CBPP Projections Show Long-Term Budget Outlook Has Improved Significantly Since 2010 But Remains Challenging
We’ve issued an updated version of this report available here.
Under current budget policies, the nation’s fiscal outlook is stable for the rest of this decade and then worsens gradually, according to CBPP’s new long-term budget projections.
Policymakers should not ignore the long-run budget problems, which remain challenging. No deficit or debt crisis looms, however, and promoting further labor market improvements remains the nation’s most immediate economic concern. Policymakers should therefore avoid too much deficit reduction too soon, which would weaken the economic recovery, and focus deficit-reduction efforts on measures that take effect after the labor market has more fully recovered.
Policymakers should avoid too much deficit reduction too soon.Under our projections of current policies, the federal debt will be virtually flat in relation to the economy for the next several years and then slowly rise. The ratio of debt to gross domestic product (GDP) — which was 74 percent at the end of fiscal year 2014 — will drop slightly to 73 percent by 2017 but grow to 92 percent by 2040, we project. That’s a marked improvement over the situation just five years ago (see Figure 1), but policymakers need to take further significant steps to address the problem.
A stable — or declining — debt-to-GDP ratio is a common goal for fiscal stability. Although a rising debt ratio is advantageous when the economy is operating well below its potential, as it was in the Great Recession and ensuing sluggish recovery, a rising debt ratio in a strong, high-employment economy, in contrast, reflects an unsustainable budget policy that ultimately jeopardizes financial stability and long-term growth. Policymakers should reduce projected debt-to-GDP ratios through carefully designed policies that strengthen the economic recovery in the near term, while putting in place equitable and balanced deficit reduction that grows in size over time. (See box.)
These long-run budget projections are not a prediction. Rather, they show what will likely happen, under reasonable expectations of how the economy will perform in coming decades, if policymakers continue current laws and policies — that is, without reducing projected deficits or expanding them (by cutting taxes or boosting spending without covering the cost).
Our new projections update those we published in May 2014 to reflect the latest Congressional Budget Office (CBO) ten-year and long-term budget projections, the latest projections by the Social Security and Medicare trustees, changes in budgetary policies, and other recent developments. On a comparable basis, our new projections are very similar to last year’s. The technical note at the end of this paper provides more information about how we made the projections.
The Revenue Outlook
Federal revenues are projected to continue at their current level of a bit over 18 percent of GDP in 2016 through 2025. After 2025, two trends — rising real incomes that push people into higher tax brackets (so-called “real bracket creep”) and growing, taxable withdrawals from tax-favored retirement accounts by an aging population — will help push up revenues gradually as a percentage of GDP. By 2040, they are projected to reach 19.4 percent of GDP, close to their level in the final years of the Clinton Administration. (See Figure 2.)
The Spending Outlook
Federal outlays are projected to rise from 20.6 percent of GDP in 2015 to 23.6 percent of GDP in 2040. Only about one-third of the rise stems from primary, or non-interest, spending — that is, spending on programs that pay benefits to ordinary Americans and carry out the functions of government. (See Figure 2.) The bulk of the rise stems from net interest, as interest rates rise from historic lows and the federal debt gradually mounts.
The composition of federal non-interest spending will also change significantly by 2040. Because of an aging population and rising health care costs, Social Security, Medicare, Medicaid, and health insurance subsidies will grow substantially — both as a percentage of GDP and as a share of total federal spending — while all other programs as a whole will shrink. Social Security and the major health programs, which today account for 53 percent of non-interest spending, are projected to reach 69 percent of the total in 2040, with all other programs representing a correspondingly smaller share.
Social Security. Benefits under the Old-Age, Survivors, and Disability Insurance programs (together known as Social Security) will rise slowly but steadily in the next two decades — from a bit under 5 percent of GDP today to just over 6 percent in the 2030s — and then stabilize. That pattern largely mirrors the aging of the population and is dampened by the scheduled rise in the program’s full retirement age — which was historically 65, is now 66, and will climb to 67 between 2017 and 2022. (Each year that the full retirement age is raised lowers benefits across the board for future retirees by about 7 percent, regardless of whether they claim benefits early or work until the full retirement age or beyond).
The Debt-to-GDP Ratio
Generally, the debt-to-GDP ratio should rise only during hard times or major emergencies and then decline during good times. That enables the government to combat recessions through tax cuts and spending increases and to alleviate hardship during bad times, while creating a presumption against policies that markedly increase the debt during good times.
