Turkey: Selected Issues


Turkey: Selected Issues

New Private Pension Automatic Enrollment: A Missed Opportunity1

A. Introduction

1. The Turkish authorities have introduced a new automatic enrollment pension system also to promote private savings. The authorities’ rationale behind the new automatic enrollment system is to raise the domestic savings rate and to encourage employees to save funds for their post-retirement life so that they can maintain the welfare level they enjoyed during their employment period.

2. This select issue paper provides policy advice to promote private savings and assesses the macroeconomic impact of the recently enacted provisions. The reform has many advantages with respect to the current voluntary private pension industry. It has also several design weaknesses that limit the ability of this reform to promote private savings. This select issue paper provides policy advice on industrial organization and the aforementioned design weaknesses to help authorities achieve their macro objective and it assesses the macroeconomic impact of the new provisions.

B. Industrial Organization Considerations

3. The new automatic enrollment pension system has many interesting features. Four main features relevant to the assessment in this section include:2,3

  • Salaried workers younger than 45 years of age are automatically enrolled in pension plans4 chosen by their employers. The system is voluntary in nature as individuals can opt out but experience shows that consumers are inert and tend to stay in the system, treating the voluntary contribution as an addition to their personal income tax. Exploiting such inertia is central to the Government strategy of raising private savings. The opt out rate is likely to be even lower than in the current voluntary system (8.3 percent of stock per year).

  • Collection of contributions, record keeping, asset management and custodianship remain the responsibility of existing voluntary pension firms and their service providers. The current voluntary pension system is based on vertically integrated pension fund management companies (now 19) which are dedicated subsidiaries of financial services companies. These collect contributions and will use their sales force to attract employers into the system and retain individual workers. The Pension Monitoring Center (PMC), governed by representatives of the companies and the Government, acts as the front-line information and monitoring hub of these companies, manages the registration and data on individual accounts and manages the 25 percent contribution subsidy provided by the Government to encourage pension saving. Takas bank, an offshoot of the stock exchange, provides centralized record keeping and custodianship of the assets—but with the account management duplicating that of the Pension Management Companies—reflecting the concern at establishment in 2003 to have an ultra-safe triple wall of protection for the assets of pension fund members.

  • The PMC will reconcile information ad contribution flows. The contribution base is the same as the one for contributions to the Social Security Institution (SSI). The PMC will cross check contribution flows to the individual pension funds with the information extracted from the data transmitted to the SSI.

  • Pension firms can only charge a capped fee levied on assets under management. The single fee facilitates performance comparison to help employers chose the firms/plans with the highest net rate of return. The cap is set to 0.85 percent of assets under management.

4. Many implementation details are yet to be defined. In particular, the Undersecretariat of Treasury needs to prepare bylaws on things such as: (i) the enrollment of an employee to a pension plan by way of their employer; (ii) the criteria to be taken into consideration by the employer when choosing the company and pension plan; (iii) the funds in which contributions will be invested; the contract to be entered into by the employer with the company; the right to withdrawal; (iv) enrollment of an employee to a pension plan by the employer in cases where the employee changes their workplace; transfer of accumulated savings in cases where the employee changes their workplace; (v) payment to the related pension plan upon the request of the employee whose employment relationship has been terminated; pausing contribution payments; (vi) abandonment of the system; and (vii) payment of Government contributions.

5. The new design has many advantages when compared with the existing voluntary pension industry.

  • The reform exploits consumers’ inertia. This is at the frontier of private pension system design and it is a central consideration for optimal industrial organization design of these quasi-markets.5

  • Employers act as a procurement board on behalf of workers. Choice is exercised by employers. This reduces competition in the market for individuals that has proved to be wasteful in other countries. Typically, competition in the market encourages large marketing expenses that act as a barrier to entry, end up being paid for by consumers through higher fees, and do not result in consumer choosing pension firms with higher net rates of return (see further for a more detailed discussion on the heuristic behavior of pension consumers).

  • The automatic enrollment applies to the stock of salaried workers. There is currently a lively debate of whether the stock design is superior to the flow design. With the stock design, the full segment of participants (the inert customers) is allowed to be served by providers procured by employers. In addition, target participants can spend their whole working career in the procured segment as they will be served by a sequence of providers selected through periodic procurement auctions. However, it requires periodic auctions for service providers to keep high the quality of services. With the flow design, only a fraction of inert participants is served by providers in the procured segment (typically, the flow of new entrants in the labor force) and periodic auctions are not necessary as the automatic allocation rule would contain clauses to encourage quality (Mexico, for instance). Generally, the stock design presents attractive characteristics, like targeting all inert participants and protecting them from dynamic predatory pricing schedules. More concretely, it maximizes the increase in private savings in the short term.

  • Only one type of fee can be charged and it is capped. This is a huge improvement from the average 2.5 percent rates charged by private pension funds in the voluntary system. Everything else equal, this is likely to contribute an additional 20 percent cash balance at time of retirement for the average career worker.

6. It has also three key design weaknesses that have contributed to unraveling pension industries in other countries. These are: (i) the procurement component is incomplete; (ii) providers can charge a fee based on assets under management, and (iii) the fee rate is capped without a link to the structure of costs of pension service providers. Experience shows that these same industry design weaknesses contributed in other countries to: (i) wasteful competition in the market, (ii) extraordinary profits for, and rent extraction by, service providers, and (iii) regulatory capture when legislated fee caps need to be changed. Forced by strong social discontent stemming from consequences of these design weakness, many countries have indeed reverted the reforms that gave birth to these industries. We discuss the nature of these design weaknesses in the next three sections.

The Problem with Incomplete Procurement

7. Limiting choice by employers may not reduce marketing incentives. It is unclear that employers are abler than individual consumers to choose the best pension firm for their workers. They, like individuals, tend in practice to adopt simple rules of thumb to solve the needed intertemporal optimization problem and this leads to systematic biases. Calderón et al. (2008) analyze the choice of pension funds providers by Mexican contributors and the role played by the sales force of each fund in this selection process. Like many other studies, this paper finds that the number of switches is highly correlated with the number of sales agents hired by the receiving pension fund manager and its marketing expenditure In their sample, nearly 40 percent of switches were made to pension funds with lower historical gross returns and higher fees (text picture).

