A Framework for Comparing Social Security Reform Proposals:

        an introduction by Dallas L. Salisbury

President and CEO, Employee Benefit Research Institute (EBRI)

Chairman and CEO, American Savings Education Council (ASEC)



DECEMBER 4, 1996

Arnold & Porter Conference Center

555 12th Street, NW

Washington, DC 20004-1202

Any views expressed herein are the views of the author and do not necessarily reflect the views of the Employee Benefit Research Institute or its membership.

I express my thanks to Kelly Olsen of EBRI for her invaluable assistance in preparing this introduction.

I. Introduction

Policy reform comparisons take the form of both numeric-based and value-based analysis. Numeric analysis quantifies "nuts-and-bolts" policy outcomes. Value-based analysis builds on the results of quantification by assessing where predicted policy outcomes fall relative to certain normative criteria. That is, value analysis asks which reforms, given their predicted outcomes, are desirable as public policy against a given set of normative policy objectives. For example, value analysis would include questions such as, "Is the retirement income provided under a given plan 'adequate'?" and "Are the costs 'fair'?"

The model used by the Social Security Administration (SSA) Office of the Actuary has been the primary source of quantified data about Social Security. Social Security policy simulation modeling in the private sector has been limited, and the models have not been continuously maintained. Private models have usually been constructed to simulate either the current system or one policy alternative's outcomes.

Furthermore, the available models in both the public (SSA) and private sectors are "static," as opposed to "stochastic." Static models can predict outcomes under a few economic and demographic scenarios but cannot present a distribution of the wide range of scenarios that could arise from different combinations of economic and demographic variable values. Only stochastic models can measure the "risk" of their performance measure values, because stochastic models, using Monte Carlo methods, are based on probability distributions. [1] Risk is an important policy consideration because no one can predict the future with certainty.

As a result of the SSA's near monopoly on analysis, its numeric analysis results are the basis on which proposed reforms are most frequently compared against normative value objectives. Hence, Social Security reform comparison, both value- and numeric-based, is primarily dependent on the numbers generated by the SSA Office of the Actuary, which has limited resources and must respond on a priority basis to public-sector officials. As a result, the ability of the private sector to fully participate in reform debates has been limited.

The Employee Benefit Research Institute (EBRI) Social Security project has contracted to work with the Policy Simulation Group to create a complementary simulation model called the EBRI/SSASIM2 Stochastic Model. This model will be available to run simulations of various Social Security reforms. Like the SSA Office of the Actuary, the EBRI Social Security Reform project does not make value comparisons between reform proposals.

The EBRI/SSASIM2 model provides quantified data that will allow others to apply their normative values and goals for the assessment of each reform's

* benefit adequacy

* fairness of benefit distribution

* fairness of cost burden distribution

* overall program economic efficiency.[2]

Preliminary results[3] suggest the EBRI/SSASIM2 Stochastic Model is a reliable means of providing quantified comparison data. In fact, when the same assumptions are used, preliminary simulations of the costs and benefits under the current system closely resemble those generated by the model used at the SSA Office of the Actuary.

Because the answers to some important questions for Social Security reform comparison cannot be quantified, the EBRI Social Security project includes a variety of other considerations in its framework for comparison. These considerations are delineated in Section III. EBRI's framework sets the stage for policy discourse and debate by specifying the major categories of consideration that should be applied to each Social Security reform proposal if it is to be comprehensively and fairly compared with others.

II. EBRI/SSASIM2 Model Performance Measures

One of the components of the EBRI Social Security project's framework for comparing reform proposals is contained in the performance measures used by the EBRI/SSASIM2 Stochastic Model.[4] This simulation model uses nine measures of benefits and two measures of costs to compare each Social Security reform plan as well as the current system:

Benefit Measures:

* Lifetime Program Benefits: Actuarial present value of lifetime program benefits (adjusted for inflation and mortality).

* Benefit-Contribution Ratio: Actuarial present value of lifetime program benefits divided by the actuarial present value of lifetime payroll contributions.

* Net Benefits: Difference between actuarial present value of lifetime program benefits and the actuarial present value of lifetime payroll contributions.

* Net Benefits as a Percentage of Earnings: Net benefits minus contributions, divided by the actuarial present value of lifetime earnings.

* Internal Rate of Return: Internal rate of return on benefits given contributions.

* Average Benefit: Average annual real benefit over retirement years.

* Replacement Rate: Percentage of final year of preretirement earnings that is replaced by the average benefit for a continuously employed person earning average wages.

* Low Benefit Avoidance: Percentage of beneficiaries expected to have retirement benefits above a low-benefit threshold.

* Real Per Capita gross domestic product (GDP): Inflation-adjusted value of the GDP.