A stable debt-to-GDP ratio is a key test of fiscal sustainability. Increases in the dollar amount of debt are not a serious concern as long as the economy is growing at least as fast. Between 1946 and 1974, for example, debt held by the public grew significantly in dollar terms but — thanks to economic growth — plummeted as a percentage of GDP, from 109 percent to 24 percent.
Some suggest that certain debt-to-GDP ratios have a particular meaning in terms of their effect on the economy. In reality, there are no absolute thresholds.
Until a few years ago, for instance, many pointed to a 2010 analysis by economists Carmen Reinhart and Kenneth Rogoff suggesting that debt-to-GDP ratios of 90 percent or more are associated with significantly slower economic growth. But the authors have acknowledged computational errors in their original work and clarified that there is no “magic threshold” for the debt ratio above which countries suddenly pay a marked penalty in terms of slower economic growth. To the extent that countries with higher levels of debt experience slower growth, there is not much evidence that the high debt caused the slow growth; the reverse is just as likely to be true — that the slow growth caused the high debt — or some combination of the two effects.
Similarly, some analysts call for a debt ratio of 60 percent of GDP or less, a goal that the European Union and the International Monetary Fund (IMF) adopted some years ago. No economic evidence supports this or any other specific target, however, and IMF staff have made clear that the 60 percent criterion is arbitrary and should not guide near-term fiscal policy in the wake of the recent financial crisis, which drove up government debt worldwide. IMF recently stated, “Our results do not identify any clear debt threshold above which medium-term growth prospects are dramatically compromised.”a
All else being equal, a lower debt-to-GDP ratio is preferred because of the additional flexibility it provides policymakers facing economic or financial crises and the lower interest burden it carries. But all else is never equal. Lowering the debt ratio comes at a cost, requiring larger spending cuts, higher revenues, or both. That is why we emphasize the importance of not only the quantity but also the quality of deficit reduction, which should not hinder the economic recovery or cut spending in areas that can boost future productivity or harm vulnerable members of society.
a Andrea Pescatori, Damiano Sandri, and John Simon, Debt and Growth: Is There a Magic Threshold?, International Monetary Fund WP/14/34, February 2014, p. 4.
Medicare. Net outlays for Medicare benefits — that is, total payments minus the premiums that enrollees pay — are expected to rise from 3 percent of GDP today to 5 percent of GDP in 2040. Medicare fundamentally faces the same demographic pressures as Social Security. But Medicare faces an extra cost pressure: the tendency of medical costs, fueled by technological advances and increased utilization, to outpace GDP growth. The cost controls and delivery system reforms in the Affordable Care Act (ACA), plus other developments in health care delivery, are expected to curb (though not eliminate) that pressure. Our projections are based on current law and assume that policymakers will retain the ACA’s cost-control provisions.
Medicaid, CHIP, and health insurance subsidies. Medicaid — a joint federal and state program — provides acute health care coverage and long-term supports and services to eligible low-income people, while the Children’s Health Insurance Program (CHIP) covers many low-income children through capped grants to states. The ACA expanded the reach of Medicaid, at state option, and created new state-based marketplaces to enable millions of people without other coverage to buy health insurance at reasonable prices and without exclusions for pre-existing conditions or other restrictions that often made coverage unaffordable.
In the short term, the ACA expansion of enrollment in state-based marketplaces will push up spending for this trio of programs from 2.2 percent this year to 2.5 percent in 2017. After 2025, demographic and cost pressures will lead this category of health spending to reach 2.9 percent of GDP in 2040.
The fact that health care costs remain the largest driver of increased future spending should not obscure how dramatically their projected costs have fallen over the last few years. As Figure 1 shows, in January 2010 we projected that debt would exceed 200 percent of GDP by 2040; we now project less than half that ratio. Much of the improvement is from lower health care costs: in January 2010 we projected that Medicare and Medicaid together would cost 11.1 percent of GDP in 2040, but (based on the latest projections from CBO and the Medicare actuaries) we now project that Medicare, Medicaid including the ACA expansion, CHIP, plus the new marketplace subsidies will together cost 7.9 percent of GDP, or about 30 percent lower than the previous estimate. (See Figure 3.) This development has substantially improved the long-run fiscal outlook.