8. Pension services should be auctioned.

9. Collection of contributions is not centralized but implemented by existing pension firms. According to interview with authorities, the existing pension firms will collect contributions and use their sales force to maintain individuals in the system. This has proved in other countries very detrimental. It is extremely costly to collect contributions. Sales force expenditures typically amount to 30 percent of total pension firms’ expenditure. Because of these high costs and revenues being caped, pension firms will need to subsidize this service through another arm of the financial group and recuperate such costs by selling complementary services to participants. In other words, the current reform creates a captive source of revenues for existing pension firms or related financial groups. In addition, as it has happened in other countries, consumer protection is severely weakened as the presence of a sales force creates the opportunity for mis-selling of financial products.

10. The SSI is in principle best placed to collect contributions. Having the SSI collect contributions would maximize economies with minimal marginal costs. The SSI already collects SSI premia through the payroll system on the same contribution base and it enforces compliance with the labor code through its labor inspectors. Additional costs would be related to designing the software that diverts contributions and information to the pension funds chosen by the employer but these are expected to be minimal. Since collection of contributions is the pension services with the highest fixed costs, it is typically provided as a public good by the government through the local social security agency or the tax authorities. The choice between the two depends on relative efficiency and governance considerations.

11. The reform does not establish a public procurement board to auction all other pension services. Other pension services should be auctioned to avoid competition in the market. In addition to collection of contributions, all other services in the pension industry (record keeping, administration, custodianship, and in part asset management) are characterized by high fixed costs. This means that centralization of services exploits economies of scale and reduces average costs for customers. While it makes sense that collection of contributions be run by the SSI as a public good, other services should be auctioned. However, the authorities plan to use current pension firms for administration and asset management; the PMC for record keeping and reconciliation of money and information flows, and Takas Bank for custodianship. Asset managed is planned to be auctioned but only to Turkish incorporated managers. it is unclear what rationale was followed to avoid a fully transparent public competitive procurement process (auctions) for most of these services that include qualified foreign providers from reputable jurisdictions. In other words, the current reform appears to create a captive source of additional revenues to the current banking and private pension fund industry disregarding efficiency considerations.

12. A public procurement board would promote competition for the market. The job of the procurement board is to assign participants to pension firms using an exogenous, public, and transparent rule; usually a contest based on prices, yields or other variables that are related to performance efficiency. The essence of procurement is that consumers are not given the opportunity to choose the board, otherwise a quasi-market is created and consumer inertia regains significance. Examples of countries that have adopted this framework are Singapore, Bolivia, and the United States for the Thrift Savings Plan. The merits of pure procurement arise directly from its primary objective, which is to deal in a radical manner with consumer inertia. When one demand block is granted to each of the firms winning the contest, the incentive of providers to spend in marketing to attract clients is removed. By establishing competition for the market, rather than in the market, consumer protection is ensured as rent extraction activities (mis-selling, cream skimming, et cetera) are eliminated.

The Problem with Fee Rates Based on Asset Management

13. Uniform fee rates are considered more transparent and equitable but can generate super natural profits. Uniform fee rates allow participants to quickly compare firms in terms of prices. However, they reduce efficiency in the supply. This, in turn, is a highly valuable policy objective as it helps align prices paid by consumers (administrative fees) to the structure of costs of pension firms. In other words, it contributes to reducing super natural profits, typical of these quasi markets. Uniform fee rates are also thought to be more equitable as they do not allow pension firms to price discriminate. But this happens only in proportional terms: the proportional fee rate is the same for all customers. Higher net worth customers pay more in absolute terms for the same services than low net worth customers. This inefficiency worsens over time as the fee base grows in a convex manner while pension service costs are essentially fixed. Over time, pension firms can earn extraordinary profits as pension consumers are inert (have low elasticity to prices) unless fees are capped and continually lowered. These amount to a redistribution from consumers to pension firms.6

14. Uniform fee rates applied heterogeneous bases promote wasteful competition in the market. When uniform rates are applied to heterogeneous bases, different participants represent a different rent for pension firms. It is therefore economically attractive for the pension firm to invest in marketing whenever the marginal rent is larger than the marginal search/contact investment needed to attract that customer. Uniform rates applied to heterogeneous bases encourage firms to excessively invest in marketing and provide an explanation for the occurrence of marketing wars. Marketing expenses typically add up to 30 percent of fees charged to participants. These charges would add value if performance across firms were highly heterogeneous and individuals could choose the pension firms with the highest net rate of return. However, since individuals are inert and unable to choose (see next), uniform fees applied to heterogeneous bases only promote wasteful competition in the market.

The Problem with Caps on Fees Not Linked to Supply Costs

15. Capping fee rates as in Turkey reduces supply inefficiencies. From an economic perspective, the standard justification for price regulation as adopted in Turkey is that it limits the price distortions generated by low demand elasticity. Low demand elasticity renders the clientele of pension firms captive: i.e., unable to “vote with its feet” and to leave costly pension firms for more affordable pension firms. With price regulation, the ability of firms to charge above average costs is largely bounded by the credible threat of political interference (through the introduction of price caps) when mark-ups become intolerably high.

16. But caps need to be adjusted frequently increasing the likelihood of regulatory capture. Caps are not linked to the cost structure of pension firms and can easily become obsolete when the fee base grows allowing firms to charge well above average costs. In the absence of a formal process to set caps that reflect the actual production costs of firms, pension firms rely on lobbying and related practices to influence policy making. In this case, regulatory capture by well-connected market players becomes likely. At the same time, pension firms are exposed to excessive regulatory risk if a populist administration decides to arbitrarily lower the caps.

Other Potential Design Weaknesses

17. Interviews with authorities suggest other potential design weaknesses. These are considered “potential” as they had not been regulated at the time of writing.

  • Auctions for asset management would be limited to Turkish incorporated entities. Current pension funds charge fees on asset management around 250bps. This suggests that they are not efficient enough to be able to bid for a 70–100bps capped fee or that lack of competition exposes consumers in the voluntary system to predatory pricing. Allowing participation to auctions of qualified foreign asset managers from reputable jurisdictions without a footprint in Turkey (i.e., that would be unable to undercut the bid and subsidize it with ancillary activities) would generate competition to existing asset managers to reduce costs.

