Significant uncertainty surrounds the estimates under the intermediate assumptions, especially for a period as long as 75 years. This appendix presents a way to illustrate the uncertainty of these estimates. The stochastic projections are intended to supplement the traditional methods of examining such uncertainty.
The Trustees Report has traditionally shown estimates using the low-cost and high-cost sets of specified assumptions to reflect the presence of uncertainty. These additional estimates provide a range of possible outcomes for the projections. However, they provide no indication of the probability that actual future experience will be inside or outside the range of these estimates. This appendix presents the results of a model, based on stochastic modeling techniques, that estimates a probability distribution of future outcomes of the financial status of the combined OASI and DI Trust Funds.
This model is subject to further development. Future improvements and refinements are expected to expand, rather than reduce, the indicated range of uncertainty.
Other sections of this report provide estimates of the financial status of the combined OASI and DI Trust Funds using a “deterministic” model. For the deterministic model, certain assumptions are made regarding levels of fertility, changes in mortality, legal and other immigration levels, legal and other emigration levels, changes in the Consumer Price Index, changes in average real wages, unemployment rates, trust fund real yield rates, and disability incidence and recovery rates. In general, each of these variables is assumed to reach an ultimate value at a specific point during the long-range period and to maintain that value throughout the remainder of the period. As mentioned above, three deterministic scenarios are developed by assuming separate, specified values for each of these variables. Chapter
V contains more details about each of these assumptions.
In contrast, the results of 5,000 independent stochastic simulations are presented in this appendix. Each of the 5,000 simulations is determined by allowing the above variables to vary throughout the long-range period. The fluctuation in each variable is projected by using standard time-series modeling, a method designed to help make inferences based on historical data. Generally, each variable is modeled by an equation that captures a relationship between current and prior years’ values of the variable, and introduces year-by-year random variation as observed in the historical period. For some variables, the equations also reflect relationships with other variables. Parameters for the equations are estimated using historical data for periods between 25 years and 110 years depending on the nature and quality of data available. Each time-series equation is designed so that, in the absence of random variation, the value of the variable would equal the value assumed under the intermediate set of assumptions.
1
For each simulation, values for most of the variables listed above are determined by using Monte Carlo techniques to randomly assign the year-by-year variations. The one exception is that net other immigration is modeled directly, rather than individually modeling its components, other immigration and other emigration. Each simulation produces an estimate of the financial status of the combined OASI and DI Trust Funds. Results shown in this appendix reflect the distribution of results from 5,000 simulations of the model.
The results from this model should be interpreted with caution and with an understanding of the limitations. Results are very sensitive to equation specifications, degrees of interdependence among variables, and the historical periods used for the estimates. For some variables, recent historical variation may not provide a realistic representation of the potential variation for the future. Also, results would differ if variables in addition to those mentioned above (such as labor force participation rates, retirement rates, marriage rates, and divorce rates) had been allowed to vary randomly. Furthermore, additional variability could result from incorporating statistical approaches that would more fully model shifts in the long-range central tendencies of the variables. The historical period utilized for most variables does not reflect many substantial shifts. The time-series modeling reflects only what occurred in the historical period. As a result, the variation indicated in this appendix should be viewed as the minimum plausible variation for the future. Substantial shifts, as predicted by many experts and as seen in prior centuries, are not fully reflected in the current model.
Figure VI.E1 presents simulated probability distributions of the annual trust fund ratios for the combined OASI and DI Trust Funds. The two extreme lines in this figure illustrate the range within which future annual trust fund ratios are estimated to occur 95 percent of the time (i.e., a 95-percent confidence interval). In other words, actual future trust fund ratios in a given year would be expected to exceed the upper bound only 2.5 percent of the time or to fall below the lower bound 2.5 percent of the time. Other lines in the figure display additional confidence intervals (80‑percent, 60‑percent, 40‑percent, and 20‑percent) around future annual trust fund ratios. The median estimate for each year indicates the trust fund ratio that this model projects will fall exactly in the middle of possible outcomes for that year. These lines do not represent the results of individual stochastic simulations. Instead, for each given year, they represent the percentile distribution of trust fund ratios based on all stochastic simulations for that year.
