Pension Systems: Alternative Designs in Latin America By Solange Berstein Jáuregui and Alejandro Puente Gómez

Transcription

1 0 Foreword By Álvaro Clarke De la Cerda 1 Pension Systems: Alternative Designs in Latin America By Solange Berstein Jáuregui and Alejandro Puente Gómez 2 Pension Plans: History and Perspectives By Vicente Lazen Jofré and Álvaro Clarke De la Cerda A Decade of Annuity Reform: Lessons, Progress, and Remaining Challenges By Alejandro Ferreiro Yazigi 3 Reserves and Solvency of the Life 4 Insurance Companies that Sell Annuities By Gonzalo Edwards Guzmán and Guillermo Martinez Barros Economic Impact of Insurance Company Investments By José Ramón Valente Vias 5 Organization of the Annuity Market in Chile By Guillermo Le Fort Varela 6 Asociación de Aseguradores de Chile a.g.

2 Vicente Lazen Jofré Vicente Lazen is a partner in the consulting company BC Fundamenta. Until 2014, he was a supervisor with the Superintendence of Pensions. He has also held several positions with the Chilean Superintendence of Securities and Insurance, including head of supervision of securities custody, clearing, and settling; head of international relations; and economist in the Research Division. He served as a member of the BIS Committee on Principles for Financial Markets Infrastructures. He has also worked as a consultant in Latin America and as professor of finance and microeconomics at the University of Chile. Mr. Lazen has a bachelor s degree in business administration with a minor in economics from the University of Chile and an MBA from the University of Texas. Álvaro Clarke De la Cerda Álvaro Clarke has a bachelor s degree in economics and business administration from the University of Chile and a master of arts in economics from the University of Leuven. He currently heads the Center for Corporate Governance at the University of Chile and is president and partner of the rating agency ICR Clasificadora de Riesgos. He is also a founding member of the Latin American Corporate Governance Roundtable, a joint World Bank and OECD task force. Mr. Clarke has served as Under-Secretary of Finance and Superintendency of Securities and Insurance and participated on the Presidential Advisory Commission on Pension Reform, the Anti-Monopoly Commission, and the Pension Fund Technical Investment Committee (as chair). At the international level, he has served as chairman of the Association of Latin American Insurance Supervisors and the Ibero-American Securities Markets Institute, vice president of the Council of Securities Regulators of the Americas, and alternate governor of the IDB. As Superintendency of Securities and Insurance, Mr. Clarke designed and spearheaded key capital market reforms, including the Law on the Public Offer of Shares and Corporate Governance and the first Capital Market Reform (MKI).

3 Pension Plans: History and Perspectives By Vicente Lazen Jofré and Álvaro Clarke De la Cerda 61 Introduction 63 1 Pension Plans in Chile 1.1 Immediate Life Annuity 1.2 Progammed Withdrawals 1.3 Temporary Income with Deffered Life Annuity 1.4 Immediate Life Annuity with Progammed Withdrawals 1.5 Pension Plans in Countries with Individual Capitalization 71 2 Conceptual Foundations of Pension Plans 2.1 Risks Faced by Pensioners Longevity Risk Investment Risk Spending Risk 2.2 Research on Individual Choice 79 3 Historical Evolution of Pension Contracts 3.1 Plan Selection over Time 3.2 Returns Offered over Time 3.3 Research on Annuity Competition (Money s Worth of Annuities) 3.4 The Implementation of the scomp System 86 4 The Current Outlook for Pensions Plan 4.1 Variable Pension Alternatives 4.2 Alternative Pension Plans for Addressing Longevity Risk 4.3 Strengthening Pension Knowledge and Brokerage 94 5 Conclusions 95 References

4 4 vicente lazen jofré and álvaro clarke de la cerda

5 INTRODUCTION For most people who are close to retirement age, the set of choices that have to be made represent one of the most important events in their lives in terms of their personal finances. This process involves, first, a choice that originates in the legal possibility of choosing between two pension products and combinations thereof that mainly differ in the allocation of the risks that have to be assumed. This chapter analyzes the different pension plans that are currently available in Chile: programmed withdrawals, life annuities, and programmed withdrawals with deferred life annuities. The analysis includes a discussion of conceptual and theoretical issues, the historical development of the pension industry, and the different commercial models applied by pension providers. Our analysis addresses two of the most pressing challenges facing the country in terms of pension products. First, as longevity has increased while the retirement age and contribution rates have remained the same, funding pensions in the later years of retirement has become progressively more costly in relative terms. Second, the low yields on fixed-income assets both locally and internationally have put pressure on pension levels and raise the need to evaluate new pension alternatives, including the adoption of variable-income schemes with longevity insurance. The chapter is organized as follows. The first section describes the different pension plans and variations offered in Chile, from a legal and technical perspective, and also reviews the products offered in other individual capitalization pension systems. The second section discusses some basic conceptual issues related to the risks that condition the individual s choice of pension plans, including longevity and investment risk. The third section analyzes historical trends in the choice of pension products over the last few years, reviewing the shift toward choosing pension plans through the online Pension Consultations and Offers System (Sistema de Consultas y Ofertas de Montos de Pensión, or scomp). We also present a study of the returns available to pensioners, concluding with an assessment of competition in the annuity segment from a historical perspective. Finally, we offer our perspective on the different alternatives for the future development of the industry, in terms of the pension products that should be evaluated as possible tools for facing the challenges described herein. pension plans: history and perspectives 5

