to take a value between 38 and 41. The period

from the end of transition period to the end of model horizon (2060) is divided into two

sub-periods to account for the tendency of life expectancies to increase. While the age ranges for

these sub-periods (2011-2023 and 2024-2060) are common across scenarios, the entitlement ages

are required to be higher in the last sub-period than the previous one. The beginning of this last

sub-period is taken as 2024 so as not to require anyone who is in workforce and covered by the SSK at the time of legislation to po tpone collection of retirement benefits any longer than

required by the actual reform act. Given that the reform act already forces the youngest workers

who began to work at the age of 15 in 2000 to postpone benefit collection until the age of 58

(instead of 38) or by the year 2043 (instead of 2023), the scenarios here require that retirement

ages exceeding 58 to be applicable only to those who are to begin working after the new law

become effective.

Table 3 presents two sets of results obtained from the genetic algorithm-based search for

optimal configurations under each simulation scenario, along with a description of the case of no

reform and actual reform of 1999. The first column under each scenario shows a sub-optimal

configuration (i.e. one at which

I

D

I

has not attained its minimum value) obtained during genetic

algorithm search, whereas the second column shows the optimal configuration.

The numbers in the second and third columns of Table 3 reveal that the pension reform act of

1999 is expected to achieve considerable success in reducing the real POV of the pension deficit

likely to be incurred by the SSK over the 2000-2060 period from 38.013 to 7,301 billion TLs.

While this i an impressive outcome to be achieved just by raising the minimum retirement age

(gradually to 51 first, and to 58 thereafter), the optimal olutions reported under fifth and seventh

column (entitled 1.2 and 2.2) show that it would have been possible to follow a smoother transition

for entitlement age increases and yet, get

I

D

I

practically equal to zero, if the values of CR and RR

Serdar Sayan

The plots in Figure 2 are generated as follows. A one-year increase introduced to the

entitlement age at period, say, I will set the relevant age for that period at A ,, making it impossible

for people previously projected to retire at the age of (A, -1) in year I (i.e., prior to the increase).

These people would then be forced to postpone retirement for a year and retire at the age of A, in

year f. If there are N uch individuals, the number of people taking retirement in year I will be

reduced by N and the number of contributing workers will be raised by the same number increasing

the active worker/retiree ratio in year f.

Figure 2: Projected Trends of Number of Active Workers per Retiree

under Alternative Paths for Entitlement Ages

Q - --- - -- - : -- --- ---: -(/) ' ' _' 8 ----r-----~- -' ' ' W 7 --------L-------- L--~

:

:

~

0 1 1 ' ' ' ' ' ' - - - - --- -_, r-- -------r --' -~

:

U 3 - - ---- -~--------L- -<l: 2 ' ' ' '

----

-

,---

.--

--

----

, --' ' _' ' _' ' ' ---., --- ---'T- - - - ----, . - --- -' ' ' Reform of 1999 ' ' ' ' '

---

.

---

--

--

-

,---

.---... ~~~~~,,:..d

'

'

'

---

-

~

·

_

-

-·::

_

·

_:-

_

'

_;,·_

·

_

-

---

-

-

-

-

-

-

-

-

'

~

-

-

-

---- _,i,~ .-.-_

--: :

-_L~o-~orm

o ~ - - ~ -- - ~ - - -~ - - - ~ - - - ~ - - ~

2000 2010 2020 2030 2040 2050 2060 Ye a rs

Within thi framework, the reform of 1999 will immediately cause an increase in the

worker-to-retiree ratio and this critical ratio will continue to rise until the entitlement age target of 58 is

reached following gradual increases in A,. As· shown in Figure 2, the ratio will start declining after

2009 again as population aging continues. In other words, the transition scheme introduced by the

pension reform act of 1999 jump starts a quick rise in worker-to-retiree ratios of SSK, thereby

inducing a rapid reduction in SSK's yearly deficits, even though the overall PDV of the 2000-2060

deficit remains high at 7 .5 trillions of 1995 TLs. The moother transition schemes that are

considered here, on the other hand, are much lower to raise the worker/retiree ratios and hence,

the yearly deficits of SSK in the aftermath of reform. These chemes manage to reduce the overall

PDV of the 2000-2060 deficit to practically zero by causing a significant increase in the

worker-to-retiree ratio after 2024, i.e., by going ea y on currently working generations but requiring higher

entitlement ages for generations that are yet to start working.