Other program spending. This category includes hundreds of programs for which Congress appropriates funding on an annual basis — known as defense and non-defense discretionary programs — as well as entitlement or mandatory programs such as SNAP (formerly food stamps), pensions for federal civilian and military retirees, veterans’ disability and education benefits, the refundable portions of the Earned Income Tax Credit and certain other tax credits, Supplemental Security Income (SSI) for poor elderly and disabled people, unemployment insurance, Temporary Assistance for Needy Families (TANF), farm price supports, and various smaller programs.
Over the next ten years, this broad category — which spiked to nearly 14 percent of GDP in 2009, during the economic downturn — is projected to fall as a percentage of GDP from 9.2 percent in 2015 to 7.4 percent in 2025. Both these figures are well below the 11.1 percent average of the last four decades. Almost all of the drop from 2015 to 2025 occurs in discretionary spending and is concentrated between now and 2021, as the caps and sequestration provisions of the 2011 Budget Control Act (BCA) squeeze defense and non-defense programs alike, and as the war in Afghanistan and similar military operations continue to wind down. Spending for the mandatory programs in this part of the budget also drifts down as a percentage of GDP, though less precipitously; unlike Social Security and the major health programs, most other mandatory programs do not face particular demographic or cost pressures, and some — such as unemployment insurance and SNAP — shrink naturally as the economy recovers.
After 2021 (for discretionary programs) and after 2025 (for the entire “other program spending” category), we assume that outlays keep pace with inflation and population growth — in other words, that real spending per person remains constant. That’s consistent with the historical pattern: we’ve generally found that these categories of spending rise faster than inflation and population growth only a) if Congress affirmatively acts to increase these programs, which is by definition not consistent with a projection of current law or policy; or b) during recessions, when unemployment insurance and similar automatic stabilizers rise temporarily but then fall back to normal levels when the economy recovers. Keeping pace with inflation and population growth implies a continued downward drift in this spending category as a percentage of GDP, from 7.4 percent in 2025 to 6.2 percent in 2040.
Interest. Unlike every other spending category, net interest doesn’t reflect explicit funding decisions by policymakers. Instead, it’s jointly determined by the amount of borrowing fueled by policymakers’ other spending and revenue decisions (in other words, by the debt) and by the interest rates set in financial markets.
Today, federal net interest costs represent 1.2 percent of GDP, almost matching the historic lows posted in the 1950s through early 1970s, when federal debt was far smaller. But today’s low interest rates, which are holding down borrowing costs, will not last forever. As a result, by 2025, net interest costs are expected to climb to 2.7 percent of GDP, even though the debt hardly rises (from 74 percent to 76 percent of GDP) during that period. By 2040, we expect net interest to reach 3.3 percent of GDP and debt to reach 92 percent of GDP.
Assuring Solvency for Social Security and Medicare
Assuring long-run solvency for the Social Security and Medicare Hospital Insurance (HI) trust funds would substantially improve the long-run budget picture. Like other organizations’ long-term projections, ours assume that full benefits will continue to be paid even after those trust funds are exhausted. Nevertheless, the programs lack legal authority to pay full benefits in that situation. Their trustees project that the HI fund will be exhausted in 2030 and the combined Social Security trust funds in 2034. In those years, incoming revenues would support 86 percent of Medicare HI benefits and about three-quarters of Social Security benefits.
Bringing the Social Security and HI trust funds into financial balance — through tax increases, benefit cuts, or some combination of the two — would forestall much of the projected rise in the debt-to-GDP ratio. If Social Security and HI expenditures equaled their revenues in each year after the projected depletion of those trust funds, federal debt would peak at 84 percent of GDP in 2033 and decline to 79 percent of GDP by 2040. The “Trust Fund Solvency” line in Figure 4 assumes that solvency is restored to the trust funds abruptly, through a sudden benefit cut or tax increase once the assets of the trust funds are depleted. Since, by law, benefit payments cannot exceed amounts available in the trust funds, it is indeed plausible to assume that, one way or another, solvency will be restored to the trust funds. Phasing in some combination of additional revenues and lower benefits more gradually, starting sooner, might produce slightly lower debt ratios than those shown here.
To summarize, policymakers can avert about three-quarters of the projected 18-point increase in the debt ratio through 2040 by restoring solvency to the trust funds through revenue increases, benefit reductions, or a combination of the two.