6 6 vicente lazen jofré and álvaro clarke de la cerda

7 1 PENSION PLANS IN CHILE In Chile, Decree Law 3,500 stipulates that in order to cash in pension benefits, pension system affiliates must choose one of the following plans: (a) immediate life annuity; (b) programmed withdrawals; (c) temporary income with a deferred life annuity; or (d) immediate life annuity with programmed withdrawals. This section reviews each of these plans from a legal and technical perspective. 1.1 IMMEDIATE LIFE ANNUITY Under an immediate life annuity, the affiliate signs an irrevocable contract with a life insurance company to receive a monthly income from the moment the contract is signed until the death of the pensioner and his or her legal beneficiaries. The life insurance company is then obligated to pay the monthly annuity, as well as the survivor s pension to the pensioner s beneficiaries. This liability is effective once the balance of the individual capitalization account is transferred to the company and allocated to pay the premium. 12 Decree Law 3,500 was legally reformed through Law 19,934 of 2004, establishing that annuities can also be offered as a variable-income plan. The law stipulates that an annuity can be constant or variable over time and that the constant or fixed-income part of the variable annuity must be denominated in unidades de fomento (ufs, an inflation-indexed unit of account commonly used in Chile), while the variable component can be denominated in the legal national currency, a foreign currency, or an investment portfolio index authorized by the Superintendency of Pensions. The immediate life annuity plan is only available to pension affiliates who can contract an income equal to or greater than the basic solidarity old-age pension. Under an annuity plan, affiliates have the option of contracting special insurance conditions to improve the situation of their survivor s pension beneficiaries. For example, an affiliate can contract a guaranteed period, where the life insurance company commits to paying 100% of the contracted pension, distributed among the pensioner s legal beneficiaries, in the event the affiliate dies before the end of the guaranteed period. At the end of that period, the survivor s pension benefits are paid according to the percentages stipulated by the law. A second special condition is the percentage increase clause, where, in the event of the pensioner s death, the life insurance company pays the spouse and 12 As a general rule, the affiliate can freely withdraw as a surplus any funds in excess of the amount necessary to finance a pension equal to or greater than 100% of the maximum pension with the solidarity contribution and 70% of the average wages received and income declared. In the case of affiliates entitled to a disability pension, the level is 70% of the base income. pension plans: history and perspectives 7

8 other survivor s pension beneficiaries a larger percentage of the contracted annuity than stipulated by law. This option can only be contracted if the affiliate has a spouse. From an actuarial perspective, the annuity is financed through the funds that are transferred to the company, the yields earned on those funds, and also any funds that are freed on the death of other pensioners. The latter component is called a cross-subsidy: affiliates who die before reaching the average life expectancy subsidize those who live beyond the average. In the early years of the annuity, investment earnings fund the majority of the annuity, whereas the cross-subsidy becomes more significant as time passes (Blake and Biffs, 2012). The annuity is equal to the annual flow (rv), expressed in ufs, that solves an equation in which a single premium (the amount transferred to the life insurance company) is equal to the annuity flow, weighted by the probability of occurrence (that is, that the pensioner is alive) and discounted at a rate that ultimately depends on the life insurance company. Expressed mathematically in simplified form, to improve conceptual clarity for a 65-year-old retiree with no legal beneficiaries, the formula for calculating an annual annuity is as follows: 110 x-65 RV Px (1+TV) (x-65) + CM where x is the age of the beneficiary in each year; rv is the annual amount of the annuity, that is, the monthly annuity multiplied by 12; Px is the probability that the pensioner will be alive in one more year, given that he or she is alive at age x, calculated through the hypothetical age of 110 year; CM is the funeral benefit; and TV is the internal rate of return that solves the equation. A lower TV rate solves the equation through a lower annuity and vice versa. From the perspective of the future pensioner, therefore, a higher rate is beneficial. This rate, which financially is the internal rate of return (IRR), is an indicator of the price of the annuity, and it is strictly related to the yields offered by the fixed-income instruments in the market and the degree of competition in the industry. Simple observation of the equation reveals the following results. First, as mentioned earlier, given a fixed single premium, a higher TV implies higher annuity flows. Second, an improvement in morality (that is, an increase in life expectancy) and thus an increase in Px necessitate lower annuity flows, ceteris paribus, which is intuitively clear: a longer survival period after retirement requires sacrificing the amount of the pension to be received so as to fund the later years of the pension. Finally, a trivial result is that a lower single premium can only be solved through lower annuity flows. 8 vicente lazen jofré and álvaro clarke de la cerda