  • The investment rules in the new system would be decided by the Deputy Prime Minister in charge of the Undersecretariat of the Treasury. This would be done on the basis of advice of an investment advisory committee. At the time of writing, it is unclear whether the yet to be designed provisions will apply to either the investment options that procured or to the investment rules that asset managers need to follow to provide the authorized investment options, or both. Irrespectively, it is unclear why a political appointee should be in charge of approving either of these. This could leave the option open for undue political influence on the asset allocation of pension funds for direct lending and or equity investment or simply to facilitate issuance of domestic debt (the responsibility of the Treasury). The presence of the envisaged advisory committee does not change the assessment as its governance is still undefined and its opinions are not likely to be published, as customary in most policy decision making in Turkey. Finally, the fact that the insurance and pension supervisor is hosted within the Treasury does not justify that a technical matter such as the one discussed in the paragraph be left to the decision of a political appointee.

  • The investment rules will include government debt. If the objective is to pre-fund pension liabilities, investing assets in government debt does not achieve that objective. Many countries have used private pension funds as source of cheap financing of budget deficits and in this way reduced fiscal discipline. Investment limits in government debt should be kept at a minimum to avoid generating such temptation/perception.

How to Strengthen the Pension I/O Weaknesses in Turkey

18. A hybrid I/O would eliminate most of the aforementioned weaknesses. A hybrid industrial organizational model separates demand in two segments: procurement and quasi-market:

  • In the procurement segment, the pension fund service providers are selected by a public board through periodic auctions. This applies to collection of contributions, administration and record keeping, and asset management. In this way competition for the market replaces wasteful competition in the market.

  • In the quasi-market segment, participants choose their own provider.7

In the most interesting types of hybrids, each participant is free to choose segment; i.e., there exists competition between organizational forms. Hybrids typically allocate undecided participants to the procured segment. Countries that have adopted hybrid industrial organization models include Mexico, Bulgaria, Chile, New Zealand (Kiwisaver), and Sweden (PPM). At the time of writing, the Turkish authorities had essentially chosen a hybrid system but with the important design weaknesses discussed in the previous section.

19. A hybrid I/O has many advantages relative to its extreme organizational forms. Firstly, it creates a performance benchmark for the board and benchmarks for the board’s suppliers via outside options. In addition, by replacing the choices of the undecided with a technically qualified public board that compares prices while controlling asset management quality, the welfare of the undecided increases. Furthermore, if the allocation to the procured segment is reasonably targeted to inert participants, the share of active participants in the quasi-market segment increases. This raises the demand elasticity faced by firms in the quasi-market segment, which could result in lower prices and less marketing expenditures in that market. Finally, the public notoriety, or “signaling effect”, of establishing a procured segment may further increase public awareness about price differences among participants further raising price elasticity. For example, in Mexico, the allocation of the undecided led the press to intensify information on prices every time a new group of participants was allocated.

20. The design of a well-functioning hybrid model needs to take into account several policy considerations among which the following five appear critical, especially for Turkey:

  • First, procurement needs to be supported by a strong governance framework. The government should allocate participants to the procured segment only when it is arguably in their best interest and protect them from marketing and wasteful competition. Given the volatility of returns on pensions in a defined contribution system, a government may face suits from individuals that find ex post that the government allocation materially reduced her pensions, even though this could not be predicted ex ante. Thus, the allocation to the procured segment needs to be transparent and well-reasoned. This requires a procurement board that assigns periodic licenses to service providers through auctions under a strong governance framework. While it makes sense for collection of contributions to be conducted by the SSI as a public good, administration and record keeping but more critically, custodianship and asset management need to be auctioned periodically. At the time of writing, the Turkish authorities had not established a procurement board but chose to discharge its responsibility by forcing employers to choose on behalf of workers.

  • Second, the targeting of participants should be made with objective rules. Involving the government in the business of finding the most convenient segment for an individual participant may lead to excessive interference, micro-management and legal liabilities. Possible objective rules include the simple requirement that the participant be undecided, in the sense that he or she does not to choose a pension fund administrator. Alternatively (or in addition), rules could be based on observable attributes of undecided participants (such as individuals with low assets) to be within a range that ensures that it is highly unlikely that they can do materially worse in the default segment, if they choose to remain there. At the time of writing, the Turkish authorities had not chosen the rules that employers need to follow to choose pension firms.

  • Third, participants that recently chose a pension firm should be exempted from the default allocation. Individuals should be free to leave the procured segment if they consider it worse than the choices available in the quasi-market. Also, to be efficient, a hybrid model must limit duplication of search costs and switching costs, and consequently, it would be desirable to exclude from the procured segment individuals that chose a provider. At the time of writing, the Turkish authorities had not chosen how individuals can elect alternative pension providers/services. Discussion with authorities suggest that this will be done by means of choosing alternative investment options within the procured segment, as it is the case for the US Thrifts Savings Plan. This appears a very reasonable and cost effective choice.

  • Fourth, heterogeneity in fee bases can destabilize the hybrid model. Heterogeneous fee bases can set the maximum welfare for participants at one of the extreme organizational forms (pure procurement or quasi-market) and the hybrid model would not be viable. If the average base in the undecided segment is much lower than in the quasi-market segment, providers in the procured segment would need a higher fee rate to collect the same fee income per period as providers to the other segment. If the ratio between the average fee base (assets under management) of participants in the procured segment and the average base of participants in the quasi-market segment is low enough, providers do not participate in auctions in the absence of a large fiscal subsidy. I.e., the procured segment is eliminated.8 By contrast, when flat fees are dominant as in New Zealand, there is no inequality in fee bases: the base is one unit of service per person per period. In other words, encouraging flat fees reduces the risk of an artificial elimination of the hybrid model and should be an additional aspect to consider when selecting fee bases. At the time of writing, the Turkish authorities had chosen to allow providers to charge a fee based on asset management which Staff would discourage.

  • Finally, there is the issue of whether a stock model is preferable to a flow model for Turkey. As previously mentioned, the stock design that allocates all undecided individuals to the procured segment has the advantage of targeting all inert participants and protecting them from predatory pricing schedules and it maximizes the increase in private savings in the short run. However, it requires periodic auctions to ensure that incumbents maintain a high quality of service or be replaced by new entrants by the procurement board. At the time of writing, the Turkish authorities had chosen the stock model but still needed to design and introduce periodic auctions for services like asset management, custodianship, administration and record keeping.