The median estimate for each year indicates that the assets of the combined OASI and DI Trust Funds would be exhausted by the end of 2035 with a probability of 50 percent. This exhaustion date is one year earlier than the year of exhaustion (2036) that is projected under the intermediate assumptions. Figure
VI.E1 shows that the 95‑percent confidence interval for the trust fund ratio in 2025 ranges from 301 to 119 percent of annual cost. In comparison, the 2025 trust fund ratios for the low-cost and high-cost alternatives are each outside this range, at 323 and 113 percent, respectively.
Figure VI.E2 displays the probability distribution of the year-by-year OASDI cost rates (i.e., cost as a percentage of taxable payroll). The range of the cost rates widens as the projections move further into the future, which reflects increasing uncertainty. The income rate under the intermediate assumptions is also included in the figure to give some indication of the patterns of cash flow for the OASDI program. Only the income rate is included, and not the cost rate, because there is relatively little variation in income rates throughout the projection period. The lines in figure
VI.E2 display the median set (50th percentile) of estimated annual cost rates and the 95‑percent, 80‑percent, 60‑percent, 40‑percent, and 20‑percent confidence intervals expected for future annual cost rates. These lines do not represent the results of individual stochastic simulations. Instead, for each given year, they represent the percentile distribution of cost rates based on all stochastic simulations for that year. The projected cost rates for the year 2035 are 14.62 percent of payroll for the low-cost alternative and 19.88 percent of payroll for the high-cost alternative. These cost rates are close to the limits of the 95-percent confidence interval (14.50 and 19.74 percent of payroll), as can be seen in figure
VI.E2. By 2085, the cost rates for these alternatives, 12.56 and 25.30 percent of payroll, are still close to the limits of the 95‑percent confidence interval (11.35 and 24.57 percent of payroll).
Table VI.E1 displays long-range actuarial estimates that illustrate uncertainty for the combined OASDI program using both the deterministic and stochastic approaches. Actuarial estimates included in the table are for the long-range period, 2011‑85. Stochastic estimates are shown for the median (50th percentile) and for the 95‑percent and 80‑percent confidence intervals. For comparison, deterministic estimates are shown for the intermediate, low-cost, and high-cost assumptions. Each individual stochastic estimate displayed in the table represents the level at that percentile from the distribution of the 5,000 simulations. However, for each given percentile, the stochastic estimates shown for the different long-range actuarial measures are generally not from the same stochastic simulation.
The median stochastic estimates displayed in table VI.E1 are, in general, slightly more optimistic than the intermediate-alternative deterministic estimates. The median estimate of the long-range actuarial balance is ‑2.14 percent of taxable payroll, about 0.08 percentage point higher than projected under the intermediate assumptions. The median year that cost first exceeds non-interest income (and remains in excess of non-interest income throughout the remainder of the long-range period) is 2016, 6 years later than projected under the intermediate assumptions. The median year that assets first become exhausted is 2035, one year earlier than projected under the intermediate assumptions. The median estimates of the annual cost rate for the 75th year of the projection period are 17.22 percent of taxable payroll and 5.74 percent of gross domestic product (GDP). The comparable estimates using the intermediate assumptions are 17.56 percent of payroll and 6.01 percent of GDP.
The 95‑percent confidence interval determined by the stochastic modeling projections can be compared to the range of variation defined by the traditional low-cost and high-cost alternatives. For three measures in table
VI.E1 (the actuarial balance, the open group unfunded obligation, and the first year assets become exhausted), the 95‑percent stochastic confidence interval is narrower than the range defined by the low-cost and high-cost alternatives. In other words, for these measures, the range defined by the low-cost and high-cost alternatives contains the 95‑percent confidence interval of the stochastic modeling projections. For one measure (the first year cost exceeds non-interest income and remains in excess through 2085), the low-cost and high-cost estimates are consistent with the bounds of the 95-percent stochastic confidence interval. For the remaining two measures (the annual costs in the 75th year), one or both of the bounds of the 95‑percent stochastic confidence interval fall outside the range defined by the low-cost and high-cost alternatives.