9 1.2 PROGRAMMED WITHDRAWALS Under a programmed (or systematic) withdrawal plan, the pension received by the affiliate is deducted from the balance held in an individual capitalization account, which is determined by dividing the real account balance by the capital necessary to pay a pension unit (cnu) to the pensioner and, in the event of death, to the pensioner s beneficiaries. The amount of the pension is recalculated annually based on the individual account balance, which depends on the withdrawals in the previous year, the share return of the pension fund in which the account balance is held, the life expectancy of the affiliate and/or any beneficiaries, and the current interest rate for calculating systematic withdrawals. Under a programmed withdrawal plan, the affiliate retains ownership of the fund and is allowed to change to another pension fund administrator (pfa) or type of pension plan. Consequently, investment risk and longevity risk are borne by the retiree, and the ownership of the funds is never transferred to a third party as long as the pensioner is alive. When the affiliate dies, the remaining account balance is used to continue paying survivor s pensions to any beneficiaries; in the event that the pensioner has no beneficiaries, any remaining funds are paid out as a bequest. The equation for calculating a programmed withdrawal pension, expressed in annual terms, for an affiliate without beneficiaries, is as follows: Saldo cci Pensión = cnu Rearranging the equation to parallel the annuity equation presented above yields Saldo cci =Pensión*cnu where Saldo cci is the balance in the mandatory individual capitalization account plus the recognition bond (a government bond issued to affiliates who made contributions to the old pension system but moved into the new system after the 1981 reform) plus the balance in any voluntary savings accounts. It is comparable to the single premium, or fixed sum, transferred to the life insurance company in the case of an annuity plan. On the other side of the equation, Cnu is the necessary unit capital per pension unit, an actuarial factor that makes the programmed withdrawal calculation more complicated than the annuity since it has to be recalculated annually. pension plans: history and perspectives 9

10 l x+t l CNU = x 11 (1+it ) t 24 where i t, is the interest rate for a period of t years, and l x is the number of people alive at age x according to the corresponding mortality table. The application of the annual recalculation derives from the need each year to incorporate the probability that the pensioner will live an additional year, based on having actually reached his or her current age. The programmed withdrawal interest rate also has to be adjusted, as it is determined annually. Moreover, since the funds in the individual capitalization account are drawn down from year to year, the new balance has to be considered in the new pension amount applicable for the following year. The expected mortality rate specifies a percentage by which the pension is reduced in the next year. Based on this factor alone, the amount of the pension will be lowered at the same rate as the probability of dying in the next year. Finally, there could be a positive effect on the pension if the rate of return on shares in the pension fund in which the balance of the individual capitalization account is invested exceeds the programmed withdrawal interest rate. Projected Pension from Programmed Withdrawal versus Annuity Plans (uf) Monthly pension, programmed withdrawal (3.3% fund return) Age (years) Monthly pension, programmed withdrawal (5.0% fund return) Monthly pension, life annuity Simulation for a 65-year-old man with no beneficiaries, with an accumulated balance of uf 2,000. Programmed withdrawal interest rate: 3.30%, with expected returns on the individual capitalization account of 3.30% and 5.0%. Annuity sales rate: 2.23%. Given the design of the programmed withdrawal plan, the pension necessarily has a negative slope that starts declining within the first few years of retirement. Consequently, depending on the investment return of the pension 10 vicente lazen jofré and álvaro clarke de la cerda

11 fund in which the pensioner s individual capitalization account is invested, the pension is cut in half within 20 to 25 years after retirement. In order to smooth pension flows during the later years of a programmed withdrawal plan, the Pension Reform of 2008 modified Decree Law 3,500 to incorporate an adjustment factor, whereby a programmed withdrawal pension that does not meet the eligibility requirements for the solidarity pension contribution (because the benchmark self-financed pension is above the maximum pension with a supplementary solidarity contribution) must maintain a special reserve containing the balance necessary to finance a pension of at least 30% of the affiliate s benchmark programmed withdrawal 13 or 30% of the first payment of the survivor s pension through 105 years. Chronologically, 30% of the benchmark systematic withdrawal has a higher probability of being reached between 89 and 95 years, depending on gender, the retirement age, and the pension fund investment return. The following figure illustrates the difference between the adjusted and unadjusted programmed withdrawal pension curves. Simulation of a Programmed Withdrawal Pension with and without the Applied Adjustment Factor (uf) Age (years) Unadjusted pension Adjusted pension Notes: Simulation for a 60-year-old woman with no beneficiaries, with an accumulated balance of uf 10,000. Programmed withdrawal interest rate: 3.30%, with expected returns on the individual capitalization account of 3.30%. 13 The benchmark programmed withdrawal is equal to the individually calculated pension at the time the affiliate retires or at the legal retirement age, whichever comes first. pension plans: history and perspectives 11