Program cost measures:

* Average Cost Expressed as a Percentage of Taxable Payroll

* Actuarial Balance Expressed as a Percentage of Taxable Payroll

Finally, for each proposal as well as for the current system, EBRI/SSASIM2 assesses the risk inherent in the individual performance measures by using Monte Carlo simulation methods that calculate different demographic and economic scenarios.[5] Risk is the likelihood that a performance measure will fall below a specified threshold value because of possible changes in human behavior, demographic factors, and economic variables. A reform that would have advantages only if demographic, human behavioral, and economic variables stay the same as they are today, for example, presumably would be less desirable than a reform that would produce advantages under a variety of possible economic and social outcomes.

III. Further Numeric Analysis Derivable from the EBRI/SSASIM2 Stochastic Model

Aside from the EBRI/SSASIM2 Stochastic Model's performance measures, there are other important considerations in making comparisons that the model is able to quantify. These considerations include breakdowns of aggregate benefit and cost data generated by the model. After identifying simulated cost and benefit distributions, these distributions can be compared with alternative program goals. The following considerations can be answered for each reform by disaggregating the model's quantified data output:

* How would reform benefits be distributed across the target beneficiary ranges under each proposal?

* Under a given reform, would all beneficiaries receive the amounts intended by program goals or would some fair unexpectedly well and some unexpectedly poorly?

* If implemented, would a given reform deliver benefits to all beneficiaries intended, or would there be unintended distinctions between those who do and those who do not receive benefits?

* What would be the cost distribution be under each reform? For example, how much of total cost would be administrative? How much would be transitional?

* Is the above in line with the reform's intent or are administrative, borrowing, transitional, operating, or other costs higher than anticipated by program advocates?

* Would cost burdens be distributed to groups and individuals according to the program's intent, or would some bear an inadvertent and disproportionate amount of burden?

* Under the reform proposal, would all persons pay into the program as intended by program goals, or would there be unintended distinctions between those who contribute and those who do not contribute to the new system?

* What would be the overall costs and benefits under each reform, not just from an individual perspective but from an intergenerational family perspective as well?

IV. Other Reform Proposal Comparison Considerations of the EBRI Social Security Project

Some policy outcomes are not easily quantified, yet they are nevertheless important considerations in making fair comparisons among Social Security reform proposals. EBRI's Social Security Project will study these potential outcomes in the months ahead. These include predicted program macroeconomic effects, economic implications of alternative revenue structures for transition funding issues, the effect of policy reforms on private defined benefit and defined contribution programs as well as on individual savings behavior, the potential effects of Social Security reform on other government programs, the likely impact of political forces on reform, and the possible effects of Social Security reform on other aspects of society.

Macroeconomic effects among various reforms have been a popular topic of debate since many reform proposals involve reallocating funds from investment in the public sector to investment in the private sector. In addition, some reforms seek to bolster the United States' personal savings rates in hopes of favorably impacting economic investment and growth. Given the interrelated nature of the components of the U.S. economy, attention must be given to each reform's impact on various macroeconomic considerations. These include each reform proposal's likely effect on:

* the bond, corporate, and equity markets

* the employment rate

* the federal deficit

* interest rates

* private borrowing

* overall economic growth rates

* labor-leisure choices

* savings-consumption choices

A Social Security system, like any retirement income program, can be designed in ways that influence the above macroeconomic variables. A system primarily accomplishes this through its incentives structures, which encourage or discourage certain human behaviors.

To demonstrate a national Social Security system's effect on individual decisions, consider that a plan offering the same benefits at age 62 as at age 65, all things being equal, would encourage early retirement, whereas a plan that raised annual retirement pensions by 10 percent for each year retirement is deferred would discourage early retirement. Due to the interconnected nature of macroeconomic systems, retirement rates, in turn, can influence a number of economic variables, from the employment rate to GDP. Retirement rates, depending on plan design, may also affect the duration of time that participants withdraw benefits from the program and the amount of benefits collectible in retirement, affecting overall program costs.

Similarly, pension plans can encourage either savings or consumption of workers' disposable income. Theoretically, if a program's minimum guarantee were, for example, a lump-sum distribution of $2 million, few workers would be likely to contribute to augment this minimum. On the other hand, if there were no minimum, workers might panic into saving every last penny for fear of starvation in retirement. In turn, a Social Security reform plan's savings-consumption incentives for workers and retirees may affect macroeconomic variables such as private investment and economic growth. However, changes in individual savings may not necessarily change overall national savings if higher individual savings rates are offset by additional federal debt. Federal debt plays an enormous role in predicting macroeconomic outcomes, as does the rate of interest accrual on that debt. In short, Social Security reform structures can have numerous effects on macroeconomic variables. Because of the interrelated nature of these variables, the effects may multiply, and hence they are important considerations when comparing Social Security reforms.

Another important concern, related to the potential macroeconomic effects of reforms (because of its possible impact on individual savings), is the possible effect of Social Security reform on the rest of the U.S. retirement system. Because many employment-based pension plans are designed to be integrated with the current Social Security system, any changes in Social Security will certainly affect employment-based savings. The following considerations arise:

* What will be the administrative burden on employers to integrate the new Social Security reform into their established systems?