C. Macroeconomic Impact of the Reform

21. We assess the impact of the auto enrollment reform for various scenarios. We want to answer the following question: “How effective is the new automatic enrollment system in reducing external imbalances”? In order to do so, we need to estimate the potentially covered population, their wage mass, make serval assumptions to project cash balances, and hence aggregate savings, going forward. We do so for different scenarios reflecting likelihood of outcomes and desired policies. We finally assess the impact of the reform on various macroeconomic aggregates using the G20 core module of the Flexible System of Global Models (FSGM) developed by the research department of the IMF.

Estimating the Covered Population and Wages

22. We use the labor force survey of 2015 to estimate gross individual earnings to estimate the initial conditions of our projections. The 2015 labor force survey contains salaried individuals for the private and public sector. A total of 14.1 million salaried workers reported a total of TL26 billion monthly net earnings for an average monthly net earnings of TL1,838 (Table 01, 2015 Net).

23. We gross up net wages also to take account of the 30 percent minimum wage increase of 2016. A total of 14.1 million salaried workers9 reported a total of TL37.5 billion monthly gross earnings for an average monthly gross earnings of TL2,757 (Table 01, 2016 Gross). We derive the distribution with the following 3 steps:

  • We move from 2015 net earnings to 2015 gross earnings by observing that:

    where WG > 0 is gross wage,10 WN > 0 is the net wage reported in the labor force survey, s = 0.15 is the sum of the SSI and UIF contribution rates, SSC > 0 is the sum of the SSI and UIF contributions, SSI = TL1,273.5 is the 2015 minimum SSI contribution wage, SSI¯=TL8,227.75 the maximum 2015 SSI contribution wage (calculated as 6.5 times the minimum wage),11 OT > 0 is the rest of personal income tax (net of the minimum living allowance) and other contributions (stamp duty, etc.), and x = 0.093 is the proportional factor of OT in WG (basically, the effective average PIT rate on gross wages).

  • We estimate the effective average PIT rate on gross wages in two ways and take the maximum of the two estimates. In Turkey, the SSI premium is 14 percent of gross wages and the UIF premium is 1 percent of gross wages. Wages net of SSI and UIF premiums are used as base for the PIT. PIT averages 17 percent of such base or 14 percent of gross wages. A minimum living allowance of about 45 percent of the PIT is deducted from the PIT. This allowance amounts to 8 percent of the PIT base or 7 percent of gross wages. A small stamp tax of 0.0795 percent is also levied on gross wages. Hence, the effective PIT rate is estimated as x = 0.14 + 0.000795 − .07 ≅ 0.08. However, these calculations underestimate the effective PIT rate as they are based on SSI population where earnings are notoriously underreported. An alternative way to estimate effective PIT is to use macro data. In 2014, 17 million individuals were categorized as wage employed with annual average gross earnings of TL27,083. PIT revenues amounted to TL73,899 million, of which 60 percent are from wage income. This implies that PIT revenues from wage income was 9.3 percent of gross wages. Hence x = 0.093.

    Hence, gross earnings are given by:
  • We increase the whole 2015 gross earnings distribution by 8.3 percent to simulate the 2016 gross earnings distribution before the 30 percent minimum wage increase. The annual growth rate the average nominal gross wage increase we use in our projections, equal to average headline inflation of 6.3 percent and a productivity growth of 1.9 percent.

  • We simulate the impact of the 30 percent minimum wage increase that occurred in 2016. In the 2015 labor force survey there are 5.8 million individuals with estimated 2015 gross earnings in the first bullet below the 2016 minimum wage. Of these, 3.8 million are between the 2015 and the 2016 minimum wages and 2 million are below the 2015 minimum wage. We simulate the impact of the 30 percent minimum wage increase that occurred in 2016 on the whole wage distribution by bringing the 2016 estimated gross earnings of individuals between the 2015 and 2016 minimum wages to the 2016 minimum wage.12

Scenario Design

24. We design 3 main scenarios: The first scenario attempts to replicate the newly enacted provisions in terms of covered population and wages. The second scenario relaxes the age participation constraint but maintains the restriction on covered wages. In this way, an additional 2 million salaried individuals are captured in the new system. The third scenario relaxes also the covered wage constraint. In this way, individuals are allowed to contribute on full remuneration. By eliminating the constraint on minim wage, 2 million additional workers are allowed to participate. This population is important as it mimics the presence of individuals with low density of contribution.

25. Scenario 1. The number of participants is restricted by age and minimum gross salary: a ≤ 45 and WG > SSI. The contribution base is capped at maximum SSI contribution wage: SSI¯=TL10,705.5. Hence the individual contribution base is given by:


Table 1 reports the initial condition for our projections with a total number of 10.2 million workers with estimated monthly gross earnings of TL28.4billion and average monthly earning of TL2,778.13

Table 1.

Turkey: Labor Force Survey Scenarios

article image
Source: IMF staff calculation.

26. Scenario 2. The number of participants is restricted only by the minimum SSI salary: WG > SSI. The contribution base is capped as in case 1. Hence the individual contribution base is given by:


Table 1 reports the initial condition for our projections with a total number of 12.2 million workers (2 million additional workers, older than the age of 45) with estimated monthly gross earnings of TL34.7 billion and average monthly earning of TL2,848.

27. Scenario 3. Neither age nor contribution base is restricted. Hence, the number of workers increases by another 2 million which reported gross earnings below the 2016 minimum wage.


Table 1 reports the initial condition for our projections with a total number of 14.1 million workers with estimated monthly gross earnings of TL37.5 billion and average monthly earning of TL2,657.

28. The age and wage distributions under the 3 scenarios are summarized in the following charts.


Turkey: Total Salaried Workers by Age, 2015


Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Turkey: Total Wage by Age, 2015

(TL million per month)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Turkey: Average Wage by Age, 2015

(TL per month)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Assumptions and Aggregate Savings Projections

29. We project these 3 scenarios using the following assumptions. The first set of assumptions in Table 2 relates to the reform. They characterize the three main scenarios in terms of covered population and wages. In addition, we assume a gross rate of return on accumulated balances of 9.3 percent, the average return in the last 5 years of the current voluntary pension system. We also assume that full coverage is achieved only in 5 years.

Table 2.