12 1.3 TEMPORARY INCOME WITH DEFERRED LIFE ANNUITY Under this plan, the affiliate chooses to transfer part of the funds in his or her individual capitalization account to a life insurance company, in exchange for a monthly annuity starting on a future date specified in the contract. The affiliate also maintains a sufficient account balance to receive a monthly income from the pension fund administrator (pfa) throughout the period prior to the start of the annuity contract. The legislation stipulates that the contracted deferred annuity cannot be less than 50% of the first monthly payment of the temporary income or more than 100% of that first payment. Projected Pension from a Temporary Income with Deferred Annuity Plan (uf) Age (years) Monthly pension from programmed withdrawal Monthly pension from annuity Notes: Simulation for a 65-year-old man with no beneficiaries, with an accumulated balance of uf 2,000. Programmed withdrawal interest rate: 3.30%, with expected returns on the individual capitalization account of 3.30%. Annuity sales rate: 2.23%. As in the case of an immediate life annuity, the affiliate has the option of contracting special insurance conditions, as described earlier. 1.4 IMMEDIATE LIFE ANNUITY WITH PROGRAMMED WITHDRAWAL Under an immediate annuity with programmed withdrawals, the affiliate uses part the balance in his or her individual capitalization account to contract an immediate annuity with a life insurance company, while the remaining balance is used for systematic withdrawals. The pension is thus the sum of the amounts received from these two sources. As in the case of an immediate annuity, this plan is only available to those who can obtain an immediate annuity that is equal to or greater than the basic solidarity old-age pension. 12 vicente lazen jofré and álvaro clarke de la cerda

13 Projected Pension from a Temporary Income with an immediate Annuity Plan with Programmed Withdrawals (uf) Age (years) Monthly pension from programmed withdrawals Monthly pension from annuity Sum of the two pensions Notes: Simulation for a 65-year-old man with no beneficiaries, with an accumulated balance of uf 2,000. Programmed withdrawal interest rate: 3.30%, with expected returns on the individual capitalization account of 3.30%. Annuity sales rate: 2.23%. The affiliate allocates uf 1,200 for programmed withdrawals and uf 800 to contract the annuity. Affiliates who choose this plan can invest the balance that is allocated to programmed withdrawals in any of the pension funds offered by their pfa. The following table summarizes the characteristics of the different pension plans available in Chile. Pension Plans en Chile Programmed withdrawals Life annuity Temporary income with deferred annuity Management PFA LIC PFA and LIC Option of changing plans Always No Only move up the deferred annuity Ownership of funds Affiliate LIC Affiliate and LIC Investment risk Affiliate LIC Affiliate and LIC Longevity risk Affiliate LIC LIC Pension amount Variable Constant Constant State guarantee or APS Yes Yes Yes Bequest Yes No Only the temporary income Abbreviations: pfa: Pension fund administrator; lic: Life insurance company; aps: Solidarity pension contribution. pension plans: history and perspectives 13

14 1.5 PENSION PLANS IN COUNTRIES WITH INDIVIDUAL CAPITALIZATION The following table shows the different combinations of pension plans available in countries in Latin America that have fully or partially adopted an individual capitalization system (fiap, 2013). As the table shows, annuities are offered in all the countries included in the table. Systematic withdrawals are also present in almost all cases, except for Bolivia, Costa Rica, and Uruguay. Argentina further incorporates a partial withdrawal plan (which has a more accelerated schedule than a programmed withdrawal plan and is used when the calculated pension is less than 50% of the basic universal pension). Benefit Argentina Bolivia Chile Colombia Costa Rica El Salvador Mexico Peru Dominican Rep. Uruguay Programmed withdrawal Life annuity Partial withdrawal Programmed withdrawal with deferred annuity Source: Federación Internacional de Administradoras de Fondos de Pensiones (fiap). 14 vicente lazen jofré and álvaro clarke de la cerda

15 2 CONCEPTUAL FOUNDATIONS OF PENSION PLANS For most people who are close to retirement age and who have the possibility of receiving a pension, the set of choices that have to be made represent one of the most important events in their lives in terms of their personal finances. This process involves, first, a choice that originates in the legal possibility of choosing between two pension products and combinations thereof that mainly differ in the allocation of the risks involved. The economic literature in this area has developed around a conceptual rationalization of the behavior of a hypothetical individual, who makes dynamic optimization decisions conditioned not only on his own willingness to take on risk, but also on individual expectations in terms of longevity, expenses, and additional future income. Much effort has gone into modeling this complex choice, but it has proven extremely difficult to compare these models with the empirical behavior of pensioners. This section reviews the risks to which a pensioner or future pensioner is exposed and analyzes the different studies on the issue and their results. 2.1 RISKS FACED BY PENSIONERS Future pensioners must face a wide range of risks deriving from various sources, both at the time of choosing a pension plan and during the retirement period. This section analyzes longevity risk, interest rate and reinvestment risk, inflation risk, and spending risk LONGEVITY RISK From the individual s perspective, longevity risk can be defined as the risk of living so long that the pensioner s asset portfolio or stock of resources eventually becomes insufficient to finance a pension that supports an acceptable standard of living that is, the pensioner outlives his or her assets. From the pension provider s perspective, longevity risk is the risk that future pension payments are greater than expected or estimated (Antolin, 2013). From the individual s perspective, the factors that contribute to the materialization of longevity risk can be either voluntary or involuntary, such as a low propensity to save in the asset accumulation phase or a longer effective lifespan than expected for that age cohort or simply an underestimation by the future pensioner of his or her life expectancy. With regard to the latter, some recent studies confirm that people tend to underestimate their life expectancy or age of death. 14 People think they will live less than they actually live, and that un- 14 For example, O Brien, Fenn, and Diacon (2005) find that in the case of Great Britain, men pension plans: history and perspectives 15