* What will be the employer's tax burden under each proposed reform?

* In what ways could each reform affect individual participation in employment-based pensions?

* Will employers need to spend increased resources on participant education as a result of reform?

In addition, Social Security reform will affect individually initiated savings such as non-employment-based individual retirement accounts (IRAs). Ultimately, Social Security reform could impact the composition of the three-legged retirement stool.

The government will be impacted differently under the various reform options. Consideration must be given to each reform's effect on other government programs, both those that affect retirement income (such as Supplemental Security Income and Medicare) and others. In a society of finite resources, opportunity costs must be weighed for each reform. For example, a fair comparison cannot be made between two reforms, one that provides spectacular benefits and is actuarially balanced and one that provides fewer benefits and is also balanced, if it is not mentioned that the former program is so costly that other important programs are overwhelmed or abolished as a result. Possible opportunity costs need to be made clear for each reform.

Any Social Security system will be subject to political pressures. These "political risks" must be assessed to make accurate comparisons among Social Security reform plans. Among many others, these "political risks" include:

* For reforms that would have the government invest a portion of the Social Security trust fund in private equities, would there be special interest pressure to invest in certain equities and public pressure to engage in "social investing."[6]

* Would people pressure the government for pre-retirement access to individual accounts under a defined contribution-style Social Security reform?

* If the reform involves private market investment, will voters be able to accept natural market fluctuations and ride them out, or will the public panic during market down-turns and demand political action?

* Aside from actual economic efficiency, how would a proposed Social Security reform effect people's confidence in their national retirement system and in the government as a whole?

Finally, there are multiple social considerations involved in Social Security reform. These social considerations involve social goods that do not have a quantifiable price tag associated with them. Some examples are the following:

* Does the program promote financial literacy?

* Does the program promote a sense of national community?

* Does the program alleviate family burdens of supporting elderly family members?

* Does the program contribute to individual peace of mind in retirement planning and retirement?

* Does the program, by design, present its risks and benefits so that all participants are aware of them?

Public reaction to reform proposals will be affected by all the factors noted. Public education will be essential, and the EBRI project seeks to make that more achievable than it might otherwise be. For example, EBRI's November edition of EBRI Notes provides a point-by-point comparison table of seven Social Security reform initiatives, based on the best information available. The article, entitled "Keeping Track of Social Security Reform Proposals," is one of a number of EBRI initiatives to assist people at all levels of understanding to learn about the current Social Security reform debate.

V. Conclusion

This project would not be possible without the prior modeling initiatives by the Quadrennial Advisory Council on Social Security and the SSA Office of the Actuary. It builds on EBRI's work prior to the 1983 Social Security Commission, as well as on EBRI's published study, Social Security: Perspectives on Preserving the System (1982). It is but one more step in what has been, and what will be, a long road to secure the economic security of retired Americans through private and public action.

VI. Endnotes

1 It is conceptually straightforward to apply Monte Carlo (or stochastic) simulation methods to the Social Security current-law projection problem. The basic idea is to recognize that the future values of the assumption variables in a Social Security model are not known with certainty. The Trustees Report recognizes this fact, but grapples with it in an ad hoc manner that has been criticized by several review panels over the past decade. Monte Carlo methods deal with the fundamental uncertainty concerning the future values of key model assumption variables in a more systematic fashion that involves describing the uncertainty with probability distributions. Each probability distribution formalizes expectations about the future value of each assumption variable. Then the simulation model is used to produce a many-scenario run, in which random sampling from these probability distributions is used to generate the values of assumption variables for each scenario. During each scenario, detailed output and summary results are collected on whatever aspect of the simulation system is of interest. After a number of scenarios have been simulated and their results have been recorded, a distribution of output values (describing each output variable's risk properties) is available for statistical summarization or visualization.

2 Economic efficiency is achieved by the program that costs society least and benefits society most when compared with policy alternatives. Economic efficiency involves normative value assumptions because of the weighting of importance of individual costs and benefits involved in computing overall economic efficiency.

3 Martin Holmer, Development Progress Reports on the EBRI-SSASIM2 Model, October 11, 1996 and October 24, 1996, Policy Simulation Group.

4 This section of the paper refers solely to those quantitative criteria which will be used in the first phase of this study to compare the relative merits of alternative reform proposals against the status quo. Other components, such as private investment returns on mandated private savings, will also be generated by the model. See sections III and IV for more detail on the other components.

5 The model may also be used to isolate the impact of a particular scenario. For example, the model has the capability of looking at different mortality assumptions and their impact on the financial health of the Social Security system.

6 Social investing involves favoring investment in equities and stocks offered by companies engaging in "socially desirable" behaviors, such as recycling waste or building schools.

Contacting EBRI | Copyright & Disclaimer