Turkey: Projection Assumptions

article image
Source: IMF staff calculations.Notes:

As a percentage of gross monthly earnings;

As a percentage of private sector contributions;

One off transfer after 2 months of participation.

The second set of assumptions in Table 2 relates to the baseline. We assume a labor force growth of 1.9 percent consistent to the historical average. Wage growth is given by the sum of our inflation forecast of 6.3 percent and historical productivity growth of 1.9 percent. The nominal GDP growth is the result of the growth of 3.5 percent for real GDP and 6.3 percent for the GDP deflator.

30. We also add 2 more scenarios. Scenario 4 is based on scenario 2 with a 6 percent private sector contribution rate and no government subsidy/contribution. Scenario 5 is based on scenario 3 with no government subsidy/contribution. They are motivated by the concern on the generosity of the government subsidy. With these scenarios we want to see what happens to aggregate savings if the subsidy is absent but we compensate for this by increasing the covered population and/or their contribution rate.

31. We project aggregate savings for the next 20 years. The next charts decompose the growth in aggregate savings as a percentage of GDP under all 5 scenarios. The increase in aggregate savings is due to four components. Firstly, the government dissaves due to the lumps sum and the Government contribution rate on the flow of private sector contributions.14 Such dissaving is going to be highest in the first years of the implementation s many people are covered by the system per year. When full coverage is achieved, the lump sum contribution is reduced to the new net entries into the labor force per year. Secondly, government dissaving is a transfer into private sector accounts and therefore amounts private sector savings. Thirdly, private sector contributions will grow over time as a function of the growth in the covered labor force and the contribution wage. Fourthly, accumulated assets will yield a return which is automatically re-invested as, by assumptions, cash balances are completely illiquid.


Turkey: Change in Private and Public Savings

(Scenario 01, p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Turkey: Change in Private and Public Savings

(Scenario 02, p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Turkey: Change in Private and Public Savings

(Scenario 03, p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Turkey: Change in Private and Public Savings

(Scenario 04, p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Turkey: Change in Private and Public Savings

(Scenario 05, p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

32. Aggregate savings increase the least under scenario 1 and the most under scenario 4. Under scenario 1 aggregate savings would increase by 1 percent of GDP in only 10 years when coverage is complete (blue line). The low growth rate is due to the fact that the least number of people with the lowest total monthly earnings are covered in the system. In scenario 2 the cap of 45 years of age is released and all workers are allowed to participate. Aggregate savings would increase by 1 percent of GDP already after 7 years (grey line). Finally, in scenario 3, all salaried workers are allowed to participate with no wage restriction. Under this scenario, aggregate savings would increase by 1 percent of GDP already after 6 years (orange line).15

33. Scenarios 4 and 5 are alternatives that attempt to mitigate the impact of higher number of individuals on government dissaving. While a faster increase in aggregate savings is desirable to help rebalance the economy, this comes at the expenses of increasing debt. Scenario 4 replicates scenario 2 but it eliminates government subsidies and it doubles the private sector contribution rate to 6 percent to compensate. The higher contribution rate more than compensates for the absence of the government transfer, Since the contribution base and rate are both larger than in scenario 1, the return on asset component (the grey area in the previous charts) grows very rapidly and over time aggregate savings accumulate more quickly than in all other scenarios. Under scenario 5, we relax both the age and salary constraints as in scenario 3 but we eliminate the government subsidy. Aggregate savings would grow by less than in scenario 2. The problem with scenario 3 is that a large number of individuals are covered but they have a low average wage as now we are adding also about 2 million workers with reported earnings below the minimum wage. Hence, since the large government subsidy is missing and individuals have a very low contribution potential, aggregate savings accumulate very slowly.


Turkey: Change in Aggregate Savings

(p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Model Simulations

34. We simulate the impact of the auto enrolment reform using the G20 core module of the FSGM. The FSGM is a semi-structural model combining both micro-founded and reduced-form formulations of various economic sectors (See Annex II). We simulate the path of aggregate savings Consumption deteriorates.16 This yields directly from the accounting identity that GDP is either consumed or saved. in the previous subsections as annual shocks in the model.17 The key results of the model when aggregate savings increase are and can be followed graphically in Annex III where scenarios 1 and 4 are simulated:

  • Investment increases. The cost of capital (the global interest rate), as perceived by Turkish firms, increases because the REER depreciates. Notice that global rate per se does not increase as Turkey’s contribution to global savings is too small to affect the cost of capital in FX. This should depress investment. However, the real interest rate decreases and this increases investment. Finally, a financial accelerator effect amplifies the short term response of investment to stronger activity. Overall, the last two effect dominate the increase cost of capital and real investment increases.

  • The REER depreciates. With inflation constant, the REER depreciates because of nominal depreciation. The exchange rate depreciates as a consequence of the level of net exports to achieve the current account balance required to support the desired net foreign asset position that reflects households’ desired wealth holdings.

  • Employed labor remains constant. With more capital there is higher demand for labor but both the income and wealth effect tend to decrease the supply of labor. In net terms, employed labor does not change while wage inflation decreases in the short run driven by the negative output gap through a Philips curve for nominal wages.

  • Potential output increases. Because investment is higher while TFP and labor remain constant.

  • A small negative output gap opens in the short run but quickly closes. In the short run real GDP decreases because of the fall in consumption while potential output remains constant. However, in the long run, real investment increases potential output and GDP increase because investment and net exports more than offset the fall in consumption.

  • The monetary authority accommodates to keep inflation constant. The decrease in consumption contributes to lowering GDP in the short run. This opens a small negative output gap and the monetary authority reacts by decreasing the policy rate to keep headline inflation constant. This decreases the real interest rate in the economy. In the long run, GDP increases due to the higher contribution of investment and net exports. The policy rate remains below the baseline over 20 years as the monetary authority continues to be hit by new surprises year after year but it returns to baseline in the longer run (not shown).

  • The private sector saves as per projections. In this model we do not have fully Ricardian agents. Two elements contribute to breaking down Ricardian equivalence. Firstly, the model uses finitely lived, rather than infinitely-lived, OLG households. This means that there is a chance that any tax liabilities associated with the government subsidy (see next) will fall due beyond their expected lifetimes. Secondly, the model contains liquidity constrained (LIQ) household. These do not have access to financial markets, do not save, and thus consume all their income each period. Adding LIQ households amplifies the non-Ricardian properties of the basic OLG household framework and it is one way to simulate liquidity constraints for households who also save.

