Adapting Social Security Policy for the Long Term

of 25
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information Report
Category:

Comics

Published:

Views: 6 | Pages: 25

Extension: PDF | Download: 0

Share
Description
Adapting Social Security Policy for the Long Term
Tags
Transcript
   - 1 - Appendix  Adapting Social Security Policy for the Long Term September, 2007 Steven W. Popper 1 Senior Economist, RAND Corporation Introduction From the perspective of policy analysis, three groups of factors affect the ability of Congress to legislate wisely for the long term. The first are those related to Congress’s institutions, processes and political interests that have been detailed in other contributions to this series. The second is the difficulty analysts face in providing definitive findings to policymakers as the time horizon lengthens. The third is the difficulty of meshing the styles and requirements for rigorous analysis with the needs and language of legislators and decision makers. The future of Social Security 2  provides the perfect storm conditions that bring all these into view.  Although this brief is based on current research exploring how to resolve the difficulties raised by this intersection, it is not a report of that research. 3  No prescriptions on specific policies will be given nor any claims made that the analysis itself is definitive. Rather, as a proof-of-principle illustration (and of only one aspect of the larger Social Security issue at that) it is intended to make concrete and demonstrate the viability of a shared vision of process for assisting Congress that is implicit in several briefs that have previously appeared in this series. This vision is worthy of serious consideration in any program to enhance Congressional capability to legislate for the future. The discussion begins by specifying the problem faced by policy analysis when confronting deep uncertainty 4  and why this poses serious obstacles to any deliberative body such as the U.S. Congress. The brief then outlines an analytical method designed to be practical while making explicit that which everyone knows: we cannot be sure what the future will bring. Instead of deciding a priori what the “best guess” future might be and then choosing policies to optimize based on those conditions, the method systematizes the type of inductive thinking that responsible and dedicated legislators and policymakers actually engage in by asking   The Legislating for the Future Project is an initiative of New York University’s John Brademas Center for the Study of Congress and the Organizational Performance Initiative, and is co-sponsored by the Brookings Institution and the RAND Corporation. The project will examine the capacity of Congress to address long-term problems facing the nation, probe the public’s attitudes towards Congress’ ability to make long-term decisions for the 21 st  Century, and analyze specific long-term policy issues. The Legislating for the Future Project will convene experts for discussions of specific long-term issues, such as global warming, and seek to generate strategies to make Congress more flexible and adaptive to future problems. The Advisory Committee for the project is headed by Former Representative Lee H. Hamilton. The project is funded by the John Brademas Center for the Study of Congress, the Smith Richardson Foundation and the Carnegie Corporation. For more information, please visit: www.nyu.edu/wagner/performance and www.nyu.edu/brademas.   The views and opinions of the author do not necessarily state or reflect those of the co-sponsoring or funding organizations.  tacitly or out loud, “What if..?” The discussion then describes briefly a reasoning process that is iterative, interactive, and inclusive of different classes of knowledge, expertise and perspectives on the world. The object is to reveal potential solutions that perhaps may not be optimal for any particular set of assumptions or values but will nevertheless meet minimal criteria for acceptability, set by the political process, across a wide range of the plausible futures we may well confront. The brief concludes with thoughts on how incorporation of robustness thinking and explicit recognition of uncertainty may affect legislative processes. While the focus of the following discussion is Social Security, the true subject is how to conduct an analysis of policies that play out over the long term. Exchanging our usual question “what is likely to happen?” for a better question (“given that we cannot reliably predict what will happen, what is our best course for the short term?”) is a subtle transformation that permits systematic, quantitative analysis of an issue notoriously resistant to such treatment. Perhaps most important, it creates an avenue for framing policies that are designed from the outset to be both flexible and adaptive, a crucial value when legislating for the future.  ANALYSIS OVER THE LONG TERM OF LONG-TERM POLICY The title of this brief would seem redundant. After all, Social Security, by its very nature is an instrument designed to operate over the long term. Current workers’ contributions are not banked in a conventional sense. Rather, they are used to pay the benefits of current Social Security recipients. Later, those who contributed earlier are paid back from those who are then currently in the work force. By design, the system is inherently intergenerational and long term.  Yet, when we consider measures to ensure the ability of this system to match its revenues and savings to current and future entitlements (“solvency”) we face a paradox. The system extends over the long term yet our ability to perceive and act over great expanses of time are quite limited. Social Security solvency is but one example of a type of issue that consistently confounds our ability to analyze and frame wise policy. For illustration we need look no further than the 2006 (or, indeed, any) annual Social Security Trustees’ Report. 5  Though only required by law to look out five years, the Trustees wisely choose to look out over a considerably longer term to scan for possible shortfalls. They possess data on demographics, statistics on insurance utilization, and sophisticated models and forecasting tools. But they do not possess perfect foresight. Small shifts in assumptions well within the range of plausibility produce dramatically different conclusions over decades. Using variable values to represent what the Trustees term the Intermediate Cost case, the trust funds 6  set aside to cover Social Security 2   2  obligations become exhausted in the year 2040. At that point, any resources to cover additional obligations would presumably have to come from general revenue sources if benefits were not to be adjusted to fit within the means available. If the High Cost assumptions are applied, the exhaustion would occur a decade earlier, in 2030. On the other hand, utilizing the Low Cost assumptions, the trust funds never go negative and there would be a positive and growing balance of nearly $8 trillion in the year 2050. Clearly, these three forecasts also have three different implications for policy. One suggests staying the present course while the others suggest the need for short-term attention while small changes may still suffice to correct our course. What makes