16 derestimation is greater among younger people, which has an impact on the propensity to save during the life stage when saving is most critical. Underestimating one s own life expectancy also leads to the choice of pension plans that provide greater liquidity and income in the early years of retirement. In other words, it fosters a preference for liquidity, which is reflected actuarially in a higher discount rate. Over and above the underestimation of life expectancy that is, even if the future pensioner assumes that his or her life expectancy is the same as the average for that age cohort (the individual s expectation is unbiased in neither underestimating nor overestimating life expectancy) there is an additional source of complexity in terms of the variability of the age of death. For example, Blake (2013) shows that if the life expectancy of 65-year-old men in the United Kingdom is 87, then 25% of them can be expected to live to less than 80 years of age, while another 25% will live to over 93. In the case of Chile, this variability can be illustrated using the 2009 annuity mortality table (updated to 2014). The table shows that Chilean women who are 60 years old today have a life expectancy of 88 years. However, only 50% will live to between 82 and 96 years of age, a fairly wide range, while the rest will die at the extremes of the distribution. tend to underestimate their life expectancy by 4.62 years and women by 5.95 years, in comparison with the estimates by the government s actuarial department. This is despite the fact that the same survey respondents tend to expect to live longer than others of the same age and gender. In Chile, the 2006 Social Protection Survey indicates that 60-year-old women expect to live 79.7 years, on average, whereas the life expectancy of pensioners at the legal retirement age was 87.9 years according to the 2004 annuity mortality tables. 16 vicente lazen jofré and álvaro clarke de la cerda

17 Distribution of the Expected Morality for the 60-Year-Old Cohort of Chilean Female Pensioners in 2014 (2009 annuity table adjusted to 2014) Life expectancy of 60-year-old cohort (women) 24,45% 24,39% A similar conclusion can be reached for 65-year-old male retirees in Chile, who, according to the same mortality table, have a life expectancy of 85 years. However, only 50% will live to between 78 and 90 years of age, while the rest will die at the extremes of the distribution. Distribution of the Expected Morality for the 65-Year-Old Cohort of Chilean Male Pensioners in 2014 (2009 annuity table adjusted to 2014) 5 4,5 Life expectancy of 65-year-old cohort (men) ,5 26,2% 26,9% 2 1,5 1 0, Also from the individual s perspective, we need to acknowledge mortality risk, that is, the event of not surviving very long after retiring and thus not using a large share of the accumulated savings. Based on unique personal information pension plans: history and perspectives 17

18 on a given pensioner s health status, this risk can be eliminated by choosing a pension plan that does not cover longevity risk. On the other hand, from the pension provider s perspective, the actuarial design of the pension constitutes a risk factor to the extent that it could systematically underestimate the life expectancy of the people in a cohort (model risk), resulting in a situation in which an improvement in life expectancy is incorporated into the mortality tables. If this latter risk is not transferred, pensioners will be facing a faster depletion of their savings. Significant medical progress, improved living standards, and better hygiene conditions, such as healthier lifestyles, together with the absence of global war conflicts or major pandemics, are key determinants of life expectancy (Gutterman et al., 2008) INVESTMENT RISK From the pensioner s perspective, investment risk involves the possibility that the amount saved for retirement ends up being insufficient because the assets in which the funds were invested had a lower-than-expected return. For our purposes, investment risk includes inflation risk namely, the reduction in the purchasing power of the pension payments or the amount saved. Investment risk should be understood not only in terms of an acceptable yield range, but also with regard to the time interval in which it is measured. Thus, many people could accept a volatility component in their pension to the extent that it is offset by the expected return on their assets. For example, the type e pension funds record a higher degree of volatility, which exposes pensioners to interest rate volatility. 18 vicente lazen jofré and álvaro clarke de la cerda

19 Real Annual Share Return of Type c and Type e Pension Funds 25% 20% 15% 10% 5% % % -10% -15% -20% -25% Type C funds Type E funds A specific case of investment risk arises when an individual, at the time of retirement, faces particularly low rates, yet must make an irreversible decision, as in the case of an annuity. The following figure shows that for six-month periods, the average half-yearly differences in sales rates for normal old-age annuities, which average around 5%, can exceed 10% in some months. For example, if, in January 2009, a future pensioner waited six months to process his pension, he would face a drop in the average normal old-age annuity sales rate of approximately 14% (from 3.55 to 3.05%) and would therefore have to accept a pension that was 5% lower. In the last few years, however, the pension interest rate risk calculation has been lower since rates have become more predictable again after settling at an alltime floor The estimation isolates the effect of a higher or lower rate. Of course, someone who waits six months also sees an increase or decrease in his or her pension due to the pension fund returns in those six months, as well as the increase in age, which increases the pension. pension plans: history and perspectives 19