  • The government dissaves more than projected and debt to GDP worsens. This is due to the presence of automatic stabilizers. When GDP growth decreases by 1 pp, the budget deficits by 25 bps.18 The ratio of debt to GDP worsens relative to the baseline as by assumption, that the deficit is debt financed. The presence of automatic stabilizers is obvious in scenario 4 where no government dissaving is assumed. Budget and debt dynamics are of course contingent on the financing choice. Should the government decide to raise taxes or use privatization proceeds, these would be different.

  • Higher aggregate savings reduce both flow and stock measures of external imbalances. The real depreciation makes Turkey more competitive and the current account improves. Imports decrease as they are positively correlated with domestic activity and are an increasing function of the appreciation if the REER. Exports increase with foreign activity, and are also an increasing function of the depreciation. With the economy rebalancing, the current account, and consequently, the NIIP improve.

35. The new automatic enrollment voluntary pension system contributes only weakly to rebalancing the economy. The following two charts report the deviations of the current account and NFA from the baseline under our 5 scenarios. With scenario 1, the improvement of the current account deficit as a share of GDP amounts to only 1 p.p. in 2027–28. With scenario 2, the CAD improves by the same amount in 2025–26. Scenarios 3 and 4 are better with the CAD improving by 1 p.p. in 2024–25 and 2020–21, respectively. The improvement in NFAs follows similar pattern and ranking under the 5 scenarios studied. In the short run, NFA/GDP worsens slightly because the current account improves at a rate lower than GDP growth. However, in the long run, the rate at which the current account improves accelerates and NFA/GDP improves. A faster implementation schedule would help reach the 1 p.p. target earlier under all scenarios but the improvement would be small (1–2 years). Clearly, a slightly larger base and a higher contribution rate is needed to rebalance the economy.


Turkey: Change in Current Account

(p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

Turkey: Change in NFA

(p.p. in GDP, deviation from baseline)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: IMF Staff Calculations.

D. Conclusions and Policy Recommendations

Industrial organization

36. The newly enacted automatic enrollment provisions has several advantages relative to the current voluntary private pension system. They exploit consumer inertia to maximize participation, fees are capped, they attempt to reduce wasteful competition in the market.

37. But they have several weaknesses that risk endangering the reform in the long run. The hybrid I/O is not complete without the establishment of a public procurement board and periodic auctioning of pension services. Employers are unlikely to be more skilled than individuals in choosing pension plans for their workers. Fees are charged on heterogeneous bases (asset under management) thus prompting competition in the market and wasteful marketing expenses. They would become obsolete over time and it will be necessary to lower them periodically exposing the authorities to the risk of regulatory capture. The SSI is not used to collect pension contributions while pension firms’ sales forces are used to retain individuals in the system generating he incentives and opportunities for misspelling. A political appointee is in charge of investment rules and/or investment options. The governance of the investment advisory committee is yet to be defined and its opinion are not likely to be publicized as customary of the style of Turkish policy making. Foreign asset managers are not likely to be allowed to participate in the auctions.

38. Hence, Staff advises to:

  • Complete the hybrid I/O model along the lines recommended by the World Bank by establishing a procurement board for pension services for undecided participants. The most important objective of the reform is to avoid competition in the market. The central benefit of the procurement board is that it removes the responsibility from employers for choosing pension providers and it establishes a centralized mechanism to select them via periodical auctions. Repeated auctions would contribute to maintaining prices close to average costs even with uniform fees applied to heterogeneous bases. In particular:

    • The Pension Monitoring Center (PMC) appears to be the ideal candidate to act as a public procurement board. The private sector/public sector representation in the would ensure sufficient check and balances during, and transparency about, the public procurement activity.

    • Collection of contributions should be given to the SSI. No private sector service provider would be able to provide the same service at lower costs. In addition, this would represent a duplication of service. It is more efficient for the Government to provide collection of contributions as a public good.

    • Initially, allow Takas Bank to provide custodian services and/or record keeping and administration. This service, especially custodianship, would need to be auctioned periodically so as to ensure maintenance of quality of service over time.

    • Asset management should be allocated through competitive auctions where foreign asset managers could participate. Ideally, new entrants would provide a limited set of investment options following passive asset management styles. Asset management and custodianship could be bundled in the same auction. The presence of qualified foreign asset managers could be beneficial for specific mandates for which local expertise is missing or weak. In addition, they would generate competition among domestic asset managers to reduce the currently very high fees.

  • For the quasi market, wasteful competition within the market would be eliminated with flat fees. A flat fee kills incentive for marketing and protects participant in the procured segment from predatory pricing by providers in the quasi market. It also protects participants in the quasi market from wasteful competition from pension firms. More critically, it facilitates the existence of the procurement segment where fee rates would be much higher than in the quasi-market segment to cover production costs.

  • Finally, replace the current proportional subsidy with a means tested flat subsidy. Flat fee would be high for low income participants or participants with low contribution density in general. Hence a flat subsidy that is means tested and comparable to the current proportional one in terms of fiscal cost, would enable low wage individuals to participate in the hybrid I/O pension system.

Macroeconomic Impact

39. The new automatic enrollment system has many positive macroeconomic implications. The increase in aggregate savings would increase GDP and potential output. As real investment would increase with the lower cost of funding and higher savings, potential output would be boosted. At the same time higher real investment and stronger net exports would compensate for the fall in consumption induced by higher aggregate savings. This would raise GDP growth keeping the output gap closed in the long run.

40. However, the impact on external imbalances of the new enrollment system would be felt only after 20 years. Under plausible scenarios, the CAD would improve by 1 p.p. only between 2024 and 2028 with the latter date being more than 10 years from today. An improvement of 2 p.p. would not be possible before 20 years from today. The increase in aggregate savings would be even smaller if the authorities’ assumptions discussed in footnotes 13 and 15.

41. Hence, Staff advises to increase the private sector contribution rate and the contribution base.

  • A contribution rate of at least 6 percent is needed to improve the CAD by at least 1 p.p. in the next 5 years; sooner if the government subsidy is maintained.