Uncertainty of Long-Run Projections
Users of these or any long-run budget projections should keep in mind that they are highly uncertain. CBO recently estimated, for example, that if productivity in the economy grew by ½ percent a year less or more rapidly than it projects, the debt ratio in 25 years would be about 17 percentage points higher or lower. Thus, the debt ratio in 2040 under current budgetary policies could easily be as high as 110 percent of GDP or as low as 76 percent, given our projection of 92 percent.
Likewise, if interest rates over the 2016-2040 period are one-half of a percentage point higher or lower than we project, the debt ratio in 2040 would be roughly 10 percentage points higher or lower than we project, all else equal. Since other critical variables such as health care costs are also inherently difficult to predict, the actual range of estimating uncertainty surrounding these long-run projections is even greater.
In addition to uncertainty about the economic future, considerable policy uncertainty surrounds our projections. As the technical note explains, our projections approximate a continuation of current laws and policies. But consider the “tax extenders,” a set of tax provisions primarily benefiting businesses that policymakers routinely extend for a year or two at a time, most of which expired at the end of 2014. Suppose that policymakers revive and continue these extenders without offsetting their cost. Suppose also that policymakers repeal or circumvent the sequestration of discretionary funding, both defense and non-defense, without offsetting those costs, and that war costs continue at current, real levels indefinitely instead of winding down. Those three additional costs would increase the 2040 debt ratio by 7 percentage points, 8 percentage points, and 5 percentage points, respectively, producing a 2040 debt ratio of 112 percent of GDP rather than 92 percent.
|Different Economic or Policy Assumptions Change the Projected 2040 Debt Ratio Percentages of GDP|
|Change in debt as a percent of GDP||Resulting Ratio: 2040 debt as a percent of GDP|
|0.5% higher annual productivity growth||-16||76|
|0.5% lower annual productivity growth||+18||110|
|0.5% higher Treasury interest rates||+10||102|
|0.5% lower Treasury interest rates||-9||83|
|New policy costs, not offset:|
|Revive normal tax extenders||+7|
|Cancel sequestration of appropriations||+8|
|Do not phase down war costs||+5|
|New policy savings, $1,650 billion over 10 years||-20||72|
Note: Figures may not add due to rounding.
Source: CBPP projection based on data from Congressional Budget Office, Social Security and Medicare Trustees, and Joint Committee on Taxation
On the other hand, policymakers could stabilize the debt ratio over the next quarter century by enacting $1.65 trillion in further program cuts and revenue increases over the coming decade. Those savings and the resulting debt-service savings would reduce the 2040 debt ratio by about 20 percentage points, to 72 percent of GDP. (See Table 1.)
Policymakers shouldn’t ignore long-run budget projections just because they’re uncertain. After all, some of the important underlying trends — notably the aging of the population and rising health costs — are highly probable, even if we can’t precisely predict their magnitude. But the uncertainty grows dramatically as the time horizon expands. That’s why we and CBO focus on the next 25 years or so for long-run budget estimates, a period that amply documents future fiscal pressures and presents a reasonable horizon for policymakers.
We base the first ten years of our projections on CBO’s most recent baseline budget projections, published in August 2015. We adjust those projections in a few respects to better reflect current tax and spending policies. Specifically, we assume that improvements to refundable tax credits (the Child Tax Credit, Earned Income Tax Credit, and American Opportunity Tax Credit) in the 2009 Recovery Act — which policymakers have extended twice — will not expire after 2017 but will be permanently extended and that the United States will reduce troop levels in Afghanistan and elsewhere in the Middle East to 30,000 by 2017. Each of these adjustments is based on alternatives published by CBO.
In addition, our projections assume that funding for non-entitlement programs will grow with both inflation and population after caps on this category of funding expire in 2021. In contrast, the ten-year CBO baseline adjusts these programs only for inflation.
Here are some important adjustments we don’t make to CBO’s ten-year projections. We do not assume that policymakers will breach the BCA caps on discretionary funding or repeal sequestration; we assume that the costs of any relief in this area will be offset. Similarly, we do not assume that policymakers will continue the normal tax extenders without paying for them. Finally, we do not remove CBO’s $61 billion for emergencies from its baseline, which we believe is a modest — and possibly low — figure for future, unexpected emergency costs.
