Lee Anderson and Tuljapurkar 2003 — Stochastic Forecasts of the Social Security Trust Fund

Summary

Lee, Anderson, and Tuljapurkar extend the stochastic Social Security finance framework introduced in Lee and Tuljapurkar (1998) into a fuller treatment of Old-Age, Survivors, and Disability Insurance (OASDI) trust fund mechanics — modeling income rates, cost rates, and the trust fund balance stochastically rather than deterministically. The core finding is that the Social Security Administration's (SSA) three-scenario approach understates the true uncertainty around 75-year solvency outcomes, the official intermediate scenario sits below the stochastic median (official projections are optimistic relative to the model's central tendency), and fertility dominates long-run uncertainty while SSA ranks it last. This paper represents the intermediate step between the 1998 American Economic Review Papers and Proceedings (AER P&P) brief application and the 2004 infinite-horizon work with Miller.

Key Claims

• Probability distribution over trust fund outcomes: Combining Lee-Carter stochastic mortality with autoregressive integrated moving average (ARIMA)-based stochastic fertility, productivity, and interest rates generates a full probability distribution over the OASDI trust fund balance trajectory — not just high/medium/low scenarios.
• SSA intermediate is optimistic: The SSA's intermediate ("best estimate") projection corresponds to a favorable-tail outcome relative to the stochastic model's probability distribution; the stochastic median implies earlier trust fund exhaustion and a larger long-term actuarial deficit than SSA's intermediate.
• Fertility dominates uncertainty at 75-year horizons: As in Lee and Tuljapurkar (1998), fertility uncertainty is the largest contributor to 75-year long-term actuarial balance (LTAB) uncertainty, followed by productivity, interest rates, and then mortality — the reverse of SSA's own sensitivity ranking (which varies one factor at a time and uses narrow fertility scenarios).
• Trust fund depletion probability substantially elevated: The probability of trust fund exhaustion before the end of the 75-year projection window is substantially higher than implied by the official scenarios. The wide stochastic confidence intervals mean there is a meaningful probability of scenarios both considerably better and considerably worse than the intermediate.
• Income and cost rate mechanics: Explicitly models OASDI income rates (tied to wages, labor force participation, covered earnings) and cost rates (tied to mortality, disability incidence, retirement behavior) as stochastic functions of demographic and economic inputs, improving the structural fidelity of the simulation relative to purely demographic projections.

Concepts Introduced or Extended

• Long-Term Actuarial Balance
• SSA Mortality Forecasting

Entities Mentioned

• Ronald Lee
• Michael Anderson
• Shripad Tuljapurkar

Quotes

My Take

This paper fills the methodological gap between the brief 1998 AER P&P piece and the 2004 infinite-horizon work, providing a complete treatment of OASDI trust fund mechanics under stochastic inputs. The core message — that official deterministic scenarios understate uncertainty and that the intermediate projection is not the median of any reasonable probability distribution — was important for the Social Security policy debate of the early 2000s, when reform proposals were frequently benchmarked against the intermediate scenario as if it were a near-certainty. The uncertainty decomposition finding that fertility dominates (not mortality) has direct policy implications: immigration and labor force participation interventions may have larger solvency effects than the official framing suggests. Note: specific numerical results in this entry are based on the methodological sequence and contextual knowledge rather than direct PDF access; readers should consult the primary source for precise figures.