  • Individuals older than 45 years of age should also be allowed to participate and the cap on their contribution salary lifted. However, a change in the contribution base would require a legislative amendment which could be impractical. It would also reduce the ability to reconcile the flows of contributions and information. Hence, maintaining the current wage cap could be compensated by a further increase in the contribution rate.

Annex I. Key Aspects of the New Auto Enrollment System

Automatic participation. Salaried employees under 45 years of age and registered with the Social Security Institution (SGK) will be included in a pension plan through a pension contract arranged by the employer. The pension company that the employer may choose must be amongst the one approved by the Undersecretariat of Treasury.

Contributions. Participants’ contribution amount will be 3 percent of their earning subject to Social Security Insurance (SSI) contributions. Such earnings have a lower limit of has its own upper and lower limits by law. The lower limit is the monthly gross minimum wage (TRY1,647 for 2016), and the upper limit at 6.5 times Gross Monthly Minimum Wage (TRY10,705.5). The Council of Ministers is authorized to double the contribution rate or lower it to 1 percent, or determine it to a fixed amount. The employee may request the employer to make a deduction in an amount greater than the amount specified in the pension contract for automatic enrollment.

Matching. The Government will provide state subsidy for employees amounting to 25 percent of employees’ paid contributions to private pension account. In case the employee stays in the plan, the government provides another one-off subsidy of TRY1,000.00. Employer’s matching contribution is not an option within the auto enrollment system.

Opting out. Participant employees may opt out from the system within 2 months of participation. In this case the accumulated amount of contributions and investment income, if any, will be refunded within 10 days.

Portability. Full portability is ensured by having individual accounts linked to the national ID number. In the event of workplace change, employee’s accumulated savings and retirement time basis gained in the system will be transferred to pension contract of the new workplace, if the new workplace has a pension plan. In case the new workplace does not have a pension plan, employees may continue to pay contribution to the contract arranged in previous workplace upon request.

Retirement benefits. Upon retirement, the employee has the choice of receiving their savings in a one-lump sum or as a monthly annuity over the course of several years. If the employee chooses to receive their pension in annuities exceeding 10 years, the state will provide an extra 5 percent contribution of a private pension customer’s total savings.

Fees. Pension companies cannot charge any fees on participants of auto-enrollment plans other than fund management fees. These will be capped.

Roll out. The plan is to start implementing the new system starting with large employers in January 2017.

Pending issues at the time of writing. Enrollment of an employee to a pension plan by way of their employer; the criteria to be taken into consideration by the employer when choosing the company and pension plan; the funds in which contributions will be invested; the contract to be entered into by the employer with the company; the right to withdrawal; enrollment of an employee to a pension plan by the employer in cases where the employee changes their workplace; transfer of accumulated savings in cases where the employee changes their workplace; payment to the related pension plan upon the request of the employee whose employment relationship has been terminated; pausing contribution payments; abandonment of the system; payment of Government contributions and other principles and procedures governing the enforcement of the new provisions are yet to be defined by the Undersecretariat of Treasury.

Annex II. The Flexible System of Global Models

The Flexible System of Global Models (FSGM) is a semi-structural model combining both micro-founded and reduced-form formulations of various economic sectors. Real GDP in the model is determined by the sum of its demand components in the short run, and the level of potential output in the long run. What follows is a brief overview1 of the components of aggregate demand, potential output, the price block, commodities, and finally monetary and fiscal policy.

A. The Real Side

Aggregate demand follows the standard national expenditure accounts identity, where real GDP is the sum of household consumption, private business investment, government absorption and exports of goods and services, less imports of goods and services.

The consumption block is micro founded and uses the Blanchard-Weil-Yaari overlapping generations (OLG) model of households. Using OLG households that have a finite expected lifetime rather than infinitely-lived households results in important non-Ricardian properties whereby the path for government debt, and thus fiscal policy actions, have significant implications for private consumption dynamics. The model also contains liquidity constrained (LIQ) households that do not have access to financial markets, do not save, and thus consume all their income each period.

In the OLG framework, households treat government bonds as wealth since there is a chance that the associated tax liabilities will fall due beyond their expected lifetimes. The OLG formulation thus results in the endogenous determination of national savings given the level of government debt. Consequently, the world real interest rate is endogenous and adjusts to equilibrate the global supply of and demand for savings. The use of an OLG framework necessitates the tracking of all the stocks and flows associated with wealth, and thus the model has full stock-flow consistency.

Private business investment is also micro founded and uses an updated version of the Tobin’s Q model, with quadratic real adjustment costs. Investment is negatively correlated with real interest rates. Investment cumulates to the private business capital stock, which is chosen by firms to maximize their profits. The capital-to-GDP ratio is inversely related to the cost of capital, which is a function of depreciation, the real interest rate, the corporate tax rate, and relative prices.

Government absorption consists of spending on consumption and investment goods. Government consumption spending only affects the level of aggregate demand. It is an exogenous choice determined by the fiscal authority. The level of government investment is also chosen exogenously, but in addition to affecting aggregate demand directly it also cumulates into a public capital stock, which can be thought of as public infrastructure (roads, buildings, etc.). A permanent increase in the public capital stock permanently raises the economy-wide level of productivity.

The real competitiveness index (RCI) is the long-run determinant of the level of net exports that adjust to achieve the current account balance required to support the desired net foreign asset position that reflects households’ desired wealth holdings. Exports and imports, individually, are modeled using reduced-form equations. Exports increase with foreign activity, and are also an increasing function of the depreciation in the RCI. Imports increase with domestic activity, and are an increasing function of the appreciation of the real effective exchange rate (REER).

The current account and implied net-foreign-asset positions are directly linked to the saving decision of households. The model can be used to study both creditor and debtor nations as positive or negative net foreign asset positions can be a feature of the well-defined steady-state in the OLG framework.

Aggregate supply is captured by potential output, which is based on Cobb-Douglas production technology with trend total factor productivity, the steady-state labor force, the non-accelerating inflation rate of unemployment (NAIRU), and the actual capital stock.

The unemployment rate varies relative to the NAIRU according to an Okun’s law relationship with the output gap.