CBO’s ten-year projections — on which we base our estimates through 2025 with the adjustments explained above — don’t extend past that date. Therefore, we base our extrapolation through 2040 on other recent sources. Those include CBO’s most recent long-term outlook (published in June 2015) and the latest reports of the Social Security and Medicare trustees (published in July 2015). Because our new ten-year projection is based on CBO’s August projections, it has slightly different values for spending and revenues in 2025 (the “jump-off” point) than CBO’s long-term outlook, which was published in June and based on earlier economic and technical budget assumptions. Likewise, our jump-off point for Social Security and Medicare has slightly different values for spending in 2025 than those used by the trustees. We therefore take our 2025 values and grow them at the same rate, as a percentage of GDP, as the applicable CBO or trustees’ projection. As a result, whatever (small) difference exists in the 2025 value will persist over time. Our approach thus avoids a sudden discontinuity in the projections.
Specifically, we assume that after 2025:
Revenues grow with the projections in CBO’s June 2015 long-term extended baseline.
Social Security and Medicare costs grow with those from the trustees’ intermediate projections.
Medicaid, CHIP, and health insurance subsidy costs grow with those in CBO’s June 2015 long-term extended baseline.
“Other program spending” — for defense and non-defense appropriations, plus outlays for mandatory programs other than Social Security and the big health programs — grow with inflation plus population, thus keeping real per-capita spending constant at 2025 levels.
For net interest, we calculate how much borrowing results from the revenue and spending totals already calculated, and we apply the overall interest rate on federal debt, assuming the continuation of the average interest rates that CBO projects over the 2021-2025 period.
Table 2 shows our projections for each major category of the budget as a percent of GDP between 2000 and 2040. Data for each year, including historical values since 1962, are posted on our website.
Overall, our new projections are very similar to those we published in May 2014, on an apples-to-apples basis. At that time, we projected that the debt would reach 101 percent of GDP in 2040. However, that projection assumed the normal tax extenders would all be continued permanently, without being paid for. Had we assumed those costs would be offset, last year’s debt ratio for 2040 would have been 95 percent, much like the 92 percent we now project. (Alternatively, if we now assumed again that the extenders would all be revived and become permanent without being paid for, our current debt projection for 2040 would be 99 percent of GDP.)
|Outlays, Revenues, Deficits, and Debt as a Percent of GDP Through 2040|
|Social Security||Medicare||Medicaid, CHIP, and exchanges||Other program outlays||Total program Outlays||Net interest||Revenues||Surplus (+) / Deficit (-)||Debt held by the public|
Source: Historical data from the Office of Budget and Management; projections from CBPP based on data from Congressional Budget Office, Social Security and Medicare Trustees, and Joint Committee on Taxation.
How Our Projections Differ From CBO’s
In June, CBO issued a set of long-term projections showing that, under its “extended baseline” scenario, debt would rise to 103 percent of GDP by 2040. Our figure is 11 percentage points lower than CBO’s, with nine of the 11-point difference occurring after 2025. There are three reasons we project somewhat lower debt than CBO.
- Roughly two-thirds of the improvement comes from lower interest rates. CBO based its June long-term projection on the economic assumptions it issued in January, while we use the interest rates CBO issued this August. CBO’s newer figures imply an interest rate on new debt of about 3.8 percent from the mid-2020s on, while its earlier figures imply about 4.0 percent. Since interest compounds over time, CBO’s reduction in assumed interest rates makes a noticeable difference.
- The remainder of the improved outlook comes from our projection methodology for discretionary appropriations. To begin with, as noted and unlike CBO, we assume war funding will phase down over the next two years as some troops withdraw from Afghanistan and the Middle East.
- In addition, we assume that discretionary spending — both defense and non-defense — keeps pace with population growth and inflation after 2021, as mentioned above. And while CBO’s projection of this category of spending does not account for population growth between 2021 and 2025, CBO assumes this spending grows with GDP after 2025; this entails faster growth than if this spending tracks population growth and inflation.
 Richard Kogan et al., CBPP’s Updated Projections Show Long-Term Budget Outlook Is Significantly Improved But Remains Challenging, Center on Budget and Policy Priorities, May 5, 2014, https://www.cbpp.org/research/cbpps-updated-projections-show-long-term-budget-outlook-is-significantly-improved-but.