20 Six-Month Variation in Annuity Interest Rates ( ) 25% 20% 15% 10% 5% 0% 01/ / / / / / / /2014-5% 01/ / / / / / / / % -15% -20% -25% SPENDING RISK During retirement, expenses are not as predictable or smooth as one might assume. In particular, medical expenses can be particularly onerous and are harder to project (or model) with any certainty relative to other expenses. No pension plan directly considers insurance against excessive expenses as a result of overspending or health catastrophes. 16 The average hospitalization rate for people under the age of 60 (that is, the probability of being hospitalized) was 9.2% in As age increases, so does the hospitalization rate. A 60-yearold man has an 11% probability of being hospitalized, which is similar to the probability for a woman of the same age. At 75 years of age, however, a man has a 25% probability of being hospitalized and a woman, 20% RESEARCH ON INDIVIDUAL CHOICE Having described the risks faced by individuals at retirement, we now explore how affiliates make decisions in terms of which pension plan they contract. As indicated in the last section, economic research has centered on rationalizing the behavior of a hypothetical individual, who makes dynamic optimization 16 To cover part of catastrophic health expenses, Decree Law 3,500 establishes that all pension, independent of the pension plan, are subject to a uniform tax rate of 7% of taxable income, which is allocated to financing health care. 17 See Superintendency of Health (2006). 20 vicente lazen jofré and álvaro clarke de la cerda

21 decisions that are conditioned not only on his or her own willingness to take on risk, but also on individual expectations in terms of longevity, expenses, and additional future income. In particular, analysts have focused on understanding the conceptual foundations that inform how individuals make decisions on contracting an annuity that transfers their resources to a third party in the case of Chile, to a life insurance company in exchange for eliminating their investment and longevity risk, versus a different type of plan that does not imply a transfer of savings to a third party. Research in this area has largely centered on the idea of the annuity puzzle, first described by Menahem Yaari (1965). According to Yaari, this puzzle arises because even though annuitization (that is, the contracting of a life annuity) is the optimal pension decision under certain assumptions (for example, that the individual has no bequest motives), very few consumers, or future pensioners, voluntarily contract an annuity. Basically, the magnitude of the demand for annuities, which has been studied especially in the United States, does not coincide with Yaari s optimization model, which gives rise to the puzzle he described. One of the explanations that has been explored is the bequest motive, or the desire to leave wealth to one s children, since annuity schemes in their purest form do no allow for this possibility. However, different studies have produced contradictory conclusions. Lockwood (2010) finds that the main reason that people do not contract an annuity is that they place a high value on leaving a bequest, in which case the optimal allocation of wealth to an annuity is zero in his model. Ameriks et al. (2007) reach a similar conclusion. In contrast, Hurd (1989) find no difference in the pension decision of people who want to leave an inheritance and those with no bequest motive. Brown and Warshawsky (2004) conclude that people who state that they want to leave their wealth to their descendants do not differ significantly in their choice of plan. Finally, Butler et al. (2007) and Brown, Mitchell, and Poterba (2001) indicate that there is not enough evidence to conclude either way on whether the bequest motive influences individual choice. The phenomenon of adverse selection has also been suggested as an explanation for the annuity puzzle. According to this behavior, the people who opt for an annuity are those who have the highest probability of needing it, that is, of living longer than the average of the rest of their cohort. Consequently, those who have better information on their health status and genetic history would tend to opt for the pension plan that best fits their reality. Those who estimate, based on this information, that they will live less than the average for their generation will choose a non-annuity plan. Finkelstein et al. (2004) and pension plans: history and perspectives 21

22 McCarthy et al. (2002) find evidence of adverse selection using data on annuities purchased in Japan, the United Kingdom, and the United States. Despite the plausibility of their results, however, in reality the vast majority of future pensioners do not have more or better information on their life expectancy. Thus, while the adverse selection hypothesis can contribute to explaining the choice of pension plans among certain population groups, it cannot provide a general solution to the annuity puzzle. Other studies try to solve the puzzle through the optimization model construction, that is, by improving the behavioral model proposed by Yaari and incorporating additional variables in the model. Examples include Horne et al. (2007), who finds that even in the absence of a bequest motive, it will not be optimal for an individual to allocate 100% of his or her savings to an annuity. We conclude that while the seminal problem proposed by Yaari has not yet been fully resolved, it has provided a key orientation for economic research thanks to its conceptual basis for understanding the mechanisms of personal choice. This field of economic research has strong potential for contributing to the design of public policies and, in the commercial area, to the development of new and better pension products. 22 vicente lazen jofré and álvaro clarke de la cerda

23 3 HISTORICAL EVOLUTION OF PENSION CONTRACTS This section analyzes historical trends in the selection of pension products over the past several years, including the trend toward choosing pension plans via the online Pension Consultations and Offers System (Sistema de Consultas y Ofertas de Montos de Pensión, or scomp). We also present a study of the returns available for pensioners, concluding with a look at the competitive performance of the annuity segment from a historical perspective. 3.1 PLAN SELECTION OVER TIME To study the selection of pension plans, the source of information that allows us to extract the most accurate conclusions is the scomp system, since the pensions contracted through offers accepted via scomp correspond solely to pensioners who could effectively choose their pension plan. 18 The following figure shows the number of regular and early old-age pensions contracted using the scomp system in each quarter, from the third quarter of 2004 to the second quarter of Simple observation of the figure shows that programmed withdrawals have not varied much in terms of their share in the pension distribution over the last ten years, with the exception of a sharp increase in late 2008 and early In any given period, a significant number of future pensioners do not meet the requirements for choosing an annuity; these consumers must receive a programmed withdrawal pension and are not eligible to use the scomp system. According to statistics from the Superintendency of Securities and Insurance, there were 93,929 new pensioners in 2014, of which just 41% used the scomp system (38,872 pensioners). pension plans: history and perspectives 23