B. Prices

The core price in all regions is the consumer price index excluding food and energy, CPIX, which is determined by an inflation Phillips curve. CPIX inflation is sticky and reflects the expected paths of import prices and the economic cycle, as captured by the output gap. In addition, although the direct effects of movements in food and energy prices are excluded, there is a possibility that persistent changes in oil prices can leak into core inflation. In addition, there is a Phillips curve for nominal wage growth. Wage inflation exhibits stickiness and allows the real wage to return to its equilibrium only gradually depending on the expected evolution of overall economic activity.

There is also a full set of prices that mimic the structure of demand: consumption; investment; government; exports; and imports. The GDP deflator itself is a weighted average of the consumption, investment, government, export, and import deflators.

The model also incorporates three types of commodities – oil, food and metals and their associated prices. This allows for a distinction between core and headline inflation, and provides richer analysis of the macroeconomic differences between commodity-exporting and -importing regions arising from commodity-based terms-of-trade shocks.

C. Policy

In the short run, the nominal side of the economy is linked to the real side through monetary policy. The behavior of monetary authorities is represented by an interest rate reaction function. The standard form is an inflation-forecast-based rule operating under a flexible exchange rate. However, due to the form of the interest rate reaction function, the model can accommodate a fixed exchange rate regime, monetary union, or a managed floating exchange rate regime.

Monetary policy can influence activity through both short-term and long-term interest rates. The long-term, 10-year, interest rate is based on the expectations theory of the term structure, plus a term premium. The interest rates on consumption, investment, government debt and net foreign assets are weighted averages of the 1-year and 10-year interest rates, reflecting their differing term structures, and allowing for a meaningful role for the term premium.

Annex III. Model Simulations of Scenarios 1 and 4


Model Simulations of Scenarios 1 (blue) and 4 (green)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A001

Source: Staff calculations using G20MOD.


  • Andrle, M., Blagrave, P., Espaillat, P., Honjo, K., Hunt, B., Kortelainen, M., Lalonde, R., Laxton, D., Movroeidi, E., Muir, D., Mursula, S., and Snudden, S., 2015, “The Flexible System of Global Models (FSGM)”. IMF Working Paper No. WP/15/64.

    • Search Google Scholar
    • Export Citation
  • Calderón-Colín, R., Domínguez, E. E., and Schwartz, M.J., 2008, “Consumer Confusion: The Choice of AFORE in Mexico”. IMF Working Paper No. WP/08/177.

    • Search Google Scholar
    • Export Citation
  • Impavido, G., 2013, “Pension Funds” Chpt 47 In: Gerard Caprio (ed.) “Handbook of Key Global Financial Markets, Institutions, and Infrastructure”, Vol. 1, pp. 52332. (Oxford: Elsevier Inc. Pub.).

    • Search Google Scholar
    • Export Citation
  • Impavido, G., Lasagabaster, E., and García-Huitrón, M., 2010New Policies for Mandatory Defined Contribution Pensions – Industrial Organization and Investment Products”. (The World Bank Pub.).

    • Search Google Scholar
    • Export Citation

Prepared by Gregorio Impavido.


A larger set of features is summarized in in Annex I. This fuller set of information is used also in the next section.


The assessment and policy advice of this section is based on chapter 3 of Impavido et al. (2010).


We distinguish in this note between pension firms (the corporate entity) which operates a pension plan (the legal contract) by, inter alia, managing pension funds (the accumulated cash balances). For a more detailed private pension taxonomy, see Impavido (2013).


Private pensions are referred to as “quasi-market”. They are “markets” because services (although not necessarily all) are provided by competitive independent, often specialized—entities. They are also “quasi” because they differ from a conventional market on both the demand and supply side. On the demand side, consumption is typically mandatory generating a captive clientele. On the supply side, providers do not necessarily maximize profits and their governance structure includes both private and public sector firms, as well as for profit and mutual associations. A quasi-market may also emerge in the absence of formal compulsion if incentives for participation lead to the creation of a de facto captive clientele, as in the case of Turkey.


This is the key rationale for the strong popular discontent with the pension industry in Chile in these days. In Chile, 6 pension fuds manage assets amounting to 73 percent of 2015 GDP and charge fee rates around 75bpsThis yields, gross revenues in the neighborhood of 50bps of GDP that have generated popular discontent as deemed excessive.


Choice does not need to be over providers, it could be over investment options provided by asset managers procured by the public board.


The difference in fee bases between the two segments can be substantial as undecided participants have also low contribution density. When Mexico switched in 2008 to asset-based only revenues, the inequality between the two segments increased dramatically. The average balance per participant in the allocated segment was US$30 compared to ten times that in the quasi-market.


We did not increase the labor force by the 1.9 percent assumed growth rate. Hence, our calculations underestimate the growth in the wage mass, albeit this omission contributes very little.


We use “wage” and “earnings” interchangeably for this work.


In November 2016, this was brought to 7.5 the minimum wage. The increase in the contribution base would approximately add an additional 10bps in GDP to the increase in aggregate savings calculated in this note.


This assumes no impact on quantities and excludes any impact on wages above the 2016 minimum wage. Although these impacts work in opposite directions, it is expected that our simplification underestimates the overall impact on the wage mass.


The authorities count 14 million individuals under scenario 1 with an average wage between TL2,300 for public workers and TL2,500 for private workers. The difference between our estimated covered participants could be explained by the fact that we excluded around 2 million individuals with earnings below the minimum wage and that we excluded individuals with zero reported net earnings. This could also explain why the authorities have also a lower average wage.


We assume that due to inertia, individuals do not opt out. Authorities assume a positive opt out rate that lowers, other things equal, their estimates on aggregate savings.


We projected aggregate savings under scenario 1 using the authorities’ assumptions for covered workforce reported in footnote 13, a gross rate of return of 7 percent and an opt out rate of 8 percent (also authorities’ assumptions). With these assumptions aggregate savings grow more quickly than our scenario 1 in the short term as more people are covered over time and the government transfer plays a larger role. Over the long time, aggregate savings grow less than in our scenario 1 as the gross rate of return is lower and (more critically) the covered population shrinks: the opt out rate is much higher than the growth rate of the labor force.


All statements should be interpreted as “deviation from the baseline”.


This, inter alia, has the effect of reducing the impact of rational expectations on nominal variables that, as a consequence, adjust much more gradually.


This assumption is standard for all EMs in the FSGM.


This is a summary of Andrle et al. (2015).

Turkey: Selected Issues
Author: International Monetary Fund. European Dept.