 Jim Nunns and Jeff Rohaly, Why Individual Income Tax Revenues Grow Faster Than GDP, Tax Policy Center, September 1, 2015, http://www.taxpolicycenter.org/UploadedPDF/2000351-why-individual-income-tax-revenues-grow-faster-than-gdp.pdf.
 See “Why Does Raising the Retirement Age Reduce Benefits?” in Kathy A. Ruffing and Paul N. Van de Water, Social Security Benefits Are Modest, Center on Budget and Policy Priorities, May 5, 2015, https://www.cbpp.org/cms/?fa=view&id=3368.
 Robert Greenstein, Joel Freidman, and Isaac Shapiro, Program Spending Outside Social Security and Medicare, Already Low in Historical Terms, Is Projected to Fall Further, Center on Budget and Policy Priorities, March 12, 2015, https://www.cbpp.org/research/program-spending-outside-social-security-and-medicare-already-low-in-historical-terms-is.
 Of course, as our nation becomes wealthier, it might choose to spend increasing amounts per person on infrastructure, research, education, and other programs in this category. But such a choice — which could keep this category from falling as a percent of GDP — is not consistent with the historical pattern, and (in the case of benefit programs) would require enacting increases in benefit amounts or expansions in eligibility, whereas we project spending based on current law. See Appendix 2 of Kogan, Ruffing, and Van de Water, Long-Term Budget Outlook Remains Challenging, But Recent Legislation Has Made It More Manageable, Center on Budget and Policy Priorities, June 27, 2013, https://www.cbpp.org/research/long-term-budget-outlook-remains-challenging-but-recent-legislation-has-made-it-more. Auerbach and Gale — who do assume in their budget projections that such spending keeps pace with GDP — clearly state that this assumes future policy shifts, reflecting a choice by a “wealthier and more populous society” to maintain spending as a share of GDP. Alan J. Auerbach and William G. Gale, The Fiscal Problem: Gone Today, Here Tomorrow, Tax Policy Center, September 2015, p. 12.
 The separate Disability Insurance trust fund is expected to be exhausted in late 2016, the much larger Old-Age and Survivors Insurance fund in 2035. Combined, the two funds could pay full benefits until 2034.
 Social Security and Medicare Boards of Trustees, Status of the Social Security and Medicare Programs, A Summary of the 2015 Annual Reports, July 22, 2015, http://www.ssa.gov/oact/trsum/.
 Congressional Budget Office, The 2015 Long-Term Budget Outlook, June 2015, p. 98, https://www.cbo.gov/publication/50250
 This simulation assumes that half of the $1.65 trillion in policy savings would come from revenue increases, one quarter from Medicare cuts, and one-quarter from cuts in mandatory programs other than Social Security. We assume these policy savings would start small in 2016, be fully phased in by 2019, and thereafter grow at the baseline growth rates of revenues, Medicare, and other entitlements, respectively. A different distribution of savings would likely produce broadly similar results.
 Congressional Budget Office, An Update to the Budget and Economic Outlook: 2015 to 2025, August 2015, https://www.cbo.gov/publication/50724. We use CBO’s August 2015 baseline because it is the most recent, reflecting the enactment of the Medicare Access and CHIP Reauthorization Act after CBO’s prior baseline was issued in March, as well as CBO’s newest economic and technical assumptions. If, however, our figures for the first ten years had started with CBO’s March 2015 baseline, our 2040 debt ratio would have been 94 percent of GDP rather than the 92 percent we use in this analysis; CBO assumed higher long-term interest rates in March than it does now.
 CBO’s baseline and the BCA appropriations caps include a special, limited disaster allowance to cover the normal operations of FEMA’s Disaster Relief Fund and related funding. Emergency funding, in contrast, is fully outside this allowance and is needed only for large and unexpected events that cannot be handled by regular disaster relief, such as Hurricane Sandy and the Ebola outbreak. CBO’s baseline projects emergency funding to equal the amount provided last year in response to the Ebola outbreak, adjusted for inflation. Continuing this funding, as we do, does not mean that we expect future Ebola outbreaks, but rather that we anticipate one or more unexpected emergencies sometime in the coming decade that will require total funding of $61 billion outside the normal disaster allowance.
 See CBO, https://www.cbo.gov/publication/50250, and Social Security Administration, http://www.socialsecurity.gov/OACT/TR/2015/index.html.