24 Regular and Early Old-Age Pensions Contracted through the scomp System, by Quarter 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% T 3Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q 2014 Programmed withdrawals Life annuity Temporary income with deferred annuity A second trend that is evident from the figure is the increase in the share of deferred annuities since However, the deferment period chosen by new pensioners is relatively short. The next figure shows that the number of retirees who chose a deferred annuity with a deferment period of four years or longer is only slightly over 1% of the total. This suggests that the use of the deferred annuity in Chile is based on a decision to delay the fixed but lower pension in order to have access to greater liquidity for a short time. Share of Deferred Annuities in Total Annuities Contracted in 2014 Deferred 4 years or more, 1% Deferred 3 years, 6% Deferred 2 years, 18% Immediate annuity, 49% Deferred 1 year, 26% 24 vicente lazen jofré and álvaro clarke de la cerda

25 Plans with an annuity guarantee period are increasingly popular. According to the following figure, 82% of the annuities contracted using the scomp system include a guarantee period. Life Annuities with a Guarantee Period Contracted in 2014 through the scomp System No guarantee period, 18% Other periods, 7% months, 28% months, 28% months, 28% 3.2 RETURNS OFFERED OVER TIME During the retirement period, the financial returns of the system reflect the yields earned on the investment instruments in which the pensioners savings are invested. In the case of programmed withdrawals, the interest rates on fixed-income instruments have a strong impact, insofar as notes and bills make up a significant part of the pension fund portfolios and, therefore, of the investment shares that back the programmed withdrawals. In the case of annuities, the interest rates on fixed-income instruments determine the yields earned by the life insurance companies on their own investments and therefore, in a competitive market, the sales rate offered to pension system members. Fixed-income returns are thus a determinant of the potential pension level, so the fact that the interest rates on long-term fixed-income instruments are currently at historically low levels is not encouraging. The following figure shows the evolution of annuity sales rates and real long-term fixedincome interest rates, represented by the market rate (mr) calculated by the Superintendency of Securities and Insurance for the purpose of establishing a market benchmark for the constitution of technical reserves. 19 As the figure 19 As of June 2015 the market rate will no longer be used for this purpose, but it was so used during the period of analysis. In addition, having a long data series on this rate provides a bench- pension plans: history and perspectives 25

26 shows, real interest rates were almost always over the range of 6% in the 1990s, but they fell sharply from 2000 to 2004, landing at around 3%. In the last ten years, average annuity interest rates, or sales rates, have not gone above a ceiling of 4%; they have been under 3% since 2013 and are currently at 2.23%. Interestingly, since rates began to drop in 2002, annuity sales rates have been higher than the market rate, reflecting the strong competition in the annuity segment, which was not the case in the earlier period. Market Interest Rate (mr) and Annuity Sales Rate (ar) ( ) 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% MR Average Regular Old-Age Sales Rate These data have a very significant impact on pension amounts, since, on average, each point of difference in the annuity sales rate translates into 9% to 11% less in the pension amount in any of these plans, depending on the contract conditions. Thus, an affiliate who received a pension of Ch$500,000 in 1993 would today receive just Ch$373,000 with the same savings, based on the declines in the sales rate from levels over 5.0% to around 2.5%. At the same time, programmed withdrawal interest rates have suffered a similar trend over the period, given that they are indirectly based on the longterm market interest rate. In this case, the calculation methodology has been modified successively. The most recent change, in December 2013, brought back the uniform rate indicator, but based on rate vectors and a quarterly frequency. 20 mark for the risk-free rate on de long-term instruments. 20 Until 1995, the superintendencys calculated differentiated rates for each pfa. Starting in 1996, they calculated rates by fund but then adopted a uniform rate for all the pfas. In 2010 they began to use a weighted rate methodology with a discount rate for each pfa and an implicit annuity rate. Subsequently, they adopted a methodology based on risk-free rate curves incorporating corporate spreads. 26 vicente lazen jofré and álvaro clarke de la cerda