While different transition schemes for gradual increases in entitlement ages have different

implications for patterns of labor supply a well a the behavior of macroeconomic balances over

time, discussing the macroeconomic implication is beyond the scope of this paper. A for the

labor market implications, the following three-dimensional figures showing the age composition of

SSK worker over time are illustrative of the changes resulting from the different strategies for

Turning Potential Retirees i1110 Workers:

Pension Rc!form Acr of I 999 and Beyond in Turkey

Figure 3a. Number of Active Workers by Age in 2000-2060: Pre-reform

Serdar Sayan

Figure 3c: Number of Active Workers by Age in 2000-2060:

Scenario 1 (Optimal Configuration in 1.2)

i

ii ..

e

a, I" J

i

1

i

'I!

i

I

A comparison of Figure 3a through 3c reveals that increasing entitlement ages not only

increase labor supply (in the formal sector) but also shifts up the age corresponding to the highest

number of workers gradually over time. 1n fact, this is what increases the worker-to-retiree ratios

facing SSK, which, in turn, reduces pension deficits by extending the contribution period and shortening the benefit collection period. While the resulting shift in the average age of workers is al o certain to have productivity implications the analytical tools developed here are not suitable for examining the productivity implications. It suffice to note here that investigating the

productivity and other consequences (such as the likely changes in saving and consumption

behavior of individuals that stay in the workforce for longer periods of time) of deficit reduction

strategies involving different paths for minimum retirement age increases would be very worthwhile

to undertake in the future.

V.

Conclusion

s

Turkey's PA YG pension system run by the state experienced erious financial difficulties during the

past decade, forcing policy maker to legislate a pension reform act in 1999. Since past reduction

in statutory entitlement ages for pen ion benefits and the resulting decline in the ratio of

contributing workers to retirees were among the primary reason behind the los es generated by the system, there was an evident need to extend average contribution periods. This made the entitlement age a critical parameter for pension reform. Accordingly, the major change introduced through the

reform act of 1999 turned out to be a significant increase in entitlement age for female/male

workers from 38/43 to 58/60 to become effective after a transition period of eight years. Yet, in February 200 l, the Turkish Constitutional Court examining an appeal by the deputies from main

opposition party ruled that the bill's scheme for gradual increases in the entitlement age during the

transition period violated the constitutional rights of workers in certain age group , and hence, the equality principle. The Constitutional Court gave the government six months to make arrangement

Serdar Saya11

R

e

fer

e

nc

es

Akmaz M. (1999), Sy temic Analy i of and Policy Generation for the SSK Pension Scheme in Turkey, Unpublished Thesis, Middle East Technical University, Ankara.

Ayas, E. (1998) 'Turkiye'de Sosyal Guvenlik Sisteminin Sorunlari ve Cozum Onerileri," lsletme ve Finans Vol. 13(150), 41-50.

Boll. S., B. Raffelhuschen and J. Walliser (1994), "Social Security and Intergenerational Redistribution: A Generational Accounting Perspective," Public Choice, Vol. 81 79-100. Gillion, C. and M. Cichon (1996), "Pen ion Reform in Turkey," Hazine Dergisi (Special Issue on

Social Security and Health insurance Reform), 40-56.

ILO ( 1996), Social Security and Health Insurance Reform Project: Final Report, Geneva: ILO.

Kohl, R. and P. O'Brien ( 1998), "The Macroeconomics of Ageing, Pensions and Savings: A Survey", Economics Department Working Paper 200, Paris: OECD.

Sayan, S. (2001 ), "Political Economy of Pension Reform: The Lessons that ME A Countries can

Derive from The Pension Reform Experience of Turkey," Paper presented at the Second Mediterranean Social and Political Research Meeting, Mediterranean Programme, Robert

Schuman Centre for Advanced Studies European University Institute Florence, March 21

-25, 2001.

Sayan, S. (1999), "Turkiye'deki Yaslilik (Emeklilik) Sigorta Prim Kar;:aklarinin Toplumsal ve

Bireysel Maliyetleri [Social and Individual Costs of Social Security Tax Evasion in Turkey],"

ASO Medya September, 40-54.

Sayan, S. and T. Kenc (1999), "Long-term Consequences of Rehabilitating a Financially Troubled

Pension System: An Overlapping Generations, General Equilibrium Analysis for Turkey,"

Economic Research Forum Working Paper o. 9914, Cairo.

Sayan, S., and A. Kiraci (2001a), "Parametric Pension Reform with Higher Retirement Ages: A Computational Investigation of Alternatives for a Pay-As-You-Go Based Pension System,"

Journal of Economic Dynamics and Control, Vol. 25(6-7), 951-966.

Sayan, S. and A. Kiraci (2001 b). "Identification of Parametric Policy Options for Rehabilitating a

Pay-as-you-go Based Pension System: An Optimisation Analysis for Turkey, Applied

Economics Letters, Vol. 8(2), 89-93.

Sayan, S., and G. Turhan-Sayan (2000), "Social Security Reform through Increases in the Statutory Entitlement Ages for Old Age insurance: Optimal Magnitude and Timing," Paper presented at

the Society for Computational Economics, 6th International Conference on Computing in

Economics and Finance, Barcelona, July (?-8. 2000.

Sirageldin, I. and E. Al-Rahmani ( 1999), "Elderly Support in Oil Economies: How Sustainable in the 21st Century, with Illustrations from Kuwait," Economic Research Forum Working Paper

o. 9919, Cairo.

Teksoz, A.T. and S. Sayan (200 I), "Simulation of Benefits and Risks after the Planned Privati ation

of Pension System in Turkey: I The Expected Boost To Financial Markets Likely?"