this even more problematic is that the range of outcomes spanned by these High, Intermediate and Low Cost cases actually cover only a relatively narrow band of the range of plausible futures. Plausible changes to certain key variables would have the power to place the future that actually transpires well beyond either the High or Low Cost cases framed by the Trustees Report. The Trustees and their analysts would be among the first to acknowledge this fact. In this sense, our usual means for analysis appear inadequate to help Congress reliably evaluate what courses of policy to consider. Generating multiple, credible forecasts is easy and the Trustees’ Report is at great pains to do so in as sophisticated a manner as present art allows. What they cannot provide are reliable predictions. Legislative deliberation is stymied if political discourse becomes an irresolvable debate over which presently unknowable set of future values are the correct assumptions for the purpose of planning short-term policy. Worse yet may be a state where the Trustees’ Intermediate Cost case, based on one set of assumptions, is taken as “truth” and the debate is over how to handle the resulting insolvency. The addition of ideological and political concerns completes a recipe for friction, inaction or ill-advised policy actions. ROBUSTDECISION METHODS AID POLICY CHOICE  A recent general audience survey article on economic modeling raised troubling questions for model builders and the consumers of model output. 7  It portrayed a Red Queen’s race in which ever more sophisticated generations of models seek enhanced predictive capabilities – only to see the prize consistently remain out of reach. Yet, even if such models were considerably more complicated than they currently are they still could not generate reliable predictions for even just a decade from now. That being the case, it is worth asking what use any 3   3  model can be for illuminating policy choices over issues as complicated as those embraced in the Social Security solvency problem. The root of the apparent problem lies not in the models themselves but in how they are used. Most models are created to play a role in a “predict-then-act” process of analysis: The steps are to first develop the most accurate possible model of the system of interest, gather data and make assumptions, generate predicted outcomes, and then apply the tools of optimization to find the best course for action. Is this a reasonable use of a model, however, when prediction is not credible? If we have optimized for one, supposedly most-likely, future and the actual future turns out differently, we can hope that our previously optimal plan will still be serviceable but we have no proof that this will, indeed, be the case. We might find ourselves bound on a course that is clearly deleterious to our interests given the way the state of the world has changed from what had been expected.  Yet, predictions themselves are rarely what we seek. Rather, what we would wish to understand is how changes in the future might affect our choice among alternative actions today and how the actions we do take will affect our chances of being successful in meeting our goals. Our true interest, once we acknowledge that we cannot be sufficiently predictive, is to understand how we can choose today’s actions most wisely in light of our long-term objectives. The shift in focus from model forecasts to informing decisions is subtle yet resolves many conundrums. Instead of determining the “best” model and solving for the strategy that is optimal (but fragilely dependent upon assumptions) we should instead seek among our choices those actions that are most robust – that achieve an agreed level of goodness across the multiple models and assumptions consistent with known facts. This is closer to the actual policy reasoning process. What we need from a model is not a prediction. Rather, a model serves as an artifact that contains what we understand about critical relationships among key factors and that can then be used to generate the myriad scenarios of the future that are consistent with our current information. As we systematically vary assumptions about factors whose future values are presently unknowable, we generate an ensemble of alternative futures -- a test bed for helping select among policy alternatives. Rather than characterizing uncertainties at the beginning of the analysis either by assigning values, assuming probability distributions, or dropping them entirely pending later analysis, we leave the uncertainties uncharacterized in terms of probabilities but nevertheless explicitly represent them in the model. The focus of the analysis then becomes not what assumptions we should choose but rather what conditions we would need to believe were likely in order to favor policy 4   4  “A” over policy “B” – and how we might construct a policy “C” that may relieve us of the need to choose. 8 HOW CAN ANALYSTS AID LONG-TERM DECISIONS? It is convenient to map factors we deem important in the analysis into four categories:  X   (or e  X  ogenous) factors outside our control that may affect outcomes in the future and render some strategies superior to others in retrospect after a decision has already been taken;  L evers constituting actions under our control that may either be combined with others by variations in composition, degree and sequencing into alternative strategies or else explored individually on their own; Cause-and-effect  R elationships between the actual state of the world (represented by those characteristics explicitly explored as  X category factors) and actions we may take (  L ) that yield the outcomes we wish to measure (  M  ); and  M  easures for assessing whether outcomes resulting from taking actions (  L ) within a particular environment (  X  ) meet our criteria for goodness or not. This is a category for exploration because no single measure may be sufficient to satisfy the criteria for goodness held by the parties to a decision. Similarly not everyone will agree on what weights+ to place on different measures of outcome success or failure. The XLRM framework serves several purposes. It is a transparent check list that will allow others to observe with precision what unknowns and assumptions are being modeled. It provides an intellectual bookkeeping system permitting further refinement by parties to the analysis of their own perception of the problem and issues to be addressed. Perhaps most of all, it serves as a template for design of the analytical tooling that will be used to support the investigation of alternative policies under a variety of emergent conditions. Table 1 presents the XLRM design being used in the RAND study. It is quite limited in scope and so offered only as an illustration. It is framed around solvency, by no means the only important Social Security issue. A full policy analysis of Social Security, even if only limited to the solvency issue, must contain a richer set of elements in each quadrant before its findings could have credible implications for policy. Possible Future States of the World (“  X  ”) The academic fields of demography and economics both wield powerful tools that often succeed in identifying trends and predicting 5   5
Recommended
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x