27 3.3 RESEARCH ON ANNUITY COMPETITION (MONEY S WORTH OF ANNUITIES) Over and above the yield trend of the fixed-income instruments available in the market for investment by life insurance companies, the companies can offer their clients a larger or smaller share of these profits in the form of an annuity pension income. It is therefore important to assess the degree of competition in the industry, by using the money s worth ratio (mwr) or the present value ratio of the money invested in an annuity to determine whether the annuitants are receiving a good income. The variables needed for this calculation are the risk-free interest rate structure, the assumptions in the mortality rates, and the actual income payments associated with different levels of pension savings. These are used to contrast the present value of the payments received by the pensioner with the alternative cost of these resources in the financial market. As discussed earlier, the single premium finances a flow of potential payments whose probability of occurrence is given by the possibility that the annuitant will be alive at the time of payment, in accordance with the following equation from section 1: 110 SINGLE PREMIUM = x=65 RV P x (1 + TV) (x-65) + CM In the mwr exercise, the affiliate has the choice of purchasing either an annuity or a hypothetical fixed-income instrument, whose flows are exactly equal to the annuity offered by the life insurance company, weighted by the probabilities of being alive, but subject to discount rates equivalent to the market discount rate structure. This is called the expected present discounted value (epdv): EPDV = RV P x (1 + i where i it is the market interest rate at time t. t t ) (x-65) 110 x=65 The mwr is defined as follows: 21 mwr = single premium / epdv. The higher the mwr, the greater the return offered by the annuity relative to the hypothetical instrument with similar characteristics. James, Martínez, and Iglesias (2006) study this issue for Chile. They find an mwr of in 1999 and in Morales, Rocha, and Thornburn 21 An alternative and perhaps more intuitive definition of the mwr is the gap between the annuity rate and the internal rate of return on risk-free instrument flows. This indicator is calculated as follows: PU= (RVxP x )/(1+r t -m) (65-x), where m is the endogenous variable the magnitude of the movement of the yield curve that makes the annuity flows equal to the single premium. pension plans: history and perspectives 27

28 (2007), who use a larger data set including an updated mortality table, find similar results for 1999 and 2003, as well as an mwr of in 2002, in 2004, and in Taken together, these works suggest that pensioners in Chile earn a good return on their savings in the form of an annuity in comparison with the mwr of other countries, such as the United Kingdom (0.896; James and Song, 2001), Australia (0.879; James and Song, 2001), or the United States (0.859; Brown, 2001). An important factor to bear in mind here is that annuities are indexed in Chile, in contrast with the annuity products available in most other markets. On the other hand, Cassasus and Walker (2013) argue que the lack of adjustments for liquidity and insolvency risk leads to an overestimation of the value of this ratio: In Chile, in particular, it is overestimated by at least 7% once this factor is taken into account THE IMPLEMENTATION OF THE SCOMP SYSTEM The changes introduced by Law 19,934 of 2004, which legally reformed Decree Law 3,500, include a requirement to implement the online Pension Consultations and Offers System (Sistema de Consultas y Ofertas de Montos de Pensión, or scomp), with the aim of providing more and better information to pension system affiliates and promoting competition in the market for pension plans. A number of studies have evaluated the effects of the scomp system on the pension industry in Chile. For example, using data from 2003 to 2006, Morales and Zucal (2009) find that price competition (the sales rate) has indeed increased, given that a high share of retirees chose one of the three best offers. They further indicate that broker fees charged by annuity brokers have converged to levels below the legal maximum. Finally, they find that price dispersion fell significantly following the implementation of the system. In another study, Halcartegaray and Miranda (2011) indicate that the percent difference between the first pensions under the two different types of plans (that is, programmed withdrawals versus annuities) has become a key determinant in the choice of plans since the implementation of the scomp system, which could help explain the reduction in the probability of contracting an annuity. Current data from the scomp system tends to support the hypothesis of stronger competition. In 2014, fully 83% of the annuities accepted by retirees were external offers (that is, a second offer by a company that exceeds its 22 The authors also argue that the Chilean regulations do not adequately measure leverage and risk-taking incentives, which could help explain the higher indicators for Chile (due to the companies offering higher sales rates). 28 vicente lazen jofré and álvaro clarke de la cerda

29 original offer made through the scomp system), while just 17% accepted the original offer. In terms of annuity broker fees, the scomp system has had the effect of bringing down commissions to below the legal limits. 23 From July 2013 to June 2014, the average percent commission charged by pension consultants was 1.0% for programmed withdrawals and 1.8% for annuities. Insurance company agents, in turn, charged a commission of 1.4%, on average, in the same period. Data provided by the Superintendency of Securities and Insurance reveals that when a pension consultant negotiates a programmed withdrawal plan, just 3% of the contracts include a commission under the legal maximum, versus 15% of annuity contracts. In the case of insurance company agents, 40% of the contracts charge commissions under the legal maximum. 23 Pension consultants can charge service fees if the affiliate (or his or her beneficiaries) contracts a programmed withdrawal plan or a sales commission for an annuity. Service fees are charged directly to the affiliate (or beneficiaries) with money from the individual capitalization account, whereas the annuity commission is paid by the life insurance company. When a pension is contracted, the allowable commission is 2.0% of the amount allocated to finance the annuity, with a maximum of uf 60; or 1.2% of the balance allocated to finance a programmed withdrawal pension, with a maximum of uf 36 (Executive Decree nº 782 of 2010). The total cannot exceed uf 60, even in the case of a temporary income with a deferred annuity or an immediate annuity with programmed withdrawals. This ceiling also holds for a switch from programmed withdrawals to an annuity: 2% of the balance of the new plan less the percentage paid for the previous consultancy, with a maximum of uf 60 less the ufs paid for the first consultancy. Annuity brokers can choose to charge a fee if the affiliate (or his or her beneficiaries) contracts an annuity in the company for which the agent works, depending on the agreement between the company and the agent. pension plans: history and perspectives 29